Matilal, Bimal Krishna - Logic, Language, and Reality - Indian Philosophy and Contemporary Issues-Motilal Banarsidass (1990) PDF

LOGIC, LANGUAGE AND REALITY
Indian Philosophy and Contemporary Issues

LOGIC, LANGUAGE AND
REALITY
Indian Philosophy and Contemporary Issues
BIMAL KRISHNA MATILAL
MOTILAt BANARSIDASS PUBLISHERS

PRIVATE LIMITEDtDELHI
First Edition: Delhi, 1985
Second Edition: Delhi, 1990
© MOTILAL BANARSIDASS PUBLISHERS PVT. LTD.

ALL RIGHTS RESERVED.
ISBN: 81-208-0008-7 (Cloth)

ISBN: 81-208-0717-0 (Paper)
Also available at :
MOTILAL BANARSIDASS
41 U.A., Bungalow Road, Jawahar Nagar, Delhi 110 007
120 Royapettah High Road, Mylapore, Madras 600 004
24 Race Course Road, Bangalore 560 001
Ashok Rajpath, Patna 800 004
Chowk, Varanasi 221 001
PRINTED IN INDIA
BY JAINBNDRA PRAKASH JAIN AT SHRI JAINENDRA PRESS, A - 4 5 NARAINA INDUSTRIAL
AREA, PHASE I, NEW DELHI 1 1 0 0 2 8 AND PUBLISHED BY NARENDRA PRAKASH
JAIN FOR MOTILAL BANARSIDASS PUBLISHERS PVT. LTD., BUNGALOW
ROAD, JAWAHAR NAGAR, DELHI 1 1 0 0 0 7
FOR
TAMAL AND ANVITA

CONTENTS
PREFACE TO THE SECOND EDITION ix
X1
INTRODUCTION ^
CHAPTER O N E : LOGIC AND DIALECTIC IN ANCIENT AND
MEDIEVAL INDIA 1
1.1 Ancient Indian Logic and the Question of
Greek Influence 1
1.2 Rules of Dialectic and Debate 9
1.3 Argument and Pseudo-Argument
(Nyäya and Nyäyäbhäsä) 23
1.4 The Early Nyäya Theory of Inference
(Nyäyasütra 1.1.5) 29
1.5 Pseudo-Reason {Hetväbhäsä) 42
1.6 What Do We Infer? 58
1.7 The Triple-Character'of Reason 68
CHAPTER Two: PROBLEMS OF PHILOSOPHICAL LOGIC IN
NAVYANYÄYA 76
2.1 Empty Terms 76
2.2 Reference and Existence 85
2.3 The Navyanyäya Logic of Property and
Location 112
2.4 The Problem of 'Universal Properties'
(kevalänvayiri) 128
2.5 Inference and Concomitance (vyäpti) 140
2.6 Double Negation in Navyanyäya 145
2.7 The'Difference'of Difference 155
2.8 Definition and Classification 164
2.9 Definition, Differentiation and Essence 176
CHAPTER THREE: PROBLEMS OF KNOWLEDGE AND
PERCEPTION 203
3.1 Awareness and Knowledge 203
3.2 Perceiving and Misperceiving 215
3.3 Knowing the External World 228
3.4 Memory 262
vijj Contents
CHAPTER FOUR: ONÏOLOGICAL PROBLEMS 268

4.1 Three Rival Ontologies: Nyâya, Buddhism and
Jainism 268
4.2 Causality 284
4.3 Substance (Jainism and Vaisesika) 294
4.4 Seven Ways of Non-Absolutism (Saptabhaiigi) 301
4.5 'Emptiness' and 'Many-sidedness' 314
4.6 Avidyä in Buddhism, Yoga and Nyäya 319
4.7 The Enigmas of Buddhism: Alayavijnanay Duhkha
and Nirvana 333
4.8 Transmigration and the Moral Enigma of
Karma 35]
CHAPTER FIVE: INTERACTION OI GRAMMAR AND
PHILOSOPHY 372
5.1 The Doctrine of Karana 372
5.2 Substance and Quality in Sanskrit Grammatical
Theory 378
5.3 Grammatical Categories: A Navyanyâya
Appraisal 389
5.4 The Notion of the Sentence (vclkya) 398
5.5 Grammaticality and Meaningfulness 416
SELECT BIBLIOGRAPHY 431
INDEX 439
PREFACE TO THE SECOND EDITION
Ï take this opportunity to note the important issues raised by

some of my reviewers and try to answer very briefly a few
constructive criticisms made by them.
The new sub-title underlines the fact that I have attempted to
reassess the problems of traditional and classical Indian
philosophy in contemporary setting. Hence it neither is a book
on history of philosophy nor a mere philological research on
texts of Indian philosophy. In the introduction to the first
edition, I emphasized upon philological issues. But that is
because it has been my firm belief that without sound philologi-
cal grounding original research in Indian philosophy is next to
impossible. I have studied the texts, however, as a philosopher.
It is true, as Professor Pranab Sen in his review in Mind
January 1989 has noted, that in Chapters Two, Three and Five, I
have dealt with what is 'particularly germane to contemporary
philosophic concerns'. However, in Chapter One, I deal with the
history of Indian logic, for history of logic in any tradition, I
believe, has, unlike the history of ideas, a separate philosophical
interest of its own. Jjn Chapter Four, I have been concerned with
certain broader question of Indian philosophy—the goal of
human life, the idea of freedom, moksa and samsara and the
notion of the self and non-self. Although these problems are
typically Indian, I still believe they ought to be of interest to
some modern philosophers (if, however, 'philosophy' is narrowly
defined to mean simply analytical philosophy or the so-called!
'naturalism' in modern philosophy then this would not be thd
case).
In view of certain critical remarks in the reviews, I have
reformulated some sentences in a couple of places. For example,
on page 211 my reformulation answers the objection raised by
Professor S. Bhättacharya in his 'Critical Notes' in J.I.C.P.R. in
1987, and others. I had said that a non-judgmental cognitive
state is not known to us. It has been pointed out that this goes
against the Nyàya view for it is said in Nyäya that such a state
is 'known' by inference. I still insist that such a state is not known
X Preface to the Second Edition
in the same way a cognitive state is usually known, i.e., by an

inner perception (called anuvyavasaya), according to Nyäya. This
type of non-judgmental cognition is not open to such inner
perception. The inference (which is given in the text) establishes
simply the fact that such an experience (a non-judgmental
cognition) sometimes takes place, even if the cognizer himself
remains unaware of it at the moment of its origin (or in the
immediately following moment). However, it is true that such a
cognitive state as a fact is among the knowables, and sometimes
such knowledge of it comes through inference.
I also accept Sen's point {Mind, 1989, p. 151) about the major
concern of the Indian logicians with 'education' (a variety of
induction) which I have called 'singular inference' following
Quine. But I hasten to add that Dinnaga and the logical theories
inspired by him did accommodate both kinds of inference: those
that have a singular proposition (sentence) as conclusion and
those that have a universal proposition (e.g., 'All noises are
impermanent') as conclusion. In Navya-nyâya, this distinction is
noted with tie lielp of the following terminology. What we call a
singular or a particular proposition today is characterized by
Navya-nyâya as a case where the inferred property (sadhya) is
co-present with the delimiting property of the subject or paksa.
Thus, in 'the hill has fire on it' the inferred property, i.e., fire
(presence of fire), is co-present with the delimiting property, hill-
ness, of the subject, i.e., the hill. This is satisfied if and only if
there is at least one locus, the particular hill in question, where
hillness is co-present along with fire. What we call a universal
proposition is treated as one where the inferred property is
present in all loci of the delimiting property of the subject of the
paksa. In other words,'all noises are impermanent' is analysed
as the case where impermanence, the inferred property, is present
in all loci where noise-hood, the delimiting property of the
subject, is present. This means that anything that is characterized
by noise-hood has also impermanence. This does not fully
answer the question raised by Sen. But I wish to discuss this
issue more elaborately in a forthcoming bqok of mine, The
Development of Logic in India,
The general form of the inference studied by both the Nyäya
and the Buddhist (as well as by Sarhkhya, Mimärhsä and Jaina)
was: A is B because of C. This can be unpacked into what is
Preface to the Second Edition xi
called nyäya-vakyä or Indian syllogism as :
A is B.
Because of C.
Whatever is C is also B, e.g. X.
{A is such, i.e., Ç,
Therefore A is so, C: i.e. B)
This is similar to a deductive argument and ensures deductive
certainty. Indian logicians however showed lack of interest in
the deduction technique, and instead concentrated upon the
examination of truth and falsity of the universal premiss. They
insisted that the supporting examples, both positive and negative,
would be necessary. Besides, they left it open that a counter-
example, if it can be found, would falsify the universal premises.
This is what I have called the empirical character of Indian
logic.
Regarding the question of universals, I wish to add the follow-
ing. I have chosen, following Quine, the 'natural kind' reading
of the Nyâya theory of universal, while Sen prefers, the following
Strawson, the 'sortal' reading of the Nyäya universals. I now
believe both readings are possible, and in either case, some
modifications of the traditional Nyâya doctrine would be requir-
ed. My preference for the 'natural kind' reading is based upon
the fact that I am not sure whether the reality of sortal universals
can be maintained and argued for against the Buddhist alterna-
tive, the apoha theory of Dinnaga and Dharmakirti. Again, the
matter requires more careful study, which I intend to do later.
Apart from logic and theories of knowledge, Indian philosophy
was intimately connected with a sort of soteriology. This may
divulge the pre-modern character of Indian philosophy. But the
issues raised inthis context are important to get a well-rounded
picture of the subject. I Mve discussed these issues in Chapter 4.
I believe this balances our modern discussion of Indian philoso-
phy, the charge of one-sidedness may thus be answered. I have
not (consciously) distorted anything, but I claim originality in my
presentation and in relating some of the topics with some con-
temporary issues. On the whole in our pursuit of modernity and
the post-modern trend, the authority of the past should not be
xii Preface to the Second Edition
forgotten lest we repeat ourselves and repeat even the mistakes of

our predecessors.
4 September, 1989 All Souls College,

Oxford
INTRODUCTION
This is a book on Indian philosophy. Its aim is to combine

the discipline of philosophy with that of philological research. It
is not, however, concerned primarily with historical research.
The book deals with the problems of Indian philosophy—pro-
blems which are necessarily philosophical, in the sense of being
logical, epistemological, ontological and soteriological. All the
problems discussed here are interconnected in some way or
other, and it is reasonable to expect that a systematic account of
classical Indian philosophy will emerge from a discussion of
such topics.
The book is meant for philosophers along with those interested
in Sanskrit, Indian Studies and what may be called a global
approach to the study of philosophy. I have begun with the
logical theories of ancient and medieval India. 'Logic' is used in
this context to mean the study of inference-patterns, rules of
debate, points of controversy, sophistical arguments etc. Some
may object to our use of the term 'logic' in this rather broad
sense. In using this term, however, I have been guided by other
recognized experts in the field. Thus,I.M. Bochenski has remark-,
ed in his History of Formal Logic:
"Formal logic (Nyäya-sästra) developed in India, as in Greece,
from the methodology of discussion. Such a methodology was
already systematically constructed in the second century B.C. The
first ideas which can be said to be formal-logical occur indeed
as early as the Vaisesika-sütra(first century A.D.), but the history
of Indian formal logic properly begins with the Nyaya-sütra
(edited in the second century A.D.). This 'logical' sütra (so charac-
terized by its very name) was the foundation of Indian logical
thought." (p. 417)
Problems of inference lead to the discussion of philosophical
logic. Historically, the early Nyâya in India was succeeded by
xiv Introduction
what we call Navya-nyäya in the 12th-13th century \ . D . Philo-
sophers of this period were involved in the discussion of such
problems as empty terms, reference-failure, double negation,
concomitance, definition, classification and essences. All these
issues constitute the second chapter here.
Logical theories of the above kind developed in the background
of a theory of knowledge. Medieval philosophers of India (of
such schools as Nyâya, Buddhist, Jaina and Mîmâmsaka) had
made extensive study of the problem of knowledge. I have given
only a very brief account of it in the third chapter. As I discuss
these issues more extensively in another forthcoming book.
Perception (Oxford University Press), I have only presented a
sketch of the Indian theory of knowledge here. Several peculiari-
ties of the Indian notion of knowledge and inference have been
noted in this connection.
The problem of knowledge leads to the question of what is
known. Chapter Four deals with oncological problems. I have
tried to present them in a somewhat new, non-traditional, way.
I have tried to bring together the three rival traditions: Hinduism,
Buddhism and Jainism. F. Kafka once said:
'The scriptures are unalterable and the comments often
enough merely express the commentator's bewilderment."
My new attempt may therefore be taken to be an expression of
my own 'bewilderment'. The discussion here inevitably leads to
the discussion of such broader questions as soteriology, the
meaning and goal of human life and the problem of suffering,
alayavijnâna, soul and karma. I have tried, within a very brief
space, to present the provisional answers given to such questions
by the classical Indian philosophers.
In the last chapter, I deal with the interrelation of grammar
and philosophy in India. This is a significant character of classi-
cal Indian philosophy. The grammarians used to have consider-
able influence upon the philosophers. Some grammarians (e.g.,
Bhartrhari) were well-known as philosophers. India was fortunate
to have Pânini and the Pâninîyas. I have tried to give some
examples to show how the insights of the grammarians were
combined with the problems of the philosophy of language.
In India, philosophy of language formed part of a comprehensive
theory of knowledge, i.e., theory of pramänas, for one of the
Introduction XV
ways of knowing what is the case is to rightly understand

what is said by an expert and trust-worthy person (äpta).
Following this line, the philosophers of knowledge in classical
India established a connection between the traditional problems
of knowledge and those in the philosophy of language. I conclude
with a few comments on modality seminaliy involved in the
notion of compatibility (yogyatâ) and grammatically of linguistic
expressions.
In several places, I have repeated one very general theme, of
which thediiferent issues discussed in this book provide only
different illustrations. The theme is the question of the relevance
of Sanskritic study in the context of modern philosophical
research. The importance of the study of the Sanskrit
philosophical texts in modern context can hardly be overempha-
sized, specially when even among the professional philosophers,
students and research-workers of India today, study of Indian
philosophy, in the way 1 conceive it here, has been much neglect-
ed and largely forgotten. In support, I wish to quote from the
writing of an erudite Indian pandit who was, in my humble
opinion, also a creative philosopher of our country. Pandit
Sukhlalji Sanghavi recommended openly "a non-partisan, histori-
cal-comparative study'' of any Sanskrit philosophical text, and
wrote in his Preface to Advanced Studies in Indian Logic and
Metaphysics (1961):
fc
T became firmly convinced thatAe study of any philosophical
system inevitably demands certain pre-requisites and that
these prerequisites include a fairly accurate under-standing of
. the historical inter-relationship obtaining between the various
philosophical systems of India."
Panditji's Preface should be read by all young scholars of our
country, specially by those who wish to work on any system of
Indian philosophy. 1 was myself deeply influenced by these
comments of Panditji, when I started, about twenty years ago,
my own research in Indian philosophy. One may add, however,
to Panditji's commeritthat a reasonably clear understanding of
modern (largely Western) philosophical problems should also be
included among the 'prerequisites'.
In writing this book, I have incorporated, either partly or
wholly, several articles written over the last twenty years and
xvi Introduction
published in various anthologies, festschrifts and commemora-
tion volumes. As they are by and large unavailable and as they
fit very well with the main argument of this book, I have included
their revised and sometimes re-written versions here. On the
whole, however, this is a new book, which aspires to be an in-
formative introduction to Indian philosophical problems—
problems that were discussed in the classical period.
Some portions of this book bear unmistakably the "vestiges"
of my "Indian" English style. I make this remark in awareness
of the fact that scholars in India as well as in other non-English
speaking countries have used such a style for a pretty long time.
In recognition of this fact, one well known Dutch scholar, C.A.O.
van Nieuwenhuijze (whose own English, by the way, was perfect)
made an elaborate comment which.is worth quoting:
"An additional remark on the linguistic appearance of this
book should be made here. Contrary to what at a first glance
may seem to be the case, this book is not ;n English. It has
been written in a lingua franca of the Western world of the
middle of the twentieth century, which by some is mistakenly
called English, particularly as far as its vocabulary and syntax
are concerned." (p. xii, Preface)
The remark has obviously been made facetiously. I wish to
add simply that my hesitation has been overcome only by my
awareness of the fact that the category of readers addressed to
in this book has largely come to expect a style like this from
Indian writers, specially those who deal with Sanskritic material.
In preparing this manuscript I have been considerably helped
by my former student, Mr. P.K. Sen of Calcutta University, as
well as by my wife, Karabi. I wish to thank both of them here.
CHAPTER ONE
LOGIC IN ANCIENT AND

MEDIEVAL INDIA
§1.1 : ANCIENT INDIAN LOGIC AND THE QUESTION

OF GREEK INFLUENCE
Unlike the Arabic tradition in Philosophy, Indian philosophic

tradition was never directly or indirectly influenced by either
Aristotle's writings or Aristotelianism. Yet there is an "uncom-
fortable" affinity between many problems discussed by Aristotle
(and Greek philosophers in general) and those discussed by the
Classical Indian thinkers. In Greek writings we hear about the
"Gymnosophists" (of India) who used to go around and debate
on philosophical, religious and moral issues. But in Sanskrit or
Pali writing we seldom hear about Greek philosophy; and never
about a Greek philosopher.
I have used the expression ''uncomfortable affinity" and per-
haps I should explain it further. It was Professor Krisel, who
remarked at the beginning of this conference, that from a dis-
tance many things, if not everything, look alike, but on closer
scrutiny, when we get to the details, they are all found to be
* different. A distinction should be made between superficial simi-
larity and what we may call essential affinity that deepens our
philosophie understanding. Superficial similarity may mislead
one into false beliefs. I shall cite a classic 'example' of such a false
belief—an example that will be of some relevance to the main
theme of this conference1.
1
Part of this section was verbally presented to the Round Table Conference
on the occasion of the 23rd Centenary of Aristotle's Death held in Paris,
organized by the UNESCO, May 31—June 3, 1978.
2 Logic, Language and Reality
In 1920, SKC. Vidyäbhüsana published his pioneering work A

History of Indian Logic (Calcutta), where in Appendix B, he
included his paper "Influence of Aristotle on the Development
of the syllogism in Indian Logic". An earlier, brief, version of
the same paper was published1 in the Journal of the Asiatic
Society of Great Britain and Ireland. Vidyäbhüsana argued, on
almost non-existent historical evidence, that Aristotle's works on
logic migrated into India at three different periods, in bits and
pieces, to "greatly" influence tho Indian logic, especially its
syllogism. This thesis of Vidyäbhüsana was preposterous, and
even Vidyäbhüsana himself was vaguely aware of its absurdity.
But the positive merit of Vidyäbhüsana's paper was to show
conformity and philosophic relevance of some aspects of what
he called "The Syllogism in Indian Logic" on the one hand, and
the "logical rules" of Aristotle on the other. Let me deal with
a couple of specific points from Vidyäbhüsana, for, they will be
instructive as far as our purpose is concerned.
Vidyäbhüsana first refers to the notion of sthäpanä and
pratisthäpanä found in the logical section of the Carakasarnhitä
(c. first century A.D.), and rightly translates them as 'demons-
tration' and 'counter-demonstration'. But then he goes on to
argue that these were taken from, or directly influenced by,
Aristotle's two species of enthymemes, viz., demonstrative and
refutative, found in Rhetoric, Bk. ii. Now it would be worth-
while to see what a 'demonstration' or a 'counter-demonstration'
stands for in Indian logic. For, then we shall be in a better
position to judge whether there was actually any influence here
from Aristotle's Rhetoric.
The standard form of demonstration, as recorded in the
Caraka, Nyäyasütra and some other contemporary texts, is called
the five-step argument procedure or 'five-membered argument'
sentence (cf., pancävayavaväkya). The steps or 'members', with
illustrations, are as follows :
Demonstration (sthäpanä)
1. Statement of the thesis or what is to be proved (pratijnä)
"The soul is eternal"
2. Statement of the reason (hetu)
"For it is non-produced"
iSee the reference in ' S.C. Vidyäbhüsana's A History of Indian Logic
(Calcutta, 1920), Appendix B.
Logic in Ancient and Medieval India 3
3. Exemplification (udäharana) or Citation of an example

to illustrate the logical relation between step 1 and 2.
"Whatever is non-produced is also eternal : witness,
the sky (space)."
4. Application of the rule to the present case (upanayd)
"The soul is also non-produced"
5. Conclusion (nigamana) : "Therefore the soul is eternal"
A counter-demonstration runs through the same five steps as
follows :
1. "The soul is non-eternaL"
2. "For, it is amenable to perception."
3. "What is amenable to perception is also non-eternal :
witness, a pot."
4. "The soul is also amenable to perception."
5. "Therefore, the soul is non-eternal."
It is clear that what we have before us has very little to do
withoratorial arguments or rhetorical enthymemes. It is also clear
that the counter-demonstration is nota refutative enthymemes,
for, nothing is refuted by it. It only establishes a proposition
which happens to be logically contradictory to the thesis of the
'demonstration'. As Aristotle clearly says, the demonstrative
enthymeme proves some affirmative or negative proposition,
but the refutative kind disproves one. The first is formed by the
conjunction of compatible propositions, while the second, by
the conjunction of incompatible propositions (cf., Rhetoric,
1396 b 25-29). Besides in Bk. ii, ch. 23, of Rhetoric, Aristotle
lists twenty-eight types of rhetorical enthymemes, none of which
seems closer to the patterns given above. Thus, I think, it is
impossible to imagine, as Vidyäbhüsana implicitly suggested,
that the author of the Caraka or the Nyäyasütra had read a
translated version of the Rhetoric. Besides, it is fairly certain
that no translated version of either the Rhetoric or any part of
the Organon, in any Indian language existed at that time.
It seems obvious, on the other hand, that demonstration or
counter-demonstration is not very far from Aristotle's concept
of dialectical argument. For, dialectic has been clearly explained
by Aristotle as follows :
"Dialectical arguments are those that reason from premises
generally accepted, to the contradictory of a given thesis."

(De Sophisticus Elenchus 165 b2-5)
Aristotle discussed the subject of demonstrative arguments in
the Analytics and that of dialectic arguments in the Topics and
sometimes in the Metaphysics. In the Topics, he notes that a
demonstrative argument is where the premises from which the
reasoning starts are 'true' and 'primary', or such that our knowl-
edge of them has originally come through premises which are
primary and true, whereas dialectical argument reasons from
opinions that are generally accepted (100 a 27-30). He further
elaborates that dialectic takes its start from plausible views
(endoxa), äs opposed to true and primary premises, and, thus,
it is able to debate both sides of the case, regardless of which, if
either, is the correct side1 (S.E.ii, 172 a 15-21). The concept of
'demonstration' and 'counter-demonstration5 in the Caraka is
similarly an argument-pattern that reasons from premises generally
accepted (see steps 2 and 3 in both cases), to prove a thesis or
its counter-thesis, and thus debates both sides of the case. It is,
for example, a generally accepted view that the soul, if it is
believed to be different from the body, etc., is non-produced (at
least, according to the Indian religious belief) and, therefore,
since non-producedness and eternalness go together, the soul is
eternal. Similarly, the soul is perceptible, for, according to
presumably the materialist, it is identical with the body, and
even for the non-materialists, it is argued that in such inner
perception "I see x" the seer (i.e., the soul) is also seen, i.e.,
mentally perceived. It may be contended by the materialist that
perceptible objects are generally perishable.
At the next stage, however, the debate moves to a different
level, for, the proponent or the first disputant may claim that
not all perceptibles are perishable; for example, uni versais
(admitted to be perceptible by Nyâya philosophers) are not. The
opponent or the second disputant may likewise answer by
claiming that such perceptibles do not really exist. Thus, we
can see that the Nyàya concept of 'demonstration' and 'coun-
ter-demonstration' is closer to that of dialectic than to that of
rhetoric, although we shall have to call it 'the Indian dialectic'
1
See alsoJ.D.G. Evans, Aristotle's Concept of Dialectic (Cambridge, 1977),
p.32.
for certain obvious peculiarities it has, viz., the requirement of

5 steps in each demonstration, the particular order of the steps
that is given here and must be observed according to its exponents
and so on.
While I talk of such parallelism, I do not intend to imply
either the influence of, or borrowing from, Aristotle. For, the
Indian theory of dialectic has, as modern researchers have shown,
an independent and indigenous history of development of its
own1. The so-called 'five-step' argument-pattern was reached
during the time of the Nyäyasütra through the process of
gradual elimination and critical modification of some elaborate
earlier model or models (with ten or more steps), which were
presented in the earlier texts (cf., reference to one model in
Vâtsyâyana and the Yuktidîpikâ; two other models in Bhadra-
bähu : Dasavaikalika-niryukti)2. Besides, it is easily discernible
that the 'five-step' demonstration differs from the syllogistic
demonstration of Aristotle in some important respects, although
the basic principle of inference involved in syllogism need not
be different from that in 'five-step' demonstration. Let me
illustrate the 'five-step' demonstration of the Nyäyasütra :
Example I
1. There is fire on this mountain.
2. For, there is smoke there.
3. Smoke goes with fire always (or, in all cases, or in all
places) : witness, kitchen2.
4. This is also a case of smoke.
5. Therefore, there is fire there (on the mountain).
We can call this to be a positive demonstration where an
affirmative thesis is established. The Nyäyasütra provides also
for a negative demonstration, where a negative thesis is reached
as the conclusion. I shall illustrate it with the help of Uddyota-
kara, ignoring Vätsyäyana's rather mistaken interpretation4.
1
One may consult E. Solomon's * Indian Dialectics, Ahmedabad, for the
historical materials relevant to such a study.
2
For Bhadrabâhu, see Vidyäbhüsana as well as E. Solomon.
3
I am taking the liberty of interpreting this step as 'embodying a universal
proposition with an example. This is generally recommended by traditional
(Sanskrit) scholarship. And I believe this to be an essential part of the
4
nyäya in the Nyäya-Sütras.
Vâtsyàyana's statement of the negative example is inaccurate and
misleading in many places.
Example II
1. This body is not without a soul.
2. For, this body is not without life.
3. Lack of a soul goes always, etc., with lack of life, wit-
ness : a piece of rock.
4. This is not f.o (not without life).
5. Therefore, this is not without a soul.
To underline the essential form of these arguments, we may
represent them schematically as follows :
I
1. A (or, A applies here).
2. Because B {or, B applies here).
3. B goes with A always, or in all cases, or in all places;
witness case C.
4. It is a case of B.
5. Therefore, it is a case of A.
II
1. Not A (or, A does not apply here).
2. Because not B (or, B does not apply here).
3. A goes with i? always, etc., witness C.
4. It is not so (not a case of B).
5. Therefore, it is not a case of A.
It is easy to see that both I and II (i.e., both examples) can
be transformed to fit into Aristotle's first figure and, thus, I can
be mapped into Barbara and II into Celarent. But, such
transformations, though certainly permissible and legitimate,
disregard certain important and specific characters of the model
of arguments examined by the Indian logicians.
Let me point out at least two significant features of the Indian
model, features that would not be immediately visible in their
mappings in the Aristotelian system, Barbara, and Celarent.
First, instead of presenting the thesis or conclusion in the usual
subject-predicate form, I have presented it, following the Indian
tradition, in what may be called a 'property-location' form. The
theory is operating mainly with the predicate and the reason,
i.e., the major term and the middle term in Aristotle's termino-
logy: The role of the so-called minor term, if there is any
such role here at all, is quite different (see F. Staal 1 ). The

major term or the 'predicate' {sädhya) is being used hereto
apply to a (particular) case or a (particular) place on the
ground that the middle term or the 'reason' (hetu) so applies.
This claim is being supported in the 'third' step by underlining
and exemplifying the invariable togetherness of the 'reason' with
the 'predicate'. In the second case, the 'predicate' is being used
not to apply to a case on the ground that the 'reason' does not
apply. And this is supported in the third step by underlining and
exemplifying the invariable togetherness of the lack of the
'predicate' with the lack of the 'reason'. The so-called minor
term is introduced here only by such expressions as 'here',
'there', 'in this case' or'in this place'. But the mapped images
of the two examples in Barbara and Celarent would be read as.
follows :
I
All S are F
This (M) is S
Therefore, this (M) is F
(where S ~ smoking things, F = things with fire,
M = t h e mountain).
II
No L is S
This (B) is L
This (B) is not S
(where L= Living things, S= soulless, B = body).
This brings us to the second major contrast. The thesis/con-
clusion in the Nyäyasütra demonstration is what is called a
'singular proposition, which is distinct from both the universal
(using such quantifiers as "all" or "each") and the particular
(using "some"; proposition. In Aristotle's system, however, the
conclusion is either a universal proposition ('A' or CE') or a
particular one (using 'some'). The singular propositions are
assimilated into either positive or negative universals, 'A'
or 'E', so that the syllogistic rules may work. In the early
F. Staal, "The Concept of Paksa in Indian Logic", Journal of Indian

Philosophy, August, 1973.
•"8 Logic, Language and Reality
Indian system, however, the conclusion is given, more often

than not, in a singular demonstrative form, such as, "This is
A" or "There is A here/there". Even a universal proposition,
such as, "All noises are temporary" or "Noise is temporary"
is assimilated into a singular form by such rephrasings as :
"There is temporariness in noise."
or "There is temporariness in whatever has noiseness."
or ''Temporarinessgoes with noiseness always (in all cases)."
Thusît is that when we focus our attention upon such details
and contrasts, the idea of Aristotle's influence upon the Indian
theory of logic seems highly improbable. In fact, we can almost
be certain and would refute such a hypothesis, when we consider
the complete lack of any evidence of any of Aristotle's work
being translated into an Indian language ( Sanskrit, Pali or
Prakrit) of that time. Besides, while various views are meticu-
lously referred to, there has not been even an indirect reference
to any Greek view in any of the texts.
Although my conclusion about the influence of Aristotle on
ancient Indian logic or philosophy is negative, I have a positive
view about the question of relevance and usefulness of the study
of Aristotle for the study of the history of ancient Indian logic
and dialectic. It is true that these two traditions, both ancient,
are independent of each other. But, because of this independent
development, study and understanding of the one is bound to
illuminate understanding of the other. Besides, it seems to be
widely recognized now that the study of the history of ancient
philosophy can be combined with the first-hand study of phi-
losophical problems and questions, to the advantage of progress
in both enquiries (as Professor Bambrough has argued in his
Preface to New Essays on Plato and Aristotle1), In this context,
I wish to add that it is also time to realize that the study of
ancient Indian philosophy, where simiiar philosophical problems
have been examined and studied with great interest in
precision and analysis, may also be relevant and advantageous
in a similar way.
bambrough, J.B., New Essays on Plato and Aristotle, London (Routledge

anc} Kegan Paul), 1965.
§1.2 : RULE OF DIALECTIC AND DEBATE
I have chosen the term 'dialectic' to refer to the art of philo-

sophic disputation in ancient India, as this was obviously the
Indian counterpart of the Greek art of discussion or logical
controversy. Philosophical thoughts in ancient times filtered
through the art of disputation and this was as much true of
ancient Greece as it was of ancient India. But as in the case
of many other concepts, the concept of Indian dialectic has
only a 'family resemblance' with the notion of dialectic found
in the writings of Plato and Aristotle. Let me first explain the
Indian concept of philosophic debate, and the kind of argu-
ments I shall be calling 'dialectic'.
During the sramana (post-Upanisadic) period of Indian
philosophy, the intellectual climate was brisk, critical and
controversial. Topics that came under fire were not only the
organized religion and ritualism of Vedic orthodoxy, but also
the established social codes and moral norms, as well as the
knowledge-claims regarding the final destiny of man. In such an
environment, debate—by which I mean controversy, question-
and-answer, and discussion—was the order of the day. No
subject was considered too sacred for criticism and refutation.
These debates sometimes degenerated, as it must, into wrangl-
ing and bitter verbal fights. That might have been the reason
for the Buddha in the Nikäyas to tell his pupils to avoid fruitless
debates1. It was, however, not very easy in those days to avoid
debates.. That is why, in the Upäyahrdaya, one of the early
Buddhist debate manuals, the author first raises the objection
that one should not enter into a debate with others (vädo na
kartavyah) and argues elaborately that it is necessary to enter
into a debate if one has to have any hope to maintain, defend and
propagate one's own religious and philosophic convictions2. Not
only that, ordinary people may be misled by a crafty debater
into immoral and false beliefs. Therefore, some professional
training in the art of debate is essential for a philosopher.
Manuals for the professional debate must have been written
for different schools for training the debater in the types of
îghanikâya. I.33.
Ûpäyahrdaya, p.3, in G. Tucci's Pre-Dignciga Buddhist Texts on Logic from
Chinese Sources, Baroda, Oriental Institute, 1929.
debate, types of arguments, tricky devices of debate and the

checks or grounds for censure or defeat. The canons of Buddhism
and Jainism contain frequent references to many technical terms
of the art of disputation. Texts like Kathävastu (at the Buddhist
council at c. 255 B.C.) report about the various topics for debate
for the Buddhist monks as well as the various ways of debating.
Early manuals for debate, however, are not extant. We have
some crystallized versions of them, probably from two distinct
sources, in such texts as Upäyahrdaya, Asanga's Yogäcärabhümi,
Caraka, and Nyäyasütras. In giving a brief exposition on the
rules and procedures of debate, I shall follow the Nyäyasütras,
because the discussion here appears to be more systematic than
others.
Originally, it seems to me, debate was a respectable vocation
and a natural expectation when two philosophers metin a friendly
session. Thus Janaka, the philosopher-king of the Brhadâran-
yaka Upanisad, asked Yâjnavalkya, who went to him one
morning, "What is on your mind today, Yäjnavalkya? Do you
wish to receive cattle as gift ? or, do you wish to have a
philosophic discussion with me on subtle matters?" Yâjnavalkya
replied, "I have come wishing both 1 ." And then followed a
friendly debate between them. The quality of the debate
obviously depended upon the debaters or participants. Näga-
sena, the Buddhist monk, made the following comment,
probably facetiously, on the types of debate in reply to King
Milinda's questions 2 :
Milinda : Reverend Sir, will you debate with me again ?
Nâgasena : If your Majesty will debate as a scholar, yes, but if
you will debate as a king, no.
M. How is it then that scholars debate ?
N. When the scholars debate one with the other, your
Majesty, there is summing up and unravelling, there
is also defeat, and yet the scholars do not get angry
at it. Thus do the scholars debate, your Majesty.
M. And how do the kings debate ? ^
N. When the kings debate, your Majesty, they state a
x
See Brhadäraiiyaka Upanisad, 4th Adhyâya, 1st Brâhmana.
*Milinda-panho, 2.6.
proposition, and if anyone differs from them, they

order his punishment saying, "Inflict punishment
upon him." Thus, your Majesty, do the kings debate.
The Jaina canon, Sthänänga, refers to four types of refuta-
tion in a "tricky" debate. First, there is the trick of confound-
ing the opponent by using verbiage and thereby trying to give
him a 'run-around' (yäpaka-hetu)1. Second, there is the direct
refutation with a valid reason by confounding one's trick
(sthäpaka-hetu). Obviously, the first kind can be countered
with the second. Third, there is the argument based upon
equivocation (vyamsaka-hetu). This can be countered by the
fourth kind, called lusaka, by exploiting the means of equivoca-
tion and thereby confounding the opponent.
Socrates seems to have referred to this type of broad
division of the debate procedure, as he mentions to Meno:
(Meno 75 c-d)2:
".... if my questioner were one of the clever, disputatious,
and quarrelsome kind, I should say to him, 'You have heard
my answer. If it is wrong, it is for you to take up the argu-
ment and refute it'. However, when friendly people, like you
and me, want to converse with each other, one's reply
must be milder and more conducive to discussion. By that
I mean ,that it must not only be true, but must employ
terms with which the questioner admits he is familiar."
The debate between "friendly people", as Socrates calls ït,
does not seem very different from the kind of debate which
Caraka described as sandhäya sambhäsä 'debate among fellow-
scholars who are friends'. And this is tobe contrasted with what
Socrates described as a debate with a disputatious person. In
Caraka's terminology, this is the other kind of debate which is
actually a verbal fight (vigraha).
Caraka broadly divides debates into two types3. The first is
held with a fellow-scholar and in a spirit of co-operation
(sandhäya sambhäsä), but the second in a spirit of opposition
and hostility (vigrhya). Caraka gives an elaborate description
. Solomon, Indian Dialectics, Vol. I, Ahmedabad, p. 5Iff.
2
Piato, Collected Dialogues (Eds. Hamilton and Cairns).
z
Caraka-Samhitä (ed. Yadava Sarma, Bombay, Nirnaya Sagar, 1933).
p.3O3ff.
about what must be done by a debater before accepting to

enter a debate, viz., examination of the strength of the oppo-
nent, the level of knowledge of the jury and the audience etc.
It is humorously described how if the opponent or the audience
is stupid, one can ensure victory by "a bag of tricks". This is
obviously with reference to the hostile debate. Caraka further
divides the hostile debate into jalpa and vitandä. In jalpa, both
sides establish their position with reason and try to refute each
other, e.g., one side maintains, "There is rebirth," while the
other "There is no rebirth/' and each side adduces reason to
support as well as to controvert the other. In vitandä, one tries
to censure the other without establishing anything.
Nyäyasütra uses the same terms in slightly different senses,
as we shall see presently1. The Näyayasütra classification of
debate was more systematic and hence carried more authority in
philosophical circles. The name for philosophic debate in
Nyäyasütra is Kathä, literally, speech, discussion. It notes
three kinds of debate, väda, jalpa, vitandä. The first kind
corresponds to the friendly and congenial debate in Caraka
(sandhäya sambhäsä). It must have the following characteristics :
(1) There should be a thesis and a counter-thesis mutually
opposing each other. Such a situation arises when mutually
incompatible attributes are ascribed to the same locus (cf., ekädhi-
karana, in Vätsyäyana). Uddyotakara further qualifies it by saying
that such contradictory attributions are to be made with regard
to the same locus at the same time and neither should be taken,
for the purpose of the debate, to be finally decided.
(2) Proving, i.e., establishing, and disproving either of the
theses, should be based upon evidence (pramäna) and argument
(tarka).
(3) Each side should mention the standard five steps in the
demonstration of one's reasoning. (Definition of these five
steps are given elsewhere in the Nyäyasütra; see § 1.1).
(4) The reasoning should not entail contradiction with any
tenet, or accepted doctrine.
This debate is usually to be held between the teacher and the
students or between friends, where each participant is a seeker
after truth (tattvabubhutsu).
1
Nyâyasûtras L2.1-3.
One may wonder whether in this type of debate there may

arise any censure or defeat-situation, for, surely it is not the
nature of a seeker after truth to humiliate somebody with
defeat. But remember what Nägasena said to Milinda. There
will be defeat or censure (nigraha) but no animosity, for it will
be a fair game. The detection of a faulty reason is also recog-
nized as a ground for defeat as Nyäyasütra 5.2.32 informs.
Since there will be refutation (disproving) of the untenable
thesis, it will entail a 'defeat-situation', tjiough based solely on
sound evidence and argument. Besides, the third and the fourth
characteristics indicate that there may be censure based on
some additional grounds. Censure arises also when the debater
fails to mention exactly all the five steps, i.e., mentions either
less (in which case, it is a censure called 'insufficient', hina),
or more (in which case, it is a censure called 'redundant', adhika).
The fourth characteristic, according to Uddyotakara, refers to
the possibility of a censure based upon falsification of an
accepted doctrine 'apasiddhänta*1.
The second type of debate, jalpa, is held between equals, i.e.,
two rival parties, and the explicit goal here is victory (yijayd)
which may not necessarily coincide with the establishment of
truth. Here we come to the discussion of 'tricky' debates that
I have alluded to earlier. According to Uddyotakara, this
debate will share only the first two characteristics of the first
type, but not the last two. For, the last two characteristics
imply that only certain types of censure are applicable here, not
all the other types. The jalpa debate will include, apart from
the first two, the following :
(J.3) Proving and rebuttal are based upon equivocation (chala),
false parity of reasoning (jätl) and censure of all kinds.
It may be pointed out that equivocation and false or
unwarranted parities can neither prove nor disprove anything.
Uddyotakara concedes the point and says that the debater
uses these tricks anyway, when he is unable to defend himself
or censure the opponent on fair grounds. Since victory is the
goal, such tricks are allowed according to the rules of the
game, so to say. The onus is on the opponent to stop him or
to "call his bluff". The Nyäyasütra lists three varieties of
lUddyotakara, under Nyäyasütra 1.2.1-3.
equivocation, and twenty-four varieties of rejoinder based

upon parity of reasoning (jäti)— (twenty in the Upäyahrdayä).
The ways of censuring a debate are given in the Nyäyasütra as
twenty-two. In other words, it notes that in twenty-two ways
a debate might be brought to a close with a decision where one
side wins and the other side loses. The debate is thus turned
into a game and the soundness of the reason by itself cannot
ensure victory, for, as in a game, the strategy of the debater
becomes an important factor. Udayana says that the debater
loses as soon as he shows his incompetence, i.e., he acts in a
way that indicates his confusion1. Uddyotakara points out that
even when the debater uses a sound reason he may not win,
for, he may very well be confronted with a sophistical rejoinder,
and being so confronted if he fails to assert his sound reason
with confidence, he will lose2.
Most of the twenty-two varieties of "checks" in the game of
debate are, however, reasons of common sense. For example,
starting a debate to prove a thesis, one cannot, in course of
the debate, abandon the thesis or cqntradict it (cf. pratijnâ-
häni or pratijnä-virodhä). Nor can one be evasive (cf. viksepa),
or, approve a contrary view (cf. maiänujnä), Bfut the most serious
offence would be to use some unsound or fallacious reason.
For, if that is detected, the debater immediately loses (cf.,
hetväbhäsä).
The third type of debate, vitandä, is more controversial in
nature and, it seems to me, philosophically more interesting. It
is said to be characterized by the lack of proving the counter-
thesis. In other words, the debater here is engaged simply in
the rebuttal of a position but does not give the opponent a
chance to attack his own position. Some have thought, how-
ever, that this is, therefore, not a 'fair' game. Some even have
maintained that since refutation (düsanamätra) is the sole
purpose of this type of debate, the debater may claim not to
have any position or belief in anything. This was, obviously,
the way out for the sceptics, or the sceptic-sophists, or the
sceptic-Buddhists (Mädhyamikas), the sceptic-monists, or even
the sceptic-materialists (Cärväka).
Udayana, pp. 627-628 (Nyayadariana, A. Thakur's edn.).

Hbid., pp. 624-625.
This raises two related questions. First, is it possible to

enter into a debate, without believing in anything, only for the
sake of rebuttal ? Second, even if it is possible, what vaûe
would it have, or what purpose (prayojana) would it serve,
save victory, when such rebuttal is done by the questionable
tricks established in jalpa. i.e., the second type of debate ?
In other words, what lasting value can we attach to a
refutation that is based solely upon either equivocation or (false)
parity of reasoning or both? The first question is more
fundamental, and I shall have to revert to it on another
occasion. I shall briefly comment on it following Vätsyäyana
and Uddyotakara after I have dealt with the second question.
Regarding the second question, it was noted that a debater
cannot simply refute the opponent by bluffs or tricks, unless
of course he is allowed by the opponent to get away with it.
But it so happens that some equivocations or reasoning based on
parities, are#the hardest things to recognize. It is conceivable
that sometimes the debater is not trying to be presenting an
unsound reasoning in refutation. Using modern jargons, we may
call it a linguistic snare or a philosophic puzzle. I can cite at
least two of the rejoinders based on parity (the last two in the
jäti list), nityasamä and käryasamä, which, under one interpreta-
tion, may generate genuine philosophic puzzles, rather than
being tricky arguments for refutation.
The Naiyäyikas were not always very happy in support-
ing the jalpa debate as serving any good purpose. Nyäyasütra
4.2.50 says enigmatically that the second or the third type of
debate is meant for the protection of one's learning when the
learner is apparently a young beginner so as not to be swept
away by some tricky debater into believing the false doctrine.
Sometimes it is emphasized that they render some negative
benefit in the sense that by studying those tricky devices one
can be ön guard and not be easily defeated by such tricks, for,
if the debater can uncover the trick of the opponent he wins.
In any case, in the Buddhist tradition, Dharmakirti (in
Vädanyäya) explicitly denounces such tricks, and he has every
reason to do so1. But as I have already noted that sometimes
philosophers use arguments, which are seemingly plausible and
1
Dharmakîrti, Vädanyäya, (ed. Dwarikadas Sastri), Varanasi, 1972.
it takes another equally gifted philosopher to uncover that

such and such arguments were based on an (unconscious)
equivocation or parity of reasoning. We need not necessarily
suspect that the philosopher in question was deliberately trying
to trick us.
The third type of debate, vitandä, has evoked much criticism.
If the goal here is also victory, as in the second type, then it is
felt that this victory is earned not only through trick but also
by motiveless malignity. Vätsyäyana argues that a debater in this
case cannot be entirely without a motive. One may point out that
since the debater {vaitandlka) does not establish a position here,
it is possible that he does not have a position to defend. But
can there be any debater without a position ? Can there be any
person who does not believe in anything ? In fact, according to
Vâtsyàyana, the third type of debate can easily be put to an end
by asking the debater what his position is or what he proposes
for debating. If he says something, then he has a position to
defend and should be asked to do so. If he declines, he should
forfeit his right to debate. If he says simply that he proposes
(motive) to refute the opponent then also he concedes a
position, viz., refutation of the opponent. In any case, he can
be shown in this way that he cannot conceivably participate In
the third type of debate when he concedes a position of his own.
However, it is by no means absolutely clear why a debater in the
third case cannot consistently maintain that he only wants to
refute the opponent and that the onus of proving anything lies
with the opponent, not with him. In fact, many respectable philo-
sophers would opt out for refutation only of rival views and
avoid proving anything, since either a) they are truly sceptics
and would like to suspend judgement, or b) their truths are
self-evident and hence require no proof.
In fact, it is not necessary that a debater in the third type
of debate is always looking for victory as the goal and using
tricky devices. It is conceivable that his business is also to find
out or seek after truth. In other words, we may not be dealing
here with a charlatan in each case, but a genuine seeker after
truth. Noticing this possibility, some Gauda Naiyäyikas, such
as Sänätam, talk about a four-fold classification of debate,
a) vCula, b) vcuh-vitandä, c) jalpa and d) jalpa-vitandä, the first
two being for the honest seekers after truth, and the last two
for the proud people who intend to defeat others.1 Tricky

devices are allowable, therefore, only in the last two and not in
the first two.
In fact, vâda-vita$4â, an honest and fair debate aimed at the
refutation only of the opponent's thesis, is philosophically a
more fruitful and powerful concept and certainly it has its
adherents. Specially, in case of the sceptics or the sceptic-
mystics, a debate of this nature is positively helpful. For, if
scepticism has to be feasibly maintained at all, the debater must
only refute the position that is positively offered and cannot
defend any position. We may add that scepticism itself can
neither be formulated, nor can it be defended, as a position in
this sense. The debater has to allow his opponent to formulate
his position before he can proceed to refute it.
Is not the negation of a position another position? If the
proposition p is refuted, does it not amount to the defence of
not-/? ? This was actually the import of the criticisms made of
vitaniä by Vâtsyâyana and Uddyotakara. But the sceptic-
debater after refuting p will, only proceed, under the circum-
stances (i.e., being faced with such a question), to refute not-/?.
This will probably imply that p and not-/? do not exhaust the
possibilities, and that might mean in turn that the law of excluded
middle does not apply here. But notice that the law of con-
tradiction as it was understood in the Indian context is not
violated here. For it is possible for p and not-/? to be both
false. It is only when both are held to be true that the law of
contradiction is violated. Or we can say, if one position is
proven to be true, the other is, thereby rendered false, but not
if one is proven to be false, the other would be true.
Alternatively, we may hold that the refutation of a position
need not amount to commitment to the 'negation' of the
proposition involved. Refutation of this debater may be taken
as an 'illocutionary' negation as distinct from a 'propositional'
negation.2 For example, Sanjaya, being asked about after-
life, said: "I do not say there is an after-life". This can be
represented in the manner of Searle:
1
Udayana, op. cit., p. 620.
2Searîe, J.R., Speech Acts, p. 32-33.
This is an illocutionary negation and should be distinguished

from the propositional denials :
"There is no after-life".
F(~p)
It would be clearly a mistake, as Searle has argued, to blur the
distinction between the two. In the same vein, Sanjaya may be
allowed to say:
"I do not say that there is an after-life,"
"I do not say there is no after-life".

~ \- ~(a*) (x is F)
Here the second would not be contradictory to the first. Notice
that the first is of the form ~ \-~{q), while the second, ~ h-(~q).
Consider the following hypothetical debate modelled after the
first verse of Nägärjuna. 1
1. Is a thing (bhävä) produced from itself ?
la. No.
2. Is it produced from something other than itself ?
2a. No.
3. Is it produced from both itself and the other ?
3a. No.
4. Is it produced from neither or nothing ?
(this is equivalent to: Is it not produced at all?)
4a. No.
It is clear that in this formulation, at least, 1 and 2 are not
contradictories, for, it is possible for a thing to be produced
partly from itself and partly from others. Hence 3 is a possible
formulation, which is not exhausted by the rejection of 1 and
2. Now, the question arises: have we exhausted all the possi-
bilities of, say, production from something or other, by the three
rejections la, 2a, and 3a? If we have, then the fourth position
will have to be a rejection of the production itself. Nägärjuna,
however, asks us, by 4a, to reject this position too. This raises
the problem about the subject that we are talking about here.
If refutation of the refutation of production amounts to produc-
tion, then we are back in the game, i.e., with one of the three
1
Madhyamakaèâsttai verse 1 (verse 3 in P.L. Vaidya's edn.)
alternatives, 1,2 or 3. But they have already been rejected.

Therefore, it would be argued that the rejection in 4a, the rejec-
tion of the rejection of production, should not be construed, at
least in Nägärjunian sense of rejection, as a position affirming
production. This can be taken to be a special case of the general
point we are discussing here. The rejection of a position need
not always amount to a counter-position. If this is agreed, then it
is quite feasible for a debater to conduct an honest (non-tricky)
form of debate consisting of refutation or rejection only. Such
a debate can be called a väda-vitandä.
In fact, the aim of the debater in this case is to reduce the
opponent's position to absurdities. In spite of what I have
argued here it is clear that not all the Buddhists were happy
about vitandä. Dharmakïrti clearly rejects any other form of
debate except väda.1 And we may imagine that in this rejection
he probably followed Vasubandhu and Dignäga.
Udayana first refers to the variety called väda-vitandä, an
honest form of debate consisting of refutation only, and then
argues that this would be an impossibility.^ He reasons as
follows : A debater who is also a seeker after truth (as he should
be if he is participating in a väda) can hardly remain content
with refutation only. For, if he simply refutes the opponent's
reason, truth is not determined thereby. Because determination
of the truth depends upon some means (sädhana). If, hq#ever,
he does not refute the opponent's reason, truth remains equally
undetermined, since doubt regarding its falsification has not so
far been removed. If he does not even care for the opponent's
reason, he cannot be said to be determining truth at all. At all
events, determination of truth cannot be done simply by
refutation.2
In any case, Udayana's criticism missed the point that I have
already made. Truth may be either self-evident or indetermin-
able, according to a debater, in which case he does not need a
reason to prove it, or determine it. Sriharsa has elaborately
criticized this argument of Udayana in his Khanânakhanda-
khädya*
1
Dharmakirti, Vädanyäya, p. 69-71.
2
Udayana, op. cit., p. 620.
3
See the introductory section of Khandanakhandakhädya. See also
This classification of debate into the one for seeking after

truth and the other for victory is reminiscent of Plato's way of
contrasting what he called 'dialectic' with 'eristic' or the art of
quarrelling. It is clearly indicated in Plato that the aim of the
procedure he sometimes calls 'eristic' is to win the argument,
whereas the aim of 'dialectic' is to discover truth (Robinson).1
The dialogue that is illustrated in Euthydemus may remind one
of the jalpa or vigrhya kathä in the Indian tradition. The debater
cares nothing for truth, but uses any and every device that pro-
vides the appearance of winning an argument. The word vigraha
in Sanskrit means 'a fight' and the description ofjalpa indicates
that it was a sort of verbal fight. The appropriate picture for
'eristic' in Plato was a verbal fight.
The art of dialectic became a technique in the hand of Aristotle
who undertook to write a handbook of dialectic, Topics. The
character of dialectic was turned into a dubious game of debate,
an exercise for the muscles.of intellect. The Topics of Aristotle,
therefore, resembles, at least in the above sense, the viväda-sästras
of ancient India. But, perhaps, this is too broad a generalization.
Let me qualify it. Aristotle deals with dialectic in the Topics
and what he calls 'syllogism' in the Analytics. One of the
distinctions is underlined by him as follows: a question is treated
'in accordance with opinion' in the work on dialectic, and 'in
accordance with truth' in the work on syllogism. Under syllogism,
Aristotle studies mainly inferences based upon class-inclusion.
But in a broader sense, syllogism, for Aristotle, stands for any
argument in which, after certain propositions have been assumed,
there necessarily results a proposition other than the assumptions
but because of the assumptions. Using this notion of syllogism,
Aristotle says that every dialectical argument is either a syllogism
or an epagoge {Topics I, 12). The general characterization of
epagoge is that it approaches the universal from the particular
(although other varieties are also recognized), and this is
reminiscent of induction of later times. Aristotle says that the
debater must admit an epagoge supported by instances unless he
can produce a negative instance.2
P. Granoff's Philosophy and Argument in Late Vedänta, B.K. MatilaPs The
Logical Illumination of Indian Mysticism (Oxford, 1977).
1
Robinson, R. Plato's Earlier Dialectic, p. 85.
2
Aristotle, De Sophisticis Elenchis, 165 b.
The most pervasive form of refutation practised by Socrates

is called elenchus. An elenchus, according to R.Robinson 1 , is, in
the narrower sense, a form of cross-examination or refutation.
In the wider sense it means examining a person's statement by
asking him questions and then further questions in the hope that
the person giving answer will finally feel that he must agree to
a position that entails the falsehood of his original assertion.
The Platonic idea of an elenchus, the one that he approved,
was a contest in which both sides openly recognize that the
questioner was trying to refute and the answerer was trying not
to be refuted.2
If the Socratic elenchus was a purely destructive instrument,
as it seems to be the case, then it was almost the mirror-image
of the vitandä type of debate. In the jalpa type, the questioner
is trying not only to refute but, as we have seen, also to
establish the contrary position, and the answerer is also
trying to defend his position as well as refute the counter-
position. Plato, however, in his middle and later dialogues,
transforms, perhaps unconsciously, as Robinson has argued3,
the destructive tool of Socrates, the elenchus, into his own new
constructive instrument of dialectic. In other words, he harnesses
the tool of his teacher to constructive purposes, thereby
incorporating it into the larger whole which he calls dialectic.
M. Kneale has remarked that perhaps Plato himself was oonfused
in this matter. 4 But even this modified form of elenchus would
hardly resemble jalpa, for jalpa can hardly be saidvto be a road
to truth or science. Jalpa is explicitly stated to be for victory,
and it may be argued that in some cases achievement of victory
and establishment of truth may coincide with each other. But
still jalpa can hardly approximate the dialectic in its Platonic
sense, for, Plato emphatically declares dialectic to be the noblest
as well as the most useful method.
The specific nature of the Platonic dialectic, however, remains
ever elusive to a reader of Plato. He recommends the dialectical
method with great enthusiasm. In the Republic (533 B), it is
said that dialectic seeks 'what each thing is', the abiding element
^•Robinson, op. cit., p. 7.
2
Ibid., p. 19.
Hbid., p. 83.
4
Kneale, W. and Kneaie, M. The Development of Logic, p. 9.
in the thing. It is a search for definitions. In fact, dialectic is

hardly distinguishable in Plato from the very intellectual type of
philosophic activity that rejects the manifold changing appear-
ances, the mundane things of this world, and searches for the
changeless essences or forms. He did not, however, distinguish,
as we do now, between methodology and metaphysics. A perfect
dialectician was, for Plato, an inspired philosopher. Platonic
dialectic seems to be vaguely reminiscent of Vätsyäyana's
comment about the methodology of a sästra, which says that
it progresses through naming, defining or characterizing and
examining.
Aristotle, however, cleared some mist that surrounded the
notion of Platonic dialectic, by transforming it into a technique
that could be learnt by itself. Dialectic is such a technique. It is,
according to both Aristotle and Plato, unrestricted in its
application. But, while for Plato it is essentially a scientific
activity, for Aristotle its lack of restriction is an indication of
its unscientific character.1 Aristotle rejects the Platonic con-
tention that dialectic involves search for definitions. For him,
it is the study of technique of argument from non-evident
premises. Thus Aristotle believed to be covering also dialectical
arguments when he was giving an account of syllogistic. In the
Topics,hs gave .rules for conducting debate (i.e., the disputatious
debate by means of valid arguments). But in De Sophisticus
Elenchus which is considered to be an appendix to the Topics,
he gives rules for detecting (as well as inventing) invalid
arguments and in this respect it resembles the Väda manuals
of the ancient Indians where the method of invention and
detection of invalid arguments are talked about.
It has, however, been argued by J.D.G Evans, against the
prevailing opinion of modern commentators on Aristotle, that
it is a mistake to represent the Topics as a manual of instruction
on how to win a debate at all costs, or even to regard it as a
first draft on the Analytics.2 The Topics, according to Evans,
is sui generis, and here Aristotle elects to treat such concepts
as intelligibility in their full complexity.
1
Evans, J.D.G. Aristotle's Concept of Dialectic, p. 50.
2
Ibid.9 p. 94.
§1.3 : ARGUMENT AND PSEUDO-ARGUMENT

(NYÄYA AND NYÄYÄBHÄSA)
It is, perhaps, Vâtsyâyana who first uses the expression nyäyä-

bhäsa, as an opposite term to nyäya. The word 'nyäya' carries
a specialized sense in philosophical treatises. It is difficult to
find in the English Language a synonym which will fully convey
the meaning expressed by the term nyäya in Indian Philosophy.
Vâtsyâyana brings out the sense in the following way. 'Nyäya',
in his opinion, is the examination (i.e., apprehension of an
object) with the help of all pramänas. 1 The etymological meaning
(ni+in niyats prapyate vivaksitärthasiddhir anena—by which the
desired end or object is attained, i.e., to say, is understood)
also contributes to such an explanation. The real sense of the
word is anumana or inference expressed fully in a verbal form,
i.e., what goes by the name parärthänumäna. In case of parärthä-
nwnäna which is constituted by the five-membered inference-"
schema (which we have already discussed in the first two
sections), one must take recourse to all the four pramänas
in some form or other. In the first member, pratijnä or
assertion, we are helped in some way by sabda pramäna
or verbal testimony. The second member, hetu or reason,
presents a skeleton form of anumana or inference. Percep-
tion performs the main function so far as the third member,
udäharana or example, is concerned. The fourth member,
iipanaya or application bears some sort of distant similarity with
the implication of upamäna or comparison. Thus, all the pramänas
become helpful in an inference meant for the other, i.e., parär-
thänumäna, which is better known as nyäya. Uddyotakara
explains the word evidently in this sense. Väcaspati, however,
indulges in a bit of twisted explanation of the lines of Vâtsyâyana.
According to him, nyäya is examining or testing the object addu-
ced as the hetu or reason of the inference. It is inference itself
which puts the hetu to test. Thus, nyäya is parärthänumäna. But
we may humbly submit that this explanation of Väcaspati does
not appear to be quite satisfactory, in view of the fact that it
takes a roundabout way of unfolding the meaning.
Pramânair arthaparïksanaifi nyäya h—Nyäya-Bhäsya under Nyäya-Sü tra

1.1.1.
2.4 Logic, Language and Reality
One of the best ways to explain definitely what is denoted by

the term nyäya is to distinguish it from what is known as nyäyä-
bhäsa, or pseudo-inference. In ultimate scrutiny nyäya stands
for those inferences as rest upon, i.e., are not at variance
with, perception and verbal or scriptural testimony.1 In this
manner Vâtsyâyana tries to introduce a distinction between
nyäya and nyäyäbhäsa. Thus, inferences, the results of which do
not agree with what is obtained from perception, or verbal
(scriptural) testimony are treated as nyäyäbhasas (or pseudo-
inference.2 From this it is clear that nyäyäbhäsa or pseudo-
inference is of two types : one that is liable to be frustrated by
perception and the other by verbal testimony (ägama).
The example of the first type is to be sought in the following :
fire is not-hot, as it is created, such as a pot. This argument
resembles an inference in all its external features. Hence, one
may take it as a nyäya, but on scrutiny it is found to possess
simply outward and superficial resemblances. So the inferential
knowledge thus derived is rejected in no time by just the opposite
judgment derived from the perception of fire that it is hot. Thus,
it is not nyäya in the strict sense of the term but nyäyäbhäsa. In
other words, here perception opposes inference.
How is the inference opposed here ? Let us understand the
position more clearly. Uddyotakara answers that this is a wrong
application of inference to an improper case. The case under
enquiry (i.e., not-hot-ness of fire) is not at all a case for infer-
ence. It disqualifies itself as the stronger proof (perception)
carries just the opposite conviction, e.g., fire is hot. Uddyota-
kara says, perception here baffles inference.3 Here another point
may also arise. The hetu, viz., createdness (krtakatva) suffers
here from another serious defect. It is non-concomitant with
the sädhya. But this defect, e.g-, non-concomitance cannot be
proved until and unless we first take recourse to perception of
particular instances. And when perception proves just the
opposite, eg., the absence of the sädhya in the locus, it is use-
less to look for other defects, if any.
1
Pratyak sägamäsritamanumänam, son vïksâ—Ibid.
%
Yat punaranumänampratyaksägamaviruddham nyäyäbhäsah sah—Ibid.
^ Yasmin visaye et at prayujyate sa pratyaksenäpahrtah—Nyäya- Värttfka, on
above.
The Buddhist logician, in Nyäyapravesa, gives a different

example in a context almost similar to this. According to him,
'sound is inaudible, since it is created, such as a. pot9 maybe
taken to be an inference which perception opposes. He really
contends that here the thesis 'sound is inaudible'is inconsistent
with perception. But Uddyotakara retorts that audibility is not
an object of perception. Audibility is but a relation existing
between the ear-organ and sound. But the ear-organ itself being
etherial (cf. gaganà) by nature, is beyond the reach of percep-
tion. Now, a relation (sambandha) becomes perceptible, if both
the relata (sambandhin) are perceptible. Here neither the ear-
organ nor sound are perceptible. Hence, the said relation is
imperceptible. Kumârila says, 'knowability-by-the ear' is not
perceptible.1 Thus, the absence of sädhya, e.g., audibility
(srävanatä) not being an object of perception, it is improper to
say here that perception opposes inference.
The example of the second type is supplied by the following :
The skull of the corpse is pure, as it is a limb (of a corpse),
such as the conch-shell.
Such an inference is invalidated by the scriptural testimony.
Manusmrti and such sacred texts speak of the human skull
as impure. Thus, it is an example of nyäyäbhäsa. Let us
make it more clear. What is the exact sense of the word
purity in this inference? Evidently, it implies the absence of
demerit on the part of those who touch the object under
question. Now, what is the authority to decide whether
the person who touches it, incurs demerit or not? If the autho-
rity of the Vedas or Revealed Texts is accepted, then such an
inference cannot stand scrutiny. Because Manusmrti, which owes
its authority ultimately to the Vedas, gives the opposite verdict.
Here, the conch-shell is treated as pure on the authority of the
Vedas, When its authority is accepted in one case, it cannot be
rejected in another case on the same issue. Thus, with due
reverence to the Vedic authority, we are to accept the conch-
shell as pure and the skull of the corpse as impure. Hence, the
said inference is rendered invalid.
When all these are stated, a question naturally arises. What
should be the position when an inference is opposed by (1 )
x
na hi srävanatä nämapratyaksenävagamyate—Sloka-Värttika of Kurnärila,
Anumänay verse 60.
another inference or by (2) comparison (upamäna) ? Will these

be the cases of nyäyabhäsa ? If so, why does not Vätsyäyana
refer to them ? The second problem can be easily resolved.
When comparison opposes an inference, it becomes a nyäya-
bhäsa of the second type, inasmuch as that inference is in reality
opposed by tabda pramäna or verbal testimony which lies at the
bottom of all cases of upamäna ( comparison).
So far as the first case is concerned, Uddyotakara says that
two opposing inferences of an independent nature are practically
impossible to be drawn about the same subject. In such a case,
both the rival reasons (hetus) are of equal strength, and hence
suffer from the serious defect known as satpratipaksatva (i.e.,
having rival a reason). Sb, none of them produces any inference
They cancel each other. The question of nyäyabhäsa (pseudo-
inference), therefore, does not arise at all.
Väcaspati, however, is inclined to admit another type of
nyäyabhäsa, where inference opposes another inference. Accord-
ing to him, where there are two inferences, one of them
depending upon the other, in that case if the dependent
inference is opposed by the other upon which it depends, the
dependent inference becomes a nyäyabhäsa. This may be adduced
as the third type of nyäyabhäsa. For example, an inference like
'God is not the creator' would depend upon a prior inference by
which the existence of God is proved. Now, by that very prior
inference, just the opposite of the second inference, i.e., 'God is
the creator' is also proved. Hence, the second inference loses its
force and becomes a nyäyabhäsa.
Lastly, the relation between nyäyabhäsa and hetväbhäsa1 (bad
or defective reason or defects of the reason) is an interesting
point to be noted in this connection. In all cases of nyäyabhäsa,
the hetu or reason suffers from a serious defect, e.g., bädha or
kälätyayäpadesa, so as to mar the essence of the inference.
Abädhitatva or non-opposition is one of the five essential marks
of a real hetu. The five essential marks of a (sound) hetu are :
paksa-sattva, sapaksa-sattva, vipak$äsattva, asat-pratiapaksatva
and abädhitatva. If a hetu lacks any one of them, then, according
to the Naiyâyikas, it becomes a hetväbhäsa. In this case, therefore,
the hetu itself becomes a hetväbhäsa or defective reason. This
1
See section § 1.5 below.
defect is contained in the foregone determination of the absence

of sädhya in the locus or subject {paksa). This disqualifies the
paksa to be a real paksa of an inference. The Vaisesikas,
however, regard this as a defect of the paksa itself. In their
opinion, a hetu possesses three essential marks, e.g., paksasattva,
svapaksasattva, and vipaksäsattva, and not five, as stated by the
leading Naiyäyikas. The Vaisesikas, in this case, would agree
with the Buddhists.
The Navya Naiyäyikas make no special mention of the term
nyäyäbhäsa, most probably because they regard it as redundant
to coin a new technical term in addition to the five hetvä-
bhäsas which are enough to cover all the cases of false infer-
ences. Thus, we see that the term nyäyäbhäsa though
mentioned by Vatsyäyana (perhaps, not with such seriousness
as his commentators intended) by way of giving a popular
definition of what is known as nyäya, illustrated, elaborately
dealt with and commented upon by his followers, viz.,
Uddyotakara, Vâcaspati and Udayana, has been completely
ignored by other systems as well as by the new School of
Nyäya.
One of the important characteristics of Indian Logic is that
it seldom develops the sort of doctrines which we class <|s
'Formal5. Formal validity in argument was perhaps looked upon
by the Indian logicians as not of much importance. So, formal-
ism in logic does not find here much scope for development.
This can be illustrated in many ways taking examples from the
books of Indian logic. The doctrine of nyäyäbhäsa as an opposite
term of nyäya is also a case in point where we may have the
advantage to look to the Indian attitude towards formal validity.
The examples like 'fire is not-hot, as it is created, such as a pot',
may be formally valid, so far as the Indian form of argument
is concerned. It may be submitted that the relation of universal
concomitance between sädhya and hetu, an essential requisite
for valid inference, is not established here, since all created
things are not not-hot. But Naiyäyikas will say that an awareness
of such a relation (be it valid or invalid) is not impossible here,
since the persons inferring may experience their co-existence in
some instance or instances (as illustrated by the case of a pot in
the said example). In the absence of any contradictory experi-
ence, this awareness, for all practical purposes, is sufficient for
the desired inference to follow. So, enquiry into this line to

prove invalidity of the argument will not be much helpful. The
Naiyäyika maintains that the conclusion, which such argument
leads to, is contradicted directly by the experience of fire
which gives the contradictory judgment, fire is hot. From the
formal point of view, the two propositions 'fire is hot5 and
'fire is not-hot' are contradictory propositions, and hence,
both of them cannot be true at the same time. Thus, the truth
of the one entails the falsity of the other. Hence, truth of the
first being well established by the direct experience of fire, the
purpose of the said inference (which is nothing but to prove the
truth of the contradictory proposition) is vitiated. So, if we infer
at all that the opposite is the case, it will not be a real inference
but a pseudo-inference, to all intents and purposes. A
pseudo-argument, therefore, is one that is based upon a
pseudo-inference which goes against the fundamental
requirement of rationality, consistency or coherence. Our
conscious beliefs must cohere orbe consistent with one another.
We may construct an argument, a sound one or an unsound
one, in support of a proposition or belief. But if that belief or
proposition does not cohere with other accepted beliefs, if it mili-
tates against some well-established and sound belief, then, without
further investigation into the soundness or unsoundness of that
argument, Nyäya would declare it as a pseudo-argument. There
may be an implicit logical fault in the argument itself, but that
fault need not be brought out to the surface in order to discard
the argument as a pseudo-argument.
We may note that we are concerned here with the constructed
argument in favour of some thesis or belief which runs counter
to some of the well-established, and least contested beliefs. This
is different from the case where two arguments (inferences) are
constructed (adduced) in support of two theses which are con-
tradictory to each pther. In the second case, both arguments
(inferences) are to be considered as inconclusive. But, they would
not be the instances of pseudo-argument in the above sense.
The latter case is technically called the case of sat-pratipaksa—
'a thesis with a counter thesis'. There may be cases when both
the thesis and the counter thesis are supported by SOUND argu-
ments (though not necessarily by formally valid, deductive
arguments), and in such cases, it would be impossible to reach
a decision one way or the other. Hence, 'having such a counter
thesis' is regarded as a logical defect of an argument, because

the case would be inconclusive.
§ 1.4 : THE EARLY NYÄYA THEORY OF INFERENCE

{Nyäyasütra 1.1.5)
Nyäyasütra 1.1.5 is very difficult to explain. We find here a

threefold classification of inference. The problem arises,
because the terms used to name the three types of inference
are ambiguous and the sütra does not mention any examples
to clarify the principle of this threefold division. Many
other logicians of that period mentioned and exemplified this
threefold classification. A careful study reveals that they
were all related in some way or other.
The examples found in the Carakasamhitä seem to be illustrat-
ing the Nyäya theory of inference. Nyäyasütras 2.1.37-38 also,
discuss the points mentioned in Nyäyasütra 1.1.5. A satisfactory
explanation of these two süiras (2.1.37-38)1 is needed for having
a correct interpretation of Nyäyasütra 1.1.5. Vätsyäyana's
commentary here seems to be confusing. The threefold classifica-
tion (in Nyäyasütra 1.1.5) was probably based upon the
principle of threefold division of time—past, present and future.
The structural arrangement of the Nyäyasütras also justifies this
point. Another possibility is that pürvavat refers to inference
based on causal relation, sesavat to the method of exhaustion
(parisesa), and sämänyatodr$ta to inference based on non-causal
relations.
When the theory of inference {anumäna) was first systematized
in India, two slightly different accounts of inference were found
in the tradition. The first account suggested a twofold classi-
fication of inference, and was upheld by the Vaisesika and the
Mïmamsâ schools. The second account spoke of a threefold
classification of inference, and was recorded in the Nyäya, in
some Buddhist texts, and in Caraka. Nyäyasütra 1.1.5 mentions
the three types of inference as pürvavat, se$avat and sämänyato-
drsta. These terms themselves are ambiguous, and as far back as
lr
This numbering has been accepted in the reading of MM. Phanibhusana
Tarkavagis. See his Nyäya-darsana, Bangiya Sahitya Parishat Series 63,
Calcutta : 1939. In Ganganatha Jha's reading, the last two sütras are
numbered 2.1.38 and 2.1.39. Poona Oriental Series No. 59, Poona : 1939.
Vâtsyâyana, alternative explanations about them were offered.

Moreover, Nyäyasütra 1.1.5 gave only the names, and no
examples, of these three types. It is unfortunate that no single
series of interpretations offered by the earlier commentators,
such as Vätsyäyana, is consistent and that those offered by the
later commentators, such as Uddyotakara and Jayanta, while
sometimes consistent, also give the impression of being twisted,
far-fetched, and anachronistic.
I shall attempt to show what might have been the Sütrakära's
understanding of this threefold division and to furnish reason-
able and satisfactory examples of the three types without
stretching our imagination too far.
We may begin by giving a synopsis of the different interpre-
tations already suggested by the ancients. I avoid the question
of the exact dates of these authors. Their chronology is approxi-
mately as follows: Upäyahrdaya (UH) or Prayoga-sära1, Vätsyä-
yana (Vät), Ts'ing-mu (a comm. on the Mülamädhyamikakärikä)2,
Gaudapäda, thePäsupata School3,Uddyotakara (Udd)4, Jayanta,
and Bhäsarvajna.5
UH. (a) pürvavat = 'as before' : From previous knowledge
of some special mark (like six fingers, etc.,) of a
particular child, to infer later that this is the same
man (grown-up child) from that mark.
(b) sesavat = "the rest will be alike' : From the salty taste
of one drop of sea-water to infer that all other drops
will be salty too. (Inference through sampling?)
(c) sämänyatodrsta : Displacement of ordinary bodies
is seen to be due to their movements. Therefore,
the celestial bodies, such as the planets and the
moon, move, because they are seen to be displaced.
1
See G. Tucci, Pre-Dignâga Buddhist Texts on Logic from Chinese
Sources, Gaekwad's Oriental Series Vol. XLIX, Baroda, 1929. My
friend Prof. Hattori informs me that a better translation of the Chinese
title would be Prayoga-sära (as suggested by Frauwaliner).
Hhid,
z
Päsupata Sütras with the Pancärthabhäsya of Kaundinya, ed. by R.
Ananfo Krishna Sastri, Trivandrum Sanskrit Series No. CXLIII, Trivan-
drum : 1940, p. 7.
4
Uddyotakara, Nyäyavärttikam, Benares, 1915.
5
Jayanta, Nyayamanjarïy Vizianagram Sanskrit Series, Benares, 1895. And
Bhäsarvajnä, Nyäyabhüsana, Svami Yogindrananda, Varanasi, 1968.
UH A. General example of inference given before the

threefold classification. Seeing fire with smoke
before, one infers fire from smoke later, (pürvavat?
See below.)
Vät I (a) Pürvavat: From cause to effect: It will rain, because
there is a cloud.
(b) Sesavat : From effect to cause: It has rained, because
the river is full and flowing swiftly.
(c) sämänyatodarsta : Same as UH (c). This unfortunate
example was criticized and even ridiculed by Uddyo-
takara and Väcaspatr. It was declared by them to be
a variety of sesavat. This is also identical with
Sahara's example of sâmânyatodrstasambandha.
VatII (a) Same as UH A. pürvavat = c as before': To infer
fire from smoke.
(b) 'sefa= the remainder': Inference by elimination of
alternatives :
Premise: Sound is either D or G or K
But Sound is not D because of X,
and not AT because of F. Conclusion:
Therefore, Sound is G.
This inference is based upon a more developed
logical principle.
But, perhaps, the Sütrakära did not have this in
mind.
(c) Another interpretation of the term sämänyatodrsta.
Example : Desire, etc., are qualities (guna) and a
quality inheres in a substance as its locus. There-
fore, the substance where desire, etc., inhere is the
self. This was also criticized by Uddyotakara who
pointed out that in this type too, an implicit use of
the principle of elimination is made to prove
the substance in question to be the self.
Ts'ing-mu (a) Same as UH A./Vât II (a).
(b) Similar to UH (b) : One grain of rice iscooked.
Therefore, all other grains of rice (in the same
pot) are cooked.
(c) Same as UH (c). Alt. example : Same as Vät
II (c).
Gaudapäda : (a) Same as Vât I (a).
(b) Same asUH(b).

(c) Same as Vât II (c).
The Pâéupata school: (a) & (c) Same as UH (a) and (c)
(b) To infer the whole cow from seeing only the
horns, etc. (essential parts). An interesting form
of knowledge (whole from part?). Kanada
used this example in three places of the
Vaise?ika Sütra.
Udd I Threefold division: anvayin, vyatirekin and anvaya-
vyatirekin. The terms pürvavat, etc., are not used.
pak$a = subject =. the place or the property-locus
where the sädhya is intended to be inferred. Sapak$a
— homologue - entities, apart from the subject, where
the sädhya is known to occur, vipakça = heterologue
»entities where the sädhya is known to be absent.
(a) No heterologue : Sound is impermanent, because
it is produced: (when it is an inference of those
who believe everything to be impermanent).
(b) No homologue : The living body is not
soulless, for, otherwise it would be lifeless.
(Comparable with a negative conclusion from
at least one universal negative premise—
Celarent or EAE in Aristotelian system. Also,
avita inference in the Sämkhya system) The
example is slightly defective. The later school
(Navya-Nyäya) gives a better example.
(c) Having both homologue and heterologue:
Usual form of inference : Sound is impermanent,
etc.
This certainly indicates logical advances of a later
period. Gangesa accepted this threefold classifica-
tion. (For this as well as for a brief exposition
of Dignâga's "wheel of reasons", see below § 2.4.)
Udd II (a), (b), and (c) : Same as Udd I (a), (b) & (c). But
some very far-fetched explanations of the terms
pürvavat, etc., are suggested so as to cover the
above cases. Sämänyatodr?ta is analysed in two
ways:Jsämänyatah + dr?ta and sämänyatahJr
adr§ta.
Udd HI. (a) & (b): Same as Vät I (a) & (b). A difficulty in
inferring the effect from the cause is pointed out,

but is also explained away in order to defend
Vâtsyâyana.
(c) To infer a from b where a and h are not causally
related: There is water nearby, because wild
geese are present. This is a new example.
Udd IV. 'Threefold' means that the hetu must have the
three characteristics mentioned in Vaise$ika sütra
3.1.10-11 (GOS-136).
(a) prasiddha = The hetu occurs in the subject.
(b) sat = The hetu occurs in the homologue.
(c) asandigdha=-The hetu does not occur in the
heterologue.
Jay an ta I (a) & (b) : Same as Vät I (a) & (b).
(c) Similar, in principle, to Udd III (c). The actual
example is perhaps, better : To infer the
nature of the taste of a kapittha fruit, from
seeing its colour.
Jayantaîl(a) & (b): Same as Vät I ï ( a ) & (b). The suffix
'VatV in püt'vavat, etc., means 'like',
(c) Sämänyatodrcta refers to the case where there
is sämänyato vyäpti, and the sädhya is not
perceivable. Example: We have sense-organs,
because we have sense-perceptions which are
acts, and all acts must have instruments.
(Jayanta admits that sesavat and sämänyato-
drsta are closely related, but he also contends
that the principle involved in each case is
different, while the examples might be similar.)
Bhäsarvajna I ( a) Pürva =-- cause. Hence, infer cause from effect.
(b) sesa = 'remainder' = effect. Hence, infer effect
from cause.
(c) Non-causal connection. Infer a specific
taste (of a fruit) from a specific smell.
II, (a), (b), (c) same as Udd I.
It seems to me that none of these interpretations fully represents
the position of the Sütrakära on the threefold classification.
Of what the Sütrakära had in mind we can only make an
intelligent guess. Such a guess should, of course, be based upon
some textual or other evidence. While making such a guess I

shall assume that at least some sütras in the beginning of
Chapter 1, Ähnika 1> and some of those of Chapter 2 r Ähnika 1
are related to each other, since they seem to develop some
coherent theories.1 In particular, it seems that sütras 2.1.37 and
2.1.38 refer back to the same threefold classification of infer-
ence that is mentioned in 1.1.5. Thus, my minimal assumption
will be that at least sütras 1.1.5, 2.1.37 and 2.1.38 were intended
to present a single theory of inference based on a threefold
classification. When these sütras are studied and understood
together, the threefold classification, it seems to me, appears
neither silly nor unsatisfactory.
The structure of these sütras also indicates that they were
developing the same theory of inference. Sütra 1.1.5 states
that inference, which is always preceded by perception, can be
of three types : pürvavat, sejavat and sämänyatodr$ta. Nyäya-
Sütra 2.131 raises an objection that inference is not a valid
means of knowledge {pramänä) , and assigns three reasons to
the three types, in order to show that individual examples of these
three types are wrong or unsound. Nyäya-Sütra 2.1.38 answers
this objection by furnishing three correct types, in order to show
that the charge of unsoundness in each case can be satisfactorily
explained away. Vâtsyâyana has explained these sütras in a
similar manner, although he has made, it seems to me,
some blunders.
Sütra 2.1.37 mentions the following three reasons related to
the supposed three examples:
(1) rodka = obstruction (See Vät I sesavat). The river may
be full because of some obstruction in its course, and
not because of rain, in which case the inference con-
cerned will be false.
(2) upaghäta = demolition or dispersal (See Vät I pürvavat).
It may be that the clouds are dispersed (by the wind),
the Nyäyasütras resulted from various strata was already stated by
Haraprasada Sastri in J. A. S. B., 1905. Tucci has proposed that we distin-
guish the section dedicated to "the pure vâda rules" from the section which
is more or less "polemical". See Tucci (riote 1 in p. 29 of Pre Dignäga
Texts on Buddhist Logic), p. xxiv of Introduction. Since Chapter 1, Ähnika
Î, does not exactly deal with the "pure" vâda rules (but only Ahnika 2 does
so), I suggest that this need not be taken along with the vädb-section.
in which case the said inference will be false. Vâtsyâ-

yana, however, mentions a slightly different example
here : It will rain, because the ants are carrying their
eggs. Thereupon he interprets upaghäta as 'destruction':
It may be that the nests of the ants have been destroyed
and they are moving to a safer place in which case the
inference of rain will be false.
(3) sädrsya=* similarity. Vâtsyâyana suggests a completely
different example here: There is a peacock, because
its voice is heard. He explains 'similarity' as follows:
The above inference will be false, if one is duped by the
similar voice of a man imitating a peacock.
The respective answers given in sütra 2.1.3$ are as follows :
(4) ekadesa = partial case. The fullness and other appear-
ances of the river (viz., swift current, etc.) due to
previous rain (in the upper region) agree with the
fullness and other appearances of the river due to
the blocking of its course (in some lower region) only
partially and not wholly.
(5) träsa = fear. When the nests of ants are destroyed, only
a few ants are affected by fear and the movement of
these ants is haphazard. But, when rain is imminent,
all ants move carrying their eggs in a systematic and
unperturbed fashion.1
(6) sâdrsya = similarity. Similarity implies difference also.
iWas it "hrâscT instead of " träsa" 1 The script "/r" 3" is similar to "Ar" "^. If
it were "hräsa" ( = "decrease") we need not consider the carrying of eggs
by the ants as the reason (hetu), but just can take the accumulation of clouds
as the supposed cause of future rain'and its decrease to account for the
possible failure of rain. Even Vâtsyâyana, to whom the ant-example is due,
neither mentions the word "träsa" nor does he paraphrase it. He only notes :
"pipïlikâpràyasyàndasancâre bhavisyati vrstir ity anumiyate,nakäsäneid iti"
Ganganatha Jha translates as follows : "when one infers 'coming rain', he
does so from the fact that whole hosts of ants are running about (calmly
and peacefully) with their eggs,—and not only a few ants..." (See his Transla-
tion, note 1 in p. 29 of the book, p.166). Could "käsäneid" be a
paraphrasing of "hrâsa" (="decrease, i.e., numerical decrease")? Anyway,
there is nothing in Vätsyäyana's expression, which could possibly refer to
"träsa" i.e., fear. But in the absence of any textual evidence in favor,
this line of conjecture, perhaps, should better be avoided, and the
reading should be taken as "träsa". For difficulties connected with
the ant-example, see also Ganganatha Jha, p. 164, the second foot-note.
Hence, the mistaking of one entity for another due t a

similarity can be corrected when their difference is
discerned.
In short, sütra 2.1.38 contends that there is a discernible
difference between the series of instances mentioned by the
objector in sütra 2.1.37 and the series of instances originally
adduced as valid types of inference.
The above explanation given by me mainly follows that of
Vätsyäyana with some variations already indicated. What is
more curious to note here is this. From the explanation that
he gives, Vätsyäyana seems to think that the Sütrakära has
reversed the order of his threefold division given in sütra 1.1.5,
so that the first-mentioned cause (either in 2.1.37 or 2.1.38) is
to be related to the example of se$avat and the cause mentioned
next (either in 2.L37 or in 2.1.38) is to be related to the
example of pürvavat. Vätsyäyana, however, does not explicitly
point out this anomaly. Now, this breaking of the order seems
to me to be a false interpretation. Rather, the order here may
lead us to a correct formulation of the examples of the three
types.
My guess about the principle of the threefold classification
of the Sütrakära and their examples is a very simple one. The
terms 'pürvaJ and 'sesa' can be taken to mean, among other
things, temporal stages. Thus, pürvavat might mean 'that which
is related to the preceding (pürva) event' (compare Uddyotakara :
"pürvam asyästi iti pürvavat".1 The term 'seâvat9 being contrasted
with 'pürvavaf might mean, accordingly, 'that which is related
to a later (sesa) or future event' (ct9se$o 'syästi iti se§avat). Thus,
Suggestion I (a) pürvavat: To infer the past event from the
present: It has rained, because the river is
full, etc. Same as Vät I se$avat. Compare
with(l) & (4) above,
(b) se$avat: To infer the future event from the
present: It will rain, because there is cloud,
or because the ants are carrying eggs. Same
as Vät I pürvavat. Compare now with (2) &
(5) above.
(c) sämänyatodrtfa : Where both the events are
lUddyotakara, p. 49, line 18.
present events, but one is perceived and the

other is inferred. Examples: There is fire,
because there is smoke. (Same as UH A.
and Vât II pürvavat). Or, There is water,
because there are wild geese: Same as Udd
III sämänyatodr$\a.
Ory It is raining, because the peacocks are
crying.
Is this the actual example which Vâtsyâyana had in mind when
commenting on sütra 2.1.37, and not the example suggested by
Uddyotakara (which I have noted in Udd III a, b) ? Ganganatha
Jha's note may be consulted in this connection.1
The term sätnänyatodrcta can be explained in the light of
these examples in the following manner: The joint presence of
the two objects or events has been observed (perceived) before,
but in the case under consideration, though both are present,
only one event is perceived by the person who thereby infers the
other. In such a general sense, this might include other examples
of sämänyatodr$ta given above. The expression "similarity" in
sütra 2.1.37 can, accordingly, be interpreted as follows: Instead
of seeing the smoke one may mistake a similar object, viz.,
water-vapour, dust, etc., for smoke, or instead of hearing the
peacock's voice, one may hear a similar voice, and, thus, the
inference of fire or the inference of rain will be wrong.
Since the notion of causal connection has been associated
with the theory of inference from a very early period (cf.,
Vaise$ika-sütra 9.18; GOS-136,)2 we might suggest another
alternative but not an altogether different interpretation. (This
is exactly what Bhâsarvajna I suggests.)
Suggestion II (a) purva-= prior event, i.e., the cause: To infer
the cause from the effect. Example : Same
as Sug. I (a), Vat I.
(b) se?a = posterior event, i.e., the effect: To
infer the effect from the cause. Example:
Same as Sug. I (b) /Vât I.
1
See G. Jha, (note 1 in p. 29 of the book) p. 165, footnote.
H follow here the numbering of Candränanda. See Vaise$ika-sütra, ed. by
Muni Sri Jambuvijayaji, Gaekwad's Oriental Series 136, Baroda, 1961.
(c) Where the two objects are not causally related-

Example : Same as Udd III sämänyatodr$ta.
There is admittedly some strength in this interpretation. Except
for their perpetuating Vätsyäyana's confusion of pürvavat and
sefavat, commentators like Uddyotakara and Jayanta indirectly
support this line of interpretation. However, I am inclined to
believe that in the tradition of the threefold classification, the
notion of causal connection was not predominant, although it
must be admitted that the evidence of Caraka goes against my
inclination (see below). Moreover, considering the fact that
inference through sampling was also a primitive form of correct
inference, the word "sepa" in "sesavat" can be happily explained
if we accept the interpretation of Prayogasära and Ts'ing-mu
(see UH (b) & Ts'ing-mu (b)).
To sum up, the following points might be considered :
First, the examples given in Suggestion I or II seem to suit well
the three objections raised in sütra 2.1.37 as well as the three
answers mentioned in sütra 2.1.38, and no breaking of the
order is necessary. Vätsyäyana's explanation implies the
breaking of the order which seems to be highly improbable,
For similar reasons, the Prayogasära example of sesavat
cannot, be accepted as that of the Sütrakära, because sütra
2.1.37 or 2.1.38 cannot, in that case, be explained.
Second, immediately after examining the definition of inference
in sütras 2.1.37-38, the Sütrakära proceeds to consider
whether it makes sense philosophically to talk of the three-
fold division of Time, viz., past, present and future, or more
particularly, whether the notion of present is distinguishable
from the notion of past and future {cf., 2.1. 39-43). This would
have been irrelevant unless the threefold classification of infer-
ence called, at least implicitly, for such a distinction of three
time-stages. In nyäya terminology the^ discussion of the
problem of three time-stages is related to the discussion of the
examination of the theory of inference by upodghäta sangati
or prasanga sangati.1 For similar reasons, I think, the
*See Ingalls, Materials for the Study of Navya-Nyâya Logic, Harvard Univer-
sity Press, Cambridge, Mass., 1951, p. 80. For a comprehensive discussion
on the notion of sangati, see Gàdâdhari-covam. on Tattvacintâmanididhiti,
pp. 442.
examination of the doctrine of avayavin (whole) follows (in

sütras 2.1.33-36) almost as a part of the examination of the
theory of perception. Thus, our guess-work will be supported
if we credit the compiler of the second chapter, at least, with
the understanding of the position of the Sütrakära (if he was
a different person) regarding this threefold classification of
inference.
Third, some of the remarks of Uddyotakara seem to lend
support, but only indirectly, to the interpretations suggested
above (specially Sug. II) Compare : *e but what is pürvavat
here ? Is it the effect or the cause ? If it is analysed as lpürvam
asyästV then pürvavat means the effect. But, then, it contra-
dicts the interpretation that it refers to the inference of effect
from the cause". 1
Although this is mentioned as the objector's view (pürva-
paksa), a view which Uddyotakara is contending against,
the remark at least shows that the interpretation of
pürvavat as referring to the inference of the cause from the
effect is a plausible one. Regarding the above, I only suggest
that this is also a possible and a reasonable interpretation. But
since Vätsyäyana has confuted pürvavat and sesavat, a confusion
which is attested by his rather odd explanation of sütra 2.1.37,
Uddyotakara has tried to defend him by a twisted interpreta-
tion.
Fourth, Vätsyäyana's concluding remark under sütra 1.1.5 can
be said to lend some indirect support to our suggested inter-
pretations (specially Sug. I). Compare : kisad-visayam pratya-
k$am, sad-asad-visayam cänumänam,...asac ca khalv atitam
anägatam ceti."2 I would interpret this, not entirely in
disagreement with the commentators, as follows : Perception
grasps objects which are present, and inference grasps objects
which are present (sat) as well as not present (a-$at, i.e., the
past and the future objects)...:A~saf is what is past and what
has not yet happened (i.e., might happen in the future).
Vâtsyâyana is apparently reporting here a view which
distinguishes perception and inference according to their objects,
Ui
püryavad ity uktarri kirfi punar atra pürvavat I kitn käryam uta kâranam> yadi
pûrvam asyästiti pürvavat käryatp. pürvavat prâpnoti/tatas ca kâranen a
2
käryönumänam iti vyäghätah. " —Uddyotakara, p. 49.
Compare with Ganganatha Jha's translation (note I in p. 29 of the book)
p. 28.
a view which he obviously supports. What he seems to miss here,

however, is to connect this view about the objects of inference
with the theory of its threefold classification.
Last but not the least, the Carakasamhiiä (which is probably
contemporary with the earlier stratum of the Nyäyasütras1)
records, in fact, three types of inference (Sütrasthäna, Chapter
XI, verses 13-14). Carakà's remark, "pratyak$a~pürvakam trivi-
dharn trikälam cânumîyate" sounds as though it were stating
almost the same theory as that found in Nyäyasütra 1.1.5, viz.,
"tat-pürvakam trividham anumänam, etc." Caraka's examples are:
(a) past (cause) from the present (effect) : sexual intercourse
from pfegnancy.
(b) future (effect) from the present (cause): future fruition
from the seed.
(c) present from present : hidden (nirgüdhd) fire from smoke.
Some remarks: The term sämänyatodrsta is rather a confusing
one. This term seems to have been already in use much earlier
than the Sütrakära. Kajpäda seems to have used this term to
denote a particular type of inference.2 Prasastapâda, following
him, used it in his twofold classification. Sabara also used this
term with slight variation (cf., sämänyatodrsfa-sambandha) in
the Mïmâmsâ tradition. Sabara's example is also identical with
Vat I (c) or UH (c). It is difficult, at the present state of our
knowledge, to learn what the term originally stood for. It might
just have meant the unsophisticated but often correct inferences
based on analogy and previous experience. Or, it might have
meant the type of inference Kanada used to show that sound
(sabda) being a guna or quality proves the existence of äkäsa
(thesky or the physical space) as a substance 3 (i.e., inference
by elimination of alternatives; compare Vât II (b) and (c)
above). But, anyway, at the time of the Sütrakära, its meaning
was vague, so that the Sütrakära might be said to have
redefined the term and used it for the third type of inference as
noted above (Sug. I &II). It is an inference through non-causal
relation, i.e., mere association.
this point A.K. Warder, "The Date of Bhämaha", Journal of Oriental

Research, Vol. XXVI, pp. 94-95.
2
Compare Vaisesika-sütras. 2.1.16, 3.2.6.
3
Cf. Vaisesika-sütra 2.1.24-25 (Gaekwad's Oriental Series 136).
Logic in Ancient and Medieval India 4Î
The term "sesa" might not have a temporal significance. Thus,

the example of UH (b) is, perhaps, more fundamental. But I
suggest that the Sütrakära might have contrasted the term with
''pürva" (which obviously has a temporal significance), and
hence, there has been what is called a change of meaning
through context (prakarana). Vât (II) (b) has also some claim to
be regarded as sesavat, because this type of inference is also
old, since Kanada used it several times in his sütras.
After writing this, I have read an article by G. Oberhammer.1
It is gratifying to see that in many areas both of us have come
to basically the same conclusions. There are two points, how-
ever, where we disagree.
Oberhammer, if I understand him correctly, wants to take
"sämänyatodrsta" to be a description of the general nature
(definition ?) of inference and not as the name of the third type
of inference (p. 70-81). My first objection to this interpretation
is that it leaves the word "trividham" unexplained, since the
third type is neither named nor exemplified. My second objection
is that the definition (laksana) usually precedes division (vibhägä)
and does not follow it. Thus, when "tat-pürvakam" gives the
definition of inference, it is unlikely that the sütrakära will again
describe the general nature of inference at the end of the sütra
without mentioning the third type of inference.
Secondly, the concluding remarks of Oberhammer seem to
suggest that Vrsagana's twofold classification of inference
into sämänyato drsta and visesato drsta (the first being subdivided
into pürvavat and sesavat) was due to the influence of the
Nyäyasütra. But, actually, Vrsagana brought the two different
traditions (twofold classification and threefold classification)
together and had the philosophical insight to see that they were
forming only parts of the whole picture. I do not think that the
sütrakära was aware of the fact that he belonged to the particular
G. Oberhammer : "Zur Deutung Von Nyäyasütram 1.1.5" in Wiener

Zeitschrift für die Kunde Süd—und Ostasiens, 10 (1966). Some portion of
this section was prepared for presentation at the South Asia Section of
XXVII International Congress of Orientalists, at Ann Arbor, Michigan
(USA) on August 18, 1967. For a far richer material on the description of
the theory of inference in ancient Indian texts see Nancy Schuster's
paper "Inference in the Vaisesika-Sütras," Journal of Indian Philosophy,
Vol 1, 1972, pp. 34-395. Some portions here pre-date Mrs. Schusters' paper.
tradition where only sämänyato drtfa inference was recognized.

He should rather be regarded as describing the tradition of three-
fold classification of inference based on the temporal signi-
ficance as I have already indicated.
As I have already said, my interpretation is a guess-work. It
is, at best, an intelligent guess. And, I think, all the later
commentators including Vätsyäyana were doing the same thing,
i.e., making intelligent guesses, while trying to interpret these
three terms. The interpretation suggested above takes into con-
sideration certain historical as well as textual facts which the
other interpretations seem to have ignored. This is all that
can be said in favour of the above interpretation (viz., Sugges-
tion I and II).
§ 1.5 : PSEUDO-REASON (Hetväbhäsa)
In Section 1.3, I have already said that while the Buddhist as

well as the Vaisesika mentions three essential characteristics or
marks of an inferential reason or evidence, i.e., hetu, the early
Nyäya thinks that the hetu should have five, instead of three,
essential marks in order to be able to generate infallible infer-
ences. The dispute is usually referred to as one between trirüpa
and pahcanipa (threefold versus fivefold inference-generating
condition). The term 'hetu' is not easy to translate. I shall
use 'reason' or'evidence' interchangeably to render 'hetu' into
English, although I am aware that 'evidence' may have a much
wider application. (Perhaps, any pramana, properly understood,
would be an evidence.) For 'hetväbhäsa' I shall use 'pseudo-
reason'. Modern writers, being partly misled by their acquain-
tance with the Western logical terminology, use 'fallacy' to refer
to the defect that a reason or evidence may have—the defect that
would incapacitate the reason. In other words, if a reason has
one of these defects (cf., hetu-dosa), it is rendered useless, for,
it would then not be potent enough to give rise to the desired
inference. Since old habit dies hard, I shall use 'fallacy' some-
times in this context, while I would assign to it the sense just
described.
The five marks are :
1. The reason (evidence) must be present in the case under
consideration ( called pak$a, the subject-locus).
Logic în Ancient and Medieval India 43
2. It must be present in another case similar to the case under

consideration (called sapak$a, the homologue, the agreeing
case).
3. It must not be present in cases dissimilar to the case under
consideration, i.e., cases where the inferable property
(the predicate-property) is not present.
4. It must be such that the proposition it tries to establish
should not be contradicted by another already established
truth.
5. It must be such that there should not be another evidence
or reason establishing the opposite thesis, to counter-
balance the thesis it tries to establish.
4 and 5 are called in Sanskrit : abâdhitatva and asatpratipa-
ksitva, 1, 2 and 3 are the same as the 'three marks' of the
Buddhist. Some Naiyäyikas even argued that the three terms in
Nyäyasütra 1.1.5, pürvavat, sesavat and sämänyato'drsta, actually
referred to the first 'three marks' of the hetu. [pürva^paksa,
sepa^sapaksa and sämänyato adrs(a = vipakse sämänyena adrsta].
The same view further maintains that the word"ca" in the sütra
1.1.5, indicates the additional two marks, 4 and 5. But, as
Bhâsarvajna has said, this is a far-fetched and tortuous inter-
pretation (cf., kasta-kalpanâ, pp. 191-92).
Those who regard the five marks as essential face the criticism
that 4 and 5 are rather context-dependent and not directly con-
cerned with the soundness of the reason (or hetu) itself. Those
who regard the 'three marks'as essential face a further criticism.
For, it may be argued that under a reasonable interpretation,
2 and 3 would be equivalent and hence, either of them would
be redundant. I would discuss this controversy in section 1.7.
Let me dwell here on the general idea about what makes a
hetu a defective hetu, a pseudo-reason.
Nyäyasütra 1.2.4 mentions five types of pseudo-reason.
Sütras 1.2.5-1.2.9 try to explain (define) them, but their mean-
ings still remain obscure. I shall try to follow Vätsyäyana here.
The first type of pseudo-reason is, perhaps, the least controversial
one. It is called sa-vyabhicära 'a deviating reason' or we may
call it an inconclusive reason. Vàtsyâyana's example is :
Sound is eternal because it is intangible.
It is explained :
A pot is an entity that is tangible (can be touched) and it is
seen to be non-eternal. Sound is not tangible in the same
way. Therefore, (one may argue) sound is eternal because it is
* intangible. (Answer :) In the example, tangibility and
eternality are not grasped as being connected by way
öf being the reason (or evidence) and what is to be proven
{sädhya). For example, an atom has touch and it is eternal.
When the soul, etc., are examples, the (purported) reason
bears resemblance with the example and hence purports to
prove the sädhya (the desired property, eternality). But the
reason, intangibility, deviates from the property, eternality.
For a cognitive episode is intangible as well as non-eternal.
Since the (purported) reason deviates from the (desired)
property in both examples (atoms and cognitive events), it is
not a (proper) reason, for the definition does not apply.
Eternality and temporality are two mutually exclusive oppo-
sites. If a reason belongs to one or the other (not both), it is
non-deviating. Violation of this rule generates an inconclusive
pseudo-reason, for it belongs to both.
This is only a clumsy way of stating the principle that a pro-
perty in order to be adequate to prove the presence of another
property in the same location must belong only to the cases
where the second property is seen to belong, and should not
belong to a case where the second property is seen to be absent.
The pseudo-reason viruddha is enigmatically defined in NS
1.2.6. Most probably, it meant something very different from
what the later tradition in nyäya understood as viruddha. If we
keep ourselves within the bounds of Nyäyasütra and Bhä$yay
we have to explain it as an 'incoherent' pseudo-reason.
NS 1.2.6 siddhäntam abhyupetya tad-virodhi viruddhah
"An incoherent reason is one which contradicts what has
already been conceded as a tenet."
Vätsyäyana gives an example which tries to show that the
Sätfikhya doctrine of change in the midst of permanence (cf., the
sat-käryaväda) is incoherent, for the reason adduced therein
contradicts what has already been accepted in the system.
The so-called 'mistimed' pseudo-reason (kälätita), the fifth in
the list, remains as another enigma of the old school. The basic
idea was probably that the proper formulation of what we are
going to prove (i.e., the proposition to be proved) would be
enough to show that it cannot be proved. Hence, the adducing
of the reason would not even be necessary. It would be mistimed.
I shall now give a detailed analysis of Vätsyäyana's comments
on the other two pseudo-reasons : sâdhyasama and prakarana-
sama. Since H.N. Rändle wrote a somewhat pioneering work on
these issues (Indian Logic in the Early Schools, Oxford, 1930), I
shall examine some remarks made by Rändle in this connection.
Scholars have usually translated the term sâdhyasama occurring
in early Sanskrit philosophical texts as petitio principii. Curiously
enough, H. N. Rändle used 'petitio principii9 to translate
prakaranasama, one of the five defects of inference mentioned in
the Nyäyasütra.1 While commenting on sâdhyasama, however,
Rändle remarked :
This clearly resembles the preceding fallacy (prakaranasama)
in being a kind of begging the question: and this kinship with
the previous fallacy is, no doubt, indicated by the word *and?
with which the present sxitra commences.2
Rändle expressed also a word of caution regarding the use of
'sâdhyasama9 in the Mädhyamika texts. Professor K. Bhatta-
charya in his recent note 3 shows that Randle's misgivings about
the interpretation of sâdhyasama in the Mädhyamika texts were
right. It is also to be noted that the usual translation of sâdhya-
sama in the Nyäya context as petitio principii (which even Rändle
seems to have suggested) is not only imprecise but also
incorrect,
Sâdhyasama, like petitio principii, is a defect in the argument
and, hence, can be better explained in the context of an argument.
First, we must be clear about what we understand by petitio
principii as well as what we understand by a defect or a logical
fallacy. It is better to clarify here the notion of fallacy.
Let us define a fallacy as an invalid or unsound form of
argument. An argument or reasoning involves a transition from a
l
Nyäyasütra I.2.4.
2
Randle, p. 197.
3
SeeK. Bhattacharya's article in Journal of Indian Philosophy (1974),
pp. 225-230.
set of premises or some evidence to a conclusion. A false belief is

not called a fallacy, since it does not involve an argument. From
a practical point of view, however, 'fallacy' is very often used
in a much extended sense. Traditionally, any mistake or confu-
sion in the context of an argument or proof is called a fallacy.
Thus, apart from fallacies that can be called 'formal' because of
their relation with some formal logical system, there are fallacies
in non-deductive inferences, as well as fallacies in philosophic
arguments or discourses. The 'non-formal' fallacies are called
fallacies actually in the extended sense. It is to be noted that an
argument with such a 'non-formal' fallacy could be wrong,
though it may be formally valid.
A petitio principii is a fallacy in the extended sense noted
above. It denotes an argument that 'begs the question'—an
argument that uses the conclusion to be arrived at as one of the
premises. What'we call'arguing in a circle' is also a petitio
principii. We argue in a circle when we defend a statement by
another statement and when asked to defend the second state-
ment we defend it by the first statement.
A petitio principii, it is to be noted, can be a formally valid
argument. A conclusion can very well be entailed by a set of
premises that happens to include the conclusion itself. Thus,
arguing in a circle can be formally valid!
A petitio principii, therefore, is very different from a fallacy
like (say) the illicit minor in the traditional theory of categorical
syllogism based upon what is called 'the distribution of terms'.
An illicit minor is said to occur when the 'minor term', i.e., the
subject of the conclusion, is distributed in the conclusion (in
other words, the conclusion is an A or E proposition) but not
in the minor premise. In short, it is a fallacy, since it violates
a formal rule of categorical syllogism.
It is also instructive to note Aristotle's comment on petitio
principii:
Petitio principii falls within the class of failure to prove the
thesis to be proved; but this may happen in several ways.
One may not reason syllogistically at all, or one may use
premises no better known than the conclusion, or logically
posterior to it. None of these constitutes petitio principii.1
Aristotle, Prior Analytics 64& 28-32, p. 143.
I shall later come back to the detailed comments of Aristotle

on petitio principii. With the above as a prelude, let us now
investigate the meaning of the term sädhyasama in early Nyäya
texts. Nyäyasüira 1.2.8 says :
sädhyävisistah sädhyaivät sädhyasamah.1
Rändle translates :
(And) a reason which is indistinguishable from probandum in
respect of having to be proved is called the reason which is
'identical with the probandum.2
I suggest the following translation :
A reason that is non-distinguished from what is to be proved
is called sädhyasama ( = one having the same predicament as
that of what is to be proved) because it (itself) needs to be
proved.
Now let us follow Vâtsyâyana's explanation of this sütra*
The proposition to be proved is : Shadow is a substance. 4
The reason is that it moves. This reason is not distinguished
from the proposition to be proved. It is in the same predica-
ment with what is to be proved because it itself needs to be
proved. (To explain :) This reason is itself unestablished
(unproved) and, hence, should be established as true (proved)
just as the proposition to be proved is supposed to be. One
has to decide (establish) the following : Whether a shadow,
like a man, moves, or whether what is apprehended is a series
of non-presence of light arising out of a series of succes-
sive coverings as the covering substance moves. (In fact)
the non-presence of those portions of light that are cut
off (covered) by the moving substance is continuously
apprehended. 'Covering' means the rejection of its (the light's)
approach (to some place).
reads a *ccC (='and') in the sütra, but no 'oz' is present in G. Jha's
reading. See Jha's edition, p. 63.
2
Randle, p. 197. For the reason explained in note 1 in p. 47 of the book,
I have put 'and' in parentheses.
3Vätsyäyana, pp. 63-64.
4
The term sâdhya in this context stands for the entire proposition to be
proved. And likewise the term heiu means the proposition used as a
premise of the argument.
Vâtsyâyana's explanation of sädhyasama, as will be clear from

the above, enjoys the benefit of both worlds. While it is suitable
as a defect of inference as it was in the sütra, it also contains
the seed of the later theory by which sädhyasama can be turned
into a well-entrenched fallacy, viz., the svarüpäsiddha of the later
Nyäya. The earlier theory of inference in the Nyäya tradition
was something like the following. An inference was treated as a
transition from one statement or cognition to another statement,
or cognition with the implicit help of an implication-rule. The
first statement ( premise) was called a hetu 'reason', and the
second statement was called rd sädhya 'proposition to be proved'.
We can illustrate the process as follows :
Hetu 'reason' :—a shadow moves.
Sädhya 'what is to be proved5 : a shadow is a substance.
The required implication-rule : Having motion implies being
a substance.
In the explicit statement of 'steps' in the argument (which is
technically called nyäyaväkya or avayava, or the so-called nyäya
syllogism), the third 'step' is called udäharana 'exemplification',
which consists in the statement of the implication-rule along
with an instantiation of it, i.e., an actual 'example' (-=drstänta).
An 'example' is defined as a case which is unanimously estab-
lished.1 Care should be taken to note that the conclusion arrived
at as the fifth 'step' in this process of argument, is neither
identical nor equivalent to the first 'step' called pratijnä, i.e., the
statement of the sädhya or thesis to be proved. The conclusion
arrived at in the last 'step' is a mixed judgement of the form :
"Since a shadow moves, it is a substance."
It is assumed in this theory of inference that the statement
called 'reason' hetu (sädhana, not sädhana-dharma) must be
established beyond doubt (i.e., its validity should be established)
before it can warrant the transition from it to the concision.2 In
other words, statement of the reason should not have the same
dubious status (as far as its validity is concerned) as the sädhya
1
Nyäyasütra î.1.25.
2
Compare Aristotle's comment in a similar context : "Demonstration
always proceeds from what is prior and more certain." Prior Analytics*
64T> 32-33.
'proposition to be proved'. It should be noted that the validity

of the proposition to be proved, before the process of inference
is complete, is doubtful, i.e., not yet established. Inference,
according to Nyâya, is a passage from the known or 'establised'
to the unknown or 'doubtful'. This point is emphasized by the
Nyäyasütra when it accepts samsaya 'doubt' as the very starting
point of the nyäya debate procedure. As Vâtsyâyana himself
points out, we use inference with regard to something about the
validity of which we entertain a doubt.1 But the reason adduced
for the required conclusion cannot be equally in doubt. It cannot
be an uncertain premise, for, then it cannot be used to certify
the thesis or the proposition to be proved. However, when the
reason adduced is less certain than the proposition it tries to
prove or establish, the fallacy called sädhyasama arises. This
fallacy, then, is not certainly what we understand by a petitio
principii. To recall the comment of Aristotle (which apparently
was the source of Randle's logical terminology) :
...but this (failure to prove the thesis to be proved) may
happen in several ways. One may not reason syllogistically
at all, or one may use premises no better known than the con-
clusion, or logically posterior to it. None of these constitutes
petitio principal
For, according to Aristotle :
It h petitio principii when one tries to prove by means of itself
what is not self-evident.3
It is clear that when the second situation mentioned by Aristotle
(see the italicized expression in the passage quoted above)
obtains, we have a case of sädhyasama. Aristotle in no uncertain
terms denies this to be petitio principii.
It is, therefore, difficult to see why modern scholars who
were supposed to be familiar with Western logical terminology
would interpret, as they did, sädhyasama as 'begging the
1
Vâtsyâyana, p. 3 : i(tatra nänupalabdhe na nirnïte^rthe nyäyah pravartate
kirn tçtrhi, sarpsayite 'rthe. "
2
Aristotle, op. cit. The expression within parentheses is mine, and the first
italics is mine for the sake of emphasis.
z
Ihid. 641) 36-37. Another version of the same is : '*...whenever a man tries
to prove what is not self-evident by means of itself, then he begs the
original question.", p. 93 of The Basic Works of Aristotle.
question' or 'petitio principiï or even 'arguing in a circle'. If I

have to express the Sanskrit term sädhyasama by an adequate
English expression, I would call it the fallacy of 'having the
same predicament with the proposition to be proved'.
The upshot of the above discussion is this. In a comparative
study of logic, if we blur the distinction rather than underline it,
this will give rise to further confusion rather than clarification.
The method of comparative study in logic and philosophy
should be used with great care, caution, and with adequate
checks and control over our tools.
In the above discussion I have interpreted the terms sädhya
and hetu as statements or propositions. It is important to realize
that in the earlier texts these two expressions were used to mean,
more often than not, statements or propositions rather than
what we call 'terms' (i.e., constituents of a proposition). Even
the expression paksa was used sometimes synonymously with
sädhya 'proposition to be proved'. However, the main use of
paksa in earlier texts was in connection with its counter-term
pratipaksa. In this context, paksa meant the thesis of the propo-
nent and pratipaksa meant the counter-thesis of the opponent.
In other words, paksa and pratipaksa are used to denote two
sides of an argument or a controversial issue. Thus, in the defini-
tion of nirnaya 'decision' in Nyäyasütra 1.1.41, it is said :
'vimrsya paksa-pratipaksäbhyam arthävadhäranam nimayah.9
'After deliberating over the thesis and the counter-thesis,
when a certainty with regard to an object is reached, it is
called a DECISION.5
When the later theory of inference was developed by Dignâga,
Prasastapâda and Uddyotakara these three expressions, paksa,
sädhya and hetu, were used to denote the thres 'terms' with the
help of which thé process of inference functions. In this 'new'
theory, paksa meant the locus of inference, the substratum in
which the other two properties, sädhya and hetu, were supposed to
reside. Sädhya meant the inferable property which was intended
to be established as occurring in the 'locus' paksa. And hetu meant
the other property which was already established as occurring
in the 'locus' and invariably connected with the sädhya as well.
Inference, under this new theory, is interpreted roughly as a
process by which we try to establish that A ( = paksa) is charac-
terized by C ( = sädhyä) on the basis of C's connection with

B (=*hetu) and B's presence in A. Certain formal rules of infer-
ence were formulated in this system. Violation of these rules
gives rise to 'defects' or 'fallacies' in our extended sense.
Thus it is that if B's presence in A is in doubt, i.e., 'not
established' (a-siddha) as certain, we would have a fallacy in the
system called svarüpäsiddha, 'the essentially unestablished reason'.
Here, again, Rändle confused the issue by comparing this fallacy
with the western schoolmen's 'illicit process of the minor'. 1 The
illicit process of the minor, as I have noted earlier, was based
upon the very confused theory of distribution of terms. But,
as P. T. Geach has shown, this doctrine of distribution of terms
was 'wholly unknown to Aristotle', and it is such an incoherent
theory that it should be discarded so that logic may pro-
gress. 2 The fallacies that were supposed to be captured by
the theory of distribution of terms, however, need not be
discarded. For, a more coherent theory can be formulated in
order to locate these fallacies in the forms of argument. It
should be noted that a doctrine parallel to the doctrine of
distribution of terms was never formulated in the Indian
tradition. Hence, although for the sake of convenience we can
describe pak$a, sädhya and hetu as 'terms', as opposed to
propositions, they do not behave essentially as 'terms' the way
the minor, major and middle terms are supposed to behave
under the theory of distribution.
To distinguish between the two senses of sädhya, sädhya as a
proposition to be proved and sädhya as the inferable property,
Vätsyäyana sometimes used the expression sädhya-dharma to
denote the latter. This procedure was also followed by Digoäga
and Uddyotakara. But as this never became a strict rule of
usage with these three authors, we have many occurrences of
sädhya which are systematically ambiguous in meaning in
earlier texts.
Let us now examine the notion of prakaranasarna, which,
according to Rändle, closely resembles the notion of sädhyasama.
In fact, Rändle calls prakamnasama directly petitio printipii.
This was another case of incorrect interpretation by Rändle, as
*Randle, p. 251n.
2
See Geach, Reference and Generality, Cornell: Ithaca pp. 3-21.
we shall see presently. Moreover, Rändle was misled by Vätsyä-

yana's example in this case and reached a very strange conclusion
that prakaranasama had 'affiliation' with the asädharana fallacy
of the later school !
The prakaranasama is defined as follows :
Nyäyasütra 1.2.7: yasmät prakaranacintä sa nirnayärtham
apadistah prakaranasamah-
Rändle translates :
When the quality from which the question arises is adduced
as proving (one of the alternative), the reason is called
prakaranasama, petitio principii, 'identical with the question'. 1
I suggest the following translation :
If the situation (reason) from which the deliberation' (cinta)
regarding a 'controversial context' (—prakarand) arises, is
adduced as a reason for reaching a decision, it is called
prakaranasama, (the fallacy of) 'indecision'.
Both terms, prakarana 'controversial context' and cintä 'deli-
beration', are used in this context in their technical sense, as
Vâtsyâyana clearly explains. I shall now translate Vâtsyàyana's
explanation of this sütra : 2
'Controversial context' means the (presence of) two sides,
thesis and counter-thesis, both of which are subjected to
doubt, and (hence) of uncertain validity (for the moment).
'Deliberation' about it means the investigative procedure
starting from the arising of doubt up to the step that is prior
to the decision. When that which gives rise to this controver-
sial context is applied (as a reason) for reaching a decision, it
is called prakaranasama, and since this reason is common (i.e.,
commonly applicable) to both the (opposing) sides, it does
not lead to a decision. To illustrate : sound (noise) is imper-
manent (non-eternal) because we do not apprehend in sound
properties of eternal things. It is seen that in pots, etc., we do
^Rändle, p. 195. .
2
Vätsyäyana, pp. 62-63. The passage enclosed in brackets (...), is not found
in Jha's printed text. But MM. Phanibhusana Tarkavagisa in his Bengali
edition notes that this additional reading is found in some manuscripts.
The additional reading obviously explains the example a little further.
not apprehend the properties of eternal things and they are

non-eternal.
[Sound is eternal because we do not apprehend in sound
properties of non-eternal things. It is seen that in the sky, etc.,
we do not apprehend the properties of non-eternal things, and
they are eternal.]
A property that is common to two (alternative positions)
causes doubt (comprising those two alternatives). When such
a common property is adduced as the reason, it becomes a
samsayasama, which is just a sa-vyabhicära (a fallacy of)
'inconclusive reason'. But the fact that doubt depends for its
(eventual) resolution upon (the awareness of) a specific pro-
perty (belonging to either side) as well as the fact that no
specific property is apprehended in either side, gives rise to
a 'controversial context'. Just as the properties of eternal
things are not apprehended in sound, so also the properties
of non-eternal things are not apprehended therein. This non-
apprehension of the specific property in either side gives rise to
the 'deliberation' regarding the controversial context. How ?
For, the controversy ends if it happens otherwise. If the pro-
perty of eternal things is apprehended in sound, the contro-
versy ends. And if the property of non-eternal things is
apprehended therein, the controversy ends as well. Thus,
this reason gives rise to both positions (in the cotroversy),
and does not lead to any decision with regard to either of
them.
If we follow the above explanation of Vätsyäyana, it is hard
to see why prakaranasama should be indentified with petitio
principii. Even Randle's literal translation of prakaranasama as
'identical with the question' is puzzling. Prakaranasama should
better be translated as 'the fallacy of indecision'. It is certainly
not a formal fallacy, and in this respect it agrees with sädhya-
sama 'the fallacy of having the same predicament'. It is necessary
to distinguish this fallacy from a well-known fallacy like sa-
eyabhicära 'the fallacy of deviation or inconclusive reason'. The
latter fallacy arises when the required implication-rule does not
hold. For example,
Sädhya 'the thesis to be proved' : Sound is eternal.
Hetu 'reason' : For, sound is intangible.
The required implication-rule : Being intangible implies being

eternal. \
But this implication-rule fails because of the citation of a
counter-example : States of consciousness which are non-eternal
(transient) are also intangible. I have already discussed this hetu-
dosa. Let me further deal with this so-called 'defect' of the reason.
'Vyabhicära9 means a counter-example which falsifies the required
implication-rule. The technical term sa-vyabhicära means a case
where a counter-example is present. When the reason adduced
allows a counter example, the required implication-rule fails.
Vätsyäyana's explanation of this fallacy under Nyäyasütra
1.2.5 was a bit involved and somewhat confusing. (See before).
He attempted to explain not only the term sa-vyabhicära but also
the term anaikäntika in the same breath. The s dira explained
sa-vyabhicära as anaikäntika. This means that anaikäntika was
more familiar and less technical in meaning than sa-vyabhicära.
The term anaikäntika was, perhaps, in use in the earlier texts
which discussed the rules of a philosophic debate or controversy
with two opposing sides, a thesis and a counter-thesis. For
example,
Thesis Counter-thesis
Sound is eternal. Sound is non-eternal.
Hetu : For, sound is intangible.
Now, being intangible can neither imply eternalness nor non-
eternalness. Or, to put in another way, being intangible can as
much prove being eternal as it can prove being non-eternal. In
fact, being intangible is the property common to both sides.
Just as an eternal thing, the sky, is seen to be intangible, so
also a non-eternal thing* a state of consciousness, is seen to be
intangible. Thus, being intangible is anaikäntika, i.e., indecisive
in favour of either side, the thesis or the counter-thesis. What was
called anaikäntika in the context of a debate, was explained as
sa-vyabhicära in a general theory of inference. This, I think,
accounts for the obscure explanation of 'the inconclusive reason*
in the early texts.
It should be noted that there is some similarity in this regard
between a sa-vyabhicära and a prakaranasama. Both resist a
decision in favour of either side. That is why, Vâtsyâyana tried
to distinguish prakaranasama from samsayasama^savyabhicära
while he was explaining the former (see above). The point to be

remembered in this connection is as follows. The savyabhicära
'inconclusive reason' consists in the employment of a common
property {säähärana dharma) as the reason which will result in
a doubt of the form : "Perhaps sound is eternal, or perhaps it
is non-eternal'.1 But, in prakaranasama 'the fallacy of indecision',
no common property is adduced as the reason so as to give rise
directly to a doubt of the above form. What is rather emphasiz-
ed in this case is the lack of knowledge about a specific property,
i.e., a decisive factor, in favour of either side. And this lack of
knowledge of a decisive factor leads.only to an indecision.
It is, however, enough for the fallacy of deviation (or,
'inconclusive reason') toarise, if there is deviation ( = vyabhicära),
i.e., if there is a counter-example. This requirement about the
absence of deviation, i.e., absence of a counter-example,
violation of which results in the fallacy of deviation, gradually
became prominent in the later theory of inference, for, it
actually supplied the universal relation needed as the basis of a
correct inference. It is to be noted that although in the Nyäya
tradition the theory of inference needed, like the traditional
Western syllogism, two premises, it contained one particular
premise (a singular proposition, which was called hetu) and an
implication-rule. However, implication with impossible antece-
dents was excluded. The theory insisted on an actual instantia-
tion of the implication-rule, a drstänta or a case where the rule
is illustrated. The requirement about instantiation made
another point clear in Indian logic, a point that was left
ambiguous in the traditional Western logic: the universal pro-
position that might express the implicative premise must not
have its subject-class empty.2
Rändle commented in the context of his explanation of sa-
vyabhicära as follows:3
The quantitative formalism of the western syllogistic is, there-
fore, quite alien to Indian logic: and the attempt to identify
the two schemes of fallacy can only, lead to confusion.
1
See Nyäyasütra 1.1.23.
2
For criticisms of such ambiguity in traditional Western logic, see Strawson,
pp. 163-179. It is important to realize the function of examples in the
Indian theory of inference in the light of these comments.
»Rändle, p. 193.
It is worth pointing out that this remark was based on a con-

fusion. Rändle was, of course, right in his second statement
here. But what he had in mind while he made the first state-
ment is difficult to understand. In a footnote to this sentence, he
added that the Vaisesika-Bauddha logician explained the fallacy
of deviation from a formal standpoint, but that the earlier
Nyäya school was concerned to examine only examples and
counter-examples. The comment was obviously based upon the
unacceptable view that the early Nyäya theory of inference was
merely that of analogical inferences based upon examples and
counter-examples. This view has been widespread among the
Indological scholars, but it should rather be discarded. It is
true that Vâtsyâyana's explanation of anaikäntika in the context
of a public debate, with examples and counter-examples, was a
bit misleading. But the truth is that Vätsyäyana was a
clumsy writer and, hence, confusingly reported the history of a
term along with its logical role. We should know, however, that
examination of examples and counter-examples does not, by
itself, make a theory of logic less formal or less general.
Let us revert to the distinction between the fallacy of
îndecision' (prakaranasama) and that of the 'samepredicament'
(sädhyasama). I have already established above that they are
not petitio principii, nor 'begging the question' in particular.
The situation in the Nyäya school was further complicated by
the citation of the same terms to denote two of the twenty-four
different ytfffs' 'sophistical refutations'. Nyäyasütra 5.1.1 supplies
the listj ojf twenty-four sophistical refutations or jätis in a
philosophic dispute or debate. (See § 1.2). The eighth and the
fifteenth of this list are also called sädhyasama and prakarana-
sama. To some, this has been a bit confusing. But, if we keep
in mind the distinction between the two general categories
under which they are included, there should not be any con-
fusion regarding their interpretation. One is the category of
fallacies (cf., hetväbhäsa), which we have discussed above. The
other is the category of sophistry (cf., jäti). A fallacy is not a
sophistry. A fallacy is a mistake in the argument, in this case,
an inference. But a sophistry is a deliberate use of unsound
reasoning, of a fallacious argument, for a rejoinder. One can
indulge in a sophistical device by intentionally using a fallacy
to deceive the opponent or to win an argument unfairly. Or
it may be used to confuse the issue for the time being. Thus, the
sophistical device (i.e., the use of jäti) is explicitly excluded from
a debate which the Nyäya calls the väda variety of debate.1 Most of
the sophistical refutations found in the Nyäya list assume that
the argument used as a rejoinder is not based upon the required
implication-rule of inference, but only on some superficial simi-
larity or analogy. The very fact that the early Nyäya school
recognized these rejoinders as sophistical and also discussed the
means by which one should solve problems posed by such
rejoinders (cf., uttara), goes a long way to show that the early
Nyäya theory of inference was not based upon analogy and
examples. And the doctrine of hetväbhäsa 'fallacious reason'
certainly required that inference in the early Nyäya school must
be based on sound implication-rules.
A brief comment on the use of the term sädhyasama in the
early Mädhyamika texts may be in order (for more on this
point see the article by K. Bhattacharya). In the Mädhyamika-
kärikä, Chapter IV, Nägärjuna notes in the two concluding
verses (8, 9) a general pattern of most Mädhyamika arguments.
" The simple meaning of verse 8 here seems to me to be this: If
a reason is adduced to refute the Mädhyamika thesis of empti-
ness in a philosophic dispute (where an opponent is present),
It will not constitute a refutation, for, it will enjoy the same
predicament along with the proposition to the proved. The
second verse, verse 9 in the context, says the same thing with a
slight variation : If a reason is adduced for questioning the validity
of the Mädhyamika thesis of emptiness in the context of an
explanation of the doctrine to the disciples (cf., vyakhyäna) ^ it
will not be a proper question, for, the reason will have the same
predicament as that of the proposition to be proved. Or, the
point may be that in either case (i.e., either in a dispute or in
vyakhyäna) the supposed refutation of\ the doctrine of emptiness
will be baseless, since it will be a f^lse refutation (a jäti) liKe
the sädhyasama. We should note here that no matter whether
sädhyasama is interpreted as a fallacy or as a sophistry, the
Mädhyamika thesis of emptiness will remain unharmed, i.e.,
unrefuted.
Thus, I venture to suggest that the term sädhyasama as it
*See Nyäyasütra 1.2.1-1.2.3.
was used in the early Mädhyamika texts was not, contrary to

what Rändle suggested, very different in meaning from the
term sädhyasama found in the early Nyàya texts. In fact, once
we have realized that Randle's interpretation of the sädhyasama
in the Nyäya texts is wrong, we can see that both the Nyâya and
Mädhyamika texts use sädhyasama in more or less the same
manner. However, not a few modern scholars of repute
uncritically assumed that Randle's interpretation was correct.
It is important to realize that in the recent history of the study
of Indian logic, once an eminent scholar commits a mistake
in interpretation (for whatever reasons), most later scholars
accept the interpretation as valid without going into the trouble
of examining it further. But, they then often worry about other
related problems that such a wrong interpretation will generate.
I am, however, in agreement with K. Bhattacharya, and also
in partial agreement with Rändle, on the point that the term
sädhyasama in the early Mädhyamika texts cannot be interpreted
as petitio principii. For, it is one thing to say that the conclu-
sion is assumed in the premises, and another to say that the
reason or the justifying example (which was in the Mädhyamika
argument the concept of rüpaskandha 'aggregate of matter')
needs to be established beyond doubt before it could claim any
justificatory force. Just as a blind person, to quote a Sanskrit
proverb, cannot lead another blind person to his destination,
so also a reason which is itself in need of being established
cannot establish another proposition.
§ 1.6 : WHAT DO WE INFER ?
In one's theory of inference, the question 'What is known

through inference?' (in other words, 'What is the nature of anu-
meya ?') seems to be a pertinent question. In developing his
theory of inference, a logician must at a certain point answer
this question. In ancient India, at a particular period, a great
deal of discussion centred around this crucial question.
Dignâga, for instance, discussed several alternative answers to
this question, none of which he found satisfactory. At the end
of his discussion, Dignäga gave his own view about anumeya.
Uddyotakara1 under Nyäyasütra 1.1.5 referred to several other
alternative answers to the above question. He criticized, as usual,.
1
Nyäyavärttika, Chowkhamba, Varanasi, 1915, p. 50-51.
Dignäga's view and incidentally mentioned a new view which

he supported. But inspite of Uddyotakara's critique, Dignäga's
view seems to have been more acceptable to later logicians
like Kumärila.1
Dignäga's discussion of anumeya is contained in the four
verses (verses 8-11) of Chap. II of Pramänasamuccaya. Although
the original Sanskrit work is lost, these four verses are found
in Vacaspati Misra>2 Nyäya-värttika-tätparya-Ukä. Dignäga's
own Vrtti (to be called Svavrtti henceforward) on these verses
are available only in Tibetan translation.3 H.N. Rändle4 translated
the Sanskrit verses into English in his Fragments from Dignäga.
Rändle did not use the Tibetan Svavrtti, but closely followed
Kumärila's rephrasing of Dignäga's discussion in his English
translation of these four verses. It will be shown here that
Rändle misinterpreted Dignäga in at least one place. It will be
shown further that while Dignäga's main thesis on the nature
of anumeya (i.e., dharma-vis ista-dharmy anumeya) was well
understood by later logicians, certain other points in his critique
of alternative views were misunderstood and wrongly interpreted
by his successors in Sanskrit philosophy. This misunderstand-
ing started, perhaps, with Vacaspati Misra or some time after
him. It will also appear that the later Naiyäyikas like Vardha-
mäna Upädhyäya did not consult Dignäga's own Vrtti on these
verses, although it is conceivable that the original Sanskrit
1
Mîrnamsa'éîokavârttika, Chowkhamba, Varanasi, 1898-1899, Anumäna-
pariccheda verses 23-24.
2
Nyäyavärttikatätparyatikä, Kashi Sanskrit Series, Varanasi, 1925, p. 180.
The verses are : {Pramänasamuccaya, Chap. II) :
kecid dharmäntaram meyam lingasyävyabhicäratah\
sambandham kecid icchanti siddhatvâd dharmadharminoh //8//
Ungarn dharme prasiddharn cet kirn anyat tena miyatej
atha dharmini tasyaiva kim-arthatn nänumeyatä //9//
sambandhe 'pi dvayam nâsti sas thi tniyeta tad-vati/
aväcyo 'nugrhitatvän na cäsau linga-samgatah //10//
lingasyävyabhicäras tu dharmenänyatra drsyate /
tatra prasiddham tad-yuktaifi dharminam gamayisyati / / l l /
3
Tibetan Tripitaka, Peking edition, Mdgho-rel XCV, Ce. No. 5701 :
Vasudhararaksita. No 5702 : Kanakavarman.
^Fragments from Dignäga, The Royal Asiatic Society, London, 1926,
p. 18-21.
Svavrtti of Dignäga was available at the time when they wrote

their own work.
It will be necessary for our purpose to give an explanatory
translation of the four verses of Dignäga in the light of his
own Vrtti on them. The Tibetan version1 of the Vrtti will be
used to settle ambiguity of the verses. Pramänasamuccaya,
Chap. II, verse 8 states that, according to some, a different pro-
perty, viz., fire, is what we infer from the known property
(smoke), because the inferential mark, smoke, is in invariable
relation with that different property, viz., fire. (Let us call the
upholder of this view Opponent A.) According to others (let
us call this group of thinkers Opponent B), it is the relation of
fire to the place, mountain, is what we know through inference,
since both the dharma (the property fire), and the dharmin (the
property-possessor mountain) are known already as established
objects. (It is understood that the object of our knowledge
acquired through inference cannot be an already known object.)
Criticizing the first view (Opponent A), Dignâga says in verse
9: If the property, smoke, is well-known as being related to fire
it would not be proper to say that we infer fire from smoke.
(It is to be understood that when seeing a body of smoke one
remembers the relation between fire and smoke, the other
property, fire, becomes an already known object, and, hence,
there will be no room for inference). If however, it is asserted
that we infer the other property, fire, as characterizing the locus,
mountain, why do we not say that the place itself which
possesses fire is what is our anumeya ?
The second view (that of Opponent B) is criticized in verse
10: The relation (sambandha) in question does not possess the
two characters (viz., fire and smoke) which an object should
possess in order to be an anumeya. (It should be admitted that
the objectof inferential knowledge, i.e., the anumeya, is con-
nected, in the first place, with the sädhana, viz., smoke, and
then with the sädhya, viz., fire). If the relation between the
mountain and fire were our anumeya, one would expect in the
verbal expression of our inferential thesis {pratijnä) a genitive
case-ending in the word expressing the adjunct of such relation.
1
I like to thank here my esteemed friend Professor Masaaki Hattori of
Kyoto University, Japan, who helped me in reading the Svavrtti of Digaäga
upon the aforesaid four verses.
In other words, the verbal expression would be of the form

"There is the relation of fire here {atra vahneh sambandhah) ",
and not of the usual form "There is fire here (atra vahnih).9*
In asmuch as the said relation is only implied, not expressed,
by the inferential thesis "atra vahnih" we have no right to call
such relation the anumeya. Moreover, the said relation is not
connected with the sädhana (viz., smoke, the inferential mark).
Rändle gave a slightly wrong interpretation of this verse.1 He
thought that if the said relation was accepted as anumeya we
would expect the genitive in the inferential thesis in order to
express that relation. He contrasted the usual form of inferen-
tial thesis "There is fire in the mountain (parvate 'gnirasti)" with
the unacceptable form "There is fire of the mountain (parvat-
syägnir asti)". He also quoted Pânini's rule "fafthi sefe" II, IIL
50 and indicated that the genitive expresses mere relation, and
not a käraka-fünction like the locative. But it is evident from
our above interpretation that these remarks of Rändle were
beside the point. Rändle was thus puzzled when he saw
that another acceptable form of the inferential thesis "The
mountain is fiery (parvato vahnimän)" does, contain the posses-
sive suffix matup ( = män) which also expresses mere relation.2
But, this confusion stems mainly from his misunderstanding of
Dignäga's point about the genitive case-ending.
Verse 11 gives Dignäga's own view about anumeya. The
invariable connection of the inferential mark, smoke, with the
property fire is seen in other places; and the inferential mark,
being well established in such other places, leads us to infer the
present property-possessor (dhanniri), viz., the mountain, possess-
ing the property, fire.
Uddyotakara inserted the discussion of anumeya while he
was explaining the meanings of the terms pürvavat, etc., in
Nyäyasütra 1.1.5. The gist of his arguments on the notion of
anumeya can be stated as follows. What we know by inference
through the inferential mark, viz., smoke, is not-another pro-
perty, viz., fire, nor the place itself, nor the existence of fire, nor
even the place as possessing fire. The anumeya, i.e., the object
of inference, must be related to the inferential mark (sädhana)
as a property to a property-possessor. Moreover, the object of
1
See Fragments from Dignâga, p. 20 and footnotes 1 and 2 therein.
2
Ibid.9 p. 20, footnote 3.
62 Logic» Language and Reality
inference should not be known in advance. None of the above

things, viz., the property like fire, the place, or the existence
of fire, can be said to be related to smoke, the inferential mark,
as its property. And all of them become known before we
reach the conclusion of the inference. The particular place,
viz., the mountain, which possesses fire (Dignäga's own view)
cannot also be regarded as the anumeya, because smoke, the
sädhana, is not known to be a property of this complex (viz.,
the mountain characterized by fire). We do not actually see the
place that possesses fire, but we see only smoke coming
out of a particular place. If Dignäga says that the observed
smoke by virtue of its invariable connection with fire leads us to
infer the particular place as possessing fire, it might be asked,
what is meant by this invariable relation (avinäbhävasambandhä)?
It cannot mean an invariable co-occurrence in the same locus,
because a body of smoke is sometimes seen hovering in the sky
without fire. It is also seen that a body of fire (in a red-hot
iron ball) is unaccompanied by smoke. Thus, since the invari-
able relation itself, of smoke with fire, cannot be established
beyond doubt, it is not proper to say that we infer the place as
possessing fire through the inferential mark, smoke, on the basis
of its invariable relation with fire. What we infer is that this
body of smoke (perceived on the mountain) possesses fire too
(as its qualifying adjunct). According to Uddyotakara, in an
inferential situation, the particular smoke is perceived and along
with it the property of its continuously rising upwards from a
particular source and such other characteristics (cf>9 sätatya-
samhatya-ürdhvagati-svabhävädayo dharmäh, Uddyotakara, p. 51)
are also perceived. What is not perceived is that this particular
smoke is also characterized by, i.e., connected with, fire. Hence,
this unperceived factor viz., this smoke possesses fire, is what is
grasped by inference. Uddyotakara repeated the same view
about anumeya under Nyäyasütra 2.1.46.1
This view about the nature of anumeyah a bit peculiar. Instead
of accepting the propositional complex, the mountain pos-
sessing fire, as the object of our inference, we have to admit,
under this view, another complex, viz., this particular smoke,
because of its certain peculiar characters, is related to fire also,
ïNyâyavarttika, Chowkhamba, Varanasi, 1915, p. 258.

as the object of inference. Jayanta Bhatta1 mentioned this view

about the anumeya in a couple of places of his Nyäyamanjarü
As far as I can see, this view seems to be a more sophisticated
rephrasing of an older view. According to the earlier logicians,
inference is what extends our knowledge by informing us of
the unperceived fact from the perceived one. Thus, inference
becomes a function of two terms only, the property to be
inferred and the inferential mark.2 In this theory of inference,
the minor term (to use Aristotelian terminology) does not play
any important role. From the existence, or presence, of certain
property (or, properties) we infer the existence, or presence,
of certain other property accompanying the former property.
Uddyotakara gave simply a refined version of this theory. Even
the well-known case of inference "Sound is non-eternal because
it is produced" was understood by Uddyotakara in this way.
Sound or noise is a known entity, so is its property of being produ
ced. Since the non-eternalness of sound is not known, we infer
non-eternalness of sound from the property of its being produced.3
Uddyotakara's critique of Dignäga's view about anumeya
sounds all right as far as it goes. It was admittedly not clear in
Dignäga's exposition whether the locus (i.e., the place), which
is the subject of the inferential thesis, should be known to us
prior to the inference or not. In fact, Uddyotakara bases his
objection upon the principle that the property-possessor
(dharmin) as well as the inferential mark becomes known to us
1
Compare : yathä agnimän ay am dhümah bahalapändutädidharmayogitvän
mahänasävadhrtadhümavad iti, dhüma evägnimattayänumlyate. ..."
Nyâyamanjarî, Kashi Sanskrit Series 106, Varanasi, 1934, p. 117. Compare
also Jayanta's remark on p. 118. Curiously enough Jayanta quoted
Kumärila Bhatta in both the places. But Kumärila, as I have already
stated, supported Dignäga's view about the nature of anumeya although
he mentioned other alternative views. Jayanta, on the other hand, had
something like Uddyotakara's view in mind.
2
The crudest form of inference was to derive the knowledge of x from
our knowledge of y. Thus, in the earliest phase, the Vaisesikas inferred
the existence of soul, mind or the physical space (äkäsa) through our
knowledge of their qualities such as cognition, desire and sound. In
Nyâya tradition also, we infer future rain from the cloud and past rain
from the fullness of the river.
3
Compare: sabdasyätmasattä prasiddhä krtakatvam ca dharmas tv anitya-
tvalaksano 'prasiddha itijtad-visesano 'yam anumiyata iti, (Nyäyavärttika,
p . 51).
(prasiddha) and hence cannot form a part of our anumeya.

Besides, Dignâga's use of the term "property-possessor (dharmin)"
was ambiguous. To avoid the Lokäyata critique of inference,
Uddyotakara pointed out that not in ali cases, our perception
of smoke will validate our inference of fire, but only of some
particular type of smoke, such as, the one that continuously
rises from a spot. If the locus or the minor term is thought to
be a part of the inferential mark (the middle term), we can
happily explain the earlier examples of inference found in the
Nyäya tradition. Thus, we infer the future rain- from the cloud
and the past rain from the fullness of the river.1 Later com-
mentators, however, analysed these examples in such a way
that we get the cloud as the minor term (paksa or the locus of
inference) in the first case and the river (or the fullness of the
river) in the second.2 The middle term or the hetu crystallized,
in the hand of the later commentators, as some specific character
or characters (visesa) of the minor term (or paksa), such specific
character being experienced to be invariably related to the
inferable property (the sädhya) or the major term.
Dignâga's theory of logic, like other contemporary systems
of Indian logic, was based upon the principle that inference is a
function of three terms, the property to be inferred (sädhya),
the inferential mark (sädhana) and the locus (paksa) or the
property-possessor. Dignâga's own view about anumeya (viz.,
we infer the propositional complex, the mountain possessing
fire) was quite consistent with this principle of inference with
three terms. Thus, Kumärila closely followed Dignäga in his
discussion of anumeya. The two views (Opponents A and B)
rejected by Dignäga were also noted by Kumärila; and he
correctly understood the position of Opponent B. Vacaspati
Misra, after quoting verses 8-10 from Dignâga's Pramänasamuc-
caya, Chap. II, inserted the following explanatory remark: nahi
sambandhadharmataya Ungarn pramlyate, api tu desa-samgatam
ity arthah. But this explanation, presumably of verse lOd of
Dignâga's work, was not consistent with Pärthasärathi Misra's
•'•See Vätsyäyana's examples under Nyäyasütra 1.1.5.
2
Compare Uddyotakara's analysis : vrstimanta etc meghäh gambhiradh-
vänavattve sati bahulabaläkävattve sati aciraprabhävaitve sati unnatimattvät
vrstimanmeghavad iti (p. 47). Jayanta also gave similar analyses. See Nyäya-
manjarï, p. 117-118.
explanation of a similar verse of Kumârila. Rändle was rightly

puzzled over this discrepancy.1 But Rändle did not notice that in
the Nyäya tradition these verses (8-10) of Dignäga were given a
slightly different interpretation. We shall see presently how
Vardhamäna Upädhyäya, the celebrated Naiyäyika of the
fourteenth century, explained these verses in his Nyäyanibandha-
prakäsa.
Vardhamäna came across these verses of Dignäga while he
was commenting on Väcaspati's Tätparya-tikä and Udayana's
Tâtparyaparisuddhi.2 Commenting on verse 8cd, Vardhamäna
remarked: paksäntaram äha / sambandham iti / agnidhümayor in
sesah krtah. This means that Opponent B of Dignäga holds
that our anumeya is the relation between fire and smoke, and
not the relation of fire with the mountain. Since Dignäga
has mentioned the term "sambandha (relation)" in his
verses without specifying the relata, it is quite possible
for someone (who has not read his Svavrtti on these verses) to
interpret this as referring to the relation between fire and smoke.
In fact, Vardhamäna construed the view of the opponent quite
differently from that of Opponent B. Accordingly, I shall call
this view that of Opponent C. In plain words, Opponent C
maintains that what we infer is the relation between fire and
smoke. In verse 10, Dignäga criticized Opponent B as I have
shown above. But here too, Dignäga's arguments are given in
such a language that one might, if one is not acquainted with
the Svavrtti, interpret them as rejecting the new view, i.e., that
of Opponent C. Vardhamäna actually followed this course of
explanation in order to be consistent with his earlier
explanation. He commented on verse lOab as follows: samban-
dham sädhyam düsayati j sambandhe 'pitijdvayam vaksyamänam /
taträdyamähaj sasthitij yadi sambandho 'nwneyah syät tadâgnidh-
ümayoh sambandho 'stlti sasthlsravanam pratijnäväkye syäd ity
arthah. Here, it has been stated without ambiguity that if the
relation was accepted as the anumeya, one would expect the
statement of the inferential thesis to be of the form "There is the
relation of fire and smoke." In other words, the two relata of
1
Fragments from Dignäga, p. 20, footnote 4.
^Nyäyavärttikatätparyaparisuddhi by Udayanäcärya with a Gloss called
Nyâyanibandhaprakcïsa by Vardhamäna Upädhyäya, Bibliotheca Indica,
Calcutta, 1911 ff., p. 748-9.
the said relation would have genitive case-ending. Vardhamäna

continued his remarks on verse 10c: apratijheyah sambandha ity
urthahlkutah dhümavattväd agnir aträstlty ata eväsmadabhimatä-
gnivisistadesasädhyäd arthato 'gnidhümayohsambandhasyävagater
ity arthah. This may be translated as 'The relation is not
expressible in the statement of the inferential thesis. Why?
Because, we come to know of the relation between fire and smoke
by implication from what we grasp as the inferential object, viz.,
the place which is qualified by fire; and this is expressed as
"there is fire here because it possesses smoke." It is evident that
Vardhamäna was always referring to Opponent C, and not
Opponent B.
If we keep the view of Opponent C in mind, we shall, per-
haps, be in a better position to understand Vacaspati's remark
under Dignäga's verse lOd. Apparently, explaining verse lOd of
Dignäga (viz., lina cäsau lingasamgatah"), Väcaspati remarked:
Furthermore, (the relation cannot be our anumeya because) the
inferential mark is apprehended not as a property of the (said)
relation, but as one connected with the place. It seems to me
that Väcaspati interpreted "relation (sambandha)" in Dignâga's
verse in the sense of the relation of fire with smoke. Moreover,
there is a variant of verse 8cd noted in the footnote of the
Kashi edition of Nyäyavärttikatätparyatikä,p. 180 which reads:
"Sambandham kecid icchanti siddhatväl lihgalihginoh" Such a
reading (or, perhaps, a misreading?) suggests that according to
some (whom we have called Opponent C), what we infer is the
relation between the inferential mark, smoke, and the possessor
of such a mark, fire. I suspect that at least some Naiyäyikas
understood Dignâga's verse 8cd in this way, and that Vardha-
mäna followed this Nyäya tradition. Commenting on verse lOd
of Dignäga, Vardhamäna remarked: na cetijasau sambandho
na lihgena dhümena samgatah sambandhah lihgadharmo na
paksadharma ity arthah. What Vardhamäna meant is, perhaps,
that the relation (which, according to him, is the relation of
fire and smoke) is not a property of the inferential mark, and
that, although the said relation is connected with smoke, it is not
connected with the particular smoke which is our inferential
mark, i.e., the body of smoke which is resident in thQpaksa (the
mountain).
To sum up: we have tried to make two distinct points—one is
philosophical, while the other is philological. Inference, in its

crudest form, was regarded as a function of two terms where we
are concerned only with the inferential mark and the property
to be inferred. B-Ut gradual awareness of logical problems led
the Indian philosophers to analyse inference as a function of
three terms. The minor term (i.e., the locus of inference)
gradually rose to prominence. Dignäga's conception of anumeya
is illustrative of this point. The philological ( or, historical) point
can be understood thus. One might ask: Was the variant
"lingalinginoh" inverse 8cd an original variant of the manus-
cript of Dignäga's text? The answer is obviously negative
because the Tibetan tradition does not show this variant. Thus,
I believe that this variant was the result of a misreading by a
later Naiyäyika. We cannot be sure when this misreading
actually took place. But this incident eventually led to the
confusion in respect of the views of Opponents B and C.
One might further ask: Was the Svavrtti of Dignäga on
these verses (PS, Chap. II, 8-11) read by Vardhamäna Upädh-
yäya when he tried to explain these verses in his Nyäyanibandha-
prakäsa ? Or, was he merely depending upon the explanation
suggested by his Nyäya teachers?! believe the second alternative
to be correct. This is rather an unfortunate fact. But, all the
same this is an interesting point, because the Sanskrit text of
Dignäga's Svavrtti might well have been available at the time
when Vardhamäna was writing his commentary. According to
scholars, the last Tibetan translation of the pramänasamuccaya-
vrtti by Kanakavarman (and Dad-pa (hi) ses-rab) was done
some time after the last quarter of the fifteenth century A.D.1
The Sanskrit text of the Svavrtti was definitely available when
Vacaspati Misra (tenth century A.D.) wrote his Nyäyavärttika-
tätparyatikä, but I doubt whether Vacaspati was directly
acquainted with Dignäga's Svavrtti on these verses. Otherwise,
Väcaspati's remark on Dignäga's verse lOd would have been
pointless. It is, nevertheless, extremely odd that a well-reputed
a
The name of the Tibetan collaborator (Dad-pa (hi) ses-rab) of Kanaka-
varman is not mentioned in The Blue Annals which was composed between
1476 A.D. and 1478 A.D. Thus, it seems that Kanakavarman appeared
later than the last quarter of the 15th century. For more on this point see
the Introduction of M. Hattori's book Dignâga, On Preception, H.O.S.,
1968, p. 12-20.
Naiyâyika like Vardhamâna should explain these verses of

Dignäga without reading his Svavrtti on them. It is, however,
conceivable that in the Mithilâ Seminary where Vardhamâna
Upädhyäya received his education, Dignâga's text was not
available.
§ 1.7 : T H E 'TRIPLE-CHARACTER' O F REASON
Inference leads to knowledge when it is based on an adequate

evidence. An evidence is adequate when it not only suggests
that something may be the case, but also excludes the possibility
of the case being otherwise. When an evidence only suggests,
we have a guess-work. When further it removes other possibilities,
we have an inference.
Imagine the following dialogue :
M—Rama was born at midnight on January 31, 1970, at
a town called Dee, (and he was the only child born there
at that moment).
N—Oh ! Then Rama must be a happy man.
M—Why ?
N—I have never seen a man born on that date in that
town to be unhappy.
M—But, have you seen such a man to be happy ?
Is the last question a stupid one ? If not, then we might
have resolved a problem that is faced by any modern inter-
preter of Dignâga's theory of inference based upon what he
called 'triple-character' hetu. For, we may rewrite the above as:
M—a has H.
N—Oh ! Then a has S.
M—Why?
N—I have never seen something having H but not having S.
M—But, have you seen something having H along with S ?
I shall explain that here M is simply insisting, as Dignâga did,
that all the three characters of the trairûpya doctrine are jointly
needed to give the sufficient condition for an adequate hetu.
To explain the Buddhist view of knowledge, we have to mention
two kinds of knowledge or knowing episode. Both are claimed
to be cases of cognitive awareness that arise as episodes. There
is no ownership of such episodes (for there is no person distinct

from the 'aggregate' of such episodes and much else besides)
but each such episode is a discrete member of some awareness-
series or other. Hence, we can say that each awareness-episode
belongs to a particular awareness-series (an awareness-sçries
is only a continuous sequence of distinct awareness-episodes
which are connected causally in some relevant sensç—the
relevant sense being such that the latter is dependent upon
the former for its 'origination'). Hence, only in a figurative
language could we say that an awareness arises in a 'person',
or that a 'person' owns the awareness.
in order to be a knowledge-episode, a cognitive awareness
nust be certain. This element of certainty is shared by both
kinds of knowledge that we shall be talking about. But there
are two ways of ensuring this certainty, the direct way and the
indirect way. "Ensuring certainty" implies removing doubt, i.e.,
all possibilities of error. It is agreed that error creeps in as we
let our mind, our fancy ( imagination = vikalpa) take over.
Hence, the direct way to ensure certainty is to prevent the play
of fancy before it sets in. Prevention is much better than cure.
This is possible only when the pure sensory awareness presents
the datum (we call it the 'percept') untainted by any imaginative
construction (or any play of fancy). This is, therefore, the first
kind of knowledge, according to Dignâga : sensation or sense-
perception. Each such sense-perception perceives also itself.
Therefore, each perceptual event, according to Dignäga, has the
following structure : [percept—perception (percept)—(self-)
perception]. Each percept is a unique particular. Perception is
knowledge because the unique particular shines here in its own
glory, uncoloured by any play of fancy, any operation of the
mind. This is the much-coveted epistemologist's foundation.
For Dignâga, it is not simply a foundation; more importantly,
it is knowledge par excellence.
There is an indirect way of ensuring certainty, according to
Dignâga. This is not a preventive measure as before, but a
curative measure. The play of fancy is allowed to set in, but
possibilities of error are gradually removed. A doubt is
transformed into a certainty, for, the grounds for doubt are all
removed or destroyed. This can happen either through the
employment of an inferential mark called the 'indicator' reason
(lifiga), or through a proper linguistic expression, a word

(sabda). In both cases we deal with a general notion of sign. It
is through the route of a sign that we are led to the object,
finally the particular. Since we are not directly confronted with
the object, we cannot take the direct route. We cannot
prevent the operation of the mind before it sets in. We, in
fact, let our fancy play, and then use it to reach the required
certainty.
How does a sign lead to the knowledge of the object ? It
would be highly uninteresting if we say that there will be a
particular sign for each particular object, so that seeing the
sign, we would know that the object is there. Seeing my friend's
car parked outside, I know that my friend is in. But it is
more interesting and non-trivial when we can talk about a
general sign for a number of particular objects. In the previous
case, we have to see not only the sign, but also, at least once,
both the sign and the object together in order to learn that it
is the sign of that object. In the latter case, we connect a general
sign with a general concept under which several particular
objects fall. In fact, the general aspect of the sign is connected
with the general .aspect of the objects concerned. Seeing, or
obtaining, a particular sign, we consider its general aspect and
from the general aspect of the sign we are led to the general
aspect of the object. Our mind, our 'imaginative' (constructive)
faculty, will take us that far. But if the connection between
the general aspects is the right one in the manner (to be
described below) the general aspect will remove, all rival
possibilities or opportunities for all errors, to lead us to the
certainty that there is a particular object there, an object
that falls under that general concept.
What is a sign? Dignäga said that any object can be the sign
for a second object, provided (1) it has been observed to be with
the second object at least once, and (2) no example of the
'contrary possibility5 has been observed or cited. A contrary
possibility would be a case where an instance of the sign is
present but not the object signified by it. The first condition
could be called suggestion of the possibility, while the second,
exclusion of the contrary possibility. Our knowledge of the sign
will lead to knowledge of the object, provided certainty is
reached through this dual procedure : the possibility is
suggested begetting an uncertain awareness and contrary

possibilities are excluded yielding certainty.
Dignâga used the above theory of sign and object to show
how, apart from sensory perception, inference and linguistic
utterance yield knowledge in the indirect way. A body of smoke
is observed with a body of fire suggesting the possibility of one
being the sign for the other. This means that sighting of a fire
or a body of smoke may lead to a doubt : perhaps, there; J $
also smoke (or fire, as the case may be) there. In such cases,
the two conditions of the triple-conditioned hetu or the
inferential mark (trairüpya) are fulfilled, according to Dignäga,.
and hence, only a dubious awareness can be generated as a
result. For certainty, we need the third condition called vipaksa-
vyävrtü or, in our language, 'exclusion of other possibilities'.
This needs awareness about the absence of any example
('counter-example5)—a case where the sign is present but the
object is not. Now, this also determines which one of the two,
fire or smoke, in the previous example, could be the sign or
the inferential mark or indicator, and which one would be the
object, the inferable object. Examples of fire without smoke are
easily available, but none of smoke without fire. Hence, our
sighting of a body of smoke suggesting the possibility of fire
makes it certain by excluding any contrary possibility, viz., that
of there being smoke somewhere even when no fire is there.
The above way of putting matters, as far as inference is concerned,
would raise problems for logicians; but with Dignäga, the
epistemologist, this would be unproblematic. For the logicians,
inference of fire from smoke would arise from the relation that
we have pinpointed as 'exclusion of the contrary possibilities' (or
'absence of a counter-example'). But, the above way of putting
matters wftuld be psychologizing logic. For logic, it does not
really matter how a person argues or arrives at other inferential
conclusion (by first noticing the suggestion of the possibility
and thereby entertaining a doubt and then arriving at a certainty).
It would be enough to say that A is a logical sign of B, provided
A is such that no case of A is a case of non-5, or, what comes
to the same thing, that every A is B. The only assumption
needed here would be that there are A's and J5's. In this way,
it will be argued, logic can be freed from the fault of the
psychologism.
While I fully approve of the way logic is to be done, or is

being done today without reference to psychological or epistemo-
logical implication. I would like to maintain that the above way of
psychologizing logic is not a totally censured procedure. For, we
are not interested here in the particular way a person infers or
derives his conclusions, but rather in the general 'impersonal*
conditions or factors that give rise to knowledge-episodes and
other awareness-episodes. Besides, each knowledge-episode is
identified by virtue of what is'contained' in it or 'grasped' by it,
and not by virtue of its ownership. And what is contained in
such knowledge is derived from what is expressed or expressible
by a corresponding utterance or linguistic expression. Logic
which seems to avoid psychologism deals, nevertheless, with
sentences, utterances, statements or propositions. To be sure,
utterances are no better than episodes (similar to our knowledge-
episodes), and propositions are no worse than abstract entities.
Conceding in this way the charge of psychologizing logic
{psychologism is not always a crime), we may return to Dignäga,
the epistemologist. One of the traditional problems, that
survived for long in the history of Indian logic, one that has
at the same time been a puzzle for modern researchers in
Indian logic, is the following : Of the so-called triple character
of the hetu, the indicator-reason—1. the indicator-reason is to
be present in the case (or all the cases) under consideration, 2.
it is to be present in a case where the object to be inferred is
present, and 3. it is to be absent from the cases wherever the
object to be inferred is absent—it seems that not all the three
are jointly necessary. Even if 2 is not interpreted as 'it is to be
present in all cases where the object to be inferred is present', it
seems clear that 1 and 3 together would be sufficient to make
the indicator-reason adequate to generate a sound inference.
This apparently falsifies Dignäga's insistence upon the necessity
of 2 along with I and 3 as constituting the required sufficient
condition of the indicator-reason.
I have already said that part of our problem arises as soon as
we switch from epistemology to logic. In epistemology, our
problem is to find how certainty is to be attached to an
awareness-episode, when the said direct route to certainty,
disallowing the mind or the play of fancy to operate, is not
available. It is to be observed that an awareness-episode may
very well be true or fact-corresponding, even when it lacks

the required psychological certainty. For, it lacks certainty
when and only when proper evidence or argument cannot be
given. But this does not affect the fact of its being true. The
epistemological enterprise is to supply the required evidence or
argument, so that we may not attach psychological certainty to
a false awareness. (Because very often we feel sure even of our
false awareness.) Thus, if the proper evidence or argument can
be adduced, we can eliminate false psychological certainty, and
arrive at what we may now call logical certainty. Psychological
certainty is simply subjective, while logical certainty is supported
by an evidence or reason.
In inference, an awareness of A (the indicator-reason) with
regard to a particular case or a set of particular cases (called
paksä) leads to an awareness of B (the inferable object property).
First, we have to grant that the awareness of A with regard to
the particular place or places must be certain, if it has to yield
certainty in our awareness of B with regard to the same place.
The situation is that : certainty of A with regard to the
particular place coupled with some additional information will
yield certainty of B occurring in the same place (paksa). This
additional information comes from our previous knowledge.
An assumption is made, namely, if a rule or pattern emerges
from previous knowledge we may hold it true also for the case
under consideration. Therefore, if previous knowledge yields
that contrary possibilities (possibilities of there being A without
there being B) are absent, we may hold the same to be true in
the case or cases under consideration. In this way, the indicator-
reason A will fulfil the third and the first condition of a proper
sign and thus we may reach the required certainty. But Dignâga
insisted that something more is needed as the additional
information from previous knowledge in order to lead us to
the required certainty : condition 2. In other words, exclusion
of contrary possibilities is not enough, information about an
actual case of co-occurrence of A and B in a place is to be
supplied from previous knowledge in order to ensure the
required certainty. Why? Is it not enough to know that there
cannot be absence of B in the present place, i.e., the case under
consideration, for there is A ? What, in other words, did
74 Logic, Languagç and Reality
Dignäga have in mind when he insisted upon the second

condition as being necessary?
A tentative answer is the following. We find it easier to collect
from previous knowledge some information about a co occur-
rence of A with B than that about the exclusion of the contrary
possibilities. Hence, we can imagine that the "citation of a case
of co-occurrence would bring us nearer to certainty.
That is, a doubt whether there is B or not would be brought
within the range of possibility. Next, the exclusion of contrary
possibilities would assign the required certainty. This answer
seems plausible if we regard Dignäga as being concerned here
only with the psychology of inference, and not with logic. But
I would now argue that this answer is wrong, for Dignäga
cited definite examples where such gradual steps, viz., doubt—
possibility—certainty, have not been marked separately. This
leads us to the consideration of those particular examples
where contrary possibilities are eliminated but it is not possible
to obtain examples of co-occurrence from previous knowledge,
for, A is such that it could be and is present only in the given
places, i.e., the cases under consideration. In other words, A is
a unique mark or character of the paksa, the case (or cases)
under consideration : For example,
PL Sound has impermanence, for it has sound-hood (or
audibility).
It does not seem counter-intuitive to say that sound-hood or
being sound (a noise) cannot be the logical mark or basis for
inferring impermanence. If, however, we reformulate the
argument as given below, as is the practice with most
modern writers of the History of Indian logic, it seems logically
impeccable.
P2. Whatever is a sound or is audible is impermanent. This is
audible (a sound). Ergo, this is impermanent.
I submit that P2 cannot be a proper reformulation of PL For
PI does not want to show, asP2 wrongly assumes, that a particular
case is a case of sound (an audible object) and, therefore, it is
impermanent. Rather it tries to show that all cases of sound
are impermanent, for, they are simply the cases of sound. I shall,
therefore, dismiss P2 as a reformulation of PI, and consider
only PI instead. It should also be noted, in the light of my.

previous comments, that the proposition 'Sound is impermanent'
may very well be true or the awareness that sound is imper-
manent may be fact-corresponding, but Dignäga's claim here
is simply that it lacks the required logical certainty (in the
sense defined earlier).
We can now face the question of justifying this claim. If the
contrary possibility of something being a sound and not
impermanent has been excluded by the information available
from previous knowledge, (i.e., by the available information),
why can't we decide that sound (all cases of sound) is
impermanent ? Here we reach the crux of the matter. We have
to remember that all cases of sound are not (at least, in princi-
ple) part of the available information. They lie outside the
domain which is constituted by available information. We are
only certain of one more thing : sounds are sounds, or have
sound-hood, (or have audibility). This is an Ö priori certainty.
But this does not guarantee that cases (instances) of sound are
the kind of things of which impermanence or permanence is
predicable. It could be that sounds are neither. Such a guarantee
is available only if we could cite a case, independently of the
present situation, where both the indicator-reason and the inferable
object exist together, and show that the present case is similar
to such a case. This is, therefore, part of the justification for
Dignäga for not being totally satisfied with the exclusion of
contrary possibilities (vipaksäsattva), and thereby insisting
upon the citation of a similar case or a case in point (sapak-
sasattva = sädhannyadrstänta). PI is, accordingly, declared as
inconclusive or uncertain. Hence, it is not a deductively valid
argument—and as P2 is). If is being declared as uncertain,
because it is quite a different sort of argument whose certainty
is not determinable.
The above discussion raises many fundamental philosophical
and logical issues—issues connected with the meaning of negation,
logical negation and contraposition, contradictories and contra-
ries, possibility and certainty. While I do not wish to enter into
such issues in the present context, I would claim that all these
issues are relevant here. Briefly, I would note a couple of points.
First, the above justification assumes that lack of togetherness
of A with non-i? does not necessarily imply togetherness of A
with 5. As Richard Hayes1 has rightly stated, while 'Every A

is 5 ' may presuppose (as it does in the interpretation of the
Aristotlian syllogistic) that there are A's, cNo A is non-5' may
not, under this theory, presuppose that there is at least one
A which is B also. For, as I have already argued, all A's may
be such things with regard to which the question of their being
either 5 or non-5 does not arise. Hence, 'an A is neither 5 nor
non-5' is a further possibility that is not eliminated by the
exclusion of the contrary possibilities. And since such a further
possibility is not eliminated, the required certainty that the case
under consideration is 5 is not reached. Citation of a 'positive5
example with A and 5 together eliminates the said third
possibility, and thereby leads us to the required certainty.
From what has been stated so far, it follows that 'not non-5'
is not always equivalent to 'B\ for, sometimes it could mean
something with regard to which the question of being either B
or non-5 does not arise. Further, 5 and non-5 are not con-
tradictories', in this way of looking at things, since they can
only be contraries in the sense that they both may fail to apply
to some cases (which are neither 5 nor non-5).
x
Richard Hayes : "An Interpretation of Anyâpoha in Dignâga's general
Theory of Inference," Buddhist Logic and Epistemology, éd. B. K. Matilaî
and R. D. Evans, 1986, Reidel.
CHAPTER TWO
PROBLEMS OF PHILOSOPHICAL LOGIC IN

NAVYA-NYAYA
§ 2.1: EMPTY TERMS
Too often Indian philosophy has been described as primarily

metaphysical or theological in character and hence opposed to
rational discussion. Navya-Nyäya, as well as the tradition from
which it stems, is a clear proof that there has existed in India
for a considerable period of time a rigorous system of thought
less concerned with "the eternal verities" than with human
knowledge open to verification and rational procedures. This
system well deserves a prominent place in any history of Indian
philosophy.
Researches in Navya-Nyäya usually take either of two forms.
The first form, mainly historical and philological, is represented
by the recent works of Erich Frauwallner of Vienna and his
students.1 The second form is usually characterized by systematic
translations of certain texts and critical analysis of some theories
and techniques. The work of Daniel H. H. Ingalls exemplifies
this second kind of approach to Navya-Nyäya, notwithstanding
the fact that he added in Section I of his book the results of his
1
Frauwallner, E. "Prabhäkara Upädhyäya," Wiener Zeitschrift für die
Kunde Süd-und Ost-asiens 1965, p. 198ff.
"Die Lehre von der zusatzlichen Bestimmung in Gangesas Tattvacintä-
manih,'" Vienna Acad. 266.2, 1970.
«Raghunätha Siromani" WZKSO, 10, 1970, pp. 86-207; 11, pp. 140-208;
14, pp. 161-208.
historical research.1 The book of Karl Potter2 and the recently

published book of C. Geokoop3 belong exclusively to the second
group. So does a book of Mohanty.4 Mohanty clearly states in
his Preface that his approach to Gangesa's text has been "that
of a critical philosopher." He also claims that this is "the best
way of dealing with so outstanding a critical philosopher as
Gangesa." Although one might wish to qualify this statement, it
must nevertheless, be admitted that anybody who wants to
explain and systematically translate an Indian philosophical text
in a European language will, knowingly or unknowingly, be
using the method of comparative philosophy. In other words,
he cannot but compare and contrast the Indian philosophical
concepts with those of Western philosophy, whether or not he
consciously does so. Thus, I wish to repeat what I have said
elsewhere; "comparative philosophy" in this minimum sense
should no longer be treated as a derogatory phrase. Rather,
when pursued with sufficient care and an informed conviction,
this method may lead to some useful results.
In The Navya-Nyäya Doctrine of Negation (Harvard, 1968),
Part I, I dealt with the basic concepts of Navya-Nyäya as well
as with some of its theories. I wish to present here the results
of some further research.
I shall first discuss the problem related to the empty terms,
i.e., expressions which are meaningful but do not refer to any-
thing real. This problem has gained increasing attention from
modern logicians. It would, therefore, be interesting to see
how Navya-Naiyäyikas, i.e., the philosophers of the twelfth-
thirteenth century India, tackled the problem. The first two
sections in this chapter will concentrate on the different aspects
of the problem, and these aspects will reappear in some places
of the remaining sections.
One of the important arguments offered by the Nyäya-Vaise-
sika Theism to prove their thesis, e.g., "God is," can be rephrased
1
Ingalls, D.H.H. Materials for the Study of Navya-Nyäya Logic, Cambridge,
Mass., 1951.
2
Potter, K.H. Raghunätha's Padärthatattvanirüpana, Cambridge, Mass.,
1957.
3
Geokoop, C. The Logic of Invariable Concomitance in the Tattvacintämani,
Dordrecht, 1967.
4
Mohanty, J.N. Gangesa's Theory of Truth, Santiniketan, 1966.
Problems of Philosophical Logic in Navya-Nyäya 79
as follows : God exists, because to deny His existence is but

to affirm Him as God. The argument is significant. The general
principle from which it derives its strength is this : we cannot
negate a non-entity whatsoever, and hence if we negate
God significantly, we are bound to accept His 'entitative
existence.' To put it in the technical language of the Naiyâyika,
in all negations that we can speak of, the negatum (pratiyogiri)
cannot be a nonentity (alïkà). The principle contains the seed of
an interesting problem of ontology which has also been reflected
in the old Platonic riddle of non-being.
The above argument, innocent as it looks, unlocks the door
to some great controversies leading to grave consequences. Let
us, for the present purpose, remove "God" from the subject
position of the above statement, (as we may be reluctant to
offend the theist) and replace it with some such words like the
hare's horn, the winged horse, etc. Regarding the non-existence
of these entities both the theist and the atheist are unanimous.
The principle that negation, if it is not to be nonsensical, (and
it must not be nonsensical) must have as its negatum an entity
and not a fiction, will logically demand the 'entitative existence of
the hare's horn' or 'the winged horse' also. For, otherwise it
would be nonsensical to say 'there is no hare's horn' and the
like. The problem becomes all the more entangled as we further
replace the subject by some such terms like 'the round-square'.
How the Nyäya-reaiist would seek to solve these difficulties
and avoid the contradictions involved, we shall see presently.
Let us, at the outset, try to understand the main problem clearly
from different points of view.
Some philosophers, being hard pressed by the above contradic-
tions, may concede to a peculiar but precarious position. They
are, however, prepared to acknowledge some form of existence,
i.e., subsistence of these entities, namely, the hare's horn, the
winged horse, etc. Entities like the hare's horn are, they will say,
unactualized possibles. When we negate them, we do nothing
more than deny their actuality, and 'actuality', according to
them, is but apredicable attribute like redness or roundness and
may be equated to what we call 'existence.' Thus negation or
rather the denial of existence to hare's horn, etc., is as meaningful
as they are, i.e., entities belonging to the "world of possibles."
But if these philosophers are pressed with such questions,
e.g., "What does your possibility consist in ?" no direct and

clearcUt answer perhaps will be forthcoming. If, again, instead
of the rabbit's horn, etc., such things as 'the round square,' are
chosen as examples, would they then, on the same argument, be
thrust into that world of possibles ? Or, should another weird
universe of impossibles be conceived to include these things, so
as not to destroy the usual sense of the word "possible" ? The
extremists, however, would like to assign 'subsistence' to such self-
contradictory things as the round-square, and would not
object to such over-population of their weird universes of
the possible—'possible', of course, to be understood in a special
sense of their own.
The opponents to the above view, however, will come forward
with a much sharper weapon in their hand to chop out the weird
entities like the winged horse, the rabbit's horn, the round
square, and a thousand others which are likewise advocated
by force of the above argument, i.e., in order to make the
negation of them or statements about them meaningful. This
weapon is traditionally named the "Ockam's razor" : entities
should not be multiplied beyoad necessity. If we can reasonably
explain and maintain thereby the meaningfulness of the state-
ments like "There is no hare's horn," "The author of Waverly
was a poet," and "the. round square table is pink," we need not
postulate such a world of strange entities as has been claimed
above. Moreover, such postulates also go against "that feeling
for reality which ought to be preserved even in the mostabstract
studies," (Russell). We shall, therefore, see how the above
statements can be meaningfully analysed and explained with the
least oddness about them.
The well-known theory of Description, which is due to Russell
has been fruitfully utilized to show how we can meaningfully use
the 'seeming names' without supposing that there may be entities
allegedly named. Thus, the statement "The author of Waverly
was a poet" is, according to this theory, paraphrased as
"something wrote Waverly and was a poet, and nothing
else wrote Waverly." Following this principle, it becomes
easy to meaningfully negate fictitious things like Pegasus, the
present king of France, the round square table. What seemed
to be 'names' purporting to apply to the one and only one
object now, on Russellean analysis, turn out to be descriptive
phrasés of the form the 'so-and-so'—i.e., the thing that Pegasi-

zes, the thing that is round and square and also a table, etc.
Moreover, the grammatical subject of our previous statement,
is not the logical subject in Russellean translation where the
subject position is occupied by what is known as a 'bound vari-
able,' 'variable of quantification,' e.g., words like "something,"
"nothing," "everything." Thus, it is evident that in affirming or
denying any predicate including existence of these fictitious
things, we need no longer presuppose, to maintain the meaningful-
ness of our statements, any such queer entities. The burden of
objective reference is thrown upon the shoulder of bound
variables like "something" etc. They do not purport to be names
at all, but refer to entities generally.
Russell's theory of Description or better, the theory of singular *
description, directly applies to the singular terms, and also, with
a little ingenuity, to such proper names as Pegasus, Cerberus,
etc. Quine, however, has taken the further step to eliminate
'names' altogether from logical sentences, and push them into
the domain of description taking recourse to trivial analysis of
such "names". Following the principle of analysis contained in
the above argument, statements about the non-existence of so-
called self-contradictory terms can also be translated into the
logical language containing variables, so as to eliminate the alleged
contradiction. "Round square does not exist, for example, may
be paraphrased as "each thing that is round, is not square".
"The hare's horn does not exist" may be written as "each
thing that is a horn does not belong to a hare". This method
bears a striking similarity with the Naiyäyika's interpretation of
the negation of the so-called self-contradictory terms. This will
concern us in the remaining paragraphs.
The problem before the Indian logicians is not simply to
reasonably interpret the above sentences, but also to give a satis-
factory analysis of the cognitions (pratyaya) which take notice
of the non-existence or absence of such self-contradictory objects.
Some logicians, (e.g., the Buddhists), however, advocate that the
asat or iictitious objects do appear in certain modes of cognition,
and thus, they tackle the above problem in a different manner.
The Naiyàyika, on the other hand, would never accept such a
position. He rather holds that even erroneous cognitions do not
reveal any object which is totally fictitious, but are contained
82 Logic* Language and Reality
in the wrong attributions of the adjectives, or qualifier. More-

over, according to him, both the negatum (pratiyogin) and the
locus of negation (adhikarand) are, as we have already noted at
the outset, real objects. This principle has been explicitly stated
in Udayana's Nyäyakusumänjali, Chap. Ill, verse 2. Thus,
ordinarily one cannot significantly negate such things as "the
hare's horn" or "the round square."
How then, should we reject the claim of such terms to entita-
tive existence ? The prima facie answer to such question may
easily be put thus : Because they are nowhere found as such.
But the Naiyäyika cannot demolish the world of such strange
entities simply on the ground that they are not felt or cognized,
or perceived anywhere. Because, he himself accepts many im-
perceptible entities like atoms, etc. Even his theistic position
would be at stake under such circumstances. Hence, the oppo-
nent may drive him to the acceptance of the theory that non-
existence of an object is established jointly on two grounds, e.g.,
(i) non-perception of it, and(/ï) its being amenable to perception
(which is technically known as Yogyatä). Yogyatä has been
defined as the collocation of all the required causal conditions
leading to the perception or cognition of an object minus the
object itself in question, along with any other thing concomitant
with it. The Naiyäyika, however, cannot accept the view that
things like the hare's horn, winged horse are yogya i.e., possible
or amenable to perception or cognition. Some, however, may
suggest that we may mistake, in insufficient light, the long ears
of a hare standing erect for its horns and we may visit in our
dreamland a winged horse, and on such evidence hare's horn
may be held as yogya, i.e.. amenable to perception. Thus,
they will take the easy way out, that of rejecting the "entity-
hood" of these objects on the ground that they are not
perceived, though amenable to perception i.e., yogya. But the
Naiyäyika will reject the suggestion outright, since it is self-
destructive in purpose, and also lends support, by implication,
to the opponent's view that errors reveal fictitious objects.
If these objects are held as yogya i.e., perceptible even in
error as stated above, then, no argument can be given to show
them to be fictitious or non-existent, according to the Naiyäyika.
Moreover, the supposed argument that they do not exist because
they cannot be seen will also fall flat. If it is said, on the other
hand, that they are not actually perceived in any case, then we
are bound to give up the idea that they are yogya.
Thus the Naiyäyika chooses to hold such terms e.g., the 'hare's
horn' etc., as impossible objects just in the sense we take 'the
round square', 'circular triangle' to be impossible. Just as a
round thing cannot by definition be square, a hare also by
definition, cannot grow horn, and a horse cannot be winged.
(See also § 5.5, the last section.) If, however, the long, erect ears
of the hare are mistaken as horns under any condition of error,
then, of course, the erring person, who knows also what 'hare'
means will never identify the seen animal with u hare. If an
animal resembling a hare in every respect is found to grow horns
in addition, then it will make a new species of animal, and not
a hare. Thus, in the Naiyäyika's view both the objects, hare's
horn and round square, sail in the same boat.
Suppose, theNaiyäyika is asked to explain the statement "There
is no hare's horn". First, he would say that here the subject of
negation, i.e., thenegatum, is not at all the hare's horn. What is
negated here is simply the horn, and the locus of such negation
is a hare, Or, from the other viewpoint, the occurrence in a hare
i.e., sasavrttitva) is being negated and the locus of such negation
is a horn. In plain language, the said sentence is to be read in
this view as "horns do not occur in a hare". Thus, "There is no
round-square" is to be read as "the round thing is not square".
Another way of meeting the problem has been suggested by an
indigenous theorist who is generally known as Sondada or Son-
dala. He tries to show that the negation of a self-contradictory term
as an entity is actually felt to occur in any loci. When an entity,
say a pot, is negated, potness resident in a pot is ordinarily
held to be the limiting property (avacchedaka) of the abstract
property, absenteehood or negatumhood (pratiyogita) tagged to
the negatum-entity pot. This is shortly spoken as the negation
of a pot limited or qualified by potness. Sondala holds that an
object limited or qualified by a property actually non-resident
in it may also be negated, and such negation will occur every-
where, because nowhere can we realize such a self-contradictory
object—an object, (say a pot or a horn) qualified by a non-
resident property (say, cotness or occurrence-in-hare). Thus,
Sondala recognizes such statements like 'the hare's horn is
absent in a cow', 'the pot qualified by cotness is absent in the
hill', 'the cupola in Berkeley College is not round square'as

valid and meaningful. According to Sondala negation of one
entity as-qualified-by-a-non-resident-property is real and uni-
versally present (Jc&valänvayin) (see Section § 2.4). So, negation of
a horn as limited by a property e.g., relatedness to-a-haremay be
experienced in any locus like acow or a horse, just as pot-limited-
by-cotness is known to be absent in a hill or in any other locus.
Gangesa rejects the view of Sondala on the ground that a
property that is non-resident in the negated entity, the absentee*
cannot delimit the absenteehood tagged to it. To explain : our
cognition of a negation of any entity is by necessity a visista-
vaisi$tyävagähi type of cognition i.e., a cognition where a quali-
fied entity is also cognized to qualify another. Such a type
of cognition is said to be necessarily conditioned by a
prior cognition with the first adjective, i.e., the adjective of the
main adjective in the resulting cognition, as the chief qualifier
or prakära. Thus, the cognition of the negation or absence of a
pot while revealing the pot-qualified-by-a-property potness as the
absentee or pratiyogin reveals also the property potness as the
limitor or avacchedaka of the abstract, absenteehood. Besides
this, a property cannot be revealed independently or in any other
way as the limitor of the said abstract. It would be a strained
distortion of the fact if we claim that the cognition of the
negation or absence of a pot may also reveal as the absentee the
pot qualified by a non-resident property cotness. But if for any
perverse reason it is claimed, in error at least, that such strangely
qualified entities are revealed as the absentee and that we do
experience an absence of the hare's horn in cow, then Gangesa
will turn his back, and we may allow Sondala to hold that the
absence of such objects is resident everywhere. (See below § 2.4).
The Naiyâyika, however, is constrained to admit such negation
where the absenteehood is de-limited or, conditioned by a 'non-
compatible' relation (vyadhikarana-sambandhäuacchinna-pratiyogi-
täkäväbha) as "real". Just as a pot that is conjoined to (satnyu-
kta) the ground, does not inhere there, (and so we can say that
the ground is the locus of the absence of pot through the relation
of inherence), so also seeing that though colour inheres in
substances, it occurs nowhere through the relation of conjunc-
tion or saipyoga (which may be called, accordingly, a relation
'non-compatible' with colour), we can also submit that the
negation or absence of colour through conjunction is real and it

is universally present (kevalänvayin). But this opens a different
issue altogether which I shall discuss later. To sum up our
speculations about the Naiyäyika's interpretation of empty terms,
we may submit the following : First, that they as such refer to
any possible entity is unanimously rejected. Secondly, so far *as
negations of such terms (e.g., self-contradictory compounds)
are concerned, Sondala claims that in such negations,
one component of the said compound is held as negatum,
while the other component is revealed as the limitor of the
negatumhood-tagged-to-the negatum. But Gangesa explains
that here one of the components holds actually the position
of negatum, while the other occupies the position of the locus
of negation in question. Thus, from what has been said it
is evident that in the analysis of such terms, the Naiyâyika
largely anticipates, in his own way of course, what a Russellean
would try to show with the help of the theory of Description.
This is, however, a very general statement which must be quali-
fied when further 'in-depth' study of both the traditions is
undertaken. I move to the next section to examine the long-
standing Nyäya-Buddhist controversy over the acceptability of
empty terms in a discourse and trace its historical origin in the
tradition.
§ 2.2 : REFERENCE AND EXISTENCE :
Non-Referring Expressions in Language

Kumärila once made a very significant remark : Word
or speech can generate cognition even of entities which
are totally non-existent.1 The fact that there are 'meaningful'
and grammatically acceptable expressions in language which
purport to refer to or to denote some entity or entities
but which actually do not refer to anything in our world of
experience, has very often proved a puzzle for philosophers
and logicians. It is somewhat paradoxical to say that we refer
to non-existent entities by such expressions as "the rabbit's
horn", "the sky-flower" or "the son of a barren woman". All
that we have here is a class of 'meaningful3 expressions which
1
Kumärila, see section of Codanä-sütra, verse 6.
share the same substantival structure in common and possess

the grammatical property of a proper name in the sense that
they can be successfully used in a context where a proper name
might have been used. These expressions have been called
* vacuous' or 'empty' terms. A problem arises when such a
term occupies the subject position in a sentence; a problem that
is both logical and epistemological.
The understanding of a substantival expression or phrase
does not imply that it has a reference; in other words, under-
standing of its meaning precedes the knowledge of whether
or not the expression actually refer to any real entity. That is
why, we are justified in calling such expressions 'meaningful',
although they fail to refer to anything.
An unusual strain of realism pervades our ordinary language
in such a way that whenever we try to refer to or express an
imaginary or fictitious object we feel constrained to admit
some kind of 'relative' reality for these fictions. We may not be
happy about this situation. One should point out that we
would otherwise face a logical problem. To put it simply, it
would be difficult, for example, to negate (logically) a state-
ment whose subject is a fictitious object. The actual formu-
lation of the antinomies of a two-valued logic will mainly
concern us here. The awareness of these antinomies can be
clearly discovered, as we shall see, in the writings of the
Naiyäyikas and the Buddhist logicians of the tenth-eleventh
centuries.
An initial note on the source material used in my discussion
is in order. The Nyâya position as expounded here is mainly
that pof Udayana. The Buddhist position is what Udayana
regarded as the view of his Buddhist opponent. The idea of
momentariness of all entities apparently belongs to the Sauträn-
tika school of Buddhism. Unfortunately, no extant philo-
sophical text of the Sauträntika school where this problem
is discussed is available to us. But a good account of the
thesis of momentariness has been presented by Dharma-
kïrti and his followers. Udayana becomes involved in the
argument concerning the status of empty terms while he
repudiates the logical proofs of the thesis of momentariness.
In the first chapter of his Ätmatattvaviveka (from which
I have largely drawn my material), Udayana is mainly
concerned with the Buddhist views as set forth by Jnäna-

srimitra in his Nibandhävali.
The Riddle of Non-being
The Nyäya school claims that a sentence whose subject term
does not refer to anything stands in need of some philosophic
paraphrasing. A sentence is a representation of some cognitive
event. A cognitive event, i.e., a judgmental one, usually attributes
a property, rather a Qualifier', to a subject or a qualificand.
And this attributable property will be called a qualifier. Now,
a cognitive judgment fails if it lacks a subject to which it can
attribute some property. Hence, a sentence which apparently has
a non-referring expression as its grammatical subject undergoes
a philosophical paraphrasing in the Nyâya system so that it can
properly represent some judgmental (or qualificative) cognitive
event. A judgmental cognitive event may be erroneous where
the representing sentence will be regarded as false. If a
cognitive judgment is right, the corresponding sentence
will be true. Thus, knowledge and error are the epistemic
counterparts of the truth and falsity of the sentences
that express the corresponding cognitive events. Proceed-
ing along this line, the Nyäya realism almost pre-judged
the issue and tried to show that a sentence with a non-referring
expression as its subject should be traced back to some kind of
erroneous cognitive event and should be explained accordingly.
In other words, Nyäya ruled that these sentences were demons-
trably false. Bertrand Russell seems to have tried to analyse such
sentences in much the same way. It has been shown by him that
these sentences can be paraphrased into such logical forms as
will make them patently false.1
Apart from this apparent similarity, the philosophic motivation
of Russell was, perhaps, not very different from that of the
Nyâya school. Among other things, Russell was worried about
ontology. Thus, he wrote :
It is argued, e.g., by Meinong, that we can speak about 'the
golden mountain', 'the round square', and so on; we can
make true propositions of which these are subjects; hence,
they must have some kind of logical being, since otherwise the
1
B. Russell, "On Denoting/' Mind, 14 (1905), pp. 479-93.
propositions in which they occur would be meaningless. In

such theories, it seems to me, there is a failure of that feeling
for reality which ought to be preserved even in the most
abstract studies.1
Whether Meinong did actually postulate such a theory or not
can be determined from his doctrine of the independence of
what he calls Sosein from Sein.2 Roughly speaking, Meinong
contended, arguing for a very broad sense of the term
'object,' that the object's having some property or charac-
teristic is not affected by its being existent or non-exis-
tent. Thus, although the round square does not exist, be-
cause it cannot exist, it is possible to make true assertions about
it or to predicate some property of it. In other words, we might
truly say, "The round square is round and square." Even at the
risk of a paradox, Meinong adds, one can very well say, "There
are objects of which it is true to say that there are no such
objects". To the criticism that the law of contradiction would be
violated if the sentence "The round square is round and square"
is held to be true, Meinong replied by saying that the law of
contradiction holds only in the case of what exists, or is real,
and is thus not violated by the sentence in question.
Even if we leave Meinong aside, we can conceive, following
Quine3, of some fictitious philosopher like McX or Wyman,
who would be willing to countenance at least subsistence, if
not existence, to such entities as "the golden mountain". In
other words, these philosophers would posit a world of
'unactualized possibles5, but would resort to the doctrine of
meanjnglessness when faced with such self-contradictory ex-
pressions as 'the round square' or 'the son of a barren woman.'
Part of the difficulty connected with sentences with non-
referring expressions as their subject terms can be transformed
into a logical riddle. This riddle has a long history in the West
and it is usually nick*named Plato's beard. In India the story
is more or less the same. The riddle of cnon-being' in some
form or other stayed alive in the controversies between the
Buddhists and the Nyâya philosophers down the centuries.
X
2
B.Russell, Introduction to Mathematical Philosophy, London (1919), p. 169.
A. Meinong, pp. 76-117, 'Theory of Objects" in R. Chisholm's Realism
3
and the Background of Phenomenology (1961).
W.V. Quine,From a Logical Point of View, Cambridge, Mass (1961),
Essay 1.
The Status of Example in Indian Logic

I shall now explain the Indian version of the riddle of non-
being, following Udayana.1 Some acquaintance with the
development of Indian logic during the time of Jnänasrimitra
and Udayana is necessary for understanding the Indian version.
The Buddhist believes that the demonstration of the proof of
his thesis (viz., everything that exists is momentary) is possible.
Thus, he tries to formulate a philosophic argument in the
standard "logical" form which is mainly non-deductive. Un-
fortunately, modern interpreters have sometimes tried to reduce
the argument into a deductive inference. (See our comments in
previous § 1*5.)
A few remarks about the nature of Indian logic may be in
order here. Logic developed in India out of two slightly
distinct traditions : (1) väda tradition, i.e., tradition of debate
which was concerned with dialectical tricks, eristic arguments
and sophistry, (2) pramäna tradition, which was concerned
with the criteria of empirical knowledge, the accredited source
of knowledge. On account of this genesis, Indian logic imbibed
an epistemological character which was never abandoned
throughout the successive periods of history.
The model of reasoning with which the Indian logicians were
chiefly concerned was not purely deductive. Modern interpreters
of Indian logic have seldom realized this point (for such over-
sight, see Rändle, 1930). As a result, there have been some
confusing and futile attempts to reduce the arguments studied
by the Indian logicians to Aristotelian syllogistic model (See
§ 1.5). It should be noted that inspite of the neatness, elegance
and precision of a deductive system like that of Aristotle, it is
undeniable that a good deal of our actual reasonings may not
follow the deductive pattern. The reasoning of an experi-
mental scientist, a historian or an ordinary man trying to
ascertain the truth of a particular matter, is a reasoning
from what we may call 'evidence5 to what we can call 'con-
clusion*. Even most of our philosophical arguments, where we
try to depend more or less upon empirical evidence, belong to
this type of inference. A purely deductive model is not always
appropriate to this kind of reasoning.
Udayana, Ätmatattvaviveka, pp. 59-89.
In a deductive reasoning, the so-called premises entail (in

some acceptable sense of this term) the conclusion in such a
way that if we accept the premise we cannot afford to avoid
the conclusion without the risk of contradicting ourselves. In a
non-deductive argument, the conclusion is not entailed in the
same sense by the premises. We should better say that our
evidence or 'premises' here justify or support the conclusion.
Evidence may be good or bad, and the corresponding argu-
ment may be sound or unsound. A good evidence strengthens the
thesis and makes it acceptable to us. But a bad evidence either
weaken the thesis (renders it doubtful) or falsifies it. Thus, it
seems to be better to talk in terms of soundness and unsound-
ness of this type of arguments (instead of talking in terms
of their validity or invalidity).
The general form of the arguments studied by the Indian
logicians is ; A is B because of C. The middle term or the
'reason' hetu Ccan be either adequate or inadequate (instead of
being strictly valid or invalid). An adequate middle term or
'reason' will establish the conclusion and the argument will be
sound. If the middle term is not adequate, the conclusion will
not be established and the argument will be unsound. Ordinary
conversation and philosophical treatises provide millions of
examples of this kind of argument.
In a non-deductive inference, e.g., "It must have rained
because the ground is wet", the second part is believed to be
the adequate ground for. accepting the first. Indian logicians
tried to frame the rules of logic from the paradigm cases of sound
inference of this kind. But the neatness of a deductive system
can easily capture our mind. We may thus be tempted to introduce
an additional premise so that these arguments will be deductively
valid. We can resort to the theory of 'suppressed premises'
and decide that we are dealing in fact with deductive
arguments in all these cases. But this seems at best to be a
distortion, and at worst the demolition of the original
structure of the actual argument. That we are inclined to
reduce these arguments to deductive arguments proves
that we are somehow assured of the soundness of these
arguments much in the same way as we feel secure about a
deductively valid conclusion. But this reduction, even if it is
sometimes justified, throws very little light upon the original
problem, e.g., how can we reasonably draw, as we obviously

do, conclusions from the so-called premises which do not
strictly entail them? We merely direct our attention to a slightly
different and a bit narrower question : How do we establish
general propositions such as these 'suppressed' premises?
Most of these premises are not admittedly necessary propositions,
i.e., analytic judgments. They are in some sense 'synthetic'
propositions, representing general beliefs which come from
common experience.1 The early history of Indian logic seems to
have been a search for an adequate model which will explain
empirical inferences of the kind mentioned above.
That it is often misleading to introduce a 'suppressed' general
premise in some cases can be easily shown by applying the same
method to a dubious case, i.e., an unsound non-deductive argu-
ment : speech sound is permanent because it is audible (5th type
in Dignäga's table of'reasons' see § 1.7, 2.4; cf., Hetucakrada-
maru). Should we construe this case as having a general premise
"Everything that is audible is permanent", and thus treat the
argument as deductively valid with a false conclusion from a
false premise ? Or, should we construe our general premise as
"Nothing that is audible is permanent", and thus make the
argument deductively invalid ? The truth is that in either case
we are turning our attention from the actual structure of the
argument to our experience and general belief. In fact, we
pre-judge the case, finding that the conclusion is definitely false.
The best thing would, of course, be to do neither. The Indian
logician says that you cannot state your general premise or the
universal proposition unless you can cite a supporting 'example'
(drstänta). A supporting 'example' can be of two types : an
'agreeing example' (sädharmya-drstänta) and a 'disagreeing
example' {vaidharmya-drstänta). An 'agreeing' example is a case
where both the 'reason' and the 'inferable property' Qietn and
sädhyä) are present together. This will at least show that neither
the 'reason' nor the 'inferable property' can be fictitious. A
'disagreeing' example is a case where both the reason and the
inferable property are absent. Suppose, I wish to infer that there is
fire on that hill which I see. An 'agreeing example' or a homologue
in this case will be a kitchen where fire is present along
x
See also my review of C. Goekoop in Journal of American Oriental
Society, 1970. For a similar point, see Strawson, An Introduction to
Logical Theory, London (1952), pp. 234-36.
with my 'reason', i.e., Smoke, A'disagreeing example', orahetero-

logue, would beany room where neither smoke nor fire is present.
With this prelude we can proceed to the heart of the contro-
versy between the Buddhist and the Nyäya logicians. The thesis of
momentariness which the Buddhist wants to prove is a universal
proposition :
Whatever exists is momentary.
Its contrapositive version is :
Whatever is non-momentary does not exist
The 'reason' in this case is 'existence', and the 'inferable

property' is 'momentariness'. Now, to prove the invariable con-
nection between the 'reason' and the 'inferable property', the
Buddhist logician must cite, in the first place, some 'agreeing'
example where the two properties (the 'reason' and the 'inferable
property') and their connection are instantiated. But the 'agree-
ing' example will not be enough unless one can support it by the
citation of what is called a 'disagreeing' example. A 'disagreeing
example' is something that instantiates the contraposed version
of the main thesis. According to the acceptable form of demons-
tration, the Buddhist, in order to prove his thesis, must cite not
only an 'agreeing example' but also a 'disagreeing example' for
making his argument sound and more convincing.
Although the proof here consists in the citation of a 'disagree-
ing example', the conclusion should not be regarded in any way
unsound or uncertain. The whole point of citing a 'disagreeing
example' is to show the actual absence of any counter-example.
If a counter-example can be found where the 'reason' is present
but not the inferable property, then the supposed thesis is
immediately falsified. Thus, the soundness of the conclusion
depends upon this absence of any counter-example, or our failure
to discover any counter-example.
Now, a difficulty arises when we try to find a 'disagreeing
example' for the above thesis (viz., an instantiation of the con-
traposed version given above). To cite an example of unreality
would be in some sense to contradict one's own position, because
it would be like saying that there is an example (or, an entity,
if we like) which, according to us, is unreal, i.e., non-existent or
a non-example. An 'example', or what is called a drstänta in
Indian logic, must be a well-established case admitted by both

sides in a philosophic debate, the opponent and the proponent.
It is implied here that there is admittedly an accredited source
or 'means of knowledge' (pramänä) with which we can establish
the example. Thus, an example can never be an unreal entity.
This will finally lead, according to Nyäya, to the destruction of
the Buddhist claim that he.can prove 'logically' the momentari-
ness of everything.
The Nyäya-Buddhist Controversy

It should be made clear at this point that neither the Nyäya nor
the Buddhist wishes to countenance the world of strange entities
like the golden mountain or the rabbit's horn. But the difference
lies in their method of approach to the problem.
Confronted with the Nyäya criticism (noted in the last section),
the Buddhist does not so easily give up his hope of proving his
thesis of momentariness. He tries to point out that the position
of the opponent, i.e., the Nyäya position, also involves a self-
contradiction. By saying that an unreal entity cannot be used as
an example, or, cannot be used as the subject of a proposition,
the Nyäya actually mentions 'an unreal entity' in his speech-act.
And this can be shown to imply a proposition whose subject
would be an unreal entity or a non-existent fiction. Thus, for
example, Nyäya would have to argue: the rabbit's horn is a non-
example, because it does not exist. And this will contradict
the original Nyäya position that fictitious entity cannot be the
subject of a proposition. Thus, the Buddhist claims to have
proved his thesis of momentariness even at the risk of an
implicit contradiction, because otherwise the rejection or
criticism of his argument would lead the opponent into a patent
self-contradiction.
Udayana remarks here as follows.1 The Buddhist wants to
avoid a patent self-contradiction (like saying "the rabbit's horn
cannot be the subject of any proposition, because it does
not exist") and thus allows that certain speech-acts, and con-
sequently certain sentences, about fictitious entities like the
rabbit's horn are quite in order. The Buddhist> in fact, does not
, Ätmatattvaviveka, pp. 64-65.

want to accept the Nyäya position that the subject term of a

sentence must refer to something actual (or real), and if it does
not, the whole sentence stands in need of some philosophic para-
phrasing. Thus, for the Buddhist, "The rabbit's horn is sharp"
is a normal sentence which we may use in our discourse for
various purposes. One of such uses is made when we cite an
example of a non-entity, viz.,
"The rabbit's horn is non-momentary and also non-existent."
Nyäya, on the other hand, wants to exclude from logical
discourses any sentence which will ascribe some property (posi-
tive or negative) to a fictitious entity. Vacaspati remarks that we
can neither affirm nor deny anything of the fictitious entity, the
rabbit's horn. 1 Thus, Nyäya apparently agrees to settle for a
'superficial' self-contradiction because, in formulating the princi-
ple that nothing can be affirmed or denied of a fictitious entity
like the rabbit's horn, Nyäya, in fact, violates the same principle.
Nyäya feels that this 'superficial' self-contradiction is less
objectionable, because it can somehow be explained away, while
the Buddhist approach to the problem is deplorable, because it
will eventually lead us to reject any discrimination between
actual and fictitious entities.
To simplify the matter for discussion, we might talk in terms
of exemplified and unexemplified properties (borrowing the terms
of Carnap with suitable modifications2) instead of talking in
terms of referring or non-referring expressions. A referring
expression, be it a name or a description, can be said to
signify a property (in a broader sense) which is exemplified
by the individual it names or the individual that answers the
description. In the light of this theory, a referring expression
like 'Aristotle' would signify such a property as Aristotlehood
(essence of the person called Aristotle ?). A non-referring
expression signifies a property, or rather a complex of properties,
which is not so exemplified. An unexemplified property (or,
property complex) may also be called an empty property.
The Buddhist argues that it is possible to talk about fictitious
objects or empty properties, because otherwise one cannot even
successfully deny their existence. Thus, it is in order when we
1
Vâcaspati, Nyàya-Vârttika-Tatparyaiïkâ, K. S. S. 24, Benares (1925),
2
pp. 172-73.
R. Carnap, Meaning and Necessity, Chicago, 1956, pp. 20-21.
purport to attribute empty properties to fictitious individuals or

when we use a fictitious individual as a 'disagreeing example' or
even as the subject of some negative proposition. A putative
answer to 'the riddle of non-being' can be given as follows : an
utterance like
"The rabbit's horn does not exist"
is perfectly all right because here we only deny the existence
or 'actuality' of a fictitious entity. Similarly, the utterance
"The rabbit's horn is not sharp"
is also in order, because here we simply refuse to attribute a
property, viz., sharpness, to a fictitious object. One may also
add that the correctness or soundness of these utterances can
be authenticated by our accredited means of knowledge (pra-
mänd). We can neither perceive the rabbit's horn, nor test its
sharpness by direct perception, nor can we infer its sharpness on
any logical grounds (such as, knowing that it cuts hard objects
easily).
But if we accept empty subject terms in order to make the
denial of existence to fictitious entities successful we shall invite
other logical problems. Udayana points out that if the negative
sentence
'The rabbit's horn is not sharp"
is held to be true, because it is authenticated by our accredited
means of knowledge, then the affirmative sentence
"The rabbit's horn is sharp"
can be argued to be also true on similar grounds. Since the
rabbit's horn cannot be known through any means, no one can
establish that it is not sharp. As long as it is not established
that it is not sharp our claim that it is sharp should also hold.
Udayana notes
If nobody has ever seen or known a person called 'Devadatta'
anywhere at any time, then the question "Is Devadatta
white, or is he black ?" results simply from some outrageous
perversion. And if, without caring to understand what this
is all about, someone answers the quesition by saying "he
is white" another person has as much right to answer by
saying "he is black." Nothing is established by such
questions and answers. In each case, the lack of our means
of knowledge (to establish the subject term) and (the con-

sequent possibility of) self-contradiction remain the same.1
The point that Nyäya tries to make through this criticism is this.
If we allow statements about fictitious entities in a logical
discourse—statements by which we purport to attribute some pro-
perty (positive or negative) to the fictitious entity—we shall
have no way of deciding whether they are true or false, for, it
will never be possible to experience the fictitious entity through
any accredited means of knowledge. But the Buddhist argues
that we do utter statements about fictitious entities We tell
fictitious stories, and we can conceive of unreal entities like the
rabbit's horn or the hair of a turtle. It is not always the
case that we have to know a thing before we may make
statements about it or attribute some property to it. A
simple cognition, an error, a conceptual construction, or even a
deliberate attempt at fiction, will be enough to prompt us to talk
about fictious entities. Statements about fictitious entities like the
rabbit's horn may also serve some useful purpose in a logical
discourse.
In fact, the Buddhist proposes a kind of pan-fictional approach
to the world of phenomena. Thus, he believes that language
creates fictions and the cognitive element behind language can
very well be the cognition of a fiction. In other words, an
unreal entity can be as much the object of a cognitive episode
(i.e., an erroneous cognitive episode) as it can be expressed by
some non-referring expression in language.
The Epistemological Significance of the Controversy

Let us consider
A "The rabbit's horn is sharp."
Nyäya says that here the subject term itself can be treated as a
complex term, in fact, a 'disguised' sentence, in which we are
attributing either the property of having horns to a rabbit, or
the property of beloagLng-to-a-rabbit to the horn. As long as
this is a wrong attribution, the whole sentence should be
regarded as false, because it can only represent a possible error,
i.e., an erroneous cognitive event.
1
Udayana, Ätmatattvaviveka, K.S.S-84, Benares, 1940, p. 69.
The issues involved here are eventually connected with an

epistemological problem, i.e., the problem of explaining an
erroneous cognition. Nyâya contends that our error consists
only in our wrong attribution of a property to a subject. But
the property itself or the subject itself must be a real entity of
this universe. The Buddhist claims that an error does not neces-
sarily consist in the wrong attribution. According to a section of
the Buddhists (viz., the proto-Mädhyamika), an error consists
in making an unreal thing appear as real. The Nyâya theory is
called the anyathäkhyäti theory of error, while the proto-Mädh-
yamika view is called the asatkhyäti theory of error.
Let us consider the situation of a perceptual error for the
purpose of comparison. When someone perceives (erroneously)
a snake in a situation where only a rope is present in the visual
field and hence is in contact with one's sense organ, we can say
that he has an erroneous perception of a rope. In fact we can
show him by examining the object further that what he saw was
not a snake but actually a rope. But if he is dense enough (or,
philosophic enough) to ask "It was not what snake?" we shall be
in trouble, because it will be difficult to successfully deny the
existence of a non-existent or imagined snake. Whether or not
he has seen a real snake in life before is just beside the point
here. The particular snake which he thought he perceived a
moment ago cannot be the subject of a successful denial, because
if that particular snake (or, perhaps, we should say, the snake-
particular) is totally non-existent, then the denial of its existence
will be pointless, and if it is supposed to have some sort of
existence (i.e.,'subsistence') then the denial will be contradictory.
One may resort to such an ambiguous position as the following :
The particular snake I experienced a moment ago is not on the
same level of reality as this piece of rope which I am experienc-
ing now, but, nevertheless, that particular snake-fiction is not
entirely unreal, because otherwise I would not have experienced
it a moment ago. This will invite a host of philosophic questions
about the nature and criteria of reality which we need not go
into here. Our purpose will be served if we remember the meta-
physical background in which the Nyäya and the Buddhist
theories of error were developed.
It will be consistent for the Buddhist to say that the object
of an erroneous cognition (or of a dream cognition), viz.,
is a snake', is unreal. The following judgment, viz., 'this is not a

snake', which destroys the error simply reveals the unreality, i.e.,
non-existence of the snake. Suppose we accept causal efficiency
as our criterion of reality. A real entity may be the object (visaya)
of a perceptual cognition by being somehow causally related
to the production ofthat perception. But if something has become
the 'object' (visaya) of a cognitive event, it does not follow that
it must have been causally related to the production of that
cognitive event. For example, the 'snake' grasped by an errone-
ous perception of the form 'this is a snake' becomes the 'object'
of this cognitive state without apparently being causally related
to it. Thus, a cognitive state is an intentional act where a non-
existent entity can very well be revealed as the 'object'. This is,
in brief, what is implied by the asatkhyäti theory of error.1
The Nyäya reply to this is somewhat theory-bound. Nyâya
realism does not admit that a totally fictitious entity can be the
'object5 of any cognitive event, even of an error. Nyäya attempts
to construct a theory of reality, a conceptual scheme, that consists
of some interconnected basic categories (viz., the scheme of
Vaisesika categories). Thus, it is asserted that corresponding to
each fundamental element of thought or cognition there is a
fundamental element of reality. The so-called fiction is always
constructed out of real elements. And these real elements can
be categorized under some basic principles.
A judgmental cognitive event combines two or more elements
of reality, and a sentence expresses such a cognition. An error or
erroneous judgment combines two elements of reality which are
not actually so combined. To be more precise, an error com-
bines two elements of reality that are not so combined in reality.
One might say that the verbal expression of an error purports
to refer to a fictitious entity. But Nyâya hastens to add that the
fundamental elements that go to constitute the object-content
of an error are all real elements or real properties exemplified
somewhere in the world. Even the oddest imagination can be broken
into elements each of which is not just 'airy nothing9, but 'has
a local habitation and a name9, i.e., is real.
p , op. cit., pp. 85-86. For the Nyäya theory and the Prâbhâkara
theory of error, see my Review of Mohanty's Gangesa1 s Theory of Truth9
Philosophy East and West 18, pp. 327-28. I have discussed the different
theories of error in a book on Perception : An Essay on Classical Indian
Theories of Knowledge : Chap. 6.
Nyäya criticizes the asatkhyäti theory of the Buddhist by saying

that what becomes the 'object' of a cognitive event is also, in
some acceptable sense, causally related to that cognitive event,
and that it is absurd to suggest that a fictitious non-entity can
cause anything. Thus, in the case of perception, what becomes the
'object' causes, at the same time, the perception itself as an event.
Even a perceptual error has an 'objective basis' (älambana)
which causes such an error. In the case of wrong inference also,
we do not deal with fictions. We are either mistaken about our
initial perception which lies at the root of such an inference, or
we fail to apply the rules of correct inference. In the case of an
erroneous cognitive state derived from someone's utterance, we
should analyse its object-content also in a similar fashion.
It may be argued that the expression 'the rabbit's horn' is
understood to refer to some non-existent entity, because no such
thing as the rabbit's horn is found in this universe. Nyâya meets
the objection in this way. Like perception, language is also a
source of knowledge. Through words or a word-complex, we know
atomic objects. This is called conception. And from sentences we
know facts or object-complexes. This is called verbal judgment.
Thus, the situation is comparable to that of a perception where
we can have either judgmental perception of facts or 'simple'
perception of (atomic) objects. Error arises in the case of judg-
mental or qualificative perception only (see § 3.1, § 3.2). It is
impossible for a non-qualificative 'simple' perception to be wrong
about anything unless we allow that the 'simple', unitary
(indivisible) object-content of such a perception can sometimes
be totally fictitious having no 'objective basis' {älambana) in the
objective world. But as far as Nyâya is concerned, this possibi-
lity is ruled out. A qualificative perception can be described as
c
a sees that x is P\ and hence there arises the chance of its being
erroneous in case the property expressed by ' P ' is not actually
present in x. Here, x is the älambana 'objective basis'. The
property expressed by CP9 may be a simple one, or it may be a
complex one in which case it should be analysable into simple
components.
The Implicit Nyäya Principle of Semantics

Nyâya asserts that a simple, non-complex property can never be
empty. We cannot conceive of a simple, non-complex property
which is not instantiated by anything in this world. Acceptance

of this principle is almost axiomatic in Nyäya realism. Thus, a
complex property is held to be analysable in this system into
'simple' components which must be individually instantiated
somewhere in the actual world. When we say that a particular
property is empty, we mean that the combination of a number
of simple real (i.e., non-empty) properties is not as such
exemplified in the actual world (at a given time). One may
combine two contradictory or mutually exclusive properties in
which case the combination will be expressed by a self-con-
tradictory expression. One may also combine two non-exclusive,
but hitherto uneombined or unconnected, properties in which
case it will be an empty property expressed by a non-referring
expression. "The son of a barren woman" will be an example
of the first kind, ''the rabbit's horn," "a man who is twenty-feet
high," and "the hair of a turtle" are examples of the second
kind. (Note that one has to take the time of utterance also into
consideration : "the present king of France" was not a non-
referring expression when France was a monarchy, nor "the
first man in the moon" is so now since the time Neil Armstrong
landed on the moon.)
A problem will apparently arise if someone insists that a non-
referring expression, such as "the rabbit's horn" or "unicorn",
is not the result of a combination, but should be treated as an
'atomic' non-divisible expression, such as unicornhood. In other
words, whatever is signified by such an expression could be
regarded as a simple, indivisible property. In that case, one can
say that we do utter expressions signifying simple, indivisible
properties which have no exemplification in the world of experi-
ence. Udayana anticipates this objection. He adds that if for any
perverse reason one wishes to admit expressions in our language,
which signify simple but fictitious (empty) properties, one will
have to find a way to distinguish between two different expres-
sions which signify two empty (but presumably different)
properties.1*
To explain : Unless we allow the meaning of '«unicorn" to be
indistinguishable from the meaning of ''goblin" (or, from that
of "the round square") we cannot hope to prove that these
HJdayana, Ätmatattvaviveka,p. 71.
Problems of Philosophical Logic in Navya-Nyâya 101
expressions signify 'simple' but different empty properties. The

meaning of such expression cannot, in the first place, be learnt
by ostensive means. We depend upon descriptions to make
their meanings understood. Thus it is that the apparently unitary
expression finally resolves itself into a series of descriptions
which will, in their turn, signify complex properties.
The way Nyâya deals with this problem seems to be close to
what Carnap has to say about his artificial language system S^ :
Generally speaking, it must perhaps be admitted that a desig-
nator can primarily express an intension only if it is exempli-
fied. However, once we have some designators which have a
primary intension, we can build compound designators out
of them which express derivative, complex intensions, no
matter whether these compound designators are exemplified
or not.
Nyäya believes that there is a class of 'simple' properties (which
are exemplified and) which are expressed by a class of simple'
(non-compound) designators. But compound designators express
complex properties which may or may not be exemplified.
Following the Nyâya principle we can resolve
A "A rabbit's horn is sharp"2
as follows : "(1) Something is characterized by horn-ness and
(2) it is characterized by the property of belonging to a rabbit
and (3) it is also characterized by sharpness." Of these con-
stituents, if (1) is true, (2) cannot be true, and vice versa. And
(3) can be true or not true according as the subject (whatever
that is) is or is not sharp. But, in no case will the conjunction
be true. And, we can resolve
B "A rabbit's horn is not sharp"
as follows : "(1) Something is characterized by horn-ness and
(2) it is characterized by the property of belonging to a rabbit
and (3) it is also characterized by the absence of sharpness."
*R. Carnap, op. cit., p. 31.

2
I am using the indefinite article 'a* instead of the definite 'the' and main-
taining the view that such statements will have existential commitment
without the uniqueness condition of the definite article. I have chosen
this procedure in order to avoid unnecessary complications.
This will be equally not true (a-pramä) as before, because both

( 1) and (2) cannot be factually true together. (3) will be true
or false according as its supposed contradictory (3) is false or
true.
One may note here that Russell has declared that a sentence
of type B is ambiguous. He introduces the notion of "primary"
and "secondary" occurrences of descriptions to explain this
ambiguity.1
Russell extended the concept of false statement to include sen-
tences with empty descriptions as their subjects. He was criticized
by Strawson on the ground that in ordinary language the existence
of something denoted by the subject of a sentence is a necessary
condition of its utterance being true or false.2 Strawson's theory
can, in fact, be systematically developed with the assumption
of more than two truth-values. One might suggest from this
controversy that a multivalued system is even inescapable if logic
is drawn closer to ordinary language. Besides, certain paradoxes
can, perhaps, be better tackled in a multivalued system.
However, the Buddhist's insistence that we can and do make
statements with empty subject terms need not be taken as a plea
for accepting a third truth-value 'neither true nor false' ('inde-
terminate') to be attached to such statements. His argument is
more like that of Meinong who wants us to accept the fact that
there are unreal objects which can be spoken about, can be
thought of or desired. We may describe an unreal object like
the rabbit's horn as having a property or characteristic, such
as non-momentariness (or, even the lack of potentiality to
produce something). Thus, a term expressing such an unreal
object can very well be used as a 'disagreeing example' or as the
subject term of the contraposed version of the universal con-
comitance between existence and momentariness.
The Buddhist, in fact, would like to put all the objects over
which our thoughts and other psychological activities may range
at the same level; and this will include not only (a) things
which do exist now (i.e., which are assumed to be existent by
the common people or by the realist) but also (b) things which
do not exist now (i.e., past and future things), (c) things which
1
B. Russell, Introduction to Mathematical Philosophy (1919), p. 179.
2
P. F. Strawson, «'On Referring", Mind, 1950.
cannot exist (viz., the rabbit's horn), and also (d) things of
which it would be a logical contradiction to say that they exist
(viz., the son of a barren woman). One point is common to all
of these four groups, and this is that we can think about them
and our mental activities can be directed toward all of them. In
their theory of objects, the Buddhists were not interested in
ontology or in the metaphysics of being. If this opens the door
to idealism, it may be welcome to the Buddhist (because that
would simply prove the Yogäcära point that objects are integral
parts of, in fact, indistinguishable from, consciousness). Even
without giving in to idealism the Sautrântika Buddhist may
maintain this theory of objects with due modification while
emphasizing that the real objects are only momentary point-
instants which are beyond the range of ordinary experience. (For
further discussion of these issues see Chapter 3 specially § 3.3.)
By way of documentation, I give below the translation of an
excerpt from Udayana's, Ätmatattvaviveka, pp. 56-69. Order of
the paragraphs here follows that of the Sanskrit text which is
reproduced after the translation.
By such arguments, the negative method of inferring
momentariness of an entity from its existence is also
rejected.
Besides, there are some other defects in this negative
inference. The minor term (the 'subject' paksa), the middle
term (the 'inferential reason' he tu) and the example cited in
such an inference cannot be established by any means of
knowledge. There cannot be any me,ans of knowledge to
establish a non-entity (i.e., a fiction, avastu). If it could be
established by some means of knowledge it ceases to be a
non-entity.
(Opponent :) If so, then your talk about the non-entity
becomes self-contradictory.
(Proponent :) Does this self-contradiction point out that
there is a means of knowledge to establish the non-entity ?
Or, (second question) does it reject the prohibitive statement
that we should not talk about non-entity ? Or, (third
question) does it imply that we must concede such state-
ments (about non-entity) which are unauthenticated, i.e.,
not established by any means of knowledge ?
The first alternative is not tenable. Even a thousand of

self-contradictions cannot conceivably show that (the non-
entity like) the stable object (i.e., the minor term) or the
absence of gradual efficiency, etc. (i.e., the hetu) or the
rabbit's horn (i.e., the example cited to support the general
premise) 'is amenable to (a means of knowledge, such as)
perception and inference. If it could, what is the use of
this silly fight over the nature of non-entities ?
The second alternative is acceptable to us, because we
admit only valid means of knowledge.
(Opponent :) If the prohibitive statement is rejected, no
statement with regard to non-entities will be possible.
(Proponent.') What else can we do but remain silent in
regard to a matter where statement of any kind would be
logically incongruent? Silence is better in such cases. (No state-
ment is better than any statement in such matters.) You your-
self may please consider as to who is the better of the two : One
who is making statements about entities that cannot be
established by any means of knowledge ? Or, the other person
who remains speechless (on such occasions) ?
(Opponent:) But, although you are a wise man, you
have not remained silent yourself. You, on the other hand,
have made a prohibitive statement with regard to our talk
about non-entities !
(Proponent :) True. In order to avoid a self-contradictory
object not established by any means of knowledge, you have
conceded that one can make statements about the non-
existent. Similarly, in order not to allow any statement
about the non-entities in our discourse on the means of
knowledge, we concede that a self-contradictory statement
(prohibiting the use of non-entities) is possible, although
it is not supported by any means of knowledge. If you
treated both the cases in the same manner, we would not
have said anything about non-entities. (We have made the
above self-contradictory statement because you first raised
the question.)
To the third question we say the following : By whose
command, may we ask you, one has to accept (in a philo-
sophical discourse statement about) an object which is not
established by any means of knowledge ?
(Opponent :) Because, use of such statement is deeply

rooted in our habit (speech-behaviour).
(Proponent :) It will be a self-contradiction to claim
that something which cannot be established by any means
of knowledge can be deeply rooted in our speech-behavi-
our. (We speak about objects that appear in our experience
and a non-entity cannot appear in our experience and hence,
cannot be spoken about.)
(Opponent :) Such entities are somehow established
(through, for example, some wrong cognition).
(Proponent :) If it is an unreal entity it can never be
established through any means of knowledge. If it is estab-
lished by a means of knowledge, then we ask you to describe
that means of knowledge. In väda, i.e., the philosophic
debate whose object is to seek truth, we expect you to state
the means of knowledge.
In the case ofjalpa 'a philosophic debate whose sole object
is to defeat the opponent', or in vitandä 'a philosophic debate
whose sole object is to defeat without establishing any
position' too, we shall not be guilty of self-contradiction,
because we simply ask you about the means of- knowledge
by which your paksa 'minor term', etc. are established. It
is impossible as well as undesirable for you to answer
by supplying the means of knowledge. If you answer with-
out supplying the means of knowledge, then your own
statement will be self-defeating, because you have to admit
yourself that there is no means of knowledge to establish
your paksa 'minor term', etc. And if you do not answer at
all you will meet the 'defeat-situation' (nigrahasthäna) called
a-pratibhä, 'silence due to lack of intelligence'. [Read
paksädisu' instead of 'prasnädhu' ; see also for Udayana's
comments on väda, etc., § 1.2.]
If you could avoid self-contradiction by only conceding
statements about non-entities (like the rabbit's horn) we
could have allowed such an expedient. But this is not so.
There is no lack of self-contradiction when we say that
something with regard to which no statement can be made,
can be the subject of a denial or a prohibitive statement.
(Opponent :) How can there be self-contradiction if we
say (rephrasing our position) that a non-entity cannot be
the subject of any affirmative statement ?

(Proponent :) Oh! (Then we ask the following question :)
Do you or do you not speak about something which can
never be the subject of any affirmation or denial ? In either
way you will contradict yourself. In either case, that some-
thing would be a non-entity because a real entity cannot be
such that we cannot make any statement about it.
If you say no to the above question, then it will contradict
the very statement "(something) can never be the subject of
any affirmation or denial". For you cannot negate unless
there is some subject of negation. If you, on the other hand,
say yes (that is, if you admit that we can speak about some-
thing which can never be the subject of any affirmation or
denial), then contradiction will appear as soon as we discuss
the nature of that object (that something). It does not hold
that something cannot be talked about, i.e., affirmed or
denied, and, at the same time, it can be talked about (in the
manner just stated).
If you insist that a non-entity can be the subject of a
denial (i.e., a negative statement), we would ask : why can
it not be the subject of an affirmation too ? The lack of an
accredited means of knowing (such a non-entity) remains
the same in both the cases.
(Opponent : ) That the son of a barren woman does
not speak can be proved (established) by pointing out
that he is not a conscious being. But there is no way of
proving that the son of a barren woman does speak. (Thus,
we prefer the negative statements.)
(Proponent : ) No. Even to prove that (the son of a
barren woman does speak) you can assign the reason that
he is a son (of someone and hence can speak). You cannot
say that the son of a barren woman is not a son because if
you do you will contradict yourself.
(Opponent :) It is a mere statement (viz., "the son of a
barren women"), in fact there is no real son of a barren
woman. (Hence no contradiction.)
(Proponent :) No. Even the lack of consciousness (which
you adduced as your reason to prove lack of speech in the
son of a barren woman) will be treated in the same manner.
(One might say : "the lack of consciousness in the son of a
barren woman" is a mere phrase, there being no real son

who lacks consciousness.) "Lack of consciousness" refers
(in fact) to another nature which is different from con-
sciousness.
(Opponent:) We simply want to deny (possession of)
consciousness here, and this is quite possible.
(Proponent :) No. In our case also we shall say then that
we simply want to deny (possession of) the property of not
being a son.
(Opponent :) Our mere denial of not being a son (human
offspring) cannot by itself prove the presence of activity and
cognition. Thus, how can we adduce a reason (hetu) without
including another competent, determinable (and positive)
object (in our hetu) ? (Denial cannot reveal an entity which
we can use as our 'reason' hetu.)
(Proponent :) No. The same principle will apply to lack
of consciousness.
(Opponent :) An object only in the form of a negation of
the contrary possibilities can be used as the reason (to
prove anything). For example, the nature of being a simsapä
tree is adduced as the reason (in the inference : it is a tree
because it is a simsapa tree) because the nature of being a
simsapä means the negation of its not being a simsapä.
The son of a barren woman, on the other hand, not simply
excludes (i.e., negates) the possibility of being a pot, etc.,
(the contrary possibilities) but also excludes the possibi-
lity of being a son (a human offspring) like Devadatta
(which is not a contrary possibility). Therefore, the
nature of being the son of a barren woman cannot be
adduced as the 'reason' hetu.
(Proponent:) The lack of consciousness (in the son of a
barren woman) can also be treated in the same manner. It
is not the case that the son of a barren woman (who lacks
consciousness) excludes only such conscious beings as
Devadatta and not also the unconscious objects such as a
piece of wood.
(Opponent :) Speech is a property which is present only
in real entities. Thus, in the face of contradiction, how can
the presence of speech be proved in a non-entity ?
(Proponent :) Through what means of knowledge can

this contradiction be established ? Is it because we appre-
hend non-entities always without speech? Or, because we
do not apprehend speech divorced from a real entity ?
A non-entity can never be apprehended through any
means of knowledge. If it could, it will not be a non-entity.
The latter alternative is also not tenable because we shall
be in the same predicament. Just as speech as divorced
from any real entity cannot be established, so also lack of
speech divorced from any real entity cannot be proved by
any means of knowledge.
(O pponent :) There is only a conceptual construction of
the lack of connection between the property speech and the
subject.
(Proponent :) What can stop a 'conceptual construction*
(vikalpa) of the connection between speech and the
subject ?
(Opponent :) To be a speaker means to produce speech.
How can such a productive power be present in a non-
entity which is (supposed to be) devoid of all power to do
anything?
(Proponent :) How can there be even lack of speech ?
For, lack of speech means the character of producing what
is not speech. ( Speechlessness implies doing something else.)
(Opponent :) It is not contradictory to say that if there
is lack of power to do anything, then there is also the lack
of power to speak.
(Proponent :) By what means of knowledge can it be
established that the son of a barren woman lacks all power
to do anything ?
(Opponent :) It lacks all power because it is a non-entity.
(Proponent :) How can you establish that it is a non-
entity ?
(Opponent :) Because it lacks all power to do anything.
(Proponent :) In this way you are only moving here and
there uttering mere words and trying to avoid the issue, just
as a penniless borrower of money tries to avoid the money-
lender. And you are not seeing the vicious circle (the defect
of mutual dependence).
Problems of Philosophical Logic in Navya-Nyäya 10^
(Opponent :) It is a non-entity because it lacks both

gradual efficiency and simultaneous efficiency. (If something
cannot produce a thing either gradually or simultaneously,
it is a non-entity.)
(Proponent :) No. In order to prove such lack of gradual
or simultaneous efficiency, you have to adduce a means of
knowledge. If the nature of being a son is applied to it, all
other properties, such as being able to speak, that invariably
go along with being a son, can also be applied to it. Thus,
how can there be any scope of proving the lack of gradual
and simultaneous efficiency in it? How can there be any
scope of proving that it is a non-entity ? And, how can there be
any chance of proving the lack of activity, etc. ? Therefore, it
is our accredited means of knowledge only which is the limit
within which our speech-behaviour should operate. If this
limit is transgressed, there will certainly be chaos.
SANSKRIT TEXT
Udayana, Ätmatattvaviveka, pp. 56-69.

Etena vyatirekapakso'pi nirastah.
Adhikas ca taträsrayahetudrstäntasiddhau pramänäbhä-
vah, avastuni pramänäpravrtteh, pramänapravrttäv alîkat-
vänupapatteh.
Evam tarhy avyavahäre svavacanavirodhah syäd iti cet.
Tat kini svavacanavirodhena tesu pramänam upadarsitaip
bhavet, vyavahäranisedhavyavahäro'pi vä khanditah syät,
aprämäniko'yani vyavahäro'vasyäbhyupagantavya iti vä
bhavet.
Na tävat prathamah, na hi virodhasahasrenäpi sthire tasya
kramädivirahe vä sasasrnge vä pratyaksam anumänam vä
darsayitum sakyam, tathätve vä krtam bhautakalahena.
Dvitïyas tv isyata eva prämänikaih.
Avacanam eva tarhi tatra präptam.
Kim kurmo yatra vacanam sarvathaivänupapannam
taträvacanam eva sreyah, tvam api paribhävaya tävat ni sprä-
mänike'rthe mükavävadükayoh katarah sreyän ?
Evam vidusäpi bhavatä na mükibhüya sthitam, api tu
vyavahärah pratisiddha evasatïti cet.
Satyam, yathâ aprämänikah svavacanaviruddho'rtho ma

prasânk?ïd iti manyamänena tvayä ca aprämänika evüsati
vyavahärah svîkrtas tathâsmâbhir api pramänacintäyäm
aprämäniko vyavahâro ma prasänksid iti manyamänair
aprämänika eva svavacanavirodhah svikriyate. Yadi tübha-
yaträpi bhavän samânadrstih syâd asmäbhir api tadä na
kincid ucyata iti.
TYtïye tv aprämänikas cäpy avasyäbhyupagantavyas ceti
kasyeyam äjneti bhavän eva prastavyah.
Vyavahärasya sudrdhanirudhatväd iti cet.
Aprämänikas ca sudrdhanirudhas ceti vyäghätah.
Kathaficid api vyavasthitatvâd iti cet.
Aprämänikas cen na kathaficid api vyavatisthate, prä-
mänikas cet tad evocyatâm iti vàde vyavasthà.
Jalpavitandayos tu paksädisu pramänapräsnamätrapravrt-
tasya na svavacanavirodhah. Tatra pramänenottaram
anistam asakyam ca, apramänenaiva tüttare svavacanenaiva
bhangah, maduktesu paksädisu pramänam nàstîti svayam
eva svikärät, anuttare tv apratibhaiveti.
Yadi ca vyavahärasvikäre virodhaparihärah syät asau
svikriyetäpi, na tv evam, na khalu sakalavyavahäräbhäjanani
ca tannisedhavyavahärabhäjanam ceti vacanam parasparam
avirodhi.
Vidhivyavahäramäträbhipräyenäbhäjanatvaväde kuto viro-
dha iti cet.
Hanta, sakalavidhinisedhavyavahäräbhäjanatvena kincid
vyavahriyate na vä, ubhayathäpi svavacanavirodhah,
ubhayathäpi avastunaiva tena bhavitavyam, vastunah
sarvavyavahäravirahänupapatteh.
Netipakse sakalavidhinisedhavyavahâravirahïty anenaiva
vyavahärena virodhät, avyavahrtasya niseddhum asakyatvät.
Vyavahriyata iti pakse'pi visayasvarüpaparyälocanayaiva
virodhät, na hi sarva-vyavahärävisayas ca vyavahriyate ceti.
Yadi cävastuno nisedhavyavahäragocaratvam, vidhivyava-
häragocaratäpi kim na syät pramänäbhävasyobhayaträpi
tulyatväd iti.
Bandhyäsutasyävaktrtve' cetanatvädikam eva pramänam
vaktrtve tu na kincid iti cet.
Na, taträpi sutatvasya vidyamänatvät, na hi bandhyäyäh
suto na sutah, tahä sati svavacanavirodhät.
Vacanamâtram evaitat na tu paramärthatah suta eväsäv

iti cet.
Na, acaitanyasyäpy evamrüpatvät, cetanäd anyat svabhä-
väntaram eva hy acetanam ity ucyate.
Caitanyanivi ttimätram eveha vivaksitam, tac ca sambha-
vaty eveti cet.
Na, taträpy asutatvanivi ttimätrasyaiva vivaksitatvät.
Asutatvanivrttimätrasya svarüpena krtijnaptyor asämar-
thye samartham arthäntaram adhyavaseyam anantarbhävya
kuto hetutvam iti cet.
Na, acaitanye'py asya nyäyasya samänatvät.
Vyâvrttirûpam api tad eva gamakam yad atasmäd eva
(vyâvartate), yathä simsapätvam, bandhyäsutas tv asutäd
iva ghatädeh sutäd api Devadattäder vyâvartate, ato na hetur
iti cet.
Nanv acaitanyam evamröpam eva, na hi bandhyäsutas
cetanäd iva Devadattäder acetanäd api kästhäder na
vyävartate.
Vaktrtvam vastvekaniyato dharmah sa katham avastuni
sädhyo virodhäd iti cet.
Sa punar ayam virodhah kutah pramänät siddhah—kirn
vaktrtvaviviktasyävastuno niyamenopalambhät ähosvid vas-
tuviviktasya vaktrtvasyänupalambhäd iti.
Na tävad avastu kenäpi pramänenopalambhagocarah,
tathätve vä nävastu, näpy uttarah, samänatvät, na hi
vaktrtvam iva avaktrtvam api vastuviviktam kasyacit pra-
mänasya visayah.
Tadviviktavikalpamätram tävad astîti cet.
Tatsamsrstavikalpane'pi ko värayitä ?
Nanu vaktrtvam vacanam prati kartrtvam, tat katham
avastuni, tasya sarvasämarthyavirahalaksanatväd iti cet.
Avaktrtvam api kathan tatra tasya vacanetarakartrtvala-
ksanatväd iti.
Sarvasämarthyavirahe vacanasämarthyaviraho na virud-
dha iti cet.
Atha sarvasämarthyaviraho bandhyäsutasya kutah pra-
mänät siddhah ?
Avastutväd eveti cet.
Nanv etad api kutah siddham ?
Sarvasämarthyavirahäd iti cet.

So'yam itastatah kevalair vacanair nirdhanädhamarnika
iva sädhün bhrämayan parasparäsrayadosam api na pasyati.
Kramayaugapadyavirahäd iti cet.
Na tadvirahasiddhäv api pramänänuyogasyänuvrtteh,
sutatve ca parämrsyamäne tadavinäbhütasakalavaktrtvädid-
harmaprasaktau kutah kramayaugapadyavirahasädhanasyä-
vakäsah, kutastaräm cävastutvasädhanasya, kutastamäm
cäkartrtvadisädhanänäm. Tasmät pramänam eva sfmä vyava-
härasädhanasya, tadatikrame tv aniyama eveti.
§ 2.3 : THE NAVYA-NYÄYA LOGIC OF

PROPERTY & LOCATION
A judgemental cognition in Navya-nyäya is analysed in terms

of property and location. Negation is always construed as
term-negation. Sentential negation is usually transformed into
term-negation of some kind or other. Negation of a property
generates another (negative) property. A negative statement is
analysed as attribution of a negative property. Properties, here,
are to be understood not simply as universals. They would
include any occurrent or attributable, specific features, which
may even be particulars.
The universe U is peopled with loci or locations where
properties are locatable. The presence-range of a property is
the set of loci where it is locatable. The absence-range is the
set of loci where it is not locatable.
A property with an empty presence-range is unbeatable. It
is ruled as fictitious (e.g., the golden mountain). Properties
with empty absence-ranges are admitted as real (non-fictitious),
e.g., knowability. They are called ever-present (see next section).
Both the fictitious (unlocatables) and the ever-present are
ruled as unnegatable, for, negation of them does not generate
real (locatable) properties. A property is unreal if it is not
locatable.
Most properties are wholly locatable, such that they are not
co-locatable with their absences in the same set of loci. But
some properties are partially locatable, such as chair-contact.
Such a property is apparently co-locatable with its absence in
Problems of Philosophical Logic in Navya-nyäya 113
the same locus. This infringes upon the generally understood

law of negation. For we can say, with regard to the same
locus, that it has as well as does not have a particular property
(in the given sense). Thus, a device is used to reparse the
partially locatable properties as wholly locatable, so that the
standard notion of negation may not be 'mutilated' in this
system.
Non-deviation and pervasion are two important logical
relations which generate inference in the system. The Navya-
nyäya formulation of these relations will be given here.
Navya-nyäya insistence on the non-emptiness of the presence-
range of properties serves the purpose of making the
existential import of general statements explicit. In this
respect, non-deviation can be contrasted with the A-relation
of Aristotle.
To explain the notion of the unnegatable as well as the
negation of the partially locatable, some concepts of a multiple-
valued system may be used with an entirely different inter-
pretation of the values. The negation matrix has been given at
the end of this section. I shall continue the discussion of the
unnegatables (i.e., the ever present properties) in the next
section.
Despite the peculiarities mentioned above, Navya-nyäya
tries to work with the standard notion of negation in a two-
valued logic. This will be clear from the two previous sections
(§ 2.1, § 2.2) as well as from this section.
The quest for good reasons which generate dependable and
acceptable conclusion is almost universal. Indian logic, by which
I mean a combined tradition of the Buddhist, Nyâya and the
Jaina, is only another instantiation of this universal quest in the
intellectual history of mankind. It represents an independent
tradition of studying inference and its soundness. Just because
of its difference as well as independence from the Western
tradition, the inference theory developed here should prove
extremely interesting for both logicians and philosophers. The
Indian theory of inference shows also a continuous development
from the pre-Christian era upto the seventeenth century A.D.
It lacks, it is true, some of the familiar logical (and mathe-
matical) notions which logicians of today have come to expect.
But then it offers a contrast in these areas with Western logical
theories that developed primarily during the last two centuries.

It is also instructive. For, it shows, at least, what other ways
are left to us for solving some logical problems in case certain
färniliar devices were not available.
With this as a prelude let me describe some features of what
has been called "Navya-nyäya logic" or just "Navya-nyäya"—
a system that developed within the Nyâya tradition beginning
from 1100 A.D. It absorbed the Buddhist criticism of the earlier
Nyâya school and reformulated its older theory of inference.
In this section, I shall first outline the Navya-nyâya concept
of property and location and the logical relations formu-
lated in terms of property and location. I shall then make
some observations to show the relevance of some Navya-nyäya
theories to certain modern concerns in the philosophy of logic.
Cognitive states
Navya-nyäya analyses cognitions in terms of property and
location or locus. More correctly, Navya-nyäya analyses what
I have elsewhere called judgemental or qualificative cognitive
states in terms of qualifiers and qualificands.1 Such a cognitive
event is usually represented by a sentence. Because of the use
of the term "cognitive" or ''cognition" here, a logician trained
in the tradition of Frege and Carnap may immediately bring the
charge of "psychologism" against Navya-nyäya. But I have argued
elsewhere that this charge is not always relevant.2 Navya-nyäya
is concerned with the objective' content of a cognitive event
and analyses the sentence that is supposed to represent the
structure of such, a content. It is not concerned with the psy-
chological act of cognition as such. Thus, in Navya-nyäya
logic when one cognitive event is said to be contradictory to
another, it is not just their psychological impossibility that
is appealed to. In other words, what is appealed to here
is the impossibility that is completely determined by the logical
relation between p and not-p.
Dignäga suggested a dharma-dhannin (/'property and locus")
analysis of a qualificative (judgemental) cognitive event. In
X
B. K. Matilal (1968), p. 12.
*See B. K. Matilal (1986), p. 121-122.
Dignâga's terminology, however, such a cognitive event is called

'constructive'; for, Dignäga, like the British empiricists,
emphasized a distinction between the data (immediately 'given'
in consciousness) and the constructs based on the data.
Existence or reality is ascribed only to the data (svalaksana
"unique particular"), and the constructs are products of
imagination (kalpanä). Navya-nyäya rejected this ontology of
data of the Buddhists, but accepted the dharma-dharmin
analysis of a cognitive event that is propositional.
Properties
A cognitive eyent is usually said to locate a property in a
locus : the form is 'x has p* or '/? (is) in x\ Simple predicate
formulations, such as 'x is FJ are noted, but only to be
rephrased as {x has F-ness' (where "F-ness" stands for the
property derived from "F"). Thus, we have here two types of
individuals—properties and locations or loci. Correspondingly,
we can talk about two sorts of individual constants : property-
terms (r, s, t, u, w, h....) and location-terms (/, m, n, o, p).
The best example of a property-term is "blue-colour" which
is locatable in a cup that is blue, or the property expressed by
"cowness" that is locatable in a cow (in any cow). Such
physical materials as a cup, lire, smoke, water, and a pot
are also treated in Navya-nyäya as properties, inasmuch as
they are locatable in such loci as a table, a mountain, ground,
the kitchen and the plate. Hence, terms expressing such
physical materials are treated as property-terms in the specific
sense of being a property-particular, that I have alluded to in
the beginning of this section. The apparent oddity of treating
such things as properties can be resolved if we conceive
anything to be a property that purports to have a location and
allow a sort of stipulative identity between having a cup on it
and cup-property. In other words, we have to stipulate a sort
of referential identity between such expressions as "cup-possessor-
hood" and "cup"'(used as a property). One may even suggest a
distinction here between two uses of the expression "cup": one
use of "cup" ("a cup" or "the cup") is to refer to the locus
of properties, the other use ("a cup" or "the cup") is to refer
to a property. Both refer to the same ontological entity but to
different logical constructs.
It may further be noted that even a so-called relation (a

connector) may sometimes be treated as a property in Navya-
nyäya. If a relation is tied in one end to the relatum, then the
whole complex can be treated as a particular qualifier of the
other relatum. Thus, the cup-contact in the case of a cup being
placed on a table can be treated as a property or a qualifier
of that table, provided we can take the cup-contact as a
particular locatable on the table, the locus.
Negation
Navya-nyäya basically recognizes two types of negation :
absence and difference. Most peculiar features of Navya-nyàya
emerge in connection with its interpretation of negation of
properties. Sentential negation is usually avoided. A negation
is construed as a term-negation in either of the following ways.
We get an absence when it is a negation of occurrence or location,
a difference when it is a negation of identity. When a negation
or a negative statement negates location or occurrence of a
property in a locus, it is construed as ascribing the absence of
a property to that locus. Thus, absence of a property is treated
as another property. "The pot is not blue" is first rephrased
here as "the pot does not have blue colour'* which is further
rephrased as "the pot has the absence of blue colour."
Using the complement sign '-' for term-negation, we can
represent the above statement (m -= the pot, s =blue colour) :
i(
m has ~.ç."
When a statement negates an identity between, say, a table
and a cup, it is construed as "a table is different from a cup"
("s^£t"). Navya-nyâya argues that to say that a table is different
from a cup is equivalent to saying "a table lacks the essence of
a cup, or simply, lacks cupness". In other words, 'difference
from a cup' is said to be extensionally equivalent to 'the
absence of cupness' (which means that both these properties
are locatable in the same set of loci).
World of loci : Presence-range and Absence-range

Let us conceive of a universe U, which is peopled with loci or
locations. Locations are called locations, because they accom-
modate 'properties', in our specific sense of the term, i.e., in
the peculiar sense that we have tried to develop here. And

similarly, properties are properties as long as they are locatable
in some locus or other. Henceforth, I shall use the term
'property' unabashedly in this specific sense.
Given a particular property t, we can find a set of locations
or loci where t is locatable or present, and another set of loci
where t is not locatable. Let us call the former set the presence-
range of t, and the latter the absence-range of t. Let us use the
notation 'r-f-' for the presence-range of /, and '/—' for the
absence-range of /. Thus, ordinarily, the two sets, r+and t—,
are supposed to exhaust the universe of loci U.
The iinlocatahles
Navya-nyäya demands that^the presence-range of a non-fictitious
(real) property should be non-empty. Navya-nyäya argues
that if the presence-range be empty then the property in
question would be unbeatable. 1 An unbeatable property is a
suspect in Navya-nyäya. It is regarded as a fictitious property
which cannot be located in our universe of loci. It is called an
a-prasiddha property "unexampled' property, i.e., 'unestablis-
hed\ imaginary property.2 Using modern terminology, we may
say that it is a property which has location in a possible
world, but not in the actual world, (i shall come back to this
problem in the last chapter.) Navya-nyäya hesitates to perform
logical operation on such a property. For example, one cannot
negate such a property and thereby obtain or derive another
(negative) property ! For, they would not be locatable in the
actual world. Thus, we have the following restriction on
negation : If s is a property with a non-empty presence-range,
then by negating it we get another property, a negative property
s; but if s is uhiocatable, it cannot even be successfully negated.
Properties in Navya-nyäya are either atomic (or 'simple') or
composite. A. composite property is formed out of atomic ones,
and, hence, such a property is analysable into atomic cora-
1
Sec before; sections 2.1, and 2.2. The 'unlocatable' is another way of talking
about the non-existent entities, and, hence, is connected with the problem
of empty terms.
2
D.H.H Ingalls, p. 61.
ponents or 'simple' properties. A 'simple' property is regarded

as fundamental. It is not analysable into components. (For
more on the notion of 'simple' property, see the previous sec-
tion).1 An example of simple property is: cowness. The absence
of cowness is a composite property. All fictitious properties like
the property of being a flying horse, that of being a unicorn,
a golden mountain and the son of a barren woman, are com-
posite properties, being analysable into a number of 'simple'
properties. And, it is argued, such 'simple' components are
always real properties in the sense that they are locatable in
some locus or other in our actual world (§ 2.1, § 2.2).
The unnegatables
If the presence-range of a property is empty, it is unlocatable.
Nyäya calls such a property fictitious. What about properties
whose absence-range is empty ? Nyäya admits such properties
as real, i.e., non-fictitious. They are called ever-present
properties (cf., kevalänvayiri). They are said to be locatable in
all loci of U. Examples of such properties are : knowability,
expressibility and provability (see next section).
An ever-present property is non-fictitious in Navya-nyâya, for,
its presence-range is non-empty (in fact, the presence-range is
the whole universe LJ). We have to assume that such a property
is locatable also in itself, for, it must belong to the universe U.
But since its absence-range is empty, Navya-nyäya regards such
a property as unnegatable ! In other words, just as an unlocat-
able property is said to be not negatable in Navya-nyàya, an
ever-present*property is also regarded as not negatable. For, we
cannot derive a real, non-fictitious (negative) property by
negating an ever-present property. Thus, we have another
restriction on the operation of negation : If s is an ever-present
property, it is locatable (i.e., real), but it is, nevertheless,
unnegatable.
It is obvious that the introduction of ever-present properties
in the system involves many logical difficulties. Thus, some
pre-Gangesa Nyäya logicians were definitely not in favour of
using such a concept. They argued that a true property should
have a non-empty presence-range as well as a non-empty
absence-range. If we rule the unlocatables as fictitious, we might
also B.K. Matilal (1971), pp. 83-91.
as well rule the ever-present properties as fictitious, for, both,

as we have seen, cannot be successfully negated. But Gangesa
rejected this view and argued that even if we do not accept
such properties like knowability as non-fictitious, we cannot
escape from admitting other kinds of ever-present properties.
For, if we believe that each locus in the universe of loci is
distinct from another, then this property, distinctness, can be
construed as an ever-present property (for more on this
argument, see next section).
Sondada disputed the position that the unlocatables can be
unnegatable. If we admit an ever-present property as real (non-
fictitious), i.e., accept such a property to be real as is locatable
in all loci, then, one might argue, by negating a so-called
unbeatable property, we obtain only a negative property which
should be locatable in all loci. In other words, such a negative
property has to be admitted as real because its presence-range is
non-empty (it is an ever-present property). Thus, if the property
of being a golden mountain is unbeatable, then the absence of
such a property is to be located everywhere ! For, it makes
perfect sense to say that there is no golden mountain, or that
all loci in our actual world lack the property of being a golden
mountain.
But Gangesa refuted Sondada's contention. An unbeatable
property, according to Gangesa, resists the operation of nega-
tion. Negation is restricted to the Iocatables and again only to such
Iocatables whose absence-ranges are non-empty. To say, "there-
is no golden mountain" means, for Garrgesa, that no mountain
is golden, i.e., made of gold. But '„the property of being a golden
mountain' as expressing a composite property is unbeatable.
Partial location
We face a further oddity about negation when Navya-nyäya
introduces the notion of partial location (cf., avyäpya-vrtti,
Ingalls : "incomplete occurrence'') of properties. Most
properties are wholly or pervasively occurrent or locatable in
their loci, but some properties are said to be only partially or
non-pervasively occurrent or locatable in their loci. (We may
imagine a 'propQrtyXdhanna) in this sense to be a paint-coating,
with which the locus is besmeared partly or wholly.)
To explain this notion we have to develop some further
logical vocabularies. Let us use a two place predicate (a

relational term), 'L' for "locus of''; we can then define some
other (logical) predicates or connections in terms of this iL\
First, let us define the connection of co-location, ' C . We can
say that s is co-located with t provided there is a locus where both
s and î are locatable. Thus, co-location is symmetrical. In other
words, one property is co-iocatable with another just in case their
presence-ranges intersect or overlap. Using the convention
of modern logic, we can say that s is co-locatable with t
provided the logical product of s-\- and t+ is non-empty.
Lotus-hood and blue are co-iöcatable in things we call "blue
lotuses". If such things did not exist in our actual world, the
said logical product would have been empty.
In the above we have noted that if ^ is a locatable property
then .s'-j- and s— exhaust the universe of loci U. But we have not
required the presence-range and the absence-range of s to be
disjoint. In other words, we have left open the possibility of
one intersecting the other. According to Navya-nyäya concep-
tion of negation, this* is not impossible : in other words,
a property and its absence may both be locatable in the same
locus. Navya-nyäya calls such properties partially or non-
pervasively locatable.
A property is pervasively (wholly) locatable provided it is
not co-locatable with its absence. But when a property is co-
locatable with its absence, it is called a partially locatable
property. To put it in another way, if the absence-range of a
property overlaps or intersects its presence-range, it is only
a partially locatable property.
Physical contact is the best example of a partially locatabie
property. When I am sitting on a chair, there are places in the
chair where my body-contact is absent. Thus, the same chair is
said to be the locus of my body-contact (as a property) and also
of the absence of my contact. Obviously it clashes with our
general notion of negation to say that the same locus is
characterized by a property and its absence at the same time.
(Remember that absence of a property means only the negation
of that property. How can we affirm and negate the same
property of the same locus ?) Thus, this doctrine of partial
location requires some reformulation of the usual notion oi
contradiction. A property and its absence cannot be
Problems of Philosophical Logic in Navya-nyâya 121
'contradictory' in this sense (cf., sahânavasthâna) unless their

loci or places of occurrence are specifically qualified in
details (using 'delimiters', etc,).
An example may illustrate some further problems involved
in the notion of partial location. Suppose, w is a partially
locatable property. Now the absende-range of w will include
not only those loci where w is absent (wholly) but also those
loci where w is partially present. In other words, the presence-
range of w includes the presence-range of w. Thus, the presence-
range of w is the whole universe of the loci U. This means
that if w is a partially locatable property, then w is an
ever-present property, for, the formal character of an
ever-present property will undoubtedly apply to it. Now,
if we negate further w, We are supposed to derive an
unlocatable property. (Remember the previous point:
negation of the ever-present generates the uniocatablc) How-
ever, Navya-nyâya accepts the law of double negation.
Udayana formulated the law as follows : the negation of the
negation of a property is identical with the property itself
(Nyaya-kusumanjali Gh. 3 verse 2). Thus, we must have : the
absence of w = w. We face here an apparently paradoxical
situation : if w is a partially locatable property, then w can
be shown to be unlocatable Î
Gangesa avoids this apparent problem by pointing out that
there are two kinds of ever-present property, one of which is
to be treated as unnegatable but the other is negatable. It is
all right to say that when w is partially locatable w becomes
an ever-present property in the above manner, for it is present
not only where w is absent but also where w is present. But w
is also partially locatable with regard to some of its loci. In
other words, the presence-range of w is actually a combination
of the two : its pure presence-range (where w is not present) and
a mixed range where w is co-locatable with w. Thus, w is a
partially locatable ever-present property, and as such, it is
negatable. The absence-range of w is non-empty; it coincides
with the presence-range (which is a 'mixed' range) of w. Thus,
we have a formal restriction on the former restriction of
negation : not all ever-present properties are unnegatable.
Gangesa saved the law of double negation by resolving the
oddity in the above manner. Some Navya-nyäya writers diifered
from him in this regard. Raghunâtha, for example, suggested

that the law of double negation be given up in the given
context, for, it is based upon only extensional identity (their
presence-ranges and absence-ranges being equal). Intensionally,
w and the absence of w are distinguishable.
Mathurânâtha suggested a different method of resolving the
above oddity. According to him, instead of treating w as
ever-present, we should treat the expression "iy" as ambiguously
referring to two distinct (negative) properties : one that is
partially locatable in its loci, the other wholly locatable in its
loci. The presence-range of the first is disjoint from that of the
second. The first is actually co-locatable with w, but the
second is locatable where and only where w is not locatable.
Thus, the problem of negating an ever-present property will
not arise in this case.
Deviation and Pervasion

In the above we have defined co-location. Let us define some
more logical predicates, such as, deviation (D), non-deviation
(N) and pervasion {V). We can say that li deviates from s just
in case the absence-range of the latter overlaps (intersects) the
presence-range of the former (cf., sädhyäbhävavad-vrttitvam
vyabhicârah), Using the modern logical, i.e., the Boolean con-
vention, we can write :
hDs iS h+ .s— =£0.
Similarly, h non-deviates from s if and only if s— does not
overlap h-\- {sädhyäbhävavad-avrttitvam avy ab hicarah) :
liNs iff h+.s— -0.
The relation of pervasion (vyäpti) is an important relation in
Navya-nyäya, since it allows valid inference of one property
from another. Thus, if h is pervaded by s then from the
presence of h in a particular locus, we can validiy infer presence
of s in it. The rule is: (hLm. sVli) ZD sLm.
The relation 'pervaded by' is identifiable with non-deviation
(defined above) as long as we talk of such properties whose
absence-ranges are non-empty. (For, we have used the absence-
range of i1 in the above definition of non-deviation.) However,
if .sis unnegatable, the above definition, according to Navya-
nyâya, becomes inapplicable. There are also several ever-present

properties, according to Nyäya, and, hence, one can be inferable
from another. Thus, Gangesa reformulates the definition of
pervasion which will be inclusive of pervasion between ever-
present (unnegatable) properties, (cf., hetuman-nisthäbhäväpraü-
yogi-sädhya-sämänädhikaranyam vyäptih) : Thus, we may say :
s pervades # if and only if (1) s is co-located with h and (2) if
the absence-range of any property t intersects the presence-range
of h, then t is non-identical with s.
sVh iff s+.h+^t 0 and if (f-.h + ^ 0 ) , then t^s.
A further problem arises when s becomes a partially locatable
property. For, we have seen that, by definition, the presence-
range and the absence-range of such a partially locatable
property do intersect. Thus, when s is partially locatable,
its absence-range includes its presence-range, and thereby
its absence-range intersects the presence-range of //. Thus,
the second component of the above definition may not be
satisfied by such an s. Gangesa avoids this quandary by
suggesting further qualification of the above definition :
sVh iff s+.h+^zO and if (t+.t— = 0 & h+.t—Ô) then t^s.
There will arise some further problems even in this formula-
tion, and commentators of Gangesa discussed them in detail.
But I shall move on to the next section without going into such
details.
I shall make some general observations in which I shall try
to connect the problems discussed above with the explicit con-
cern of modern logicians. This is by way of answering the
criticism, viz., why these theories would form part of a study that
has been called 'logic*. Let us note, first, that non-deviation and
pervasion relation may be compared with the A-relation of
Aristotle, for all three share the common logical feature, i.e.,
transitivity. For contrast, we may say that the Navya-nyäya
formulation of non-deviation (or pervasion relation), while it is
narrower in its scope, does not suffer from the same ambiguity
which the A-relation of Aristotle seems to have suffered from.1
It is often pointed out, for example, that the existential
import of the A-proposition should be assumed, in order that
1
P. F. Strawson, Introduction to Logical Theory (Reprinted 1960), p. 196.
all the laws of the traditional (Aristotelian) system might be

satisfied. Strawson has discussed three possible interpretations
of the four propositions of Aristotle, and has shown that all
the traditional laws can be satisfied under the third. In the
context of Indian logic, we are primarily concerned with a
general (affirmative) proposition which is used as the major
premise. Richard S. Y. Chi has rightly pointed out (against the
common misinterpretation of many modern writers on Indian
logic) that the exemplified major' in the Indian variety of
syllogism is actually to be interpreted as 'an existential major
premise' 1 By 4an existential major premise' Chi has obviously
meant a general affirmative proposition where the non-empti-
ness of the class denoted by the subject term is presupposed.
The contrast between non-deviation (or pervasion) on the
one hand and ihz A-relatio.i of Aristotle on the other can be
brought about in the following way. Navya-nyäya says that
non-deviation of It from s holds when the following conditions
are satisfied :
i) h and s have non-empty presence-ranges;
ii) s is not unnegatable,i.e.,its absence-range is non-empty;
and iii) the absence-range of s does not intersect the presence
range of A.
And pervasion of s with/2 holds when :
i) the non-empty presence-ranges ois and A intersect;
and ii) if A is locatabie in the absence-range of any t, then tzfis«
Following Strawson, we can represent the three interpretations
of the A-relation and contrast them with non-deviation and
pervasion as follows:
xAy (1st inter.) ah = 0
(2nd inter.) ah = 0.
(3rd inter.) ab = 0
hNs (h+.s—=*0). A-f ^ 0 . .y-
sVIx (A+.s+Ô). Not ( A + . f - ^ O • t=s).
From the above it is clear that tiie third interpretation of the
A-relation is closer to the concept of non-deviation in Navya-nyâya
except for the fact that the latter requires an additional condition.
j
R.S. Y.Chi, Buddhist Formal Logic, pp. xxx-xxxi.
Problems of Philosophical Logic in Navya-nyâya. 125
Navya-,iyäya's insistence on the non-emptiness of the presence-

range or absence-range of properties pays dividend in the long
run, inasmuch as it makes the presupposition of a general
statement (involving non-deviation or pervasion) explicit. It
should, however, be noted that both non-deviation and perva-
sion are much stricter relations compared to the A-relation.
(This will also support our contention stated in the first section
of Chapter 1 of this book : The notion of vyäpti is not derived
from the A-relation of Aristotle.)
Second, let us note that most inferences studied in Navya-
nyäya try to locate a property (called sädhya 'inferable property'
s) in a particular locus (called paksa) with the help of another
property (called hetu 'reason' h). Thus, the predominant
inference-pattern of Navya-nyäya corresponds to what W. V.
Quine has called 'singular inference'.1 Hence, contrary to the
belief of some modern interpreters of Indian logic, the Navya-
nyäya inference is not exactly a Barbara, buta singular inference.
Chi has distinguished the standard Barbara from the singular
inference by calling the latter Barbara-A and the former Barbara-
B.2 Navya-nyäya, however, allows inferences corresponding to
Baibara-B, for it notes that the 'pervasion' relation is transitive
(cf., tad-vyäpaka-vyäpakasya tad-vyäpakatvam, tad-vyäpya-vyä-
pyasya tad-vyäpyatvani).
The Navya-nyâya restrictions on negation are instructive in
many ways. To recapitulate briefly the Navya-nyäya position on
negation : A property with an empty presence-range is called
fictitious or unreal. We have called it unlocatable. Negation is
viewed as an operation on real (non-fictitious) properties
generating further real (i.e., locatable but negative) properties.
Thus, a property with an empty absence-range is considered
unnegatable in this system. For, although such a property is held
to be real (since it is locatable) ,its negation would not generate
a real (i.e., locatable) property.
It is possible to use some notions of multiple-valued logic
under a special non-standard interpretation in order to represent
the domain of properties in Navya-nyäya. Using "property" in
the widest sense, we can construct the following tree to
represent the branching of properties (particulars).
1
2
W.V. Quine, Methods of Logic, p. 196.
R. S. Y. Chi, p. 13IT.
Property
real fictitious
positive negative unnegatable
In ordinary three-valued system, such values as T, F and / are

usually interpreted as "truth", "falsity", and "intermediate" (or,
"undecided" or "neither true nor false"). Let us propose a
completely different interpretation of values for the representa-
tion of the so-called real properties of Navya-nyäya.
Our proposed three values are : P (for "positive"), N (for
"negative") and U (for "unnegatable"). Now, we can have a
standard three-valued negation as the following table will
show :
w not-w
p N
N P
U U
This has the desirable outcome, viz.,
The presence-range of w = The absence-range of H\

The absence-range of w = The presence-range of w.
But, by negating an unnegatable we get only another unnegatable

(a fictitious one). Further, since combination of an unnegatable
with positive yields, for Navya-nyäya, a positive property (and
disjunction of a positive with an unnegatable yields an unnegat-
able), the corresponding tables for " A N D " and "OR" can be
constructed accordingly. But these tables will differ from the
standard tables in some respects.
The problem of negation of the partially iocatable properties
can be tackled in another way. Let us construe the negation of
a partially Iocatable property as both partially and wholly
Iocatable. Then, we can agree with the following fourfold
classification of properties :
Property
! _
locatable (real) uniocatable (unreal)
partially wholly both partially and

locatable locatable wholly locatable
We have seen, for example, that negation of the body-chair-
contact (a partially locatable property) yields a (negative) property
which is both partially locatable (in the same loci e.g., my body)
and wholly locatable in other loci. Here, using the notion of a
multiple-valued system, we can assign value 1 for the wholly
locatable, 2 for the partially locatable, 3 for those which are
both partially and wholly locatable, and 4 for the uniocatable.
Thus, we can construct a four-valued system with non-standard
interpretation of all values, and the negation matrix can be
written as :
w not-w
1 1
2 3
3 2
4 4
Finally, we may note that despite the above oddities, the
Navya-nyäya doctrine of negation is not very different from
what is usually called "classical" or standard negation. The
law of double negation, which roughly combines the law of con-
tradiction and the law of excluded middle, is always satisfied
by what Nyâya calls wholly locatable properties. (Only Raghu-
nätha, a commentator öf Gangesa, disputed this position which
I have mentioned above.) Thus, within the domain of
wholly locatable properties, our standard notion of negation is
not ''mutilated" (to use a term used by Quine).1
Since the notion of partial location creates difficulty in
interpreting negation in the standard fashion, Navya-nyäya
recommends the use of the technique of delimiters (cf., avac-
chedaka), by which a partially locatable property can be parsed
1
W. V. Quine, Philosophy of Logic, p. 7.
as a wholly locatable one so that negation can be given the

desirable standard interpretation. By declaring the unlocatables,
as unnegatable, Navya-nyäya solves another problem that may
possibly arise due to what is called "truth-value gaps" of such
propositions as : "There is no golden mountain" or "The son
of a barren woman does not speak" (Udayana's example).
Thus, inspite of the oddities encountered in Navya-nyäya
theories, attempt has constantly been made here, with regard to
negation, to follow what Quine has called the maxim of
minimum multilation.
§2.4 : THE PROBLEM OF UNIVERSAL 'PROPERTIES'

(KEVALÄNVAYIN )
We have seen in the previous sections that certain problems are

rather peculiar to Navya-nyäya. They arise in the discussion of
the Nyäya-Buddhist logical theories because of certain particular
doctrines that were already propounded in the tradition. The
concept of universal or ever-present properties is one such
doctrine. As I have already noted, these universal properties
cannot be equated with the notion of the universal class. For,
to be sure, knowabillty and namability are held to be non-
identical properties, although they are said to occupy the same
set of entities as loci.
That certain properties could be present in everything was an
idea that was already implicit in the 'wheel of reasons' (hetucakra)
of Oignâga and the theory of inference propounded therein. If
inference is the establishment of an object (or property in our
sense described before) through an already known object occurr-
ing in a subject-locus (which is again another object), then what
we have is a three-term operation. The first object is what we
prove (to be precise, whose presence or occurrence we prove)
by inference, and it is, accordingly, called sädhya. The second is
what proves (or to be precise, whose presence in the third object
as well as its relation with the first, proves), and, hence, it is
called sädhana or hetu. The third object is called the paksa. (In
this way of putting the matter, no distinction will be made bet-
ween "object" and "property", for, both are alike members or
items of the so-called universe of discourse.) Due to the above
reason, most modern writers have translated 'sâdfiyà* as probandum
Problems of Philosophical Logic in Navy a-nyây a 129
and 'hetii' as probans, and I have sometimes followed them.

However, obviously, the terms, probandum and probans, are not
at all familiar to those who today write and read philosophical
treatises in English. There is, therefore, some argument in favour
of retaining these terms, sädhya and hetu, in the English ver-
sions. My advice is this. If 'probandum' and 'probans' seem
almost as opaque as *sädhya' and *hetu\ one may very well leave
these two terms untranslated. In what follows, if the reader
finds the probans-and-probandum pair unacceptable, he may
substitute it by the sädhya-and-hetu pair.
Now, to sketch Dignâga's 'wheel or reasons', we can define
the class of agreeing instances ($apak?a) as the class a of all
objects x such that the probandum is present in*. In symbols:
*x(x possesses the probandum). Similarly, the class of disagreeing
instances (vipaksa) can be defined as class ß of all objects x such
that the probandum is absent from x. In symbols ;x—(x possesses
the probandum). Thus, any member of a is a sapaksa1 and any
member of ß is a vipaksa. Now, the probans as a property can
be present in all, some or no members of a. Similarly, the
probans can be present in all, some or no members of ß.
Combining these two sets of cases we get nine possibilities, of
which only two cases are cases of valid inference.2
The above is a rough sketch of Dignâga's system of logic as
iThere is, however, one difficulty here. The paksa, i.e., the subject
of inferential conclusion, possesses also the probandum, if the
inference is a valid one. Thus, paksa should also be a member of a. But to
avoid a petitfo principii, the paksa should be considered as being in
a twilight zone during the process of inference. In other words, we are not
sure whether the probandum is present in the paksa or not. Dignâga
defined paksa as: sâdhyatvenepsitah paksô viruddhârthânirâkrtah(quoted by
Vacaspati (1925) p. 273). In his system, paksa differs from sapaksa in that
in the latter the probandum is already established while in the former the
probandum is not yet established. See Dignäga (1968) Chapter III, verse
18cd and the vrtti; Kanakavarman (fol. 130a-130b), Vasudhararaksita (fol.
45a). I owe this information to my friend Prof. M. Hattori. Navya-nyâya
tackled this problem by its doctrine of paksatä. See Manikantha [1953] pp.
109-115; GANGESA [1926] pp. 1079-1176.
2
The second and the eighth in Dignâga's table are valid forms of inference.
In one case the probandum and the probans are equal in extension, in the
other case the class of probans is included under the class of probandum.
See next note.
found in his Hetucakradamaru (c. Ä. D. 500) l For our purpose

it is important to note here that one of the nine possibilities
demands that the probans be present in all members of a as
well as ß. Now, if a and ß are taken to be two complementary
classes in the sense that taken together they exhaust the whole
universe of discourse, then the probans in the above case will
be a universal property which is present everywhere. UDDYOTA-
KARA [i915] argued that in some cases of inference even our
probandum can be a universal, i.e., ever-present property. This
implies that with regard to certain cases of inference class ß may
be a null class, class a being a universal class. 2
In Navya-nyäya school, however, the concept of ever-present
property appears to have been taken very seriously. Navya-nyäya
writers like Vallabha, Manikanthaand Gangesa, rejected all such
definitions of vyäpti (invariable concomitance between the
probans and the probandum) as were based on the notion of
non-deviation {avyahhicaritatva), because such definition would
be inapplicable to cases of inference with an ever-present pro-
perty as the probandum. 3 The siddhäntalaksana 'conclusive
definition' of vyäpti is formulated in such a way that it becomes
logically applicable to all cases of inference including those in
which some ever-present property is the probandum. 4 I have pre-
sented my version of this definition of vyäpti in the previous section.
lf
This is a very short but illuminating manual of logic written by Dignäga.
It consists of a table of nine forms of inference" and only 17 explanatory
verses. The Sanskrit original is lost but the Tibetan translation is available.
2
I am using the term "everpresent" to translate the Sanskrit "kevalânvayiri"
although there is a touch of neologism here. "Omnipresent" and ''ubiqui-
tous" are not acceptable for this purpose because they express just the
opposite sense. The physical space, for instance, is omnipresent or ubiquitous
because everything exists in space, but it is not kevalânvayin. " Kevalânvayin"
means a property which is present in everything. "Unnegatable" (used by
Ingalls) may be all right inspite of the periphrasis, but I fear that it is more
suitable to be an adjective of "term" rather than of "property". I refrain
from using "universal property" in order to avoid confusion with "universal
class".
3
Thus see the previous section. Also see VALLABHA [1927-1934] (c.A.D.
1175) p. 500, line I; MANIKANTHA [1953] (c. A.D. 1300) pp. 45-46; and
GANGEéA [1926] p. 141. For an English version of the argument see INGALLS
[1951] pp. 61-62, 86 and 151.
4
See MANIKANTHA[1953] p. 62; and GANQESA [1926] p. 391. Ingalls gave a
rough idea about the structure of this definition (Ingalls [1951] p. 62).
First, an ever-present property, in the sense I am using it here,

cannot be identified with the notion of universal class for the
following reason. Using the convention of modern class logic
we can say that classes with the same members are identical.
Thus, eü>^ü>" may be written as a convenient abbreviation of
\x) (jc8ü)- = -xeo/)'. But a property or an attribute, in its non-
extensional sense, cannot be held to be identical with another
attribute, even if they are present in all and only the same
individuals.1 Properties are generally regarded by the Indian
logicians as non-extensional, inasmuch as we see that they do
not identify two properties like anityatva (non-eternalness) and
krtakatva (the property of being produced or caused), although
they occur in exactly the same things.2 In Udayana's system,
however, such properties as are called jäti (generic characters)
are taken in extensional sense, because Udayana identifies two
jäti properties if only they occur in the same individuals.3
Following the older tradition of the Nyäya school (noted in
Uddyotakara), Gangesa classified the types of inference as
follows4: 1. kevalänvayin, cases in which the probandum is
an ever-present property, 2. kevalavyatirekin, cases in which
the probandum is a property unique to the subject (paksa) so
that no agreeing instances are available, 3. anvaya-vyatirekin,
cases in which the probandum is a property present in some
examples but absent from others. The third type includes the
commonest forms of inference where both classes a and ß (i.e.,
sapakfa and vipaksa) are neither universal class nor nail class.
We shall be concerned here mainly with the first type, in which
1
See QUINE [1961] p. 107. Particularly significant is the remark of Quine
[1963] p. 2, "If someone views attributes as identical always when thöy
are attributes of the same things, he should be viewed as talking rather of
classes".
2
Those familiar with the Western logic may recall Carnap's excellent
illustration of the distinction between class and property : the class of
humans and the class of featherless bipeds are identical but the
property humanity is distinct from the property of being a featherless
biped. See CARNAP [1956] p. 12.
3
This is the significance of the condition called iulyatva(equipollence) found
in the list of six jäti-bädhakas (impediments to generic characters) mentioned
by UDAYANA [1885-1919] p. 33, lines 7-8. This will be discussed in the last
part of this chapter.
4
GANOESA [1926] p. 1326. See in this connection UDDYOTAKARA [1915] p. 46.
there cannot be any vipak$a> i.e., class ß is a null class, (see

before §1.4.)
Uddyotakara's example (taken from Dignâga's1) of anvayin
inference (corresponding to the first type here) was "Sound is
noneternal because it is a product {anityah sabdah krtakatvät)".
Here the probandum noneternalness will be a universal property
for those thinkers who hold to the doctrine that everything is
noneternal. Note here that the universe of discourse in this case
will include only noneternal things and hence class ß will be a
null class. Vacaspati cited a better example of this type of
inference: Vi§e$a (particularity) is namable because it is know-
able. In a slightly modified form, this example was accepted as
a paradigm in later Nyâya school: The pot is namable because
it is knowable.
Gange&i defined this kind of inference as one where there is
no disagreeing instances (yipak$a). Since everything in the
universe of discourse is (at least, theoretically) namable or
expressible in language, the property namabiiity (abhidheyatvd)
is a universal property and in no individual is there an absence
of namabiiity. To cite an instance where namabiiity is absent is
ipso facto to demonstrate that this instance is not inexpressible.
If, however, the opponent does not cite such an instance where
namabiiity is absent, but, nevertheless, believes it to be existent,
then as far as the logicians' inference is concerned it is as good
as non-existent, since inferential procedure demands the use of
language. The opponent may argue that although a disagreeing
instance in this case is not expressible in language, it can still be
a communicable concept in the sense that it is conveyed by the
meaning of some linguistic expression. But this would run counter
to the Nyâya thesis that there cannot be any instance which is
not namable.
Gangesa argued that from the opponent's viewpoint, the no-
tion of everpresent property invites the following paradox. If p
is asserted to be an everpresent property then one can infer
validly from this premiss that p is not everpresent. It is
*Dignäga anticipated the possibility of kevalänvayin inference and discussed
the issue in DIONÂGA [1965] Chapter III, verse 20. See Kanakavarman
(fol. 131a-131b; Vasudhararaksita (fol. 45a). But this created little problem
in Dignäga's system. According to him, «to be absent from vipaksa* can
conveniently be interpreted as 'absence of vipakfa*.
observed that with regard to each property (ciharma) it is legitimate

(according to the Indian theory) to assert that each property
is such that it is absent from something. Using quantificationai
notations and interpreting 'Fx' as *x is a property' and
'Oxf as cx is present in f we may represent this premiss as:
Now, since p is a property (which we have assumed to be ever-

present) , it follows (by universal instantiation and truth-func-
tional tautology) that p is such that it is absent from something.
In other words, the conclusion is \r&y) — Opy\ This implies that
there is an instance^ where/? (i.e., knowability) is not present.
Thus, our original assumption that p is an everpresent property
is contradicted.
Gangesa tried to answer this objection as follows. If the
property 'to be absent from something', i. e., the property re-
presented by the propositiona! function '('&y) — Opy\ is said to
be a property which is not absent from anything then the same
property becomes everpresent. If, however, this property (i.e.,
'to be absent from something') happens to be not present in
something x then that x becomes, in fact, everpresent.1 Let us
try to understand the implication of this argument. Let class w
be defined as x (jy)—Oxy. Now, if we assume that (coeco), it means
that the statement '(3J>)—O^f is false, i. e., '—(37)—O^f is true.
This implies that the class property of w is something which is
not absent from anything, i. e., it is everpresent. In an indirect
way, this means that w is a universal class. If, on the other
hand, we assume that o)£w then the statement '(3>0—Ovy* becomes
true. This means that there is something y from which the class-
property o) is absent. But to deny the class-property <a of some-
thing y means to admit y as an everpresent property. (Notice
that no type-difference of properties is being admitted here).
GangeSa's argument was exactly similar to this, although he
did not use the notion of class. Instead, he used his notion of
constant absence (atyantäbhäva) and its counterpositive-ness or
the absenteehood (pratiyogitä). A constant absence is arrived at by
hypostatizing the negation illustrated in the matrix 'there is no
1
This is however different from the famous class-paradox, viz., if the
class of all classes which are not members of themselves is a member of
itself, then it is not a member of itself. Note that Gangeéa's philosophical
motivation was different.
x in y or 'x is not present in y\ Thus, y is said to be the locus

which possesses constant absence of x, and x is said to be the
counterpositive or absentee of an absence which is present iri
locus y1. In fact, the constant absence of jcraay conveniently be
regarded as a class-property of the class which is defined as y (3*)
(x is not present in y). The mutual absence of* (illustrated by the
matrix 6yis not x') may likewise be regarded as a class-property
of the class which is defined (using usual symbols for identity and
negation) as y fax) (x^y). This interpretation of absences in
terms of the class-concept of modern logic gets indirect support
from the fact that Navya-nyâya, in most cases, identifies two
ab&eriees which occur in the same loci.2
Gangesa argued as follows. If the property of being the
absentee of a constant absence does not become the absentee of
any constant absence then the same property can be taken to be
everpresent. And if, on the other hand, that property is regarded
as the absentee of some constant absence say, the constant
absence of x in locus y, then the locus y where such a constant
absence resides becomes itself an everpresent property. The
upshot of Gangesa's argument is that if something x is a pro-
perty it does not necessarily follow that there is something else y
wherefrom x will be absent. This is so because there are ever-
present properties which will not be absent anywhere. An ever-
present property can now be defined as:
Dl. x is an everpresent property if and only if x is not the
absentee of any constant absence3
1
For the notions of counterpositive and constant absence, see INGALLS
[1951] p. 54-58. They have aiso been explained in detail in my book, see
MATILAL [1968] pp. 52-61, 94-95.
2
Thus the prevailing view of Navya-nyäya writers is that the two absences
which are samaniyata (equipollent) are identical.
3
MANIKANTHA [1953] gave a similar definition of everpresent property
•'"tatrâtyantàbhâvapratiyogî dharmah kevalanvayV' ' (p. 126). MAINJIKANTHA'S
three-fold classification of parämarsa corresponds to Gangesa's three-fold
classification of inference. But there seems to be a difference. Manikan{ha
seems to consider the nature of the probans while Gangesa, as he has
been explained by Raghunätha, considers the nature of the probandum.
Thus, an inference with an everpresent property as the probandum and
an ordinary (non-everpresent) property as the probans will be included,
under the first type, i.e., kevalânvayin inference. This is, at least, what
Gadâdhara seems to think. See Gadädhara's sub-commentrry on
To develop the next point in Gangesa's discussion we have to

understand what Navya-nyàya calls a non-pervasive (avyäpyavrtti)
property (see § 2.3). A property is cailed non-pervasive if and only
if it occupies only a part of the locus such that in remaining parts
of the locus there is the constant absence of that property. 1 Thus
properties like a pot or contact-with-a-monkey (in fact, almost
all properties except certain abstract ones like cow-ness), with
respect to their loci, such as a piece of ground or a tree, behave
as non-pervasive properties. Now, the constant absence of a
property p is regarded as another property, say qy which is present
in all things except where p is present. But the constant absence
of any non-pervasive property, it has been already argued, will
become an everpresent property simply because such an absence
is not only present in all loci except where the non-pervasive
property in question is absent but also in locus where the same
non-pervasive property is present. This follows from the very
definition of non-pervasive property. However, Gangesa pointed
out that as soon as we introduce the notion of delimitors
(avacchedaka) in our discourse the constant absence of a
non-pervasive property (say, a pot) can no longer be, strictly
speaking, an everpresent property. Thus, a pot cannot be said
to be absent from the locus ground if it is actually present there.
In simple language, this only means that right in the space of
the ground occupied by the pot there cannot be any constant
absence of the pot. Hence, such a constant absence is not ever-
present. For there is a locus, as we have just referred to, where
pot-absence is not present. Note that the notion of delimitor
here serves to dispel the vagueness of ordinary uses of 'locus
{adhikaranay and 'occurrence (vrtti)\
Another suggestion for constructing an everpresent property
can be given as follows. The ubiquitous physical space (gagana)
in the Nyâya-Vaisesika system of categories is held to be a non-
occurrent entity in the sense that it does not occur in any locus.
All entities of the Nyâya-Vaisesika system are properties (in the
sense that they occur in some locus or other) except entities like
Raghunätha's commentary on GANGESA [1926] p. 1327, lines 6-7: "...

kevalänvayisädhyaka-vyatirekihetoh keValänvayyanumänatayä...**
*Seo INQALLS [1951] p. 73. The meaning of "coatact-with-a-monkey" has
been explained by Ingalls, See also MATILAL [1968] pp. 53, 71-72.
the ubiquitous space. Thus, since there is no entity where the

space might occur as a property, the constant absence of the
space becomes everpresent. But this procedure eventually leads
to some difficulties. Technically speaking,the constant absence of
the space can very well be the absentee {praiiyogin) of another
constant absence, viz., the constant absence of the constant
absence of the space (which, according to Nyäya, is just identical
with the space itself). Thus, the above definition of everpresent
property cannot be applied to the constant absence of the space.
This eventually landed Gangeéa into the puzzling discussion of
the Nayva-nyäya school, viz., what constitutes the absence of
an absence ?l (See below §2.6, §2.7).
The constant absence of x is constantly absent from all things
except those that have no x. Hence, the constant absence of the
constant absence of x is present in all and only those things
where x is present. Applying the principle of identification of
the indiscernibles, Udayana, and following him Gangesa, iden-
tified the constant absence of the constant absence of x with x
on the gound that
A-. (y) (y has the constant absence of the constant absence of
r = 7 has x).
The mutual absence of or difference from pot is constantly
absent from all things that are called "pot", i.e., from all
things that have pot-ness. Thus, the constant absence of the
mutual absence of pot is present in all and only those things
that have pot-ness. Therefore, as above, one can identify the
constant absence of the mutual absence of pot with pot-ness
on the principle that
B. (y) (y has the constant absence of the difference from pot
• = y has pot-ness).2
Note that we are identifying here two class-properties on the
ground that the corresponding classes are identical by virtue of
their having the same members. This indirectly supports my
earlier suggestion that absences in many contexts can conveniently
be taken to be class-properties suitably chosen. Properties, in such
*See Ingalls [1951] pp. 68, 71-72.
a
See GANGESA [1926] p. 1350 : atyantäbhävätyaniabhävah pratiyogy eva;
anyonyäbhävätyantäbhävas tu pratiyogivrttir asadhärarto dharma Hi. See
UDAYANA [1926] Chapter HI, verse 2.
contexts, are used in their non-intensional sense. I shall discuss

these issues further in § 2.6, § 2.7.
Navya-nyäya, however, regards the constant absence of the
ubiquitous space as an everpresent property, and, accordingly,
Gangesa developed a technical sense of 'everpresent property'
by rephrasing Dl as follows:1
D2. x is everpresent if and only if x is not the absentee of any
occurrent (yrttimat) constant absence.
Although the constant absence of the space may be said to be
the absentee of the constant absence of the constant absence of
the space, the second absence is not occurrent because it is to be
identified with the space and the space is, by definition, not
occurrent anywhere. Properties like knowabiiity and namability
are not the absentee of any occurrent constant absence and hence
they can be called everpresent. This is one of many possible
interpretations of Gangesa's rephrasing (which was ambiguous
in the original). But, according to Raghunätha, this was just
Gangesa's way of being polite to the opponent (cf., abhyupaga-
mamätram). Actually, the constant absence of the constant
absence of the space cannot be identified with the space
because the above principle A is not applicable here. Since
in the Nyäya-Vaisesika system there is no entity which
has the space as a property, we cannot identify it with
the constant absence of the constant absence of the space
under principle A. The significance of the adjective "occurrent
(vrttimat)" was explained by Raghunätha as follows. When
something is said to be present in something else, it is present
there always through some relation or other. Thus, in speaking
of something as everpresent one should specify the relation
through which it is considered present everywhere:
D3. x is ever present through relation r if and only if r is the
delimiting relation of the absenteehood of some constant
absence and x is never the counter positive of such absence.2
To expose another logicaL difficulty involved in the notion of
everpresent property, we have to go back to the definition of
1
GANGEéA [1926] p.1353'.vrttimad-atyantäbhäväpratiyogitvaiiikevalänvayitvam
2
Raghunâtha comments on GANGESÀ [1926] p. 1354 : evatn ca yatsamban-
dhävacchinapratiyogitäkähhäväpratiyogitvam yasya tasya tena sambandhena
kevalänvayitvam iti sücayitum vxttimad Hi.
kevalänvayin inference (type 1 above). First, it is odd to say that

the probans does not reside in disagreeing instances, when there
is, in fact, no disagreeing instance. It is further odd to say that
there is no disagreeing instance, when "disagreeing instance
{vlpak$d)" is a mere indesignate or empty (nirupäkhya) term, for
one tends to argue that to make such denials meaningful our
acceptance of the existence of such non-entities is in order.
Väcaspati puzzled over this problem because, according to the
Nyäya theory, each negation, in order to be meaningful, must
negate a.real entity arid must denote an absence which usually
behaves as a property occurring in some locus. Thus, an absence
is always determined by its absentee (i. e., the negatum) on the
one hand and by the locus (ßdhärd) on the other. 1 Väcaspati
tried to solve the above puzzle by saying that the prudent
course is silence, i. e., not to deny or affirm anything (including
existence) of the non-existents. The denial sounds odd because
its contradictory, i.e., affirmation, sounds odd too. 2 Udayana
suggested a better method of answering such problems as we
have seen in § 2.2. According to him, a statement like
(1) "the rabbit's horns do not exist"
does not affirm or deny existence of anything, but simply expresses
an absence not of the rabbit's horns but of horns, an absence
which occurs in a rabbit.^ Note that having horns is a real
property such that one can meaningfully speak of its absence
(another real property for theNaiyäyikas). This analysis is related
to the epistemological theory of error of the Nyâya school, which
is technically known as ahyathäkhyätL The structure of this
analysis as I have already noted may remind one of B. Russell's
analysis of similar statements with his theory of description.4
Applying Udayana's principle of analysis, Gangesa tried to
make sense of the statements which make use of such indesignate
1
VÂCASPATI [1925] p. 172 : sadbhyâm abhâvo nirûpyate naikena satety
uktam.
2
See VÄCASPATI [1925] pp. 172-173.
3
Compare UDAYANA [1957] p. 331: kas tarhi sasasrngam nâstîty asyârthah ?
êase adhikarane visânahhâvo' stiti. In UDAYANA [1940] Udayana discussed
at length the example "bandhyä-suto na vaktä (the son of a barren woman
does not speak)*' (see pp. 64-73).
4See RUSSELL [1919] pp. 168-180
expression as "the absence of an ever-present property like know-

ability", viz.,
(2) "the absence of knowabiiity is not present in y" (a true
one).
(3) "the absence of knowabiiity is present in y" (a false one).
Note that "the absence of knowabiiity" is, as it stands, an empty
term and on par with "the present king of France". According
to Gangesa, we can rephrase (2) and (3) as :
(4) Knowabiiity is not the absentee of any absence that may
occur in v.
(5) Knowabiiity is the absentee of an absence which occurs in
y-
Here, (4) predicates of knowabiiity the absence of the property
of being the absentee of any absence occurring in y, while (5)
predicates of knowability the absenteehood of an absence
occurring in y. Thus, (4) expresses a trivial truth (see Dl before)
while (5) expresses a falsehood. Note that "an absence which
occurs in y" will denote a real absence occurring in the thing
substituted for *y* and that its absentee will be a real entity.
Hence the property of being such an absentee is also a real pro-
perty which characterizes certain things (viz., things which are
really absent from y) but not knowabiiity.
Gangesa used this method of analysis in order to make sense
of the doubt or uncertainty (samsaya) of the form "perhaps it is
knowable, perhaps it is not". This statement which expresses a
doubt can be said to be a meaningful statement if it is rephrased
in the above manner so as to avoid the use of any empty term-
complex such as "the absence of knowabiiity" (which refers to
nothing as in 3 above). Note that the second part of the statement
expressing doubt, viz., "it is not (knowable)", would have con-
tained |uch an empty term-complex, if it were straight-forwardly
analyzed in its logical form : it has the absence of knowabiiity.
It should be noted in this connection that, according to
the Navya-nyâya theory of inference, an inference (as. an effect, i.e.,
kärya) must be preconditioned by what Navya-nyâya calls paksatä.1
The condition of paksatä, according to the view of the old
Nyâya, involves in the presence of a doubt or uncertainty which
X
MANIKANTHA [1953] pp, 109-115 ; and GANGESA [1926] pp. 1079-1176,
should be expressed in the form "perhaps the subject possesses the

probandum, perhaps it does not". This postulate is based upon
the simple fact that we do not infer something which we already
know with certainty unless we wish to prove it again. Now, if
inference of an everpresent property like knowabiiity has to be
an actual event, it should be pre-conditioned by an uncertainty
of the form described above. Thus, the statememt which
expresses this uncertainty or doubt must be a meaningful
statement so that the required doubt {sarnsaya) may, in fact,
arise. Gangesa pointed out that when the second part of the
statement expressing doubt is interpreted as (5) above, we can
retain its meaningfulness and avoid using empty terms that
refer to nothing. 1
While studying Indian logic, scholars will find themselves
concerned with issues of two different kinds. The first are those
problems which are bounded by the Indian tradition itself, that
is, those which arise out of the peculiar yet rich tradition of
India's scholastic past. They are partly conditioned by the
Sanskrit language and partly by the fundamental concepts and
philosophical attitudes that Indian logicians inherited. The
second set of problems which we face here could be called
universal. They are, in essence, the very same problems faced
by the Western tradition, though often, because of the parochial
and tradition-bound interest of both sides, this fact has been
either ignored or badly misunderstood.
2.5 : INFERENCE AND CONCOMITANCE (VYÄPTI)
With the advent of Navya-nyâya methodology, the notion of

invariable concomitance or pervasion {vyäpti) became increasingly
the centre of interest of most Naiyäyikas in India. Even before
the time of Gangesa, there were numerous definitions of perva-
sion offered by different writers, the difference of one definition
from the other being at times very subtie and theory-bound and
1
Ia the above, I have chosen the term "absentee" and "absenteehood" for
pratiyoginandpratiyogitä, instead of "counterpositive"and "counterpositive!*'
used by D.H.H. Ingalls. I have followed Prof. Ingalls in my earlier writings.
But recently, a suggestion from a former student of mine, Dr. Arindam
Chakrabarti (see his Oxford thesis, Our Talk About Non~existents, 1982)
has persuaded me to accept this new terminology.
at other times trivial. Even a cursory glance at Gangesa's text

(he notes as many as twenty-one definitions ail of which he
rejects for some reason or other, and then goes on to give seven
or eight more definitions each of which he seems to accept) will
convince one how much interest was created regarding the
explication of the concept of pervasion. This interest continued
even after Gangesa with much gusto, and as a result, we find
numerous commentaries and sub-commentaries written particular-
ly upon this portion of Gangesa's text. It is no wonder, therefore,
that in the traditional seminaries of India today a beginner in
Navya-nyâya usually starts with one or two sub-commentaries on
some subsection of the Vyàpti section of Gangesa. Why do we
find this rather unusual interest in the definition of this concept
among the Indian logicians ? The history of logic in India
had its own unique nature of development. A brief review may
be enlightening.1
Early attempts to study the inferential relation can be found
in the Vaisesika sutras 3.1.8, and 9.182 as well as in the Sämkhya
school (viz., Sa$titantrd). The former speaks of four types of
inferential relation beginning with causal relation (in the Vaisesika
sense of the term 'cause'), while the Sämkhya speaks of seven
types of relation beginning with part and whole (rnäträmätri-
kabhäva)? It was felt at the time of Prasastapäda and Dignäga
that this type of classification was not exhaustive or could not
have been so. 4
Kumärüa used the term vyäpti 'pervasion' for the inferential
relation and tried to develop a sort of logic based upon the
relation of class inclusion and extension of terms. The pervaded
(vyäpyä), i.e., the middle term, is either co-extensive with (sama)
or included in (nyfma) the extension of the pervader (vyäpaka),
i.e., the major term. Inductive generalization, according to
Kumäriia, is based upon multiplication of empirical evidence,
x
For a brief outline of the relevant history, see § 2.2 before.
2
See Vaise$ika-sütra of Kanada (with the Commentary of Candränanda),
ed : Muni Sri Jambuvijavaji, Baroda, 1961.
3
Quoted by Väcaspati Misra in Nyäya-värttikatätparya-fikä, ed : Rajeswara
Sastri Dravid, Benares, 1925, p. 165.
4
See PraÈastapâda-bhâ?ya, Varanasi, 1963, pp. 503-505.
and an undiscovered or unnoticed 'associate condition' (upädhi)

may falsify the supposed generalization.1
Dharmakîrti provided a much neater scheme for classifying
pervasion. Pervasion or inferential relation may be based upon
identity relation, which is actually a relation of class inclusion
(viz., it is a plant, because it is an ivy). This is called identity,
because the two terms here refer to the same thing. Pervasion
may also be based upon causal relation, which should be an
inseparable relation (effect being inseparably connected with
its cause) between two different entities (viz., there is fire
there, because there is smoke). In fact, in the former case we
get what we may call today an analytic judgment as our
major premise, the whole argument taking purely a deductive
character. In. the latter case we get a synthetic judgment (in some
sense) as our major premise which combines two different entities
through causal relation. Whether Dharmakîrti envisioned a real
distinction similar to the one that we make today between
analytic and synthetic propositions is however very difficult to
say. The matter is not easily decidabie. Dharmakirti also noted
various other types of inseparable relation, which were, in essence,
ramifications of these two major relations combined with nega-
tion and contradiction. 2
This neat scheme of Dharmakîrti was severely criticized by the
Naiyäyikasas being insufficient on obvious grounds. 3 Some very
common forms of inference (e.g., inference of sunrise tomorrow
from today's sunrise) can hardly be assimilated under this neat
scheme. Triipcana, the Naiyâyika, thought it proper to define
pervasion as the natural (sväbhävika) relation. A natural relation
is explicated as an 'unconditional' relation (anaupädhika), a
1
Mimâttisâ-sloka-vàrttika, ed : Rama Sastri Tailanga Manavalli, Benares :
Chowkhamba, 1898, verses 4-6, of Anumäna-pariccheda.
a
See Nyaya-bindu, ed : Chandrasekhara Sastri, Benares, 1954, chap, 2 and
3. In an earlier paper, I suggested that Dharmakirti's distinction of tädä-
ttnya and tadutpatti may be compared with the analytic-synthetic distinction
in the Western tradition. E. Steinkellner disagreed with me on this point.
See his "On the Interpretation of the Svabhava-hetuh" WZKSO 18 (1974)
pp. 117-129. While Ï have modified my view on this issue due to his
criticism, I still believe that there exists a tension in Dharmakirti's system
here.
3
See Vâcaspati Misra, op. c//.,pp. 158-164. See also Srïdhara's Nyàya-kandall
(comm. on Prasastapâdabhâsya, op. cit.), pp. 492-502.
Problems of Philosophical Logic in Navya-rsyâya 143
relation which is uncontaminated by any 'associate condition'

npädhi.1 Udayana favoured a modified version of unconditiona-
lly as a definition of pervasion.2 Vallabha registered a note of
caution. For him, pervasion means accompaniment of all the
cases of middle term with the major term. The differentiating
mark (laksana) of pervasion relation is, however, the absence of
upädhi 'associate condition.' An 'associate condition' is defined,
according to Vallabha, as the property which accompanies all
cases of the major term i.e., what is to be inferred (sädhya), but
only some cases of the middle term i.e., the hetu or the 'reason'. 3
By the time Navya-nyâya method was developing and greater
attention was being paid to the precise formulation of the defini-
tion of different concepts, there were several alternative definitions
of the concept of pervasion as well as several alternative form-
ulations of the definition of upädhi 'associate condition' (which
was well recognized by this time as a negative mark of pervasion).
Thus, Manikantha Misra (who preceded Gangesa) mentioned as
many as eleven different definitions of pervasion, each of which
was rejected by him on various grounds. He accepted what seems
to be a modification of his eleventh definition.4
Gangesa's twenty-nine different formulations of the definition
of pervasion (twenty-one of which being unacceptable and
eight being acceptable to Gangesa) were largely based upon
Manikantha's and Sasadhara's discussions of pervasion.5 The
following are the eleven alternative definitions of pervasion
found in Manikantha: Pervasion 1. is 'any kind of relation'
sambandha-mätra (view of Bhüsanakära =» Bhäsarvajna ?),
2. 'non-deviation' avyabhicaritatva (found inSrîdhara's Nyäya-
kandali and in many other places), 3. 'the property of not
occurring without (the other)' avinäbhäva (Dignâga, Prasastapâda
and many others), 4. 'natural relation' sväbhävikasambandha
1
Trilocana's view has been quoted by Jnânasrïmitra. See an illuminating
discussion of this point in Gerhard: Oberhammer's article "Der Sväbhävika-
sambandhüy ein Geschichtlicher bei trag zur Nyäya-logik," Wiener Zeits-
chrift für die Kunde Süd-und Ostasiens, Band Vlll, 1964, pp. 131-18Î.
2
Ätmatattva-viveka, ed : Dhundhiraja Sastri, Benares : Chowkhamba, 1940,
z
p. 403.
NyäyaIilävati, ed : Dhundhiraja Sastri, Benares : Chowkhamba, 1934, pp.
496-502.
^Nyäya-ratna, ed: V. Subrahmanya Sastri and V. fCrishnamacharya, Madras,
5
1953, p. 42-61.
Saéadhara's Nyâya-siddhânta-dipa was printed in The Pandit journal. A
new edition of this book has been prepared by the present author.
(Trilocana), 5. 'relation of the effect to its efficient cause' nimitta-

naimlttika-bhäva (the Sämkhya view ?), 6. 'identity' tädätmya
(Dharmakîrti), 7. 'relation of the qualifier to the qualified'
visi?ta-vaisi$tya (?), 8. 'the property of being the counter-
positive of an absence which (absence) is pervasive of the
absence of the major term' sädhyäbhäva-vyäpakäbhäva-prati-
yogitva, 9. 'accompaniment of all cases of one term with the
other term' kärtsnyena sädhana-sädhya-sahabhäva (Valîabha),
10. 'unconditional relation' anaupädhikasambandha (Udayana
and others), 11. 'co-occurrence with something that is never the
counter-positive of a constant absence which (absence) is co-
occurrent with the other term (the hetu) in the same locus'
sädhanatväbhimatasamänädhikaranätyantäbhävapratiyogisämänä-
dhikaranya.
Gangesa first takes the second definition of Manikaritha's list,
viz., non-deviation, and gives seven different formulations of
this definition and rejects each of them mainly on the ground
that it fails to include the pervasion relation existing between
two 'ever-present' kevalänvayin properties, such as knowability
and namability.1 An incidental discussion is introduced here on
the point whether the absence of 'unactuaiized possible' entities
could be regarded as an ever-present property or not. I have
noted the question already in the previous discussion. I shall
discuss certain related issues in the last chapter (§ 5.4 and § 5.5).
This is followed by four different ramifications of the definition
of pervasion, some of which can be located in Sasadhara's
Nyäyasiddhäntadipa. Then Gangeéa examines two different
formulations of the notion of unconditionally (definition 10 of
Manikantha) and four different formulations of the pervasion
relation by making use of a universal quantifier (krîsna, yâvat;
definition 9 above). Next we find brief mention of definition 4
(sväbhävika-sambandha), definition 3 (avinäbhävd) aad definition
1 {sambandha-mätrd) from the above list.
The siddhänta-laksana, i.e., the definition acceptable to Gaii*
gesa, is only a modified version of Manika$tha's final definition.
This formulation takes care of the cases where the major term
is such that both its absence and its presence can be truly
asserted of the same locus (i.e., avyäpyavrtti-sädhyakä). A similar
definition is also found in the list of Sasadhara. This definition
1
See the previous section §2-4.
does not use any universal quantifier, but makes use of a generic
absence, i.e., an absence whose absentee is qualified by a generic
property. Gangesa inserts here a discussion to show how and
why the generic absence must be regarded as separate from the
integration of specific absences. Gangeéa next offers three
different formulations of the definition of pervasion where no
use of the notion of generic absence is made. Gangeéa finally
accepts definition 10, i.e., 'unconditionally' as pervasion, as an
alternative definition, and gives four acceptable formulations of
this definition. This is followed by three different formulations
of the notion of 'associate condition' upädhi.1
§ 2.6 : DOUBLE NEGATION IN NAVYA-NYÄYA
The phrase 'double negation' immediately calls to our mind

what is now known in the Western logical tradition as the
classical law of double negation. This law is actually derivable
from the law of excluded middle and the law of non-contradic-
tion, and it is maintained as a law under the 'classical' or
standard account of negation in logic.2 The logical law of double
negation, however, attracts more attention than usual when a
proposal is made to suspend or restrict it. Thus, it is well-known
in Western logic that a non-classical or non-standard account
of negation should be proposed for constructing non-standard
multiple-valued logical systems. W. V. Quine has called them
'deviant logics' and described the non-classical negation as a pro-
posal for the 'mutilation'of our standard sense of negation.3
In the context of Indian logic, the phrase 'double negation'
is, however, reminiscent of an important and very controversial
x
An analysis of Gangeéa's Vyâpti section has been done in English by
C. Geokoop in his The Logic of Invariable Concomitance in the Tattva-
cintamani. This book, inspite of all its defects, should be read and a
better analysis in modern terms should be undertaken as a long-
term project by scholars today. On upädhi* 'associate condition' see
M. K. Gangopadhyay, "The Concept of Upädhi in Nyaya Logic,"
Journal of Indian Philosophy. 1, 1971, pp. 146-66.
2
The ordinary law that is expressed in Latin as duplex negatio affirmât has
its counterpart in Sanskrit, dväbhyäm nUedhäbhyäm prakrty-arthadärdhya-
bodhanam% which means that two negatives lead to a forceful assertion of
the original thought. See Nägeia's Paramalaghumanjüsä, p. 71.
3
See Philosophy of Logic, Ch. 7. We have touched upon this point in § 2.3.
doctrine of Dignäga—the anyäpoha *the exclusion of the contrary*,

as a possible nominalistic substitute for universals that are reified
as meanings of general terms. Dignäga's proposal was met with
scepticism in opposite camps, and the followers of Dignäga
apparently complicated the issue by proposing various tentative
solutions to the problem. Perhaps, the relevant problem cannot
be solved adequately unless one proposes (as Prof. H. Herzberger
has done)1 another non-standard account of negation. My own
feeling is that to make sense of the use of negation in Buddhist
philosophy in general, one needs to venture outside the pers-
pective of the standard notion of negation. This is another way
by which, I believe, one can provide an alternative interpretation
of the so-called Mädhyamika tetralemma (catuskoti).2
In Navya-nyäya the problem of 'double negation' arises in an
entirely different context. The treatment of the problem is, as
may easily be understood, essentially conditioned by the peculiar
Nyâya doctrine of negation. It is obvious that the Navya-nyäya
writers were not at all perturbed by the limitations of the two-
valued logic, nor did they worry much about the so-called
Vedäntic 'challenge' to the law of excluded middle. Moreover,
the Buddhist apoha theory was not directly relevant to the problem
of'double negation' in Navya-nyäya. There is, however, a minor
detail in the Navya-nyäya theory of double negation, where one
might detect a reaction to the apoha doctrine of the Buddhist.
I shall mention it in a note towards the end of this section.3
Navya-nyäya, as I have shown elsewhere,4 is not in favour of
the affirmative-negative dichotomy of propositions. Instead, it
speaks of contradictory pairs of qualifiers, viz., blue-colour and
the absence of blue-colour, or pot (i.e., pot-presence) and pot-
absence. Thus, the contradictory pairs of propositions (or
1
See 'Double Negation in Buddhist Logic,' Journal of Indian Philosophy
3(1975), pp. 1-16.
2
For example, I have argued once that the Mädhyamika advocates an
extreme form of negation where we have only a denial without any 'com-
mitment' aspect. The commitment aspect of a denial is usually derived
from its presupposition. See Epistemology, Logic and Grammar in Indian
Philosophical Analysis, p. 164.
3
See note l , p . 153.
4
See Matilal, The Navya-nyâya Doctrine of Negation, pp. 92-3. See also next
section § 2.7.
qualificative cognitions, to use strictly the Nyäya terminology)

are formulated with such contradictory pairs of qualifiers. But
each qualificative cognition is seen as attributing (instead of
affirming or denying) some property or qualifier to a qualificand.
Nyäya admits mainly two kiiids of negation, relational absence
and difference, and this twofold division is rather intriguing.1
Both are called abhäva or absence in Nyäya, and Professor
Ingalls, accordingly, translated them as constant absence (atyan-
täbhäva) and mutual absence (anyonyäbhäva). For the sake of
convenience I shall use 'absence' for the former and 'difference'
for the latter. Again, for convenience, I will use c ~ ' for the first
and '—' for the second. The twofold division of negation seems
to be based upon an implicitly maintained2 twofold division of
propositions : viz., those expressing qualification and those
expressing identity. In the first case, a qualifier qualifies a
qualificand and by negating it we get an absence of that
qualifier (which is regarded as another qualifier) qualifying the
same qualificand.
For example,
"The ground is qualified by pot-presence,"
and its negation,
"The g. is qualified by pot-absence."
But in the second case, two singular terms are joined by the i s '
of identity and by negating it we get a proposition of the form
l
ajfzb\ The typical Nyäya example is: "(4 or the) pot^£(a or the)
cloth." This is an example of negation being construed as
difference. Nyäya, however, will also recommend rephrasing of
this sentence so that we shall get a proposition wherein cloth-
difference will appear as the qualifier qualifying the qualificand
pot :
"The pot is qualified by cloth-difference."
x
To simplify matters for discussion, I shall ignore the four-fold division of
negation, viz., prior absence, destruction, constant absence and difference.
2
I use the term 'implicitly' here because the two-fold division of propositions
based on the two-fold interpretation of the copula 'is' has no explicit
parallel in Navya-nyäya. The problem did not arise explicitly probably
because the copula is never explicitly used in Sanskrit. Nevertheless, the
point about attribution and identity arose in connection >vith the Nyàya
analysis of negative propositions.
It is easy to see that absence of blue-colour and difference from

blue things are closely related properties. For, the absence of
blue-colour qualifies only the non-blue things, and since each
non-blue thing is different from what is blue, difference-from-
blue-things will qualify the same non-blue things. In Nyâya
terminology, however, the two negations, absence of blue-colour
and difference-from-blue-things, are as properties, samaniyatd
'equi-locatable' (see Section § 2.8). In other words, the class of
loci of one is equal to the class of loci of the other. From this,
Nyäya generalizes :
The absence of any property is equi-locatable with the differ-
ence-from-the-possessor-of-that-property: ~;tis equilocatable
with- (the possessor of x).
The Naiyâyikas generally agree that because of this equi-locat-
ability of the above two properties one should be held as equiva-
lent to the other. This equivalence should not be confused with
equality or identity, but should be seen as a relation of mutual im-
plication. In other words, presence of one in a locus implies the
presence of 'the other, and vice versa. Some Naiyäyikas will be
in favour of identifying the absence of x with what is called the
difference from the possessor of x just because of their equi-
locatability. This obviously amounts to an 'extensional' inter-
pretation of the (negative) properties. But the prevailing orthodox
view was against such a move.
Udayana, as we have already noted, formulated the law of
double negation as: 1
"The negation of the negation of x is identical with x."
In the same context, Udayana has insisted that one cannot
negate what is not there, nor can the locus of the negative property
(absence or difference) be a fiction. In other words, both the
praiiyogin or negatum and the anuyogin or the locus which the
negative property is supposed to qualify should be real entities.
This is rather an important feature of the Nyäya concep-
tion of negation, and we have already discussed it. 2 Let us
concentrate on the so-called double negation.
Navya-nyâya gives the following versions of double negation :
x
See Nyäyakusuniänjali, Çh. Ill, verse 2cd.
2
See section § 2.Î, § 2.2.
I The absence of the absence of #=;*:; svätyantäbhävätyan-

täbhävasya svätmakaivam ; ~ ~ x =» x.
II The difference from what is different from that which is
delimited by x — x; svävacchinnabhinnabhedasya svätmaka-
îvam.
III The difference from what has absence of x=x; sväbhäva-
vadbhedasya svätmakatvam.
I shall discuss them one by one.
The first version is almost universally accepted, and it is a very
well-known formulation in Nfavya-nyäya. I say 'almost', for there
was some controversy over it, which Professer Ingalls has nicely
described in his book. 1 Most authors accept I as a valid interpre-
tation of Udayana's comment on double negation. I is, however,
justified as follows : Whatever may be the locus of x must also
be the locus of the absence of the absence of x9 for as long as x
is known to be there x-absence is not cognized to be there,
and, hence, the following statement will always hold true 'there
is absence of the absence of x there.' Note that this simply
amounts to the class of loci of x being identical with the class
of loci of~ ~x. Thus, from the equi-locatability of x with ~ ~ x
we can derive their equivalence, and this equivalence can be
thought of as a relation of mutual implication in the sense I have
already noted above. But remember, in the previous case, the
prevailing Navya-nyäya opinion refused to equate the two proper-
ties; absence of x and difference from what possesses x, were treat-
ed as distinct. In the present case, however, the prevailing opinion
is in favour of identification of * with ~ ~ x . It sounds quite
anomalous. Raghunätha, however, in his intensionalist vein,
argued against the identification ofx with~ ~x. For, he thought,
the notion of negation conveyed by the second can never be
conveyed by the first, and, hence, it is difficult to think of them
as non-distinct. Raghunätha's point may have the following
implication. There is cognitive difference between how i{x" and
< c ^ ~ x " present their referents. In other words, the mode of
presentation of the property by "JC" varies from that of by
« ~ ~x". The first presents it as qualified by x-ness, while the
second as qualified by a complex of properties of which not only
jc-ness is an element but also some negative facts, absences and
*See Ingails, p. 68-9.
absenteehood, are elements. Hence, Raghunätha says, we cannot

strictly equate them. I would ask modern Fregean philosophers
to judge whether Raghunätha was talking about what we call
today Fregean Sinn.
The second version, II, has not been universally accepted in
Navya-nyäya. Here to simplify, we may replace "that which is
delimited by x" by "that which has x or is qualified by x*\ or
only "the possessor of x". Using '—' for 'difference' and 'Locus'
for 'possessor or that which possesses', we can represent the
complicated expression as follows :
(parentheses will be used to indicate the scope)
H. «—(Locus of (—(Locus of x))) = x."
The argument that is usually given to justify II is exactly
similar to that for I above : Wherever x is present it is true to
say of that locus that it is NOT DIFFERENT from what is
delimited by x. In other words, the two properties are equi-
locatable. And we have already seen that although equi-
locatability is not a sufficient condition for identity, it is at
least a necessary oire. We shall presently see the view of
those who are sceptical about II.
The third version is equally met with scepticism. It can be re-
written, using the convention of the previous paragraph, as follows :
IE. "—(Locus of(~x))«x."
The reason for justifying III is also similar to that for justifying
II or I : Wherever x is present as a property, it is true to say of
that locus that it is DIFFERENT from what has absence of x.
So, we have, as before, only a necessary condition for identity,
not a sufficient one.
Those who hesitate to accept II or III not only point out that
equi-locatability is not a sufficient condition but also formulate
some counter-examples where even the equi-locatability condition
will not be fulfilled. The major problem arises in connection
with those properties which are called avyäpyavrtti 'non-perva-
sive.' See § 2.3 above.
Contact or contact-with-a-particular-monkey is said to be a
typical example of this kind of properties. Such a property (I shall
represent it by 'c') is, by definition, co-occurrent with its absence
in the same locus. In other words, as long as one can say that
a tree k in contact with a monkey at its top but not so at the
bottom, Nyâya will claim in such cases that the tree is qualified
by c as well as by the absence of c. Now, if V is substituted for
'x' in III we get an obvious falsehood. For, the same tree is
also regarded as a locus of the absence of c (by Nyäya) and,
hence, "difference-from-the-tree" cannot possibly qualify the
same tree which is qualified by c. We cannot say of the tree that
this is NOT what possesses the absence of c. One may wonder
that II can possibly be salvaged even when one uses V for 'x\
For, it is quite possible to say of the tree that it is NOT DIFFE-
RENT from what has c. But Gadâdhara in many places disallowed
the identification of c with the DIFFERENCE from what is
different from the locus of c. For, it is argued, although the tree
is qualified by c (so that it is a locus of c) it is not (known to be)
DELIMITED by c. (Roughly, a delimiting property has to be
pervasively occurrent, not non-pervasively occurrent.) Thus, the
original version of II uses the notion of delimitation. Hence, it is
concluded that both II and III fail to be satisfied by a non-per-
vasive property, and this should be regarded as a sufficient argu-
ment against their acceptance as rules of identification, i.e., as ver-
sions of the law of double negation (in Udayana's sense, of course).
Some might have the following rejoinder : The above counter-
examples may require us to narrow down the scope of application
of II and III; they may not force us to reject them totally. Thus,
one may hold without embarrassment that as long as we speak
in terms of pervasive properties, II and III are acceptable. If,
however, such delimitation of the scope of application is not
allowed, then, so the argument goes, it would be difficult to
maintain even I as acceptable. For, using the Nyâya ontologicaî
scheme, it is not hard to find a value of x which will render the
status of the universal acceptability of I dubious. For example,
gagana or the sky, in Nyâya ontology, is regarded as a non-
occurrent entity. (In other words, while most entities qualify or,
are occurrent in, some entity or other, the sky does not qualify,
or is not occurrent in anything.) Now, if we substitute 'the sky'
for 'x' in I, we shall face a peculiar Nyäya problem, which will
render I invalid ! To wit : since the sky is not occurent in any-
thing, its absence will be occurrent in everything including
itself (for, as long as the sky is not occurrent in itself, the
absence of the sky can be said to be occurrent in the sky). Now,
since the absence of the sky is occurrent in everything, the absence
of the absence of the sky will not be occurrent in anything. This

amounts, in Nyâyà theory, to saying that the absence of the
absence of the sky is fictitious! Therefore, I does not work in
such cases (Q.E.D.)!
In other words, if something is occurrent in everything includ-
ing itself, it is an 'ever-present' property (in the technical sense
defined by Nyäya, see § 2.4 above), and absence of such a
property cannot be instantiated.
The point of this rejoinder, stated simply, is this. If we reject
II and III because of their lack of universality, we should be
prepared to reject I also on the same ground. But many Navya-
nyâya authors, such as Mathuränätha, reject this rejoinder and
maintain I as an acceptable version of Udayana's comment on
double negation. They also stipulate that the absence of the
absence of the sky is identifiable with the sky, and hence I has
universal application. In fact, if we accept the notion of null-class
or empty class (of modern Mathematical logic),—a notion that is
not easily accepted by Navya-nyâya—then the above stipulation
is easily defensible. For remember, the sameness of the class of
the loci of one with that of the other was a necessary condition
for the equality of the two properties in the given context. Thus,
one can say that both the sky and the absence of the absence of
the sky are occurrent in such loci as are members of the null-
class. It should also be noted that as long as the absence of the
absence of the sky can be identified with the sky, a real entity,
the orthodox Nyäya will never accept the absence of the
absence of the sky as fictitious !
Let me conclude after discussing an incidental question which
is often asked in this connection. If absence and difference are
the two main kinds of negation, then one may expect the
following combinations of them :
(i) ~ ~ x (using * ~ ' for 'absence').
(Ü) x («_» for 'difference').
(iii) - - * .
(iv) —~x.
Navya-nyâya authors generally discuss (i) and (ii) but ignore (iii)
and (iv). I have already discussed the meaning of combination
(i);There is a general agreement among Naiyäyikas regarding the
status pf (ii), viz., absence of the difference from xis identifiable
Problems of Philosophical Logic ia Navya-nyâya 153
with an inseparable property of xy i.e., #-ness (may be9 the

essential property of x).1 Ingalls has given the Nyâya reason for
this identification,2 and, therefore, I shall not go into it here. One
may only add that the argument in its esence is bassed on the
condition of equi-locatability as before. This identification,
however, leads to a difficulty which has already been discussed
by Ingalls.8 It may be remarked that if we use suitable paren-
theses to clarify ambiguity and keep our rule I operating, we
would not face the anomaly which apparently Mathurânâtha had
to face (and which Ingalls, accordingly, pointed out). For example,
we can write :
(a) ~ ( ~ — x)= ~xx (where 6xx stands for 'jc-ness').
(b) „„(-x)=-x.
Note that the parentheses mark out the distinction between the
left-hand side of (a) and that of (b). Thus, abbreviating 'absentee'
or 'negatum' (pratiyogiri) by 'd' we get :
(c) d of ~ ( ~ - x ) = ~ — x = x v
Now, if we stipulate that rule I should be first applied to cancel
the uninterrupted succession of two absences before we apply the
operation with d9 we can easily obtain :
(d) doî ~ ~ ( - ; c ) = dof (-*) = *.
Thus, we can avoid the undesirable consequence of accepting
"a double nature" (ubhaya-rüpatva) of~— x, that ofx and that
Why does Navya-nyâya ignore a discussion of (iii) and (iv)?

The reason will be obvious if we think about the meanings of
these two combinations. Suppose, 'blue' (i.e., 'blue colour') is
1
This is an interesting result from the point of view of the Buddhist apoha
doctrine. Under one interpretation, 'agovyâvfttV means the absence of the
difference from the cow, and this is exactly what the Buddhist would like
to substitute for the universal cowness. Thus, instead of saying *cow* means
cowness, one may say that 'cow* means the absence of the cow-difference.
2
See Ingalls, p. 149; Mathurânâtha uses this principle and says, "svàvacc-
hinna-bhinnabhedarüpasya...."'
3
Ingalls, p. 71-2.
4
While Mathuränätha admitted that sometimes** — x=*x and sometime
~ — *=*.x-ness, he was probably confused about the fact that these
two results are different because they are derived in different ways.
substituted for '*' in (iii). Now, '— blue' will refer to difference-
from blue-colour which is a property shared by all and only
non-blue things. Let us call this property eulb.1 Now, there are
many other properties (and things) besides eulb, and all of them
will belong to the class defined by the property difference-from-
eulb, that we may express by c —eulb' (i.e., by *—-blue'). In other
words, the last property cannot be a unique property of only
blue things. Hence, it cannot be identified with blue, for, the
necessary condition, equi-locatability, is not satisfied. Similarly,
it can be shown that 'difference from the absence of blue'
(derived from (iv)) expresses a property which is shared by every-
thing except the absence of blue. Hence, from Nyâya point of
view, these two properties are not very interesting.
Let me emphasize that the above has not been an exercise in
what may be called 'comparative logic'. My attempt here has
been an humbler one. I have tried to understand (myself) and
explain some of the issues which Navya-nyäya (and by the same
token, Indian logic in general) thought important, interesting and
relevant to their way of philosophizing. In my exposition I have
made use of certain modern (logical) terms. Some of these terms
have acquired very specialized and technical meanings in the
writings of modern logicians. I have, however, taken some liberty
with these terms (such as double negation and equivalence) for
the purpose of my exposition and translation. I think their
intended meanings will be clear from the context of my discussion,
and I sincerely hope that such uses will at least help our under-
standing of the problems of Indian logic.
x
The idea of'eulb' is from a comment of A. J. Ayer, 'Negation', p. 51-52
(See his Philosophical Essays, 1963 edn.) One note for caution : Taking
for granted the object/predicate distinction, as is normal In first order
logic, and using 'Fy for 'pot\ we have to say thlflo say that x is —F (in
our sense of ' — ') is not to say that (y) (Fyz^y^x). Rather to say that x
is—F is to say that x={y : NOT Fy}. Note that the sense of «—F' is distinct
from the sense of 'Not F' or 'is Not F' as used normally in predicate
calculus. Thus it is that if eulb is taken to be the class of non-blue things,
then things (classes, individuals, or properties) that are distinct from eulb
will form the class of -eulb (or - - blue), i.e., to say that the —eulb is the
class of things different from the class of non-blue things.
This note owes its origin to a criticism of Mr. Christopher Peacocke of
All Souls College, Oxford, who very kindly read my typescript, and whom
I wish to thank here.
On the other hand, I deplore the view that maintains that

Indian logic is too exotic and too remote from what we under-
stand by logic today to be relevant to our modern style of
philosophizing. For, I firmly believe that the study of Indian logic
is almost as much relevant today as the study of Aristotle, Plato,
and the Scholastics. These will certainly help us to gain insight
into the problems of the philosophy of logic. It is also important
to understand the exact nature of the questions the Indian
logicians asked themselves, and thus to ask ourselves why they
fail to solve certain problems satisfactorily. I wish to end with a
quotation from Professor Ingalls on this issue.1
There are striking similarities between Western symbolic
logic and Navya-nyäya. They deal with similar problems and
are often guided by a similar spirit of inquiry... (but) the
complexities of the one tradition are not readily translatable
into the other. For all the similarities of subject-matter and
spirit, the structures of symbolic logic and Navya-nyäya
are radically different. They differ not only as wholes; the
difference permeates each detail of the architecture. *
Let me add also that such details and their differences are
always philosophically interesting and certainly important.
§2.7 : THE 'DIFFERENCE' OF DIFFERENCE
Navya-nyäya accepts, as I have already noted, two types of

negation : absence and difference. This means that the meaning
of the negative particle NAN is construed, depending upon the
context, either as implying the absence of a property in a
certain thing or a place, or as denial of identity between two
objects. This distinction reflects, as I have noted in the previous
section, the Navya-nyäya understanding of the underlying
difference between two types of cognitive event. What is
grasped by a cognitive event is usually expressed in a sentence.
But the sentence cannot fully represent the underlying
structure as well as the force of the cognitive event. In other
words, some parts of what we may call the structured content
of the cognitive event are not directly visible in its conventional
linguistic representation, i.e., the sentence, although it is only
*See Preface to D.Ct Guha's book, Navya-nyâya system of Logic, p. xviii.
the sentence that we have accessible to us for our analysis study,

introspection, and eventual reconstruction of the structure of
the initial cognitive event. For, these events, as the Nyâya
insists, are fleeting states, but to the extent they are verbalizable
they leave their imprints on the language.
In analysing a cognitive event, Navya-nyäya invariably
analyses the sentence that is supposed to verbalize it-—a met-
hod that I have been following here. In fact, Navya-nyäya
derives the structure of the cognitive event involved, from the
verbal, i.e., linguistic representaion of it. Consider :
(1) "This is white."
(2) "This is a cat."
For the sake of simplicity, we are dealing with simple, declarative
sentences. According to Navya-nyäya, (1) represents a cognitive
event which identifies an object x as "this" and ascribes a
property (called its qualifier) to it, viz., the property of having a
particular white. The object so identified is called the qualificand
in that cognition in relation to the property ascribed, i.e., the
qualifier. Even the object x is not grasped by the cognitive event
as such (without qualification), as the expression "this" indicates
to the contrary. Navya-nyäya says that "this" has a sense, viz.,
that of being present in front of the cognizer, and, hence, the
object x is also qualified by this sense. For simplicity again, I
shall ignore this and other complexities. The cognitive event
connected with (1) would have the following structure, according
to Navya-nyäya.
Cl. The qualificand referred to by "this" is qualified by the
property of having a particular white texture.
Accordingly, (2) has :
C2. The qualificand referred to by "this" is qualified by the
property of being a cat.
But (2), as it stands, is ambiguous, for it can also represent
another cognitive event having a different structure.
C12. The qualificand or the object referred to by "this" is
identified with the qualificand cat that is qualified by the
The linguistic representations of both C2 and C12 will be the
same, but they are two distinct cognitive events. (In modern
terminology, one may say that the above-mentioned ambiguity
of (2) is reflected, at least in part, in the suggested alternative
ways of symbolizing the structure t>f (2), e.g., "Fa" or "a=6").
A Navya-nyäya theory insists that a säbdabobha or a hearer's
cognition or knowledge of (2) would always be like C12,
while the speaker's cognition could be either (G2 or C12). (see
also § 5.4 and § 5.5).
In its affirmative form, the ambiguity of (2) is seldom resolved.
But, as soon as negation is involved, Navya-nyäya resolves the
ambiguity by talking about two types of negation, absence and
difference, which the cognitive episode concerned is said to have
taken into account.
(3) "This is not white."
(4) "This is not a cat."
If we follow the previous model of Cl and C2, we shall have
cognitive events having the following structures :
C3. The qualificand, etc., is qualified by the lack of being
white, i.e., absence of being white.
C4. The qualificand, etc., is qualified by the absence of the
But there is a strong feeling in the utterance of (4), that the
cognitive event involved is given by.
C14. The object referred to by "this" is different from, i.e.,
non-identical with, the object or qualificand cat that is
qualified by the property of being a cat.
C4 recognizes an absence as the qualifier (i.e., the main quali-
fier) while C14 recognizes a difference as the main qualifier.
Very roughly this amounts to saying that C4 involves a denial
of a purported relation (other than identity) between the object
and the property concerned, while C14 involves a denial of the
possible relation of identity between the two objects. That is
why, Navya-nyäya says that the first involves a relational
absence {samsargähhävä), while the second a difference or (as
it is sometimes called) mutual absence {anyonyäbhävd).
One curious connection between absence and difference,
according to Navya-nyäya, is this : Wherever one finds absence
of potness, i.e., the property of being a pot, one can find, or
cognize difference from a pot. In other words, whatever lacks

potness is also different from a pot and vice versa. Thus, absence
of potness and difference from a pot are two distinguishable
properties while they occur in the same locus or loci. But when we
talk about double or repeated negation, Navya-nyäya uses almost
the same criterion to eliminate innumerable negative formulations.
Therefore, absence of absence of a pot is identified with a pot,
mainly because wherever the first is present the second is present
and vice versa. Hence, to avoid unnecessary multiplication of
entities it has been said :
"absence of absence of x=x."
Using almost similar argument, absence of difference from a
pot is identified with potness (i.e., the property of being a pot).
The difference from a pot is absent in all and only those things
that are pots (a pot cannot be different from itself), and potness
is present in all and only these things. All these and related
problems have been discussed by D.H.H. Ingalls in his pioneer-
ing work on Mathuränätha. 1
The combination "difference from difference from x" has created
an understandable confusion among the modern interpreters of
Navya-nyäya. I wish to clear this muddle. First, we should note,
Navya-nyäya does not use this combination very frequently,
because it does not serve any useful purpose in the discussion
of its logical theory. But it certainly raises an ontological issue
and, hence, there is a brief reference to this problem in
Raghunätha's Padärthatattvanirüpana (PTN), which has been
translated and commented upon by FC.H. Potter. 2 We should
further note that the expression "difference from difference
from x" is distinct from "different from whatever is different
from x." In Sanskrit, one would be expressed as "ghatabheda-
bheda", while the other "ghafabhinnabhinna." The second
expression clearly refers to the same thing as x. In other
words, Navya-nyäya accepts the following identity.
"*--=different from whatever is different from x."
But the property expressed by "difference from difference from
x" is an entirely different matter, as we shall see below.
1
Ingalls, p. 68, pp. 71-72.
2
Potter, p. 68.
In his commentary on the fifth definition of vycipti in the

Vydpti-pafwaka, Mathuränätha uses an identity principle that
is formulated as follows.1
S1. ' 'svävacchinnabhinnabheda == s va",
In literal English this would be
El. "Difference from whatever is different from anything
that has x or is delimited by x - x."
This is certainly not what is stated as
"The mutual absence of the mutual absence of x ~-= x"
or "The difference from difference from x - x . "
Let me explain El. There are two different principles implicit
in the acceptance of El. The first principle is non-problematical,
for, it is usually assumed in modern class logic. Two classes are
held to be identical if and only if their members are so. In
symbols, class a and class ß are identical if the following
condition holds :
(x) (xsa—xsß).
It is also part of the theory that in a closed universe class y,
the complement of the complement a, is identical with a.
But, Navya-nyäya accepts, over and above this, a second
principle with some reservations. It postulates that if there is an
essertial property belonging to all and only members of a class
a and if there is another property belonging to all and only
members of class ß, then those two properties are also identical,
provided the classes are identical.
Now, we can proceed to explain E l . Suppose, ix> is sub-
stituted by 'blue-colour.' Then the class of all and only blue
things is given by the description "anything that is delimited by
blue-colour".
Let us call this class a. Now, the class of whatever is different
from blue things is given by the complement of a. At the next
step, let us define class ß as the class of anything that is different
from whatever is different from blue things. Now, it is clear
from these definitions that both a and ß will have as members
all and only blue things, and, Iieace, they are identical, [u fact,
1
IngaUs, pp. 149-50
ß is the complement of the complement of a. So far we have

said nothing against what is very common in the logic of classes.
Our proposition so far can be stated as follows :
"All and only blue things are different from whatever is
different from anything blue."
The second principle which Navya-nyäya implicitly accepts,
derives the following.
Premises : 1. a = ô
2. Ail and only members of <£ have the essential property p.
3. All and only members of ß have the essential property q.
Therefore 4. p = q
What is it that all and only members of a have ? Any member
of a is at least essentially blue, and, therefore, it has blue-colour.
In other words, '/?' is 'blue colour'. What is it that all and only
members of ß have ? Any members ofß is at least different
from whatever is different from anything blue. Therefore, each
has what is describable as difference from whatever is different
from anything blue. That is, q is such a difference. Therefore,
"difference from whatever is different from anything blue =
blue colour. Q.E.D.
Now, I shall turn to a passage from Raghunätha's PTN.
Potter in offering his translation of this passage has emended
the text, for, he thought, mistakenly I believe, that the author
in this passage was talking about the identification of the
above kind, viz.,1
"x is indentical with what is different from whatever is
different from x"
However, as I shall argue here, the text does not need the
suggested emendation. The text of Raghunâtha (as given in the
Pandit edition) is translatable as it stands, provided we do not
think that the author was talking about the above identity in
this passage :
"anyonyäbhävasyänyonyäbhävo bhävatvam, samsargäbhä-
vatvam ca."
Translating "anyonyäbhäva" as 'difference' we can say that
1
Potter, p. 68.
Problems of Philosophical Logic in Navya-nyàya Î61
Raghunätha was talking about 'the difference of difference of

x9, and not about 'the difference from whatever is different from
anything that has x\ We have to understand the problem by
introducing, as is customary in Nyäya, the notion of property-
locus or superstratum-substratum. Suppose, we are talking about
the difference from a ( = blue-colour). Anything different from
a would then be the substratum of this difference from a.
Suppose, we denote this property or superstratum by "Da".
Then
"Da = difference from a."
"Da" names a property which occurs in everything in the
universe except a. (A further qualification would be needed, if
we do not allow that a property can occur in itself. For, in that
case, we have to exclude from the class of loci of Da not only
a but also Da. But I shall ignore this complexity here.) Now it
is obvious that Da is one of the properties of this universe and,
hence, different from other properties and other things including
a. What about the property denoted by "difference from
difference-froni-a", i.e., "difference from Da"l Let us denote the
second property by "DDa*\ Then
^difference from difference from a,"
DDa, as before, is said to occur in eveything in the universe
except Da9 but it also occurs in a. In other words, a is only one
of the members of the class of loci of DDa.
A problem arises due to what may be called the 'recursive
property' of the language. The phrase "difference from" can be
added to a word not just once or twice, but indeed innumerable
times. Therefore, it is possible to have phrases like "DDa" and
"DDDa" formed indefinitely, Are we supposed to say each time
that each such phrase denotes a new, different property of the kind
denoted by "Da"? In other words, Nyäya accepts the property
denoted by "Da" as a negative entity included under the category
they call abhäva. It is to be noted that Navya-nyâya accepts
seven types or categories of entities : substance, quality, action,
generality, particularity, inherence and abhäva (negation). The
first six are grouped under bhäva category, i.e., the positive
category, which is contrasted with the seventh category called
abhäva or negative category. The negative category is sub-
divided into two types, absence and difference. We can look

upon these categories as classes consisting of individuals or
properties (for Navya-nyâya, there is very little difference bet-
ween the two concepts) as members. Suppose, members of the
first six classes {bhäva categories) as well as those of the class
of absence (=samsargäbhävä) are nameable by the following
expressions :
«tj , a2 5 «3 <Xn ,
In that case the members of the class of difference would be

nameable by
«<Dai'\ "Da,", "Da 8 "...... "Da*".
Raghunätha says that Navya-nyäya accepts the first group of
entities and the second group of entities as well. What about
the entity named by "DDaî Would it be a member of the
second group? The answer is no. For, the entity denoted by
"Da/' does not fall under the first group. Therefore, by
definition, the property named by "DDa" cannot fall
under the second group. Would it form a separate sub-
category of what is called difference. The answer is no, for
(a) it is unnecessary and (b) it goes to add innumerable sub-
categories to the second group. Therefore, Raghunätha says
that the property denoted by "DDa-f* must fall under the first
group, for, there is no other group left. The first group, as we
have already noted, comprises all the positive (bhävatva) and
one sub-category of abhâva, i.e., samscrgäbhäva.
Raghunätha's text (without the suggested emendation) is
translatable, as :
"The difference (mutual absence) of difference is (identical
with) the property of being a positive entity {bhävatva) and
that of being a relational absence as well. For, there will be
infinite regress if we admit it to be (identical with) another
difference."
The commentary (in the Pandit edition, p. 56) adds :x
"Idaip cänyonyäbhävatvävaccinnasyanyonyäbKävam avalam-
byoktam ghatänyonyäbhävasya tu anyonyàbhâvo bhàvatvam
1
Raghudeva, Coram. On Raghunätha, The Pandit, p. 56.
samsargâbhâvatvam patädyanyonyäbhävatvam cety api

bodhyara."
I translate :
"This has been said (by Raghunàtha) haying recourse to the
difference from the set of all differences. The difference
from the difference-from-a-pot, however, is (identical with)
(a) the property of being a positive category, (b) the property
of being a relational absence, and (c) the property of being
(particular) difference from a cloth etc. This is understood,"
Obviously, for each thing x, if x is a difference (such as, the
difference from a pot, the difference from a cloth), then x is
also DIFFERENT from any positive entity as well as from
any relational absence. Suppose, the d is defined as the set of
all differences :
(dp d2, d3 dn ).
Now, if the difference from all such differences is denoted
by "ZW", then such a property (the one denoted by "Dd") is
present in all and only those that are either a positve entity or
a relational absence. The property denote^ by ".Dd", therefore,
is said to be identical with either the property of being a positive
entity or a relational absence. A complexity will arise, if,
instead of talking about the difference from all kinds of
difference (first order) as above, we talk about the difference
from a particular difference (say, difference from a pot^Jj). In
that case, the second difference, namely, that denoted by ccDdx",
will reside not only in all positive entity and in all relational
absences, but also in other (first order) differences, such as d2
and d3. This explains the second statement of the commentary.
The commentary goes on to note another view under "kecit"'.
This maintains as against Raghunàtha, that the difference
from the difference from x is another difference, i.e., should
fall under the category of difference. This, it is claimed, will
not lead to an infinite regress. For, infinity will not be a fault
when it is based on evidence or arguments (prarnäna). The
commentary continues :
anyathâ anyonyâbhâvasyaivàtiriktatvavilopaprasangât.
Sarvatra tadadhikaranamätravrttidharma eva tatsvarüpatva-
kalpanâ-sambhavâd iti vadanti.
î translate :
"Otherwise, difference itself need not be considered as an
additional category. Everywhere we can imagine a property
p identical with any other property q which occurs in all
and only substrata of/?. This is what they say."
I pointed out earlier that if the same principle on the basis
of which the absence of absence of x is identified with x, or the
absence of difference from a pot is identified with potness, is
insisted upon, then one may also identify difference-from-pot
with absence-of-potness, for, these two properties have also the
same locus. If this is conceded, then one may as well eliminate
the category (or sub-category) difference and talk of only one
category, absence. But apparently, Navya-nyäya feels strongly
about a separate category difference; for, there is, Navya-nyäya
argues, a stronger epistemie consideration in its favour. How
about properties denoted by "DDDa" or "DDDDa"! Navya-
nyäya does not discuss them, as far as I have read, and so I
cannot tell either.
§2.8 : DEFINITION AND CLASSIFICATION
F. P. Ramsey in his The Foundations of Mathematics: (1929

edition p. 263) said :
"Essentially, philosophy is a system of definitions, or only
too often a description of how definitions might be given."
If by 'definition' in philosophy we moan something that is
similar to the concept of lakçana in Sanskrit, then, surprisingly,
the above comment would be true of several classical philosophical
systems of India. The term laksana in Sanskrit has been used in
many senses1 but in logical treatises it mainly stands for "défini-
*In the grammatical tradition, for instance, laksana stands for sütra or
grammatical rules, and laksya stands for words. (Cf. Mahâbhâ$ya, Pas-
paeähnika, Värttika 14). J. F. Staal has discussed this in his paper "The
Theory of Definition in Indian Logic," JAO$9 33 (1961), pp. 122-6. One
remark may be added to his discussion. The scheme for a samjnä sütra of
Pänini roughly corresponds to the notion of "nominal or syntactical
definition" of the modern formal logicians. Such definitions are explained
as "conventions which provide that certain symbols or expressions shall
stand (as substitutes or abbreviations) for particular formulas of the
system". They "may be theoretically dispensed with and all formulas
tion". The general problem of "definition" as it is found in

Sanskrit philosophical texts has been studied by many scholars.1
Mile. Biardeau has skilfully handled the question emphasizing the
Nyäya and the Vedântic traditions,2 and recently J. F. Staal has
attempted a ' Normalization of the theory of definition" in Navya-
nyäya3 with the help of the notations of Boolean algebra.4 In his
concluding remark, Staal notes that unlike a large number of
cases in Indian logic the doctrines of laksana and sarhkara "show
a marked extensional character".6 I consider this to be slightly
misleading. In any case, I shall examine here in detail the pro-
blems connected with lak$ana in Sanskrit.
From a very early period the Nyäya school considered three
elements to be the main concern of a philosophical treatise :
uddesa (enumeration of the philosophical concepts), laksana
(definition), and pariksä (examination of those concepts).
The technique of definition, however, in which the Navya-naiyâ-
yikas evince so deep an interest, was not clearly developed in the
early school. Vätsyäyana stated that the purpose of definition
(lakçana) was to differentiate an entity from that which does not
possess the nature or essence6 (tat t va) ofthat entity. Thus, insofar
as the term "nature" or "essence" remained vague in meaning,
the notion of laksana also remained vague. In the new school, an
written in full." See Alonzo Church, Dictionary of Philosophy, (D. D.
Runes) (New York), pp. 74-5.
1
See A. B. Keith, Indian Logic and Atomism, pp. 153-4, S. Radhakrishnan,
Indian Philosophy, H, p. 47, A. Foucher, La Compendium des topiques, pp.
7-14, &D. H. H. Ingalls, Materials for the study of Navya-Nyäya Logic
(to be abbreviated as MAT), 1951, pp. 80-1.
2
See her "La définition dans la pensée indienne*', JA, 1957, pp. 371-84.
3
Navya-Nyäya is usually considered to have begun with Gangeéa, whose
date is now usually placed in the 14th century A.D. See D. C. Bhatta-
charyya, Bange Navya-Nyäya Carcä (Calcutta, 1952).
4
J.F. Staal, op. cit., p. 126.
5
He also refers to I. M. Bochenski, who holds that Indian logic is intensional
and thus differs from Western logic which is more or less extensional in
character.
6
Two different readings of the Nyäya^Bhä§ya are available : udditfasya
tattvavyavacchedako dharmo lakçanam, and uddi$tasya atattva-vyavacchedako
dharmo lakfanam. I agree with Miss Biardeau in favouring the latter as
stating the general principle of definition less vaguely than the other
reading, but I think that the first reading may be more in keeping with
Vätsyäyana's style. See next section § 2. 9.
attempt was made to avoid this vagueness by specifying that the

purpose of definition is to distinguish the 'definiendum' (laksya)
from all entities that are different from it (itara-vyävartakatvatn).
Or, to put the pragmatic value of definition more clearly, it was
said that its purpose was to facilitate the inference that the
definiendum is distinct from any other entity (itara-bhedänumäpa-
katvam). A Nyâya definitiod1 (laksanavakyd) usually consists of
two parts 2 : a lak$ya (definiendum) and a laksana (definiens). In
each case of a true definitiod, it will be possible to formulate
a sort of miniature syllogistic inference, of the form : 'bA (is) B,
because C", 3 where the definiendum will occupy the subject
position (paksa), "distinct from others" 4 will be the sädhya, and
the definiens will be the hetu.
The content of the above inference evidently points to a con-
struction of the definiendum and the definiens as co-extensive
terms and "co-extensive" may seem to have the same meaning as
the term "samcmiyatd"* in Sanskrit. Here one may note that the
Nyâya theory of definition does not generally rest on synonymity,
but on sameness of reference. W. V. O. Quine in his Two Dogmas
of Empiricism has shown that definition, in some of the best-
known senses of the term, rests on "the theory of meaning", i.e.,
on synonymity,5 but that in another sense, definition may rest on
sameness of reference or extension.6 This latter sense is found
iBy definitiod is meant a sentence which expresses a definition.
2
Good Sanskrit never employs a copula in the present tense; it distinguishes
attribution from predication by word order or by choice of structure.
3
The use of such types of sentences in representing the inferrential process
in Indian logic clearly indicates that its language is somewhat intensional
in character. Cf. R. Carnap's The Logical Syntax of Language (1959),
Part IV, p. 246, where the author, while enumerating some of the important
examples of intensional sentences, mentions also "Because A, therefore B".
4
It may be noted here that this form of inference apparently fails when the
definiendum is a kevalänvayin (unnegatable) term like vâcyatva ("being
nameable") etc. To avoid unnecessary complications I wish to leave aside
such exceptional cases in the present discussion.
5
Cf. From a Logical Point of View (Cambridge, Mass., 1961), Essay II, pp.
24-7.
Q
Op. cit., Essay VIII, p. 312. "A general term / is said to be definable in
any portion of language which includes a sentence S such that S has the
variable x in it and is fulfilled by all and only those values of JC of which
/ is true. Definability so construed rests only on sameness of reference-
sameness of extension on the part of / and S". It is evident that S here is
frequently in mathematical literature. Insofar as the Naiyâyikas

claim that the definiendum must be co-extensive with the definiens,
the Nyâya theory of definition may be described as extensional
and Staal deserves credit for having noticed this anomaly. But
here some caution must be exercised. In translating "samaniyata"
as "co-extensive" we are liable to miss one important point. The
word samaniyata contains the notion of niyama which is usually
explained as a vj>ö/?f/-relation (cf., niyamas cätra vyäpakatä).
Thus, samaniyataivam has been analysed by the Naiyâyikas as
follows :
A. x is samaniyata with y if and only if x is pervaded by y and
also the pervader ofv. (tatsamaniyatatvam tad-vyäpyatve sati
tad-vyäpakatvam).
The following preliminary remark is necessary to explain this
condition. The Naiyâyikas in their logical analysis use a language
structure which is carefully framed so as to avoid explicit men-
tion of quantification, class, and class-membership. Consequently,
their language structure shows a marked difference from that
of the modern Western logicians. In class logic we use the
notation "g" for class-membership and thus "xsa" is read as
"x is a member of a" or "x belongs to the class a". Thus, we
can express the fact that two general terms F and G have the
same extension by using quantification and class-abstraction as
follows :
(x) (xsa if and only if x&$).1
But the Naiyâyikas instead of classes use properties and in lieu
of the relation of membership they speak in terms of occurrence
what is called an "open sentence" of the form "x is F". Thus schematically,
the definitiod may be represented as " / = F A " . Open sentences, like terms,
have extensions, and the extension of open sentences is the class of all
objects of which the open sentence is true. (See W. V. O. Quine, Methods
of Logic, New York, 1961, pp. 89-104). Accordingly t and S have been said
to have the same extension here.
1
Herea and ß stand for the class-abstractions corresponding to the terms F
and G. That is, a designates the class of all objects of which the term F is
true. This is but another way of saying that the extension of F is the class
a. Similarly with ß. Properties are, however, regarded in Nyâya, as par-
ticulars or individuals, as I have noted earlier. I shall avoid this complica-
tion in the present context, for this point would not change anything
materially here.
(vrttitvä) and its reciprocal, possession. In other words, the

simple predication "a is an F" (in symbols; Fa) which is inter-
preted in class logic as "asx\ would be interpreted in Nyäya
metalanguage as "/occurs in d\ where the small letter "/" is a
property-abstraction (as opposed to a class-abstraction) of "F".1
Moreover, instead of quantification, the Naiyâyikas use "double
negatives and abstract substantives" to accomplish the same
result.2 Note also that "negatives" are frequently embodied in
what I shall call "termini"* Any noun substantive in Sanskrit
can be taken as what I shall call a terminus9 inasmiich as-it
designates what may be treated as a dharma (property)
occurring in some locus and also as a dharmin ( a property-
possessor) in which some dharma or property occurs. To the
Naiyâyikas, just as "ghatah" (a pot) can function as a dharma,
just so can "ghaßbhävah" (pot-absence, i.e., lack-of-pot or
otherness-than-pot). Now, the following procedure is taken to
state the condition for the co-extensheness of the terms F and G.
If " F " and "(?' are not already substantive expressions, so that
they cannot be taken as termini in the above sense, then let "a"
and "6" be the corresponding substantivized termini.4 Now we
state, (see also § 2.3) :
B. a is samaniyata with b iff5 (1) a does not occur in a locus
where absence-of-& does occur, and also (2) absence-of-a does
not occur in a locus where b does occur.
Here (1) gives the condition for the portion "a is pervaded
(vyäpya)hy b" and(2) gives the condition for the portion ua is
1
This can also be put as "a possesses/". The Naiyâyikas while not objecting
to the latter form, usually find it more convenient to adopt the former one,
" / " actually designates the "essential** property of the entity to which F is
applied. Perhaps we are here faced with some form of "essentialism". See
2
next section § 2.9.
Thus Prof. Ingalls has said, "The Navya-Nyäya begins by removing all
quantified statements by the use of two negatives and an abstract" in
"The comparison of Indian and Western Philosophy", Journal x
of Oriental
8
Research (Madras), Vol. xxii, 1954, p. 7.
I use the Latin word "terminus" and its plural "termini" in place of "term"
to furnish the special sense of a substantivized word which may serve as a
term in a Nyäya definitiod. Thus, in this special sense "gamanam" is a
terminus whereas "gacchati" is not.
^ o substantivize a verbal expression the appropriate bhävärtha krt sufiSxes
will be used, e.g., for gacchati, gamana (=gam + lyut).To substantivize an
adjective or descriptive noun (e.g., "father", "wife", etc.) one uses the
tadßita suffixes -tä or /va(-ness, -hood in English).
^This is a common abbreviation for the expression "if and only if".
the pervader (vyäpakä) of £". (See definition (A) above). IriNavya-

nyâya language the conditions (1) and (2) are stated as follows :
a is non-occurrent in the locus of the absence-of-è and is also
not a counter-positive of any absence that occurs in the locus of
b {Tad-abhävavad-avrttive sati îad-van-nisthâtyantâbhavapra-
tiyogitvam).
Some further remarks on the nature of abstraction in Sanskrit
are required. Navya-nyäya language, as we know, is generally
intensional. Thus, in their scheme, relations, such as samaniyatatva9
vyäpyaiva, are really relations-in-intension as opposed to
relations-in-extension. The internal structure of this language
can better be represented with the notations for intensional
abstractions than those of class abstraction. I propose to use the
symbol "#[...*:...]" for the (intensional) property-abstraction,
where the prefix "x9\ like the quantifiers, is a variable-binding
operator, but unlike a quantifier produces a singular abstract
term when attached to a sentence. Thus, "being fire" or "fireness"
can be represented under this convention as i6x [x is fire]'9. But
it is important to note that when, instead of terms like fire, etc., we
x
By using the transformation rules of class logic and truth-functional logic
one can easily show that the two schemata given below are equivalent.
(b)—(g*) ((xSoc). — (xSß)).—(a*) ((XBP).—(*6a)).

Now if we interpret "a" as "the class of all objects or loci in which a
occurs" and " ß " as "the class of all such objects or loci in which h occurs"
then the first half of schema (b) represents condition (1), and the second
half represents condition (2) of (B) above. In (b) we have eliminated the
universal quantifier (xy in favour of the negative and the existential quantifier.
The manipulation is designed to show that the language of the Naïyâyikas
can be translated into extensional language involving quantification. The
point to be noted is this : The Navya-nyäya language may be translated
into extensional language although it is not by itself extensional. But this
is not a case in favour of the Thesis of Extensionally, which (in its strongest
form) holds that all intensional languages are translatable into extensional
language. For a brief exposition of this thesis see R. Carnap, The Logical
Syntax of Language, ILP, 1959, pp. 245-6. We can also eliminate "a" and
"ß" and express (b) in quantificational language using only "predicates"
like F, G etc.
(c) — fax) (Fx.—G*).—(a*) (Gx.~¥x)
But note that neither (b) nor (c) reveals all the essential features of Navya-
nyäya language, and that the variable x ranges over the domain of the
loci.
use a "fragment of description1'1 like laksya (definiendum), sâdhya

etc., the notation "x [x is laksya]" will behave more like what
is called an incomplete symbol2 than like a singular abstract term
naming an entity. In fact, an expression such as "x is laksyd" is
to be developed fully only in a descriptive phrase like -'the object
x such that xis the laksya of such-and-such a definition" (in
symbols : (lx)Fx).z When the Naiyâyikas use such descriptive
phrases for x without naming x, it is clear that they cannot talk
directly about the abstract property of x by substantivizing (with
suffixes (i-tva", or "-tä") the term that describes x. The usual
practice of the Naiyâyikas under such circumstances is to use
another descriptive phrase to refer to such an abstract property.
Thus, if x is described as the laksya in a given context, the
property-abstraction from x will be described by the Naiyâyikas
as the laksyatâvacchedaka, i.e., the limiting property of the
relational abstract laksyaiä, i.e., in symbols, the limiting property
of x[x is laksya or definiendum ] in such a context. For instance,
if "fire" is described as the laksya, we can describe the property
"being fire" or "fire-ness" as the limiting property of the rela-
tional abstract laksyatä in such a context.
To extend the above convention for representing intensional
abstraction, I propose to use the symbol "xy[.. .x«..y...]" for the
relation-in-intension.4 Thus, the relation of samaniyatatva, etc.,
will be written as "xy[x is samaniyata with 7 ] " , where the order
of the variables in the prefix will determine the pratiyogin and the
anuyogin of the relation in question, namely, the first will be
taken as the anuyogin or "subjunct" and the second as the prati-
yogin or "adjunct" of the relation.5
x
Prof. Ingalls has introduced this convenient term in his MN, p. 45.
2
By "incomplete symbol" is meant a symbol which has no meaning in isola-
tion, but is only defined In a certain context. "They have a meaning in use,
but not in isolation." Cf. A. N. Whitehead & B. Russell, Principia Maihe
matica, Vol. I Ch. Ill, pp. 66-7.
3
Since Pea no it has been customary in logic to use this iota- opera tor to
represent the descriptive phrases like "the object x such that".
4
AU the above notations for intensional abstraction have been adapted from
W. V. O. Quine. See his Word and Object (Cambridge, Mass., 1960), pp.
164-6.
5
Each Nyäya relation has a "direction", i.e., it can be conceived as proceeding
from one relation to the other, in other words, a relation can always be
expressed in the form "relation of.y to jt". In such context the Naiyâyikas
describe y as the pratiyogin, and .vas the anuyogin of the relation in
question. On this point see Ingalls, MN, p. 40.
Now, to return to the problem of the relation between definiens

and the definiendum. Let us illustrate the point with this defini-
tiod : bhäryätvam vwähita-närltvam.1 One may fairly translate
this as "a wife (is) a married woman", whereupon the laksya or
the definiendum would be "wife" (bhäryä) and the laksana or
the definiens would be '/married woman" (viuähita-näri). Now,
"wife" and "married woman" can be said to be co-extensive,
i.e., they have the same extension, because they can be taken as
"predicators" (of degree one) having as extensions the correspond-
ing classes which are equivalent.2 But since "wife" and "married
woman" cannot be taken as the relata of the relation of sam-
aniyatatva, we ascend to the next level of abstraction, i.e., to our
termini "wife hood" and "married-womanhood". Actually, these are
the two relata of the relation of samaniyatatva, and not "wife" and
"married woman". It may be noted that this point will be missed
if this relation is translated without reservation as "co-exten-
siveness".
Now, to modify our previous remark, a definitiod in Navya-
nyâya substitutes a laksyatâvacchedaka for the laksya of popular
language. Accordingly, the laksana undergoes a corresponding
change. This is clearly shown by the definitiod given above.
Instead of vivähita-nän (the ordinary definiens in English), the
abstract vivähitanäritva is taken as the laksana. This justifies
the Naiyäyikas' claim that in a correct definitiod the laksana
should be samaniyata with the laksyatâvacchedaka (instead of
saying that the definiens and the definiendum should be co-
extensive). We can formulate the general definition of the rela-
tion of samaniyatatva in intensional notation as follows :
xy [x is samaniyata with y ] = dz xy[x is pervaded by y]. xy[x
is the pervader of y] = d xy [x does not occur in a locus where
absence of y does occur], AT [X is not absent from a locus
where y does occur].
This corresponds to our conditions (1) and (2) given above.
Now, it is easy «to see that the violation of condition (1) gives
1
Taken from Gadädhara Bhattächäryya's Vivähavädärthah.
2
In thus stating the condition for co-extensiveness I have roughly followed
R. Carnap. See his Meaning and Necessity (Chicago, 1956), pp. 14-25.
3
I shall use the ordinary sign " — d" for definition. This can be read as "is
defined as*'.
rise to the defect of ativyäpti (too wide a definition), and that of

condition (2) to the defect of avyäpti (too narrow a definition).1
A third kind of dosa (defect) mentioned by the Naiyâyikas is
asambhava which is but an extreme case of avyâpîi. It takes place
iff x is absent from all the loci of j \ 2 The corresponding extreme
case for aüvyäpti, though conceivable, is not important in this
context. But there is another peculiar case of avyâpti noted by the
Naiyäyikas which is philosophically significant. This is technically
called ''aprasiddhyavyäptV9**
If both conditions (1) and (2) are violated, both the defects
avyäpti and ativyäpti take place. The situation is as follows : x
occurs in a locus where y does not occur and x is absent from
a locus where y does occur.4 This brings us close to another
problem, viz,, sarhkara, which we shall examine presently. At
any rate, the presence of two defects in one case is nothing to be
wondered at9 for this only amounts to saying that the hetu of the
inference concerned5 is defective in two ways. Actually, the
Naiyâyikas can construct an inference where the hetu is defective
in all five possible ways.6
x
The Sanskrit definitions for these two defects are a-lakfya-vrttitvam and
iak?yävrttitvam respectively. We can represent them in intensional symbols
as follows : (1) xy [x is ativyâpta by y]=d xy [x does occur in a locus where
absence of y does cccur]. (2) xy [x is avyâpta by y]=d xy [x is absent from
a locus where y does occur]. Using the convention discussed in footnote 1
of p. 169, we can express these conditions in the language of class logic :
(I) (3*)(*Sa.~(*Sß))
(II) (g*) (— (*Sa).*Sß)
Note here, that the variable x% as before, is to be taken as ranging over the
2
domain of loci, which is not the case in the former expressions (b) & (c).
The Sanskrit definition is laksya-mâtrâvrtiitvam. In symbols of class logic
the condition is presented (using the convension of footnote 1 of p. Î69)
as follows :
(III) (x) (*SoO —(*Sß))
Note that the word "mätra" in Sanskrit accounts for the universal quanti-
fication in symbolic expression.
3
For further discussion regarding this subject see Ingalls, MAT, pp. 80-1.
4
In notations of class logic this can be put as follows : (g;c) (x8a.—(*Sß)).
(H*) (—(*€*)• *Sß). J. F. Staal has explained {op. cit.) samkara or "over-
lapping" using the notations of Boolean Algebra as follows :
) y ) y ^ )
5
It has been already noted that the Naiyäyikas want to reduce all definitions
to an inference of the form : A differs from other entities because B. See
above.
®E.g. gaur asvo ghatatvât (A cow (is) horse, because of potness). Here all
the five kinds of hetvâbhâsa occur.
Samkara is a problem related to the notion of logical classifica-

tion. The situation which the Naiyayikas call sämkarya, can be
treated in the same manner as samaniyatatva, vyäpyatva, etc.,
viz., as a relation-in-intension between two termini (as before),
belonging to the same level of abstraction. But samkara by itself
is in no sense connected with the defects of definition (laksana).
The notion of samkara is philosophically significant, only when
one is trying to determine whether a particular terminus abstrac-
ted from an ordinary (general) term can be regarded as jäti (a
generic real) or not. 1 Since our present concern is not to inves-
tigate the problem of jäti or sämänya proper, I shall proceed,
after a brief remark on the notion of jäti, to examine the two
problems connected with samkara, e.g., how to formulate an
exact definition of samkara, and under what condition samkara
or sämkarya will disqualify a terminus from being a jäti.
Since the second question is directly related to the problem
of jäti, I shall take it up first. See also § 4 . 1 .
The nyâya concept of jäti, in its earlier phase, was undoubt-
edly related to the familiar problem of universals. The earlier
Naiyayikas betrayed a spirit of realism in upholding the doctrine
of jäti. The pragmatic ground for accepting jäti, namely, the
need to explain why a general name is applicable to different
individuals, was, however, furnished almost from the beginning.
It was argued that there must be some permanent, characterizing
entity or essence corresponding to the general name which
accounts for the name being true of different individuals. Such
a theory was, however, subjected to the classic objections which
have been raised against Realism. Moreover, the idea of essence
invited still more difficulties. A hierarchy of jäti was admitted.
Sattä (existence or being-ness) was accepted as the highest in the
scheme. Under sattä9 dravyatva, gunatva, and karmatva (answer-
ing to the three general names, dravya or substance, guna, or
Hi is highly probable that the term samkara originally belongs to the Smrti
literature which deals with problems of varna-samkara etc. Udayana in his
Kiranâvali on Prasastapäda-bhäsya develops the notion of samkara at some
length describing it as one of the five jâti-bâdhaka?s. This theory however,
met trenchant criticism at the hands of îSrîharsa and other philosophers.
Even Raghunätha disputed it in his. Padärthatahvanirüpanam and also
briefly in his Guwkiranâvali-phikàsa-dïdhiti (particularly the problem of
samkara with relation to bhütatva and mürtatva). This specific problem has
been discussed by many scholars since. See Vädaväridhi (ascribed to Gadä-
dhara), Padärthamandanarn (of Venidatta) (Prince of Wales Text Series,
No. 30) etc.
quality, karma or movement) were accepted as the three jätiys

which together are considered to be exhaustive and also mutually
exclusive. Points to be stressed in this connection are these :
the question of jäti cannot be compared with the medieval hierarchy
of classes, where the class animal, for example, is divided into
man and beast, or rather, manand not-man,nor with the modern
hierarchy of classes and sub-classes of the mathematical logicians
which invites the paradox of classes (Russell) and the antinomies
of class and sub-classess (Cantor). 1 The Naiyâyikas thought of
this jäti as something real and indestructible, occurring in indivi-
duals (vyakti).2 It is as real as an ordinary particular object, say,
a pot. Just as an ordinary object, a pot, is determinable by
space-time co-ordinates, and any other object of a similar "type"
cannot occupy the same space-time co-ordinates, unless the one
is totally included in and, therefore, becomes a part of the other,
so also only one jäti can occur in one individual and no other
jäti can occur in that individual, unless it is either included in or
is inclusive of the former jäti. This principle was implicitly
followed with the result that only a few termini abstracted from
general names can betaken to be naming jätVs. We can rephrase
the condition for jäti in a specific case as follows :
C. If x and y are two termini naming two general properties
shared by more than one individual, and if those two proper-
ties are said to occur in one and the same individual (vyakti),
then they can both be regarded as instances of jäti, iff one of
them is either included in or inclusive o/the other.
Violation of this rule gives rise to the situation known as samkara.
A terminus under samkara is disqualified from being a jäti since
it violates this principle.3 This is, in effect, a defect of cross-
classification or overlap.
^These antinomies are usually named after their discoverers. For a good
exposition of them one may consult S. C. Kleene, Introduction to Meta-
mathematics (New York, 1962), Ch. Ill, § 11, § 12.
2
The traditional Nyäya holds the curious doctrine that the jätVs are real
(sat, satsvarüpa) but do not possess reality (sattä) by inherence. They were
forced into this position by difficulties of hierarchy and of the relations
between categories. If sattä could possess sattä there would be the faults of
ätmäsraya and of infinite regress. Cf. Kiranâvali.
3
The principle in Nyäya language has been stated as follows : "sva-sämänä-
dhikaranya-svâbhâvavadvrttitvâitad-ubhaya-sambandhena jâti-visisfa-Jâtitvâ-
Now, to formulate the definition of samkara is not at all

difficult, if only we make certain that the relation of "inclusion"
as expressed in the above condition be refrained using intensional
language, I shall examine a definition of samkara offered by
Gadädhara Bhattäcäryya in his Tattvacintämani-didhiti-vivrti.
The definition runs as follows.1
Taj'jäty-avyäpakatve sati taj-jäti-vyabhicärirve sati taj-jäth
sämänädhikaranyam samkarah.
Using the notation for intensional abstraction as before, this
can be written as follows :
x[x possesses samkara] = x[x is not the pervader of a jäti a}.
x[x is not pervaded by the samojati a]. x[x occurs in a locus
where a occurs].
Example : by substituting "indriyatva" (sense-organness) for x
and "prthivïtva" (earth-ness) for a we get a true statement,
which proves that the former is disqualified from being a jäti by
the latter. It is, however, clear that samkara is here defined with
relation to a jäti term. But there is another kind of samkara
which the Naiyâyikas call "paraspara samkara", for instance, the
relation between bhütatva and mürtatva. In order to apply the
above definition of samkara here, it is necessary to deliberately
ascribe jätitva to one of the two, so that the other can be tested
for samkara. But such an ascription (called äpädana) is entirely
deliberate, and is made only for the purpose in question. Thus,
here also the whole notion of samkara amounts to this : it is a
sort of relation-in-intension between two termini, and it deter-
mines that not both of them can be jäti, i.e., at least one of
them is not jäti. Samkara, in Nyâya theory, may disqualify a
terminus from being a jäti whose status as jäti is yet to be
proved, but it cannot do so where such status has been proved
through some stronger means.
Thus, from what has been said it can be seen that the doctrines
of laksana and samkara can be translated into extensional
language. And one will readily grant that such translation is a
vaçchedena sva-samänädhikaranatyantabhäva-pratiyogitväbhäva-niyamah"
Note that the expression "jâtitvâvacchedend" has the force of a universal
quantifier.
iSeeGâdâdharî{^Chowkhamha Sanskrit Series, Vol. 42>, p. 91.
valuable aid to our understanding of Navya-nyäya. But such

translation does not always reveal, and, in fact, if not used with
constant caution may actually obscure, the essential features of
the language-structure adopted by the Naiyâyikas. In their analysis,
as. we have seen,, the Naiyâyikas cling to intensional abstraction,
and speäkjrf their abstract properties as intensions of the predica-
tes and sentences, rather than of classes and truth-values.
§ 2.9 : DEFINITION, DIFFERENTIATION AND ESSENCE
A. N. Whitehead once described some philosophers as "poor

definition-cutters with logical scissors." Similar derogatory
phrases have been applied to the Navya-naiyâyikas by their
compatriots in India. This only shows the importance Navya-
nyäya has put on clear and precise formulations of laksana (of
'definitions') of logical and philosophical concepts in any
systematic treatises (cf., sästra). I have translated "laksana" by
"definition" in English, although I do not thereby claim that
the function of laksana in Sanskrit philosophic activity is exactly
and neatly matched by the-function of "definition" in Western
logic. In fact, few philosophical (and logical) terms in Sanskrit
are translatable in this way into English or in Western philoso-
phical language. But using the principle of charity we select a
translation that comes close enough so that our discussion of
classical Indian philosophy can proceed in modern terms. Such
a procedure seems quite acceptable (in fact, is current among
modern writers) provided care is taken to explain the concept
in question in intelligible manner in the course of our discussion.
Sometimes laksana is understood to be the defining property
or properties. However, "lakçana" or "definition" is also used
ambiguously to denote an act that the philosophers perform when
they utter a definition-sentence. Arguably, laksana in this sense
may belong to the class of 'illocutionary acts', the kind described
by Austin and Searle.1 As an illocutionary act, it is also correla-
ted with what has been called a 'perlocutionary' act. In fact,
some sort of a speech-act analysis of the act of definition, i.e.,
laksana, may be fruitful, although this will not be my concern
here.
Austin, (1962) pp, 98-99, Searle, p. 54ff.
In ordinary Sanskrit "laksana" generally means a mark, an

indicator or even a characteristic, and "laksya", accordingly,
means the indicated entity. Philosophers have taken
these two terms to use them in a rather technical sense,
which is not entirely unrelated to their ordinary sense. The
grammarians (Pâninïyas) use the same pair of terms in a slightly
different sense. The sütras of Pänini which embody the rules of
grammar are called laksana, and the correct forms derived in
accordance with these rules are called laksya. This has given
rise to some confusion in the minds of some modern schplars.
But G. Cardona has set the matter straight by pointing out
that there is a much simpler explanation of this phenomenon.1
"Laksana" means "explanation through characterization". A
grammatical rule or a sütra is called a laksana, because it
characterizes (i.e., explains) the correct speech. In fact, we should
note that a laksana does not mean a sütra or a grammatical
rule in the sense of stating a regularity of the word derivation.
(This was the confusion of Al-George.)2 But a laksana, of course,
stands for a sütra which states a grammatical regularity. By
formulating a sütra, Pânini does not simply state a regularity
of word-derivation (in a 'descriptive sense'), but also recognizes
that there are correct word-forms (cf., sädhu-sabdd) which are
explained through derivation, etc., by the sütra in question in
such a manner that they are thereby distinguishable from in-
correct, ungrammatical forms (cf., apasabda). In other words,
one of the functions of the sütra is to indicate, i.e., it serves as
an indicator of, the forms that are grammatically correct (so that
the incorrect forms are excluded). Viewed in this way, it does
not come as a surprise that the sütra will be called laksana
indicator' and the correct forms laksya 'the indicated'. I suggest,
therefore, that this (grammarians') use of the pair 'laksyorlaksand*
is also connected with the logicians' use of the same pair, for
which one may adopt the translation 'defined-definition' in the
absence of any better terminology.
Vätsyäyana states under Nyäyasütra 1.1.3 : tkThis sästra
proceeds in three ways. Enumeration or naming (of the cate-
gories), (their) definitions {laksana) and examination (of those
Cardona, p. 328.
2
À1-Georgc, pp. 213-21. ["Laksana, grammatical rule", Festschrift K. C.
Chaitopqdhyaya,\91\.}
definitions)."1 What Vätsyäyana say? here is, in fact, true generally

of philosophical treatises of all other schools of classical India. A
philosophical treatise would, according to this style of philoso-
phizing, formulate definitions or laksanas of the concepts or
categories accepted in the system, and then examine, i.e., ascertain,
with the help of pramänas (means of knowing) whether the
defined object (laksita) conforms, in fact, to the definition as
given. After formulating a definition, the philosopher proceeds
to show why and how his definition adequately meets the
acceptable standard or the necessary requirements of a definition
and how the defined objects and only such objects are
characterized by the definition, i.e., the defining property or
properties suggested by him. This eventually leads him to a
discussion, and a critique, of the rival theories and inadequacy
of rival definitions of the same concept.
Let us now see whether we can regard lukfana (in Vätsyäyana's
sense) as an illocutionary act. What is a laksana or what [
have called a definition here ? Vätsyäyana answers : It is the
property that distinguishes the object to be defined from what
it is not. 2 This answer underlines the fact that a philosopher
gives a definition by assigning to the object or objects the
defining property (laksana) and by his act of definition he
intends to produce some effect or consequence, i.e., an awareness
in his students or readers so that they would be able to recognize
or identify the object or objects defined as distinct from others.
This is, it seems to me, comparable to what is called a 'perlocu-
tionary-effect' of an illocutionary act.
Uddyotakara makes this point clear in answer to a slightly
different question. It is asked why Ak§apäda (the supposed
author of the Nyäyasütras) formulated a separate statement
(sütra 1.1.3) for classifying the pramänas (means of knowing)
into four, while it would have been obvious that there were
(according to Aksapäda) only four pramänas from the fact
that he formulated definitions of only four such pramänas.
Uddyotakara's answer amounts to saying that the intended
effect of the act of definition is different so that the students or
iVätsyäyana, p. Î81 noted also in Section 2.8.

2
Ibid., p. 181. See also Staal, pp. 122-6.
readers would not have understood from such definitions that

there are only four pramânas. I quote i1
"Such a contention could have been conceded ; since four
(pramânas) have been defined (by Aksapäda), it is understood
that there are only four pramânas (according to him). But
this is not so. For, a definition (laksana) is the ground/
reason for differentiating the object (to be defined) from
others. Certainly, the definition differentiates the object
(padârtha) from those similar to it as well as from those
dissimilar. But, it cannot perform the function of restricting
(cf., niyama) the number (so that one would understand that
there are only four pramânas). Such restriction is intended
by another act (different from the act of definition). That
there are no further pramânas (besides the four defined) is
not understood from the definition....Therefore, the
classificatory statement (in Ns. 1.1.3) is justified, as it is
intended to resolve the doubt (whether there are further
pramânas or not)."
What Uddyotakara tries to explain here is easily grasped by
distinguishing between two different kinds of illocutionary acts
with two different intentions and conditions. One is the act
consisting of a classificatory statement (vibhägoddesä) performed
by uttering Nyäya-sütra 1.1.3 intended, among other things, to
restrict the number of pramânas to four, and the other is the
act of definition of the four pramânas performed by littering
Nyâyasûtras 1.1.4-1.1.8, intended to allow the readers, among
other things, to recognize them and differentiate them from one
another as well as from those that are not pramânas. One
further point to be noted in this connection is this. If the act
of definition in this sense is an illocutionary act and its per-
locutionary effect is differentiation of the object to be defined
from similar and dissimilar objects, then the above Nyâya
definition of definition in terms of assigning a differentiating
property may be seen as an acceptable attempt to define an
illocutionary verb in terms of its intended perlocutionary effect.
Vâcaspati Misra comments2 : "Defining is speaking of the
unique reason or evidence (what is technically called kevala-
1
Uddyotakara, pp. 183-4. (A Thakur's edn. Nyäyadarianä).
2
Vâcaspati, p. 186.
vyatireki-hetu in Uddyotakara's system). That (unique evidence)

determines the items or things to be defined by separating or
differentiating them from similar things as well as dissimilar
things"..'-In other words, Vàcaspati thinks of definition as a
means of supplying a property, a 'unique evidence', for making
a valid inference regarding the distinction of a set of things
from what they are not. The nature of this valid inference
is discussed below.
A well-known controversy in the West, as far as the
theory of definition is concerned, is usually formulated in the
form of a question : What do we actually attempt to define ?
Things ? or, words? This point is reflected in the usual distinc-
tion that is made in the West between a real definition and a
nominal definition. If we define things, it is a real definition, and
if we define words, it is a nominal definition—so goes the
standard interpretation. A third alternative, which steers clear
of both, is added by asserting that we define concepts,
neither words nor things. What about the well-known defini-
tions offered and discussed by the respectable ancient and
classical (Western) philosophers ? There is a strong philo-
sophic tradition, initiated by the Greek 'triumvirate',
Socrates, Plato and Aristotle, which regards real definition
as part of an important style of philosophizing. It may
be surmised that they were not the target of such attack by
A.N. Whitehead as I have already mentioned in the beginning
of this section. It is believed that when Socrates asked "What
i% Knowledge" or "What is the' Soul?" he was asking for a real
definition. He was not merely concerned with the 'meaning' or
'use' of the word 'knowledge' or 'soul'. Thepurpose of a nominal
definition is roughly to establish or inform the meaning of a
word or symbol. This process of real definition has been discre-
dited in modern times in the West and the nominal definition has
been preferred with the growing recognition of the importance
of language in philosophy. Attempts have been made to focus
upon nominal definitions, even to rephrase the older question
about things, "What is X ?" into the question about meaning,
"What is meant by ' X ' ? " Philosophers, according to this theory,
is supposed to formulate a nominal definition, lexical or other-
wise. This does not, however, discredit the so-called real defi-
nitions practised by the classical philosophers. For, in spite of
their misconception about what they were defining, things or

names, it is possible to derive philosophic insight from such
definitions as well as to understand or rephrase them as nominal
definitions of the useful kind or as legislative definitions for
their systems.1
The dispute between real and nominal definition has been
somewhat counterproductive in the Western tradition. And the
by-product of the dispute has led to such views as that definition
is tautology and, therefore, fruitless or that definition is only
'an insistence on hard-headed clarity.'2 In fact, barring some
technical as well as some trivial uses of the term 'definition', one
can regard, with some ingenuity, most definitions as a very
useful sort of philosophic activity by which a word or a symbol
is explained in the sense that its meaning is rendered precise for
a discourse, to enable one (the student, the reader) to correctly
apply it to things or items it is supposed (or intended) to apply.
If definition is understood in this way, then it will seem that
the Indian theory of laksana is not a strange sort of activity
entirely unrelated to the Western notion of definition.
Having made this point, I wish to concentrate on the Indian
philosophers and the problems and disputes they engaged them-
selves in with regard to laksana?
1
Robinson(1954), pp. 12-34.
There is one particular use of definition that is most current in modern
logic. It is sometimes called contextual definition for eliminative purposes.
This is how the logical connectives of the propositional calculus, for
example, were defined in Principia Mathematica, These connectives were
given a 'definition in use'. They are what the authors of the Principia
called 'incomplete symbols' and, hence, defined 'contextually'. What is
given meaning by the definitions are these signs in their contexts.
Quine sharpens the notion of what he calls an eliminative definition as
follows. ''What is required of an eliminative definition of a word W, on
the basis of some accepted defining vocabulary F, is that it'explains how
to paraphrase every sentence -S, in which the word W occurs, into a new
sentence that contains only words of V and Mother than W" "Vagaries
of Definition" in Ways of Paradox and other essays, 1976 ed., pp. 54-55.
2
Whitehead A. N., Adventures of Ideas, p. 91.
3
One significant distinction between the Indian (nyäya) notion of laksana
and the Western (say, the Socratic) notion of definition is that the former
allows the possibility of a number of parallel defining properties of the
same set of objects, while the latter insists on presenting the essence (cf.,
real definition) of the objects to be defined. Cf., Socrates's reply in Theaeta-
tus, "What the thing itself is," Plato 146e. Nyäya looks for some (one of
Let us come back to the specific problems that arise in

connection with laksana in the Indian context. One usual objec-
tion (pwvapaksd) raised against the activity of definition is
given as follows. It is either useless, or it leads to an infinite
regress. It is useless, for, the defining property does not certainly
create the items to be defined. Nor can it generate a precise
understanding of such items or things, for, we are mostly
familiar with the things themselves even before we are presented
with a definition. If we insist, however, that precise understand-
ing comes only through a definition, then that definition i.e.,
the defining character itself, should need to be defined by another
definition so that we can assign it as the defining character after
having a precise understanding of it. This will then lead to infinite
regress, for the process will never end. Therefore, the conclusion
is that definition is useless.
Vâcaspati in reply to such objections has said that definition is
meant for such persons as are unable to clearly identify the
objects or things concerned as distinct from those that are not,
due to some confusion (yyämohä) or other. 1 Udayana has
explained :2
It is not that everybody is confused, so that there could be
no teacher and none to be taught. Nor is it that there is
confusion (imprecise understanding) with regard to every-
thing, so that there would be infinite regress. Nor is it
that every item causes imprecise understanding, so that
there would be no knowledge. Nor is it that every item is
(entirely) unknown so that the defining character will lack
any substratum to begin with.
Udayana's last point is rather technical in nature and hence
needs some explanation. The logical defect called äsrayäsiddhi
(literally, Mack of a proper substratum') arises if a property is
assigned by a sentence to such a locus or substratum as is
entirely unknown (to both the speaker and the hearer). The
several) unique properties in a definition {laksana), while Socrates looks

for the essential property. I believe that essential properties, if there are
any, should be only one for each set of objects, whereas there may be
several unique properties belonging to the same set of objects.
1
Vâcaspati, p. 209, line 6.
2
Udayana, Parisuddhi, pp. 242-3.
defect is connected with the problem of empty subject term in

logic (see § 2.1, § 2.2). Udayana's solution here is that what
we define cannot be entirely unknown, for, that would render
the defining act impossible. What we define is, at least, vaguely
familiar to us, or we have a general, though imprecise,
acquaintance with it (cf., sämänyajnänä). For, without such
an acquaintance, there cannot arise the required desire to
know or understand it precisely. And without such a desire
which leads to the question "What is X" (cf., jijnäsä), the
act of definition does not follow.1
The situation is, it is argued, comparable to the initial doubt
with regard to the specific nature of a thing and the eventual
resolution of it when a decision has been reached on the basis
of some specific evidence. Doubt cannot arise, according to both
Nyäyasütra 1.1.23 and Vaise$ika-sütra 2.2.16, unless there is
some acquaintance with the object of doubt. From a distance,
we see something and then wonder, i.e., doubt, whether it is a
man standing there or a dead tree-trunk. When we have further
evidence, such as, arms and legs, we decide that it is a man.
We need not insist, unless, of course, we are indulging in
Cartesian doubt, whether our so-called evidence is really
an evidence, for, our ordinary doubt is resolved, for
all practical purposes, in the above manner. Similarly, if a
defining character that has already been assigned seems to be
vague or imprecise to the hearer, we may resort to further
definitions. This certainly does not regress to infinity, for, it is
possible to soon reach such characters as are clearly recogniz-
able by both the speaker and the listener. Thus, it is said
by Udayana, "na hi sarvaira sandihänäh purusäli\ "It is not
that people entertain doubt with regard to everything."
Bhâsarvajna points out that it would be impractical to
entertain a doubt or an indecision in thecas^ of each definition.
For, the person addressed to in the act of a definition is a
listener in a situation where some communication is supposed
to take place. To make communication possible, it is assumed
that he understands certain basic categories, or is familiar with
them. In other words, he understands, at least, which things
1
Cf. Aristotle in Posterior Analytics : "When we know that it is, we inquire
what it is." 89 b 34.
certain words apply to or what is meant by them. Therefore,

when the definer defines something in terms of such basic words,
it would be improper to claim that he still generates doubt in
the mind of his listener. In such cases, there is no further
regress. If, however, he does not have a listener who can
understand the application of certain basic words, it would be a
worthless enterprise to utter words for the purpose of defining
anything. Such extreme cases, it is argued, may be ignored, just
as we ignore the ramblings of a mad man. A mad man, for
example, may utter anything pointing at a tree, we need neither
take it to be a question for giving a definition of the object,
nor can we use the utterance for defining that object.
It is sometimes argued that one can possibly define (giving
some unique character) everything else but not what we call
a definition. Bhäsarvajna replies that it is untenable. For the
definition of definition can be reflexive. Just as the unique
character of definition (cf., laksana-laksana) characterizes each
item that is called a definition, it can, at the same time,
characterize itself uiiiquel", i.e., differentiate itself from what
is not a definition. It is argued that reflexivity does not always
generate a paradox. If one utters ''all sound (noise) is imper-
manent", one does not exclude the utterance itself which is also
a sound. "The scriptures should be read" can also be part of
the scriptures which are prescribed to be read. And light can
make other things visible including itself. All these examples
are used by Bhäsarvajna to remove the alleged paradoxicality
of the definition of definition.1
The parallelism between the knowledge of the defining
character and that of the specific evidence (visefa-vijnäna) is
more than peripheral. For, according to the theory we are
trying to expound here, the following holds true. If a
specific evidence, say a, is what allows us to ascertain that
something is A ("This is an A" is concluded on the basis of
our awareness of a through an inference), then a would be a
defining character of A. In other words, with the help of a, we
would be able to differentiate any A from what is not A, and a
definition-sentence could be given as :
"A is what has tf."
l
Bhäsarvajöa, Nyäyabhüsarta conini. on his Nyâyasâra, p. 6.
This is, at least, the considered opinion of Bhäsarvajna in his

Nyäyabhüsana r1
" ''defining character", "mark" and "evidence" are
synonymous."
We are now at the heart of a very significant controversy
over the theory of definition or laksana between two most well-
known Naiyäyikas, Bhäsarvajna and Udayana. I shall briefly
discuss this controversy here, for it throws a flood of light on
the understanding of the nature and purpose of definition by
Indian philosophers in general. Bhäsarvajna argues that if a
characteristic is specific enough so that the recognition of it
leads to the correct identification of an object or a set of
objects, then there is no harm in calling the said characteristic
the defining character or laksana, A problem arises, however, if
the character in question over-extends to an object or objects
not members of the set in question. This is technically called the
fault of 'over-extension' (ativyäptl). In that case, the character-
istic in question does not qualify to be a defining character,
although it may be used for identification in some given
circumstances. We may identify a number of bulls by their
horns, but the property of having horns cannot be used to define
the class of bulls; for, others, buffaloes and goats, have horns
too. But if we can recognize some specific types of horns that
are only found in bulls, there is no harm, according to Bhäsarvajna,
in allowing possession of such horns as the defining character
of bulls. We cannot counter that, if we recognize the specific
type of horns by which we have already identified the bull,
then, the purpose of definition (identification of the object as
distinct from others) has been defeated. For, first, we do not
recognize usually all the bulls in the world at a time. Second,
the definition is intended to resolve uncertainties about the
borderline cases. It is quite possible that after seeing some bulls
we may be familiar with the particular type of horns (if at all
such a type exists), and then on future occasions, noticing such
a particular type of horns, we may call the object "a bull".
This is how, for example, the jeweller examines and differentiates
a real gem from a fake one. Therefore, the procedure is a valid
one, neither circular nor useless.
^Bkâsarvajna, Nyäyabhüsarja, pp. 7-9
A property that is over-extensive in the above sense cannot

be accepted as the defining property without some qualifications.
How about the property or character that is neither over-
extensive nor co-extensive, but, to indulge in another neologism
in English, under-extensive (avyäpta), i.e., too narrow? An
example that is frequently discussed is derived from the Vaisesika-
sütra 1.1.6. The Vaisesikas apply the term "dravya" = "substance"
not only to the four elements, earth, air, water and fire, but
also to such ubiquitous entities as the sky, space, self or soul,
and time, as well as to mind. The above sütra introduces several
alternative definitions (Jaksanas) of substances (see §4.1). One that
is non-problematical says, "A substance isthat which has qualities
(guna)." A list of qualities is then given, and in fact, it is shown
that each substance (or, what is called "substance", according
to them) has some quality or the other. But a problematical
statement is, "A substance is that which has action (kriyä)"
Since action is identified, in the system, with any sort of move-
ment, it is clear that some substances, such as, the ubiquitous
space or time cannot have action. Thus, if the statement is
interpreted as 'whatever is a substance has action' then it would
be wrong. Bhäsarvajiia argues that it should mean, "Whatever
has action is called a substance." This will be true, since action
cannot be found in a non-substance. Bhäsarvajna insists that
action, even though it is under-extensive or narrow, can be
regarded as the defining character of the substance. For, its pre-
sence can still help us in determining a substance, although it
cannot exclude all and only non-substances from the set of
substance. This was a liberal view about the defining property,
which Udayana and others refused to accept. This view supports
a weaker relation between the defining property (laksana) and the
objects to be defined (laksyà), which can be expressed as: If there
is the defining property present, it is the object to be defined,
but not vice versa.
Udayana argued for a stronger relation between the defining
property and the objects to be defined. It is called the relation
of ' co-ex tensiveness' or samaniyatatva, expressed as : There is
the defining property present, if and only if it is the object to
be defined. Therefore, any property, over-extended or under-
extended in the above sense, will not qualify to be a defining
property according to this view. This has been the prevailing
Problems of Philosophical Logic in Navya-riyäya 187
view of the later Naiyâyikas. 1 I shall discuss this view

in some details below after I have raised and answered another
objection against the liberal view of Bhäsarvajna. We have
already seen, in the previous section, how samaniyatatva and
c
co-extensiveness' may not be the exact parallels, although they
are very similar.
Udayana criticized Bhäsarvajna as follows.2 According to
the liberal interpretation, any evidence or mark, a, as long as
it is invariably connected with A, the object to be denned, will
be regarded as the defining character of A. (I use 'evidence' as
a convenient translation of hetu here.) Therefore, we have the
following inference :
"Something is an A because it has a
And Whatever has a is an A, such as, the case /?."
If this is conceded, then, Udayana argues, not only the dis-
tinction between an evidence or 'mark' (Jingo) and a defining
character is eliminated, but also the purpose of definition is
defeated. For, the purpose of definition (laksand) is : either (a)
to differentiate AJs from not A9s9 or (b) to warrant
successful application of the expression ''A99 to the relevant
objects. One is called technically vyâvrtti, and the other vyava-
hära. These two can also be called the 'perlocutionary' effects
of the act of definition. Now, differentiation is possible if we
reach a negative conclusion on the basis of the evidence <?, such
as, "something is not a non-A." However, if we depend, for
the required inference, upon the invariable concomitance bet-
ween a and the property of being an A, and if we insist at the same
time that such concomitance is derived from positive examples,
(i.e., cases where both are present), then we usually obtain a
positive or affirmative conclusion of the above kind ("something
is an A"). We cannot, however, differentiate certain entities from
non-yl's on the basis of such a mark or evidence, a, for a may
not be present in some A's, according to this theory. Secondly,
Udayana argues, he who has recognized some A's can successfully
apply "A" even without depending upon the recognition of
a. And he who has recognized no /fs cannot apply 'M"
successfully any way.
1
Annaipbhatta, p. 121. See also Staal, op. cit.
2
Udayana, Kirwjävali, p. 29.
Udayana's second objection has, I think, already been

answered above by our "bull" example. Udayana's first objec-
tion can also be answered partially. For, it can be pointed
out that differentiation as the purpose of definition is not
defeated as long as "it is an A" is held logically equivalent to
"it is not a non-,4."
Udayana uses definition in a restricted sense, as I have already
rioted. Following the line of Uddyotakara and Vacaspati, he
thinks that the defining property a should only be 'co-extensive'
or cco-pervasive' with A's, it should neither be present in any
non-A, nor fail to be present in any A. In other words, the
definition should avoid the fault of over-extension as well as
under-extension. Therefore, an under-extensive, or narrow, but
non-promiscuous (avyabhicärin), property like action cannot be
regarded the defining property of substance. Besides, Udayana
argues, the act of definition should make only a particular type
of inference possible, not just any type. This type is called
keva/avyati'rekin, and can be illustrated as follows :
''Everything that is an A differs from non-A.
For it has 0,"
where the definition sentence reads
''A is what has a."
It may seem that the above inference requires the following
general statement to be true (supplying the required relation of
vycipti or 'pervasion' between a and the property of being an A):
"All and only ^ ' s have a."
If this is admitted, then the inference amounts to either a
tautology or periphrasis. Indian logicians have usually tried to
exclude such tautologous inference from the scope of their study
oï inference. Udayana, however, formulates his objection to the
general statement (the premise) of the above kind as supporting
the above inference in a slightly different way. One of the
cardinal requirements of a general statement of the vyäpti
relation, according to Nyäya (in order that it may give rise to an *
empirical inference), is that it must be supported by an empirical
i.e., existential, example, where both the properties, a and that
of being an A, are instantiated. It is further required that such
an example must neither be identical nor form a part of what is
expressed by the subject-term of the inferred conclusion in
question. The implication of the first requirement is that the

general statement must have empirical certification. The problem
was implicit in both Dignâga's and early Nyâya theory of
inference, as we have seen in chapter one.
The second requirement implies that this process of inference
should not involve a petitio principle for, we would then only
beg the question if the certifying example already assumes the
truth (even partially) of something which we are supposed to
prove by the inference. We cannot choose as our certifying
example a case that belongs to a set of cases which we intend to
prove by the inference concerned. It may be noted that the
Buddhists (in the post-Dharmakïrti period) allowed the required
example to be part of what is expressed by the subject-term
(i.e., paksd) and called the procedure antarvyäpti. But Udayana
did not accept antarvyäpti.
In view of the above considerations, Udayana argues that the
general statement (the vyäpti), to support the above kind of
inference, should be formulated negatively, i.e., as follows :
"Whatever does not differ from non-^4, does not have a,
e.g., the case that is B"
In fact, the standard five steps in the inference are given as
follows (see § 1.1 before) :
Thesis, step 1 : Everything that is an A differs from non-A
Reason, step 2 : For each of them has a.
General rule, step 3 : Whatever does not differ from non-A
does not have a, e.g., that which is B.
Upanaya, step 4 : Everything that is an A has the absentee
(pratiyogin), a, whose absence pervades the absence of
'difference from non-^4'.
Conclusion, step 5 : Therefore, everything that is an A
differs from non-A.
This formulation, according to Udayana, avoids the previous
objection. For, the negative general statement can easily be
certified by an example that does not form part of what is
expressed by the subject term. Let us take the illustration which
Udayana discusses. This is what is called an inference based
upon an 'exclusive' evidence—or mark (cf., vyatirekiri), and any
such exclusive evidence or mark can be regarded as the defining
property. The Vaisesika used the word "earth" for every naturally
solid material substance as distinct from "water", etc., and
believes that smell is the natural property of such substance.
Thus, it is that, earth is defined in the Vaisesika system, as that
which has (natural) smell. This definition, according to Udayana,
will make the following inference possible :
Step I : Earth differs from non-Earth.
Step 2 : For Earth has smell.
Step 3 : Whatever does not differ from non-Earth does not
have smell, e.g., (pure) water in this glass.
Step 4 : Earth has the absentee (pratiyogin), smell, whose
absence pervades the absence of difference from non-
Earth.
Step 5 : Therefore, Earth differs from non-Earth.
"Whatever is earth differs from such non-earth as water and
air." Udayana claims that in Step 3, the example is easily avail-
able meeting the two requirements stated above. He further claims
that the conclusion is here in the required negative form so that
it directly records the 'perlocutionary' effect, viz., differentiation
(vyävrtti), of the act of definition. It may be argued that the
property we infer (predicate of the conclusion) is called'difference
from non-earth', and such a property is unfamiliar and, hence,
unknown to us, unless we observe it instantiated in a particular
earth, a pot, for instance. If we are asked to leave such a pot,
etc., out of our consideration, for, they are included in the
domain of what is expressed by the subject term, then we might
face another difficulty. Nyäya requirement is that where we infer
a property as qualifying something in an inferred conclusion, that
property cannot be, prior to the inference, entirely unfamiliar
to us. In other words, we must have some general acquaintance
with the property before we can infer it as qualifying something
else. The point is technically called the fault of unfamiliar pro-
perty as the inferred predicate (cf., aprasiddhasädhya). I cannot
infer something to be circular in shape, if I do not have even the
vaguest notion of what 'circularity' is. Udayana, however,
counters this objection by saying that familiarity with the pro-
perty 'difference from non-earth'is to be derived from acquain-
tance with such individual earth, a pot or a table; and we do
not beg the question here, for, j h ç coiiclusian we derive is
Problems of Philosophical Logic in Navyanyäya 191
regarding all instances of earth, including earth atoms

(including also those we have not seen and will never see).
The alternative purpose of a definition, according to Udayana,
is to establish the use of the word (cf., vyavahära-siddhi)? "earth".
This is also accomplished thro ugh an inference of the above kind.
Step 1: AH these things under dispute are called 'Earth'.
Step 2 : For they have smell.
Step 3 : Whatever is not called 'Earth' does not have smell,
e.g., (pure) water in this glass.
Step 4 : There is no absence of smell in these things, which
absence pervades the absence of designation by the
term 'Earth'.
Step 5 : Therefore, these things under dispute are called
'Earth'.
Udayana's description of this type of what we may call
'negative' inference has many loose ends. Gangesa devotes a
whole section of his Tattvacintämani to explain some of the
problems, but we need not go into them here.1 Udayana's main
point, however, is that we need the negative formulations of the
general statement in the above manner and, consequently, the
conclusion derived would also have a negative formulation. This
is due to the peculiarity of the cases discussed. In all such cases
of inference we have sort of a relation of 'co-extension' or 'equi-
valence' not only between the property we use as our evidence
or mark for the inference, and the property we infer in the
conclusion, but also between these two and the extension of the
subject term. If we ignore the function that an exampleis required
to play in the Nyäya system, we might proceed as follows :
Step 1: "Whatever has a is an A."
Step 2: "All ,4'shavea."
Step 3: (Conclusion) "All ,4'sare ,4's."
This is patently tautologous or repetitious, for, what is said in
Step 1, is repeated more or less by the next two steps taken
together. Udayana's negative formulations, it may be argued,
are not at least patently tautologous in this way.
It is argued sometimes that to avoid this problem, the alter-
native purpose of the definition is emphasized, viz., to establish
x
Gangesa, Kevalavyatirekin Section.
the use of the word (cf., vyavahära-siddhi). Under this model we

may rephrase the steps as follows :
Step 1 : Whatever has miscalled "an .4."
Step 2 : Each of these items has a.
Step 3 : Therefore, each of them is called "an A."
This, apparently, avoids the problem of patent tautology.1
One may ask : how does the negative formulation avoid
tautology? For, as we have seen in § 2.6 and § 2.7, what is
different from non-A is identical with A. The answer is that if
the same entity is presented in two or three different ways, the
cognitive contents vary, which makes the inference non-
tautologous.
I shall now discuss some related questions that are raised in
connection with the Indian theory of definition. First, one may
ask : what is the status of definition of definition (in this sense) ?
If each defining character (i.e., definition in our sense) is suppo-
sed to differentiate the corresponding objects to be defined from
those that are not so, the defining character of definition itself
must also differentiate instances of definition from non-defini-
tions. The defining character of definition, a la Vatsyäyana,
Uddyotakara and Vacaspati, is differentiation (vyävrtti) from
objects not covered by the definition concerned. Thus, we have
the following puzzle :
Each defining character differentiates, (i.e., has differentiation)
Differentiation is a defining character.
ergo Differentiation has differentiation.
Nyâya solves the puzzle by referring to many analogous situa-
tions that we admit generally. For example, mimd is said to
know everything, including itself. Bhâsarvajna gives a list of
such situations, (see above). Therefore, we may say that
1
What 1 am calling tautology is to be distinguished from the rather techni-
cal use of the term in modern logic, where any sentence that is necessarily
true by virtue of its truth-functional form is called a tautology. Roughly,
I am calling such sentences tautologous here as ''A is A" or "A has ,4-ness."
An inference which uses such a premise or such a conclusion is sometimes
cailed siddha-sadhana.
differentiation is self-differentiating to avoid any triviality or

regress to infinity.1
A second question is raised regarding the relationship between
the defining character and the means of knowing (pramäna). We
have seen above that an evidence, which is also a means of
knowing an object, can be used as its defining character. Could
all means of knowing be defining characters ? Bhâsarvajna says
that all defining characters are also evidence (or can be used as
evidence), but not all means of knowing are defining characters.2
A defining character, according to Bhâsarvajna, is what specifies
the object (and in specifying, differentiates), but a means of know-
ing can simply give a general notion of the object; it may reveal
the object without specifying.
Some may argue that the act of definition is redundant, and
hence, useless as a philosophic exercise, for all knowables are
established and distinguished by evidence or different means of
knowing, and once this is done, definition serves no useful
purpose. Udayana says that those who argue in this manner
follow the policy of drinking liquor while at the same time con-
demning it (cf., nindämi ca pivämi-ceti nyäyafi).3 For, an evidence
or an inferential mark that is free from the faults of over-exten-
sion or under-extension in the manner described above is called a
defining character by Nyäya, says Udayana. He concedes the
point that the act of definition does not establish any new truths
(or facts), but repeats what has already been established by other
means of knowing. The defining act is, in this sense, called
technically an anuväda or repetition of what is otherwise
established. But it is not, argues Udayana, purposeless. Udayana
says that the function of definition in philosophy is similar to
that followed in medical texts or in grammatical texts. The
medical texts give the defining character of particular diseases,
and the text on grammar gives defining characters of correct usage
or correct speech.
How is this similarity to be understood ? I suggest the follow-
ing. Both medical science and grammar are examples of what can
be called empirical science. A medical student observes certain
bhâsarvajna, p. 9. See note 1, p. 184 above.

2
Ibid.,p. 8.
3
Udayana, Kiraiiàvaliy pp. 29-30.
characteristics and then tries to identify the particular disease

with the help of the definition found in the text. For, the author
or the master observed the past cases and extracted certain
characteristics from them and noted them in the definition to
help future identification of the same disease. Similarly, the gram-
marians, observing the correct speech-pattern, extracted certain
characteristics and noted in the text to help the student to identify
such correct speech-pattern on future occasion. The student of
grammar learns thereby to differentiate correct speech-patterns
from thousand incorrect ones, while the student of medicine
learns to distinguish one particular disease from another. In both
cases, therefore, the act of definition is neither redundant nor
useless. Philosophers' definitions are to be regarded, according to
Udayana, useful in the same way- Students may be familiar with
the object but not with its specific or unique characters, just as
a medical student may be familiar with the disease but not with
its specific marks, or a student of grammar may be familiar with
the correct usage but not with its specific characteristics. Defini-
tion, in all such cases, is helpful in identifying and differentiat-
ing the object, the disease, or the usage, from what it is not.
The above consideration also answers the question whether
definition imparts any new knowledge, or any progress in know-
ledge is made by definition. People may argue that a definition
gives only a true proposition, and hence, an a priori or an analytic one
(for, it is true by its meaning alone), and, therefore, it is trivial, for, it
does not add to our empirical knowledge. This is reflected in such
claim as is often made, for example, that such and such statement
is true by definition and empty of factual content. But, if defini-
tion is understood in the above manner and as an illocutionary
act with a perlocutionary effect (as I have suggested here), it will
be a mistake to call it useless. An act of definition is often
intended for the'uninitiated', for whom the resulting statement
is hardly trivial, and if it is already trivial for him, then there was
no need for the act of definition for him in the first place. There-
fore, this objection against definition arises out of a mistake by
taking definition out of the context in which it appears, for, it
has been conceded already that among the "already initiated"
people there is no need for definition.
The Nyâya 'definition' is not related with the analysis of the
Problems of Philosophical Logic in Navya-Nyâya Î95
'meaning' of the word, but rather with the application of the

word to things (and by "thing" I mean anything, not necessarily
the real, the particular or the material). This may presuppose
some familiarity with the thing or things to be defined, but the
definition is supposed to make the hearer understand the things
defined better by helping him to distinguish them from others.
Philosophers have sometimes called this to be a transition from
having a clear idea to a distinct idea. One has, for example, a
clear idea of a thing when one can recognize an example of it,
but one has a distinct idea when one can understand its distinc-
tive feature and distinguish it from others.
Nyâya sometimes accommodates what we call lexical definition.
For, it has been argued by both Uddyotakara and Väcaspati
under Nyäyasütra 1.1.15, that supplying proper synonyms is
also another respectable way of supplying a definition. The sütra
gives two synonyms to define buddhi 'consciousness'. Uddyota-
kara explains that the object (padärtha) expressed by these two
synonymous words is called buddhi 'consciousness'. One may
argue: How can a synonym supply a definition? Uddyotakara
answers r1
It is so in so far as it helps to differentiate (the object from
the rest). For, all definitions differentiate the object from
the rest. As long as these synonymous words do not apply
to other objects, they express the unique property that can
be the defining property.
Even Socrates seems to have approved sometimes a definition
where the logos is interpreted as that in which the thing differs
from all other things. In other words, definition in such cases
is the effort to differentiate and distinguish.2 This is the closest
he came to the Indian idea of a definition.
Definition of the kind that Indian philosophers accept need
not amount to the search for the essence of the thing. For
Aristotle, as is well-known, definition was the statement of the
essence of a thing.3 A property is distinguished from essence by
Aristotle in Topics, 1.5. It is a predicate that does not indicate
the essence of a thing, yet it belongs to that thing alone, and is
1
uddyotakara, p. 436.
2
Robinson (1953), p. 56.
3
Aristotle, Topics, I. .5.
predicated convertibiy of it. Such a distinction is useless for the

Indian Theory of definition, for, obviously any unique property
(unique in the sense that it belongs to the thing alone) can be
regarded as the defining property or definition. It has also been
suggested, as I have already noted, that if the thing to be defined
has more than one unique property, there can then be several
alternative definitions. It is not usually allowed, even by those
who admit the essence of things, for a thing or a set of things to
have more than one essences.1 However, the doctrine of essence,
as a metaphysical reality, does not have many proponents in the
West today. It is often asked, as it should be, why it is essential
for man to be a rational animal, and quite accidental for him to
be a featherless biped. Essence under such criticisms levels down
to our own stipulations, i.e., what meaning we choose to assign
to a word as its primary or principal meaning. And this then
becomes a Humpty Dumpty way of choosing an essence.
If by 'essence' of a thing F we mean a property by which it is
what it is, then the nearest equivalent of an essence in Indian
terminology would be a svabhäva 'own being' or 'own nature'.
The Äbhidhärmika Buddhists say that the 'own being' of fire is
heat and that of water is coldness. In this system, obviously, the
essence (own-being) of a natural kind is one and not many, and,
accordingly, the Äbhidhärmikas say that even the lakçana or
defining property of fire is heat. It is for this and other reasons,
the Abhidharma can be construed (in my opinion, wrongly) as a
3
See note 3, p. 181 above. Also consider the following :
Can an act of definition amount simply to supplying a list of instances
of what is being defined ? Socrates rebuked his interlocutors on several
occasions for answering questions of the form "What is xV with enumera-
tion of instances. This criticism of Socrates may imply that since the
Socratic definition "calls for a single formula" even a complete enumera-
tion of objects or kinds of objects falling under the concept will not do.
The Sanskrit philosophers use the term srngagrâhaka to express a some-
what similar operation, where instead of giving a defining character, a list
of instances is given. This activity, as the name indicates, is probably a
way of answering " What is a bull?" by pointing them out by seizing
(grahana) their horns (srirïga). The touch of humour shows probably that
the Indian philosophers, while they did not think it to be a wrong proce-
dure, certainly consider it useless or impractical. An exhaustive enumera-
tion of kinds is called, however, uddesaov uddesa and vibhäga. We may
recall that the Nyäya commentators argued that the purpose of this
operation may not coincide with that of the act of definition.
form of 'essen tialism', and hence Nâgârjuna has directed his

relentless criticism against such an interpretation of Buddhism,
but we cannot go into that problem here (§ 4.5 below).
Another possible equivalent of the word "essence" would be the
real universals, such as, the so-called waterness of water, spoken
of by the Nyäya-Vaisesika. In this theory also there is only one
essence, say cowness, belonging jointly to all and only such
objects as are called "cow". The doctrine of real universal, how-
ever, has a number of difficulties that are not easily answerable.
Besides, if we correlate essence with the real universal (jäti), we
may unsuspectingly allow that as long as the Nyäya-Vaisesika
is committed to say that 'existence' (^satta) is a real universal, it
is also committed to say that if a thing exists, it exists essentially.
This will raise a host of problems which we shall be well advised
to skip in the present context. (See § 5.5)
A word in our ordinary language very seldom has only one
sense. In fact, all words are more or less ambiguous. Most words,
even important words, are loosely applied to not only different
things, but also to different types of things. Using such and
other considerations, Wittgenstein and his followers have argued
that words like "game" are applied to a variety of items so that
there may be only some 'family resemblance' among them, but
there is no unique property characterizing all of them. 1 This,
however, poses a more serious objection to the philosophic
enterprise that I have called definition here. If there is no defin-
ing property to be found, the act of definition would be impossi-
ble. But I think the notion of family resemblance can be
effectively directed against the doctrine that holds universals as
real entities or against the notion of essence. The same con-
sideration, however, need not ask us to reject the philoso-
phical enterprise of definition. For, a definition can be both
stipulativeand legislative. A philosopher might wish to narrow
down the use of a particular word to such an extent that
it would not apply to very disparate types of things and its meta-
phorical and extended uses would be ignored. In the course of
doing so, he might find some recurring aspect (or, even, aspects)
of the things to which it is applied (not necessarily the essence,
however) and then he can proceed with his definition, using such
Wittgenstein, Philosophical Investigations pp. 66-67.

an aspect. I wish to emphasize that the Indian theory of defini-

tion may be understood as being neutral to such other philoso-
phic worries—the doctrine of essence, the doctrine of universals
as real entities, the problem of abstract entities, the problem of
populating the world with unnecessary entities—and not being
afraid of the "razor" of Ockham. Such philosophic worries, of
course, do exist, but in justifying and following the philosophic
act of definition in the above manner, we need not be entangled
and baffled by them. The problem is one of finding a unique
characteristic or trait in an object, or a group of objects to
which, or to each member of which, a term applies.
Am 1 not trying to avoid the important philosophic issues,
the question of ontology, etc., while talking about definition?
Perhaps I am. But I submit the following consideration. We
sometimes become aware of a new general element, a form or a
pattern in our experience. Such awareness, whether we always
give it a name or not, is helpful in interpreting many future
experiences. When we give it a name and use it in a discourse
it is called the process of abstraction. And in this process,
perhaps, we try to extract the inextractible from experience. But
it is undeniable that the process is helpful for further progress
in knowledge and teaching.
Discussion of 'essentialism' has been revived In modern times,
and, of course, it has a direct relation with the theory of defini-
tion. The nominal definition deals with meanings. Hence, modern
philosophers (such as, Quine), who have misgivings about the
ontological status of meanings and intension, would find even
in nominal definitions "vestiges of essentialism". For, "meaning
is," according to them, "essence divorced from the thing and
wedded to the word." In fact, in our talk of a distinction between
a real and nominal definition, Quine sees what he calls "the old
vacillation between sign and object." The distinction that is
usually found in the dictionary between meaning and colateral
information is, in the light of this view, a reflex of the old
distinction between essence and accident, and to say that a defini-
tion should be analytic is also to yield to "essentialism". This
philosophical stance recommends that, ideally, the domestic
dictionary definition should be of each of only those words that
are "unclear" in terms of words of "higher frequency", and
there will be a set of words of highest frequency which would

go undefined (for, presumably, they are clear).1
Quine's fear of 'essentialism' seems to be overs tressed. For, it is
not always clear what is it that he wishes to avoid talking about.
And what is more problematic is that in his attack he wishes to
tie all the three concepts together : essence, analyticity and
necessity. Although Quinegoes to the extremein his rejection of
essentialism (it was "conceived in sin", he says), it should be no-
ted that heallows, forthesakeof science, what I would call some
minimal form of 'essentialism'. He calls it "a respectable vestige
of essentialism" which he would like to keep. It consists "in
picking out those minimum distinctive traits of a chemical, or of
a Speeres, or whatever, that link it most directly to the central
laws of science" (p. 52, emphasis mine). Gaining knowledge and
imparting that knowledge to others are important parts of the
activity that we call 'science' (apart from its predictive aspect).
Such activities can hardly proceed successfully and effectively,
unless we have available to us a mechanism that allows us to
pick out "distinctive traits" of the items that we intend to study.
The distinctive traits of a chemical are important in chemistry,
that of a species in biology, and that of a disease in medical
science (and this does not seem very different from the function
of a definition or laksana that Udayana had in mind, see above).
What seems to me to be the issue in Quine's empiricism here
is that we need to make a distinction, at least, provisionally
between what is considered (wrongly, tor there is no such thing or
we need not accept such a thing, according to Quine) an essence
and which are unique properties, or distinctive traits or features.
Essence carries a metaphysical overtone (it is regarded as some-
thing that is supposed to make a thing what it is)—which modern
empiricists (Quine uses the term 4*new empiricism" or "empiri-
cism externalized" to distinguish it from the views of Locke et ai)
would like to avoid. Distinctive or unique features are, presum-
ably, empirically recognizable or identifiable, and, hence, fall
within the domain of knowledge and so can be depended upon
and be allowed to contribute to the progress or propagation oi
science.
1
Quine's attack on "essentialism" is well-known. Quotations here are from
has "Vagaries of Definition" as well as from '*Two dogmas of empiricism*'
in From a logical point of view. p. 22.
Modern interest in some form of "essentialism" stems from

Kripke's theory of meaning and rigid designator.1 What seems
relevant to our problem here is, however, that the new semanti-
cists, Kripke and Putnam, have argued that even the common
names, the names for putative natural kinds, like "gold", can be
used as rigid designators. Putnam shows that Quine's pessimism
may be allowed as far as the utility of the traditional notion of
meaning is concerned.2 If a general term like "gold" is given an
analytic definition with the help of a conjunction or cluster of
properties, then it is not very illuminating. However, Putnam argues,
it is possible to empirically determine certain "core fact" about
such natural kind terms as "lemon" or "tiger", or to tell what a
tiger is, or to teach how a "tiger" is to be used. Similarly,
Kripke has argued that "gold" rigidly designates the element with
atomic number 79. It seems to me that Quine's talk of natural
kinds and "minimal distinctive traits" of chemicals is not in-
compatible with, at least, part of what Kripke and Putnam try
to justify here as some form of essentialism, and this "part"
will be, in Quine's language, a "respectable vestige". A sort of
scientific realism about natural kinds is, perhaps, the "common
core" among the divergent views of modern empiricists, and I
do not see that "common core" to be in discord with the spirit
of Nyäya realism and its theory of laksana or definition.
Putnam says that it is not a defect of dictionaries to give, for
example, colour-samples and stray ("chatty" in Quine's expression)
pieces of information and not to distinguish them from "purely
linguistic" information, for, this is "essential to the function of
conveying the core fact in each case." Nyäya considers giving
synonyms (cf., NS 1.1.15), or samples or other empirical criteria
("cow" is defined with the help of its dewlap, "pot" or "vase" is
defined by a particular shape, kambugrïvâdi)—all part of the act
of definition or giving laksana as long as it helps us to distinguish
the objects or learn the use or application of the word to objects
(vydvrtti or vyavahära).
X
S. Kripke, "Identity and Necessity," in Sehwartz, S. P. Naming, Necessity
and Natural kinds, Ithaca, 1977. Also "Naming and Necessity," in David-
son and Harman's The Semantics of the Natural Language, Dordrecht,
1972/
2
H. Putnam "Is Semantics Possible ?" pp. 111-8 (in Schwartz, S.P.).
No matter in which way we may resolve the problem of the

traditional essentialism (some philosophic problems have hardly
any agreed solution, and such "unsolvable problems are not for
that reason unreal"), it is arguable that the Indian (Nyâya)
definition will not be seriously condemned by moderners. Modern
" essen tialists" may maintain that the definition sentences are
synthetic (being empirically discoverable), but, nevertheless,
necessary. As long as we can argue that there may be some
unique (asädhär'ana) property or characteristic or aspect of a set
of things or objects or items—aspect that is discoverable by
experience but not necessarily by experience—we can formulate
definition or laksana of the term that is supposed to denote
the set of items concerned. Our problem arises when we
raise such questions as whether this unique property is also to
be essential or necessary. Nyâya empiricism claims that the
unique property need not always be the essential property, if
"essential property" means what makes the thing (say, a tiger)
what it is. We may define, following the Nyäya principle, a tiger
by the black-marks on the yellow skin, and a cow by its dewlap.
But certainly it is not believed here that the essence of a tiger is
its black-marks or that of a cow is its dewlap. To wit : Nyâya
does not identify cowness with the property of having dewlap, or
tigerness with that of having black-marks.
By admitting that a unique property may be given by experi-
ence, we have admitted what I call the Nyâya fallibilism. Some
laksatjas are easy to formulate within some given contexts.
Bankimchandra's Kamäläkanta, for example, recognized, i.e.,
defined, or identified, a cow by the particular taste of her milk,
and a lawyer by his gown. (In Kamalâkânta's world no other
person wore gowns). True, such definitions may be fallible, for,
one can always be tricked or mistaken, in the last resort tricked
by Descartes' evil genie or by Putnam's Martians. Putnam lias
argued that it is possible that the cats we see are only Martian
robots and the black-marks on tigers are only black paint; and if
we do discover these facts some day, even then we would not say
that cats are not cats or tigers are not tigers any more, but we
would simply conclude that "cat" and "tiger" have changed their
"meanings". I think Nyâya would readily agree with this point.
But if one has to choose between fallibilism and the impossi-
bility of formulating a laksana or definition, it is not a bad
idea to choose the first. For, the other alternative is a

message of despair about formulating any definition. If we
have to give up definitions, we may be asked to stake the
claim that the words we use in philosophy (or, in any
systematic discourse) have precise meanings more often than not.
If we give up this claim, we may be invited to the game of the
Alice-in-the-Wonderland croquet to roll those concepts with
mobile hoops.
CHAPTER THREE
PROBLEMS OF KNOWLEDGE
AND PERCEPTION
§ 3.1 : AWARENESS AND KNOWLEDGE
It would be useful to begin with a note about translation. Pra-

mäna (along with its possible interpretation as pramä) is a very
important term in Indian philosophy. The sense of the word seems
to be so comprehensive in some contexts and so limited in other
contexts that it defies all our attempts at finding a happy English
translation. The same is true of-another term, jnäna, which is
related to pramâna. I use the term 'cognition'to translate the
Sanskrit jnäna and 'true cognition' to translate pramä. This is,
however, purely arbitrary. I shall often use 'truth' to translate the
Sanskrit pramätva, which is one of the two senses of prämänya and
in this I shall follow J. N. Mohanty. The term pramana will fre-
quently appear here without a direct translation into English,
when it means simply various instrumental causes leading to true
cognition.
It was most probably the Mîmâmsakas who started thinking
seriously about the question : How do we apprehend the truth
pertaining to a cognition which may arise from various sources
and come to us through various means and media ? This was the
central theme of the jriapti section of the Prämänyaväda of
Gangesa.
The above question was quite pertinent to the early Mïmâipsâ
system which sought to establish the indubitable validity and
authority of the. Vedic scriptures. Scriptural statements are only
statements (forms of speech). It was held that cognitions derived

from such statements should be intrinsically true. Thus, it was
claimed by the Mïmâmsâ that speech (or sabdd) has the unique
power of generating cognitions which present objects of the past,
present, and future, gross and subtle, remote and near. Other pra-
mânas, such as perception, do not enjoy this unique power which
sabda (speech) possesses.1 To maintain the absolute validity of
the Vedas it was further claimed that cognitions arising from
speech are to be accepted as true as long as there is no defect (phy-
sical or intellectual) of the speaker concerned or as long as there
is no contradiction (bädha). Thus, falsity of a cognition arising
from a statement is always found to be due to some defect of the
person making such statements.2 Now, as the Mïmâmsaka claims,
if there can be any impersonal statement, or a series of statements,
which does not belong to any person or human being, we can say
that cognitions arising therefrom will be necessarily true. The
Vedic texts are revealed texts, according to the Mîmârpsakas ; they
are a-pauruseya, that is, they do not come from any person. Hence,
the early Mîmâmsakas claimed that the Vedas are eternal, and
cognitions derived from Vedic statements must be intrinsically
true. This philosophical position was later developed in the fol-
lowing manner : All types of cognition are to be accepted as true
unless and until they are proved otherwise. Almost all the schools
of Mïmâmsâ accepted this philosophical thesis and justified it by
claiming that we apprehend the truth of a cognition along with
our apprehension of the cognition itself without depending upon
any extrinsic condition. This, in short, was called the svatahprä-
mänya theory of the Mîmâmsà system. The origin of this theory
can be traced as far back as Sabara.3
Navya-nyâya writers have connected the prämänya theory (i.e.,
the theory of the apprehension of truth) with another theory
called prakäsa (the theory of the apprehension of cognition itself),
and has given a fourfold classification based upon their possible
lli
Codanâ hi bhütam bhavantatyi bhavisyantam süksmam vyavahitatfi viprakrs-
tarn ity evamjâtïyakam artham saknoty avagamayitawnänyat kincanendri-
yam" Sâbara-bhâsya on Mïmâmsà-sûtra 1.1.2
2
Cf. "tasmâd yasya ca dust am karanam yaira ca mithyeti pratyayah sa
eväsamicinah pratyayo nânya ///." Säbara-bhüsya on Mimâtpsû-sutra 1.1.5.
3
Thus, note the following statement : "vipratisiddham idam ucyate vraviti
vitathaifi cetir Sàbara-bhâsya o n Mimämsä-sütra 1.1.5. S e e also the remark
quoted in n. 2, p. 205.
Problems of Knowledge and Perception 205
1
combinations. Following Kumärila Bhatta it is possible to sug-
gest a parallel and equally useful classification of conflicting
theories regarding the apprehension of truth and falsity in Indian
philosophy :
(1) Truth is apprehended intrinsically (svatah), but falsity
is apprehended through some extraneous means (paratah).
(2) Both truth and falsity are apprehended intrinsically
(svatah).
(3) Both truth and falsity are apprehended through some
extraneous conditions (paratah).
(4) Falsity is apprehended intrinsically (svatah), but truth
is apprehended through extraneous means (paratah).
The first view is that of Kumàrila Bhatta as well as of the other
Mimärnsakas. The second view is ascribed to the Sâmkhya system.
Although this view appears to be quite consistent with the Säm-
khya theory of satkäryaväda, according to which nothing can be
newly generated unless it exists potentially in its cause, yet I do not
think this view of prämänya has been explained or explicitly refer-
red to in any available text on Sämkhya. Nor did Kumärila men-
tion the name of Sänikhya when he described this view.2 Mädha-
väcärya, in his Sarvadarsanasamgraha, identifies this view as
belonging to the Sänikhya system.3 The third view belongs to the
Nyäya system. The Naiyäyikas gradually became the formidable
opponent of the Mïmâmsakas on this particular point of the ap-
prehension of truth and falsity of cognitions. This is well eviden-
ced by the writings of Gangesa and his followers. Regarding the
last view, we are again in doubt about its origin. According to
Mädhaväcärya, this is the view of the Buddhists.4 But we do not
know which branch of Buddhism adhered to this view. Kumärila
elucidated this position as follows. Falsity, being an absence of
truth, is not an object (vastu\ that is, not a real item which can
be caused or produced. It is on a par with the fictitious rabbit's
horn (§ 2. 2). Truth, being a vastu (object), should be produced
x
Mlmättisä'Sloka-värttika, codanä-sütra 2, verses 33-61 (Benares : Chow-
khamba, 1898).
2ii
Svato'sätam asädhyatvät kecid ähur dvayain svatah" Ibid., verse 34ab.
z
"Pramânatvâpramânatve svatah Sänikhyäh samâsritâh." Sarvadarsana-sant-
graha (Poona, 1951), p. 279.
Aii
Saugatäs caramam svatafr. Ibid., p. 279.
by some real cause. But Säntaraksita rejected all these four posi-
tions. Kamalasîia, while commenting on Sântaraksita, elucidated
the position of the Buddhist as the fifth possibility.1 According
to this view, both truth and falsity can very well be intrinsic in
one case and extrinsic in another. In the case of our first acquain-
tance (anabhyäsadasä), that is, in the case of a new cognition,
truth and falsity should be apprehended extrinsically. In other
words, we depend on an extraneous condition for confirmation
or disconfirmation of such cognitions. In the case of our repeated
acquaintance (abhyäsadasä), however, truth or falsity of
cognition becomes intrinsically known. For example, because
of bur repeated acquaintance, the truth of such cognitions as
"water will quench our thirst," and falsity of such cognitions as
"the conch shell is yellow," will be revealed automatically with,
the cognitions.
A few wordson the prakäsathQory—the theory of the apprehen-
sion of an instance of cognition—may be in order. The main
problem under consideration here is "How do I know that I
know?" In order to precisely understand what the Indian phi-
losophers meant by raising and suggesting possible answers to
this philosophic question, we should keep in mind that some
philosophers hold cognition itself to be a perceptible object.
However, they differ as regards the nature of the perception of
an instance of cognition. There are yet other philosophers who
maintain that cognition is not at all a perceptible object. We
know a cognitive event through inference.
Sâlikanâtha, an exponent ofthePrâbhâkaraMimânisaka school,
holds that cognition is self-revelatory (sva-prakäsa) in the sense
that a cognitive state in the form *I know this' always reveals not
only the object expressed by "this" but also the knowing subject
as well as the knowledge in question. Thus, 'I know this' becomes
logically equivalent to 'I know that I know this.' On the analogy of
light revealing also the light itself, 'reveal' is claimed to be a reflexive
relation when applied to knowledge. The Vijnänaväda school of
Buddhism also regards cognition as self-revelatory, but they
maintain that a cognitive state reveals nothing but itself. Their
1
Cf. na tu bauddhair esârri catuniâm ekatamo 'pi pakso' bhi§to niyamapaksas-
yestatvät, Tathä hi ubhayam apy etät kincit svatah kincit paratah iti pürvam
upavarnitam." Tattvasantgrahapanjikä under verse 3123, p. 811. For Säntara-
ksita's critique of different prâmânya theories, see Tattvasarngraha (Baroda,
1926), verses 2813-3123 (vol. II). '
philosophical conviction has been, understood to be that there is

no other reality but a series of cognition-moments or cognitive
states. The so-called object of cognition is nothing but the cog-
nition itself appearing (for some metaphysical reason, to be sure)
as the object. Thus, we may say that 'reveal' operates, according
to Vijnânavâda. as a totally reflexive relation (see my Perception,
Ch. 4.) The Advaita Vedänta also holds to the doctrine of self-
revelation of cognition, but I shall not discuss their position here.
ââlikanâtha argues that if cognition is not admitted to be self-
revelatory, then it should not be credited with the revealing of
objects. This would imply that revelation of any object is impos-
sible, and consequently, our vyavahära (verbal behaviour) with
regard to an object or reality would be impossible. Sälikanätha
concedes that under this theory each cognition will be of the
nature of perception insofar as the cognition apprehends itself.
Thus, the classification of cognition into perception, inference,
and so on, operates only with regard to the objects (prameya). All
instances of cognition are perceptually revealed to the cognizing
subject.1
The Nyâya system maintains that cognition is by nature a per-
ceptible object. We perceive a cognitive event by an inner percep-
tion, technically called anuvyavasâya, which arises in the wake of
the first cognition. In other words, if Kx is a cognitive state which
apprehends an object a and K 2 is another cognitive state which
apprehends K1? then K i ^ K 2 . It is, however, conceded that K2
may not happen at all after Kx in some cases if any counteracting
situation (pratibandhaka) develops. This theory is called paratah-
prakäsa, which means that cognition is not self-revelatory. For
the Sâmkhya-Yoga system too, cognition is not self-revelatory.
According to them, a cognitive state is a modification of huddhi,
and it is perceptually cognized by pure consciousness (caitanya)
which constitutes the essence ofapurusa (soul). (See § 3.3)
To oppose the Vijnânavâda school of Buddhism, Kumârila
Bhatta ( with whom the Bhätta school of Mîmâmsâ started) main-
tains that cognition is by nature imperceptible. Kumärila also
supports the theory that cognition is not self-revelatory, but in a
quite different sense. He asserts that a cognitive state is inferen-
1
See Säükanätha, Prakaraqapahcikä (Benaras : Banaras Hindu University,
1961).
tially cognizable only through a property called known-ness

(jfiätatä), which an object of cognition acquires when cognized.
The prakäsa theory of the Indian philosophers seems to contain
some valuable philosophical insights, and a detailed analysis of
different theories regarding this problem may be as useful as the
analysis of the prämäryya theory. However, I must add here,quoting
a remark of Professor Mohanty (see Preface to his Gangesa s
Theory of Truth) that one must also "know where to draw the
line between what is living and what is dead, a task which has yet
to be done with regard to much of Indian philosophy."
The Prâbhâkaras maintain that all cognitive states are necessari-
ly true. Following Rämänujäcärya, Mohanty discusses three differ-
ent kinds of truth, yäthärthya, prämänya and samyaktva. The
test of samyaktva is the absence of contradiction in practice. The
revival of memory (an episode of remembering a past experience)
is not accepted as &pramä (a piece of knowledge) in any Indian
school of philosophy except the Jaina school. I shall deal with this
problem in Ch. 3.4. Some historical remarks might be added in
this connection. The almost proverbial ambiguity in the use of
the term prämänya by the ancients might have been responsible,
to some extent, for the attitude of the philosophers who refused to
accept memory-cognition as a separate pramäna. Pramäna some-
times meant an authoritative source of knowledge, sometimes
simply an authority. Some kind ofindependence is usually attached
to the notion of authority. A memory-cognition or remembering
is nothing but a reproduction of some previous experience, and the
causal conditions which produced the previous experience are not
necessary for this reproduction. Thus, if the previous experience
was an authority, the memory reproduction would only be a copy
of such authority. In other words, "authoritative" will only be a
transferred epithet when applied to the memory reproduction of
a previous experience. Realization of this sort of difference bet-
ween memory and direct experience prevented the ancient writers
from calling memory 'à pramäna. Later on, it crystallized simply
into a matter of linguistic decision. Thus, Väcaspati makes the
right point when he says that the relation of word to its meaning
is conventionally established through public usage (Jokavyava-
hära), and since pramäna is not used to include memory (smrti)
we should not call memory & pramäna.1 I have noted that in the
1
See Nyäyavärttikatätparyatikä, Kashi Sanskrit Series, 24 (Benaras, 1925),
p. 21, lines 3-6.
Jaina tradition memory is accepted as a pramäna. The Jaina

philosophers began discussions of the pramäna theory at a rela-
tively later period, and they were right in exposing the vulnerabi-
lity of the position which maintains that memory-cognition cannot
be claimed to be true (pramä). If perception and inference are
regarded as true, because of their agreement with facts (samväda),
then a memory-cognition, when it agrees with fact, cannot be
denied that privilege.1 It should also be noted that those philoso-
phers who were reluctant to accept memory as a pramä, implicitly
admitted that memory can very well be correct when it reprodu-
ces a true experience. In any case, a more detailed and critical
examination of the controversy over whether or not memory is a
pramäna may give us more insight into the exact meaning of the
term pramäna{§ 3.4 following).
The Prabhâkara account of error makes a very interesting study
in contrast with the Nyäya theory of error. For the Präbhäkaras
there is no error in our cognitive states. The so-called error con-
sists in our non-apprehension of the fact that two cognitive events
(or the two cognized objects) are not related {asarrisargägraha).
Thus, to mistake is to miss something, namely, unrelatedness
(asamsarga), and not to take or grasp anything wrongly. Thus,
when we mistakearope for a snake, we have two cognitive states:
the one which refers to the rope is presented by perception, and
the other which refers to the snake or snake-ness is presented by
memory. Error consists in our not seeing their difference, Even
when a conch shell appears yellow to one's jaundiced eyes, the
yellow colour actually belongs to the disease, that is, jaundice in
the eye, and not to the conch shell, which lies outside our body.
Our error, in this case, consists in'our missing this disconnection
of the two objects, conch shell and the yellow colour. This
account of error is opposed by the Nyäya theory. The Nyâya takes
a positive attitude and maintains that we directly perceive the
wrong thing which is not really present in our field of perception.
Our wrong perception may be aided sometimes by memory, and
sometimes by the diseased or defective sense-organ. But the
resulting cognitive state is one whole where we actively connect
(or construct a connection between) two different objects, instead
of passively failing to notice their unrelatedness or disconnection.
*See Hemacandra, Pramanamïmàmsâ (Ahmedabad, 1939), pp. 33-34.

It may be that for the Nyäya theory the model was the typical
rope-snake example, where there is a psychological basis (viz.,
intervention of memory or the snake-idea) for error, while for the
Präbhäkara theory the model was the case of a defective or disea-
sed sense-organ, where there is a physical basis for error. Thus,
for the Nyäya it was, to begin with, the misconstruction of a
relation, whereas for the Präbhäkara it was a mechanical failure
obscuring the factual disconnection. But both theories were ex-
tended and modified to cover all patent as well as recalcitrant
cases of error.
Mohanty has given a penetrating explanation of the Nyäya con-
ception of jnäna.1 The postulate of nirvikalpaka perception, that
is, prejudgmental cognitive state, was associated with a number
of important philosophical problems. The controversy over this
seems to have started with Dignäga (ca. 500 A.D.), although the
definition of perception stated by Gautama (ca. 100-200 A.D.) in
Nyäya-sütra 1.1.4 was, under the simplest interpretation, applic-
able more directly to our sense-perception rather than to our
perceptual judgment-The peculiar interpretation of Nyäya-sütra
1.1.4 offered by Väcaspati Misra (ca. 900 A.D.) was at best an
innovation, as evidenced by Vacaspati's own remark that although
the explanation he offered was not to be found in the earlier com-
mentary of Vätsyäyana ( ca 350 A.D.) or in the sub-commentary
of Uddyotakara (ca. 600 A.D.) he had followed his own teacher
(guru) Trilocana in this matter. 2 Thus, the clear-cut classification
of perception into non-judgmental (nirvikalpd) and judgmental or
constructive (savikalpa) was an aftermath of Dignäga's criticism
of the Nyäya theory. For Kumârila Bhatta (ca. 625 A.D.), a pre-
judgmental cognitive state is the direct grasp of the reality with-
out differentiation and conceptualization. This cognitive event
is held to be psychologically prior to our perceptual judgment.
The Nyäya-Vaisesika writers like Väcaspati and Srïdhara follo-
wed mainly Kumärila in their exposition of the pre-judgmental
state. But in Navya-nyäya, the status of this non-judgmental
cognitive state is reduced to a logical construction. According
to Navya-nyäya, the non-judgmental cognitive state is never
revealed to us, that is, never apprehended by our inner per-
*See J. N. Mohanty: Gangers Theory of Truth, pp. 25-37. See also
Matilàl : The Navya-nyäya Doctrine of Negation, pp. 6-21.
2
See Nyäyavärtiikatätparyaiikä(Benzres, 1925), p. 133, lines 9-15.
ception or reflective awareness (anuvyavasäya). Thus, its

content is not describable in ivords because, unless a cognitive
state is revealed to us ( by inner perception or by any other
means), there cannot be any vyavahära or verbal behaviour
(such as communication through speech) with regard to it.
In simple language, although this cognitive state has a con-
tent (because a çontent-less cognition would be meaningless for
Nyâya) it is not expressible in words. Again, the content of this
cognitive state does not admit of any philosophic analysis because
it is only our inner perception {anuvyavasäya) that gives us the
datum for such an analysis, and it has already been stated that
the non-judgmental cognitive state does not admit of such inner
perception. Since all this amounts to saying that the non-judg-
mental cognitive state is never known to us by our inner perception
one can very well entertain doubt with regard to its actual occurr-
ence. To answer this doubt, Navya-nyäya gives the following argu-
ment: A constructive or judgmental cognitive event, namely, a
perceptual judgment of the form * that which is characterized or
qualified by/?,' presupposes a cognition of the qualifier/?. This rule
is evidently based upon the common experience that in order to
know something white one must first cognize what white colour is.
Thus, a perceptual cognition which describes something as a pot,
that is, which in Nyäya analysis attributes pot-ness to something,
presupposes a logically prior cognition of pot-ness. If this logical-
ly prior cognition of potness is not presented to us by any other
means, we have to admit the hypothesis of its being presented by
perception.1 In the case of our first (perceptual) acquaintance
with a pot, the required cognition of pot-ness is not present at
the previous moment. Only the conditions for the preception of
pot-ness as an object can be said to be present in the given situa-
tion. Therefore, our cognition of the attribute pot-ness must be
a perceptual cognition in this case.
There is, thus, a shift of emphasis in the Navya-nyäya theory
of non-judgmental perception. This is not exactly the same as
what is called älocanajhäna in Kumärila's theory. We arrive at
this cognitive state by a process of logical construction described
above. Following a similar logical argument, it is proved that
1
Cf. Gangesa : "prâthamikam gaur iti pratyaksam. jnänam janya-visesana-
)Mna~janyarix janya-viàista-jnânatvât anumïtivat." Tattvacinâmani, Part I,
Bibliotheca Indica (with Mäthuri) (Calcutta, 1888), p. 817.
this non-judgmental perceptual state is by nature not amenable

to inner perception or reflective awareness {anuvyavasäya), and
thus its content is never revealed tous. This position is quite con-
sistent with the Nyâya theory of knowledge, according to which,
not all our cognitive states are necessarily revealed to us, that
is, cognized by us. 1 Thus, a cognitive state is perceptible (prat-
yaksayogya) but may not be perceived if all the causal conditions
are not present, just as a pot is a perceptible object but may not
be perceived if it is hidden from the eye. The content of a non-
constructive perceptual state can only be inferentially established
because such a perceptual state itself is, in Navya-nyâya theory,
a matter of logical presupposition, rather than a discernible
psychological event. Besides theelaborate Navya-nyâya argument
to show that a non-constructive perceptual state is never percei-
ved, the very attempt to accord a logical status to this perceptual
state springs from the feeling that the existence of such a per-
ceptual state can never be proved by our inner perception or
reflective &\varenes$(anuvyavasäya). In other words, the non-per-
ception of such a perceptual state follows from its very definition
in Navya-nyâya. When it is explained in this way, we can be in a
position to answer the charge of obscurity brought by Mohanty
against the Navya-nyâya "metaphysician." He remarks: "Though,
however, the contents of savikalpa become objects of reflective
awareness, the contents of nirvikalpa do not,—a fact, for the
explanation of which the metaphysician could only postulate
some obscure pratibadhya-pratibandhaka relation" (p. 34). In
spite of some obvious difficulties involved in the notion of a non-
constructive cognitive state, I believe that one can give a reason-
ably good explanation of this notion in Navya-nyâya following
Gangesa's remarks in the Nirvikalpakaväda of his Tattvacintä-
tnani.
It might be useful to add a few words further to clarify the
Navya-nyäya concept of a construction-free perception. It is con-
ceded that there are certain simple irreducible objects which are
perceptible butwhich can only be perceived through our construc-
tive perceptual state. Thus not all elements which form the
content of our constructive or judgmental perception are necessa-
rily perceived at our non-constructive state. An abhäva, that is,
1
Thus, cf. Gangesa's remark : "vitter avasyavedyatvâbhavenânavasthâvîga-
mätr Ibid., p. 798.
the absence of a pot, which is an irreducible entity in the ontologi-

cal schemeof the Nyâya-Vaisesika school, is perceived only in our
constructive cognitive state. It can never be an object of our non-
constructive perception.1 It is even conceded that constructive
(savikalpa) and non-constructive (nirvikalpa) states should better
be explained as qualificatory (saprakärakä) and non-qualificatory
(nisprakäraka) states respectively. Gangesa seems to allow that
even the same cognitive state may be partly qualificatory and
partly non-qualificatory, that is, partly constructive and partly
non-constructive.2 Thus, the same visual perception of a cow is
qualificatory from the point of view of the individual cow (which
is perceived) because cow-ness orcow universal appears as quali-
fying the cow; and it is non-qualificatory from the point of view
of cow-ness (which also forms the content of perception) because
nothing appears as qualifying cow-ness. The property of being
constructive in character and that of being non-constructive in
character are not two mutually exclusive properties like two
natural class characters, such as cow-ness and horse-ness. In
other words, the same individual cannot be both a cow and a
horse, but the same cognitive event can be both in part construc-
tive and in part non-constructive.3 From another point of view, the
non-constructive cognition is said to be neither true nor false be-
cause truth and falsity, in Nyäya theory, can only apply to a
constructive state. This falsifies the idea that the Nyäya theory
of truth needs only an agreement or correspondence between the
content of a cognition and the reality. The content of a non-
constructive perception may very well be said to agree with the
reality, but for Navya-nyäya such a perception is neither true
nor false. The question of truth and falsity arises when and only
when the cognitive state is propositional, that is, qualificatory. If
the qualifier which appears in the cognition as qualifying some
thing x also qualifies that x ontologically or in reality, then the
cognition is true; otherwise it is false.
Mohanty's brief reference to the "total unanalysable visayatä"
pertaining to a non-erroneous cognition is also very important
x
See Gangesa : "ata evàbhâve na nirvikalpakam...." Ibid., p. 822.
2
See Gangesa : ilgaur iti savikalpakam api gotvättise nirvikalpakam eva tatra
prakäräbhänät bhäne vänavasthänirvikalpakäsiddhisca*. Ibid., pp. 823-4.
3
See Gangesa : "ata eva nirvikalpakatva-savikalpakatve na jâtî cäksusatvä-
dinä sankaräpattehy Ibid., p. 823.
(n.5 p. 36). By introducing two distinct types oîvisayaîa (content-

ness), the Naiyâyikas tried to mark off the content of error from
that of a true cognition. I have tried to explain this problem in
1
t some detail elsewhere.
Mohanty's analysis of Gangesa*s definition of truth is brilliant.
In a previous publication he succinctly remarked that truth,
according to Gangesa, is a "hybrid" entity.21 was not quite sure
about the exact significance of this remark, and accordingly, I
expressed doubt about its validity.3 But ia the present book
Mohanty has made his point nicely (pp. 43, 45, 74-75). I wish to
add only one comment here.
Gahgesa's conception of truth contains both epistemic and
ontological components, and this, according to Mohanty, is its
main difference from the semantic concept of truth of A. Tarski.
While I am now quite in agreement with this view, I wish to point
out also that Tarski's notion of truth primarily applies to sen-
tences and is always to be related to a specific language.4 But
Gangesa's concept of truth (pramätva) applies directy to parti-
cular occurrences of cognitive events {jnäna). Although this point
seems obvious, it needs to be stressed because the well-known
antinomy of the liar (i.e., Epimenides' paradox) can be shown to
be quite meaningful in Tarski's semantic conception of truth. In
a language system we can always find some device by which we
can construct a sentence whose subject term will refer to the
sentence itself. And thus, with Tarski's postulate of object-lang-
uage and meta-language, we can try to solve the paradox of the
liar. However, under Gangesa's theory of truth, I am not sure
whether we can even formulate the liar paradox successfully. As
long as Gangesa follows the Nyâya-Vaisesika system and main-
tains that no cognitive state is self-revelatory, we cannot think
of a cognitive state the content of which will consist of the same
cognitive state itself in order that we may attribute truth and
falsity to it; that is, we would cognize it at the same time as true
x
Cf. The Navya-nyäya Doctrine of Negation, pp. 27-29.
3
J. N. Mohanty, "On the Nature of the Prâraânya Theory," Our Heritage
[Calcutta], 8 (1960), pp. 43-47.
3
Op. cit., p. 17, n.
*See A. Tarski, "The Semantic Conception of Truth," in H. Feigl and
W. Sellars, eds., Readings in Philosophical Analysis (New York : Apple-
ton Century-Crofts, 1949), p. 53.
or false. Another way of formulating the liar paradox would be

to imagine a possible cognitive state which can be of the form
"all (my) cognitive states are false." But here too, in order to
avoid the self-reveiatoriness {sva-prakäsa) of this cognition, we
have to exclude this present cognitive state from the exten-
sion of the term "all (my) cognitive states." Gangesa would
perhaps say that our perception with regard to all cognitive states
refers to all cognitive states only through the perceptual (or ap-
prehensionai) connection of the class character jhänatva (i.e, the
essential property for being a cognitive state). But such, an
indirect self-reference does not apparently go against the Nyâya
theory of the non-self-revelatory character of all cognitions
(paratah-prakäsa).1
In Gangesa's theory, a cognitive state which relates a to some-
thing b is true if and only if a is actually (i.e., ontologically)
related to b. Mohanty distinguishes the ontological component
from the epistemological component in this definition. But the
so-called ontological component, namely, 'a is actually related toè,'
which refers to an ontological situation, seems to satisfy the mini-
mum condition of adequacy which any account of truth must
satisfy. In other words, this component preserves what is perhaps
a tolerably clear sense of the much-debated formulation (by
medieval European philosophers) : Veritas est adaequatio rei et
intellectiis (truth is the correspondence between fact and intellec-
tion).
§ 3.2 : PERCEIVING AND MISPERCEIVING
Dr. Samuel Johnson attempted to demonstrate the existence of

a stone by kicking it. G. E. Moore, in his Cambridge lectures,
1
Cf. Gangesa's remark : ^jnânaiji prameycun eveti vyäptijnänatp jnänatva-
pratyâsattyci svavisayam, ïsvarajnânam ahetukatvena sarvavisayatvàt sva-
prakäsam." Tattvacintamani„ Part I, p. 798. Thus, Gangesa allows self-
reference (svavisayatva) of a cognitive state provided the cognition is about
all cognitive states including itself and the predicative attribute is anything
but truth or falsity. Truth or falsity of a particular cognitive state is not
apprehended by the same cognitive state, nor even when that cognitive
state itself is apprehended. This is the main contention of the paratah-
prätnänya theory of Gangesa. Fora somewhat closer formulation of the liar
paradox in the Indian context, see Bhartihari, Vàkyapadiya, Chap. Ill,
3/25,27,28.
raised his hands and said, "Here is one human hand, and here is
another. Therefore, two human hands exist. And therefore, two
physical objects exist." Both Johnson and Moore were concerned
with a very persistent philosophic problem: When we are said to
be seeing an external thing, do we perceive what it is that we think
we perceive? One may ask further whether the external things are
at all what we tend to believe that they are, and further whether
there are any external things at all, These problems exhibit them-
selves embarrassingly in almost every discussion which is philoso-
phic in nature. My purpose here is to discuss some theories of
Indian philosophy concerning them.
Modern philosophical studies of perception have often centered
around a controversy over what are called ''sense-data." The
sense-datum theory had its heyday in the past decades, although,
much like logical positivism, it is now on the decline. But a re-
ference to it can hardly be left out in any modern discussion of
perception, and what I will have to say about the views of Indian
philosophers here may, from time to time, be reminiscent of
points made by both" the sense-datum theorists and their critics.
This is so because sense-datum theory, like positivism, reacted
against a form of idealism. The positivist, for example, has tried
to imagine an ideal "protocol language" which is evidentially
prior and hence forms the basis of scientific empricism. Most
Indian theories of perception were also attempts to combat a
radical form of idealism which theMaitreya-Asanga-Vasubandhu
school presented. The fact that the two versions of idealism differ
does not always cause a corresponding difference in the philoso-
phies of the anti-Idealists. Moreover, perception, illusion, and
hallucination are global enough to form a common ground on
which philosophers of different persuasions and traditions may
converse and debate.
It is difficult to find an exact counterpart of the term "sense-
datum'5 in Sanskrit, although it is true that Indian philosophic
reflection on perception and knowledge began as it did in the
West, with questions about the reliability of the senses. The main
problem these questions lead to is that of justification of our
belief in the existence of the external world, although in the
Indian context it takes on a slightly different shape with some
l
See next section § 3.3.
different underpinning. But first let us see how the notion of

sense-data was introduced in the Western tradition.
From the basic idea that perceiving involves sen se-experience,
using many complex and involved arguments, of which the
"argument from illusion" is the most important one, it is
shown that there exist disparities between the way things appear
and the way they are. In this way it has been claimed that we do
not perceive a physical object, like a table, but that what we
perceive is a different sort of "thing" altogether (for example,
on various occasions, various sizes and shapes of a table), which
they call sense-data.1 With the introduction of this new term
the sense-datum philosophers introduced a way of approaching
the problems of perception which has generated a good deal of
modern philosophic literature.
The meaning or connotation of the term "sense-datum" has
not. remained unchanged in the hands of modern philosophers.
What is generally agreed is that a sense-datum is "not a physical
reality." But while for G. E. Moore2 and his followers a sense-
datum is the immediate object of perception which may or may
not be identical to a part of the physical object, for C. D. Broad
and others it is the immediate object in perception, taken to be
nonphysical.3 The denotation of the term, however, includes,
in any case, such items as the elliptical appearance and the
circular appearance of a penny, as well as such things as mirage-
appearance in the desert or the double-moon appearance.
What the Sanskrit philosophers call pratibhâsa (appearance)
is not strictly a sense-datum in many respects, for, they did not
pay much attention to the variability of the shapes or appear-
ances or components in the perception of the same physical object,
although they felt that the problem of explaining nonveridical per-
ception, hallucination, and error was one of the central focuses of
their philosophy. In a perceptual illusion, say a mirage-illusion,
they noted the duality or disparity between what appears in a
perceptual experience (that is, a pratibhâsa) of water, and what we
1
G.A. Paul, "Is There a Problem About Sense-Data ?" Aristotelian Soc.
Suppl. 15 (1936): 84. '
2
George Edward Moore, "Some Judgements of Perception," in Philosophical
Studies, (New York : Harcourt, Brace & Co., 1922), pp. 231-2.
3
C.D. Broad, Scientific Thought (London : K, Paul, Trench, Trubner &
CO., 1923), p. 244.
are confronted with in the situation, the interplay of hot air and
the sunrays.1 This appearance of water in a mirage-illusion may
be a sense-datum, but the perceptual consciousness here already
involves a judgment, an interpretation of the datum. The philoso-
phic motivation for introducing the sense-datum or even the
"appearance" language was the Cartesian search for certainty,
for the data are supposed to be incorrigible and indubitable.
They are also called "the given" because they are felt to be
independent of our judgments. They are self-evident and our
beliefs, to fulfil their claims to be knowledge, must be based
upon them in some way or other. But the pratibhäsa (appearance)
is not an indubitable datum, for the Indian epistemologists
argue that as long as it is describable in language it becomes the
datum interpreted, or taken to be something F (for example,
water). So our "F-appearance" in a state of (perceptual) con-
sciousness implies that something is being identified as an F, or
the property of being an F is attributed somewhere. If what we
seem to cognize on a particular occasion is expressible in langu-
age as "this is F " or "this is an F" then the cognition in question
is said to have an "F-appearance." The dispute among the Indian
philosophers centered, to a great extent, on the exact (ontological)
status of the "F-appearance", mainly the "F-appearance" of
what we call a nonveridical perceptual experience (such as, a
mirage-illusion of water).
The word "älambana" is a flexible term in Sanskrit. It is not
the sense-datum, but rather the "foundation" or "support" of a
(perceptual) sensory experience. The Buddhists call it a pratyaya
(causal condition) on which the arising of a state of perceptual
consciousness depends. For example, a visual perception depends
upon what may be called the visible {râpa). This is enough for
the Buddhist to call it a sort of "causal" dependence.* But it is
not depended upon in the way the visual organ (the eye) is depen-
ded upon for the arising of a visual perceptual state. It is depen-
ded upon by way of being its object (visaya), to use, again, an
already problematic term. The Abhidharmakosa notes thefollow-
^ e c Vätsyäyana, Nyâyabhàsya 1.1.4.
2
For the Buddhist use of the term pratyaya as "causal condition" see § 4.2.
Also see Vasubandhu's Abhidharmakosab/wsya, Varanasi (éd. S warn i
Dwarikadas Sastri), Î97Î, pp. 279-82.
ing distinction between a visaya and an älambana} A visaya is

the potential object of a perceptual consciousness, in the sense
that when a cognizing state arises it has to select its visaya. For
example, the eye-consciousness has to select the visible (râpa).
But a particular älambana is actually depended upon, it is &rüpa
which has been causally responsible, whenever a particular state
actually arises. For all other practical purposes, the distinction
between them may be suspended.
My perception of red-colour depends upon red-colour as its
älambana (objective ground), and my perception of the table
depends upon the table as its älambana. But if my perception of
red-colour is a product of some hallucinogenic drug, would I
still be justified in calling the red expanse its älambana—its
objective support? And if the perception of the table is due to
some neurophysiological trick in the brain would it still have an
alambana'! Apparently a class of Indian philosophers was inclined
to say yes! As long as there is an appearance (pratibhäsa) (for
example, red-appearance or table-appearance) it must be rooted
in an älambana (objective ground). In fact, they put forward
the following thesis : älambana = pratibhäsa. In other words,
älambana and pratibhäsa are only two ways of referring to the
same thing. The implications of this doctrine are manifold, as
we shall see presently.
The pramäna theory of the Indian philosophers, much like the
sense-datum theories of Western epistemologists, seeks to ground
human knowledge in a mode of experience that is immune to
such failures as error, illusion, and hallucination. If there were
an indisputable way of distinguishing veridical perception from
the nonveridical ones, the problem of knowledge would be easily
solved. However, in view of the embarrassing lack of an agreed-
upon device, the idealists argue (and, in fact, Vasubandliu has
argued)2 that errors, hallucinations, and dreams provide paradig-
matic examples in which an "object" {visaya or artha) appears
but is not "out there." The "object" in all such cases is at best
located in the mind, and at worst is nonexistent (asat) as external
reality. From this point, it is further argued that what appears in
every state of our consciousness is likewise mind-dependent or
1
Abhidharmakosabhâsya, ibid.
2
See Vasubandhu's Vijnaptimätratäsiddhi, Vithsatikä Kärika, first verse. See
next § 3.3.
internal to consciousness itself, there being no justification to

suppose that it is rooted in-an external object (älambana). Thus,
Vasubandhu reached his philosophic conclusion which was dev-
astatingly simple: The so-called external world is only a creation
of the mind; the " stuff" of the world is made of consciousness
only (vijnäptimätratä). What is emphasized mainly here is the
essential dependence of what "appears" in a state of conscious-
ness upon that consciousness itself. And this finally leads to a
radical form of idealism (from which a short step would be
solipsism) : The world around us is the world within us. 1 I shall
discuss this thesis in more detail in the next section § 3.3.
This is one extreme where älambana (objective ground) is not
distinguished from thepratibhäsa (appearance). There is another
extreme view of a different kind where again älambana is not
distinguished from the pratibhäsa. Embarrassed by the persistence
of nonveridicai perception, the realistic wing, represented by the
Prâbhâkara Mïmâmsakâ wanted to combat scepticism, idealism,
and solipsism—all in OIIQ blow—by denying completely the pos-
sibility of error or illusion or even hallucination. According to
them, all perceptions are veridical; it is an error to think that
there could be any error. The so-called perceptual error is ex-
plained by them as a fusion or confusion of two different and
distinguishable cognitive states : one is a memory state while
the other is a perceptual state. In the usual mirage-illusion, the
appearance of water actually belongs to the memory state (for
we have already experienced water on many previous occasions),
while the appearance of "this" or "there [it is]" belongs to the
perceptual state. Thus, pratibhäsa is also the älambana here, for
x
The situation is not very different with the sense-datum theorists in the
West, although very few philosophers today would go as far as Vasuban-
dhu would ask us to go. Thus, W. H.F. Barnes criticizes, "...once the
sense-datum theory is developed in the form stated above, it follows that
even if physical objects exist, they are never present in perceptual experi-
ence; and it becomes an open question whether they have any existence
at all." ( 'The Myth of Sense-Data," Proceedings of the Aristotelian Society
[1944-1945], included in R. J. Swartz's Perceiving, Sensing and Knowing
(New York : Anchor Books, 1965), p. 142. Most modern philosophers
would like to maintain a 'double-standard' that would enable them to eat
their cake and have it. Thus, it is claimed that when 1 "directly see" the
sense-datum, I also "see", in (to use Moore's term) a Pickwickian sense,
the physical object.
the water-appearance in memory is rooted eventually in the

actual water experienced. This was the position of the Prâbhà-
kara Mimänisakas who were realists or anti-Idealists. They cut
the Gordian Knot of idealism and solipsism in this amazingly
simple way.
Let us go back to the Yogâcâras. The philosophic argument
of the Buddhist Idealist is something like this :
A cognition or a state of consciousness does not need a
ground or älambana external to itself (witness either hallu-
cinations or dreams). What is possible in one case must also
be possible in every other case, for they are merely states of
consciousness in essence. Thus, the thesis about the external
world is: the external world does not exist, or if it exists, we
have no way of knowing that it exists,
It is instructive to understand the Sauträntika Buddhist position
in this context. For, the Sauträntika seemed to have allowed the
possibility of the external world which the Yogâcâra Buddhist was
unwilling to allow. In this connection we have to discuss also what
is known as the Buddhist theory of atoifiism, paramänuväda.
Atoms were posited as something like the data of sensory per-
ceptual consciousness. If it is the datum of eye-consciousness
based upon visual perception, it is given the blanket name râpa
(the visible). Similarly the data of other sense-experience are
identified : smell, taste, touch, and sound. The material object,
under this view, is a fictional construction out of these sensory
"atomistic" data. This view seems to be close to phenomenalism in
the West, although caution is necessary in order not to forget its
peculiarities. The data (the atoms) are, as a Sauträntika would
put it, "substantially real" (dravya-sat), whereas the material
objects, such as the pot or the table, are only imagined to be
real (samvrti-sat). This can be compared with the acceptance of
basic units or primitives in logical atomism, out of which the rest
of the world can be constructed. This system is basically pheno-
menalistic in the sense that the basic units, these atoms,
are phenomenal elements rather than physical elements. The
atoms are not, one must note, the material atoms of the Vaisesikas.
Of course, the claim is that these basic units are "obtainable"
or "perceptible" individuals, in the sense that they cause per-
ceptions and are cognitively immediate. This claim is probably
indicative of an underlying philosophical attitude, a desire to

show that nothing beyond the phenomenal need be countenan-
ced in order to explain everything including the physical things
—a further desire to show that if the first claim is successfully
substantiated the so-called physical world, as we commonly
understand it, can at best be dispensed with as non-ultimate or
non-final, and hence, it does not merit the importance or value
we usually attach to it. Very few Western phenomenalists would,
however, make such a value judgment about the physical world.
The choice of a phenomenalistic basis is usually argued for in the
West, as far as I know, on the ground that the phenomenal by
its very nature comprises the entire content of immediate
experience and therefore all knowables must eventually be
explicable or analyzable in terms of phenomena. However,
analyzability or explicability of a complex concept can
hardly be a criterion of its falsity or even comparative unreality.
Rather, what is claimed is epistemological priority or proximity
and immediacy for the phenomenal, and consequently, its indu-
bitability over the other, that is, the set of constructed elements.
The Sauträntika argues that the important difference between
these two classes of entities, the substantially real atom and the
gross object, is that the function of causality is assigned to the
first but not the second. An entity which is only imagined to be
real cannot really cause anything. We cannot ride an imagined
horse.
The atomicity of the Sauträntika sensory data, I repeat, should
not be confused with that of the material atoms of the Vaisesikas.
These data are called "atoms" most probably because of two
reasons: (1) their subtlety and uniqueness to each occasion, and
(2) their unanalyzability into further data. The Sauträntika also
admits that these atomic data are in perpetual flux. Moreover,
although each atomic datum is coordinated to some sense-organ,
it is by no means the content of any sense-experience; rather it is
said to determine causally the content of such experience. Thus,
it would probably be a mistake to assimilate the Sauträntika
view into a form of phenomenalism (as has sometimes been
suggested).
Let us ponder again over the relation between the Western
sense-datum and the atomic datum of the Sautràntika. Most
sense-datum philosophers agree that sensing is a from of know-
ings tind that a sense-datum is what we know immediately. It is

also argued that a sense-datum in some sense must "belong to"
a material object and it is a central problem of epistemology to
determine or explain in what manner this takes place. The
"causal theorists" (like Russell and others) claim that we make
a causal inference and are thus led from sense-data to material
objects. The phenomenalists, such as Mill, would regard material
things as permanent possibilities of sensations. Yet H. H. Price
has argued that a phenomenalist must hold that sense-data are
neither mental nor physical, and that they are not caused at all
and they are not even real in the ordinary sense of the
word.1 Price continues :
According to him [the Phenomenalist] we must simply take
the sense-given continuum as a going concern. There it is,
and all statements in which material things and events are
mentioned, are ultimately statements about it—about the
manner in which it does or could develop itself, whether now
or in the past or in the future.
According to Price, the phenomenalist is right in rejecting the
idea that sense-data are causally dependent upon the thing, (that
is, a table) as their "source." For, if by "thing" we mean the
"complete thing," then this complete thing is a combination of
the family of sense-data and the physical occupant of the parti-
cular space, and thus it would involve the absurdity of saying
that A (the sense-datum) is causally dependent upon AB (the
complete thing, that is, family of sense-data and the physical
occupant). If by "thing" we mean, however, the physical occupant
only, then Price allows that the table or other physical occupant
may well be the remote cause of the sense-data composing the
family. In the same view, it can be claimed that sense-data are
also causally dependent upon the organism of the sentient. The
atomic data of the Sautrantika, however, are stated to be indep-
endent of the mind or consciousness. They are not mental, and it
would be also difficult to call them physical. But they are
claimed to be "external" to consciousness. Certainly they are
not caused by the material object, the table. Rather it is believed
that they cause the so-called appearance of the false table, the
"material object." Thus, the Sautrantika disagrees with an
*H. H. Price, Perception (London, 1932).
imporant part of the thesis of the Causal Theorists: namely. '*M

(a material object) is present to my senses" is equivalent to "M
causes a sense-datum with which I am acquainted." Besides, the
Sauträntika believes that the existence of the extramental reality,
that is, the atomic data, is only inferable from the appearances
(pratibhasa) of the gross material object in our perceptual con-
sciousness. For (a) the atomic data must have caused the
arising of this perceptual consciousness; and (b) by so causing the
perceptual consciousness to arise, they have also caused indirectly
the appearance of a gross material object, which is a mere
appearance, only imagined to be real.
If we have followed the Sauträntika argument so far, we would
be in a better position to appreciate and understand Dignäga's
arguments in his Examination of the Alambana, where he rejects
the view that the älambana is something external to consciousness.
The opponent of Dignäga, presumably a Sauträntika, has argued
thftt these five kinds of atomistic data would act as the älambana
to give "causal" support to the ûvo kinds of perception due to the
five kinds of senses. This "atomist" (Dignäga'sopponent) concedes
that what appears in consciousness, or what constitutes the
appearance (pratibhasa or äkärä) in consciousness, is different from
these atomistic data. For, after all, a perceptual consciousness is
described as that of a table or a pot (and so it refers intentionlly to
a pot or a table). The problem before the atomist is this :
although a visual perception can be "causally" dependent (as
its älambana) upon the atoms of the visible (rüpa-paramänü),
the resulting cognition does not have atom-appearance, for the
atomic data are impartite, discrete, and many, while what
appears in consciousness is a unitary object, a table—also a
gross object.
Dignäga rejects this view on the following ground. 1 An älambana
of a perceptual consciousness must fulfil at least two conditions:
(1) it is "causally" responsible for the arising of that piece of
consciousness (tat-kärana); and (2) it is also what constitutes the
appearance (pratibhasa) of that piece of consciousness. In other
words, Dignäga supports the basic position to which I have already
referred. What appears in consciousness is also non-distinct from
1
Dignâgaïs Älambanapctriksä (Tib.) ed. with Sanskrit reconstruction by
Aiyaswanii Sastri (Madras Î942), verses 1-8.
that in which that piece of consciousness is objectively grounded

(pratibhäsa^älambana). Now, the so-called atomistic data may
fulfil the first condition, but not the second. Hence, they
cannot be called the älambanaofs. piece of consciousness. The
"phenomenal" objects, that is, the objects which are only "imagin-
ed to be real" (samvrti sat), such as a table, may constitute the
appearance {pratibhäsci) of a piece of consciousness, but they can-
not be "causally" depended upon (as & pratyaya) for the arising
ofthat consciousness. For remember, you cannotgeta sprout out
of an imagined seed ! (See § 3.3)
From Sântaraksita we learn that Bhadanta Subhagupta1 tried
to support the Buddhist "atomistic" theory by arguing as follows:
A gross body, such as a table, may be thought of as an atom cluster.
The Buddhist doctrine of momentariness requires that these
atomic data emerge at one instant (moment) simultaneously and
without gaps (and disappear at the next moment). Just as (a
follower of Dignäga agrees that) the instantaneous (momentary)
object such as a pot-moment emerges and perishes instantly,
making room for a new and similar pot-moment at the next
moment, and thus, as a result, we seem to perceive a pot that is
supposed to persist through time, similarly the data called the
'atom-visibles' {rüpaparamänu) arise in space together and without
seeming gaps, and this generates our seeming perception of the
generic nature (sämänyo) of an object, that is, a gross physical
body, a pot, that seems to have parts, and hence, is divisible (that
is, extends in space).
A modern example would be that of a cinema show where dis-
tinct films of different postures of a horse running are run before
our eyes fast enough to generate the illusion of continuous motion-
picture of a horse running. The modern photography also pro-
vides a better example of spatial continuity or spatial extension,
where discrete dots or points on the screen being put together
without gaps create the picture of an extended material object,
such as a table. Subhagupta's argument may thus be rephrased as
follows: Just as a follower of Dignâga must accept the appear-
ance of temporal continuity (temporal extension) of an object
like a pot, although he knows that it is an illusion, so also he
should admit the appearance of the spatial continuity (spatial
1
Säntaraksita's Tattvasariigraha, ed. Swami Dwarikadas Sastri, (Varanasi).
extension) of a pot in spite of its being an illusion. But the Yogä-

cära Buddhist accepts the lirst and rejects the second.
The following account of the dispute between the Yogäcära
and the Sautrântika can be gleaned from Sântaraksita :
Yogäcära : If the so-called atom-stimulants generate percep-
tual consciousness out of their own function, why do they not
appear in consciousness (in other words, why do we not say,
"I see atoms or atom-clusters," through parämarsa) ? And since
this is not so, how can we say that they are perceived ? The
momentariness (lack of temporal continuity) of everything is
established on separate evidence {prarnäna),i\i'c\i iss the evidence
of reason (compare anumäna), and therefore, we call the appea-
rance of temporal eontinuity,that is, persistence of things through
time, to be illusory. But what is the independent evidence for
establishing the reality of such atomic data, the extramental
atom visibles, such as the yellow stimulant or the white stimu-
lant ? If they cannot be established (independently), the
appearance of the atom-clusters as gross objects out there
cannot also be established.
Sautrântika : Here is an argument (evidence of reason) to
prove that the atom-stimulants are real. A gross object is always
made up of a cluster of smaller or more subtle objects : witness
a mountain range which is formed by a number of small moun-
tains (hills) put together. Hence this gross pot that is visible
must be made up of the cluster of the subtler, atomic, con-
stituents. They are what we call the atom-visibles, that is, the
visible atomic data.
Yogäcära : Your argument is faulty. The so-called grossness
(sthülatva) of the object is exactly in dispute here. Does this
grossness really belong to the object out there ? If so, then you
have assumed already what you intended to prove originally,
namely, the externality of objects, or the existence of objects
"out there." For grossness to belong to the outside thing, you
must first establish that there are outside things for such gross-
ness to belong to. If, however, you say that grossness is that
which appears as such invariably i n the experience of all persons
including the fools, the illiterate, and the educated, alike, then
such appearance of grossness {sthülatva-praiibhäsa) is present
even in a dream-object (for example, a dream-elephant) or in
errors-like "this is a piece of silver."
Thus, just as the appearance of grossness in a dream-object

(a dream-elephant) cannot lead us to infer the independent
reality of the constituent atomistic data (the atom »visibility) of
such objects, similarly the appearance of grossness in an ordinary
cognitive state cannot be used as a reason for establishing the
constituent atomistic data as independently real.
The above is a good illustration of the philosophic dispute bet-
ween the Sautrântika and the Yogäcära. But perhaps we are back
to the old problem—the problem of finding a distinguishing mark
that may serve to distinguish the class of veridical perceptions
from that of nonveridical ones, such as dreams and hallucina-
tions. As far as the pratibhäsa (appearance) is concerned, no
distinction is possible, and to the extent an älambana (objective
ground) is to be identified with the pratibhäsa it is impossible to
find a distinction. Thus, the Yogäcära concludes that the älam-
bana is as much internal to a piece of consciousness as is the parti-
cular pratibhäsa ofthat consciousness. If there is any datum like
the atom-visible, to act as the objective ground, it is internal to
our perceptual consciousness, just as any pratibhäsa in that
consciousness is also that in which it is said to be objectively
grounded.
The Sautrântika develops the concept of arthakriyä-sawväda
(accordance with the function of objects) in order to distinguish
the veridical perception from the nonveridical ones. What is this
"accordance with the function of objects" ? Well, preceiving a
gem on the floor, I may rush to pick it up, but if it is a false
perception (illusion), I will never be able to pick it up. In one
case, there is accord with the "function of the object" (arthakriyä),
in another there is discord, and thus, a nonveridical perception
is distinguished from the veridical one. But the Yogäcära is wrc-
impressed by such a "pragmatic" theory of truth. For, according
to him, it begs the question. The argument assumes the externa-
lity of objects without really proving it. It is like the Johnsonian
way of proving that a stone exists by kicking it. Or, even it
resembles the Moorean way of waving a hand before the oppo-
nent to say that there exists a human hand, for otherwise what
else can be waved ?
The Yogäcära, however, assimilates the concept of arthakriyä
into his own theory. For, if the concept means to be congruent
or coherent (or the potentiality to be so) with the expected
behaviour pattern that invariably follows the cognitive state,

then it is possible for a veridical perception to meet this require-
ment even if we do not assume that the object is external to
consciousness. In fact, this is also the Yogâcâra answer to the
question of a distinguishing criterion between a veridical percep-
tion and a non veridical one—the former has arthakriyci, while
the latter does not. The difference between a true perception of
a gem and a (perceptual) illusion of it is much like the difference
between a real gem and a fake one. In the latter case, you can
trade only the real gem for money, not the fake ono. The difference
in what follows is accountable by reference to the causal history
of the origination of the real gem and the fake one. Similarly,
the causal ancestry of a veridical perception, as well as that of a
nonveridical one, accounts for the differences in what follows in
either case. There is congruence with the expected behaviour
pattern (compare arthakriyäsamväda) in one case and the lack of
it in the other.
If the real gem and the fake gem are so much alike that they
agree in all conceivable patterns of behaviour (for example, both
can earn you a decent sum of money, both are equally beautiful),
then the Yogâcâra will argue that there is little point in calling
one real and the other fake unless we are already prejudiced with
the idea that one of them is certainly real and the other is not.
In other words, we disqualify ourselves to judge the real from the
unreal, for, we have prejudiced the issue. Jf, however, we are
already familiar with the causal ancestry of both which deter-
mined the issue, the issue has already been resolved for us, and we
do not need any further arbiter of truth. We may just kick, with
Johnson, the stone in front to prove that it is real.
§ 3.3 : KNOWING THE EXTERNAL WORLD
It is necessary to specify the sense oï the term 'idealism* in

which it is generally used to describe the Yogäcära-Vijnänaväda
school of Buddhism. I shall apply this term here particularly to
the philosophical school of Asanga and Vasubandhu, of which
Dignäga and Dharmakïrti became later exponents. For the sake
of convenience and feasibility, I shall ignore the later develop-
ments of this school to be found in the writings of Ratnâkara-
sänti and Jnânasrïrnitra in the tenth and eleventh centuries A.D.
Philosophical idealism is usually characterized as a denial of

the commonsense view that material/external objects exist inde-
pendently of the mind, i.e., independently of their being perceived.
This general character of idealistic philosophy was presumably
present in the Vijnänaväda theory of reality. Thus, without the
risk of gross misunderstanding the term 'idealism' can be properly
applied to it. In what follows I shall be concerned with the
peculiar nature of Yogâcâra idealism, and it is hoped that, as a
result of this exposition, its difference from other forms of ideal-
ism will eventually be made clear.
Two noteworthy points can be underlined at the outset. Firsts
unlike its Western counterpart, idealism in the Buddhist context
was not used as a support for philosophical theism, according to
which God is a perfect spiritual being who created every thing else
and hence more fundamental than any material thing He has
created. Nor was it used to support a sort of pantheism, accord-
ing to which nothing exists except God and his modes and attri-
butes. Second, like its Western counterpart, Buddhist idealism is
also a rejection of what may be called materialism, and an
assertion of a spiritualistic metaphysic that is favourable to
religious belief.
The second point underpins also the soteriological significance
of the 'Consciousness only' (vijhaptimätratä) doctrine. Just as the
'Emptiness' doctrine (sünyaväda) supplies the philosophic basis
for the therapeutics of Nirvana, so also idealism or vijhaptimä-
tratä doctrine provides the metaphysical rationale for pursuing
the religious programme to realise Nirvana in the form of pure,
translucent consciousness, the 'absolute'. The vijnapti doctrine
was a reaction, on the one hand, against the Mâdhyamika inter-
pretation of prajnä (wisdom, insight) as essencelessness and em-
ptiness, and, on the other, against the realistic overtone of the
Sauträntika-Vaibhäsika metaphysics. For Maitreya/Asanga, sun-
yatä becomes dhannaîâ, and this dharmatä is sat 'existent* in so
far as it represents the essence of all dharmas, as well asasat 'non-
existent' in so far as it negates the duality (dvaya), viz., subject
and object, which is always inherent in our experience. The
verse that sums up the fundamental doctrine of the school is
ascribed to Maitreya and handed down to Vasubandhuby Asanga.
abhütaparikalpo ysti, dvayam tatra na vidyate

sünyatä vidyate tv atra9 tasyâm api sa vidyate
The imagining of the unreal' exists; (but) the (supposed) dua-
lity is not present in it. Emptiness, however, is present here
(and) even in emptiness that (i.e., imagining of the unreal') is
present.
Vasubandhu explains: The imagining of the unreal here is the
discrimination of the forms of the apprehensible (grähya) from
the form of apprehension (grähaka). "Duality" refers to the ap-
prehension and the apprehensible. "Emptiness'* means the
nature of the imagining of the'unreal': this nature is devoid of
the apprehension-apprehensible distinction.1
Two more verses from the Madhyäntavibhäga will establish
clearly Asanga's position against the Mâdhyamika :
Consciousness arises having the appearance (pratibhäsa) of ob-
jects, beings, self, and cognized states: [but] objects [etc.] do not
exist. And because of the non-existence of objects [etc.], that
[consciousness of objects, etc.] is also unreal 1.3. Therefore, the
nature of its being [merely] 'the imagination of the unreal' is
established. That [position, i.e., non-existence of everything]
cannot be maintained while admitting total non-existence
[of even the imagination of the unreal]; [but] salvation is inten-
ded to be realized from the disappearance of this [imagination
of the unreal, which, therefore, should be admitted to exist]2
1.4.
The abhütaparikalpa consists in the wrong ideation or assump-
tion of existence and essence of objects. It is represented by a
continuous series of mental states which have no beginning, but
will end with Nirvana. In fact, it corresponds to the process of
sarrisära (worldly existence or life experience). For Nâgârjuna, one
x
Madhyäntavibhäga-bhäsya (ed. by G.M. Nagao), pp. 17-18
"tatrâbhûtaparikalpo gräliya-grähaka-vikalpahl
dvayam grähyam grähakamjsünyatä tasyâbhùta-
parikalpasya gràhya-gràhakabhàvena virahitatà"
2
Ibid., pp. 18-19.
"artha-sattvätma-vijnapti-pratibhäsamprajäyate
vijnânaqi nästi cäsyärthas tad-abhävät tad apy asat" 1.3
"abhüta-parikalpatvain siddham asya bhavaty atah
na tathà sarvvathâ *bhävät tat-k$ayän tnuktir isyaie" 1.4. See Translation 5.
may say, samsara and nirvana cannot be sharply distinguished

since both are empty or void. For Maitreya/Asanga, both exist,
abhütaparikalpa and sünyatä, wrong ideation and emptiness, sam-
sara and nirvana, yena sünyam and yat sünyam} Wrong ideation
co-exists with the ultimate emptiness or citta, although the former
covers or defiles the other. By removing the former, the pure citta
is realized.
Apart from the cittamätratäy two other important characteris-
tics of the Maitreya/Asanga school are : (1) the älaya dqctrine
('storehouse' or 'home' consciousness), and (2) discrimination of
the three aspects (natures) of reality. The influence of the Saut-
rântikas, specially their concept of 'seed' and 'maturation' (bija
and vipäka), is visible in the älaya doctrine. 2 The classic exposi-
tion of the 'triple-nature' of reality is found in the Sandhinirmo«
canasütra? As my main concern here is the philosophic position
which refutes the external/material world as unreal, I will skip
here the soteriology of the 'triple-nature', or the abhüta-pari-
kalpa. It is conceded that the main purpose of the Maitreya/
Asanga school is to describe the process and progress of medita-
tion towards the sarväkärajnalä, the complete wisdom of the
Buddha, which represents the final goal of the Way (märga). The
'ideation only' or 'consciousness only' theory has been inserted
into a framework of the mystic ascension of the saint, and the
philosophic speculation here is of only secondary importance.
However, I am concerned here with the philosophic implication
of this very attractive theory—the same philosophic implication
that concerned a majority of later Yogâcârins and their critics.
One historical point before I proceed any further. I believe that
some form of philosophic idealism was concomitant with the rise
of Mahâyâna Buddhism. Thus, in my opinion, the expositions
and critiques of idealism found in the Nyäyasütra (or eên
in the Nyäya-bhüsya and the Säbara-bhäsya) need' not neces-
sarily be placed after the rise of Asanga and Vasubandhu
(i.e., 300-400 A.D.). For, the specific critique of, and distinct
reference to, Vasubandhu is to be found in Uddyotakara and
Kumârila, and not earlier. Although it was concomitant with the
rise of Mahay ana Buddhism, idealism became crystalized as a
1
G. Tucci, pp. 34-35.
2
See § 4.7 following.
3
See Translation 1.
religio-philosophic system in the hands of Asanga and Vasuban-

dhu while Dignäga and Dharmakîrti developed it further as a
philosophy with their special doctrine of sva-safjivitti 'self-
consciousness'. But more on this later.
I shall now concentrate upon Vasubandhu's philosophic argu-
ment in favour of vijnaptimätrata, 'consciousness only'. Vasuban-
dhu refuted materialism and defended idealism on the basis of
two major arguments. His first argument was as follows : Since
it is possible for unreal/non-existent things to 'appear' (cf. ava-
bhäsä) in consciousness, we may conclude that everything that
seems to exist is just a'making of consciousness' (vijnapti). The
fact that consciousness reveals the non-existents as external ob-
jects is like the fact of our perceiving non-existent strands of hair
or the non-existent second moon under the influence of the eye-
disease called timira (ophthalmia 7)1 Under delusions and
dreams, we perceive objects that are not there. Our knowledge
of the external/material world is actually of a piece with the
universal delusion {avidyä) because an external world does not
exist apart from, or outsjde, our consciousness of it.
Vasubandhu formulated three objections to this thesis and
answered them accordingly. If external objects did not exist and
hence were not related causally to our consciousness of them,
then (a) what determines the fact that we have one particular
consciousness at a given place and time ; (b) how is it that onp
state of consciousness is not limited to just one person but ex-
perienced by many alike (e.g., why a number of people can see
that there is a chair in this room), and (c) how can a non-existent
object function as it is expected to function (e.g., how can a non-
existent apple satisfy hunger, or an absent woman evoke amor-
ous feelings) ? The first and the third objections are answered by
referring to the example of dream experience. One dreams of a
particular thing at a given place and time and of another thing at
another place and time without there being any (ostensibly)
external factor to determine it. A dream object, viz., a dream-
woman can satisfy the experience of sexual hunger in dream too,
i.e., a dream-object can be libidinally cathected.
l
Vijnaptimätratä-siddhi> Virhsatikä, p. 3.
"vijnapti'mâtram evaitad asadänhävabhäsanätl
yathä tdimirika&yäsat'keaacandrädidctrsanam*'
To answer the second question, Vasubandhu referred to the

example of an illusory experience of hell. In books like the Mahä-
vastu (from which Vasubandhu obviouly derived his example) the
Buddhist conception of hell is described as being the fact of some
extremely painful experiences (such as that of being chopped into
pieces, or swimming in a lake of filth) which is shared at the same
time by a number of pudgalas (persons) having a common per-
sonal history. In reality, however, there is no hell outside the
mind of these individuals.1 Even if we are hesitant about the
credibility of this type of example, i.e., the fact of a common
psychosis described in the Buddhist mythology, Vasubandhu's
argument does not lose its force. In a world of scientific technology,
it is not impossible to create conditions for some common delu-
sion which can deceive a number of persons. It is possible to think
of Descartes's all powerful evil demon, or a modern mad scientist,
who might stimulate several men's brains with various electrodes
and employ the process so ingeniously that all of them will ex-
perience enjoying a sumptuous dinner while no food is actually
present. Witness the renewed interest in the theory of hypnotic
states. Thus, some form of Cartesian doubt might even lead us
to the conclusion of the unreality of the external world.
Several points can be noted in this connection by way of con-
trast. First, ordinarily we can make a distinction between two
types of illusion or rather between illusion and delusion. In one
case, one is wrong about what one is in fact perceiving. In the
other, one is mistaken about the very fact that he is perceiving at
all. It is the second kind of illusion, or delusion, that is most
pertinent for arguing in favour of the radical sort of idealism that
Vasubandhu hoped to establish. His example is that of the eye-
disease in which one visualizes non-existent strands of hair. This
is like the experience of pain in a limb which has actually been
amputated and no longer there. Such subjective experiences pos-
sess the vivacity of the genuine perceptual experience. And thus,
the idealists claim, they are phenomenologically indistinguishable
from those that occur in so-called authentic perceptions. If this is
conceded, Vasubandhu can easily drive the point home that the
so-called material/external world is only an appearance in, and
hence an integral part of, our modes of consciousness.
., pp. 3-4
Second, in the empiricist tradition of the West, the view that is

rather contemptuously (and, to my mind, somewhat unjustly)
labelled as Naive Realism is often refuted by what is generally
called The Argument from Illusion. In plain language, 'Naive
Realism' stands for an innocent prejudice (as it is alleged by
philosophers) of the ordinary man that the sensible qualities that
we perceive around us, colour, shapes, sounds, touch, smells,
etc. actually belong to the material objects outside and that we
not only see colour but also the coloured object directly. Such a
claim, it is believed, can be easily upset by pointing out the fact
that we are sometimes deceived by our senses, as in the case of
illusions and delusions. But this 'Argument from Illusion' led the
empiricists of the West to devise the disastrous theory of sense-
data—a veil or screen or representations between us and the
external world—which, for some time, enjoyed the confidence of
well-known philosophers in the West. From the premise that
what we immediately perceive are sensations or ideas, George
Berkeley was led to his celebrated thesis: esse est percipi. One
acceptable sense of the thesis is that what we call'material things',
such as trees, rocks and tables, are orderly groups or bundles or
collections of sensations. 1
Vasubandhu's idealism appears to be similar to that of Berke-
ley, but it, nevertheless, differs from Berkeley's idealism quite
radically. For one thing, in Berkeley's thought the notion of a
material substance as distinct from sensible qualities is unimagin-
able and inconceivable (and if they were conceivable they would be
problematic existents). But Vasubandhu's example of eye-disease
not only carried religious and moral overtones with it but also
emphasized the non-duality and finality of consciousness. It is the
non-dual, fluctuating consciousness which splits itself into a
duality, an experience and the object-appearance of that experi-
ence, the forms of apprehension and the apprehensible (grähaka
lr
The classical formulation of the empiricist's thesis in the West is found in
J. Locke's An Essay Concerning Human Understanding, G. Berkeley's The
Principles of Human Knowledge and D. Hume's An Inquiry Concerning
Human Understanding. Descartes'* point about an 'evil demon' is found in
Meditations. For remarks of modern philosophers on the problem, see B.
Russell's Human Knowledge and Problems of Philosophy, J.L. Austin's
Sense and Sensibiliay and H.H. Price's Perception. A good anthology of
twentieth-century sources in the philosophy of perception is to be found in
Perceiving, Seeing and Knowing (ed. by R.J. Swartz), New York 1965.
and grähya), although this appearance of duality is a disease to be

cured by prajnä, the ultimate insight. For another, the Representa-
tive and Causal theories of perception, as they are found in the
Western empiricist tradition, concede at least that the external
world, for all we know, may be out there, although it is not direct-
ly known by our senses. In Buddhist idealism, however, even this
concession would be fatal, because then the claim of absolute
reality to be ascribed to pure, transcendental consciousness would
be at stake. Vasubandhu justified his idealism by another argu-
ment. If we admit that the gross material/external object exists,
it leads to some absurd consequences. The so-called material
object (e.g., a table) is necessarily thought to be composite in
structure, and hence, cannot be substantially or independently
real. In Vimsatikä, verse 11, it is said :
The [material] object is neither a unitary whole, nor a multi-
tude of atoms, nor even a combined form of atoms, because an
atomic constituent cannot be justified as a real constituent [of
matter].
An atom is usually conceived as the smallest, impartite, con-
stituent of matter. The theory of creation of divisible bodies
out of indivisible and impartite atoms (i.e., the Vaisesika
theory of atoms) goes bankrupt when its inner contradiction is
exposed. If the different atoms are connected (i.e., conjoined)
with one another to create a new body bigger in size, then those
atoms must be big enough to have different parts in order to
make conjunction possible. But this will contradict the impartite
nature of atoms. If, on the other hand, atoms being impartite and
indivisible do not have any spatial extension, no conjunction of
them will produce bodies bigger in size than atoms. The Kâémïra
Vaibhäsikas (a school of Buddhist realists) used to hold that
atomic constituents are not physically conjoined (as the Vaisesi-
kas thought) but are merely combined {samhata) to create impres-
sions of solid, material bodies. Vasubandhu rejected this view too
on the ground that such a combination of atoms would be either
indistinguishable from the Vaisesika notion of a body as a 'whole'
(avayavin) or indefensible because one such combination could
not have physical conjunction with another such combination. It
should be noted here that according to the Buddhist realists, while
bodies or wholes are imaginative constructs (cf. prajnapti-sat),
their atomic constituents are ultimately real (cf. dravya-sat).

But by exposing the inner contradiction of any atomic theory of
matter Vasubandhu reached his desired conclusion : Nothing
exists but consciousness.1
The Sautrântika might take atoms to be externally existent and
lending 'objective support' to the corresponding cognitive event,
i.e., the state of consciousness. In other words, the atoms being
substantially or ultimately real (dravya-sat or paramärtha-sat)
can cause thecorrespondingconsciousness. But, Dignâga argued,
the atoms being formless themselves cannot contribute the form
or image to the corresponding cognitive state. Thus, the atoms
do not satisfy the second condition for being the'objective support'
of consciousness. The Vaibhâsikas, on the other hand, might take
the combination of atoms in the form of a chair, etc., to be the
objective support. For such a combination possesses a particular
form, and hence can contribute its form to the corresponding
cognitive state. But, according to Dignâga, such a combination
fails to satisfy the first condition. In other words, this combina-
tion being only a construct and hence only conventionally real
cannot cause the corresponding consciousness. For Dignâga, a
conceptual construct is as much real as the second moon appear-
ing in our diseased eye-organ (cf. timira).
Dignâga noted a third view (ascribed to Vagbhata by the com-
mentator Vinïtadeva). According to this view, the collection
(satncaya) of atoms is capable of producing the cognitive state,
and the image or form can becontributed jointly by the multitude
of atoms. Dignâga rejected this on the following ground. Under
this view, it would be difficult to distinguish between the cognition
of an earthen tea-pot and that of an earthen pitcher, both varying
in shapes and sizes from each other, but constituted of atoms of
the same kind (viz., earth atoms). If the distinction is maintained
on the basis of their difference in size or shape, it would be un-
acceptable because size or shape does not belong to-the atoms,
the real entities, but to the constructs, the pitcher and the tea-pot.
Thus Dignâga reached his conclusion: the so-called älambana or
objective support does not lie outside consciousness. But in the
Pramänasamuccaya, Dignâga developed his doctrine of svasam-
vitti 'self-consciousness' or .'self-awareness', which must have
1
Vijnaptimâtratàsiddhit pp.6-8.
inspired Dharmakirti's elaborate defense of idealism in his Pro-

rnänavärttika. Dharmakïrti took issues with the Sautrântikas.
The Sautrântikas believed in some kind of representation'
theory of consciousness. For according to them, representa-
tions or to use the Sanskrit metaphor, the 'forms' or 'images' (cf.
äkära) of consciousness, are contributed by the external object,
and external objects do exist although what we perceive directly
are only their 'forms' in consciousness.1
Briefly, the doctrine of 'self-consciousness' states that in each
act or moment of consciousness there are not only two 'forms'
or appearances, the forms of comprehension (grähaka) and the
comprehensible (grâhya), but also a self-comprehension or self-
cognition where both these forms are registered. For Dignäga,
self-cognition is a kind of mental perception which constitutes an
integral part of each episode of awareness. Each episode of
awareness in the consciousness series is ultimately real and self-
revealing in character while it reveals both the object-form and
its comprehension.2
One of the classic arguments of Dharmakïrti is based upon
this principle of self-cognition. Although a formulation of
this argument is found in the Pramänavänüka? the verse, which
sums up concisely this argument, and which has accordingly
been quoted very often by later philosophers, is located in his
PramänaviniscaycL By using the principle of identification of the
indiscernibles, Dharmakïrti argued :
The cognition of blue and the blue (as an object) must be non-
different (from each other), for they are apprehended together
always (i.e., when one is apprehended the other is also
*See Älambanapariksa, verses 1-8.
2
The concept of äkära 'form' is very intriguing. In the Prajnäpäramitä
literature, the wisdom of the Buddha is described as sarväkärajnänaiä. The
so-called objective support or älambana of a consciousness-moment is
nothing but the idea that develops, at a particular moment, in our mind
from mind itself, there being no external, independent object. The form of
this älambana is called äkära. In the spiritual ascent of a person through
meditation, a new äkära replaces the old äkära. To the srävaka, for
example, the äkära of a rüpa-dharma is anitya * impermanent'. In other
words, the 'form' under which nlpa appears is anityatâ 'impermanence*. To
the bodhisattva, however, the älambana will be anitya and its äkära will be
anabhinivesa 'unattachment'. See Tucci, pp. 23-24.
3
Pramänasanuwcaya, Hattori's note pp. 106-13.
238 Logic» Language and Reality
apprehended and vice versa). The (apparent) difference (bet-

ween them) is due to (our) delusions, just as one sees two
moons (under the influence of an eye-disease)1.
In each event of awareness, according to the doctrine of self-
cognition, there are three aspects, the object-aspect, its cognition
and self-awareness (cf. grâhya, grâhaka and svasaryivitü). AU the
three are integral parts of one momentary state of consciousness.
Since there is the invariable fact that the object-form is never
apprehended without its cognition-form being apprehended, it
is argued that one should not be distinguished from the other.
What, then, accounts for the particularity and variety of our
different cognitive states ? The answer is amazingly simple. It is
not the external world, but the internal väsanä, individual backlog
of our personal history, our mental dispositions, the residual
forces of our past actions, coupled with the ingrained, beginning-
less, avidyä 'wrong notions'. (See § 4.6)
The Yogâcâra idealism can very well lead to a solipsistic posi-
tion, when it is carried to its logical extreme. In this respect too,
Western empiricist tradition offers a parallel, where the question
of avoiding solipsism became a recognized problem. It was
Dharmakirti who tried to refute the charge of solipsism in a sepa-
rate treatise, Santänäntarasiddhi.2 The main point of his argu-
ment can be summed up as follows. The method by which the
idealist would prove the existence of other minds (i.e., vijnäna-
santäna or consciousness-series other than one's own) is not
much different from the method by which the realist would hope
to establish the existence of other minds. If, for the realist, the
physical acts and speech belonging to other bodies are regarded
as evidence of the existence of other minds, the idealist can very
well adduce the ideas or representations of the same visible
physical acts and speech as evidence of the existence of other
santäna or consciousness-series. Although this argument is not
very convincing, it, nevertheless, exploits the weakness of the
argument of the opponent, for, the realist is also in the same
uncomfortable position when he tries to prove that there are
other minds.
x
Cf. Sahopalambhaniyamâdabhedo nïlaîaddhfyoh, as quoted in texts iike
Nyäyamanjari and Nyayabhüsana. See also Translation 2.
2
See H. Kitagawa, in Appendix A, I, pp. 407-39. Kitagawa has given a
translation of Santànânîarasiddhi*
We may now look into the other side of the story. Since Yogâ-
cära idealism was a challenging philosophical position, it evoked
criticisms from all sorts of realists, the Nyäya-Vaisesika, the
Mïmârnsakas, the Jainas as well as the Buddhist realists. In what
follows, I shall limit myself to the criticisms of some Nyäya-
Vaisesika philosophers, as well as of Kumârila, the Mîmàrnsaka.
In most Nyäya-Vaisesika texts, one finds an elaborate defence of
their theory about the constitution of matter out of atomic parts.
The notion of atoms can be made compatible with creation of
gross bodies. For example, the Nyäya posited the dyads as inter-
vening between the impartite atoms and the gross particles, the
motes. The dyads are atomic in dimension and size, but are
divisible and have parts. Thus, conjunction of several dyads can
give rise to the gross motes or bodies without running into the
absurdities that Vasubandhu mentioned. Elsewhere I have discus-
sed the Nyäya answer to the Buddhist critique of atoms, and
hence Ï will not elaborate the point further.1
For a critique of idealism on epistemological ground let us first
turn to Uddyotakara. The oldest, perhaps pre-Vasubandhu, form
of idealism cited dream-experience etc., as examples where non-exis-
tent objects appear in consciousness. Uddyotakara countered : on
what grounds are the objects of dream-experience etc., thought to
be non-existent ? In fact, there are two senses of 'non-existence'
which are distinct, and should not therefore be confused. One is
what we can call temporal non-existence, and the other is non-
temporal non-existence. When a thing does not exist before its
coming into being and after its going out of existence, we can say
that it has this temporal non-existence, i.e., it is temporally non-
existent. But this temporal non-existence is hardly equivalent to
what we understand by the term 'unreality'. The non-temporal
non-existence is, on the other hand, equivalent to unreality.
An unreal object, e.g., a sky-flower, is non-existent only in
the second sense. To prove his contention, the idealist must
show that dream-objects or objects of memory are non-existent
not in the first sense, but only in the second sense. Usually, we
declare dream-objects to be non-existent, because they are not
1
See my Epistemologyt Logic and Grammar in Indian Philosophical Analysis,
pp. 52-4. My arguments there were based particularly on Uddyotakara;
pp. 213-16.
apprehended by our waking experience. Uddyotakara analysed

this proposition :
(1) Ifx is not apprehended by our waking experience then x
does not exist.
Uddyotakara's first objection here was that the qualification
•waking' in the antecedent was redundant. The modified version
(needed to prove the consequent (x does not exist* by asserting
the antecedent) would be :
(2) If x is not apprehended by our experience than x does not
exist.
Then Uddyotakara proceeded to show that this conditional was,
in fact, a contrapositive version (cf. viparyayä) of the following
conditional :
(3) ïf x exists then x is apprehended by our experience.
Now it is a truism in logic, Uddyotakara argued, that the con-
trapositive will be true provided we previously proved the original
sentence true. In Uddyotakara's words :
\ï apprehension proves existence then it is warranted that
non-apprehension proves non-existence.1
In other words, if non-apprehension has any power to prove non-
existence, it is dependent upon the 'positive' relation, viz., upon
the proposition: apprehension proves existence.
The idealist contended: the so-called object is not different
from its consciousness, because it is merely an apprehensible
(grähya) like the feeling of pleasure or pain. The point is that
the feeling of pleasure or pain although apprehensible through
consciousness is not differentiate from it, and we can apply this
argument to all objects of consciousness. Uddyotakara argued
that this point would lose its force against the Nyâya view of
pleasure and pain. For Nyâya, consciousness is only a particular
cognitive state, and pleasure or pain is what is apprehended, i.e.,
cognized through that cognitive state. A particular state of feeling
is what is an internal object as opposed to, for example, a table
which is called an external object. But an 'object', whether
x
Nyayavâriiika, pp. 521-4.
See Translation 4.
internai, or external, must be distinguished from its cognition

inasmuch as the latter is an act (i.e., meaning of a verb) while the
former is the accusative or patient toward which the action is
directed.
Vasubandhu explained the particularity of, and distinction
between, different cognitive states on the analogy of dream-states.
Uddyotakara argued that even in dream experience, particu-
larity and distinction are based upon the particular nature of
comprehensible objects. If it is pointed out that the apprehended
object is temporarily non-existent at the time of dream experience,
Uddyotakara countered that this is simply the nature of any
erroneous cognitive state, and the dream is only one type of many
erroneous cognitive states that we usually have. The idealist may
claim that all our cognitive states are, in a sense, erroneous,
inasmuch as they externalize an object-form which is not exter-
nal. Uddyotakara argued : this will be impossible to prove. A
cognitive state is considered erroneous only when it is contrasted
with a correct cognitive state, i.e., with apiece of knowledge. If
there were no example of knowledge, how could there be any
example of error ?
To underpin the distinction between knowledge and error,
the earlier Nyâya analysed a piece of cognition, an event of
awareness, as follows. In each piece of cognition, there are two
different types of objects formulating that cognitive state. One
is called pradhäna, the 'principal' object, and the other is called
tattva, the 'contextual' object. In the wrong apprehension of a
rope as a snake, the principal object is the snake (or, the snake-
universal), whereas the contextual object is the rope which is
present in the context. There are no cases of error without there
being one principal object corresponding to it. Dream-experience
is an erroneous cognitive state where the principal object is the
common feature shared by both dream-objects and known
external objects (or objects of our waking experience). Moreover,
one can describe one's dream to the other, and thus a dream-
object is also communicable through words. The expressibility
through words, by virtue of which a dream-object becomes a
public property instead of being exclusively private like a feeling
of pain, is, in fact, grounded upon the common feature of parti-
cular things, i.e., universals. Elsewhere I have discussed the old
Indian theory which states that 'the ground for application'
{pravrtti-nimitta) of a word for expressing a thing is the universal

or generic feature of that thing. Now, to the extent that we admit
universals to be mind-independent or extramentally real, we
have to admit that even our dream-experiences are grounded
upon an external world. In short, according to Uddyotakara, the
objectivity of our dream-experience is proved by its expressibility
through words.1
Vasubandhu justified the non-privacy of our ordinary experi-
ences by referring to the hell-example, the case of a common
psychosis where the corresponding object does not exist. Accord-
ing to Uddyotakara, this explanation will run into contradiction.
A particular hell-experience, viz., experience of a pool of filth
(or blood), is also what Nyâya would call an erroneous experience,
and thus it too depends upon what we have called the principal
object, viz., the blood-universal and/or the pool-universal. And
such 'principal' objects, as already noted, are grounded in wak-
ing experiences and thereby grounded in outside reality.
Further, according to Uddyotakara, an error or a delusive per-
ception is usually in the form of a certitude (niscaya), a decision as
opposed to a doubt (samsaya). A false certitude of the form 'This
is a man* (where in semi-darkness there lies a tree-trunk in the
visual field) is removed (i.e., destroyed) by the true certitude or
knowledge of the form 'This is a tree-trunk'. But the object, i.e.,
the principal object, of the false certitude is not destroyed there-
by because it does not happen that when the true certitude arises
a man ceases tobe man. Besides, according to Uddyotakara,
there are definite causal antecedents to a false certitude. They
are, broadly speaking, (a) perception of the common feature,
(b) non-perception of the specific feature or distinguishing
mark, and (c) superimposition of a definite specific feature which
is actually absent from the context or situation. In this way,
knowledge and error can be distinguished.2
lbid., p. 522.
"na cittena parah pratipädyate ity artho 'syo na
sidhyati na hîtarasvapnam anâkhyâtam itaro
vijânâtîti
For the Indian theory of meaning and universals see my Epistemology...
pp. 106-9.
2
Ibid., pp. 524-5.
cf. "niscayahuddher nimittam asti kirn punas tat-1
sàmânyadarsanam viéesâdarsanam avidyamâna-
viéefâdhyâropa ///."
I shall next discuss the arguments of Kumärila, Jayaata and

Bhäsarvajna, as' they grappled with the problems raised by
Dignäga and Dharmakîrti in defense of idealism. Kumärila
refuted Buddhist idealism rather elaborately in two sections,
Nirälambana-väda and Sünya-väda, of his Slokauärttika. Al-
though some of his refutations amounted to sophistry, he never-
theless offered important criticisms of Dignâga's theory of self-
cogmt\on(sva-samvitti). In Pramänasamuccaya, Dignäga argued in
favour of the 'self-cognition' theory by first asserting that a state
of consciousness must have two appearances or forms: object-
appearance and cognition-appearance. If this two-fold appear-
ance of a cognitive state is not admitted then it would be difficult
to explain, according to Dignäga, the recollection of a past
cognitive state. Recollection is caused by the memory-impression
of a previous experience. But we see that we recollect not only
the object of previous experience but also our previous ex-
perience itself. Unless this two-fold appearance was present in
our previous experience, we could not have recollection of the
two in this manner. Dignäga argued that this proves further that
each cognition is self-cognized. For, unless a cognition is self-
cognized it can never be recollected. The Naiyäyikas hold that a
cognition is not self-cognized, but cognized by another piece of
cognition. According to Dignäga, this theory leads to an infinite
regress because to make recollection possible the second piece
of cognition must be cognized by a third piece of cognition and
so on ad infinitum. Thus, Dignäga established that there is a self-
cognition aspect in each state of consciousness where both the
apprehensible and apprehension are revealed.
Kumärila disagreed on this point. The fact of recollection
cannot establish the two-fold appearance of a cognitive state
because, for Kumärila, what is recollected is always the object
and never the previous cognition itself. According to Kumärila's
theory of knowledge, a cognitive state is never perceived (i.e.,
apperceived) directly, but its occurrence is only inferred from
what he calls the cognized-ness (jnâtatâ) of the object cognized,
(see § 3.2.) Thus, when the object of previous experience is
recollected, the previous experience may then be inferred indirectly
(through arthäpatti 'indirect implication'). This, however, should
not be confused as perception or even direct recollection of the
cognitive state.
Besides, if apprehension and the apprehensible were in fact

identical^ it would be difficult to explain some recollection, for
example, of this form: "I do not remember what I apprehended
at that time." Here is a recollection of the occurrence of an
apprehension but the form of the apprehensible, the object, is
for some reason left out of the recollection. If apprehension
and the apprehensible were identical, this would not have
happened.1
Jayanta repeated this point of Kumâriîa while he tried to refute
Dharmakïrti's classic argument mentioned above, an argument
based on 'togetherness.' Both, the Naiyâyikas like Jayanta and
the Jainas like Akalanka, mounted an attack on Dharmakïrti
by pointing out that the expression ctogether' used in Dharma-
kïrti's reasoning implied difference rather than unity. But this is
only a verbal slip because Dharmakïrti's meaning is quite clear:
Blue and the cognition of blue are identical because they are
apprehended always by the same apprehension. As against this
position, Jayanta held: a piece of cognition does not cognize
itself. A cognition or a state of consciousness is not self-lumi-
nous and hence self-cognition is not an acceptable theory. It
is possible to cite many counter-examples where the apprehen-
sible blue appears in the cognition 'This is blue", but its appre-
hension, i.e., cognition, does not appear there. Hence it is possible
to claim that there are cases where blue is apprehended but its
cognition is not apprehended. An apprehension of the cognition
of blue will be of the form "I know that this is blue." Using
Kumârila's example "I do not remember what I actually
apprehended at that time," Jayanta argued that we have a
counter-example where the mere apprehension is recollected
without the recollection of the apprehensible object.2
'Kumârila, p. 221.
cf. "na smarämi maya ko 'pigrhito "rthas îadeti hi
smaranti grahakolpada-gràhyanïpavivarjitam
tasmâd abhinnatâyâm ca grùhye 'pi smaranam bhavet
grâhakasmrfinirbhâsât taîràpy esaiva grhyate"
tadatyantavinâbhâvân naikâkâram hi jäyate
anvaya-vyatirekäbhyäm siddhaivam bhinnatä tayoh
Sünyaväda, verses 83-85.
%
NyäyamahjarU p. HO.
kvacic ca grahyakaranupaslitfakevalagrahaka-
Dharmakîrti has argued that perception of an object is not

established for him who has not apperceived the perception.1 But
the Nyäya argues that perception of an object is, in fact, possible
even when the apprehension (of the object) is not apperceived. For
as soon as the object is apprehended we can say that perception of
the object has taken place; we do not need to wait until and
unless apperception of that perception has taken place. In other
words, the object is revealed as soon as the perception arises,
and not when the perception is perceived (i.e., apperceived).
Arising of perception and apperception of that perception are
two different facts which may happen at two quickly succeeding
moments but they should under no condition be confused as
one and the same fact. Perception can reveal objects when it
itself is not revealed.
Jayanta also pointed out that the analogy between a lamp
and a cognitive state to prove their self-revealing character is
really faulty. "Cognition reveals itself" involves only a bad
metaphor for "cognition cognizes itself" or "cognition makes
itself known". It is not also proper to say that a lamp reveals
itself. In fact, a lamp too depends upon other conditions to
reveal its own self. The eye-organ receives or apprehends other
objects depending upon the lamp, i.e., light, whereas it appre-
hends the lamp more directly. But to make its own self known,
the lamp has to depend upon the eye-organ, etc. And strictly
speaking, this would not establish the self-revealing character
of the lamp or light.2
Dignäga pointed out that there would be an infinite regress in
the Nyâya theory of apperception. But as explained above, origin
of perception (and not apperception of perception) amounts to
the revealing of the object. Thus, we do not need to have all our
perceptual (in fact, all our cognitive) states apperceived. In
order to make our recollection of previous experience possible,
vamarèanam api drfyate 'lna smarämi

maya ko *pi grhïto 'rthas tadeti hïti".
1
Cf. apratyaksopalambhasya närthadrstih prasidhyati. Quoted in Nyàyabhù-
sana-y p. 108, and in Nyâyamanjarî, pp. 104 and 108. Jayanta notes^
apratyaksopalambhasya ca pratyut art hudrsfih siddhyati, upalambhotpâda
evärthadrsfir na punar upalambhadrstih /
p. 108
we have to admit that the previous cognitive state was therefore

apperceived. Apperception of apperception is, however, ruled
out because although we recollect that we know such-and-
such, we seldom recollect that we knew that (we) such-and-such.
Dharmakïrti's argument based upon 'togetherness' of blue and
cognition of blue has been criticized by Bhäsarvajna as follows:
Does the apprehension, which, according to Dharmakirti, is said
to apprehend both the cognition and the object blue, apprehend
also their difference or not? If it does not apprehend their dif-
ference, we should never be able to refer to them differently by
two different expressions. In other words, if the difference is not
registered in the apprehension, it cannot be registered in the lin-
guistic expression of such apprehension. If, however, the difference
is registered in the apprehension, why should we mistrust it? If it
is argued that the said appearance of difference is wrong, like the
appearance of two moons before our diseased eye, Nyäya coun-
ters that this is untenable. The appearance of two moons is
wrong but their two appearances (one real moon and one unreal)
are certainly distinguishable. Similarly, the object-appearance
and the cognition-appearance should certainly be distinguishable.
The appearance of two moons is considered wrong because we
can cite a case where there is the appearance of only one (viz., in
true perception of the moon). But the object-appearance and the
cognition-appearance cannot be declared as identical unless there
is a case where the two appear as one. Thus, self-cognition, if it
registers difference between object and cognition, cannot prove
their identity.1
Jayanta has repeatedly pointed out that even the doctrine of
self-cognition is not immune to the fault of infinite regress. If the
object-form being the apprehensible is apprehended by the cog-
nition-form, and this cognition-form is apprehended by the self-
cognition, then by the same token the cognition-form has also
the form of an apprehensible and self-cognition would have
the form of its apprehension. If the self-cognition obtains a form
in this manner then it is liable to be apprehended by another
apprehension and so on ad infinitum. The point is that the charge
of infinite regress is only a formal fault and does not further the
argument in either case.2
x
Nyäyabhüsana, pp. 132-3.
%
Nyayamafijarî, p. 15.
To come back to Bhäsarvajna. It may be argued that if a

cognition is not self-cognized, it loses its power of revealing any-
thing (cf. aprakâàa). Bhäsarvajna countered: i£>the rule is that the
loss ofthe power ofrevealing anything means the loss ofthe power
of making itself known, then we have eye-organ etc., as counter-
examples. The eye-organ can reveal something without being
cognized itself. If, however, the loss of the said power means loss
of illumination (as in the case of a lamp), it is a case of bad meta-
phor. Only light has illumination and cognition is not the same
thing as light or lamp. It is only poetic to say that cognition has
illumination. If, however, the loss ofthe said power means the
loss of its nature (essence of being consciousness), then it is
untenable. For certainly a thing does not lose its nature (or
essence) if it goes unrecognized. Flowers will have fragrance
even if there were nobody to smell them. It may be asked: with-
out self-cognition what else could constitute the nature (essence)
of a piece of cognition? The Nyâya answer is that the essence
or nature of a piece of cognition is simply the common pro-
perty (a generic property) of all pieces o^ cognition, viz., the
cognition-universal or the cognitionhood. This property can as
much be identified with the feature of self-cognition as the
nature of lire can be identified with the property of burning its
own self, or as the nature of a knife can be identified with that of
cutting its own self.1
Before we conclude our critique, let us ponder a little over the
doctrine of väsanä and that of älaya. The Buddhist mechanism
of vasanä can be explained as follows. Each experience leaves its
trace behind, which becomes a part of the ever-increasing volume
of such residual forces. This volume is, in its turn, said to deter-
mine the nature and content of our next experience. The Nyâya
criticizes it as follows: Since there is no soul or individual per-
sonality to carry this ever increasing volume or burden of the
traces ot past experience, the question arises: is this burden
different, from, or identical with, the consciousness series? If it is
held to be different then the theory of'consciousness only' will
beat stake. If-it is identical with consciousness, then we would
have to admit that this stroehouse of latent traces is also ever
fluctuating along with the consciousness series. In fact, we have
l
Nyäyabhü}anat pp. 137-9
to concede the oddity that millions oï latent traces are also dying
every moment, and creating new latent traces. Asanga and Vasu-
bandhu developed the älayä doctrine, 'storehouse consciousness,'
to account for the continuity of latent traces in the form of seeds
through ever-fluctuating states of consciousness. But this alaya
again, has to be conceived as an ever fluctuating series receiving
new seeds and dispensing with some old seeds through fruition
at every moment. For further discussion see 4.7.
Realists like Kumârila and Jayanta found this concept of âlaya
very difficult to accept. For, under this theory, not only one has
to imagine the momentary, fleeting states of consciousness (no
matter whether it is the base consciousness or alaya, or the
surface consciousness or vijhapti) as the container or substratum
of innumerable latencies or traces, but also one has to concede
the actual and instantaneous arising and destruction of such
innumerable latencies. Only when a latency or seed reaches its
maturity (vipâka), it produces a particular form of experience.
But the question is: why should we talk of the dying of innumer-
able latencies when we will have to say that these latencies as
latencies come to life again in the next moment.1
Some Buddhists draw the analogy of 'perfuming' one thing
with the other. In other words, latencies are transferred from
one consciousness state to the other just as the perfume of one
thing is transferred to the other by association. Jayanta argued
that this analogy was misconceived. Perfume is a (physical) part
of the thing that possesses it, and transference of the perfume is,
in actuality, the transference of this subtle 'physical' part of one
thing to the other. But the Buddhist postulates that one state of
consciousness is totally annihilated before the other state arises
in its place (cf. niranvaya-vinäsä). But the notion of 'perfuming'
will involve not total destruction but only partial destruction with
traces left behind. Besides, if the state of consciousness is only
momentary, how can it transfer its perfume to the other state.
For, ordinarily, a thing has to stay longer than just a moment in
order to transfer its perfume to another thing.2
Wyäyamanjari, p. 114. na câlayavijnânam
kincld astiy saty api tasminn a§esaväsanäsahasra-
samâsraye îatk§anikatvât sakrdeva
tathâvidhavasanànukfdajnânavinasah syâf.
%
Nyâyamanjari, p. 113, continued next page.
I have given the philosophic formulation of the Yogäcära idealism

and also shown formulations of some classical objections to such a
philosophic position. Whatever might have been the motive or
driving force behind this refutation of external/material world as
unreal and unfinal, it was received with all philosophic serious-
ness in India (as the above sketch tends to show). The opponents
who disagreed with idealism tried to provide philosophic argu-
ments, rather than religious or theological skirmishes for the
refutation of idealism. Very seldom was an appeal made, in this
philosophic dispute, to religious faith, either Buddhist or non-
Buddhist. Thus, I see in this controversy between idealism and
realism, in their formulation as well as refutation, one of the
most interesting developments in the history of Indian philos-
ophy. After considering the period in history as well as the
sophistication of the arguments, I am inclined to believe that this
constitutes a very important part of our heritage, that is, oui-
philosophie heritage in the global sense. I wish to underline this
global sense of our philosophic heritage by repeating a belief of
mine which was expressed in the preface to my first book: "India
should not, indeed cannot, be left out of any general study of
the history of logic and philosophy."1
To briefly review the issue at hand: There is a sense in which
the arguments of the idealist seem to be very convincing. They
do make a very strong case for the position that we are indeed
shut up in our ideas, or that the essence of a thing consists in its
being experienced. Beyond this, however, i would not be inclined
to concede anything further. It would, for example, be wrong to
assume that the idealist has deductively proved his thesis. All that
the idealist has succeeded in doing is to warrant a claim of this
type: As far as the available evidence goes, the assumption of
the reality of the material world may be false. But to cast a
reasonable doubt over our uncritical assumption of the existence
of the material world is not the same thing as proving the falsity
or unreality of the external world. For implied in the notion of
unreality of the material world is a theory of a supersensible
world which is believed to be real and even more valuable. But
this becomes more a matter of religious faith than a philosophic
nir-anvaya-vinâsâc ca na tadarnso 'nuvartate

yatah kathancid väsyeta pürveijta jnânam itttaram
l
The Navyanyâya doctrine of Negation, Preface, p. X.
point. As long as we are on philosophic grounds, we can accept

the fact that the idealist can within reasonable limits challenge
our beliefs in the external world. But we may again point out
that to be able to challenge a belief does not amount to destroy-
ing that belief, let alone replacing that belief with another, equally
questionable, belief, i.e., belief in 'consciousness only'.
The idealist is also guilty of an illegitimate generalization.
First, a doubt whether the external world exists independently
of mind or not is simply used by the idealist to challenge a
certitude about the independent existence of the external world.
But it is also taken (illegitimately, I believe) here to establish the
opposite certitude that the external world does not exist indepen-
dently of consciousness. This is, in fact, an illicit extension of the
function of doubt. Nyäya seems to be justified in pointing this
out.
Second, the Idealist relies on avery questionable generalization
step. From an examination of some of our experiences where
the so-called object of experience is actually an object in experi-
ence (such as dreams, delusions and hallucinations), the idealist
takes a big step forward to conclude that all objects of our
experience are, in fact, objects in experience, and hence, do not
exist independently of it. This is a generalization based on in-
sufficient grounds, and hence, logically not beyond censure. Even
as a persuasive argument it is not adequate. For, the notion of
the external objects is so ingrained in us that it will need much
stronger persuasion in order to dislodge it. But if the idealist's
arguments are intended simply to awaken a religious awareness
in us to follow the path of meditation and nirvana, I would not
subject it to logical scrutiny. I can admit that the arguments
could be more persuasive to a religious mind.
A Jain philosopher, Akalanka, had the insight to point out
that the reasoning of the idealist is as fallacious as the reasoning
used by the Naiyâyika to prove the existence oi% God as the
intelligent creator of the world. Having seen that certain things
that have structures and composite forms, such as a chair, a pot,
are produced by an intelligent or sentient agent, the Nyàya
generalizes that all things that possess a structure must have an
intelligent creator, i.e., God. The idealist, according to Aka-
lanka, takes a similar illicit step. 1
p. 43, continued next page.
Problems of Knowledge and Perception . 251
Last but not the least, if the idealist challenges the realist to
prove conclusively and beyond doubt the reality of the material
world, it might only be a good strategy in public debate. For even
if the realist fails to prove his thesis conclusively (as he may very
well fail if he falls into the trapset by the idealist), this will not, in
my opinion, constitute a good vindication of idealism. A realistic
interpreation of error, hallucination and dream, is indeed possi-
ble, and the Nyäya theory in this regard is not, in my opinion,
very far from the truth. But construction or consistent description
of a realistic metaphysic is another matter. Partial failure in
this respect will not constitute a justification of idealism. A
realist may even refuse quite consistently to describe an
ontological system. For, it is the idealist who has challenged
the commonly shared beliefs, and so it behoves him to establish
his thesis conclusively.
By way of documentation, I wish to present here excerpts from
certain relevant texts in translation:
Translation of Relevant Texts

1. Sandhininnocana-sütra (pp. 60-61)
[The translation is by the author from Lamotte's Sanskrit
reconstruction.]
"There are, Gunäkara, three dhar ma-characters. What are
these three?
They are, to be sure, (1) the imagined character, (2) the
dependent character and (3) the perfected character.
"Now there, Gunäkara, what is the imagined character
of dharmas? Tt is, to be sure, the determination of descriptive
names of dharmas (elements of reality) through their essen-
tial or peculiar attributes so that there may be linguistic
usages and conceptual communications.
"What is, Gunäkara, the dependent character of dharmas?
It is surely the dependent origination of dharmas: That
being there, this arises';'from the origin of this, that ori-
ginates'; 'satnskära' (mental forces) arises depending upon
sannivesädibhir drsfair gopurättälakädisu
buddhlpürvair yaihâ tattvatti negate bhüdharädisit
tathä gocaranirbhâsàir drstair eva bàhyâdisu
abâhyabhavanûjanyair any at ret y avagamyatäm.
a-vidyä (wrong notions) and so on. In this way, this great

mass of that consisting exclusively of sufferings (duhkha)
arises.
"What is, Gunâkara, the perfected character of dharmas?
It is, to be sure, 'thusness' of dharmas. Understanding of
this (character) is the cause of the strength of a bodhisattva,
the ground for contemplation or fixation of the mind upon
the ultimate root. As meditational process takes place as a
result of this understanding, there arises the unsurpassable
climax of enlightenment.
"Just as, Gui.iâkara, there is the timira defect in the eye of
a person who is suffering from timira disease, in the same
way we should understand the imagined character.
"Just as, Gunâkara, due to //7w/ra-disease there appear
before that (diseased) person strands of hair or black bees
or (black) sesame seeds, or dark objects, or yellow objects,
or red objects or white objects, in the same way we should
understand the dependent character of dharmas to be.
"Just as, Gunâkara, there are natural appearances of
correct objects before that eye of that very person when the
eye-sight is purified and cured of the //m/Va-disease, in this
very fashion we should understand the perfected character
(of dharmas) to be."
2. Dharmakïrti: Pramänaväritika
[The translation is by the author.]
Since the (cognizable) object is invariably apprehended
along with its cognition, in what manner could we establish
the object as distinct from its cognition?
Distinctness is apprehended (however) by persons having
illusory vision, just as the two moons appear to a person
with the diseased eyesight. There is no invariability in the
apprehension of two (really) distinct things, viz , the blue
thing and the yellow thing.
There is no (cognizable) object which is unapprehended,
and it is not seen that there is a cognition which is being appre-
hended without the (cognizable) object. Thus, there cannot
be any distinction between these two (cognition and object).
Therefore, it is irresistibly concluded that the (cognizable)
object which appears at the time of cognition is indistinguish-

able from its cognition.
Pratyaksa pariccheda, verses 388-39Î.
3. yâtsyàyana
Nyäyasütra 4.2.31-32
Supposition of'the means of knowledge' (pramäna) and
4
the objects of knowledge' (prameya) resembles the supposi-
tion of dream-objects.
Vätsyäyana : Just as in a dream objects do not exist but are
supposed to be there, similarly the means of knowledge and
the objects of knowledge do not exist but we (only) suppose
that there are objects of knowledge and means of knowledge.
NS 4.2.33 [Answer] This argument is not established because
it is not based on a reason.
Vät. :
There is no reason to prove (conclusively) that the means
and the objects of knowledge are only supposed to be there
like the supposition of objects in a dream, and that they are
not known like the objects of waking experience. Therefore,
the (above) argument is not established for there is no
reason. (See NS 4.2.33)
Moreover, there is no reason to maintain that unreal
objects are cognized in a dream experience.
[Objection] (The reason in the latter case is:) "Dream objects
are not cognized when one wakes up."
[Answer] Since the objects of waking experience are cogni-
zed their existence is notrefuted. (To explain): If the objects
in a dream are proven to be non-existent because they are not
apprehended when one wakes up, then (at least) those objects
that are apprehended by one at one's waking state are proven
to be existent because of their being apprehended (at the wak-
ing state). For, the reason becomes adequate in the 'contra-
positive' case. If existence is established through apprehen-
sion then non-existence can be established through non-
apprehension. If not the first, then non-apprehension cannot
be an adequate reason for proving non-existence. For
example, when the lamp is absent, visible colour cannot

be seen. Thus, here the negated version is justified by the
positive case: (e.g., when there is a lamp present, we see a
visible colour, and thus, non-perception of visible colour
can be said to be due lo the absence of a lamp).
One must also assign a reason for regarding dream-objects
as (products of) imagination. (To explain:) He who says
that the means and objects of knowledge are like the sup-
position of dream-objects should speak about a reason on
the basis of which the dream-objects are (said to be mere)
imagination. Certain dreams are inspired by fear, certain
dreams are inspired by joy, some dreams are different from
both of these two types, and sometimes we do not (even)
dream (while asleep). Supposition of dream-objects have
their 'causal antecedents' (nimitta) and the imagination of
these objects is justified by the imagination of the causal
antecedents.
MS 4.2.34 Supposition of dream-objects resembles supposi-
tion of objects in an act of remembering or desire (or
determination to obtain).
Vàt. :
Such objects as have been previously known (or apprehen-
ded) become the objects of supposition in a dream, just as
they become objects of supposition in an act of remembering,
etc. Just as both an act of remembering and a desire to
obtain have as their objects things which have been previously
apprehended (in consciousness), and hence they cannot
invalidate (the existence of) such things, so also the cogni-
tion of objects in a dream has for its objects previously
apprehended things and such a cognition likewise cannot
invalidate (the existence of) such things. Thus, objects of
our dream-experience are already apprehended by our waking
experience.
He who sees a dream while asleep surely wakes up and
remembers the dream-experience—"I saw this (in a dream)."
There he becomes certain that his supposition of dream
objects is false on the basis of his experience when he wakes
up. (To explain:) When dream objects are being remember-
ed, one becomes certain by virtue of the waking state of
one's consciousness that the supposition of dream objects

is false.
If there was no distinction between the two (dream and
waking experience), the reasoning will be meaningless. (To
explain:) When one does not accepta distinction between
the dream and the waking experience, the reasoning that
the means and objects of knowledge are like supposition of
objects in a dream will be meaningless because the basis of
that reasoning will'(thereby) be rejected.
Certitude (of a person) that it is x with regard to some-
thing that is non-x has as its (objective) basis a 'principal'
object (pradhäna). (E.g. :) Certitude that this is a man with
regard to a tree-trunk which is not a man depends upon a
'principal object'. Surely, if some man or other were never
known before, one cannot have, with regard to something
which is not a man, a certitude that this is a man. Thus,
certitudes of dream objects which are expressed as "I saw an
elephant", or "I saw a mountain", should have as their
(objective) bases some'principal' objects.
NS 4.2.35 False apprehension is destroyed by true apprehen-
sion, just as the supposition of a dream object is destroyed
at the waking stage.
Vat. :
Certitude that this is a man with regard to a tree-trunk is
called a false apprehension—it is the cognition that it is x
with regard to that which is non-x. Certitude that this is a
tree-trunk with regard to a tree-trunk is called true cogni-
tion. It is the false apprehension that is rejected by the true
cognition, but not the object which is the generic character
of either being a tree-trunk or a man. Just as the state of
consciousness at the waking stage rejects the supposition of.
objects in a dream, but not the generic character of objects
is thereby rejected, so also where there is a cognition of a
magic show or castles in the cloud, i.e., the certitude that
it is x with regard to what is a non-x, the false apprehension
is destroyed by the true cognition, but the object is not
thereby rejected.
False cognition in a magic show and the like has also a
'material* condition. When aman taking a piece of subs-
tance, whose character is familiar, uses tricks and creates

a false certitude (that it is a different thing) in the mind
of others, it is called magic. Configuration of thick
mist, etc., in the manner of a city or castle generates from
a distance the cognition of a city or castle. For, when
such mist, etc., are not present, the impression of a city or a
castle (in the cloud) does not appear. The impression of
water is generated by the vibrating sun-rays connected with
the terrestrial heat, for we apprehend here the common
character (i.e., vibrating brightness). Thus, the false cogni-
tion is not without its causal antecedents because it happens
at some place at a certain time in a certain person.
It is seen that these are two distinct cognitive states, one
belongs to the magician and the other is that of the spectator
—one is that of the distant person with regard to a cloud-
castle and mirage, and the other, that of the person who
is standing close-by; with regard to a dream object, one
cognition is that of the sleeping person and the other is
that of the same person while he is awake. This distinctness
of cognitive states cannot be explained if there were nothing
that exists, or if eveything were unreal and without essence.
MS 4.2.36 A cognition is to be thus understood, for we
apprehend its causal antecedents as well as its coming into
being.
Vât. :
Just as the object cannot be rejected, so also the false cogni-
tion cannot be rejected. Why ? For we apprehend its causal
antecedent as well as its coming into being. The causal ante-
cedent of a false cognition is certainly apprehended by us.
And the false cognition itself is also apprehended (appercei-
ved by an individual) as it originates in that individual, for
it is subject to mental perception (since it is a mental
state). Therefore, a false cognition also exists.
NS 4.2.37 Since there is a distinction between tattva, the
'contextual object' and pradhäna 'principal object', the
duality of a false cognition is justified.
The 'contextual object' is the tree-trunk, the 'principal
object' is the man. With regard to the tree-trunk there arises
a false cognition that this is a man because while the
distinction between the 'principal object' and the 'contextual

object' is not eliminated their common character is (never-
theless) comprehended.
Thus, there arises false cognition of a circle of flying
ducks in a flying banner, of a (flying) pigeon in a (falling)
pebble. There are no false cognitions co-present with the
objects (they reveal) because they (false cognitions) are depen-
dent upon the comprehension of a common character
(among several objects). For him according to whom every-
thing is without essence and unreal there may arise false *
cognitions which are co-present in the same object.
4. Uddyotakara: Nyäyavärttika (pp. 521-525)
V.P. Dvivedin's edn., 1916.
It is not true that the means of knowledge and the objects
of knowledge are supposed (to be existent) like (the objects
in) a magic show, or the appearance of a castle in the cloud,
or a mirage. For this (thesis) cannot be proved.
No reason is given when it is asserted that unreal objects
are being supposed (to be existent) like dream-objects. What
reason is there to maintain that the objects in a dream-
experience are unreal ?
[Opponent:] The reason is what is called khyäti (appearance
of objects in experience). The reason for maintaining the
thesis, viz., objects that are apprehended at the waking
state of consciousness, do not have a separate existence from
consciousness, is the appearance of objects as in a dream.
[Answer:] No. Your example (i.e., the dream-example)
stands in need of a proof just as does your thesis.
What is the reason for believing that the objects that
appear in a dream-experience are not distinct from the
(dream-) consciousness ?
[Opponent:] The objects do not exist since they are not
apprehended by the waking person.
[Answer:] If you think that these objects do not exist just
because they are not apprehended by the waking person,
then you are wrong. For you have added a qualification
(unnecessarily).
It is implied that those objects that are apprehended by
the waking person exist. And the qualification, those "that

are not apprehended by the waking person" is superfluous.
If you argue that the objects that are apprehended dther
in the waking experience or in a dream are unreal but the
consciousness (of such objects) exists, (we ask :) what could
be your reason ?
The fact of being not apprehended by the waking person
cannot be cited as a reason because it is incapable of
establishing the 'contrapositive' (viparyaya; viz., apprehen-
sion of objects by the waking person proves their existence).
If apprehension proves existence only then it is warranted
that non-apprehension can prove non-existence.
It is contended that the reason should be capable of
establishing the 'contrapositive' thesis (viz., apprehension
proves existence).
[Opponent:] The objects are not distinct from the conscious-
ness (of them) because they are subject to comprehension
(by the consciousness) just as a particular feeling (of pain)
is. Just as that which is apprehensible (grähya) like a parti-
cular feeling, is not distinct from consciousness, so also the
objects are not distinct from consciousness.
[Answer:] A feeling is a feeling of either pleasure or pain.
Consciousness is a cognitive state (of mind). A cognitive
state is distinct from pleasure or pain. Therefore, your
supporting example is not correct. Pleasure and pain are
what are apprehensible, and their apprehension is a cognitive
event. Thus, the apprehensible must be distinct from its
apprehension.
If you argue that consciousness is identical with the feeling
(i.e., pleasure or pain), we still say that there is no supporting
example where the apprehensible and the apprehension are
(undoubtedly) identified with each other. For, an act and
the object to which that act is directed cannot be identical
with each other.
If you wish to establish identity (between apprehension
and the apprehensible) without caring for the evidence, the
means of knowledge, you establish (only) four 'personality'
aggregates (instead of five, by equating feeling with con-
sciousness); and this will contradict your doctrinal position
(that there are five 'personality' aggregates, viz., aggregates
of matter, feeling, perceptions, mental forces and conscious-

ness).
If you do not accept even four aggregates but admit con-
sciousness only that is (commonly) experienced, we then
counter: since there are neither internal not external objects
to account for the distinctness of the states of consciousness,
how can you explain the distinction of different states of
awareness?
[Opponent :] One state of awareness is different from the
other just as one dream is different from another.
[Answer:] Even in that case, the distinction of awareness
states should be explained as owing to the comprehension of
(different) things, that are observed and experienced. If, in
the case of dreams also, the distinction of awareness
states is asserted to be owing to the difference of compre-
hension, then we counter by pointing out the distinction
of comprehension from the comprehensible. The compre-
hensible cannot be identical with comprehension.
All the dream-experiences that reveal cities, birds, gardens
and chariots distinctly, are erroneous cognitive states. And
these erroneous cognitive states originate, sharing the generic
features of the waking cognitive states.
[Opponent:] In my theory, all cognitive states will be held
to be erroneous.
[Answer:] If you say that, we ask you to point out the'principal'
object (pradhäna) in such errors. There cannot be an erron-
eous cognition which is without any 'principal' object.
Moreover, he who does not accept that the objects are
distinct from consciousness, should be countered with the
refutation of this reason based upon the examination of its
nature.
Is the nature of the .object (in a dream) that of being
external ? If so, then 4he position becomes contradictory
(viz., what is internal in consciousness is said to be external).
Is the nature of the dream-object that of being conscious-
ness ? If so, then the dream-object cannot be established
because the states of consciousness cannot be communicated
to others. For, if the dream of one person is not expressed
in words, the other person would not be able to know it.
[Opponent:] The states of consciousness taking the'forms' of
words are conveyed to others.
[Answer:] He who says this should explain the meaning of the

word 'form', äkära. The 'form' (in this context) is the appre-
hension by which something non-x appears as x depending
upon the similarity of the 'principal'object, (e.g., "This is a
snake"—this wrong apprehension has taken the 'form' of a
snake with regard to something that is not a snake but has
similarity with the 'principal' object, i.e., a snake).
Also, in your view, words are not real, and hence the
sentence "the consciousness-state takes the 'form' of words"
is a statement without a reference.
He who does not accept that objects are distinct from
consciousness, should also be asked to explain the distinction
between a dream and a piece of waking experience. The objects
are non-existent as much in the waking experience as in the
dream; hence, where from could there be such a distinction
as reflected in the statement "This is a dream and that is a
waking experience ?"
Even the discrimination of what begets merit and what
begets demerit cannot be established (under your view). For,
just as one does not incur dement by having illicit sexual
relations in a dream, so also one should not (in your
theory) incur demerit by doing so in the waking state.
[Opponent:] The (said) discrimination is maintained by the fact
that in one case the person is overwhelmed with slumber,
while he is not so overwhelmed in the other case.
[Answer: ] This is also wrong. How does one know that this
odd behaviour of the mind is caused by the influence of
slumber ?
[Opponent :] The clarity and unclarity of the states of conscious-
ness distinguish the dream stage from the waking stage.
[Answer:] You will have to explain the notion of clarity and
unclarity without taking recourse to the objects (which are,
according to you, non-existent).
[Opponent:] It has been seen that states of consciousness can
be distinct even when their objects are non-existent. For
example :
The departed spirits originating from the maturation of
similar 'residual forces of action' {karma) see a river full of
filth. No river is actually present there, nor any filth. One
thing cannot be many. But the states of consciousness are
seen to be distinct, for some spirits see the same as a river

full of water, and some others see it as a river full of
blood. Thus it is ascertained that consciousness alone arises
in different forms in the manner just described, depend-
ing upon its internal conditions while the external objects as
its conditioning antecedents do not exist.
[Answer:] This is wrong because it runs into contradiction. If
you say that consciousness appears in different forms in this
manner while the external condition does not exist, we then
ask : How is it so ? •/
If consciousness takes the 'form' of blood, you should
explain the status of this blood. What is. the status of this
blood ? In the same way, one should raise questions about
the 'form' of water or the 'form'of the river. When each
word in the sentence "They see a river of filth", is
examined by analysis, the sentence becomes devoid of any
objective reference if the aggregates of matter etc., did not
exist.
Regulation of place and time is also not possible. The
departed spirits see the river full of filth in a particular place,
not in just any place whatsoever. If objects did not exist,
the reason for such regulation of place would have to be
explained.
If, for a person, the existent object is determined by, fbme
'form' or other, then it is proper to hold that some of his
cognitive states are erroneous. False cognitive states do not
reject (the existence of) the 'principal' object. Thus, one
should point out the'principal'objects in the cognition of
the river of filth etc. What applies to the cognition of the
river of filth etc., is also applicable to the cases of the magic
show, appearance of the castle in the cloud, and mirage-
water.
[Opponent:] "You imagine the 'perfuming' of action {karma)
to be in one place and the result of it in another place."
(Vasubandhu, Vimsikä). The meaning of this (UnQ) is: The
(corresponding)result should be there where the action takes
place. For him who takes objects to be différent from their
consciousness, the action will happen in one place while its
result will be in another place, and thus the action and its
result will have different substrata (but cause and effect are
supposed to have the same substratum).
[Answer:] This is not true, for I do not concede the point. I do
not concede that the action and its result are occurring in
different substrata. The action is located in the self (ätman)
and the result is also in the same locus. Thus, the (supposed)
fault (in my position) is rejected.
The objects (of my consciousness) are different from my
consciousness because they possess generic features as well as
spécifie features. They are distinct just as my consciousness-
series is distinct from the consciousness belonging to another
personality-series (i.e., another person).
The objects, are different from consciousness because they
can be established (as external) by means of knowledge. Also
because they are 'effects' (kärya, 'causally conditioned'),
they are temporal, and they are preceded by residual traces
of action (dharma).
5. Madhyänta-Vibhäga-bhäsya 1.4
"na tathä sarvathä bhävät"
The meaning is admittedly not clear. Sthiramati comments:
na tathä 'stitvän na ca sarvathä nästitvät tadgrähyagrä-
hakapratibhäsam utpadyate. I wish to thank Dr. S. Katsura
for pointing this out to me. The possible translation
would be ^{Abhutaparikalpatva is established), because
it does not exist as (it appears) and because it is not non-
existent in every respect."
§ 3.4 : MEMORY
One of the main disagreements of the Jaina pramäna-thsortet

with all the non-Jaina philosophers was concerned with the status
of the memory-experience. Memory-experience was never regar-
ded by any non-Jaina philosopher (a Naiyayika or a Buddhist) to
constitute a piece of knowledge, a prama, a cognitive awareness
which amounts to truth. Or, to put it in another way, while per-
ception and inference were regarded as valid means or ways of
knowing, memory was never regarded as a means of this kind.
The Jaina philosopher, on the other hand, contested this position
and regarded memory as another source of non-perceptual know-

ledge by refuting the arguments of the Naiyäyikas and the Bud-
dhists. Pandit Sukhlalji argued, in his Advanced Studies in Indian
Logic and Metaphysics, that this dispute was primarily due to
the reluctance of the non-Jaina philosophers to extend the use of
the term 'pramä9 to memory-experience. All philosophers agree
with the Jainas on the point that if a memory-experience
happens to be a revival of a veridical past experience,
perceptual or non-perceptual, then it is also veridical. But they
apparently want to use the term pramä in a restricted sense* so
that a veridical experience would be called a pramä, only if it is not
the repetition or revival of a past experience. To quote Sukhlalji:
That mnemic cognition is true of facts is acceptable to all
(Indian logicians), and so there is no material difference of
opinion on this issue; the difference only arises when some
agree and others refuse to call memory a pramäna. (p.46)
Panditji, however, tried to give a historical explanation of this
reluctance on the part of the Hindus, and a doctrinal explanation
of the same on the part of the Buddhist. In the Hindu tradition,
smrti, the term for memory-experience, was also used to denote
the dhannasâstras as opposed to sruti, the Vedas. Now, since it is
the cardinal doctrine of the Hindus that the dhannasâstras are
dependent upon the Vedas for their authoritativeness on dharmas
and are not independent sources of knowledge about dharma9
smrti cannot be called a pramäna. To wit : There is a systematic
ambiguity in the word pramäna (=pramä), for, it can mean either
a means of knowing or an authority, or a source for knowledge.
Therefore, if smrti which ftieant Dhannasâstras was not an inde-
pendent pramäna, then by extension smrti which also meant
memory-experience, could not also be a pramäna.
The Buddhists, however, had a different reason, according to
Sukhlalji. In the Buddhist theory, any cognitive experience that
involves thought or construction (vikalpa) would be excluded
from being a pramä or pramäna. Thus, since memory involves
thought, it cannot be & pramä.
While Sukhlalji's explanation is ingenious, it does not certainly
seem to be the whole story. If from above one surmises that the
dispute between the Jaina and non-Jaina philosophers on the
status of memory was mainly terminological, it would be wrong.
I believe that was not certainly the intention of Sukhlalji. I shall

try to focus upon the deeper reasons for the dispute over memory-
experience, and the consequent difference in theories of knowl-
edge between the opposing parties.
There is something odd in calling a memory-experience an
event of knowing, for the description of this experience is usually
prefixed with "I remember". What I remember is another experi-
ence, another (past) cognitive event. If the past event amounted
to knowing and if my present memory is Hot "playing tricks" on
me, I can now remember correctly what I had experienced. My
present experience is also aware of the fact that what is coming to
my mind along with my awareness of it is a past event. But an
event of knowing is different from an event of remembering the
first event of knowing. If the first event amounts to knowing, it
does not follow that the second would be veridical, for, I may
remember incorrectly. The converse also is not true. If I remem-
ber correctly, i.e., my memory is "fully" revived, it does not follow
that the first event was an event of knowing. If veracity is allowed
to function as a qualifying property of a cognitive event when and
only when it amounts to an event of knowing (apramä), it cannot
be regarded as automatically transmissible from the first type of
events (events of cognition) to the second type of events, events of
remembering a past cognitive event.This is, I think, at least one of
the good reasons for the reluctance of the non-Jaina philosophers
to regard a memory-experience as a pramä, an event of knowing.
I wish to connect the above argument with the traditional
arguments found in the sästras. The tradition of the non-Jaina
philosophers (in this, the Mimämsakas, some Naiyäyikas and the
Buddhists agree, see Sukhlalji1) argues that a cognitive event
becomes an act of knowing, if it grasps or reveals a fact that has
not been revealed or grasped before (cf. a-grhïtagrâhitva). In
other words, a fact not known before is what is supposed to be
grasped by an episode of knowing. An episode of remembering,
therefore, can hardly qualify to be an episode of knowing unless,
of course, the very fact of my knowing the original fact was not
known to me before. If the veracity of a cognitive event is made
dependent upon its grasping anovel fact, then another event which
repeats the first in thô sense that the fact grasped in the first is the
1
S. Sanghavi, p. 45.
same as that in the second, cannot claim the property 'veracity'.

For, we cannot kill a bird more than once. An event of correct
remembering is thought generally to be a repeat performance
in the above sense. But the property 'veracity', as we have already
seen, is not transmissible from the first act to the second. The
second act may copy or repeat the first as far as the grasping of
the same fact is concerned, but it cannot copy the other pro-
perty, viz., that of grasping a hitherto ungrasped fact. For, then
it would not be' a copy or repeat performance, and hence, not an
act of what we call'remembering'.
Take the case of an original painting by one of the masters.
There may be bad copies or even a set of cperfectv copies of the
painting. But the 'perfect' copy can copy everything of the original
but not its originality, for, then it could not be a copy by defini-
tion. Remembering in this way can never have the 'novelty' that
is expected of an act of knowing. There is something more to this
point. Suppose, in our example, a doubt arises whether the first
painting, which has been copied by several persons (and there are
good or perfect copies and bad ones) is fake or not i.e., whether
it is by one of the masters or not. Now, nothing will be gained by
looking at the set of copies, to investigate whether it is a true
replica or not. To resolve the doubt one way or other, one has to
investigate the first painting. Thus, by making sure that a memory-
experience is a correct and "full" revival of a previous cognitive
event, we do not gain any insight into the problem of deciding
whether the original event was a knowing event or not. The pro-
blem of an exact remembrance, like the problem of an exact
reproduction, is quite separate from the problem of ensuring the
first act to be an act of knowing. This analysis, therefore, shows
that there is a good reason, not just a terminological dispute, for
resisting the inclination to call a memory-experience a knowing
event.
What I have argued here can be well supported by quoting a
passage from IJdayanzLsNyäya-varttika-tätparya-parisuddhi1. This
passage was Udayana's comment on Väcaspati's rather enigmatic
statement in reply to the question why memory-experience is not
regarded as a pramä. {Tätparyatikäf
^dayana, p. 110
2
Vacaspati, p. 35
The relation between word and object is determined by peo-

ple's convention (loka). And people call such cognitive event
pramä as is non-promiscuous with the object or fact (artha) and
different from such memory-experience as is produced only
from mnemonic impression {samskära).
This might have given the impression that it is a matter of
arbitrary choice of the language-users that memory is not to be
called a pramä. But Udayana sets the matter straight as far, at
least, as the Nyâya view is concerned. A pramä is a cognitive
awareness that is in accord with the object or fact, but memory
can hardly be said to have such an accord, and hence, it is not
a pramä. I quote:
Moreover, how can memory-experience be in accord with
the object/facts ? For, it is not true that when an object is rem-
embered in a particular way, it is in that same state at that
time. For, the previous state has now ceased. If it did not, it
would not be called 'previous'. It is also not true that memory
hangs on to that object as one whose previous state has
ceased. For, we do not have the awareness of the cessation of
the previous state. If we do not have (prior) awareness of
something, we cannot have a memory of it. If we did 'remember'
such a thing, it would not be a memory. Besides, we need to
search for another unique (causal) condition [for memory, viz.,
first impression => samskära]. But we are not aware of it (i.e.,
such a condition), for, there is no past impression of it.
[Opponent :] How is it that although both a (prior) cogni-
tive awareness and a memory-experience have the same object
(revealed in both alike), we say that the prior cognitive aware-
ness may be in accord with the object but not the (later) re-
membering of it?
[Answer :] At the time of (prior) awareness, the object was
in that state in which it was, but at the time of (later) re-
membering of it, it was not in the same state.
[Opponent :] Our later cognition (i.e., remembering) may be
said to be in accord with the object if it cognizes that the
object was in that state before as it was.
[Answer :] No. Then our (present) awareness of dark-colour
belonging to an earthen pot that [was dark before but now] is
red due to its being baked (with fire), would be said, by this
argument, to be in accord with the object.1

[Opponent :] A cognition that dark-colour has ceased is cer-
tainly in accord with the object.
[Answer :] This is true. For, that object is in that state at
that time. But the remembered object is not in the same state
at that time. Therefore, memory-experience is certainly not in
accord with the object. But a cognitive (non-mnemonic) ex-
perience may be in accord with the object.2 If, however, a
cognition is in accord with the object and we have a memory-
experience of the same object, then such memory-experience is
said to be in accord with the object. Similarly, if the (prior)
cognition is not in accord with the object, the exact (undistor-
ted) remembering of it is also not so. For example, when a man
has fled after cognizing a rope as a snake, he remembers it as a
snake. Therefore, the memory-experience has 'veracity' (the
property of being in accord with the object) only to the extent
of its being borrowed from a prior veracious cognitive experi-
ence; it is not natural ( = äjänika) to memory. This (unnatural-
ness of veracity with regard to memory) is what is expressed
as (memory's) 'dependence upon another', and this has been
confused by some philosophers who were lazy to make the.
point explicit. (I think this is an oblique reference to Vacaspati
by Udayana.)
Udayana, in fact, has given two arguments in the above. First,
he has argued that memory-experience cannot be said to be in
accord with the object in the strictest sense in the way an ordi-
nary (non-mnemonic) cognitive awareness can be. Next, he has
shown, in recognition of the point that we may use such expres-
sions as 'true memory', that the memory-experience can have accord
with the object in a less strict sense, but such a property is only
a transferred epithet from the original non-mnemonic past
awareness in which the present memory is grounded.
What then is the sense in which the Jaina philosophers have
argued that memory-experience is to be called a prama, a true
cognitive event ? Does it simply mean that the Jaina philoso-
phers use the term "pramä" in a less strict sense ? It is tempting
to say so, but I would suggest another way to understand the
*Read "yathärthä" for "yathàrtà."
2
Read *'yathärthä nubhava" for "yathânubhava."
problem. If I had seen the pot to be dark when it was unbaked

and now, when it is red after being baked, I truly remember that
it was dark, the claim of the Jainas is that it is a 'true' memory,
and hence, a pramä. But Udayana has argued that this claim
hides a confusion. For, if 'true memory' means, as it should, an
exact reproduction or full revival of the past experience, then the
verbal report expressed as "it was dark" cannot be a report of
what we call a memory-experience. For, the portion of the ex-
perience expressed by "was", i.e., the pastness of the fact, cannot
be any part of the past experience (the verbal report of the past
experience was "it is dark"). And if it cannot be a part of the
past experience, it cannot be apart of the present memory. There-
fore, the verbal report "it was dark" is not that of memory, but
a present experience aided by memory. I think the dispute here
lies mainly in deciding what experience we should call memory,
my remembering a past fact (that the pot was blue) or a present
experience that the pot was blue based upon such remembering?
We can also ask : whether these two are at all distinguishable
experiences in the sense of being two cognitive events ? I shall
avoid giving an answer to this question, but instead, I shall point
out that the ordinary use of'remember' is ambiguous enough to
cover both.
There is a further point which takes us into the heart of this
dispute. The problem ofdetermining the truth of a non-mnemonic
cognitive experience is quite different from the problem ofdeter-
mining the truth of a memory. Truth may be seen as a property
of a cognitive experience, a property that is generated by factor
or factors that are either concomitant with (if we accept paratah),
or included in (if we accept svatah) the set of factors that gene-
rates the experience in question. But the correctness or accuracy
or "truth" of a memory is generated, not by a similar set of
factors, but by different ones, such as, the intensity of the previ-
ous experience, so that passage of time would not render it vague
and inaccurate. If, however, it is argued that a memory in exactly
copying a past true experience can also copy its truth, then
we have to say that it is only a copy of the property truth or
pramätva, and not the property truth itself.
CHAPTER FOUR
ONTOLOGICAL PROBLEMS
§ 4.1 : THREE RIVAL ONTOLOGIES :

NYÄYA, BUDDHISM AND JAINISM
The term 'ontology' came to be used to indicate the most general

part of metaphysics in the seventeenth-century Europe, although
for the origin of ontology as a general theory of real entities, or
as a theory of being as being, one has to go back to Aristotle as
well as the pre-Socratic philosophers of ancient Greece. Aristotle
did not use the term 'ontology'just as ho did not use the term
'logic' either. But the history of logic as well as ontology in the
Western tradition seems to start with him. Aristotle talks about
a 'first philosophy', which, he says, is about being as being,
and this is taken in later Western tradition to be the nearest
analogue of 'ontology'. For our present discussion, I shall
assume ontology to mean a general theory of 'what there is'
and try to apply it to the Indian tradition. There are many
other problems usually discussed in connection with ontology
in the West, such as, the doctrine of distinction of essence and
existence, the theory of transcendental properties of ail entities,
but these questions will not directly concern us.
The Nyâya-Vaisesika ontological problem was connected with
the Vaisesika doctrine of categories (padârtha), and the category
of substance was the focal point of this doctrine. The system of
Vaisesika categories is generally regarded as a classification of
real and fundamental entities. It is also possible to view it as an
analysis of the 'concrete' objects of our experience into their
various parts in order to form a theoretical basis for our philoso-
phical discussion. The Nyàya-Vaisesika philosophers, however,

believe that if we can analyse and classify the concrete object
of our experience in this manner into substance, quality and
action, we would achieve a satisfactory explanation of 'what
there is', i.e., an explanation of what is meant when we say, 'that
object exists'.
The Buddhists, on the other hand, think that the so-called
concrete object of our experience is at best a synthetic object, and
hence, is analysable into a number of fundamental properties or
elements called dharmas. The Buddhist conception of a dharma
is that it is by nature a non-substance (cf. anätmari).1 The question,
'What is there?' can be answered, according to the Buddhists, if
we can prepare a satisfactory list of such non-substances or
dharmas, which we can refer to, while we are accounting for and
analysing the objects of our experience. The dharmas are also in
perpetual flux, 'in a beginningless state of commotion', and
nirvana is posited as the ultimate cessation of this 'commotion'
for a person. Nirvana is also said to be the ultimate reality, the
ultimate nature of things, to be contrasted with the phenomenal
existence of dharmas, but, as I have already indicated, this pro-
blem will not be our concern in this context.
Our pre-philosophical commonsense tells that there are around
us things which somehow undergo change. Our philosophical
worries start along with our recognition of the phenomenon
of change vis-a-vis our feeling for continuity and sameness
underlying change. In India this problem was reflected in the
old dispute over Sto-cosmogony versus Asat-cosmogony (found
in the Rgveda as well as in the Upanishads). The philosophic
resolution of this dispute is to be found in the two rival theories
about causation and creation in ancient India: 1. sat-kärya-
väda 'the theory of pre-existence of the effect in the cause' and 2.
asat-kärya-väda 'the theory of new creation of the effect which
was non-existent before'. For those who préféra comparative
approach, it is significant to note that the so-called paradox of
change and permanence, of being and becoming, was as much
a live issue for the early Indian philosophers as it was for the
Greeks, i.e., the pre-Socratics. Those who were inclined toward
permanence not only posited the notion of an enduring substance
iStcherbatsky, The Central Conception of Buddhism, pp. 24-5.

Ontological Problems 271
but also argued that change was only superficial transformation

of "the existent (the substantial) from one state to another. The
Sâmkhya and the early Vedänta belonged to this group insofar
as they gave prominence to Sat 'the existent'. The Vaisesikas
belonged to the group of Asat cosmologist inasmuch as they
admitted change to be real and the function of the cause to be
the creation of new things, effects. But they also posited the
doctrine of substance, in fact, plurality of substance, and their
substantial elements were said to be persistent through changing
states. The Buddhist Asat cosmologists were very radical, for,
they argued that change alone was real and the notion of con-
tinuity or persistence was illusory, and the notion of soul-
substance was a myth. The Jaina theory, as we shall see later on,
was a compromise between the Buddhists and the Nyàya-Vaisesika.
The ontological positions of Nyäya-Vaisesika, the Buddhists
and the Jainas were necessarily influenced by their respective
stands on the problem of change and continuity. The Buddhists,
for example, were pre-eminently anti-substantialists in the Indian
tradition. This anti-substantialism culminates in their 'flux'
doctrine, according to which the components of every object,
all dharmas, change completely from moment to moment. A
comparativist might be reminded here of the anti-substantialism
of Heraclitus of Ephesus, who held, contrary to Parmenides'
denial of change, that change was incessantly occurring. But it
is not certain that the Heraclitean acceptance of change as
reality amounted to the'flux' doctrine, as it was understood by
both Plato1 and Aristotle.2 The 'flux' doctrine may be due to an
interpretation of the Heraclitean position by the philosopher
Cratylus. This would at least give credence to the anecdote
related by Aristotle about the ••'•river' example of Heraclitus.
Aristotle says that Cratylus "rebuked Heraclitus for saying
that you could not step twice into the same river; he (Cratylus)
thought you could not even do so once."3 Thus, probably
Cratylus was much closer to the Buddhists in this regard.
"Socrates : Heraclitus is supposed to say that all things
are in motion and nothing at rest; he compares them to the stream of a
river, and says that you cannot go into the same river twice."
^Metaphysics 1010a : "It was this belief that blossomed into the most
extreme of the views above mentioned, that of the professed Heracliteans,
such as was held by Cratylus..."
3
Ibid., 1010a, 1. 13-14.
The Nyâya-Vaisesikas, on the other hand, were substantiaiists

while they also accepted change much in the same manner as
Aristotle. We need not proceed in this comparative vein any
further.lt is important to understand now the Nyäya-Vaisesika
doctrine of existence as well as their notion of substance.
Vaisesika-sutra 8.14 asserts that what exists can be analysed into
three categories, substance, quality and action.1 Existence in this
system is regarded as a generic property common to the members
of the three classes, substance, quality and action. Each of these
classes has a class-property or generic property, viz., substance-
ness, quality-ness and action-ness; but these generic properties are
to be distinguished from'existence'as a generic property.2 Candra-
mati, in fact, regarded existence as a separate category (padärtha)
altogether while class-properties like substance-ness were inclu-
ded under the category of'generality' (cf. sämänya or sämänya-
vtéesà). Prasastapâda interpreted 'existence' as the highest generic
property and thus brought both existence, on the one hand, and
other class-properties like substance-ness and quality-ness, on
the other hand, under one category {padärtha) called generality.3
But still a special place was accorded to existence as the all-in-
clusive generic property which should be distinct from the inclu-
ded (vyâpya) generic properties, such as, substance-ness and
quality-ness. A particular substance is characterized by the being
of substance or substance-ness much as it is also characterized
by many qualities and probably by some actions. But it is also
characterized by 'existence' (inasmuch as it exists) which is not
to be identified with its substance-ness or with any of its quality.
The best way to explain the notion of existence in this system
is to contrast it with the notion of 'real' as well as with that of
the non-existent. Existence and the other included generic
properties are themselves REAL but not EXISTENT. For, other-
wise, we shall have to indulge in talking about the existence of
existence and so on ad infiniiwm. Similarly, the important relation
called samaväya that combines the generic property existence
with the particular existants, such as, a substance or a quality or
x
Vai$esikä-sfttra 8.14 ; "artha iti dravya-guna-karmasu".
2
Consuit Vaisesika-sfttras 1.2.7-1.2.18.
Thus it is that Prasastapâda explains sâmânya as being of two types : para
and apara. Parasämänya is existence. Substance-ness etc., are aparasâmânya.
See Prasastapâda, p. 15. For contrast, see Candramati, pp. 99-101.
an action, is also regarded as REAL but not EXISTENT.1 Thus,

the generic properties and theic inseparable relation with the
particulars are posited as real, as means of explicating the notion
of existence. Hence, they themselves should not be construed as
'existents9, to avoid the problem of self-dependence and regress
but, nevertheless, these notions, existence, generalities and sama-
väya, are claimed in this system to be real in the sense of their
being independent of our thought or mind, and thus being distinct
from a non-entity. A non-entity is NON-EXISTENT, and hence
unreal, for example, the sky flower, the son of a barren woman,
the rabbit*s horn and the unicorn (§ 2.1, § 2.2.)
Briefly stated, the 'Existents' in this system (early Vaisesika)
are equivalent to the particulars, such as, a chair, a particular
colour, a particular action. The class of existents is a sub-class of
the class of reals (see also my book on Perception, 1986/
Chap. 12). Universals (including relations) are thus not EXIST-
ENTS but REALS. Prasastapäda used two significant notions in
order to separate the class of existents from that of universals:
sattä-sambandha and svâtmasattva. The first notion characterizes
each existent, for, it means that EXISTENCE resides in the parti-
cular entity by samaväya relation. The second notion became a
bit puzzling for the later commentators. Udayana explains it as
'lacking existence' (sattä-viraha). Sridhara gives almost the same
interpretation, but also points out that 'existence' could be ascri-
bed to the universals only by mistake. Vyomasiva says that while
the first notion means that EXISTENCE is correctly applied to
the class of particulars the second notion means that EXIST-
ENCE is only metaphorically applied to the class of universals/2
The riddle of existence and non-existence is further compli-
cated in the later Nyâya-Vaisesika by the acceptance of negative
properties as real. Our negative statements, according to the later
Nyâya-Vaisesika, are expressions of something, some negative
facts. The affirmative-negative dichotomy among judgments
is interpreted differently in this system. (§2.2, § 2.3) Thus, just
as a positive judgment attributes positive property to a thing,
so also a negative judgment attributes another property, a
H am using the term REAL for the Vaisesika term padärtha, EXISTENCE
for the Vaisesika term sattä.
2
Prasastapäda, pp< 20-21; see also Udayana, p. 21; Sridhara, pp. 49-50;
Vyomaéiva, pp. 118.
negative one, to the thing denoted by the subject-term. Just

as a positive property predicated by a judgment can be
construed as a real property, so also a negative property,
absence of some positive property, predicated by a judgment can
be construed as a real property. Thus, 'the room is dark' can be
interpreted as expressing the room that is characterized by the
property of absence of light. Now, this property, absence of light,
and the like, are regarded by the Nyaya-Vaiéesika as REAL,
inasmuch as they are to be distinguished from the unreal, such
as, the round square, and unicornhood. But again, care should
be taken to note that the absence of light is not, however,
EXISTENT in this system in the sense a substance or a quality
or an action is existent. It may also ba noted that although the
negation of an entity is construed in this system as expressing
absence of that entity, a so-called negative property, no non-
entity like the sky flower or the unicorn can be negated (in other
words, absence of such non-entities will not be an acceptable
negative property in this system).1
Leaving aside the riddle of existence and non-existence, let us
concentrate on the Nyäya-Vaisesika doctrine of substance and
quality, which was at the focal point of their ontology. Several no-
tions of substance have been emphasized in the Vaisesika at some
time or the other: 1. substance as the locus of qualities and ac-
tions,2 2. substance as the substratum of change,3 and 3. substance
as capable of independent existence.4 It is difficult to say whether
the concept of substance as the logical subject was at all implied in
early Nyäya-Vaieesikadoctrine, for, it was never thoroughly work-
ed out. Later Nyâya and Buddhist logicians (notably Dignâga) de-
veloped the concept of dharmin 'property-possessor' which was
the nearest Indian analogue for a 'logical subject'. This concept
was regarded as neutral to the ontological beliefs of the logicians.
The concept of substance as the unchanging 'essence' was
1
See§2. 1, §2.2 above.
2
Vaiéesika-sûtra, 1.1.7 : "kriyâ-vad guna-vat samaväyi-käranam iti dravya-
laksartam"
8
Vaisesika-sûtra, 1.1.17 : "dravya-guca-karmanaw dravyam karapaW
sâmânyam."
^This is implied by the Navya-nyäya doctrine that a substance can exist by
itself a t the moment ft is produced, without being joined by qualities and
action (cf. utpattik$anavacchînno ghafo nirgu&onifkriyai cà).
Ontologicaî Problems 275
prevalent in the Sâirikhya school as well as in the early Vedânta

(cf. the spiritual substance), but this concept was not treated
seriously in the Vaisesika school. It is also to be noted that the
Mädhyamika Buddhists were uncompromising critics of the
doctrine of sva-bhäva 'own-nature' which was analogous to the
notion of essence or inner immutable core of things.
The doctrine of substance as the substratum of change needs
further elaboration in the present context, for this will throw much
light on the Vaisesika theory of causation and change. For any
effect, the Nyâya-Vaiéesika will identify a particular substratum
cause (samaväyi-käranä) in which that particular effect is supposed
to inhere. If the physical conjunction of two material bodies are
taken to be the effect in question, its substratum cause will be
the two bodies themselves. If the taste of a fruit is regarded as
the effect, its substratum cause will be the fruit-stuff itself. But
when the effect is nothing but a concrete individual like a pot,
its substratum cause will be the pot-parts, or in final analysis,
the atomic constituents of the pot material. Thus, the substratum
cause of an effect need not be an ever-unchanging substratum. We
do not have to posit an unchanging substantial core as the locus
of change. What is needed is only the temporal stability, persis-
tence through a period of time, of the substance which acts as
the locus of the effect.
The substances are, according to the Nyâya-Vaisesika, either
impermanent (having origin, stability and decay) or permanent
(without origin or decay).1 Material bodies of intermediate size
(called avayavin 'whole' in this system) like a pot or a table are
of the first type. They have temporal stability and can be the
loci of qualitative change. These substances are breakable into
parts and those parts into further parts. But the atomic con-
stituents of these substances along with other non-material
substances, such as, soul, sky, time and space are of the second
type, i.e., permanent. An important part of this doctrine of
substance is the ontology of the 'whole' {avayavin) as distinct
from the assemblage of parts. A material body, e.g., a piece
of chalk, is a whole which is a distinctly existent entity to
be distinguished from the integration of its parts or combination
of its atomic constituents» It is a new entity that is created as
1
See Pra&islapäda, p. 22. Udayana comments : "anitya-dravyatvarii cânyatra
nîr-avayavadravyehhya iti."
soon as the parts or atoms are put together. Moreover, draw a

line on the board with this piece of chalk and you have created
a new piece, for, some parts of the old one are lost. The seeming
identity of the new one with the old piece works for all our
practical purposes, but ontologically the two are distinguishable.
The Buddhist anti-substantialism finds its extreme expression
in the Sauträntika doctrine of momcntariness. According to this
doctrine, a seemingly stable object like a chair is dissolved into
a cluster of continuously fluctuating chair-moments or chair-
stages. The real entity is a point-instant, an exclusive particular,
an essentially unqualifiable, ineffable 'here-now' subject. Every-
thing else in this system is only a conceptual construction—an
interplay of the commonly shared imagination. In what sense
does a moment exist ? A moment exists insofar as it functions
in some way or other. Thus, Dharmakirti has argued that
to be is to be capable of functioning in some way or the other.1
If a thing does not have causal efficacy, it does not exist. Starting
from this initial position, Dharmakirti and his followers have
formulated a proof of their'flux'doctrine. It will be interesting
to note the crucial steps taken by the Buddhists in proving the
'flux'doctrine:2
1. To be is to do something, i.e., to function or to have causal
potency.
2. To have causal potency means to be actually doing what
is supposed to be done.
3. If something has causal potency at a particular moment it
must do its work at that moment. (This is a rephrasing
of 2.)
4. If something does not do a work at a given moment, it
must be causally impotent to do that work. (This is a
contraposition of 3.)
5. The same thing cannot be both causally potent at one
moment and impotent at another {next) moment, for
potency and impotency are contradictory properties, mu-
tually incompatible.
a, Ch. 2, v. 3 : "arthakriyàsamarthaw yat tad atra para-

mârthasat, p. 100.
2
For the most elaborate presentation of the 'flux' doctrine, see Jnänasri-
mitra, pp. 1-159.
6. Therefore, the thing at the moment of its potency must be

held to be oritoiogically different from the thing at the
moment of its impotency. A difference in qualities implies
difference in the thing itself!
7. Everything, in this manner, can be shown to be in perpetual
riux. We cannot step twice into the same river !
The most crucial step is taken by the Buddhist here when he
identifies causal potency with actuality or actual doing. In other
words, the notion of potentiality is completely rejected. If a thing
exists and it is capable, it must function without lying in wait
for anything to come and help. If we posit two different func-
tionings at two different moments, we have to construe them
as belonging to two different things or objects. In each moment
a new object (bhäva) emerges when a new functioning sets in and
the old functioning perishes. Thus, what exists is the ever-
fluctuating here-and-now. Even the ontology of stages or mo-
ments is not quite satisfactory to the Buddhists. For, moments
or stages are also hypothetical abstractions in the face of conti-
nuum. Thus, we have to say that there is only process, only flux,
without something being there to fluctuate. There is only trans-
migration without there being any transmigrating soul (cf. the
'non-souF doctrine).
Udayana, setting forth a defence for the Nyâya-Vaiéesika
doctrine of substance, has criticized the above, argument of the
Buddhist by pointing out that it is essentially dependent upon
the total rejection of the notion of potentiality. Why, asks
Udayana, is it to be assumed that the causally potent cannot
(and should not) 'waif for its accessories. Causality operates with
two mutually compatible notions: suarüpayogyatä 'potentiality*
and phalopadhäyakatä 'actuality'; the former relates to the
general while the latter relates to the particular. If the Buddhist
equates potentiality with actuality then, Udayana argues, part of
the Buddhist argument is reduced to tautology, for, he would have
to say that x is actually functioning, because it actually functions.
And tautology is not a good philosophic argument. In fact,
potentiality is explained by Udayana not as an essential consti-
tuent of the thing, but as the mere presence of the thing coupled
with the absence of some accessory or the other and the consequent
absence (or non-arising) of the effect.1 Thus, Udayana argues, if

x does not cause y when and only when z is absent then it follows
that when z is present x produces y. This is only another way of
saying that z is an accessory to x in bringing about y. Besides,
the properties of causing y and not causing y are not two
mutually incompatible characters like cow-ness and horse-ness.
A cow, of course, can never be a horse. But a thing, if it is not
just a flux, can cause y at time tx and may not cause y at time f22.
In fact, what Udayana says is reminiscent of Aristotle's rejection
of potentiality :3
There are some who say, the Megaric school does, that a thing
'can' act only when it is acting, and when it rg not acting it
'cannot* act, e.g., that he who is not building cannot build,
but only he who is building, when he is building; and so in
all other cases. It is not hard to see the absurdities that attend
this view.
It is rather significant that the arguments and counter-arguments
of Dharmakîrti and Udayana were presupposed much earlier in
the Megaric school as well as in Aristotle.
Dharmakîrti's argument to prove his *flux' doctrine was not
entirely an innovation in the Buddhist tradition. He must have
derived his idea from Nâgârjuna's dialectic. Nägärjuna, for
example, has argued that if something exists it should exist
always, and if it does not exist at one time it cannot exist at any
time.4 This is how Nägärjuna has criticized the concept of exis-
tence and 'own-nature'. Dharmakîrti first assumes that to he
means to have causal potency. Then he argues: if something
has causal potency it must be functioning all the time, and if
something does not have the causal potency at one time it would
never have it at any other time. But Nâgârjuna's philosophic
conclusion is rather different from that of Dharmakîrti. With the
above argument Nägärjuna wishes to avoid the extremes of
eternalism and annihilationism and follow the Middle Way.
Dharmakïrti, on the other hand, intends to conclude that since
1
See the first chapter of Ätmatattvaviveka, specially pp. 16-25: "yad yad-
abhäva evayan na karoti tat tat-sadhhâve tat karoty eveti"
2
Ibid., p. 24.
^Metaphysics, 1046b, 28-32.
é
Mâdhyamikci-éâstrat Ch. 7, verses 30-31.
functioning is instantaneous, existence is also instantaneous.

And when we think of Udayana's counter-argument, we are
again reminded of Aristotle.1
Again, if that which is deprived of potency is incapable, that
which is not happening will be incapable of happening; but
he who says of that which is incapable of happening either that
it is or that it will be will say what is untrue;....But we cannot
say this, so that evidently potency and actuality are different
(but .these views make potency and actuality the same, and
so it is no small thing they are seeking to annihilate)....
The Jaina ontological position is influenced by both the
Buddhists on the one hand, and the Nyäya-Vaisesika on the other.
The Jainas were also substantialist, but in a very qualified sense
of the term. Their conception of existence (sat) is intimately
related to their doctrine of substance. The Tatlvärthasütra 5.29
asserts: "What there is, has the nature of substance."2 And the
next sütra (5.30 in the Digambara tradition) adds: "What there
is (the existent), is endowed with the triple character, origin,
decay and stability (persistence)."3 The TaWvärthabhäsya explains
that whatever originates, perishes and continues to be is called the
existent; anything different is called the non-existent.4 The
next sütra asserts that the existent is constant, for, it never gives
up its being (essence?).
In sütra 5.37, the substance is again characterized as the pos-
sessor of qualities (guna) and modes (paryäya). Here the broad
category 'attribute' is apparently broken into two sub-categories,
qualities and modes. The distinction between qualities and modes
is not found in the sütra. Umäsväti points out that qualities are
permanent attributes of the substance while the nïodes are only
temporary attributes which are subject to origin and decay.5
In the above analysis of the Tattvärthasütra, two compatible
notions of substance are emphasized: 1. substance as the core of
change or flux and 2. substance as the substratum of attri-
^Metaphysics, 1047a, 10-20.

2
"Sat dravya-lak$anant\ Tattvärthasütra 5.29.
Bit
Utpâdavyayadhrauvyayuktam sat", ibid., 5.30
4
See Umâsvàti, under sütra 5.29
ß
lbid., under sütra 5.40.
butes. Kundakunda combines these two notions as he defines

substance in his Pravacanasâra:1
They call it a substance, which is characterized by origin, per-
sistence and decay, without changing its 'own-nature', and
which is endowed with qualities and accompanied by modi-
fications. For, the 'own-nature' of the substance is its existence
(sad-bhäva) which is always accompanied by qualities and
variegated modes, and at the same time, by origin, decay and
continuity.
The Vaisesika school, as we have already seen, emphasized
both these aspects of substance, but did not equate the * own-
nature' of the substance with EXISTENCE. Aristotle, who, in
fact, suggested several notions of substance either implicitly,
or explicitly, remarked in Categories.21
The most distinctive mark of substance appears to be that,
while remaining numerically one and the same, it is capable
of admitting contrary qualities.
In Metaphysics, Aristotle also implied that the substance is
what is independently existent, for, existence, in the proper sense
of the term, applies to substances only, and qualities and relation
have a secondary existence, a parasitic mode of being.3
Therefore, that which is primarily, i.e., not in a qualified sense
but without qualification, must be substance.
The Jainas too identify the notion of 'it is' or 'it exists' with
that of substance, and they then explain that *it is' means that it
is endowed with the triple character of origin, decay and stability.
In fact, the Jainas explicated the notion of substance in such a
way as to avoid falling between the two stools of being and
becoming. It was a grand compromise of flux and permanence.
The Jainas inherited from Mahâvïra and his later followers the
well-known doctrine of 'many-natured' reality (cf. anekänta-väda),
and thus a 'compromise' position was only an important trait
of their creed. The substance, in their analysis, is being, it is also
becoming. Kundakunda observes that a substance has both
natures: from the standpoint of one's 'own-nature' it is being (sat,
1
Kundakunda, Pravacanasâra, Ch. 2, verses 3 and 4.
Categories 4a, 10-14.
^Metaphysics 1028a, 29-30.
unchanging), and from another standpoint it has triple character,

origin, decay and continuity, i.e., fluctuations.1 Siddhasena
Diväkara repeated the point more forcefully:2
There is no substance that is devoid of modification, nor is
there any modification without an abiding something, a
substance. For, origin, decay and continuance are the three
constituents of a substance.
It should be noted that the notion of continuity involved in
the triple character of the substance is not identical with the
notion of permanence x>f the substance. The former notion
means persistence or continuance (cf. pravähanityatä). The later
notion means immutability. It is the notion in the background
of which the triple character of origination, destruction and
continuity becomes understandable. The notion of continuity,
on the other hand, is essentially dependent upon origin and
decay. Thus, Kundakunda observes :3
There is no origin without destruction, nor is there any des-
truction without origin, and neither destruction nor origination
are possible without what continues to be.
The Jainas were well aware of the Mâdhyamika critique of the
'own-nature' concept as well as the problem involved in the doctrine
of the permanent substance. It is true that the immutability of
own-nature invites a host of problems. But the notion of flux,
the Jainas points out, is not sacrosanct. Thus, just as the Buddhists
argue that there is only fluctuation, there being no permanent
being, the Jainas take the bull by the horns and answer that if
there is no permanence there cannot be any change or fluctuation,
for, it is only the permanent that can change. It is only the persist-
ing soul that can transmigrate 1
When the Tattvärthasütras defines substance as the substratum
of qualities and modes, it was probably influenced by the Vaise-
sika school. Thus, Siddhasena points out that the rigid Vaisesika
concepts of substance and quality were not compatible with the
Jaina ontological principle of anekäntatä 'many-naturedness' or
<non-onesidedness\ In fact, it would be as good as a heresy
1
Kundakunda, Pravacanasâra, Ch. 2, verse 7.
2
Siddhasena, Sanmati, Ch 1, verse 12.
3
Kundakuöda, Pravacanasära, Ch. 2, verse 8.
in Jainism if one intends to maintain a rigid distinction bet-

ween substance and quality. The notion of triple character,
origin, decay and continuity, embodying the principle of (con-
ditioned) reality, was derived from the Buddhist source. The
Buddha, for example, predicated this triple character of all the
conditioned (samskrta) entities. Thus, in the Ahguttara I, the Bud-
dha said i1
Of the conditioned entities, monks, the origin is conceived,
even so their decay and their stability (persistence).
Nägärjuna, however, directed his dialectical attack against
the notion of the conditioned (sarriskrta), and concluded :
Since the notion of origin, persistence and decay cannot be
established, the conditioned does not exist. And if the condi-
tioned is not established, how will the unconditioned be esta-
blished ?
But why then did the Buddha speak about the triple character
of the conditioned entities ? Nägärjuna replied:2
Just as magic, dream and the cloud-castle are unreal (but,
nevertheless, are spoken about) so also origin, stability and
decay have been described.
The Jainas postulate the triple character in the case of each
event, each happening or change of state. Each fluctuation
embodies origin, continuity and decay. Samantabhadraillustrates
the point as follows : if a golden pot is destroyed and a golden
crown is made out of it, destruction, origination and continuity
—all three—happen simultaneously and give rise to sorrow, joy
and indifferent attitude in the minds of three different kinds of
people, those in favour of the pot, those in favour of the crown,
and those in favour of the gold stuff.3
Siddhasena has shown great philosophic insight in expounding
the Jaina ontologicai problem. According to him, reality can be
viewed from two important standpoints, being and becoming,
permanence and change. That is why Lord Mahàvïra acknow-
ledged only two nayas or standpoints : 'substance exists' and
1
Quoted by Candrakirti, see Nägärjuna, Mädhyamika-tästra, p. 73.
*Mädhyamika-sästra, Ch. 7, verse 34.
3
Samantabhadra, Äptamlmüitisü> Ch. 3, verse 57.
'modification exists'. If x is an element of reality, then, according

to Siddhasena, x can be viewed as a SUBSTANCE from the stand-
point of being, and as a PROPERTY from the standpoint of
becoming. The standpoint of 'becoming' (modification) reveals
that everything originates, stays and perishes; the standpoint of
'being' ('it is') reveals everything as existent, eternally without
birth or decay. And, Siddhasena asserts, there cannot be being
without becoming, or becoming without being; therefore, a sub-
stance (=* reality) is defined as the combination of being (the
existent) with becoming (origin, stability and decay)1
The 'being' aspect is, according to Siddhasena, the result of
generalization while the 'becoming' aspect isthat of particulariza-
tion. In our ordinary description of things, we necessarily combine
the general with the particular. From the point of view of the
highest generalization, a thing is described as 'it is' which reveals
the permanent being, the substance. But when, in ordinary des-
criptions, a thing is called a piece of wood, or a chair, or a red
chair, we have an intermixture of'being'and 'becoming' aspects.
Insofar as the thing is identified as a non-fluctuating substance,
it is the cbeing' standpoint. And insofar as the attributes of the
thing, such as, being a piece of wood, being a chair, or redness,
are revealed by the description, it is the 'becoming' standpoint.
Qualities are nothing but modes or states of the substance. In any
characterization or description of the thing there is thus an over-
lap of 'being' and 'becoming' standpoints, until we reach the
ultimate particularity, pure BECOMING, i.e., the point-instants
{k$anas) of the Buddhists,2
Thus, the Jaina conception of reality, in bringing together the
opposing viewpoints of the Buddhistsand the Nyäya-Vaisesika,
comes very close to that of Whitehead, according to whom the
chief aim of philosophy is the "elucidation of our integral experi-
ence" of both the flux and permanence of things. Whitehead has
said that philosophers who have started with 'being' have given us
the metaphysics of 'substance' and those who have started with
'becoming' have developed the metaphysics of flux. But White-
head points out the inseparability of the two:3
Siddhasena, Sanmati, Ch. 1, verses 11 and 12.

2
Ibid., Ch. 1 verse 9.
3
A.N. Whitehead, Process and Reality, pp. 240-2.
But, in truth, the two lines cannot be torn apart in this way;
and we find that a wavering balance between the two is a char-
acteristic of the greater number of philosophers. Plato found
his permanences in a static, spiritual heaven, and his flux in the
entanglement of his forms amid the fluent imperfections of the
physical world.... Aristotle corrected his Piatonism into a some-
what different balance. He was the apostle of'substance and
attribute', and of the classificatory logic which this notion
suggests.
In the Indian context, one may observe that the Buddha's search
for nirvana, the unconditioned state, freedom from suffering
or duhkha, spelt out a philosophy for the later Buddhists, accord-
ing to which the flux of things, impermanences, "the fluent
imperfections of the physical world" are identical with suffering
(duhkha, cf. whatever is impermanent, is suffering). And NIR-
VÄIÂ, the unconditioned state, is actualized with the cessation
of this duhkha. The Vaisesikas, on the other hand, were, much
like Aristotle, the apostles of substance-and-attribute duality.
§ 4.2 : CAUSALITY
Causality implies a host of philosophic problems which are

pertinent to different branches of philosophical studies. My aim
hereis, however^ a modest one. While explaining the notion of
cause as it was understood by the different philosophical schools
of classical India, I shall try to show that (a) concern for the
notion of causation was taken more seriously in some Indian phi-
losophical schools than it had been in some of their Wertern
counterparts, and (b) that the meaning of'cause' (kärana) is rçuch
wider in Indian philosophy than it is in the West.
To substantiate the second point first, I shall first very briefly
refer to the classification of 'causal conditions' as found in the
Abhidhannakosa of Vasubandhu and then to another classifica-
tion found in the Vaisesika school. Two different terms, heiu and
pratyaya, are used to coverall causal notions in the Abhidharma
system« There is, however, no essential difference between these
two concepts, the six types of hetu being included under the
broader notion of pratyaya.
The Abhidharmakosafirstexplains six types of hetu, of which
the first is called the käraria-hetu, the 'prototype' of a hetu. The

definition of a kärana-hetu reflects the attitude of the Abhidharma
school toward the general notion of a cause. In fact* to be a cause
in this sense means to exist without being a 'hindrance 5 (vlghna) to
the production of the effect in Question. Thus, if x is a dharma that
originates, then any dharma other than the 'own-being' (svabhäva)
of x may be called its käraria-hetu.1
Four types of pratyaya are : (1) hetu-pratyaya, (2) älamhana-
pratyaya,(3) samanantara-pratyaya, and (4) adhipati-pratyaya.
Of these, the first includes the 'remaining' five hetus in the forego-
ing classification while the last includes the 'prototype 9 , that is, the
kärana-hetu.2 The second and the third are found useful only
with regard to the states of consciousness. The älambana is the
'causal basis', that is, the object, of a particular state of conscious-
ness which is regarded as an eifect or event (See § 3.2). The samanan-
tara is the 'causal precedent', that is, the preceding moment of
consciousness conditioning the succeeding moment. It should also
be mentioned that in the remaining five hetus (a discussion of
which I skip here) of the six types mentioned earlier are included
hetus like a sahabhü hetu, where two dharmas originating simul-
taneously are said to condition each other mutually.
This breathtaking classification of the Abhidharma will be
enough to show that hetu or 'cause' is being used here in its widest
possible sense. It is sometimes pointed out that we should not
probably translate hetu or kärana in this context as a 'cause' or
a 'causal condition5. But I am in favour of retaining such transla-
tions simply because such terms as 'hetu9 and 'pratyaya" mean
almost the same thing (and, by the same token, share the same
ambiguity) in ordinary Sanskrit as terms like 'cause' and 'causal
condition' do in ordinary English. It is a philosopher's privilege
to use an ordinary term in a highly specialized sense.
If to cause an effect in the Abhidharma means not to obstruct
the production of that effect (that is, that dharma), to cause an
effect in the Vaisesika system means to be its immediate (and
'unconditional') antecedent. The Vaisesikas speak of a causal
substrate or a 'substantial' cause {samaväyikärana) which is the
substance where the effect occurs (through the relation of inher-
1
See Abhidharmakoto and Bhäsya of Vasubandhu, Part I, ed. Swami
Dwarikadas Sastri (Varanasi : Bauddha Bharati, 1971), pp. 279-2.
2
Ibid., p. 349.
ence). Ordinarily, this is the material cause of an effect, even

comparable to the 'material' cause in the fourfold division of
Aristotle. But strictly speaking, this is a wider notion in the
Vaisesika system, since such 'nonmaterial' substances as the soul
and the physical space are said to be causal substrates of suitable
effects. For example, a state of consciousness as an event is said
to occur in the soul (that is, the person) which is its causal sub-
strate.
All causal conditions other than the causal substrate that are
relevant to the effect are classified by the Vaisesikas into two
groups : asamaväyin cause (non-substantiai, literally ôn-inhe-
rent' cause), and the nimitta 'efficient* cause. The former forms
an artificial group which includes only qualities (and relational
qualities like 'conjunction' sarriyoga) that inhere in the causal
substrate and are causally relevant to the effect.1 For example,
the colour of the threads causes the colour of the cloth, or con-
junction of different parts of a table causes the table in this
manner. The group of 'efficient' cause includes everything else
that is causally relevant to the effect in question. Thus, the potter,
along with the potter's wheel, rod, water, etc., is called the 'effi-
cient'cause of the effect, pot. Aristotle's notion of the 'efficient5
cause, in a liberal interpretation, can match this Vaisesika notion.
But nothing like Aristotle's notion of the 'final' cause or the
'formal' cause can be found in the Nyâya-Vaisesika school.2
The distinction of a 'non-substantial' cause from the group of
efficient causes seems to be artificial. It was probably based upon
the awareness that as long as the effect, the colour of a cloth,
exists its 'non-substantial' cause (that is, the colour of the thre-
ads) should, like the causal substrate, also exist. But an 'efficient'
cause like the potter's wheel may wither away (after the effect is
born) without affecting the effect in any way. In other words,
the father may die after the son is conceived but what inheres in
the son's body and limbs must stay as long as the son is alive.3
A notion parallel to that of a 'non-substantial' cause of the
Vaisesikas is hard to find in the Western tradition.
1
SeePraéastapâda, Padärthadharmasamgraha, ed. Durgadhara Jha (Varanasi:
Sanskrit Visvavidyalaya, 1963), pp. 244, 246.
2
This is a general comment, which may be subject to qualification.
3
For this notion of nUnittakärana, see Vyomavati of Vyomaeiväcärya, com-
mentary on Prasastapädabhäsya, ed. Gopinath Kaviraj and Dhundhiraj
Sastri, (Banaras : Chowkhamba, 1930), pp. 140-2.
Concern for a causai notion was fundamental in almost all

schools of Indian philosophy. We should remember that philoso-
phic activity in India arose out of the cosmogonie speculations of
the Vedas and the Upanisads. The all-important business of phi-
losophy was to attempt to discover some simple, unitary cause
for the origin of this complex universe. Various alternative theo-
ries were propounded from the very early period, as is well evid-
enced in the passages of the Svetâsvatara Upanisad and in the
Nyäyasütra, Chapter 4.1 Some of the main views about the origi-
nal cause of the universe were: (a) time (b) nature or €own-nature\
(c) pradhäna'the un manifest' matter, (d) god, (e) atoms, and
(f) cyclical dependent origination. Of these, I shall briefly allude
to the three important views, the Sämkhya. the Buddhist, and
the Nyäya-Vaisesika views, (see also § 4.1 )
The earliest critique of the notion of cause is to be found in
India in the satkäryauäda doctrine of the Sämkhya school, which
was historically the earliest of Indian philosophic systems. The
doctrine means that the so-called effect pre-exists in its cause,
causation being merely a change or transformation from one
state to another, while the original'thing' (cf. dharmin) remains
constant and unchanging. An effect means a change in only the
attributes or characteristics of the thing, a new state of affairs
means manifestation of what was potentially present (cf., sat) in
the early state of affairs, that is, in its so-called cause.2 The Säip-
khya metaphysics posits an ultimate, original matter, the pradhâna
or avyakta (the 'chief or the 'unmanifest') from which the whole
material world evolves. It is also admitted that the potential
becomes actual at every moment and thus transformation of the
world is automatic and instantaneous. This aspectof the Säipkhya
theory might have influenced the Buddhist doctrine of universal
flux (§4.1).
The Buddha's doctrine of impermanence (anityatä) was deve-
loped by the later Buddhists as the doctrine of universal flux.
Causation in Buddhism is spanned by its doctrine of conditioned
(or dependent) origination (pratitya-samutpäda). In Abhidharma,
for example, every samskrta dharma, that is, every effect, is said
*See Nyâyasûtra and Nyäyabhösya, ed. Ganganath Jha, (Poona : Oriental
Book Agency, 1939), pp. 246-70.
2
See Sämkhyakarikä and Yuktidipikä,ed. Ramasamkar Tripathi, (Varanasi :
B. Tripathi, 1970), pp. 59-69.
to originate merely depending upon what is called its pratyaya

(or hetu). The Buddhists reduced causal dependence to nothing
more than a mere sequence, a very loose relation usually expres-
sed in the formula : if x is, y arises (cf., asmin sail idam bhavati
hrasve sati dirgharp, yathä). The Mädhyamikas rejected the notion
of svabhäva 'own-being' of things as well as origin (udaya) and
decay (vyaya)of things (bhävas).1 Thus, they might have paved the
way for the doctrine of non-origination (ajäti-vädä) ofGauda-
pâda.2 The popular belief that an effect is brought about by its
causes is reduced to an absurdity by the Mädhyamika dialecti-
cians.
The doctrine of universal flux is the result of the Buddhist
rejection of the notion oî potentiality. To be causally potent
means here only actual production of the effect. Everything is in
a flux, there being no stability (sthiratva). Causation is simply
incessant succession of events. Unlike the Sämkhya, the Buddhists
rejected the notion of any unchanging core of things, any un-
changing dharmin underlying the seeming fluctuations.3
The Nyâya-Vaisesika school opposed the Sâipkhya by its doc-
trine of a-sat'kärya-väda, according to which an effect is a new
creation, and hence, numerically different, from its cause. The
description of causation in this school is closer to the common-
sense view of a cause. True to the spirit of empiricism, causal
relation is described here also as one of invariable sequence.
Udayana asserted, in answer to the scepticism about causality,
that a causal explanation of an event is needed unless we want
to settle for a total accidentally or 'whimsicality' of every-
thing. A particular effect happens at a particular moment, not
always. This is what is called the 'temporality' (kädäcitkatua)
of an effect, and this temporality implies dependence of the effect
upon something other than itself. Causal relation is nothing more
than this obligatory dependence.4
x
See Mädhyamikäkärikä of Nägärjun^, ed. P. L. Vaidya (Darbhanga :
Mithila Institute, I960), chapts. 1 and 15.
2
See Ägamasästra of Gaudapäda, ed. Vidhusekhara Bhattacharya (Calcutta:
University of Calcutta, 1943), pp. 101-12.
3
For the Buddhist criticism of potentiality {sâmarthyà), see Ratnakïrti-
nibandhâvalï, ed. Anantalaî Thakur (Patna : K. P. Jayaswal Research
Institute), pp. 62-76.
4
See Nyäyakusumanjali of Udayana, eds. P. Upadhyaya and D. Sastri,
(Varanasi : Chowkhamba, 1957), pp. 41-60.
Otîtologica! Problems 289
The Nyäya school rejected also the notion of saktiy 'power',

'efficacy' or 'force', connected with causation. The Mïmàrrisakas,
being consistent with the common belief, argued in favour of an
efficacious power or sakti present in the cause to produce the
effect—a power which can be destroyed by the presence of an
'antidote' (pratibandhaka) and can conceivably be resuscitated
by an'antidote to the antidote'(cf., uttejaka). Thus, fire burns
because of its power to burn—a power which can be either des-
troyed or resuscitated under suitable conditions. While refuting
the notion of sakti, Udayana solved the problem presented by
the influence of antidotes by boldly asserting that causation
implies presence of not only'positive' causal conditions, but also
of relevant Negative' conditions. For a particular effect to happen,
the absence of the relevant antidote is also needed as one of its
causal conditions.1
The Navya-nyâya treatment of causation is interesting in many
ways. The notion of invariable sequence is explained as holding
between generalities rather than between particular events. Thus,
Sasadhara defined a cause as one belonging to a class, individual
members of which invariably precede individual members of
another class, the relevant effect-class (cf., käryaniyatapürva-
vrttijätiyatvä).2 A potter's wheel is said to be a cause of a
particular pot, because it belongs to the class of those wheels,
members of which are seen to invariably precede the production
of members of the pot-class. The notion of invariable sequence
is, however, to be derived from experience, from what is called
anvaya and vyatireka (seeing cases of association and absence).
Although in rejecting the notion of 'efficacy' or 'power' Nyäya
resembled the position of David Hume, the doctrine of invari-
able sequence was not propounded here exactly in Humean
spirit. For Hume, it is only the mind that spreads itself on
external objects and conjoins them as cause and effect while
nothing really exists between them to be so conjoined.3 This
x
Ibid., pp. 103-7.
2
See Nyâyasiddhântadïpa, ed. V. P. Dwivedi and Dhundhiraj Shastri,
(Benares Cantt : Lazarus & Co., 1924), pp. 89-92. A critical edition of
this text has already been published by me in the L.D. Indoiogical Series,
Ahmedabad, 1976.
^ a v i d Hume, A Treatise of Human Nature, ed. L.A. Selby-Bigge (Oxford:
The Clarendon Press, 1951), Book I, Part III, See.* XIV, pp. 166-72.
is more like the Buddhist view than the Nyâya view. For Nyäya,
invariable sequence is discovered by the mind but it exists
between extramental realities like universals or class characters
Perhaps Nyâya shuns the Humean empiricism while it asserts
its doctrine of real universals. In fact, although the early Nyâya-
Vaiéesika doctrine of universal was modified by Navya-nyâya,
it still maintained that certain class-characters were real in
order to explain, among other things, the relation of cause and
effect.1
Thus, it was felt in Navya-nyäya that the criterion of invari-
able sequence was not enough for distinguishing causal conditions
of a particular effect. For example, the production of a pot is
preceded invariably not simply by the potter's wheel, but also by
the colour and circularity of the potter's wheel. But the colour of
the wheel is immaterial and irrelevant to the production of the pot.
To exclude such irrelevant items from being considered as causes
of the particular effect in question, such Navya-nyäya authors as
éaéadhara and Gangesa introduced the notion of what they
called ananyathäsiddhatva 'unconditionally'. The invariable
presence of the colour of the wheel before the pot is produced is
conditioned by the presence of the wheel itself, and hence, it need
not be taken into account while we consider the relevant (causal)
conditions for the effect in question. By the same token Navya-
nyäya excludes the cause of a cause from being considered as a
cause of a particular effect. This takes care also of a conceivable
case where each time a pot is produced on the potter's wheel, a
donkey always walks by immediately before the event. Experience
of invariable sequence may demand that we construe the two
events as causally related. But Nyâya claims that the 'unconditio-
nality' criterion can save the situation, since it is possible to find
a reasonable explanation of each case of such appearances of a
donkey. If such a reasonable explanation is found, the donkey's
appearance will no longer be an 'unconditional' antecedent.2
If the Navya-nyâya analysis of the causal relation seems to be
somewhat embarrassing to a Humean empiricist, we may intro-
1
See my Epistemology, Logic and Grammar in Indian Philosophical Analysis,
(The Hague : Moutoii, 1971), pp. 71-77.
2
See, for a discussion of ananyathäsiddhatva, Tattvacintâmani of Gangeéa,
ed. Kamakhyanatha Tarkavagisa (Calcutta : Asiatic Society of Bengal,
1884-1901), Part 2, vol^J, pp. 154-5.
Ontoîogicaî Problems 291
duce here a brief discussion of some contemporary problems of

causation. David Hume's critique of causality brought into focus
two important questions in contemporary philosophy. The first
question is : whether there is, after all, any necessary connection
between a cause and its effect ? The second, and perhaps the
more basic question is : whether the notion of cause is at all a
viable concept in philosophy, useful for explanation and under-
standing of what are called 'events' ?
Because of the muddle and complexity associated with the no-
tion of 'cause', some modern philosophers (notably Bertrand
Russell ) despaired of making any sense of the word 'cause', and
hence recommended "its complete extrusion from the philoso-
phical vocabulary.' 1 The rise of modern theoretical physics, some
doctrines of which throw doubt even upon the once universally
acclaimed universality principle of causation (the principle that
states, "every event must have a cause"), has contributed further
to the modern despair about the notion of cause. But whatever
may be the situation in theoretical and higher physics, it is almost
undeniable that the concept of a cause is quite usefulinthe com-
mon affairs of life, in applied technology, in moral fields, in law
and jurisprudence. Thus, I believe, a philosopher can hardly
afford to be totally indifferent to this concept. It will be enough
to point out here that respectable modern philosophers, such as
A.J. Ayer, R.G. Collingwood, CJ. Ducasse, and G. E. von Wright,
have found the concept of causality to be useful.2
Modern despair about the rrotion of cause can partly be ascri-
bed to Hume's rigid demand for the empirical analysis of causa-
tion. Hume used the notion of similarity or resemblance in
giving an empirical explanation of cause. A cause is defined as
"an object precedent and contiguous to another, and where all
the objects resembling the former are placed in a like relation of
priority and contiguity to those objects that resemble the latter." 3
Unfortunately, the notion of similarity invities a number of phi-
losophic problems in empiricism. For one thing, an event today
B.Russell, Mysticism and Logic (1917; reprint ed., Garden City,
N.Y.: Doubleday Anchor Books), pp. 174-201.
2
A.J. Ayer, Foundations of Empirical Knowledge (London : Macraillan &
Co., 1951), Chap. 4. R.G. Collingwood, An Essay on Metaphysics (Oxford,
The Clarendon Press, 1940), Part 3-C. CJ. Ducasse, Nature, Mind and
Death, Part II (La Salle, Illinois : The Open Court Publ. Co., 1951).
8
David Hume, op. cit., p. 172.
must be similar to an event of yesterday in some essential res-

pects in order to be caused by a similar cause. We may recall
that Navya-nyäya introduced the notion of real universals or class-
character. To avoid this quandary of empiricism, a follower of
Hume might argue that similar effects, in order to have similar
causes, must be similar only in certain crucial or relevant res-
pects. It is, however, easy to see that the notion of such "rele-
vance" cannot be explained without resorting to the causal
notion, and thus, it will involve us in a circularity.
The question of relevance may enter in the discussion of causal
relation in another way. If, following Hume, the notion of neces-
sity is completely eliminated and causation is reduced to mere
invariable sequence, then, as Thomas Reid pointed out against
David Hume, we have to admit that day is the cause of
night and vice versa, since day is seen to be invariably followed
by night and night by day.1 J.S. Mill referred to this criticism of
Reid and defended Hume by introducing the notion of'uncon-
ditionality'in defining the notion of cause. Mill argued as follows :
"If there be any meaning which confessedly belongs to the term,
it is unconditionalness.. ..The succession of day and night evident-
ly is not necessary in this sense. It is conditional on the occur-
rence of other antecedents." 2 Since it is conceivable that there
could be circumstances (for example, sudden stopping of the
rotation of the earth) under which day will not be followed by
night and vice versa, Mill said, the two are not unconditionally
conjoined, and hence, not causally connected. It is unfortunate
that Mill here abandoned, quite unconsciously, the very point
of the empirical analysis of Hume, andreintroducedthe notion of
necessity disguised as 'unconditionally'. Besides, if, as Mill ex-
plained, conceivability of the circumstances under which a known
invariable sequence may not follow is enough to upset the causal
connection between events, then very few events could be said
to be causally connected.
Mill's own explanation of the'unconditionally' criterion was
thus hardly satisfactory. However, one may justify the 'uncon-
^ h o m a s Reid, Essays on the Intellectual Powers of Man, Essay 4, included
in Works, ed. Sir William Hamilton, (Edinburg : Maclachlan & Stewart,
1846-1863).
2
J. S. Mill, A System of Logic, 8th ed., Book III, Chap. V, Sec. 6 (1843;
New York : Harpar, 1881).
ditionality' criterion in order to distinguish irrelevant antece-

dents from the causally relevant antecedents. But then, we are
thinking of what Navya-nyäya calls ananyathasiddhatva (which Ï
have translated above as the criterion of 'unconditionality'), and
not Mill's 'unconditionally' criterion.
The point at issue may be made in the following way. Suppose,
there is a unique tribe, each adult member of which has produ-
ced, after going through the usual tribal training period, a parti-
cular craft. This craft is unique to this tribe, and let us assume
that no one else on earth has succeeded in producing that craft.
Now, suppose that each such adult member, after his training
period and before producing the craft, has got up from bed at
6 A.M. on two successive Sundays. If invariable sequence is eno-
ugh to establish causal connection, then not only the tribal train-
ing period, but also their getting up at 6 A.M. on two successive
Sundays should be considered causally relevant ! In fact, to be
true to the Humean spirit of empiricism, one has to accept such
an apparently absurd consequence. The question of what is rele-
vant, namely, the training period, and what is not relevant to the
production of the craft cannot be decided unless some notion
like the Navya-nyäya idea of ananyathasiddhatva is introduced.
It may be noted that an Abhidharma Buddhist can nicely tackle
the problem here by taking the bull by the horns. As noted
earlier, he can claim that when something has been produced,
nothing that was prior to it was really irrelevant to its production.
Vasubandhu exemplified the point as follows : When the village
folk have successfully organized a feast without interference
from the proverbial village troublemaker, they can say that the
success of the feast was also owing to the help of that trouble-
maker. In other words, when an event is caused, almost every-
thing, through its non-interference, can be causally responsible
for it. Thus, the question of excluding irrelevant items from the
domain of cause does notarise.
For Navya-nyäya, the fact of their getting up at 6 A.M. on
two Sunday mornings will be what is called anyathäslddha 'con-
ditioned otherwise', and hence, it cannot be causally relevant.
In other words, if *we investigated each case separately, we could
find in each case reasonable explanation of why that particular
adult in that case got up at 6 A.M. on two successive Sunday
mornings before the said craft was produced. And such an
explanation would reveal that each of these facts was conditioned

otherwise, anyathäsiddha. In the same way, Navya-nyâya would
declare that while being dry, that is, the lack of dampness, is a
'negative' causal condition for the matchstick to ignite, the
colour of the matchstick is anyathäsiddha, being immaterial to
its ifciting.1
§ 4.3 : SUBSTANCE (JAINISM AND VAISEÇIKA)
Jaina philosophy has not attracted the attention of modern

writers in the same way as Buddhism and Nyäya. This has
undoubtedly created a lacuna in our modern discussion of Indian
classical philosophy. Being aware of this lacuna, I have tried to
combine the remarks of the Jaina teachers with those of others
in several places of the previous chapters as well as in the first
section (4.1) of this chapter. In this and the following sections,
I shall concentrate upon two particular doctrines of Jaina
philosophy: Substance and 'sevenfold predication* method.2
In the history of Jaina philosophical literature, two teachers
(äcäryas) were pioneers in composing short treatises in the sütra
fashion. They were Kundakunda and Uraäsväti (orUmäsvämin).
The former composed for the first time several authoritative
(philosophical) works in Prakrit while the latter wrote the first
authoritative Sanskrit treatise in Jaina philosophy. It is not
known to us whether Kundakunda composed any work in San-
skrit, nor is it known whether Umäsväti ever wrote in Prakrit.
But both authors depended heavily on the Jaina scriptures and
supported the scriptural tradition with philosophic arguments.
These two authors belonged to a period that covers the first two
or three centuries of the Christian era. About 200 to 400 years
later, there appeared two other important Jaina teachers, Siddha-
sena Divâkara and Samantabhadra, who closely followed Kun-
1
In fact, the ananyathäsiddhatva criterion cannot, in final analysis, success-
fully exclude ali irrelevant factors from the relevant causal factor. GarigeSa
discusses the problem in Tattvacintämani, Part 2» vol. 3, pp. 154-5. The
important question, namely, why, of the two invariably conjoined factors,
one is taken to be the causal factor rather than the other, depends partly,
according to Gangesa, on our intuitive notion of simplicity {lâghava) and
relevance.
2
See also Matilal (1981).
dakunda and Umäsväti, and who were responsible for the

development and expansion of the Jaina philosophical method.
I shall concentrate here on the substance-attribute controversy
found in the writings of these four philosophers.
The Jaina conception of existence (sat) is intimately related to
the Jaina conception of substance. In fact, the concepts of sub-
stance and attribute are at the focal point of a number of philoso-
phical problems that have their origin in the early philosophical
tradition of India. As I have already said, the paradox of per-
manence and change, of being and becoming, was as much .a live
issue for the early Indian philosophers as it was for the Greeks,
i.e., the pre-Socratics. The doctrine of substance found favour
with those who were inclined toward permanence in the midst of
fleeting states or moments. However, those who gave primacy
to change and flux were always suspicious of the notion of 'being'
or substance. In the Indian context, the Buddhist belonged to
the second group, and their 'non-soul' doctrine was, in fact*
explained as a variety of the 'non-substance' doctrine. The
Vaiéesikas believed in the reality of substance and attribute.
Broadly speaking, the Jaina position was a compromise between
these two extremes.
Let us start with the early sütras. The Tattvärthasütra as-
serts.1 "What there is, has the nature of substance/' and "What
there is (the existent), is endowed with the triple character,
origin, decay and stability (persistence)."2 The commentary says
that whatever originates, perishes and continues to be is called
the existent; anything different is called non-existent.3 The next
sütra continues, the existent is constant, for it never gives up
its being (essence ?).4
In sütra 5.37, it is again said: "The substance is possessed of
qualities (guna) and modes (paryäya)." While we do not have
any definition of modes (paryäya), Sütra 5*40 defines quality
(guna) as: U-What reside in a substance, and are themselves
devoid of any quality, arecalled qualities." The Tattuärtha-bhäsya
adds:5
x
Sai dravya-laksanam, Tattvärthasütra 5, 29.
2ii
Utpâdavyayadhrauvyayuktarn sat" ibid., 5.30
8
See Umäsväti, under sütra 5.29.
4
"Tadhhâvâvyayant nityam" Tattvärthasütra 5.31.
5
See Umäsväti, under sütra 5.40
Though modes too reside in a substance and are themselves

devoid of any quality, they are subject to origin and destruc-
tion. Thus, they do not always reside in a substance. The
qualities, on the other hand, are permanent, and hence, they
always reside in a substance. This is how qualities are to be
distinguished from modes.
Püjyapäda, in his commentary Sarvärthasiddhi, is more specific
about the distinction of qualities and modes r1
A quality is (actually) the distinguishing character of one
substance from another. For example, the person (soul) is
different from matter (non-soul) by virture of (its possession
of) cognition, etc.; the matter is distinguished from soul by
virtue of qualities like colour. The generic attributes common
to souls are cognition, etc, and that of non-soul are colour,
etc. The modifications of these qualities, viewed in their
particular nature, are called modes (paryäyä), such as, cogni-
tion of a pot, anger, pride (in a soul); and intense or mild
odour, deep or light colour in the case of the non-soul.
In the above analysis of the Tattvärthasütra, we have at least
two compatible notions of substance: (1) substance as the core
of change or flux, and (2) substance as the substratum of
attributes. Kundakunda combines these two notions:2
They call it a substance, which is characterized by origin, per-
sistence, decay, without changing its 'own-nature', and which
is endowed with qualities and accompanied by modifications.
For the 'own-nature' of the substance is its existence {sad-
hhäva), which is always accompanied by qualities and varie-
gated modes, and at the same time, by origin, decay and
continuity.
The Vaisesika school emphasized rather the second aspect of
the substance, substance as the substratum of qualities and
action. Thus, Vaisesika-sütra 1.1.14defined substance as follows:
The definition of a substance is that it possesses qualities
(guya) and action/motion (kriya), and it is the substratum-
cause.3
1
See Püjyapäda under sütra 5.38, p. 199.
2
See Kundakunda, Pravacanasâra, Chap. II, verses 3 & 4.
Zti
Kriyàvadgunayat samavayikaracam id dravyalakfàriam" Vaiée§ikasutra
1.1.14.
The notion of "substratum-cause" {samaväyi-kärana) is explained

in this context as that which as substratum gives 'causal' support
to the changing attributes, qualities and actions. (§ 4.2).
To refer to Aristotle again, he seems to have been emphatic
about both the above notions of substance: 1. as a core of
change, and 2. as a substratum of attributes. In Categories, he
wrote r1
The most distinctive mark of substance appears to be that,
while remaining numerically one and the same, it is capable
of admitting contrary qualities. From among things other than
substance, we should find ourselves unable to bring forward
any which possessed this mark.
This comment underlines both the notions of substance mentioned
above. Aristotle, however, suggested also three other notions
of substance, all of which became very influential in later
Western philosophy: 3. substance as the concrete individual
thing, 4. substance as essence, as one having independent existence,
and 5. substance as the logical subject. From his remark that
examples of substance can be "the individual man or horse",2
one can infer the third notion of substance, substance as
the concrete individual. But admittedly, Aristotle's remark was
too vague to give us any definite conclusion.
The Vaisesika theory of substance included the concept of
the 'concrete' individual, but it was extended to include such
non-concrete 'things' as the bodyless soul, the sky, time and
space. Thus, the notion of substance as a concrete individual
thing is too narrow to accommodate the Vaiéesika view. Besides,
one may reasonably ask: what constitutes the concreteoess ?The
criteria of identification and individuation are clear enough with
regard to the standard things like man, table and horse, but
very unclear and problematic with regard to such non-standard
things as cloud, water and iron.
The idea of substance as the essence or the immutable core
seems to have been suggested by Aristotle in his Metaphysics.
A natural corollary to this notion is that a substance is indepen-
dently existent.3
Aristotle, Categories, (4a 10-14), p. 13.
2
Ibid., Categories, (2a 13), p. 9
3
Jbid., Metaphysics, (1028a 29-30), p. 783.
The Jainas too, identify the notion of "it is" (existence) with
that of substance, but they add also that "it is" or "it exists"
means only that it is endowed with the triple character of origin,
decay and stability.
The idea of 'own-nature' (svabhäva) in the early Indian philo-
sophy was, perhaps, the nearest analogue of the notion of'essence'
found in the Western tradition. Nâgârjuna and his followers,
the Mâdhyamika Buddhists, were the champion critics of the
'own-nature,' the immutable core, as well as of the notion of
existence which is necessarily dependent. "It is," as the Mâdhya-
mikas will argue, should mean that 'it always is,' and "it is not"
should mean that it always is not. The Mädhyamikas, in fact,
argued: all things (beings) are empty of their own-nature (own-
being) because they are dependently originating (pratttya-samut-
panna).1 It is thus clear that the 'emptiness' doctrine has been
propounded by Nägärjuna in order to avoid the paradoxical
situation under which existence or own-nature would have to be
admitted as dependent and conditioned. It is probably because
of Nägärjuna's trenchant criticism that the Vaiéesikas never
explicitly stated the 'own-nature' theory of substance,
although such a theory might very well have been implicit
in their doctrine of substance. The Vaiéesikas in fact held a
modified position. There are, according to the Vaisesikas, two
types of substance: permanent, and impermanent. The atomic
constituents of the material substance, and the non-corporeal
substances, such as, the sky, time and space, are regarded as
permanent substances and independently existing, but the things
like the clay-pot and the wooden chair are impermanent and
existent only being causally dependent.
The idea of substance as the logical subject was another im-
portant notion which became the concern of Aristotle and his
followers in Western philosophy. But this notion is no longer
appealed to by the logicians today, although some modern
philosophers have revived the issue in a different way. Strawson
does not mention the word "substance" but shows that material
bodies and persons are central notions of our conceptual scheme,
and he calls them the basic or fundamental particulars. He also
shows the connection of the notion of a 'particular' with that
1
Cf. tkpratïtya-samutpaniiatvâii-nilisvabhavam, uilisvabhävatvät sQnyam
iti upapaanain." Nâgârjuna, Vigrahavyavaria ni, under verse 22,
1
of an object of reference or logical subject. This can be seen
as the explication of the old problem of substance as the logical
subject. In the Indian context, the Vaisesikas never explicitly
considered the connection between the logical subject and the
notion of substance, although it was probably implied in their
early doctrine of substance. The later Indian logicians (includ-
ing the Buddhists and the Naiyäyikas) coined a new term, dharmin
("property-possessor"), which was their nearest analogue for
"the logical subject." The Indian logicians developed the notion
of dharmin independently of that of substance, and, thus,
some of the problems connected with the notion of substance
as the logical subject were somehow not raised in their discussion.
Amrtacandra Süri, commentator of Kundakunda, explains
that when a pot is produced from a lump of clay, both the
origin of the pot and the destruction of the lump amounts to
maintaining the persistence of the clay-substance. In order to
prove his contention, Amrtacandra uses the following reductio
(prasanga): (For Kundakunda's remark see § 4.1.)
If we do not accept it as true, origin, decay and continuity—all
three, none will then be really different from one another. In
that case, when the mere origin of the pot is sought after, then
either it will not originate, for, there will not be any (real) cause
for its origin, or there will be origination of the non-existent
(an untenable paradox). If the pot does not originate, no
bhävas (things) will originate. If there is origination of the
non-existent (asat), then sky-flower, etc. will come into being.
Similarly, if mere destruction of the lump of clay is attempted
(to the exclusion of the production of the pot), then either
there will not be any destruction of the lump for want of any
(real) cause for such destruction, or there will be destruction
of the existent or being (an untenable position).2
We may again refer to the triple character of the substance.
The Jaina authors have dealt with this point very frequently. I
have mentioned Samantabhadra's comment in the previous
section (§ 4.1). Kumarila, the Mïmâmsaka, elaborated the point
in this way:
l
Strawson P.F., Individuals, p. 136ff.
2
Ami tacandra Suri, in Kundakunda, Ch. II,v.8, p. 125.
If the (gold) plate is destroyed and (instead) a (gold) necklace

is made, then the person who wanted the plate will grieve,
and he who wishes the latter will be happy, but he who wishes
for the gold stuff (only) will neither grieve nor be happy.
Thus, the triple nature of an entity is proved."1
Turning to the second conception of substance in the Tattvär-
thasütras (according to which substance is the substratum of
qualities and modes), we can say that it was probably derived
from the Vaise§ika school. In fact, Tattvärthasütra 5.41 defines
quality:
Qualities are located in substances, and are themselves devoid
of qualities.2
This seems to be an echo of the Vaisesika definition of guna
or quality. It is also significant that one of the most important
Jaina concepts, mode or modification, is not even defined in the
Tattvärthasütras. The Jaina ontologicai principle of anekänlatä
'non-onesidedness\ however, is not compatible with the rigid
Vaisesika notions of substance and quality. Thus, Siddhasena
has added that it would be as good as a heresy in Jainism, if one
intends to make the notion of substance absolutely different from
that of quality.3 Moreover, Siddhasena has argued, the supposed
distinction between qualities and modes (tacitly accepted by
both Umâsvâti and Kundakunda) should also be discarded
altogether in order to remain true to the Jaina spirit.
The main contribution of Jaina philosophy was, however, its
doctrine of the 'many-sidedness' (anekäntatä) of reality along
with a unique method that was used (by the Jaina philosophers)
to formulate this doctrine. This unique method is called sap-
tabhahgi or the seven ways of formulating each metaphysical
thesis with a view to exposing its non-determinate or non-
absolute or provisional or relative character. I shall devote
the next section to explain this method in order to clarify
certain points that are usually misunderstood by the moderners.
1
Kumàrila, Mîmàmsâêlokavârttika* p. 613.
2
Tattvärthasütra 5.41 : "Dravyäsrayä nirgunägunäh"
3
Siddhasena, Sanmati, Ch. Ill, verses 8-9.
OntologJcal Problems 301
§ 4.4 : SEVEN WAYS OF NON-ABSOLUTISM (SAPTABHANGÏ)
The unique contribution of Jainism to the philosophic tradition

of India consists in its doctrine of Anekäntatä or Anekänta-väda
(the theory of non-onesidedness). The metaphysical counterpart
of this doctrine can be described as the theory of 'many-sided
nature' of reality. The Anekänta philosophy, in fact, amounts
to a philosophy of synthesis, an integrated approach to the onto-
logical problems. And thus, it has both advantages and disad-
vantages. Conflicting philosophic theories are treated by Jainism
as viewing the same reality from different points of view. A thing
has infinitefold character, and hence, the philosophic truth can
be grasped only by integrating and synthesizing all the view-
points or standpoints.
In order to prove the Anekänta doctrine, the Jaina philoso-
phers developed a philosophic methodology that was unique to
Jainism. It consisted of the dual doctrine of the Jainas : Naya-
väda 'the doctrine of standpoint' and syädväda (saptabhangï)
the doctrine of sevenfold formulation of a metaphysical question.
These two doctrines or methods of Jainism have been the two main
pillars on which the Jaina philosophy of Anekäntatä rests.1 In
fact, these two methods are complementary to each other. Thus,
using the age-old analogy, one can say that the 'standpoint'
doctrine is blind without the sevenfold predication, and the latter
is immobile (lame) without the former.2 Their joint operation
makes the Anekänta philosophy feasible.
We shall concern ourselves here with only one of these two
methods, the doctrine of sevenfold predication. This is a
controversial doctrine that has been much criticized by the op-
ponents, and I shall argue here that the Jaina position has often
been misunderstood in this respect.
The Historical Background

The sevenfold predication was historically a later development
in Jainism, for, we do not find it clearly mentioned in the early
canons. A. N. Upadhye, however, has located references to the
x
The "pillar" analogy is due to Y. J. Padmarajiah; see his Jaina Theories of
Reality and Knowledge, Bombay, 1963.
2
The parable of the blind and the lame guiding each other is found in the
Sämkhya tradition.
three primary predicates (instead of seven) in the Bhagavati-

sütra.1 Umäsväti (c. 1st century A.D.) did not make any explicit
reference to the seven alternative predicates.2 But Kundakunda
(who most probably belonged to the second or third century A.D.)
mentioned the full-fledged seven alternative predicates in his
Pancästikäya*
As forerunners of the sevenfold formula of the Jainas, we
have two similar formulas explicitly mentioned in the earlier litera-
ture. The first was the fivefold formula of Sanjaya found in the
Pâli canons. In the Sämannaphala-sutta of Dïghanikâya I,4
Sanjaya is reported to have developed a fivefold formula to ans-
wer some metaphysical and moral questions, such as, "whether
there is another world or not ?" or "whether something is right
or wrong ?" For example,
(1) Question : "Is it this (or so) ?" Answer : "No".
(2) Question : "Is it that (or thus) ?" Answer : "No".
(3) Question : "Is it otherwise (different from both above) ?"
Answer : "No".
(4) Question : "Is it not (at all there) ?" Answer : "No".
(5) Question : "Is it not that it is not (at all there) ?"
Answer : "No". 5
The first three alternatives in the above formula, "this-that-
otherwise", can be easily reduced to two alternatives, if we use
the contradictories, suchas, "this-or not this, "or "this-or-other-
wise". Thus, the fourfold alternatives of the Buddhists (later of
the Mâdhyamikas) can be seen as an improved and more precise
formulation of the earlier, rather imprecise, fivefold formula.
The Mâdhyamika denial of the fourfold alternative was :
. Upadhye, Introduction to Pravacanasâra of Kundakunda, Bombay,
1955, p. 83.
2
Some scholars point out that Umäsväti made an "implicit" reference to the
saptabhangi in the sOtra : arpitänarpitasiddhe (Tattvärthasütra 5/32).
s
See Pancästikäya, ed. A. Chakravartinayaras, Sacred Book of the Jainas,
Arrah, 1920.
4
See Digha Nikäya, ed. T. W. Rhys Davids and E. J. Carpenter, 3 vols.,
London, 1890-1911.
^ h e pâli text runs thus : (1) evem pi me no; (2) tathâ pi me no; (3)anfiathâ
pi me no; (4) no ti pi me no; (5) no no ti pi me no. I have translated it
in the form of questions and answers.
Ontologïcal Problems 303
(1) Question : "Does the effect come out of itself?"

Answer : "No."
(2) Question : "Does it come out of the others ?"
Answer : "No".
(3) Question : "Does it come out of both itself and others ?"
Answer : "No".
(4) Question : "Does it come out of neither (self or other)?"
Answer : "No". 1
It should be noted that the Buddhist answers to all these alter-
native questions were, like the answers of Sanjaya, in the negative.
Scholars like Herman Jacobi have surmised that Mahâvïra esta-
blished the sevenfold ijw/predication in opposition to the "Agno-
sticism" of Sanjaya.2 There seems to be some truth in this claim.
For, Mahâvîra adopted the method of answering all metaphysi-
cal/philosophical questions with a qualified yes? But, as I have
already noted, there is no textual evidence (either in the Pali or
in the Prakrit canons) to show that Mahâvîra had actually used
the sevenfold syät predication. K. N, Jayatilleke has apparently
been very critical of Jacobi's view in this matter. He has been
eager to show that the two (the Jaina formula and the Sanjaya
formula) "seem to have a common origin".4 In his eagerness to
show this "common origin" Jayatilleke has mistranslated syät as
"may be". I find the argument of Jayatilleke unconvincing as a
rebuttal of Jacobi's thesis, viz., Mahâvïra's philosophy was for-
mulated in opposition to the philosophy of Sanjaya. It is un-
deniable that while the former preferred conditional affirmation
of the questions about after-life, etc., the latter preferred a straight
denial.
Although Sanjaya resembled the Buddhist in giving negative
answers to the metaphysical questions, we should note that
Sanjaya's philosophic conclusion was different from that of Nâg-
àrjuna. Out of respect for truth and out of fear of, and distaste
Mädhyamika-kärikä of Nägärjuna, Ch. I, verse 1, e& P. L. Vaidya,

Darbhanga, 1960.
2
See p. xxvii, Introduction, Jaina Sutras, Tr. H. Jacobi, Sacred Book of the
East (Dover Publication, New York, 1968).
3
See my Anekânta-vâda, L.D. Institute of Indology series, Ahmedabad.
4
See p. 139, K.N. Jayatilleke : Early Buddhist Theory of Knowledge, Allen
& Unwin, London, 1963.
for, falsehood (cf. musâvâda-bhayâ) Sänjaya adopted a non-com-

mittal attitude towards questions about after-life, etc. His posi-
tion was that definite knowledge about such matters as after-life
was impossible to obtain, and he had the boldness to confess it.
Thus, I think the Pâli commentator was a bit unfair when he
called him an "eel wriggler*'.
Nägärjuna's position was slightly different from that of total
non-commitment. From the denial of the fourfold alternative,
Nägärjuna was led to a definite philosophic conclusion that these
questions about after-life, cause, etc., were only pseudo-questions
or that the concepts (regarding which such questions Were asked)
were only pseudo-concepts. They are, therefore, *'empty" of its
'own-nature', of its essence. In this way, Nägärjuna was led. to
his "emptiness" doctrine, while Sanjaya was at best a sarhsaya-
vädin, an agnostic.
In fact, it can be asserted with some confidence that the "three
termed" doctrine (cf. trairäsikä) of the Ajïvakas foreshadowed
the sevenfold predication of the Jainas.1 ThisÄjivaka sect, esta-
blished by Gosàla, declared that every thing is of triple character,
viz., existent, non-existent and both; living, non-living and both
living and non-living. This doctrine of triple character of every
entity is more akin in spirit with, and logically closer to, the later
Jaina doctrine of sevenfold formula as well as the anekänta 'non-
onesided' view of reality. For, basically, the Jaina considers only
three possibilities: positive, negative and both positive and nega-
tive. The seven possibilities,as we shall see presently, were deve-
loped out of the three basic possibilities along with a more
subtle distinction introduced in the third possibility, viz., both
positive and negative.
In the fourfold alternative of the Mädhyamika, the fourth
possibility is that of a "neither...nor...". The question was for-
mulated as "Is it neither A nor not-A ?" The answer was given
in the negative by Nägärjuna (as well as Sanjaya). In the Jaina
scheme, however, this question is not even formulated. Thus, we
may say that "neither...nor..." is not even accepted as a possi-
bility in Jainism. The reason may be that the "neither...nor..."
alternative is one of strong denial or negativity (cf. prasajya-
X
A. L.Basham made this suggestion. See his History and Doctrines of the
Ajivikas, pp. 274-5, London, 1951.
1
pnttisedha). But since Mâhâvîra, unlike the Buddha, did not
follow the line of direct denial, but rather the line of conditional
acceptance, the followers of M ahavïra were certainly true to the
spirit of their master in leaving the "neither...nor..." alternative
out of their consideration. Besides, this point underlines another
logical distinction between the Jaina position on the one hand
and the Buddhist or the Sanjaya position on the other. The
former apparently violated the Law of Contradiction (since it
accepted contradictory possibilities) while the latter, in conced-
ing a "neither...nor..." possibility, seemed to run against the
Law of Excluded Middle.
The meaning of Syät

The uniqueness of the Jaina formula lies in its use of the "syät"
particle in the predication. That is why, the sevenfold predica-
tion of the Jainas is sometimes called Syäd-väda. In ordinary
Sanskrit, "syät" is sometimes used to mean 'perhaps' or 'may
be.' In fact, it is one of the three words used to answer a direct
question : Ts AB ?" viz., "Yes" or "No" or "Syät (may be)".
But the Jainas used this particle in a very special sense. It is a
particle that indicates the anekänta nature of a proposition.2
With a syät modifier, a metaphysical proposition is given only a
qualified acceptance.
Etymologically, "syàt" is derived from the root as+potential/
optative third form, singular. Bhattojï Dîksita explained the opta-
tive suffix, lin, in one context, as expressing probability (sambhä-
vanä). Thus, under Päninisütra 1.4.96, the example "sarpiso 'pi
syät" is explained as : "there is even a chance of (a drop of)
butter." But the Jaina syät is even different from this use of syät
in the sense of probability. The Anekänta doctrine, to be sure,
is neither a doctrine of doubt or even uncertainty, nor a doctrine
of probability. Thus, "syät" means, in the Jaina use, a condi-
tional YES. It is like saying, "in a certain sense, yes". It
amounts to a conditional approval. The particle syät, in fact,
acts as an operator on the sentence in which it is used. It turns
a categorical ("A is ß") into a conditional : "If p then A is ET\
1
See my Epistemology, Logic and Grammar in Indian Philosophical Analysis,
pp. 162-5, Mouton, Hague, 1971.
2Cf. siddhih syäd-vädät (1.1.2) in Hemacandra's Sabdanusäsana, ed. Vijaya-
îavanya Suri, Rajnagar, 1954.
There is also a concessive use of "syât" frequently found in

philosophical Sanskrit, viz., "syâdetat".This expression means:
"let is be so, (but)..." The use of syät in this context implies
that the author (or the speaker) only provisionally concedes the
position of the opponent, for he tries at the moment to raise a
different (and perhaps, a more serious) objection to reject the
opponent finally! But the Jaina use of the particle syät in the
sevenfold formula is a much more refined sort of concession to
the opponent. It concedes the opponent's thesis in order to blunt
the sharpness of his attack and disagreement, and at the same
time, it is calculated to persuade the opponent to see another
point of view or carefully consider the other side of the case.
Thus, the Jaina use of "syat" has both; it has a disarming effect
and it contains (implicitly) a persuasive force.
Samantabhadra (c. 500 A.D.) has commented upon the mean-
ing of "syät" as follows.1
When the particle syät is used by you (Mahâvïra) as well as
by a érutakevalin (e.g., a saint) in a sentence, it indicates, in
connection with other meanings, non-onesidedness; it qualifies
(since it is a particle = nipäta) the meaning (of the sentence
concerned).
In the next verse (V. 104), Samantabhadra notes that syät is
ordinarily equivalent to such expressions as ''kadäcit" and "kath-
ahciV\ But even these terms, "kadâcit" or "kathancit", do not
have in this context such vague meanings as'somehow'or Some-
times'. They mean *in some respect' or 'from a certain point of
view' or 'under a certain condition'. Thus, the particle "syät"
in a sentence modifies the acceptance or rejection of the proposi-
tion or predication expressed by the sentence.2
Explanation of the seven answers

" F r o m a certain point of view, you (Mahâvïra) accept, "K is,"
and from another point of view, you accept, "It is not". Simi-
larly, both "it is" and "it is not", as well as "it is inexpressible".
1
Verse 103 of Samantabhadra's Äptamimärrisä, ed. Yugalkisore Mukhtar,
Delhi, 1967.
2Compare Bhartrhari's comment on the significance of nipäta, Väkyapadiya,
kända 2., verse 204. Bhartrhari says that a nipäta qualifies or modifies
relation (sambandhasya tu bhedakah).
Onfologicaî Problems 307
All these (four) are approved (by you) with reference to the
doctrine of standpoint (naya) only, not absolutely". (Äptamimä-
msä, V. Î4).
In this way, Samantabhadra has formulated the first four of
the seven alternative predicates. We can symbolize these four
basic propositions as ' + ' , ' — ' , ' ± ' and ' 0 \ The fourth predica-
tion, "it is inexpressible", is actually interpreted as the joint
(combined) and simultaneous (cf. sahärpanä) application of both
the positive and the negative. It is distinct from the third pro-
position, because in the latter there is joint but gradual (one after
another, non-simultaneous =*kramärpana) application of the posi-
tive and the negative, Since it is believed that the language lacks
any expression which can adequately express this simultaneous
and combined application of both the positive and the negative
characters, the Jainas say that they are obliged to name this
predicate "inexpressible", and we have symbolized it by c 0'
accordingly. The idea is that since we cannot say "yes" and " n o "
simultaneously in answer to a question, we can name thisvariety
of our answer as 'inexpressible'.
Although the predication "inexpressible" (or <Q') has been
reached in the above manner (as is evident from the Jaina texts),
the Jainas, however, still regard it as a unitary predicate, a unit,
like the positive or the negative (i.e., "it is" or "it is not").
Probably, it was thought that since the two components, positive
and negative, are here perfectly balanced and totally neutralized,
being applied simultaneously (in the same breath), the predica-
tion had lost the compound character and melted into one uni-
tary whole. In other words, a predicate that was compound in
character in its inception (or when it was first thought out)
turned into a non-compound, primary predicate, because of its
internal structure, so to say. I have thus used the neutral symbol
'0' to indicate it.
The Jainas have, in this way, three primary and non-compound
predicates, positive, negative and the neutral ( + , —, Ö). Nowit is
easy to see how the Jainas reached the seven possible varieties.
Let the three predication-units be represented by x, y and z. A
simple mathematical computation will generate only seven varie-
ties, if we use these three units in three ways, one at a time, two
at a time and three at a time :
xy y; z, xy, yz, zx9 xyz

+, -, 0, ±, -0, +0, ±0
Note that the combination in this formula is comparable to
the arithmetical conjunction or the truth functional 'and' such
that the internal order in a combination is immaterial, there
being no need to distinguish between 'xy' and *yx\ In mathema-
tical terminology, this is called the commutative property of con-
junction.
The Jainas, however, enumerate the above combinations in a
slightly different order (adding "syât" to each):
1. "From a certain point of.view, or in a certain sense, the
pot exists." +
2. "From a certain point of view, the pot does not exist." -
3. "From a certain point of view, the pot exists and does not
exist." ±
4. "From a certain point of view, the pot is inexpressible". 0
5. "From a certain point of view, the pot both exists and is
inexpressible." +0
6. "From a certain point of view, the pot both does not
exist and is inexpressible." —0
7. "From a certain point of view, the pot exists, does not
exist, and is also inexpressible." ±0
One maynote that the third predicate in the abovelist is not the
third *"neutral" predicate, but a compound one combining the
first and the second. In the fourth place in the above, we come
across the third primary predicate, "inexpressible".
While explaining the seven predicates, Vidyänanda has noted
as follows r1
"Someone says, let there be only four types of formulations.
This is not tenable. For, there are three (further) possibilities
by combining the positive, the negative and both of them with
the 'inexpressible*. Thus, we have sevenfold formulation : (1)
affirmation, (2) denial, (3) both affirmation and denial, (4) the
joint and simultaneous affirmation and denial, (6) denial, and
the joint and simultaneous affirmation and denial, (7) affirma-
tion, denial, and the joint and simultaneous affirmation and
denial.
x
See p. 125, Vidyänanda's Atfasähasri, ed. Vamsidhara, Nirnayasagar,
Bombay, 1915.
It is obvious, however, that the fourth evaluation here ('the

joint and simultaneous affirmation and denial', which is Vidyä-
nanda's explanation of the term "inexpressible") must be taken
to be a unitary whole, a primary predicate. For, otherwise it
would be difficult to explain the sevenfold combination with
mathematical computation. And Vidyänanda himself has empha-
sized that there are seven and only seven alternatives in the
Jaina system.
A common objection against the Jaina sevenfold formula has
been that instead of accepting only seven alternative predicates
in this manner, one might go up to a hundred or a thousand (i.e.,
to an unlimited number). Thus, a critic like Kumärila had said,
"Even one hundred alternatives can be generated through gen-
erous use of the method used (by the Jainas) to generate only
seven alternatives". 1
But certainly this is not a fair criticism of the Jaina method.
It is based on a misunderstanding. Thus, Vidyänanda goes on
to point out that there may be an infinite number of properties
or predicates that are ascribable to a subject. The Jaina Anekänta
doctrine of reality only welcomes such attribution. For, according
to the Anekänta doctrine, a thing or entity is supposed to possess
infinite or innumerable aspects or characters. But the sevenfold
formula (i.e., the seven alternative formulations using the three
principal modes, positive, negative and the neutral) will be
applicable to each attribution of a property, i.e., to each indivi-
dual predication. In other words, as long as we accept only
three basic qualities of OI\Q individual predicate (positive,
negative, and the neutralized), we shall get only seven possible
combinations.2
Traditional Objections
Critics of the Jaina sevenfold formula have mentioned many
faults or anomalies that are supposed to arise if the doctrine
is accepted as a philosophical method. The Jaina writers
beginning from Akalanka and Vidyänanda have analysed these
objections and tried to answer them in detail. Let us make a
brief survey of these objections and answers.
1
Cf saptabhahgï-prasâdena éatabhangy apijâyate; Kumârila's Slokavarttika,
ed. R. S. Tailanga, Benares, 1898.
a
See Vidyänaiula, op. cit., p, 126.
310 Logic, Language and Reaîijty
âamkara in his Brahmasütra-bhäsya1 mentions, among other

things, two specific problems involved in the Jaina position:
virodha 'contradiction' and samsaya 'doubt' or'dubiety'. §änta-
raksita adds another, sawkara 'intermixture'.2 Akalanka notes
seven faults for the anekänta doctrine in his Pramänasarngraha:
dubiety, contradiction, lack of conformity of bases (yaiyadhikar-
anya), 'joint fault (uhhayadosa)\ infinite regress, intermixture,
and absense (abhäva), Vidyänanda gives a list of eight faults; he
omits "joint fault" from the list of Akalanka, but adds two
more: 'cross-breeding' (yyatikard) and the lack of comprehension
(apratipatti)3. Prabhâcandra mentions also a list of eight, but he
replaces 'lack of comprehension' by the earlier 'joint-fault'.4
Vädideva drops "absence" (abhäva) from the list of Prabhâ-
candra and makes it a list of seven faults.5 Most of these faults
or defects are only minor variations of the three major problems
faced by the Jaina doctrine of the sevenfold predication: inter-
mixture, dubiety and contradiction.
Vyomasiva has mentioned another unique problem of the
anekänta doctrine.6 He says that a free (liberated --mukta)
person will not really be liberated under the anekänta doctrine.
For, he will be considered, from one point of view, both libera-
ted and not liberated, and, from another point of view, simply
not liberated. Besides, if the statement "the thing has anekänta
nature" involves an unconditional predication, then it falsifies
the anekänta doctrine, for, according to the anekänta principle,
no philosophical predication should be unconditional or unqua-
lified. But if the above predication is conditionalized with the
syät operator following the Jaina anekänta principle (viz., "in a
certain sense, the thing has anekänta nature", and so on), then
we shall be led into an infinite regress or circularity.
The above problem of anekänta is reminiscent of a similar
*See Samkara's Bhäsya under Brahma-sütra 2.2.33, pp. 559-62; ed. M.A.
Sastri, Nirnayasagar, Bombay, 1938.
2
See verse 1722 of Säntaraksita's Tattvasamgraha, ed. E. Krishnamacharyya,
Oriental Institute, Baroda, 1926.
3
See p. 227 of Vidyânanda's Astasâhasrî, op. cit.
4
See p. 156 of Prabhäcandra's Prameyakamalamärtanda, ed. Mohendra
Kumar Shastri, Bombay, 1941.
5
See p. 738 of Vädideva's Syädväda-ratnäkara, ed. Motilal Ladhaji, 5
parts, Poona.
6
See p. 20 of Vyomäsiva's VyomavatU ed. Gopinath Kaviraj, Benares, 1930.
problem or paradox posed against the "Emptiness" doctrine of

the Mädhyamika. Nägärjuna discussed this problem in the be-
ginning of his Vigraha-vyävartanL If the statement "everything
is empty" is itself empty, then it falsifies the "emptiness" doc-
trine, and if that statement is not empty, then there is at least one
thing that is not empty, which also falsifies the doctrine. Nägär-
juna explained this paradox and answered the objection against
his doctrine quite satisfactorily from the Mädhyamika point of
of view.1 As far as I can see, a similar defence of the Jaina
doctrine of anekänta philosophy or syäd-väda is not impossible
to construct in order to answer the criticism of Vyomaéiva.
In Defence of the Jaina Position
Of all the charges against the anekänta philosophy or the sevenfold
syät, the charge of contradiction or self-contradiction is certainly
the most serious one. For a philosopher to contradict himself
is like writing or stating something and then cancelling it alto-
gether. Do the Jainas really suffer from this offence? Could the
Jaina view be defended against the charge of self-contradiction
or inconsistency?
Let us focus our attention on the sevenfold predication. It is,
however, clear from the interpretation of the syät particle given
above, that the first predication does not really contradict the
second. The Jainas avoid contradiction by adding the syät parti-
cle. The syät operator turns the categorical proposition into a
conditional, and thus the logical forms of the first two are:
(i) If p then a is F.
(ii) If q then a is non-F.
Or, more fully;
(iii) For all x,if x is considered from standpoint 1, x is
eternal :
(iv) For all x9 if x is considered from standpoint 2, x is not
eternal :
It is clear that neither (i) and (ii), nor (iii) and (iv) are, in any
sense, contradictories. Thus, I think that when the Jainas say
that from the standpoint of persisting substance, .the person is
eternal, but from the standpoint of modal changes (cf. paryäya),
l
See also my Lpistemology* op. cit, p. 146-67.
the person is not eternal, they do not make any self-contradictory

assertion.
How about the third and the fourth predications? The third,
to be sure, is the joint (but non-simultaneous) assertion of the
first and the second. But if the first and the second are not con-
tradictories, then the third (which is only the truth-functional
conjunction of the first and the second) will not be self-contra-
dictory. In other words, the third predication can easily be seen
to be free from contradiction in this way. The fourth predication,
however, presents a problem. For, it seems to apply two incom-
patible predicates; eternal and non-eternal, to the subject
in the same breath or simultaneously. Although the statement is
conditionalized with the syät operator, it only means that under
certain conditions a thing will have two contradictory characters.
Thus, the speaker here may be taken to have contradicted him-
self and said nothing. (This may partially justify the use of
"inexpressible" to denote this predication, for two contradictory
predicates are supposed to cancel or erase each other).
In defence of Jaina doctrine, we can make two points here.
First, by simple application of contradictory predicates to a
thing in the same breath ( or simultaneously) the speaker does
not land himself into a self-contradiction. For, there is always
the chance of there being some hidden meaning which the*
speaker can explain in order to resolve the apparent self-con-
tradiction. For example, we can say of a man, "He is both over
six foot tall and under six foot tall", and then explain that he
has a disease which makes him stoop, but that if he were cured
and were able to stand upright, he would top the six-foot
mark.1 Mahâvîra himself followed a similar line of explanation
in order to elaborate upon the apparently contradictory asser-
tions like 'the person is both eternal and non-eternal'.2 In this
way I think the Jainas can somehow answer the chargé of self-
contradiction against the fourth predication.
This leads to our second point. The basic assumption in
Jainism seems to be the anekänta (non-onesided) nature of
reality. A thing is supposed to have infinitefold character or
iThis example is taken from P. F. Strawson : Introduction to Logical
Theory, pp. 16-19. University Paperbacks, London.
2
Cf. Bhagavatî-sutra, 2.1.90, ed. S.C.P. Jhaveri and S. Kesarimalaji, 2 vote.,
Surat and Jamnagar, 1937, 1940.
innumerable aspects or properties. If this premise is conceded

then, of course, it becomes possible to apply all kinds of pre-
dicates ( including contradictories) to the thing depending upon
one's point of view or standpoint.
One obvious difficulty in the above concession is this : If it
becomes possible to apply incompatible predicates to the same
thing, then it defeats the purpose of predication. For one im-
portant function of describing a thing or a person with pre-
dicates is to distinguish it from other things, to exclude it from
other groups.
The Jainas,/however, might reply that the fourth predication
"the thing is, in a sense, inexpressible" is not intended to dis-
tinguish the thing from other things, but to include it in every-
thing else. For, remember the Jainas would be prepared to apply
this predicate "Inexpressible" (if we can call it a predicate) to
everything without exception. This statement is actually in the
same level with statements of other schools like "everything is
empty" or "everything is existent (sat).'9 The idea of the Jainas
is probably that in such predication the purpose of description
might fail, but the purpose of stating a truth will not fail.
The anekänta-väda is thus a philosophy of synthesis and re-
conciliation, since it tries to establish a rapprochment between
seemingly disagreeing philosophical schools. Jaina philosophers
contend that no philosophic proposition can be true if it is
only unconditionally asserted. They say that the lesson to be
drawn from age-old disputes and controversies regarding the
philosophic or metaphysical propositions is the following. Each
school asserts its thesis and claims it to be the truth, and thus
it does not really understand the point that is being made by
the opposite side. Thus, rival schools only encourage intoler-
ance in philosophy. This, according to the Jainas, is the evil of
ekänta 'one-sided' philosophies. Even the conflicting proposi-
tions of rival schools maybe in order, provided they are asserted
with proper qualifications or conditionaiization. This is what
exactly the Anekänta doctrine teaches. Add a "syät" particle
to your philosophic proposition and you have captured the
truth.
Non-violence, i.e., abstention from killing or taking the life
of others, was the dominant trend in the whole of sramana
movement in India, particularly iu -Buddhism and Jainism, I
think the Jainas carried the principle of non-violence to the

intellectual level, and thus propounded their anekähta doctrine.
Thus, the hallmark of the anekänta doctrine was toleration.
The principle embodied in the respect for the life of others was
transformed by the Jaina philosophers, at the intellectual level,
into respect for the view of others. This was, I think, a unique
attempt to harmonize the persistent discord in the field of
philosophy.
§ 4.5 : 'EMPTINESS' AND 'MANY-SIDEDNESS'
One may facetiously put : Both the Mädhyamika and the Jaina
fi
cook the metaphysical goose9 quite well enough, so that all
metaphysical doctrines can be eventually disposed of. While the
former says that all beings or things ((bhäva) are empty of its
own-being or own-nature (essence ?), the latter says that they
can have only a 'non-absolute' nature. Both deny that there
could be any absolute conception of Reality (any God's eye view
of reality that we humans can possess, except of course the
omniscient Mahavira), but while the former is emphatic
in. his denial and rejects ail such attempts (to construct
a metaphysical doctrine) to be futile and fundamentally flawed
(cf. they grasp the snake at the wrong end), the latter is more
receptive, for, he accepts a plurality of such 'conflicting* concep-
tions of the absolute reality. In fact, the Mädhyamika claims
that the sooner we can get rid of all such wrong conceptions
the better. The Jaina claims, on the other hand, that by pro-
visionally accepting all such 'partial' (but not absolutely wrong)
and 'non-absolute' conceptions of reality we get closer to the
truth, for, plurality is the nature of the reality. The Mädhyamika
view is exclusive (it is not even a view, and hence the designation
'emptiness'), while the Jaina view is inclusive (it is, in fact, ail-
inclusive), for it aims at totality. If a view includes everything,
even the conflicting views, then, in a way it also ceases to be a
view.
It will be interesting to compare the Jaina dialectic with the
Màdhyaniika dialectic. The Jainas argue that different philsophers,
when they construct durèrent philosophical systems, represent
only different 'standpoints.' The Jainas further point out that as
long as we emphasize one aspect or standpoint (say the stand-
point of 'substance'), while being fully aware that this is only
one out of many, equally viable, standpoints, we employ a naya

'a right philosophical method/ But when we emphasize only one
standpoint by excluding all others, we employ a dur naya 'an
incorrect philosophical method.' The business of the anekänta
philosophy is to expose a durnaya, and isolate and identify the
nayas.
To illustrate Nägärjuna's philosophic argumentation, let me
quote two verses from the Mädhyamika-Kärikä r1
The 'own-nature' (of a thing) cannot be generated by causal
conditions Qietus and pratyayas). For, if the 'own-nature' is
generated by causal conditions, it would be (artificially)
created.
Now, how could 'own-nature' be (artificially) created? For,
'own-nature' is what is non-artificial (un-created) and indepen-
dent of others.
Nägärjuna, thus, carries this point to its logical extreme:
If the nature or essence (of a thing) does not exist, what is it
then that will change ? And if the nature does exist, what, again,
is it that will change ?2
Consistent with the attitude of the Buddha, who refused to be
dragged into the quicksand of philosophic disputations, the
Mädhyamika rejects most philosophic positions by exposing their
inherent contradictions and anomalies and points out that tattva
(truth) is not to be arrived at through such philosophic dis-
putations, for, it is only revealed to the prajnä or insight.
Similarly, consistent with the attitude of iVîahâvîra, who tried to
resolve the philosophic disputations by analysing various shades
of meaning and implications of the concepts involved (see above),
the Jâinas tried to reconcile the different philosophical schools
and showed that the difficulties involved in their ekänta positions
resulted from their hidden assumptions and tacitly accepted
standpoints.
A comment from Siddhasena is particularly illuminating in
this connection. He observes :3
All the standpoints (nayas) are right in their own respective
^Mâdhyamika Kârikâ XV, 1-2.
2Ibid.,XV, 9.
zSantnatiturka, I. 28.
spheres—but if they are taken to be refutations, each of the

other, then they are wrong. But a man who knows the 'non-
one-sided' nature of reality never says that a particular view is
absolutely wrong.
It should, however, be noted that Nâgârjuna's position of non-
commitment was not always expressed through negation or rejec-
tion. On rare occasions, he seems to betray what may be called
the Jaina spirit of concession and neutrality. For example, consi-
der : Mädhyamika Kärikä, chap. 18, verse 8.
"Everything is true; not everything is true; both, everything is
true, and not everything is true; or, neither everything is true
nor is everything not true. This is the teaching of the Buddha."
It has been said that both the Mädhyamika and the Jaina
share the same sort of scepticism about all metaphysical positions.
However, neither of them can be called SCEPTIC in the accept-
able sense of the term. How to distinguish a sceptic from Jaina
or a Mädhyamika ? I shall refer to Candrakîrti's account of the
distinction between a sceptic or a nihilist and a Mädhyamika.
Suppose, a man has committed theft. Another man, who
does not, in fact, know whether the first man has committed
theft or not, comes along and declares that this is the thief, simply
because he happens to have a dislike for that man. Then, a third
man, who has actually seen the first man committing theft, comes
along to declare that this is the thief. Now, both the second and
the third man make the same assertion about what happened in
actuality, but yet the difference between them is very significant
and important. It is the distinction between a liar and a truthful
person, between falsehood and truth—a discrimination about all
that matters in Ethics. If we have understood this distinction bet-
ween the third and the second person, we have then understood
something important about a Mädhyamika who declares every-
thing to be empty. This is the way Candrakirti wanted to explain
the distinction between the 'emptiness' doctrine and scepticism,
the essential difference between a Mädhyamika and a sceptic.1
The above parable, meant to underline the distinction between
the 'emptiness' doctrine and scepticism, is, in a sense, somewhat
superficial and may be even misleading. For, it might be argued
iFrasantiapadü pp 156-7.
that the assertion in both the cases is identical and the difference
lies merely in what motivated such assertions. Thus, as a report
oh what the case is, both assertions will enjoy the same 'truth-
value'! But this kind of argument only exemplifies how much
one can be misled by over-extending the point of a parable. In
fact, the usefulness of a parable no longer holds as soon as we
stretch it too far.
We have to understand, with great care and caution, the
implication of the 'emptiness' doctrine. For, it was Nâg-
ärjuna himself who gave the following warning against any
misunderstanding of the doctrine : "Like a snake caught at the
wrong end, or like a craft learnt in the wrong manner, the
'emptiness' doctrine may destroy the stupid person when it is
misunderstood by him!"1
That the pluralistic order of the universe is only a convenient
myth and lacks essence or svabhäva in the ultimate sense is well
expressed by the following Lankâvatâra verse :
Since the essence or 'own-nature5 of things, when they are
critically examined, cannot be established, such things have
been declared (by the Buddha) to be inexpressible and without
essence.
The Mâdhyamika comes very close to the spirit of the Advaitin
with regard to his attitude toward phenomenal plurality. But the
Advaitin seems to me to be more committed to a metaphysical
absolutism in relation to which he seeks to evaluate ordinary
thoughts and experience. The Mâdhyamika, however, tries to
maintain a non-committal attitude in ontology.
The ultimate truth, according to the Mâdhyamika, always
eludes our ordinary experience and conceptual thought. But it is
admitted to be accessible only to a direct but somewhat mystical
experience, a sort of penetrating insight or prajnä. If this unlocks
the door to mysticism in philosophy, be it so. 'Mysticism', at
least 'cognitive mysticism', need not be, it is argued, treated as a
derogatory term. For, as we realize more and more the limits of
language in our analytical struggle, the idea of something in-
expressible may well dawn in our mind, although it would be
difficult to make a logical appraisal of this 'inexpressible'.
My point is that proper understanding of the Mâdhyamika
position ought to produce an incentive to strike a middle course
iMädhyamika-KärikäXXlV.U.
318 Logic, Language? and Reality
between excessive naivete and excessive scepticism. The doctrine

of 'emptiness', sünyatä, is usually presented as the critique of all
views, all philosophical systems. But the implication of this pro-
position can be misconstrued in two ways : one by the opponent
and the other by the so-called proponent. An opponent might
think that the Mädhyamika position amounts to nihilism. But
this is wrong. A proponent might, on the other hand, think that
the Mädhyamika disproves all views, all philosophy. But this
too, is, in our opinion, wrong. If anything, the Mädhyamika
critique is an attempt to show that it is neither proper nor strictly
justifiable to regard any particular metaphysical system as
absolutely valid. Perhaps in the same vein, T. R. V, Murti has
remarked :*
The Mädhyamika dialectic is not refutation;...Refutation is the
rejection of an opponent's view by an interested party having
a view of his own to establish. A critique is the disinterested
analysis of Reason by itself.
Turning again to Jainism, one may say that the doctrine of
'non-onesidedness' (anekânta) asks us to avoid the extremes al-
most in the same way, as the Mädhyamika (or Buddhism in
general) seeks to avoid the extremes (cf. anta) of eternalism and
annihilationism. Hence, it is necessary again to underline the
distinction.
Mahâvïra was not strictly a follower of the Middle Way. For
him, the 'middle' was also an anta, a side> as is evident in the
scheme, the left, the right and the middle. Thus, from the Jaina
point of view, the Buddha would still be an ekänta-vädin, al-
though he was a follower of the middle course. With regard to
the doctrine of the four Noble Truths and the impermanenc© of
the five personality-aggregates, the Buddha held a definite posi-
tion. In other words, with regard to these questions the Buddha
was an ekäntavädin. Similarly, I think the 'dependent origination'
theory of causality in Buddhism is asserted to refute the evil of
both extremes (another illustration of the middle course): sat-
kâry a (the effect pre-exists) and asat-karya (the effect is'newly
created).
In fact, one can follow the 'middle' course in either of the
two ways. First, one can accept the middle course and reject the
iSee The Central Pilosophy of Buddhism, pp. 145-6.
Ontologicaî Problems M9
two extremes (anta). Thus, own merely suggests a third alternative

which excludes the other alternative already suggested. Second,
one can accept the'middle'course without necessarily rejecting
the two extremes. In this case, one's alternative does not com-
pletely exclude the other alternatives. He merely expands himself
to embrace the two alternatives while himself remaining in the
middle. The first 'middle' way is based upon rejection and
exclusion, the second upon acceptance and inclusion. We may
call the first the 'exclusive' middle, and the second the 'inclusive'
middle. The Middle Way of the Buddha was of the first kind.
Mahâvïra's anekänta was of the second kind.
§ 4.6 AVIDYÄ IN BUDDHISM, YOGA AND NYÂYA
The term avidyä is one of the most important terms not only in
Buddhism but also in all important schools of Indian philosophy,
it is thus an extremely difficult task to find an adequate English
(or Western) equivalent of this term. In modern writing inter-
pretation of this term ranges from ambiguous (and evasive)
explanation to total misinterpretation. A satisfactory inter-
pretation of this important concept in the context of Bud-
dhism is hard to find. In the present section, I shall try to clarify
certain misconceptions that are usually generated around our
understanding of the meaning of the term avidyä. It will be a
preliminary attempt to find a satisfactory explanation of the
concept of avidyä. To put it facetiously (but not incorrectly, I
believe), the present section is being written in complete aware-
ness of our present avidyä of avidyä. We may not be able to
remove the second avidyä, but we can hope that more light be
thrown on the first.1
I shall concern myself here mainly with the notion of avidyä
that is explained and analysed in the Buddhist and related texts.
For comparison and contrast, I shall refer to other interpreta-
tions of this term in non-Buddhist sources. But to maintain
uniformity and to avoid complication, I shall refrain from
1
Several students in my class first drew my attention to the fact that a reader
of even the Venerable Rahula's excellent book, What the Buddha Taught,
may be in doubt as to the exact distinction between avijfâ and micchu
diîîhi (false views); p. 3.
commenting on the comparatively late but quite elaborate dis-

cussion of the notion of avidyä in late Vedänta schools.1
To add one preliminary remark : We are concerned here
essentially with the problem of translation of ancient philosophic
terms in English or modern languages. It .will not be out of place
to note a few words about the general problem of conceptual
translation. The American Heritage Dictionary of the English
Language (New York, American Heritage, 1969) defines 'trans-
lation' as a process or an act of expression in another language,
systematically retaining the original sense or meaning. The ideal
of translation, as it is expressed here, is to 'retain the original'.
Let us not worry too much about the notion of'meaning' in this
context. Let us assume instead that we have a clear understanding
of the meaning of the term we are trying to translate. Then, the
point of the above definition may be stated as follows. If we have a
clear understanding of the meaning of the philosophic term we
are trying to translate, then it is expected that we shall be able
to find a suitable term in the second language which will 'retain'
the original sense, provided, of course, such a term is available
in the latter. But if the suggested term in the second language
has certain shades of meaning that are not compatible with those
of the original term, we should not recommend such a transla-
tion. For, such an incautious translation is likely to generate
confusion in the minds of those who are unacquainted with the
first language. Besides, even the weaker requirement of translation
may not be met under such conditions. For, if we translate a
key-term inadequately, even the truth-value of the original
sentences where such term occurs systematically may not be
retained in translation.
Let me begin by noting a few English equivalents that have
been usually accepted by the modern scholars without reservation.
The Venerable Rahula has suggested three alternative renderings
of avidyä in his 'Glossary'; ignorance, illusion, delusion. But he
has generally preferred to use 'ignorance'. E. Frauwallner noted
'ignorance' and 'nescience' as two possible equivalents of avidyä,2
lOne may consult E. A. Solomon : Avidyä—A Problem of Truth and Reality

(Gujarat University, Ahmedabad, 1969), for a discussion of avidyä in late
Vedänta schools.
^History of Indian Philosophy (translated from original German into English
by V.M. Bedekar; Motilal Banarsidass, Delhi, 1973), p. 398.
Both R. Robinson1 and A.K. Warder2 have shown preference

for ignorance'.
The word 'nescience* has a metaphysical overtone, and hence,
it is popular with some scholars. But, in fact, it can hardly be a
correct translation of avidyä. The dictionary meaning of 'nesci-
ence' is just 'ignorance' or 'absence of knowledge' {The Shorter
Oxford English Dictionary), or 'lack of knowledge' (The Random
House Dictionary of the English Language). But there is another
special meaning of the term 'nescience'. It stands for 'agnosti-
cism'.3
The concept of nescience was much in use in the nineteenth
century writings on Agnosticism in the Christian context. It was
supposed to indicate the 'limits of religious thought' and refer to
an undogmatic dubiety or ignorance about the existence of God.
Sometimes, nescience is supposed to be hypostatized as the engulf-
ing darkness (lack of knowledge) that surrounds the illumina-
tion of science. Thus, in view of these peculiarities and overtones
associated with the term nescience, I think, we can unhesitatingly
reject it as a possible equivalent of avidyä. For one thing, when
a Buddhist talks about avidyä, he does not talk about the un-
dogmatic dubiety regarding the existence of God.
I have shown above that there is agreement among several
scholars of today about the suitability of the term 'ignorance' for
avidyä. In fact, some scholars have accepted it without showing
the slightest uneasiness. I shall argue below that even the term
'ignorance' is quite unsatisfactory as a translation of avidyä.
If 'nescience' is divested of its second meaning, i.e., agnosti-
cism, then it becomes a term synonymous with "ignorance'. My
main objection against both 'nescience' and 'ignorance' is tnat
they express a predominantly negative meaning : 'lack of know-
ledge'or'absence of knowledge'. Thus, both translations are
too vague to make avidyä a meaningful concept within the
system of concepts found in Buddhism. Besides, and this is my
second point, the Sanskrit term avidyä (Pali : avijjä), although
it is 'grammatically' negative (having been formed with a nega-
Buddhist Religion—A Historical Introduction (Dickenson Pub., Bel-

mont, California, 1970), p. 22.
Hndian Buddhism (Motilal Banarsidass, Delhi, 1970), p. 103.
3D.D. Runes (Ed.) : Dictionary of Philosophy, (Philosophical Library, New-
York, Revised Edition, 1960), p. 209.
tive particie), does not mean negation (or absence or lack) of

anything. For, it is well-known in Sanskrit grammar that the
negative particle in a Sanskrit compound does not always express
simple negation or absence.
Let us deal with the second point first. It is undoubtedly a
question of grammar and semantics. iAvidya> is a negative com-
pound consisting of a negative particle (a = nan) and 'vidyä\
Patanjali discusses, under Päninisütra 2.2.6 'NAN\ the semantic
problem connected with the interpretation of a negative com-
pound. He uses the example 'a-brähmancf ('not-a-brahmin.).1
Patanjali seems to imply that compounds like 'a-brähmana'
(and ca-vidyä') are tat-purusa compounds where the meaning of
the second member takes predominance over the first. If this
view is accepted, then negation, i.e., the meaning of the first
member should be subordinate to the meaning of the second mem-
ber. Patanjali explains that the word 'a-brâhmana' is applied to
someone who is liable to be mistaken ÜS a brähmana. 'A person
becomes a brâhmana by all these three : birth (in a brähmana
family), spiritual education and spiritual exercise.'2 If one is
devoid of any one quality of a brâhmana or more, he will be
called an a-brâhmana. Thus, people say : 'He who eats (his meal)
while walking (in the streets) is not-a-brähmana.' Here people
refer to a person who is a brâhmana by birth only but does not
behave as a good brâhmana should (or lacks qualities of a good
brähmana). One may also apply the term 4a brähmana' to one
who is not a brähmana by birth but possesses likeness of a brâh-
mana. In such cases, Patanjali asserts, the person referred to
is usually a person who is liable to be mistaken as a brâhmana
due either to false appearances or to false information from
another source.3
What Patanjali says about 'a-brähmana'may be applicable to
'avidyâ'. Thus, following Patanjali, we have to say that 'avidyä'
refers to what can be mistaken as vidyä. If 'vidya' means know-
ledge of reality, or ultimate knowledge, or simply, knowledge,
Wyâkarana-Mahâbhâsya, (Ed. F. Kielhorn, Bhandarkar, Third edition,

1962), pp. 410-12.
2Ibid., p. 411 : tapah srutarjß ca yonis cety etad brähmana-kärakam.
31 bid., p. 411 : <Evam ayam samudäye brähmana-sabdah pravrtto *vayave?v
api vartate jâtihïne gunahîne ca...jâtihîne sandehâd durupadeéâc ca bräh-
maça-sabdo vartate.
'avidyä' will mean something that is liable to be mistaken as

such.
There are two other possible ways by which we can analyse
a negative compound. We can regard it as an avyayïbhava com-
pound where the meaning of the first member of the compound
predominates. In that case, we can understand 'avidyä' as just
lack of vidyä (knowledge). Needless to say that in such a case
English terms like'ignorance' or'nescience' will be suitable equi-
valents of avidyä. But unfortunately, avidyä would in that case
lose its significance in Buddhism. For, 'avidyä' as an avyayïbhava
compound would mean a non-entity, or non-existence (non-
existence of knowledge). But in Buddhism a non-entity or non-
existence cannot (causally) condition another thing. Thus, since
avidyä is regarded as a condition (pratyaya) of samskära, it
would be wrong to interpret avidyä as non-existence of knowledge.
If, on the other hand, we take 'avidyä' to be a bahuvrïhi com-
pound, we have to interpret it as anything that is not vidyä. For,
in this case the compound acts as an adjective of anything that
is not called vidyä. But this also will falsify the notion of avidyä
in Buddhism. For, such things as the eye-organ (and many other
things in the world) would then have to be designated as avidyä.
But 'avidyä5 is not applied to such things in Buddhism. Thus,
it seems better to accept Patanjali's first interpretation of the
negative compound, according to which 'avidyä' will mean only
those items that can be mistakenly thought as vidyä.
A verse is usually quoted by later grammarians to show that
at least six different meanings of the negative particle are derived
from various negative compounds. Vâsudeva Dïksita, in his
Bälamanoramä commentary on Bhattojî's Siddhäntakaumudu
quotes this verse and ascribes it to the pracinäh 'older gramma-
rians' i1
'They describe six different meanings of the negative particle :
similarity, absence, otherness (difference), diminution, impro-
priety (or reproach), and opposition (or contrariety).'
(i) Similarity : 'He is not-a-brähmäna' (see Patanjali's example),
(ii) Absence : 'a-päparrf ('absence of sin').
Waiyäkarana-Siddhäntakaumudi (with Bälamanoramä and Tattvabodhini;
eds. G. Sharma and P. Sharma, Motilal Banarsidass, Benares, 1941, Part 2),
p. 75. See also my The Navya-nyâya Doctrine of Negation (Harvard Univer-
sity Press, Cambridge, 1968), p. 148.
(iii) Otherness : 'an-asvah* (i.e., a cow is other-than-a-horse').

(iv) Diminution : 'an-udarä kanyä9 (cf. 'Here is a girl with no
waist, i.e. with very thin waist'),
(v) Impropriety or reproach : 'a-pasavo 'nyë ('Others are not
animals' meaning 'other animals are not proper for this
sacrifice'),
(vi) Opposition : 'adharmah9 ('It is the opposite of dharma').
The above ramification does not leave any doubt in our minds
regarding the fact that the negative particle in a Sanskrit com-
pound does not always mean mere negation or lack or non-
existence. Vasubandhu has noted that the compound 'avidyà'
is to be analysed in the manner of 'adharmah'. 1 Thus, it should
belong to the last of the above six categories.
We may now examine the philosopher's explanation of the
term avidyä as a concept within a system of concepts. The
Buddha, in his famous first (official) sermon at Benares, is sup-
posed to have emphasized the doctrine of Thirst or Desire or
Craving (tr$nä)9 and pointed out that the origin of duhkha
(unhappiness, anguish or suffering) is due mainly to thirst. W.
Rahula has rendered the relevant passage in English as follows :2
'The Noble Truth of the origin of suffering is this : It is this
thirst (craving) which produces re-existence and re-becoming,
bound up with passionate greed. It finds fresh delight now
here and now there, namely, thirst for sense-pleasures; thirst
for existence and becoming; and thirst for non-existence (self-
annihilation).
The Noble Truth of the Cessation of suffering is this : It is
the complete cessation of that very thirst, giving it up, reno-
uncing it, emancipating oneself from it, detaching oneself
from it.
The Noble Truth of the Path leading to the Cessation of suffer-
ing is this : It is simply the Noble Eightfold Path, namely,
right view; right thought; right speech; right action; right
1
Abhidharmakoèa (Ed. Svami Dvarikadas Sastri, Bauddha Bharati, Varanasi,
in 3 Parts, 1970, 1971 and 1972), p. 463 : 'Adharmänarthäkäryädayas ca
dharma-pratidvandvabhütah. evam avidyäpi vidyäyäh pratidvandvabhüta-
dharmäntaram iti drstavyam.'
2
What the Buddha Taught. W. Rahula; pp. 92-94 (Gordon Fraser, Second
Edition, 1967).
livelihood; right effort; rightmindfulness; right concentration.

(Samyutta-nikäya, LVI).
Thus, there is no doubt that in the first sermon at least, thirst,
and probably thirst alone, was regarded as the chief cause of
entanglement in the cycle of births or existences (which is
duhkha). F. Frauwallner held the view that the doctrine of Thirst
that the Buddha emphasized in his first sermon was in sharp
contrast with other teachings found in many other dialogues—
teachings where avidyä was given prime importance as condition-
ing, through a chain of conditions, suffering and perpetuating
re-birth cycles. Thirst appeared in these teachings beside avidyä
only as a subordinate condition. Frauwallner suggested that the
Buddha, during the propagation of his teachings, had reckoned
with the problem of avidyä causing suffering, and thus decisively
remodelled his doctrine on this point.1 In other words, when
knowledge of the Doctrine, of the Ultimate truth, was asserted
to be a means for the cessation of suffering, it was natural to
think of the opposite of such knowledge, i.e., avidyä, as one of
the contributing factors for the origin of suffering.
Whether Frauwallner's view is acceptable or not, one point is
undeniable. The Buddha's doctrine of dependent origination
or conditioned origination is only a further elucidation of the
connection between the origin of suffering and thirst. It is in
the elucidation of the so-called twelve causal chains of dependent
origination that the notion of avidyä gets prominence. The
Mahänidäna Sütra of the Dlgha-nikäya describes the dependent
origination doctrine as follows :
Old age and death (jarä-marana) (1) arise through the condi-
tion of birth (jäti) (2) Birth or re-birth arises through the
condition of becoming (bhava) (3) Becoming arises through
the condition of attachment or objects of attachment (upä-
däna, literally : 'grabbing') (4) Attachment arises through the
condition of thirst or desire (tr$nä) (5) Thirst arises through
the condition of sensation of feeling (yedana) (6) Sensation
arises through the condition of 'contact' {sparsa) with sensory
and mental stimuli (7) contact arises through the condition
of psychophysical complex (näma-rüpd) (8) the psycho-
physical complex arises through the condition of cons-
istory of Indian Philosophy, vol. 1, (E. Frauwallner), pp. 150-69.
ciousness (vijnäna), and (9) consciousness arises, in turn,

through the condition of psychophysical complex. (See also
§ § 4. 7 and 4.8).
Two significant points may be noted in this connection. First,
thirst or desire is mentioned in the middle of the series and is
connected with the origin of old age and death by a series of
conditions. Old age and death are typical forms of duhkha, and
they are said to lead to grief, pain, lamentation and misery. Tt
is thus an elaboration of the conditioned origin of suffering, the
second Noble Truth. Second, the series is not yet fully developed
here, for certain links found in many other accounts of the
doctrine are significantly absent here. When these missing links
are added to the description of conditioned origination, we
reach back to avidya, which is put at the beginning of the series
of conditions.
Scholars have surmised from different accounts of the con-
ditioned origination doctrine that the complete chain of twelve
conditions (of which avidyä becomes the twelfth condition) re-
presents a later development of the theory, but since the com-
plete chain is found in the teachings of all schools of Buddhism,
it is impossible to know whether this was worked out later by
the Buddha himself or it was added in the earliest period after
the death of the Buddha and before the schisms.1
In the complete twelve-membered chain of conditions, it is
asserted that sensation (vedanä) arises through the condition of
'six spheres' {sad-äy at and) (6). These six 'spheres' or bases stand
for five inner sense-organs and the mind. 'Contact', which is said
to intervene as a condition between sensation and six spheres,
thus, gains a new meaning, being recognizable as a contact bet-
ween six senses and six objects of senses (such as, rüpa 'visible
form'). The six spheres are said to arise through the condition
of the psychophysical complex (8), which, in turn, arises through
the condition of consciousness (9). Then consciousness is said
to arise through the condition of samskära2 'mental formations
or residual forces of karma (see § 4.8 for interpretation of this
term). And samskära is said to arise when there is the condition
of avidyä.
1A.K. Warder : Indian Buddhism, p. 114.
2Frauwallner deliberated at length on the exact meaning of this rather
ambiguous term, samskära. See History of Indian Philosophy, vol. 1,
pp. 156-60.
Now the question arises : what is avidyä ? Avidyä is explained

in this context as misunderstanding or misconception of the
Four Noble Truths, i.e , understanding of falsity as truth. The
Samyutta-nikäya notes in the first of the sütras on the Way
{märgd), that avidyä leads to wrong theory (mithyädrsti), wrong
intention and wrong speech, and the Right Way is to get back
to the right theory, right intention and right speech. Thus also
we can understand the connection between avidyä and samskära.
For, false beliefs and wrong convictions give rise to the pro-
pensities or forces to act wrongly, and to act under misconcep-
tion is to get involved in the cycle of rebirth, in the chain of condi-
tions, in duhkha and bondage. In this context, avidyä can hardly
mean mere lack of knowledge, ignorance. For, wrong actions,
to be sure, proceed from wrong beliefs, wrong convictions, wrong
understanding of the nature of reality, not from simple lack of
knowledge. We do not, for example, act in deep sleep which is
presumably characterized by complete absence of knowledge.
Next let us analyse the Abhidharma-kosa account of the term
avidyä) with the help of Vasubandhu's Abhidharma-kosa-bhäsya.
The notion of avidyä arises in the Abhidharma-kosa mainly in
two different contexts. First, it is listed as one of the six princi-
pal anusayas or klesas 'defilements or taints'. 1 Second, it is listed
as the last member of the chain of conditions in the dependent
origination theory (as we have seen above).
All conditioned dharmas (elements of reality) are classified in
the Abhidharma as sasrava 'conducive to defilements' except
when they fall under the Way (märga, i.e., the Buddhist Way
leading to cessation of suffering).2 Those falling under the Way
and the unconditioned ones are called anäsrava 'unaffected by
defilements'. In fact, in the Abhidharma, the following terms
have closely connected meanings : klesa, anusaya and äsrava.
'Klesa' and 'anusaya5 refer to the same group of items, but they
cannot be strictly synonymous. For, kleêa 'defilement' is said to
be the items in conjunction with which karma 'action' perpetua-
tes our painful existence or becoming, our suffering. But the
latent forms of these defilements are called anusaya. 'Latency' is
further explained in this context as potentiality or the existence
of the defilements in seed-forms. Their potentiality to generate
lAbliidhannakosa, Chapter V, p. 761 tf.
2Ibid., Chapter 1, p. 16.
further defilements is itself the result of the previous defilement.1

The word 'âsrava' is used to refer to any kind of 'taint' genera-
ted by a dharma. But unless and until these taints receive also
'nourishment' and are allowed to perpetuate in seed-form, the
corresponding dharma cannot be called 'conducive to defilements'
or säsrava. That is why the dharmas under the Way might
generate 'taints' but they are unaffected by defilements.2
In the fifth chapter of the Abhidharma-kosa, six major types
of defilements or defilement-potentialities (anusayd) are men-
tioned : passion, hate, conceit (egotism), avidyä, wrong convic-
tion (drtfi), and dubiety.3 Let us direct our attention to the last
three in this list, 'Wrong conviction' has been analysed further
into five types : i) satkäya-drtfi or the view that the five aggregates
are soul or self for us; ii) antagräha-dr$ti or the view that one
of the two extremes, eternalism or annihilationism, is correct;
iii) mithya'dftfi or the conviction that the Four Noble Truths
do not exist; iv) dr$ti-parämarsa or the conviction that regards
the despicable and the inferior as praiseworthy and superior;
and v) sllavrata-pdramarsa or the view that regards the wrong
way as the right way, God as the cause and the Lord of this
universe, and so on.
Vasubandhu notes under mithyâ-drî (false view ?) that all
the other (four) views (mentioned above) can also be designated
as mithyä-drsti 'false view'. But false view as one type of drsti
'wrong conviction' consists in the rejection of the right view
or truth, while other types of wrong conviction consist in
the superimposition of the wrong characteristics on right
things. In fact, all wrong convictions fall under the general
category of false view and the fivefold classification serves only
a soteriological purpose. But what is more important for us to
note that avidyä, being listed as one of the six major defilements,
should apparently be distinguishable from the group of five
wrong convictions. Five wrong convictions are also referred to as
ku-prajM 'bad wisdom'. And in another context it has been
ilbid., p. 763 (Ch. V) : «Prasupto hi kleéo 'nuéaya ucyate, prabuddhah
paryavasthänam, Kä ca tasya prasuptih? asammukhibhütasya bijabhävänu-
bandhah.*
aibicL, p. 17 : «Kämarp nirodha-märgasatyälämbanä api äsravä upajäyante,
na tv anuserate tatra iti na tayoh säsravatva-prasangah.' (Gh. 1).
3Jbid., Ch. V, p. 761.
asserted that avidyä is only a mental dharma concomitant (cf.

sawprayukta) with cbad wisdom', and hence, it is not to be
identified with bad wisdom.1
In the second chapter of the Abhidharma-kosa, a synonym for
avidyä is suggested : moha 'confusion'. Moha'is mentioned here
as one of the six mental concomitants (cf. sawprayukta), which
accompany any defiled mind-moment (cf. klesa-mahäbhümika).2
Vasubandhu notes here that 'mohcf stands for avidyä or ajnäna or
asarpprakhyana. Now, if avidyä is confusion or misconception,
which is to be distinguished from wrong convictions or false
views, what kind of confusion or delusion would it be ? I think
we have to call it 'cosmic confusion' or cosmic misconception
which itself being beginningless is at the root of our involvement
with duhkha and the re-birth cycle.
Explanation of avidyä as a terminal member of the twelve
chain (See § 4.8) of conditions is given by Vasubandhu in
Chapter III of the Abhidharma-kosa. Avidyä is interpreted here
as referring ambiguously to all the klesas 'defilements' (enumera-
ted above) of the previous birth or existence in relation to the
present existence. In other words, avidyä here is a blanket expres-
sion to refer to the group of defilements carried over from the
previous existence to the present existence. Vasubandhu takes
pains to show that use of avidyä here is only a stylistic device, and
should not be construed as equating avidyä with all other defile-
ments. Avidyä is selected from the group of defilements, because
it invariably^ accompanies all other defilements.3 Yasomitra
explains that only under the influence of cosmic confusion or
misconception (cf. müdha) all other defilements arise and per-
petuate. Thus, following the dictum, prädhänyena vyapadesä
bhavanti 'designation is determined by the principal factor', all
the defilements of the previous birth are called avidyä* Or* one
may say, avidyä is used here in order to practise economy of
words. Instead of mentioning all the defilements, only the chief
on£ is mentioned. For, when one says 'The king is coming' it is
implied that his followers (bodyguards, etc.) are also^poming
d., Ch. I l l , p. 463.

2
Ibid.,Ch. II, p. 191.
3
Ibid., Ch. Ill, pp. 436-7.
4Ibid., Gb. Ill, p. 437 ; (Yaéomitra :) 'müdhasya hi kleeasamudäcäräh
nâmCUjhasya.'
along with him. And for the sake of the economy of words, one
does not mention the followers.1
Vasubandhu also deliberates upon the etymological meaning
of the term avidyä. Let me fully quote the discussion from the
Ahhidharma-kosa-bhäsya.
Now, what is meant by 'avidyä' ? If it means that which is
not vidyâ then even the eye, etc. will be designated as avidyä.
If it means lack of knowledge, then it would be a non-entity.
And it is not proper to call avidyâ a non-entity. Therefore,
Avidyä is a dharma (element of reality), a different one, which
is the opposite (reversal) of vidyä. The negative compound
here is to be explained in the manner of amitra ('a non-friend')
or 'an-rta' (untruth*) (v. 28cd). A person who is the reverse
of a friend is called a non-friend (an enemy). We cannot
call just anybody an enemy. Nor can we designate the
absence of friend or friendship as enemy. 'Rta' means truth.
'An-rta' means untruth, and any statement that is opposite
of the truth is called untruth. Similarly, terms like 'adharma',
'anartha' and 'akärya' are explained as the opposite of dharma,
artha and kärya. Thus, avidyä is to be understood as the
opposite of vidyä9 Why so ? Because avidyä has been described
as a pratyaya 'condition' (in the chain of conditions).2
Vasubandhu has quoted further evidences from the sästra to
show that avidyä cannot refer to mere negation of vidyä. The
word 'avidyä' has been used in the sacred texts in connection
with many action-words such as, 'combining'. A mere negation,
Vasubandhu argues, cannot be the object of any such action-
words, and hence avidyä should be construed as a positive
dharma (cf. dharmäntara).3
Since in the Abhidharma, avidyä is used as a blanket expres-
sion for all the defilements (klesas) of the previous birth, a
genuine confusion may arise regarding the import of the term
avidyä as such. Thus it is that Bhadanta Srïlâbha held to the
view that avidyä actually means all the defilements. But Vasu-
bandhu has explicitly rejected this view of Srïlâbha for reasons
we have already discussed above.4
ilbid., p. 437 : 'Räjägamanavacane tad-anuyätrkägamanasiddhi-vat.'
2lbid., p. 463.
3ibid,, Ch. Ill, p. 4G3 : 'Tasmäd dharniäntaram evävidyä.'
4
Ibid., Yaéomitra's Comm., Ch. Ill, p. 465.
It might be useful at this point to compare the list of the five

klesas 'defilements' given in Patanjali's Yogasütra 2.3, with the
list of klesa in the Abhidharma. Patanjali's list runs as foliows :
avidyä, egotism or conceit, passion, hate and abhini vesa 'bewilder-
ment'. 1 Here abhinivesa takes the place of vimati(dubiety' of the
Abhidharma list. It is significant to note that the item called
drsti 'wrong conviction' or 'bad wisdom' is absent from the list
in the Yogasütra. For, it is already comprised by the comprehen-
sive term avidyä in the context of Yogasütra. Yogasütra 2/5
defines avidyä as the understanding of the impermanent, the
impure, the painful and the non-self as permanent, pure, plea-
surable and self. In other words, it stands for the misconception
or misunderstanding of the relevant doctrine and the consequent
misunderstanding of the real nature of things which cause per-
petuation of suffering in the form of birth and re-birth cycle.
Thus, typical false views or wrong conviction can be brought
under this liberal notion of avidyä.
It has been shown above that avidyäcannot be a negative
concept in Buddhism. In a general context, avidyä may stand
for false beliefs or a false belief-system which we all grow up
with in the worldly environment. But in Buddhism it obtains a
specialized meaning. For, not all false beliefs are technically
called avidyä in Buddhism. The ultimate concern in Buddhism.
was for nirvana^ cessation of suffering. If we can call nirvana
salvation, then avidyä becomes essentially a soteriological concept.
It stands for the opposite ofthat ultimate knowledge or insight
which brings about the goal o[ nirvana, final freedom from
suffering- It is the inherent misconception, a beginningless,
cosmic confusion in all of us, which perpetuates our painful
existence.
For a more mundane interpretation of the term avidyä in the
ancient texts, we shall have to consider the Vaisesika system.
Prasastapâda in his systematic presentation of the Vaisesika
system in Padärthadharmasamgraha, notes that there are two types
of bitddhi or cognitive states : vidyä 'true cognition' and avidyä
'false cognition'. Avidyä 'false cognitive state' is thereafter classi-
fied into four classes : doubt, error, uncertainty and dream.
iThe Yogadarsana of Patanjali (ed. M.G. Bakre, Nirnaya Sagar Press, Horn-
bay, 1917), pp. 36-37.
Under 'error' Prasastapäda includes the typical false beliefs (for

the Vaisesika school) such as, believing in what is impermanent
as permanent, what is not-self as self, and what is not good as
good.1 (Vätsyäyana in the Nyâya school gave a similar interpreta-
tion of what he called mithyä-jnäna.) Thus, avidyä in the Vaise-
sika school stands for a more general concept. The important
point to note is that it is definitely asserted to be a cognitive
state (being subsumed under buddhi 'cognitive state'). It is a
cognitive state, but only an untrue one. It is not merely an
absence of knowledge. It includes not only wrong conviction,
errors and delusions, but also doubtful attitudes, uncertainties
and dream states. And it is also asserted that avidyä with regard
to a particular thing is eliminated (destroyed) by the corres-
ponding vidyä 'knowledge' of that thing.
That avidyä can hardly mean lack of knowledge can be pro-
ved with a very simple example. Avidyä is said to be the guiding
force of our action. But certainly mere lack of knowledge does
not motivate us to act. It is knowledge or something mistaken
as knowledge that motivates us to act. A child, for example,
has what we call lack of knowledge or ignorance about many
things, such as, the theory of relativity. But his lack of know-
ledge about the theory of relativity can hardly be interpreted
as his wrong attitude, or false beliefs or misconceptions about
it. For, it will be ridiculous to say that the new-born child has
a misconception about the theory of relativity. We, adult human
beings, on the other hand, not only lack knowledge about the
ultimate truth or the ultimate goal or how this ultimate goal is
to be obtained, but also it is correct to say that we have a
million false beliefs or wrong conviction about that ultimate goal
or the ultimate truth or the Way. The doctrine of avidyä,
divested of its sectarian meanings, simply states that knowledge
begets freedom by destroying our avidyä, our false beliefs, mis-
conceptions and wrong convictions.
iPadärthadharmasahgraha, included in Udayana's Kiranävali (ed. J.S. JfeÜy,

Gaekwad Oriental Series, Baroda, 1971), pp. 171-82,
§4.7: THE ENIGMAS OF BUDDHISM : ÄLAYAVIJNÄNA, DUHKHA

AND NIRVANA
The concept of 'pudgala'

The Brahminical idea of 'self or 'soul was rejected by the Buddha.
This aspect of his teachings is too pronounced tobe missed. But
even at a very early stage of Buddhism, there arose within the
tradition much controversy over the idea of a 'person' (pudgala).
The Kathävatthu supplies ample evidence of this fact The 'Bhära'
dialogue of the Samyutta-nikaya1, where the 'burden' (bhära) is
represented as saijisära and the carrier as the 'person', is often
referred to as the Buddha's concession for some persisting entity
through the cycles of transmigration. Of the early Buddhist
schools, the Vâtsïputrîyas and the Sämmitlyas maintained the
notion of a persisting entity called 'pudgala'. This 'pudgala'
comes very close to the doctrine of'soul', which would natur-
ally be regarded as a heresy in Buddhism. In fact, the Vâtsï-
putrîyas and the Sâmmitîyas were accused of such a heresy. But
the Sâmmitïya's defence of 'pudgala', as K, Venkataramanam
informs us, need not be regarded as a heresy. The pudgala here
is not exactly the ätman of the Brahmanas. The argument is
rather for an entity persisting through the ever-fluctuating states
of transmigration.2
The Buddha accepted a highly complex notion of transmi-
gration without, a transmigrating soul.3 The analogy is that
of a river or stream where there is a continuum but no
persisting entity. Although transmigration can conceivably
1
This sütra is alternatively referred to as Bhârahâra-sutta or Bhära-sutta.
See Sarnyutta-nikäya, Part III, pp. 25-26. Cf the following lines :
Bhärä have pancakkandhä
bhärahäro capuggalo
bhärädänarri dukkharri loke
bharanikkepanam sukham p. 26.
2
The word "transmigration" is, perhaps, an unfortunate translation to
express the Buddhist sense of samsara. Prof. A. K. Narayan drew our
attention to this point at a symposium. I have, however, retained this
translation in the absence of a better word that might be acceptable to
most of us.
3
For the Sammitïya explanation of pudgala see K. Venkataramanam's trans-
lation of the Sammitiyanikäya-sästra, Visva-Bhärati Annals, vol. V, 1953,
pp. 153-242.
be explained without resorting to a persisting soul, the idea

of an underlying 'link' running through the fluctuating states
arose quite naturally in many later systems which tried to ex-
plain the Buddhist notion of transmigration. The pudgala of
theSàmmitïyas was no doubt posited as such a 'link'. TheYogà-
câra's concept of älayavijnäna was another, and perhaps a
more sophisticated, approach to explainaway this 'link' problem.
It is somewhat paradoxical to see that there was also an
important trend in Buddhism which recognized an absolute
spiritual principle. This trend must have created an internal
conflict in Buddhism because the 'no-soul' doctrine, the pre-
dominant trend in Buddhism, could not easily be reconciled to
it. In Mahäyäna, this principle is usually called the Tathägata-
garbha. In Yogäcära system, this principle was attributed to
älayavijnäna.
Vijnäna in the five personality-aggregates

The group of 'five personality-aggregates' (panca skandhäh) is
usually substituted for 'self or 'person'in Buddhist literature.
In interpreting them I shall mainly follow Vasubandhu's Abhi-
dharmakosa-bhäsya. The 'aggregate of matter' refers to the visible
forms of the material world, The 'aggregate of feeling' (vedanä-
skandha) stands for the experiences of pleasure, pain and neutral
feelings. The 'aggregate of knowings' {samjhä-skandhd) is to be
explained as the awareness (or the conceptual construction) of
the 'specific' characters of objects (cf. Abhidhannakosa 1/14).
The 'aggregate of sarriskärä> includes all other mental acts
(citta-dharma) as well as other acts and 'forces' which are not
attendant upon an act of consciousness (citta-viprayukta-
dharma).1
The 'aggregate of consciousness' is usually divided into seven
items. They are: consciousness of the five senses, 'mental' con-
sciousness (manovijnäna) and the mind. But in the Sarvästiväda,
although 'mind' (mano-dhätu) is mentioned, it is not considered
an additional entity.2 It is just any one of the six types of
iFor a good discussion of this item, see P. S. Jaini, pp. 88-98.
2The usual trend in the Ägamas as well as in the Abhidharma is to analyse
dharmasinto skandha, äyatana and dhätu ('aggregates', 'bases' and 'basic
elements'). Thus, 'mind' is included in the vijnäna-skandha. Among the
äyatanas, there is one called mana-äyatana, and among the dhâtus, there is
Ontologîcal Problems 335
consciousness. Tlie consciousness in the immediately preceding

moment acts as the 'locus' (cisrayd) of the consciousness in the
next moment and is designated as 'mind.' This explantion is
related to the Abhidharma concept of samanantara-pratyaya}
In the Sthaviraväda school, however, 'mind' is distinguished
from the six vijnänas (which include mano-vijhand). In Abhidhcun-
matthasamgaho, three peculiar functions (pahcadvärävajjana, or
adverting of mind toward any of the five 'doors', i.e., senses,
and two classes of sampaticchana, or mental 'acceptance' of
impressions—S.Z. Aung's translation) are ascribed to manodhätu
'mind'. 2 Some other functions like santlrana (investigating) and
votthüpana (determining) are ascribed to mano-vijnâna 'mental
consciousness'. Another entity, viz., hadaya-vatthu 'the heart
essence", is sometimes posited by the Sthaviras as the seat or
locus of consciousness, and as locus it is distinct from con-
sciousness.3 But the 'heart essence' is described as a subtle mat-
erial form (süksma-rüpa) and, as such, it is different from citta-
dharma 'mental form'. The Dhammasahgani omitted this 'heart-
essence' from its list, so did the Sarvâstivâdins as well as the later
Yogäcärins. But the Yogäcärins speak of the 'mind' or the
'ego-shrunk mind' (klista-manas) as different from the six
vijnänas. Thus, surprisingly, the Yogäcärins agree with the
Sthaviras in this respect. Another 'subtle material form' in the
list of the Sthaviras was jlvitendriya 'life' or 'life function'. In
Sarvästiväda and Yogäcära, it was included in the citta-vipra-
yuktasamskära 'non-material, abstract elements not concomitant
with any consciousnss'.4
one called mano-dhätu or manovijnäna-dhätu. But mana-âyatana and mano-

dhâtu are considered to be the same. And the Sarvâstivâdins consider the
mano-vijnâna and mano-dhätu to be the same. Cf. Abhidharma-kosa, 1/16.
•3Cf. Abhidharma-kosa-hhâsya, I/16, 17; Abhidharma-dïpa, 1/6, 7.
2For the position of the Sthaviras, see S. Z. Aung's note on dhätu and
äyatana in Compendium of Philosophy, pp. 254-9. See also Aung's note on
pp. 108-9. 3.
3See Yasomitra's Sphutärthä under Abhidharma-kosa, 1/17. See also Abhid-
hammatthasangaho, Chap. Ill, p. 82 :
vatthusangahe vatthuni näma—cakku-sota-ghänä-jUivä-kaya- hadaya-vatthu
ceti chabbidhani bhavanti /
For the Sthaviras' position on the connection between the mano-dhätu (as
well as the vijnänas) and the hadayu-vatthu, see, ibid., p. 83.
4
See note 1, p. 334. For the Yogäcära notion of Jïvitendriya, see Asanga's
Abhidharmasamuccaya : jivitendriyam katamatjnikäyasabhäge pürvakarmä-
viddhe sthitikälaniyame äyur itiprajnapti I p. 11.
More interesting is the concept of bhavähga vijnâna in the

Sthaviravâda. It has been explained as the passive state or 'cur-
rent' (srota) of consciousness unking the fluctuating and trans-
migrating stages. It is contrasted with the active stream of
consciousness (cf. vlthi-citta in Sthaviravâda, and pravrtti-vijhäna
or visaya-vijnapti in the Yogäcära). As S. Z. Aung has aptly put
it, "it is, as it were, the background on which thought-pictures
are drawn" (p. 11). This current is said to be bounded by birth
(pratisandhi) and death (cyuti\ but as death is but a prelude
to another birth, according to the Indian theory of samsara,
this current flows from life to life, from existence to existence.
It is also said to create the false notion of 'personal identity'. 1
Consciousness moments or the flow of active thoughts are
usually classified in Buddhism into 'good' (fcusala) and 'bad' or
'eviP (a-kusala) types. Since good cannot spring from evil or
vice versa, there arose a problem in the Buddhist explanation
of the causal continuum of the flow of consciousness moments:
How can a good citta 'thought' arise out of a bad one? The
Sthaviras explained the causal process taking recourse to the
notion of bhavänga-vijnäna, which is a-vyäkrta 'indeterminate'
(i.e., neither good nor bad) and, therefore, can very well inter-
vene between the emergence of a good thought and a bad one.2
The Vaibhäsikas posit two additional entities called präpti
and a-pröpti (two citta-viprayuktasamskäras), which are merely
two 'forces' one controlling the collection of particular causal
conditions and the other preventing such a collection. Thus,
emergence of a bad citta can be succeeded by that of a good
one through the operation of these two 'forces'—one preventing
the bad while the other causing the good to arise and vice
versa.3
The Sautrantikas.criticize both these views and posit their
theory of'seed' and 'maturity'. Seeds of evil are said to co-exist
with seeds of good side by side, in the form of 'subtle' seeds.
iThe process of active thoughts arising out of the 'passive' mind is descri-
bed in detail in the Abhidhammatthasangaho, Chap. IV. See also Aung's
Introductory Essay, pp. 27-30.
2See P. S. Jaini, Introduction, r>p. 101-10.
3For Vasubandhu's critique of prapti, see Abhidharma-kosa-bhasya under
verses 11/35-36. Yasomitra, explains in Sphutârthâ why this notion is not
exactly the same as the samyoga 'conjunction' of the Vaisesikas.
Only one of them operates at a given time, the one that has
reached its vipäka i.e., maturity. 1
Now, we can take a close look at the Yogäcära classification
of vijiïâna. Asanga, in his Yogäcärabhümi, studies the 'plane of
mind' (rnanobhümi) in its five aspects: its'own nature' (s vabhäva),
its locus, its intentional reference, its accessories, and its action.
Its 'own nature' has three forms: citta, manas and vijhäna. Citta
is älaya-vijnäna, manas is the 'ego-shrunk' or 'ego-centered'
mind, and vijnäna is the 'consciousness of six senses'.2
The locus of consciousness can be of three types: a causal
concomitant (cf. sahabhü), or a causal precedent (cf. samana-
ntard), or a causal seed (cf. bïjâsraya). The causal concomi-
tance of the five types of sensory consciousness are respectively
the five senses themselves. The causal precedent is the 'mind',
which is interpreted, following the Sarvästiväda principle, as
the immediately preceding consciousness moment. 3 The locus
which is the causal precedent to the 'mental' consciousness is
distinguished as the 'ego-shrunk' mind {klistam manah), which is
attendant (samprayukta) with ego-sense, pride, etc. (asmimäna,
etc.). The causal seed, however, for all types of consciousness,
is älayavijnäna. Thus, it is clear that the idea of a causal seed
of the Sauträntikas and the idea of a subterranean current of
consciousness continuum which was vaguely present in the
Sthaviras' talk about the bhavähga-vijnäna, and contributed to
the development of the notion of älayavijnäna in the Yogäcära
system.4
iFor the Sautrântika position, see Vasubandhu's Abhidharma-kosa-bhâsya

under verse 11/36, p. 64, and verse V/2a, p. 278 (see also Yasomitra's
comment on this section). For the Vaibhäsika critique of the 'seed' theory,
see Abhidharma-dïpa, and the commentary, pp. 220-5.
2See Yogäcärabhümi, p. 11. Citta, manas, and vijnäna are usually taken to
be synonymous in the Sarvästiväda school. Compare Ghosaka's comment
in Abhîdharmâmrta : Cittam mano vijnânam ity anarthântararp / niruktâv
evântaramj p. 55.
3See Yogäcärabhümi : Manah katamatjyat sannäm api vijnänakäyänäm
anantaraniruddhaip klistam ca mano y an nityam avidyätmadrstyasmimäna-
trsnälaksanais caturbhih klesaiîi samprayiiktamHp. 11.
4
Cf. Vasubandhu's Trimsikci : Taträlayäkhyam vijnânam vipâkah sarvabïja-
kamlverse 2cd.
338 Logic, Languages and Reality
The Sauträntika School: bija {seed) and vipäka {maturity)

A few historical comments on the origin of the Sauträntika
school and its relation to the Yogäcära school will be pertinent
here. The Sauträntika school might have originated some time
in the second half of the first century A.D. (with Dharmaträta).
It was definitely an established school in the third century A.D.
with such exponents as Srïlâta. Both the Pali and the Sanskrit
accounts agree that this school branched off from the old Sarvästi-
väda school and it was the last of the eighteen Nikâyas. 1 The
very name 'Sauträntika' indicates that this school emphasized
the authority of the Sütras over the Abhidharma. 2 Vasumitra
noted the other name of this school as samkräntiväda. This
name is explained with recourse to its special doctrine of trans-
migration. This doctrine holds that the skandhas transmigrate
(i.e., are transferred) from one existence to the other. 3 This
special 'transmigration' doctrine was further developed into the
doctrine of 'seed' bija, and a discussion of this bija doctrine is
to be found in Vasuhandhu's work and Yasomitra's comment-
ary.4 It has been suggested (e.g., by J. Masuda, p. 66) that
the above Sauträntika doctrine was probably the outcome of
the criticism of the pudgala doctrine (of the Vâtsîputrîyas) by
the Sarvâstivâdins and the early Mahiéâsakas. The Sarvästivädins
and the early Mahîsâsakas maintained the theory of perpetual
flux of dWsamskäras. The Vätsiputriyas, on the other hand, held
that some samskära are momentary while others can persist
for some time.5 They also held that dharmas can transmigrate
(cf. samkänti) from one existence to the other only along with
the pudgala, not by themselves.
The Sarvästivädins and the early Mahîsâsakas protested and
claimed that no dharmas can, in fact, transmigrate, since they are
momentary. In this context, I think, the Sauträntikas argued
that the skandhas transmigrate. J. Masuda has conjectured that
!The Sanskrit source is supplied chiefly by Vasumitra's treatises. The Pali
sources are Kathävatthu, Dîpavamsa etc. See J. Masuda, p. 66.
2Cf. Yasomitra's comment in Sphutärthä : kah Sautràniikàrthah\ye sutraprâ-
mänikä na sästraprämänikäs te Sautrântikâhfj^. 11.
»See J. Masuda, p. 66.
4
See Vasubandhu and Yasomitra under Abhidharmakosa, verses II/36 and
V/2a."
5
See J. Masuda, p . 54.
the *skandha' here probably meant, as the Chinese commentator

once interpreted, the 'seed' btja, or perhaps, it meant what the
Sauträntikas called ekarasa-skandha. The Siuträntika idea of
the ekarasa-skandha was interpreted in the Chinese commentary
as the "subtle consciousness," and Masuda thinks that this
might be due to an influence of the Mahäsämghikas. 1 In any
case, the idea of 'subtle consciousness' must have found its way
in the Yogäcära idea of älayavijnäna, as did the idea of 'seed'
bfja.
The 'seed' theory, in plain language, means that the past pas-
sions leave behind their seeds in subtle forms which possess the
power to produce new passions. In this way the causal chain is
to be maintained. This theory was severely criticized by the
Vaibhäsikas.2 However, there were undoubtedly several 'uncon-
scious' states in the santâna or consciousness series where the
Buddhist faced the problem of explaining the causal continuity
between the immediately preceding vijnäna 'consciousness moment'
and the new 'waking' vijnäna following such a state. For example,
there may be a (i) seizure or swoon (mürcchä) or (ii) a state of
extreme inaction (middha); or, there may be (iii) méditâtionaI
cessation of the 'ego-shrunk' mind-stream or (iv) some other
'higher order' meditational cessation (cf., nirodha-samâpatti, etc.).
The Vaibhäsikas explain that the causal precedent of the
'waking' vijnäna is the vijnäna immediately preceding such an
'unconscious' state. However, this called for some modification
in the usual definition of the'causal precedent' (cf. samananiara).
Ordinarily, there should not be intervention of any moment bet-
ween the causal precedent and its resultant vijnäna. But the
Vaibhäsikas interpreted 'non-intervention' in this context as the
'non-intervention by a sajätiya (similar) moment'. Since the
'unconscious' states described above do not involve any conscious-
ness moment, the said problem is thus avoided
In the Yogäcära system, however, with the introduction of
älayavijnäna it was easy to explain the causal sequence satisfacto-
rily. The flow of älayavijnäna continues in all the above-men-
tioned states. All vijnänas leave behind their residual 'seeds',
which await their respective 'maturities' (vipäka) to generate
further vijnänas. Thus, in the immediately preceding moment of
ilbid., pp. 66-69.
2See note 1 above, p. 337.
any one of the 'unconscious' states described above, all vijhänas

dissolve into manovijnäna, which, in turn, dissolves into älaya-
vijnäna retaining the results (phala) in the form of 'seeds'. The
'waking' consciousness arises out of one of these maturing 'seeds'.
Thus, the âlayavijnana is called the locus of the 'seeds' of all
vijnänas.1
The causal continuum in Samsara transmigration

In Asanga's Yogâcâra system, älayavijnana became almost the
central concept. Thus, Sthiramati asserted that, because of the
presence of älayavijnana, transmigration {samsara) and its cessa-
tion (nirvana) could become possible.2 The implication is that
without resorting to the notion of âlayavijnana it would be diffi-
cult to explain the causal chain in birth and re-birth as well as
the causal sequence in the attainment of nirvana. In this way,
âlayavijnana rose to a unique prominence, which we shall see next.
The perpetuation of existence is usually explained in Buddhism
by the causal chain with twelve members, which is known as the
dvadasânga-pratîtyasamutpâda($ÇQ §4.8). In this causal chain each
succeeding member arises with the preceding member as its 'con-
dition' (pratyaya). Vijnäna is the third member (in this chain) which
comes after samskära, i.e., 'traces' of action in the former birth,
which in turn depends upon avidyä, i.e., wrong beliefs or wrong
tendencies. After vijnäna, comes näma rüpa, the formation of
the mind-body complex, i.e., the five 'personality'-aggregates, at
the time of conception in the womb. The näma rüpa gives way
to the six sense organs and so on until rebirth and old age and
death in the next birth. 3
Sthiramati claims that the third member in this causal chain,
vijnäna, is nothing but âlayavijnana. The Vaibhâsikas explain it as
the pratisandhi-vijnäna, which is interpreted as the five 'personality'
aggregates just at the moment of conception.4 It is called vijnâna
'consciousness' only in a metaphorical sense. Even if we ignore
iCf. Yogäcärabhümi : Sarvabîjakam ààrayopàdàt vipâkasatfigrhitam âlaya-
vijnânaifi bïjâsrayahjp. 4.
2See Sthiramati under Trirnsikâ pp. 37-39.
3The twelve members in the causal chain of transmigration are the following:
avidyâ, samskära/ vijnäna, nämarüpa, sadäyatana, sparéa* vedanä, tr$riät
bhavaj jäti, jarämarana. For the Sarvästiväda explanation of this chain,
see Vasubandhu under Abhidharmakosa verses III/19-28, pp. 121-40.
4
Cf. Vasubandhu's remark : mätuh. kuksau pratisandhik§ane pancaskandhä
vijnânamjv. 131 {Abhidharmakosabhâçya).
the metaphorical sense and consider it simply as the 'conscious-

ness aggregate' (vijnäna-skandha) dit the moment of conception,
it will be difficult to explain how 'traces' of the former birth (cf.
samskära) can givQ rise to such a consciousness aggregate at the
time of conception. The 'traces' from a former birth are not
stable, and hence, will cease long before the time of conception.
And something which has ceased to exist becomes non-existent,
and hence, cannot be a pratyaya, 'causal condition'.
Besides, at the moment of conception the mind-body complex
is also to appear. If the'traces' give rise to 'consciousness' vijnäna
(at the time of conception) they should likewise give rise to the
'mind-body' complex at that time, lïnama rüpa is interpreted as
the 'mind-body' complex of the succeeding stage arising after
th&pratisandhi-vijnäna {vijnäna belonging to the 'conception' time,
the moment of conception), how does this stage differ from its
previous stage (i.e., the 'conception' time) when the 'conscious-
ness aggregate' is there along with the 'mind-body' complex
(panca-skandhäh five aggregates)? In fact, if the initial 'mind-body*
complex including the 'consciousness aggregate' arises depending
upon the 'traces' we do not need to posit vijnäna, the third member
in the causal chain, as intervening between samskära 'traces' and the
näma rüpa. Thus, Sthiramati argues that to maintain consistency
and retain the 'twelvefold' causal chain of Buddhism, we need to
assume älayavijnäna as that subterranean stream of consciousness
where the 'traces' or 'forces' leave their 'seeds'. The causal chain
can now be explained as follows : 'Wrong belief (avidyä) condi-
tions the appearance of 'traces' or 'forces', and the 'traces'
condition the stream of the 'deeded' älayavijnäna, and when the
'seeds' reach maturity (vipäka) the 'mind-body' complex arises.
The same flow of the seeded älayavijnäna starts other continuous
activities in a similar manner at the end of such 'unconscious'
states as nirodha-samäpatti (described above). 1
Sthiramati further argues that without resorting to the älaya-
vijnäna doctrine it would be difficult to explain the process of
cessation in nirvana. His arguments can be briefly stated as follows.
Perpetuation of existence (or samsara) is due mainly to what is
known as klesa 'menial blemishes' or 'passions', such as, attach-
iSee Sthiraraati under Tnipsikä verse 16 Cf. Evam äsatjijnikädifii manovijiiäne
niruddhe tadapagame punah kuta uipadyate yat tasya külakriyä na bhavati /
tat punar älayavijnänädevotpadyate / tad hi sarvavijnänabijakamiti p. 35.
ment (räga) and ego-sense, and secondarily to karma (action)

Since action by itself cannot condition the future existence unless
it is engineered by klesas, we have to consider klesa to be the
root (mûla) of samsara. Thus, only with the cessation of klesa
'mental blemishes', the 'forces' leading to the future existence will
cease to operate. Now, a klesa may be in the process of taking its
course, or it may exist in the form of a 'seed' Only the 'seed' form
of a klesa can be destroyed by an opposite mental state (pratipa-
kfa-citta), other klesas must take their courses in order to generate
further klesas and action. Now, the seed of a klesa must be
located in consciousness, and this locus must be the älayauijnäna,
the subterranean consciousness. Otherwise, to make the destruc-
tion of the seed possible, we have to admit the impossible situa-
tion that the same 'mental' state or the state of consciousness
(i.e., the pratipaksa-citta) acts as the locus of the seed and at the
same time brings about its destruction. Thus, the älayavijnana
doctrine solves the problem of the locus and explains the causal
process in obtaining nirvana. Sthiramati informs that a detailed
discussion of transmigration and its cessation with the help of
the älayavijnana doctrine can be found in the Pancaskandhaka}
Älayavijnana, Tathägatagarbha and äsrayaparävrtti

Apart from the pudgala controversy, there was another stream
in early Buddhism which recognized an absolute spiritual princi-
ple existing in every being. In Mahäyäna, this was usually called
the garbha theory which we find expounded in the Ratnagotravi-
bhäga. An unidentified prâkrt verse2 sums up the garbha as
follows :
Yathä pattharacunnamhi jätarüparji na dissatij
parikammena tad diftharn evam loke tathägata\\
This means that the Tathâgata lives invisibly in living beings like
pure gold in stones and sands and by purification it becomes visi-
ble. Such numerous comparisons of the Tathâgata with pure
gold, with an impenetrable diamond, and with the immutable
ilbid p. 39. This was probably a work of Vasubandhu translated by Hueng

Tsang into Chinese, and Sthiramati probably summarized the book under
the same title (Sylvan Levi).
2E.H. Johnston described it as a Prakrt verse. J. Takasaki mentions it as a
Pâli verse.
gem, point to the positive aspect of this spiritual principle. In the

Yogäcära school, the Tathägatagarbha doctrine became connect-
ed with the älayavijnäna doctrine in a very interesting manner,
which we shall presently see.
As a critique of the nihsvabhävatä doctrine of the Prajnäpära-
mitä the Sandhinirmocana-sütra established the tri-svabhävatä
doctrine and the älayavijnäna.lThsRatnagotravibhäga expounded
the garbha theory also as a criticism of the sünyatä or nihsvabhä-
vatä 'emptiness' doctrine. Thus, comparison between the garbha
doctrine and the älayavijnäna was obvious and natural. Besides,
there was undoubtedly mutual influence in the development of
both doctrines. J. Takasaki has collected important references to
the confusion and the later amalgamation of these two doctrines.
Special mention may be made of the Lankävatära-sütra, Chap.
VI, where the Tathägatagarbha has been identified with the
älayavijnäna. I think that these two doctrines were connected
even in their origin. Thus, in the Mahäyänasüträlankära, Chap.
IX, we find an early fusion of the two. And this must have
influenced the later Yogäcänns like Vasubandhu and Sthiramati.
The following verse from the (Mahâyâna) Abhidharma-sûtra is
quoted in both the places : in the Ratnagotravibhäga to justify
the garbha doctrine and in the Tritpsikä-bhäsya to justify the
älayavijnäna :
Anädikäliko dhätuh sarvadharmasamäsrayahl
tasmin sati gatih sarvä nirvänädhigamo'pi väjj
"It is the beginningless dhätu and the locus of all dharmas,
all causal sequence results from it, even realization of nirvana
is due to it." 2
It should also be noted that the Sauträntika 'seed* theory
might have played some part in connecting the alaya doctrine
with the garbha doctrine. Some element of the 'good' is said to
persist through the series of existence. This Sauträntika idea of
a subtle and incorruptible kusala-müla 'root of good action' is
strikingly similar to the Mahâyâna garbha theory. The incorrupti-
ble element of the good to be found in every being corresponds,
as P. Jaini rightly conjectures, to the 'seed of salvation', moksa-
1
Sandhinirmocana-sütra, Chap. VI, 4-6 (E. Lamotte's translation) pp. 60-65
See above § 3.3.
2See J. Takasaki, pp. 40-45.
bija which we find in later writings. Yasomitra quotes the follow-

ing significant verse where the Buddha compares the 'seed of
freedom' with a seam of gold hidden in a mineral rock. 1
Moksa-bijam ahani hy asya susüksmam upalaksayej
dhätupäsänavivare nilïnam iva kähcanam\\
"I notice the extremely subtle seed of freedom of this man
hidden (in him) like gold in the vein of a mineral rock."
This saying is associated with one of the ten powers of the
Buddha (for which he was called dasabald) viz., the power to
realize the pure and indestructible element, the dhätu or the
gotra or the bija, of every being. Thus, in many contexts 6gotra\
'dhätu9 and 'bija9 were interchangeably used. In Yogäcära, this
pure element was called theprakrti-prabhäsvara-citta 'mind which
is essentially pure and translucent'. And this was further identified
with the tathatä.2
The flow of älayavijnäna continues until nirvana? But, if the
älaya doctrine 4s to be brought closer to the Absolutism of the
tathatä, we have to investigate the state of älayavijnäna at the
realization of nirvana. The Trimsikä says that in nirvana this
subterranean base (the älaya) 'returns to itself, resulting in a
transcendental (lokottara), non-conceptual (nirvikalpa)-jnäna.
And this 'returning to itself happens at the removal of the two
types of dausthulya 'blemishes' ('turbulence'—S. Levi; 'Verder-
btheit—E. Frauwallner) or two types of coverings (cf. vrti in
Mahayänasüträlankära. p. 35).4
The key-terms to be investigated in this connection are
äsray&parävrtti and äsraya-parivrtti (seeTakasaki for references,
pp. 40-45). In the Mahayänasüträlankära, they are often used
interchangeably. E. H. Johnston noted (p. xii) a difference bet-
ween the garbha and the älaya doctrines reflected in this dif-
ference in terms. In the latter, tiiQäsraya or älayavijnäna returns
to itself in nirvana, while in the former, as the pure tathatä is
freed from the sheaths of klesas in nirvana (compare Sthira-
mati's explanation of the process of nirvana given above), a
i Yasomitra quotes the whole episode along with this verse. For P. S. Jaini's
comment, see his Introduction, pp. 115-6.
2Cf. Trirnsikä, verses 29, 30.
3
ibid verse 5a : tasya vyävrttir arhatve,;
4
See Sthiramati under Trintsikä verse 29.
metamorphosis of the asraya takes place. I would note that this

kind of distinction was not always maintained in the literature.
There is, indeed, an obvious and essential similarity between
the two doctrines. Thus, in Sthiramati's explanation of âlayavi-
jnäna, the notion parivrtti 'metamorphosis' seems to have been
assimilated with parävrtti 'returning to itself. Thus, in short, at
nirvana the vijnâna becomes jnäna, and it is called the anäsrava
dhätu 'incorruptible element' (cf. Trimsikä, verse 30), and the
äsrava-kmya or vimaläsraya (cf. Ratnagotravibhäga, I, 44).
It should also be noted that the above view is consistent with
the 'triple nature'doctrine of reality (cf. tri-svabhavatä) of the
Asanga school. The Sandhinirmocana speaks of the triple aspect
of the reality as opposed to its emptiness: the dependent {para-
tantrd) aspect, the imagined (parikalpita) aspect and the perfected
aspect (parinispanna). This doctrine is explained with the help
of an illustration of a crystal ball appearing red due to its pro-
ximity to a red object. The red appearance is called the imagined
aspect, and the red crystal ball is the dependent aspect. The
perfected aspect is the crystal ball itself when considered indepen-
dently of its red appearance, the imagined aspect. Thus, the
doctrine says that the 'dependent' nature is empty of the imagi-
ned nature and it is the 'dependent' nature which turns into the
'perfected' nature when the 'imagined' nature wipes itself out.1
Thus, the älayavijhäna turns into the 'perfected' jnäna when the
'blemishes' wipe out themselves.
The garbha theory, despite obvious dissimilarities, was no
doubt, influenced by the Upanisadic Absolutism, and it might
have influenced, in turn, the ajätiväda 'the doctrine of non-
origination' of Gaudapâda. The significant term 'ajäti9 occurs at
least twice in the Ratnagotravibhäga.2 Now, if the âlaya doctrine
is identified with the garbha doctrine, it will be easy to confuse
the älaya with the 'soul' of the Brâhmanas. This might have
been the reason why the Dinnäga school of Yogäcära ignored
the älaya doctrine.
iSee Sandhmirmocana-sütray Chap. IV. Cf. Tatragunäkaranimittasambaddha-
nâmâni nisritya parikalpitalaksanarnprajnâyatej paratantralaksarie parikalpi-
talaksanâbhinivesani nisritya paratantralaksanaw prajnäyate // paratantra-
laksane parikalpitalaksanâbhinivesâbhâvani niûritya parinispannalaksanarti
prajnäyate / p. 63 (E. Lamotte's translation). See above § 3.3.
2Seepp. 12. 47.
In the Upanisads, the 'soul' is to be finally realized as the

Brahman, the non-dual reality, the Absolute. The Trimsikâ says
that when the älaya returns to itself it becomes the dharmakäya
of the Buddha, the 'non-duaP (advaya) principle, the Absolute.1
But the important difference between the two principles
(although both are called Absolute) should not be overlooked.
The 'souF is the static, unchanging and all-pervading substance;
it is called nitya 'eternal'. The älaya, on the other hand, is the
ever-changing dynamic Mink' like the subterranean current of
water in the ocean; the Trimsikâ calls it dhruva 'an ever-chang-
ing constant' (verse 30). It is the ever-changing ever-lastingness.
(Compare kütasthanityatä and parinäminityatä of the later phi-
losophical literature). Thus, the warning comes from the Sandhi-
nirmocana-süira.2 (The Buddha says:)
ädänavijnänagabhlrasüksmo ogho yaihä vartati sarvabljoj
bäläesäm api na prakäsite mohaiva ätmä parikalpayeyuh\\
"the Receptacle' consciousness is the locus of all seeds, deep
and subtle like the ocean. I have not revealed this notion
lest fools construe this as the 'soul' out of confusion."
Duhkha and Nirvana

These two are admittedly bewildering concepts for us moderners
to understand. F. Kafka once said:
"The scriptures are unalterable and the comments often
enough merely express the commentator's bewilderments."
It is well-known that the doctrine of duhkha or universal
suffering is an essential part of the central teaching of the
Buddha. The four Noble Truths taught by the Buddha are : duhkha
(suffering, pain, anguish), origin or cause of suffering, cessation
of suffering, and the Way. The Yogasütras also talk of a similar
model, which shows that the model was not simply unique to
Buddhism. Vyâsa for example, says while commenting upon
Yogasütras 2:15: 3
x
The sense of 'Absolute' that I have in mind here is usually expressed in
Sanskrit by such expressions as iadvaya-tattva\ ianapeksatva\ 'svatantratvcf
and 'tathatä.'
2This verse is quoted in the commentary of Sthiramati on Trimsikâ; see
p. 34.
3Yogasütras2.15.
Just as the medical science is a system of four items—disease

or sickness, causes of sickness, health (cessation of sickness),
and the medicine to cure, this sastra is likewise a system of
four: sämsära (suffering in the form of repetitious rebirth),
causes of samsara, its cessation (freedom) and the means for
achieving the state of freedom.
The model was obviously borrowed from the therapeutics of
the medical science. Most religious/philosophical systems of
ancient India agree that the world, as it is, is nothing but duhkha
or suffering, and each darsana (philosophical world view)
presents a Way (a prescription !) to avoid or transcend this
duhkha.
Now I shall express one by one my reasons for bewilderment
as a commentator. First, what is duhkha? What is so unique
about suffering and pain that it should be given the status of
the first Noble Truth? If duhkha was the common suffering of
humanity (physical or psychological), then why do we need a
Buddha to discover this truth for us? A.C= Danto has used his
ingenuity to answer this question as follows i1
Everyone in a way knows the first Noble Truth, but it took
a virtual act of genius to see it as the sort of truth that the
Buddha did. Everyone suffers. But not everyone knows that
he suffers...What the Buddha recognized is that knowledge
of fact can be a step towards its mitigation. The first, and in
some ways the hardest, step for a certain sort of sick man to
make is towards the knowledge that he is sick.
This answer does not seem satisfactory to me. 2 For, I think, it
smacks too much of a modern Freudean psychoanalytic model.
I shall raise another set of questions. Why should everything be
regarded as duhkha? Why should even the happy moments or
experience be called suffering or duhkha?
In the Buddhist texts, such as, the Visuddhimagga, three differ-
ent aspects of duhkha are mentioned. The first is called duhkha-
duhkha under which all ordinarily unpleasant things and un-
happy mental states are classified. The second group, called
viparinäma-duhkha, includes every transitory object, even pleasant
iDanto, A.C. Mysticism and Morality., p. 68
2
See for my own approach to the problem of duhkha, Logical and Ethical in
Religious Belief, Calcutta, 1983, ch. 1.
things and happy feelings. They are called duhkha, for, they
change and decay and do not stay long enough to impart happi-
ness. At this level impermanence and duhkha become co-exten-
sive concepts. The third type of duhkha is called samskära-
duhkha, the most pervasive type of unhappiness. It is the reali-
zation of the essentially conditioned nature of our existence.
Our entire psychosomatic existence is called suffering from this
point of view. For, to a man with wisdom, it appears to be
a self-perpetuating imprisoned state dotted by craving drive for
pleasure, the agony of search, dissatisfaction and further craving
happening in cyclic order. In Yogasütra 2:15, where it is declared
that to a person with discernment everything is duhkha, a similar
threefold classification of suffering has been made. It is interest-
ing to see that the samskära-duhkha is regarded there also as
the most pervasive concept.
Duhkha is sometimes explained by the modern Buddhist
scholars as a mental state born out of, probably, frustration due
to the transcience of objects of our craving. If this is accepted,
sukha or .happiness can also be deemed as duhkha. But however,
I think, such an Interpretation refers only to the second category
of duhkha mentioned above. Danto's characterization (see
above) of duhkha as a profound realization fits well in such a
second level interpretation of duhkha. An adequate explanation
of the notion of duhkha requires us to reach the third level—
the notion of samskära-duhkha. An adequate explanation of
duhkha must encompass an adequate explanation of nirvana,
both constitute a whole, and thus one cannot be understood
or explained independently of the other.
Duhkha, as the first Noble Truth, is thus neither physical
suffering nor mental frustration, nor is it simply the obsession
(or a state of paranoia) with the uncontrollable transcience of
our pleasures. It is a profound awareness—a realization that
our existence is necessarily conditioned; samsara is a prison-
house. Cessation of duhkha implies that unconditioned state of
freedom called nirvana. The remark just made about nirvana
brings our discussion closer to an ago-old controversy. I am
sure I shall be criticized by others for giving, perhaps inad-
vertently, a positive characterization of nirvana. Let us formu-
late the controversy and see what iesson, philosophic lesson to
be sure, may we derive from this controversy.
Ontological problems 349
Is nirvana, the nirodha-satya, a positive state or a negative

state? I shall show, at least from one point of view, that any
one who formulates such a question is already a guilty party.
For, the Buddha himself would have forbidden us to formulate
such questions. This is indeed implied by the Buddha's treat-
ment of the avyäkrta questions. Most Buddhist texts would
say that nirvana is neither positive nor negative, and that we
do not have to decide this question at all. But the matter does
not really end there as testified by the controversy for over
two thousand years? I shall divide the issue roughly into three
components.
First, we have to face the logical problem. We shall have to
decide whether the description "neither positive nor negative"
makes any real sense. If "positive" and "negative" are regarded
as contradictory pairs, then we shall have to sacrifice at least
the law of the Excluded Middle in order to allow ''neither
positive nor negative" as a feasible characterization of nirvana.
And this will require, as I have already argued, a modification
in our standard or classical notion of negation. But, perhaps,
this is not an odd claim, for we have come to recognize multiple-
valued logical systems where the problem of redefining negation
has been successfully tackled (§ 23).
We face a further logical problem when it is denied empha-
tically that nirvana is nothingness or mere negation. For, such
denial usually implies that nirvana is something. We are, in
fact, led into the second, i.e., ontological problem, in this way.
Nirvana may be posited as an ontological reality even if we
disallow its characterization as either positive or negative. Or,
to put the matter in another way, nirvana can be said to be an
ineffable reality which transcends (in the sense of 'surpass') the
process of categorization as positive or negative. To a %>od
Buddhist, even this position will be unsatisfactory. For nirvana is
usually regarded as the deontologizing of the self. It is hard to
see how such a concept can be claimed to be an ontological rea-
lity. As a good Buddhist would do, I would advise everyone to
resist all such attempts at the ontological reduction of nirvana.
The third problem can best be called "the Theological
Problem". This problem arises if we regard, as we must under
certain circumstances, Buddhism to be a religion. While I am
fully aware that there is no theos in Buddhism, I use the term
"theological" on the basis of what I venture to call the process

of extrapolation. Tf Buddhism is a religion, it must share some
common feature that is supposed to characterize religion in
general. Man's religious belief seems to be founded upon the
assumption that our earthly existence, our duhkha experience,
is not all that there is. For, there must be another, a better
mode of existence, a possible cessation of the perpetual duhkha
experience. Religion in this sense seems to be born out of man's
resentment against the existing state of affairs and yearning for
something different. Modern historians of religion agree that
there is, in almost every religion, an implicit distinction of the
sacred and the profane. Using a very liberal interpretation,
one can say that Buddhism as a religion implies a provisional
distinction between duhkha and its nirodha (cessation), between
samsara and nirvana. To put it in another way, our duhkha
existence, which is conditioned existence, gains its meaning and
significance only when viewed against the backdrop of the
unconditioned nirvana.
If Buddhism is a religion, we can posit nirvana as the intended
or intentional object of the Buddhist Way, the Buddhist religious
practices or the eightfold path. The intentional object need
not be an ontological entity (F. Brentano called it 'intentional
inexistence') much in the same way "unicorn" need not refer to
an ontolog'cal entity in the expression "he is hunting a unicorn".
Notice that with regard to such intentional objects, there is a
sense in v/hich one can say that the question of their being
existent or non-existent, positive or negative, bhäva f or abhäva,
does not arise. This seems to me a less mystical interpretation
of nirvana, for, we can thus make good sense of the essential
doctrine of the Buddha, the four Noble Truths, or the eightfold
path. The Buddhist religious practices, meditation, etc. may thus
be viewed as being aimed at the intentional goal, nirvana, that
does not actually exist. There is a distinction between our think-
ing (meditating about nirvana or cessation of duhkha) and our
thinking about nothing. In the former case we are intentionally
related to an object (which happens to be an intentional object
that does not exist), in the latter case we are not. In the first
case we have a thought (which may be meant for "killing all
thoughts"), in the second case we have none. And it is a
characteristic of our intentional attitudes that they are, more

often than not, directed towards non-actual objects.
I make the above comment in full awareness of the fact that
Nägärjuna examined and rejected the concept of four Noble
Truths and nirvana in Mädhyamika-Kärikä, chs. 24 and 25.
Buddhism, as it is reflected in the emptiness doctrine of Nägär-
juna, appears to me not as religion but a philosophy of religion.
And, as a good Buddhist would do, I would wholeheartedly
accept Nägärjuha's philosophic interpretation. However my
above comment is conditional upon our acceptance of Buddhism
as a religious tradition.
§ 4.8 : TRANSMIGRATION AND THE MORAL ENIGMA OF KARMA
The doctrine of transmigration is peculiar to India's religious

philosophy. It is a common doctrine shared by Hinduism, Bud-
dhism and Jainism, though each gave its own explanation of
the doctrine to make it consistent with its particular metaphysi-
cal belief-system. There were different ramifications of this
doctrine in each sub-school, as I have already indicated with
reference to Buddhism (§ 4.7). From Hinduism I shall choose the
Nyâya system, and discuss its theory of transmigration in order
to compare it with that of Buddhism.
It is rather unfortunate that we usually render the Indian term
samsara by the English word "transmigration". Even the phrase
"causal chain", when applied in the context of a Buddhist
theory, shares, to some extent, this quandary. The Shorter Ox-
ford English Dictionary explains "transmigration" as "transition
from one state or condition to another; esp. passage from this
life, by death", and as "passage of the soul at the death into
another body; metempsychosis."1 Although the latter meaning
can be acceptable in some non-Buddhist (Vedic) theories of sam-
sara, it is obviously false in the Buddhist theory. The former
meaning is, however, vague enough to be a tolerable equivalent
of the Buddhist idea of samsara.
The Buddhist explains "transmigration" as simply the con-
1
The S hortet Oxford English Dictionary on Historical Principles, revised and
edited by C T . Orions 3rd ed., (Oxford.'Oxford Clarendon Press, 1955). The
third and the fourth meanings are cited.
tinuous succession of one state by another without resorting to

the notion of a transmigrating soul or a persisting link running
through these fluctuating stages. Here we meet the obvious ana-
logy of a river-stream, a continuum without a persisting entity.1
The philosophical (ontological) consequences of such a theory
were very important and interesting, but we shall not go into
them here in any detail. Briefly, the conclusions are : there are
only ever-fluctuating states but no enduring substance; there is
motion from one life-stage to another, but nothing that moves
from one to the other. Thus, perpetuation of personal life was
explained in early Buddhism as well as in its various later rami-
fications in terms of pratyayas "causal conditions" forming a
chain called the twelve-membered 'dependent origination' theory
(dvädasänga-pratityasamutpäda-väda). I wish to make two com-
ments here. First, the karma doctrine, which was so much
ingrained in most philosophical schools of ancient India, under-
went a singular systématisation in early Buddhism. I shall argue
that the karma doctrine was initially an alternative to fatalism
or determinism (cf. the doctrine of the Ajïvakas) on the one hand
and theism (all-powerful God as the Creator and preserver of
the universe) on the other.2 Later on, the Karma doctrine was
combined with theism and as a result, we have such Vedic
philosophical systems as the Nyäya-Vaisesika and the religious
systems like Saivism and Vaisnavism. My second comment is
that although a "Vedic" system like the Nyäya-Vaisesika became
theistic at a certain stage of its development and, unlike Bud-
dhism, admitted the notion of a soul perpetuating through differ-
ent lives or existences, its interpretation of samsara "transmigra-
tion" based on the karma doctrine was not much removed from
the basic idea of a causal chain of several members where one
member depends upon the other for its origin. Curiously enough,
the notion of „God did not play any part in the early Nyäya
explanation of the saipsära.
I shall try to substantiate my second comment. In another
place, I have tried to show that despite the early origin of
theism in India, non-theism was a very dominant trend in the
XOne is reminded of the old problem ascribed to Heracleitas : "you cannot

bathe in the same river twice, for new waters are ever flowing in upon you".
2
See below.
ancient Indian philosophical thought. All the Sramana schools

as well as some ßrähmana schools like the Sämkhya and the
Vaisesika were atheistic in outlook. Here I shall assume that the
early Nyâya was also not seriously interested in theism.1 Thus,
in its discussion of the process of transmigration, the early Nyäya
paid all attention to the mechanical notion of karma and its resi-
dual forces.
To expose the structural similarity between the early Nyâya
explanation of the problem of samsara and the Buddhist explana-
tion of the same, I shall first turn to the Nyäyasütras of Gau-
tama and Vätsyäyana's commentary. Then, for the representative
Buddhist view, I shall draw from Vasubandhu's Abhidharmakosa-
bhâsya, although I am fully aware that different schools of
Buddhism were hardly unanimous in their detailed interpretation
of the Buddha's answer to the samsara puzzle.
Nyäya doctrine of transmigration and causal chain

Nyäyasütra 1.1.2 : Final release arises as the succeeding item
disappears consecutively with the disappearance of the preceding
item (in the following list) : L suffering, 2. birth, 3. motivational
acts (or the merit and demerit derived from motivational acts),
4. evil dispositions3 and 5. wrong conceptions. 2 Here we find a
causal chain with five members or items, which are accountable
for the perpetuation of life.
The first item is duhkha, "suffering". This includes all the
individual experiences of pleasure and pain. Each human experi-
ence is regarded as <cpainful" (and hence it is duhkha), because
we cannot have an experience of unmixed pleasure or joy. Even
our most joyful moments are invariably shadowed by a feeling
of pain. The Buddhists say both, (a) all feelings or experiences
are suffering, and (b) attachment to experiences leads to duhkha,
because all things are anitya. And (b) is sometimes assigned as
the ground for (a). Just as the Buddhist argues that all our
experiences of pleasure and pain are, in fact, suffering mainly
iSee D.H.H. Ingalis, "Human Effort versus God's Effort in the Early Nyäya
MS 4.1.19-21", in Dr. S.K. Belvalkar Felicitation Volume (Banaras, 1957),
p. 228-235. See also Gopinath Kaviraj 'Theism in Ancient India", in Aspect
of Indian Thought (Burdwan, Tndia. 1966). pp. 45-71. See my Logical and
Ethical Issues of Religious Belief Calcutta, 1982, Ch. 2.
^Nyäyasütra (hereafter abbreviated as NS) 1.1.2 : duhkha-janma-pravrtti-
dosamithyâjnânânâm uttarottarâpâye tad-anantarâpâyâd apavargah.
because they are necessarily impermanent (anitya) and condi-

tional, the early Nyäyä similarly argues that all our experiences
are indistinguishable from suffering, because they fall short of
the ideal of pure joy or happiness.
The ideal of joy or happiness, as envisaged here, is compar-
able with the logician's ideal of truth. The "ideal of truth" requi-
res that if a true sentence or cognitive event is compounded truth-
functionally with some falsehood, the resulting truth-functional
compound is always false. The ideal of happiness or joy, as the
Naiyäyika would formulate it, says that if an element of joy or
pleasurable experience is mixed with an element of unhappiness
or suffering, the resulting "whole" experience is to be declared
as suffering. In fact, Nyâya makes a much stricter claim : all our
factual experiences are somewhat like compounded wholes
where at least a part will necessarily be painful, thus making the
whole painful by the above general rule. ït is the indispensabi-
lity of the painful element from the "whole" individual experi-
ences, rather than the conditionally of the experiences as was
advocated by the Buddhists, that makes life-experience suffering»
duhkka. On the other hand, both the Buddhist nirvana and the
Nyäya apavarga can be translated as "cessation", and they are
comparable in the sense that in both of them, cessation of the
duhkha "suffering" or painful experience is the ideal or goal.
However, since in Buddhism there is no soul or personality,
cessation of practical experience of suffering implies cessation
of the false personality notion, but in Nyäya there is a perma-
nent soul, and hence, cessation of life-experience means not only
the end of the false personality notion, but also unimpeded
continuation of real personality, i.e., self.1
The second item is janma, which means the process of birth
and rebirth. This item, I think, is equivalent to the pretyabhäva,
which is listed as one of the twelve prameyas "relevant items of
study for the Nyäya system" in Nyäyasütra 1.1.9. It is further
explained as "rebirth" punarutpatti in NS 1.1.9. Vätsyäyana
glosses over this as the cycle of life, death and rebirth (cf. jan-
mamarana-prabandha) under MS 1.1.19. In short, this iteitf is the
iFor the Nyäya ideal of happiness, consult Nyäyabhäsya under NS 1.1.2 :

"Tad-yathâ r/iadhu-visa-samprktânnam anâdeyam iti / evam sukham duhkhßr
nu sah tarn anödeyam Hi /".
Ontoîogical Problems 355
continuation of life process, which essentially conditions the

previous item, viz., suffering.
The third item combines, a la Vätsyäyana, two principles,
motivational acts and the residual merit and demerit derived
from such motivational acts (cf. dharma and a-dharma). The
motivational act can be of three types : those pertaining to body,
those pertaining to mind, and those pertaining to speech. Exam-
ples of good and bad acts are respectively : protecting and
killing, kindness and greed, and truthfulness and lying. The
Nyäya ethics notes that good motivational act produces merit
and bad act produces demerit.1
In NS. 1.1.9, where the list of twelveprameyas is supplied, the
item pravftti is again mentioned, and in NS 1.1.17, this notion
is explained as the above-mentioned three types of motivational
act. Vätsyäyana, under NS 1.1.17, refers back to his interpreta-
tion of this term under NS I.I.2.2 Our point here is that the
motivational act as well as the result derived from such acts is,
in fact, the condition for the third item, continuation of life
cycles.
One note on the conceptual translation of a technical termi-
nology. Although Vätsyäyana describes the demerit and the
merit as the causes of motivated acts rather than the results of
them, I have preferred the above interpretative translation for
the simple reason that merits and demerits in the Nyäya system
are usually explained as residues of our motivational acts, which,
in turn, give rise to further motivational acts, as the doctrine of
karma suggests. In fact, this notion holds the key to the whole
doctrine of transmigration or sequential succession of lives. I
shall return to this notion in connection with my interpretation
of the Buddhist theory.
The fourth item is do$a, 'evil disposition'. This includes all
our mental tendencies to produce evil. This is the motive force
behind each motivational act. It is broadly classified as räga
"attachment" (to something) and dvesa "aversion" (from some-
thing). Presumably this will be the wrong kind of attachment as
well as the wrong kind of aversion, which will condition the
1
Nyäyabhäsya on NS 1.1.2 : "seyarp päpätmikä pravrttir adharmâya / atha
éubhâ... / seyarti dharmäya /".
2Nyäyabhä?ya on NS 1.1.17 : "So' yam ârambhah àarirena vâcâ manasâ
ca punyah papaà ca daiavidhah / tad état brtabhasyaty dvitïyasùtra ïti /",
motivational act, which, in turn, will perpetuate the cycle of life

and rebirth. But why will attachment and aversion always be of
the wrong kind ? The answer to this question brings us to the
fifth item in the above list.
The fifth or the last item is mithyäjnäna, "wrong conception".
Since we have wrong conceptions of things (or reality), we are
prone to be attached to the wrong thing and the evil, and aver-
ted from the right thing and the good. Our wrong conceptions,,
like the Buddhist avidyä or anädi-väsanä, are ingrained in us
from beginningless time, and this explains the continuation of
birth and rebirth cycle from beginaingless time. These wrong
conceptions are, in a sense, self-perpetuating unless they are
destroyed by perfect knowledge or knowledge of the things as
they are (tattvajnäna). With the disappearance of wrong concep-
tions, the motive force orevi! disposition (dosa) disappears; with
the disappearance of the motive force, motivational activity
ceases, and with the cessation of motivational activity and the
consequent exhausting bankruptcy of the fund of demerit and
merit, rebirth cycle ceases and along with it our sufferings are
put to a stop. Thus, this is like the old prescription of striking at
the root in order to get rid of the growth of a poisonous t r e e -
obviously reminiscent of a similar prescription of the "great
physician" (mahävaidya), the Buddha.
One of the fundamental wrong conceptions is called ahaifikära,
misconception of the self or soul, in the Nyâya system. It con-
sists in the wrong attribution of selfhood to things that are not
self or soul. Thus, in NS 4.2.1, it is said that ahamkära disap-
pears at the appearance of the perfect knowledge with regard to
the causes of the evil forces. This misconception of self is again
said to be beginningless.1 #
But to call something beginningless is, I think, an indirect
acknowledgement (or, should we say, "philosophic" acknow-
ledgement ?) of the failure to answer the question of origin. The
Buddhist gives almost the same answer to the question of the
origin of avidyä "wrong conception" which is adduced to be at
the root of suffering and the rebirth cycle (samsara). If this is
noted as a failure to answer, I would like to point out that
those who have claimed to find a successful answer to the ques-
l/VS'4.2-1; i6Bocanimittänäw taitvajnänäd aharflkäranivrttih /"
lion of origin of everything by positing the notion of God's

will and creation, and so on, have fared no better. In fact,
instead of calling 'the wrong conception' beginningless, the
Indian thinkers should, in fairness, have given a straightforward
answer to the effect that this question about origin is unknown,
and perhaps, unknowable.
To compensate for this weakness in their explanation* the
Indian thinkers (Nyâya as well as the Buddhist) have developed
the notion of a cyclic movement. The so-called causal chain is,
in effect, a causal circle, the constituents of which mutually
condition each other. The wrong conception conditions the
life-experiences and the sufferings therefrom, and they, in turn*
feed back the body of wrong conceptions. And in this way, the
cyclic movement perpetuates by the procedure of output and
feedback until it is stopped by the final release.
Buddhism's Twelve-linked Causal Chain

Let us now consider another exposition of the same Karma
doctrine found in a non-Vedic school, viz., early Buddhism. For
the following discussion, I shall mainly draw materials from
Vasubandhu's Abhidharmakosa (see also § 4.7.)
The twelve members of the causal chain as propounded by the
Buddhist are, wrong conception (avidyä), residual forces (sams*
kära), consciousness (vijnäna) dit the tirne of birth, name and
visual form (näma rüpa), six bases fof six senses [sadäyatana),
contact (sparsa), experience or feeling (yedana), thirst or desire
(trsnä), obtaining or means of obtaining (upädäna), life-process
or action (bhava), rebirth (jäti) and ''old age and death" (jarä-
marana). There is a relation of dependent origination among
these twelve members. Each succeeding item originates depending
upon the preceding item. This notion of "causal" dependence
is to be interpreted very loosely as is well-evidenced by the
Buddhist use of the term pratyaya. The "dependence" relation is,
here, merely sequential, there being no essential causal connection
between any. Here, then, is a possible contrast with the Nyâya
system where causal relations are accepted as factual. But, never-
theless, I like to note here an often confused view that the
Nyâya defines causa! relation only in terms of essential connec-
tion and necessity. The Nyâya too, develops the notion of causal
relation in terms of empirical determination. (See § 4.2).
358 Logic, Language aûd Reality
The first two items in the above "causal''chain of the Buddhist

pertain to the former birth, the last two to the future birth,
while the rest in the middle pertain to the process in the present
birth. Thus, Vasubandhu notes : "This is the birth-cycle which
is beginningless; it is the (system of) dependent origination with
twelve limbs (and) having three parts. Two belong to the former
birth, two to the future birth; eight in the middle fulfil the
present existence."1
Let us concentrate for the moment on the eight items in the
middle. The first three in this group of eight, accounts for the
formation of the mind-body complex, arising out of a pseudo-
personality, which, I think, is comparable to the item called
"birth" (janma) in theNyâya scheme discussed above. The fourth
item is intriguing and some interpreters found it difficult to
explain. Vasubandhu interprets it as the "coming together"
(sannipäta) or the togetherness of the triad, the consciousness
series, the sense-organ (i.e., eye-organ) series, and the so-called
object series (such as, the visual form for the eye-organ). This
"togetherness" is conditioned by the formation of the mind-
body complex, and it conditions, in turn, the experiences or
feelings. To avoid any suggestion of a necessary causal connec-
tion, this fact should be better expressed as follows : when there
is the formation of the mind-body complex, there arises together-
ness, and when there is togetherness, there arise experience|
(yedana). Now, these experiences are bound to be bad experiences!
since at the initial stage of causal chain we had "wrong con-
ception" to begin with. Thus, when there are such experiences,
there arises thirst or desire, comparable to the motive forces in
the form of attachment and aversion in theNyâya scheme. When
there is this thirst or the driving force there arises motivational
act, i.e., act to obtain (or to avert, as the case may be) the so-called
object or the means of gratification (of desire). The last item
bhava ''life-activity" arises, when there is obtaining of the means
of gratification.
lAbhidharmakoéa.
Chap. HI, Verses 19d, 20 : (Ed. Dwarikadas Sastri)
"ity anâdi bhavacakrakam /19
sa pratitya-sanwtpädo dvädasängas îrikândakah \
pwrvâparântayor dve dve madhye'sfau paripürinah //" 20
Ôntoiogicat Problenis 359
Vasubandhu describes all the twelve items as "limbs pertain-
ing to some avasthä or state" (cf. ävasthika).1 Yasomitra explains
that these twelve are, in fact, twelve different states of the per-
sonality series (i.e., series of the five personality aggregates).2
Thus, in fact, the third item, consciousness, is said to refer to
the stage of the "five aggregates" just at the time of conception
in the mother's womb.3 The fourth item, "name and visible
form" (näma rüpa) refers to the immediately succeeding stage
of the same five personality-aggregates; "six bases" (sacläyatana)
refers to the next succeeding stage; and "contact" (sparsdy to
the next succeeding stage which precedes the stage called "ex-
perience" (vedanä) where experiences of pleasure and pain are
possible.4
More intriguing is Vasubandhu's classification of the twelve
items into three groups, viz., kiesa (defilement, evil dispositions),
karma (action or motive forces derived as residues from previ-
ous action to generate further action), and vastu (=phala,
"things" or "resulting process").5 Wrong conception, thirst (or
desire), and means of gratification (upädäna) are called klesa
"evil dispositions \ This is obviously reminiscent of the Nyâya
item doça "evil dispositions". Although in NS 1.1.2 "evil disposi-
tion" and "wrong conception" are separately mentioned (ashas
also been done in the Buddhist enumeration, viz., avidyä of the
former birth, and "desire" and "means of gratification" in the
present birth), under NS 1.1.18, Vatsyàyana clearly states that
dosa means the triplet, räga "attachment", dvesa "aversion" and
moha "wrong ideas or conceptions". "Aversion" and "attach-
Mbid Chap. Ill, Verse 25 :

"âvasthikaïi kiletfo 'yaw prädhänyät tv ahgakïrtanam\
pürvaparäntamadhye §u sammohavinivrttaye //" 25
2
Yaéomitra, Sphutärthä, Vols. MI (ed.) U. Woghihara (Tokyo, 1932-6),
evam ävasthikah j dvädasa pancaskandhikä avasthä ityarthahl" p. 286
z
Abhidharmakosai Chap. Ill, Verse 21c : ^sandhiskandhäs tu vijnänam"
The Yogäcarin calls this item älayavijnäna. See § 4.7.
*Abhidharmakoéa, Chap. Ill, Verses 21d, 22 :
"nämarüpam atah param /"
präk fadäyatanotpadäi, tat pürvam trikasamgamät /
sparsah, präk sitkha-dulikhädi-karanajhänasaktitahr
«Ibid., Verse 26 :
"kleéâs trini dvayam karma sap ta vastu phalam tathä /
phala-hetv-abhisamk?epo dvayor madhyânumânaîaji //"
ment" can easily be assimilated under the Buddhist category

"desire", and upädäna technically refers to a very advanced stage
of "desire", where "means of gratification" is being sought after.
One should remember here Vasubandhu's general interpretation
of all the twelve items : they are just different states or stages
(cf. ävasthikd).
Under the second group, karma is included in samskära
"residual forces (of the previous birth)", as well as bhava "life-
activity" or the "residual forces" of the present life-activity.
If the first group, the klesas, can be described as passive forces,
these two, i.e., the group of karma, may be said to be forces in
action or forces derived from activity and then engendering
further activity. The residual forces of the previous birth
are found in a frozen form called satYiskära. With regard to the
present birth, we are concerned with both the acts and the
residual forces derived from such acts to generate further acts.
This notion is covered by the term bhava. One is again reminded
of the Nyäya item pravrtti which covers both motivational acts
and the residual forcés, such as, merit and demerit. |
The third group, vastu, includes the remaining seven items.
Five of them (vijnäna, näma rüpa, sadäyatana, sparêa and
vedana) pertain to the present birth, viz., the five different states
of the personality-aggregates (as described above) beginning
from the time of conception in the mother's womb till the arising
of "desire" (which is a klesa). The other two items (jäti .and
jarä-marana) roughly cover the similar stages of the personality-
aggregates partaining to the future birth. "Rebirth" (jäti) refers
to the state of consciousness {vijnäna) at the time of birth, while
"old age and death" refers to the other four stages, until the
arising of "desire".1 The members of these three groups, evil
disposition (klesa), action or residual force (karma), and "things"
(vastu), mutually condition each other.2
ilbid., Verse 24cd :
* 'pratisandhiij. punar jäti, jarä-maranam ävidah j
Yoéomitra : "ävidah ä vedanähgät I jarä-maranam näma-rüpädi-caturanga-
svabhävamj vineyajanodvejanärtham bhagavatä jarämaranasabdena cat-
väri angäny uktäniT P • 285. See also § 4.7.
zAbhidharmakosa : Chap. Ill, verse 27 :
^Kle&at klesah kriyä caiva tato vastu tatah punah /".
vastu klesäs ca jäyante bhavängänäm ayam nayahjj".
The Notion of Karma

The notion of karma or residual force holds the key to the
understanding of the whole life-process. This seems to be essen-
tially an Indian understanding. We act because the motive forces
drive us to act, i.e., perpetrate motivational acts, and the results
of our acts necessarily feed back those very motive forces
leading to further acts #nd thus perpetuating the process of
life, death and rebirth. Both the Vedic and the Buddhist schools
agree on this basic notion. The cycle of birth and rebirth has
been set in motion from beginningless time because of the be-
ginningless avidyä, ''wrong conception". The Buddhist also
upholds the doctrine of beginningless avidyä—avidyä which is
ingrained in us. The process of output and feedback in the
karma machine is basically similar in both theories, Nyâya and
Buddhism. Their major differences, however, have been pointed
out before, and I need not go into them here.
There is another striking similarity of this sort between Nyäya
and Buddhism. The well-known Buddhist doctrine of four
"Noble" Truths (cf. äryasatyäni) has a parallel in Vätsyäyana's
explanation of the Nyäya system. The four Truths are: 1 duhkha
"suffering", 2. duhkha-mûla "conditions for suffering", 3. duhkha
nirodha, the cessation of suffering, and 4. mârga, the eightfold
path leading to the cessation of suffering, or nirvana. The group
of four is the Buddha's therapeutics for the avoidance of
suffering.
Vâtsyâyana, perhaps keeping the Buddha's therapeutics m
mind, described the four arthapadas "realities" or "footing of
realities" of the Nyäya system: 1. heya "what is to be avoided*' or
"suffering" i.e., "future suffering" (cf. anägata duhkhä), 1. häna
''avoidance of suffering'*, 3. upäya "means or way leading to
such avoidance", 4. adhigantavya "the position to be achieved,
i.e., the state of absolute freedom from suffering, i.e., the final
release". 1
iNyâyabhâsya on NS 1.1.1: "âtmâdeh khalu prameyasya tattvajnänän nihére

yasädhigamah l tac caiiad ultarasütrenänüdyata iti / heyarjt tasya nirvvart-
takam, hänam ätyantikam^ tasyopäyo 'dhigantavya ity etäni catväry
arthapandäni samyag buddhvä nihêreyasam adhigacchatir •
For Nyâyabhasya, see Nyäyadarsana, Ganganath Jha, Ed., (Poona :
Oriental Series).
Does the above discussion show that the Buddha's model of
explanation and his method of analysis and reasoning penetrated,
perhaps unknowingly, into the minds of some non-Buddhist
philosophers also? I am prepared to give an affirmative answer
to this question, There is, of course, the danger of over-simpli-
fication on this issue, because the question of who influenced
whom among the ancient thinkers becomes very tricky when
sufficient textual evidence is not available. But with regard to
the above, a common pattern is so clearly discernible that one
cannot escape the obvious conclusion. In fact, the well-known
problem of life that the Buddha wanted to solve was also the
problem for which other Indian thinkers wanted to find a
solution. And if a common pattern emerges from the different
solutions offered by these thinkers, this should not come as a#
surprise to us. I
General Remarks on Karma and Samsara

There have been many studies of the concept of karma and
samsara1. Confusions and misinterpretations are only to be
expected with regard to the explanation of such complicated
notions. These two terms are very much in use even by modern
Indians. Social anthropologists of today may apply their ethno-
semantic principles to determine the meaning of these terms as
they are understood in modern Indian society. While I do not
question the efficacy of their methodology, I would like to draw
attention to the fact that even educated modern Indians use
these terms in a multiplicity of senses and very few really care
to know the origin of these concepts and their original meaning.
The above remarks should not be taken as a prologue to my
attempt at giving the original meaning of these concepts.
Neither can I claim to give their original meaning, nor do I
*For a survey of this concept. I refer to P. V. Kane's History of Dharmasâ-
stra, Vol. V, part 2, (Poona, 1962), pp. 1531-1612. Kane gives also a brief
list of modern works dealing with the subject (pp. 1604-5).
A note on the spelling of the word karma. The correct Sanskrit spelling is
karman, nom, sing, karma. But since the most commonly found spelling
in modern works is karma, I have adopted it here. It is interesting to
see that Webster's New Collegiate Dictionary gives the following meaning
of the word karma: "1. In Hinduism and Buddhism, the whole ethical
consequence of one's act considered as fixing one's lot in the future exis-
tence. 2. Hence» loosely, destiny, fate."
hold any brief for it. My purpose here is a very modest one :
1 would like to bring to light some simple assumptions and
considerations implied in these concepts, some simple facts
which are too often omitted in the discussion of karma and
samsara, because they are too obvious ! I believe that some of
the criticisms of these concepts are due to the fact that the 'too
obvious' points are not often mentioned!
I shall indulge in some vast generalizations. Unfortunately,
this will be the nature of the present discussion in this section,
as the sub-title shows. In the same vein I shall try to see whether
any worthwhile lesson can be derived from the KARMA
doctrine in the modern context.
The concept of samsara, which means the cycle of birth,
death and rebirth going round and round, embodies the Indian
notion about the problem of life. The problem, as the Indians
understood it, can be stated as follows. We are caught up, as it
were, in a meaningless routine of birth and rebirth which de-
generates into pain and boredom and from which we seek
release. We seek release because although we are essentially
free individuals, we are, by accident, in a state of bondage. The
drudgery of our daily routine encroaches upon our freedom, and
the dry, repetitive routine of samsara cripples our cosmic free-
dom. The cosmic freedom that is mentioned here is not quite
the freedom to act according to our will but a more basic form
of freedom. It is a freedom from the situation under which
we are bound to act in certain determined sorts of ways, a
situation which does not leave us any choice.
The karma doctrine may be rendered simply as our moral
responsibility for whatever we do. Or, to put it in another way,
it means that our destiny is our own making. The present comes
out of the past and the shape of the future depends upon what
we do in the present.
'Rebirth' in the Indian sense is just what it means. It does
not convey any reference to the spiritual awakening in the
Christian sense. The law of karma is a principle of moral
causation. This principle of moral causation is an extension of
the principle of causation in the field of physical or natural
science to the field of ethics and moral responsibility. Since philo-
sophic activities in India started with the systematic speculations
about the origin of the universe and different causal theories
about it, 1 it is not surprising that the concept of causality should

play such an important role in the sphere of human conduct
and morality. Whether this extension from the physical to the
moral field is justified or only uncritically arrived at, is a separate
question with which we are not concerned here.
Both the KARMA and the SAMSARA doctrines seem to
be gradually taking shape from the time of Yäjnavalkya in the
Bfhadäranyaka UpanisacL Of many significant passages there, I
would like to mention particularly one in the story of Yäjna-
valkya and Ärtabhäga. The story runs as follows. In the court
of the king, Janaka, Yäjnavalkya was answering various 'phi-
losophical' questions put to him by other brähmanas. To Ärta-
bhäga's question as to what happens after death, Yâjnavalkia
chose to discuss it m private with Ärtabhäga alone. The Upajii-
sad states that they went to a secluded place to discuss fhe
question in private and what they discussed was KARMA alone,
what they praised was KARMA alone viz., that a man becomes
good by good works and evil by evil works'. 2 1 think the need for
this privacy arose because the KARMA doctrine was not yet
a generally acceptable doctrine. Again in the same Upanisad,
we find the rudiments of the SAMSARA theory suggested by
such poetic examples of a caterpillar reaching from one blade
of grass to another blade,3 and of a snake-skin lying dead and
cast-off in an anthill while the snake moves on with a new skin.4
It should be noted, however, that in old Rgvedic hymns we meet
notions which approximate the Western concepts of Heaven and
Hell,—-concepts which were greatly modified when the new
doctrine of KARMA and SAMSÄRA was introduced. 5
In the Sramana schools like Buddhism and Jainism, and in
other philosophical sütras, the karma theory appeared as a well-
established doctrine and the samsara hypothesis is presupposed
everywhere. The Lokâyata Sramana6, however, preached a
l
l beleive that primitive philosophical speculations started in India with such
cosmogonical hymns as Rg Veda, X. 129. Nâsadîya hymn. See also § 4.2.
2
Brhadäranyaka Upanisad, III. 2. 13.
3
Ibid.,IV.4.4.
4
Ibid.,IV.4. 7.
5
See P. V. Kane, History of Dharma&âstra, Vol. IV, pp. 154-71, and Vol.
V, Part 2, p. 1532.
ô
The Lokâyatas were referred to as both brähmanas and sramanas in the
hedonistic doctrine and rejected the samsara hypothesis and

the moral responsibility of action. The Ajïvika âramana pre-
ached complete determinism and fatalism (cf. niyati-vâda).1
Although the samsara hypothesis was accepted by the Ajîvika,
human actions were thought to be meaningless, because they
would not affect in the least the inexorable world-process. In
fact, the term 'human action' was a misnomer for the Ajïvika
(cf. a-kriyä-väda). The Jatnas and the Buddhisms opposed this
view and established the efficacy human action (cf. kriyä-väda).
Thus, the Brahminical karma doctrine found its root in the
Sramana tradition.
I have already called the karma-cum-samsâra doctrine a
hypothesis for the explanation of certain phenomena. Let us
examine the background more closely. The ultimate question
in almost all our primitive religio-philosophical traditions takes
»the following form: What happens to us when the body dies?
Does everything end with death? (Or, do we have to think of an
'after-life'?) The answer of the Lokâyata to this question is that
when the body is reduced to ashes after death nothing remains
that will re-appear.2 The usual answer of Christianity is in terms
of the Day of Judgement and eternal reward or eternal dama-
nation. This answer presupposes not only the immortality of
our souls, but also the existence of God, the all-powerful, as well
as His Grace. The karma-cum-samsära hypothesis is just another
possible answer to the above question. But the important thing
is that this answer does not necessarily presuppose the existence
of God as the all-powerful. In fact, the karma theory is inde-
pendent of theism, and on a stronger interpretation it becomes
opposed to theism of the usual form. Thus, Buddhism, Jainism
and the early Sämkhya school—all accepted the karma doctrine,
but were atheistic, or non-theistic.
Let me explain. The karma theory presupposes that there
early sources. For example, in the Rämäyana, they are called brähmanas
and in the Pâli Canons they are called sramanas. See Gopinath Kaviraj,
Aspects of Indian Thought, "Theism in Ancient India", pp. 45-47. Inciden-
tially, this article of G. Kaviraj is, contrary to what the title seems to
suggest, a very good and illuminating survey of all the atheistic doctrines
of ancient India.
iSee Kaviraj, pp. 54-60.
2
Cf. bhasmïbhutasya dehasya punarâgamanam kutah. See Sarvadaraanasarii-
graha (Abhyankara's edition, Poona, 1951), p.2.
is survival after death or, at least, there is perpetuation of

existence. The idea of immortality of soul might have come
from the Egyptian culture. The idea is found in Vedic India as
well as in the West. But the main distinction is that the Christians
believe in the post-existence of the soul after the death of the
body, but not its pre-existence. The Samsara theory of the
Indians extended this notion of perpetuation into pre-existence
and post-existence, and this was necessitated by the karma doc-
trine in a manner which we shall presently see. Buddhism rejec-
ted the idea of a soul but maintained perpetuation and continuity
of life through samsara on the analogy of a river-stream con-
tinuum and thus made the karma theory the very basis of such
samsara or transmigration.
The karma theory in its simplest form means that every act,
whether good or bad, produces a certain result or consequence
which cannot be escaped by the perpetrator of the act. It is
posited as a moral necessity. It is neither a mechanical law nor
an experimentally verifiable principle obtained through induction.
Most of its criticisms are, in my opinion, based upon the
wrong assumption that it is so. It is at best a hypothesis and at
worst a supposition.
I would like to make three points at this stage: (?L)T\IQ karma
theory refuses to admit that the world is arbitrary and that
there is anything called chance or caprice, (b) It is opposed to
the theory of destiny or fatalism, (c) It attempts to explain the
world process without resorting to the grace of a personal
God or an all-powerful Creator. It also avoids the problem of
evil in a novel way, but I shall not go into it here.1
When the Sramana movement (i.e., the movement of the 'free
thinkers') against the established Brahminical ideas was gaining
ground in India, the Lokäyata school accepted the 'chance'
doctrine to explain the origin and perpetuation of life on earth.
They preached complete hedonism. And as a logical consequence
of their 'chance' doctrine, they accepted the theory of complete
annihilation after the death of the mundane body.2 The karma
theory opposed the Lokäyata position, which proves my first
point in the preceding paragraph.
iSee Matilal (1982), op. cit. ch. 2.

2This is referred to as ucchedavâda in philosophical literature. E.%.\Mädhya-
mikakärikä, XV,10-11.
The Ajîvika âramaça school rejected the 'chance' doctrine,

but went to the other extreme. They preached complete deter-
minism and fatalism. Under the Niyati doctrine of the Ajïvika,
'human action' became a meaningless phrase and our freedom
to act was denied. The karma doctrine opposed this position
too, and allowed freedom of will and freedom to act. This
proves my second point above.
The varieties and inequalities of life and human experience
were accepted in ancient India as empirical 'facts' standing in
need of some causal explanation. The short-cut way would
have been to posit an all-powerful God and then make His
'mysterious ways' responsible (i.e., causally responsible) for all
these empircial facts. Most of the early philosophical systems in
India hesitated to accord such a place of all importance to
God. The Sânkhya system rejects the idea of God as the Creator.1
The Vaisesika-sütra forgets to mention God as a possible category.
Even later, when Praéastapâda incorporates God in the
Vaisesika system, God's will or God's grace appears more or
less as an inessential appendage to the theory of creation and
destruction (in terms of atomic conjunction and disjunction).2
The Nyäya-sütra considers the possibility of God's being cau-
sally responsible for this universe and ends with a sceptical
note.3Even the Vedänta-sütra or Brahma-sütra poses the question.
How God can be causally responsible for this universe unless
He is also thought to be either unjust or full of whims and
caprices.4 The dilemma was solved there by making God depen-
dent upon man's KARMA for his creative activity. Buddhism
and Jainism were avowedly atheistic. But they accepted the
karma theory to explain the varieties and inequalities of human
experience.
Man's own past actions are responsible for his present joys and
sorrows, his present 'experiences* (bhoga). His 'character' is his
l
See Sänkhyakärikä Verses 31, 56.
2
See Padârthadhannasamgraha, with Nyâya Kandali (Ed. Durgadhar Jha,
Varanasi, 1963), pp. 121-31.
z
Sce Nyäyasütra 4.1.19-21. For an illuminating discussion on the problems
involved see D.H.H. Ingails* "Human Effort versus God's Effort in the
Early nyâya (N$. 4.1.19-21)," in Dr. S.K. Belvalkar Félicitaiion Volume,
Banaras, Motilai Banarsidass 1957, pp. 228-35.
4
See Brahma-sütra II. Î. 34-36.
own 'destiny'. The present inequalities are due to inequalities in

his past behaviour. This is, thus, a doctrine of complete self-
resportsibility. Not only that. Man is also free to work his way
out of the mess in which he finds himself through his own actions.
Thus, reduced to its bare essentials, the karma theory breathes a
rather healthy spirit of self-determination, and looks at the facts
of life with a much straighter eye. This explains my third point
above that the karma theory reduces God's role to an unimpor-
tant detail, if not entirely unnecessary, in the matter of an ex-
planation of the world-process.
It is, however, a highly paradoxical result that the uncritical
acceptance of the karma theory led, in fact, to conservatism and
inaction in India, that it favoured acceptance of the status quo
rather than a vigorous attempt to remedy injustice and interfer-
ence with the 'law of karma' ! This is a historical fact. To under-
stand the connections one needs to analyse cautiously and care-
fully many other contributing factors. One needs to investigate
many sociological, psychological and environmental points in
detail^ which is beyond the scope of the present book. But it is
highly interesting to see that the law of karma which was origi-
nally formulated to oppose fatalism and encourage free will
became later on a plea for a disguised fatalism. The blind force
of Niyati easily stole into the system under the guise of the
'unseen' forces (adrsta) of actions perpetrated in previous births 1
Even the meaning of Niyati underwent slight change and came
to be used in the sense of'unseen' forces.1
This unusual about-face was brought about with help from
the closely associated samsara hypothesis. If the varieties of our
experience of pleasure and pain, inequalities of our situations,
environments and capabilities, cannot be explained with reference
to our actions in the present life, the full-fledged 'moral causa-
tion' theory seems to dwindle away. Particularly, problematic
situations arise when one wishes to rationalize the tough luck
of new-born babies and their apparently unaccountable sufferings.
It was in this way, I think, that the idea of pre-existence was
forced upon the karma theory to avoid the other unwelcome
iSee Kaviraj, pp-56.

For example, even in the Bhagavadgïtâ freedom of will has often t>eeu
emphasized. Cf, IV., 12, III, 20.
alternative, i.e., that of 'chance' or 'bad luck'. Much in the same

way, it seems to me, the idea of post-existence, or existence after
death, was recognized in most ancient religious traditions when
it was seen that many individuals were receiving neither rewards
for their good actions nor punishments for bad actions.
Once the 'transmigration' hypothesis is thus arrived at, it
began to both uphold and sustain the karma doctrine. We have
already noted that it rendered the IÉVARA doctrine (God as the
cause of the universe) somewhat inessential and made the 'chance'
or 'fate' doctrine generally unacceptable. The Buddhist rejected
the idea of a transmigrating soul but accepted transmigration as
a 'process' perpetuated by the KARMA principle.
The general mechanics of the karma-cum-samsära doctrine is
as follows : Our present situation, our environmental circum-
stances and our dispositional propensities are simply the results
of our previous action (karma). We act under a given circum-
stance and in so acting we enjoy some amount of freedom. (This
element of freedom to act is an important but often neglected
point.) And because of this freedom of choice we must share the
moral responsibility of our actions. How? Some acts produce
their 'entire' results immediately while other acts, after giving
some immediate results, retain some distant or non-immediate
results. This residue takes the form of karma or unused merit
and demerit, and persists unless and until we have enjoyed or
suffered from its fruits. Thus, karma is like a rolling snow-ball
which is ever melting because of the heat but retaining its size
or increasing in volume because of a feedback process. We use
up accumulated karma by undergoing experiences of pleasure
and pain and by acting in a particular way, and our actions earn
more karma for us. And hence comes the advice of the wise or
the philosophers to try to break this monotonous repetition, this
vicious circle of samsara, by seeking freedom or emancipation.
Moksa or freedom is that state in which karma will not be able
to cripple you—says the Indian philosopher.
I have already referred to the existence of the element of free-
dom of choice in the karma theory. Not everything has been
totally determined or fixed by our previous karma, or to use a
more neutral terminology, by our background. Alternative courses
of action are possible and open to mankind.1 Otherwise, the
il wish to refer to the dispute between kriyäväda and a-kriyäväda that was
Buddha's teaching about the way to Nirvana or Sankaia's pres-

cription about the way to moksa would have been pointless.
The whole moksa-sästra would be an uncalled for endeavour.
This 'freedom of choice' aspect of the karma theory is often
more intriguing than it is usully supposed to be. Given a certain
external situation and a set of internal dispositions and propen-
sities, a man is generally believed to act predictably. This is what
we often tacitly assume. But the situation becomes more interest-
ing when our prediction proves to be wrong. That a man may
act or react unpredictably under a given situation., proves his
freedom of choice among alternatives. And this freedom of
choice also necessitates the moral responsibility for the choice
made.
Part of what is called sämagri or 'the totality of causal condi-
tions' is determined, but part of it is not. And in this latter part
man uses his discretion. But once he has made a decision about
a particular course of action, he must reap his own harvest.
There is no getting away from this moral obligation. Such a
hypothesis about the moral causation and responsibility coupled
with the metaphysicafnotion of pre-existence and post-existence
may not satisfy our modern mind, since this may not be the
most reasonable explanation of the phenomena of diversities of
our experience and inequalities of results. But it certainly tries
to uphold the principle of justice and fairness of things, which
we value even today.
The difficulty consists not so much in maintaining the karma
theory as in maintaining the samsara hypothesis. The most
disturbing factor is the fact that the gradual increase in the
world population makes the whole samsara hypothesis highly
dubious.1 A satisfactory answer to the question of the increase
prevalent among the Sramanas of ancient India. This was the bone of con-
tention between the Ajïvaka leader Makkhali Gosâla on the one hand and
the Buddha or the Mahâvîra on the other. The general slogan of the kriyä-
väda was : although the present condition of a person is determined by his
previous activities or karma, he is nevertheless FREE to choose between
a life-style that would finally lead to nirvana and a life-style that would not.
Short of this assumption all the teachings for the practice of austerities,
morality etc. would be pointless.
!This objection is frequently raised in modern works dealing with samsara.
For example, J.E. Sanjana's Dogma of Re-incarnation p. 81.20. For such
inadequate answers see P.V. Kane, p. 1611.
Onîoîogical Problems 371
of the world population under the samsara hypothesis seems to

be impossible. Various answers that are usually given are too
obviously inadequate to receive any comment.1 My job here is
not to justify the samsara hypothesis. Note that the karma doc-
trine is not the same as the samsara doctrine, although in Indian
history one inevitably led to the other. My point here is to
emphasize the moral principle which the karma doctrine tried to
uphold in spite of opposing evidence. To get around such oppos-
ing evidence the karma doctrine helped to develop the samsara
doctrine as a supporting hypothesis. If, however, the same moral
principle can be maintained without necessarily implying the
samsara hypothesis (and at the same time without resorting to
such hypotheses as the 'will of God' or 'the caprice of Nature'),
then, I think, the new doctrine would be just as well acceptable
to those who first proposed the karma doctrine almost two
millenniums ago.
While discussing the problems of ontology in the Indian context
we have got irresistibly and inevitably involved in the discussion
of certain broader issues, such as : what is life? what is the
meaning (goal) of human life? and, what explanation, if any, can
be given for our problems of life, sufferings, inequalities, and
evils ? Whether or not, professional philosophers today (those
who are interested in logic and ontology) should engage them-
selves in a discussion about such broader issues, is however,
another matter. I have only tried here to give an impressionistic
account of the various ways in which Indian classical philosophers
dealt with such questions. In the next chapter, however, I shall
revert to logic and ontology, particularly to the relationship of
Sanskrit grammatical theories with the logical and ontological
theories of classical India.
iFor such answers one may again consult P.V. Kane, History of Dharmaéâ-
stra, Vol. V, Part 2, the chapter on Karma and Transmigration (punar-
janma).
CHAPTER FIVE
INTERACTION OF GRAMMAR
AND PHILOSOPHY
5.1: THE DOCTRINE OF KARATSIA
The problems of Nyâya logic are intimately related to Sanskrit

grammatical theories. The so-called 'logical' (nyäya) procedure
includes analysis of language (worcfj and sentences). It is im-
portant to see how grammatical issues influenced the arguments
of the philosophers in classical India. Modern writers have
sometimes used the term "logic" to designate the pramäna-
sâstra. What is apramänalThe answer in brief is this. APramâna,
as the etymology of the word indicates, is a Pramä-karana, i.e.
the instrumental cause of what is known as Prama. The suffix
'lyut' has been used here in karanaväcya to designate instrument-
ality. Thus, we come to another question, What is a Karana? Or,
to be more precise: what actually is this Karanatä or instrumen-
tality by which a thing is designated as a Karana of something?
A Karana from the point of view of Grammar is one
of the Kärakas or 'cases'; and from the point of view of
ontology, it is a special type of Kärana or cause. Thus Karan-
atä is a Kärakatvavyäpyadharma (a property pervaded by the
generic property Kärakatvä) and also a Käranatvavyäpyadharma
(a property pervaded by the generic property Käranatva). By
Kärakatva, the grammarians understand certain relations that
the nouns have with the verb in a sentence. Generally, as many
as six relations or connectors of nouns to the verb are expressed
in a sentence. These six specific relations or connectors are
Interaction of Grammar and Philosophy 373
grouped under the generic name, Käraka. One cannot exactly

say that these six relations exhaust the list of all the possible
relations of nouns to a verb in a sentence. It is safe to say that
Käraka is a technical name given to these six specific relations
almost arbitrarily chosen. Karana as a Käraka is directly related
to a verb. And as a Karana or cause it refers to some non-
eternal entity which is called a 'Kärya' the effect or result of a
complex of causes. There are causal conditions to produce an
effect or event, but not all causal conditions are Karana. Nyâya
believes in the plurality of causes, but only one of such causes
would be a karana. Let us take some concrete illustration of a
Karana. We use sentences like "kuthärena chinatti" ("He is cutting
a tree with an axe"), taking the axe as the instrumental cause
and, therefore, adding instrumental case-ending to it to express
the instrumental sense. In English, prepositions like "with" or
"by" are generally used to denote the instrumental sense.
Thus, the axe is regarded as atypical example of a Karana with
reference to the relevant event e.g., felling of a tree. As a product,
this event is a result of a complex of causes. But of those so-
called causes the axe enjoys a special position by virtue of which
it is regarded as a karana, and not simply a karana. A Karana
has, thus, been defined as the asädharana Kärana, or the unique
or uncommon cause. But wherein lies its uniqueness or uncom-
monness? To be precise, what is the actual position of the axe
in relation to the effect? In what relation does it stand to the
felling of the tree? Enquiry into this line, I think, will reveal
the true nature of a karana.
That a karana is a käraka par excellence, or more clearly, a
karana par excellence, is unanimously accepted by almost all
scholars. But regarding the true character of this excellence or
supremacy of karana over other causes, opinions vary. One
school of thought (propounded, perhaps, by Uddyotakara first
and then expounded by other scholars), assuming karana as the
most efficient of all causal conditions, defines it as that cause
which is immediately connected with the effect or Phala (Phalo-
padhäyakam käranam). The most efficient cause, according to
them, is the most effective cause, i.e. to say, that which finally
precedes the effect. Thus, a karana, in this view, is the final
(carama) cause, a cause which finally produces the effect. To be
more precise, a karana is that cause which is distinguished from
all other causes that are not immediately connected with the I
effect (Phaläyogavyavacchinna-käranam). It follows that a karana
is such that, it being present, the result must necessarily and
immediately follow (karanamcatatyasmin-sati kriyä hhavatyeva)1
To take for example, in the case of felling a tree, the effective
contact between the axe and the tree will be the final cause and,
hence, it is a karana in this view. For this particular type of
contact {vijätiya samyogä) causes the event itself without waiting
for other conditions. Similarly, in the case of a visual perception,
the final sense-object contact is regarded as the karana.
This view takes the karana t o b e the chief cause. There is,
however, one difficulty connected with this view. In popular
language, we take the axe as the karana in felling a tree, and the eye
as the karana of visual perception. According to this view, the axe
or eye cannot be the karana as they are not the most immediate
cause. In logical analysis we see that it is the axe's peculiar con-
tact with the tree that finally causes its felling, and the sense-object
contact that finally reveals an object of visual perception. Scholars
subscribing to this view may come forward to meet this objection
with a modified definition of kararia. It is that which causes the
action denoted by the verb, but not through the operation
(vyäpära) belonging to a case (käraka) other than itself (svabhin-
na).2 Thus, neither the agent (cutter) nor the object (tree) causes
the action directly in this specified sense. The cutter does so
only by making the axe function well, while the tree by making
the function of the axe possible (because without the tree, the
axe cannot function upon it). Only the axe has been directly
(i.e., through its own operation) the cause of the felling of trees
and not through any other means. So it is a karana and, thus,
the popular usages are justified.
A flaw in this argument is not far to seek. The axe cannot be
the karana, according to the above definition.
A relation is conditioned all alike by both the relata. The
vijätlya contact between the axe and the tree is a relation (a
connector) which is also regarded as the operation (vyäpära)
under the present context and, hence, is necessarily conditioned
by the tree also. Thus, the axe becomes the cause through the
iNyâyakosa, p. 149.
zKâraka-Cakra., Bhavânanda p. 40
medium of the operation (vyäpärä) belonging to a. käraka (e.g. 9ths

tree) other than itself. Such a definition of karana and along
with it the attempt to include axe, eye, etc. (popular examples
of karana) under such definition would be baffled, since, accord-
ing to this view, the most immediate cause immediately preceding
the action is regarded as a karana. In popular usage, the axe is
regarded as karana in a merely secondary sense, the primary
sense of karana being the final cause (kuthärädau karanapadam
gaunam)}
The other school of thought (almost generally accepted by
the Navya-naiyäyikas) overcomes the said difficulty by interpret-
ing the notion of excellence or asädhäranatva or uniqueness over
other causes in a slightly different way. Karana is that operating
cause whose operation is directiy effective, that is to say, im-
mediately connected with the effect or Phala.
Phaläyogavyavacchinnavyäpäravat käranatvam2.—Operation or
vyäpära is technically defined as that which produces the final
product, itself being produced by the operator or vyäpärin in its
turn (cf., tajjanyatve sati tajjanya-janakatvam). It follows that
karana causes the effect through its own operation or vyäpära,
which intervenes between the effect and its karana. Thus there is
no difficulty with the axe or the eye to be called karana in popular
language. The axe is vyäpäravat, its phalopadhäyaka-vyäpära or
directly effective operation being the vijätiya contact between the
axe and the tree, and the eye is so, its contact with the object being
its vyäpära. Thus, well-known examples of karana are justified,
according to the present definition, while, according to the pre-
vious definition, the last cause (which is generally viewed as
vyäpära in the present definition) is the karana purely in its
technical sense.
These two different theories are equally predominant among
the scholars. The reason for their difference lies chiefly in their
respective ideas of the excellence or efficiency (prädhänya or
atisaya) of karana as a cause. One tries to justify the popular
examples of karana like an axe, while the other neglects popular
usage as a mere grammatical convention. There is another view,
hinted at by Jayanta Bhatta and others, which asserts that it is
iKâraka-Cakra p. 42.
*ibid., p. .42.
the Sämagri (the sum total of all causal conditions) which

should be regarded as the karana. No other causal condition,
but the 'whole cause' or Sämagri deserves to be considered
as the chief cause or the supreme cause. The sämagri being
present, the product must follow. It follows from the very
conception of Sämagri. Sämagri, therefore, is the karana,
if karana be viewed as the chief cause. This view, of course,
involves certian logical difficulties regarding the exact nature of
the Sämagri. These have been pointed out by the Jain writer
Prabhäcandra (in his Prameyakamalamärtanda). Other writers
of Nyäya-Vaisesika school do not pay much attention to
this view of Jayanta, which proves that it does not find favour
with them. It is clear that, according to this view, we are to
conceive of an 'ideal cause' composed of all the causal conditions
—positive or negative, remote or proximate—which may reason-
ably be called 'a cause'. Whether the notion of karana can de
jure be applicable to this, is another matter.
One thing is clear. The notion of karana has been taken from
grammar, which in turn follows the usages of the language. And
usages roughly crystallize the idea behind them. In logical ana-
lysis, the 'final cause argument' may be thought of as the best
one and 'the Sämagri argument' may also gain some strength
from an ideal viewpoint. But 'the Vyäparavat argument' or 'the
axe argument' stands, in our opinion, certian crucial tests. The
well known instances of karana are generally classed under
the Nimitta-kärana or 'efficient cause'. (They can best be known
as 'extrinsic and efficient'.) Their importance in the production
of an effect should not be underestimated.
Let us consider certain typical cases of käryas that generally
require a karana. Käryas are technically defined as the absentee
of a prior absence (prägabhävapratiyogin) and, as such, they are
either some substance or dravya (cf., pot), or some quality or
guna (cf., jhäna or cognition), or some activity (cf., throwing a
stone) when positively considered and a negation or destruction,
i.e., posterior absence of a pot (ghatadhvamsa) when negatively
considered. Take the case of a kärya-dravya, i.e., a pot. Its
material cause (samaväyi-kärana) and along with it the so-called
non-inherent causes {asamaväyUkärand) must be there be-
cause the pot is a material substance. (See § 4.2 for asamaväyi
kärana.) For its production again some sort of activity should
be generated in its materials producing their final conjunction

(ärambhaka-samyoga), which implies that a willing agent (who
is a nimittakäranä) must be there to effect the starting push to
these materials which are intrinsically without motion. (For our
present discussion we leave aside the general nimittakäranas like
adrsta, God's will, etc.). All these follow from the general nature
of the case under consideration, but they are not always suffi-
cient for the production of the effect. What more is wanting?
There must be- the karana along with its operation. What I intend
to stress upon is this: the very notion of production of a dravya,
say X, necessarily implies that there must be material along with
its qualities as well as a producer, but that instruments like a rod,
threads, etc. along with their operations should not necessarily
be there. But without them the particular production will not
follow. Therein lies the causal efficiency of instruments like a rod,
etc. which are called the karanas. Similarly, in case of perceptual
knowledge, a knower must be there (for knowledge as a quality
must need the substance to inhere into) and so also the object
(for knowledge is always related to an object, and without a
reference to an object, knowledge is meaningless). But is that
all? No. What more is wanting is the sense-organ—the karana
(even manas may be included here). In the same way comes the
utility of the hand as an instrument in throwing a stone. Now
the case of ghatadhvamsa, the destruction of a pot by a rod, is
also easy enough to explain. Destruction is necessarily related to
its Pratiyogin or the absentee, (i.e. the pot), without which it is
meaningless. So the presence of the pot, a nimittakärana, is ex-
plained. Again, the notion of 'destruction' implies an activity
(kriyä) generated into the parts of the pot resulting in their final
disjunction (vibhäga) by destroying their ultimate 'productive'
conjunction (ärambhaka-samyoga). This activity needs an
active agent as before. But what more is needed? The instrument
rod has also to be utilized, and, thus, deserves the designation of
karana. This is the asädhäranatva of a karanay A karana is
a käranapar excellence.
A popular illustration from ordinary life, by a distant analogy,
may, to a certain extent at least, clarify my foregoing argument.
Suppose, Mr. Q, who was ill, gets weil now and finds four
persons (Mr. A. B, C, D) responsible for this restoration of
health. They are to be thanked. Mr. A happens to be his father
who calls the doctor, Mr. B happens to be his servant who prepares
the medicine, Mr. C is the brother who administers the medicine,
and Mr. D? the Doctor. While going to thank them he finds the
first 3 related to him in some way or other. Their relations per-
suaded them to do their respective duties. But Mr. D, not related
to him, comes from outside. Thus, he is the proper man to get
thanks. So also our karana gets thanks. This may not hold under
close scrutiny, but this is how our pre-philosophical intuition
singles out one particular item from the 'causal complex' and
gives it a preferential treatment.
§ 5.2 : SUBSTANCE AND QUALILTY IN SANSKRIT GRAMMATICAL

THEORY
Credit goes to the Indian grammarians, particularly to Bhartr-

hari, for emphasizing the theoretical indivisibility of a sentence
as an expression of thought.1 The modern linguist realizes that
sentences are initially learned as wholes, although at a later
stage we build up new sentences almost unconsciously from the
learned parts by analogical substitutions and combinations.
Although Bhartrhari, the great grammarian-philosopher of
c. fifth century A.D., appreciated the primacy of the sentence, he
recognized, nevertheless, the pragmatic value of the concept
of word and word-meaning (padärtha) as an aid to the analytical
method in grammar and philosophy. Padärtha 'word-meaning'
is also used in Sanskrit, perhaps, with a shift of emphasis, to
mean philosophical categories such as substance (dravya) and
quality (guna). Thus, it is that Bhartrhari discussed the nature of
certain philosophical categories and their possible relevance to
the grammatical theories of the Päninian school.
It has been suggested by some scholars that the system of
Vaiéesika categories of substance, quality and action was much
influenced by the grammatical categories of the Sanskrit language.
Thus, it will be instructive to see how the ancient Indian gram-
marians defined the so-called categories of substance and qua-
lity. We shall also consider how Bhartrhari gleaned philosophical
1
See John Brough, "Some Indian Theories of Meaning", Transactions of the
Philological Society (Oxford) 1953, pp. 164 167. See also Gaurinath Sastri:
The Philosophy of Word and Meaning (Calcutta) 1959, p. 83ft\
materials from different portions of Patanjali's Mahäbhäsya and

tried to incorporate them into his system.
Bhartrhari expounded his theory of Sabdädvaita, the philos-
ophy of transcendental reality, in the first book (kända) of his
Väkyapadiya. In the second book (kända), he examined the
notion of empirical reality, by which he meant the notion of
words and sentences. In the first chapter of the third book (kända)1
he restated his position as follows : Just as the padavädins (those
who regard the word as the primary indivisible unit) consider
word-constituents, such as roots and suffixes, to be mere fic-
titious abstractions from words, so also the väkyavädins (those
who regard the sentence as the indivisible unit) consider words
to be imaginary abstractions from the sentence.2 What holds at
the syntactic level also holds at the semantic level. Thus, in the
beginning of the fourth chapter of the third book (kända), Bhartr-
hari rephrases his theory as follows : For the grammarians (i.e.,
for the väkyavädins), the word-meanings are fictitious con-
structions from the sentence-meanings (which are to be regarded
as wholes), just as for the padavädins (the Mlmämsakasi), the
meanings of roots and suffixes are mere abstractions from the
indivisible word-meanings. The sentence-meaning, according to
Bhartrhari, is an indivisible samvit—one whole cognitive content
—from which different word-meanings—also called ontological
categories—are abstracted by imaginative construction.3
Our main concern here is with the two principal categories,
viz., dravya 'substance' and guna «quality', and their claim to
be regarded as padärtha 'word-meaning5. This will obviously be
related to our discussion in the previous chapter, particularly
§ § 4.1, 4.3. Incidentally, the term dravya 'substance' in Bhartr-
hari's system has a very special metaphysical sense—a sense with
which we will not be concerned here. Bhartrhari discusses this
metaphysical sense of dravya 'substance' in the second chapter
ofthethirdbookofhis Väkyapadiya. AsHeläräja, commenting on
the third book, points out, the term dravya 'substance' is used in
two senses : (a) the ultimate reality or the ultimate substance—
iThe third book is usually called Prakîrnaka. Whether this formed a part
of Väkyapadiya or a separate work by itself is a debatable question. All
three books are sometimes referred to as Tri-kändL
^Väkyapadiya, Book III, chap. I, verse 1.
^Väkyapadiya, (Benares Sanskrit Series), Book Hi, chap. IV, verses 1 and 2.
päramärthika-dravya and (b) the empirical substance—sämvyava-

hârika-dravya.1 The ultimate substance is imperishable and is
identical with what is called ätman.2 The notion of empirical
substance has been discussed by Bhartrhari in the fourth chapter
of the third book (kända). It is this notion of substance that
can be contrasted with the idea of 'empirical3 quality discussed
in the next, i.e., fifth chapter. And it is this notion of substance
which the grammarian Vyâdi (mentioned by Kätyäyana) believed
to constitute the primary meaning of words. In the present
section we will be concerned with this notion of substance.
Panini (c. B.C. 400-300) implicitly made use of certain onto-
logical categories, such as substance and quality, while he was
developing his grammatical system.3 Bhartrhari indicates that
the nature of these categories can be inferred from an analysis
of Pänini's grammatical rules. The commentator Helârâja adds
that the rule prescribing ekasesa (Pänini-sütra 1.2.64) implies
the following notion of substance : A substance is that which
is to be distinguished from other substances by using suitable
words referring to them. This is a very significant statement. Let
us examine it briefly.
Ekasesa is a grammatical operation of the Sanskrit language
by which two or more words (having identical phoneme sequen-
ces) referring to two or more objects or substances are (omitted
and) reduced to one word (with dual or plural inflections as the
case may be). By extension, the operation ekasesa is also
made to cover the cases of homonyms (identical phoneme
sequences having different meanings) and many other cases.4 For
, ca dvividham päramärthikam sämvyavahärikatp ca I tatra dvitiyam
bhedya-bhedaka-prastävena guna-samuddese vaksyati vastüpalaksanamyatre-
tyädinäj anena ca dravyena vyâdi-darsane sarve sabdâ dravyâbhidhclyino
bhavanti"—Heläräja's commentary on Vâkyapadiya. Benares (Benares
Sanskrit Sesies (1905), p. 85.
zVàkyapadïya, Book III, chap. II, verse 1.
3The two terms dravya and guna have, in fact, been used in various senses
on different occasions in Pänini's system of grammar. But setting aside
the controversies, we can say that what we will be discussing here are the
chief senses of these two terms.
4
Thus, for example, we have ekaêesa in words like pädäh (which combines
three different pâda words used in three different senses, foot, quarter of
a verse and one quarter in general) as well as in words like pitarau
'parents' (which combines pitr and matr by the special rule of Pänini
"pitâ mâtrâ" 1.2.70).
our purpose, it is important to note that the prescription of

ekasesa by Pänini was motivated by the idea that normally
we should utter or use words separately in order to refer to
separate individuals or objects. In other words, nominal words
should be treated as proper names or singular terms referr-
ing to one individual or substance at a time. Thus, a single
utterance of vrksa 'tree' refers to a particular tree, and another
utterance of vrksa 'tree' refers to another tree, and a third
utterance to a third. Now, tree 1, tree2 and tree 3 are ail different
individuals as far as their substance is concerned. To put it more
forcibly, we may define the word X referring to A and the word
X referring to B as 'one word5 X only for our practical con-
venience, but we should not forget that they are, in fact, two
different tokens referring to two distinct entities. This is the con-
clusion to be derived from Pänini's rule 1.2.64 prescribing
ekasesa. The philosophical implication of this linguistic theory
is, according to Heläräja, that the substance of an individual
object, say, a tree, is that which is distinguishable through our
use of the word 'tree' and in the absence of the eka§esa opera-
tion people are expected to use two or more "tree" words to
express the individual substances of two or more trees.
The general concept of quality was assumed by Kätyäyana
(c. B.c. 300-200) when he explained the rule 5.1.119 of Pänini.
He said that suffixes like'-taa' and '-fa' are expressive of quality.
Hence, according to Kätyäyana, if Y is a quality which is posses-
sed by a substance A and if because of /1's possession of Y, A
is expressed by the word *X\ then the word 'X-tva' or cX-tä9
(comparable to 'J^-ness' in English) will denote the quality Y.1
This statement, according to Heläräja, presupposes the definition
of quality as- that which is dependent on some substance or
other as its substratum.
Bhartrhari proceeds to define the notion of substance as
follows : Substance is that which is referred to by pronominal
words like 'that' and 'this' and which is purported to be dis-
tinguished by virtue of its being expressed by some name or
other.2 Heläräja makes the following comments on this verse of
i"yasya guriasya bhâvâddravye sabdanivesas tad-abhidhäne tva-talau" Kätyä-

yana under sütra 5.1.119.
zVäkyapadiya, Book III, chap. IV, verse 3.
382 Logic, Lanauage and Reality
Bhartrhari. There aie two types of pronouns. Some pronouns

like 'everything', 'something', 'this' and 'that' range over our
whole ontology (cf. vastumäträbhidhäyinah). Other pronouns like
'another'and'other than'are used in a comparative sense. Of
these two groups, those belonging to the first are, according to
Bhartrhari, the basic media of reference to substances.1 For
example, 'this' refers to a substance which is presented to our perce-
ptual field, and'that'refers to something which is absent from our
perceptual field but whose existence can presumably be proved by
some other means of knowledge (pramänäntara). Thus, 'this' and
'that' range, in fact, over the whole domain of entities, i.e., subs-
tances, either perceived or unperceived. It is contended that the
nature of each entity, i.e., each substance, is so peculiar and speci-
fic that we cannot expect to arrive at any general characteristic
possessed by all of them except that they are objects of pronominal
reference (of the pronouns of our language). Heläräja goes on
to quote Yäska, the author of Nirukta (c. B.C. 400), in order to
support this definition of substance by Bhartrhari. 2
The above definition of substance is based on a rather complex
notion, the relation of reference. But Bhartrhari approaches the
problem of defining substance also from a commonsense point
of view much in the same manner as Patanjali did in his Mahä-
bhäsya. Substance is regarded in this view as a relative concept
being dependent upon our concept of quality (guna). A substance
is that which is bhedya 'purported to be distinguished', and a
quality is the bhedaka, i.e., that which distinguishes the substance.
Patanjali, under Pänini-sütra 5.1.119, develops the concept of
substance and quality in the following manner. By quality, Patan-
jali means primarily the sensory qualities like sound, taste,
colour, touch and smell. Substance, according to him, is some-
1
We can get an important philosophical insight from this remark of
Bhartrhari. Any entity will be called a substance in this system. To be a
substance means, according to Bhartrhari, to be in the range of reference
of a pronoun. Compare this with the remark of a modern western philos-
opher, W.V. Quine. "To be assumed as an entity is, purely and simply, to
be reckoned as the value of a variable. In terms of the categories of
traditional grammar, this amounts roughly to saying that to be is to be in
the range of reference of a pronoun." From a logical point of view (Cam-
bridge, Mass.) 196!, p. 13.
2<<
etad eva nirukia-kärenäpy uktam, ata Hi y ai pratïyate tad dravyam iti"—
Heläräja's commentary, pc 141-42.
thing different from these qualities. It .may be objected that we

do not experience anything over and above these sensory quali-
ties. If, for example, we cut an animal (that is, a so-called sub-
stance) into a million pieces it may be contended that we do not
find or experience anything, apart from these sensory qualities.
Thus, the concept of substance as something different from
quality is merely a linguistic myth. Patanjali answers this objection
by saying that substance is not only something which we can
perceive, but also something of which we can prove the existence
through inference. The kind of inference Patanjali uses to
prove the existence of substance is called sämänyatodrfta, in
the terminology of ancient Indian logic.1 Briefly we can describe
this type of inference as a method by which we come to know
of the existence of an unperceived object or event through parity
of reasoning and analogy. Thus, the growth and decay of trees
and plants are inferable from the gradual changes in their sizes,
i.e., expansion and contraction of their bodies. A better example
would be the inference of the movement of the sun, stars and
planets, The movement of the heavenly bodies is not visually
perceptible. But, by observing their positions and displacements
at differnt times, we can infer their movement. The analogy
which works behind this inference is this : we do see that the
displacement and different positions of human bodies or earth-
ly bodies result from their movements and we believe that the
same law must hold good in the case of heavenly bodies.2 In or-
der to infer the existence of substance, Patanjali recommends us
to proceed as follows :3 If one takes an iron bar and a cotton-
wool pad of identical size and shape, and weighs them, one sees
iNyäyasütra 1.1.5 mentions three types of inference ipürvavat, éesavat and
eämänyato-drtfa. Opinions vary regarding the exact significance of this
classification See § 1.4 above.
2Astronomy tells us that the sun and the stars are stationary objects.
But here we are concerned with the commonplace inference of motion.
We may substitute motion in this context by 'apparent motion*.
In the case of planets such substitution is not even necessary. Another
word of caution comes from Nägesa. Here we have to neglect the philoso-
phical view according to which all types of motion lie beyond the scope
of sense perception.
3
Kaiyata comments here that although substance is perceptible according to
some philosophers, Patanjali is assuming the position of those who would
deny the perceptibility of substance.
that the weight of the one is quite different from that of the
other. What accounts for this difference in weight should be
called substance. In other words, iron and cotton are two different
substances which have different weights. We can distinguish
between the substance of one object and the substance of another
object also with reference to other properties. Thus, one object,
viz. a sword will cut our body if it touches it, while another
object, viz. a cotton garment, does not behave in the same way
under similar conditions. This difference in behaviour should be
ascribed to their difference as substance. Some object (say, a
hammer) can break a body by a single stroke, while another
object (say, cotton string) cannot break it even after a second or
third stroke. Such differences^ are also to be ascribed to their
differences as substance.
The above argument for the existence of substance has many
flaws. It may be asked why the difference in behaviour or even
in properties should be ascribed to the substance-hood of the
objects. It is equally tenable that only sensory qualities or other
qualities of objects are responsible for their different behaviour
under similar conditions. Perhaps, what Patanjali meant by sub-
stance here is what is called svabhäva 'own nature' or inherent
nature of objects—something which is unique to each object and
consequently accounts for its peculiarities. Later commentators
like Kaiyata were deeply influenced by the Vaisesika notion of
substance and hence tried to explain this passage of the Malta-
bhasya by following the Vaisesika définition of substance as a
substratum of quality.1
Patanjali's second attempt to define substance can be stated
as follows : Substance is the unchanging state of objects. Quali-
ties emerge and disappear in order to make room for new quali-
ties. What stays permanent amid ail changes is what is called
substance. It is the tattvaov 'essence' of an entity. Thus, a mango
fruit becomes green at one time and yellow or red at another
time, but it still remains the same mango or, at least, we call it
'the same mango'. What remains unchanged is the mango-sub-
stance. This seems to be a more sophisticated attempt to esta-
blish substance as the permanent core of things. We have already
discussed this notion of substance' in § 4. 1, 4.3.
iSee Kaiyata's Mahäbhäsya-pradipa, commentary on Patanjali's Vyäkarana-
mahâbhasya (Gurukul Jhajjar, Haryana, India) 1963, vol. IV, p. 87.
The third attempt of Patanjaii to define substance is, perhaps,

more successful. A substances is a bundle of qualities, an integ-
ration of qualities.1 The implication is thatthe integrated whole
or the 'bundle' should be conceived as different from the con-
stituent parts. This is, thus, consistent with the original assertion
of Patanjaii that substance is something different from quality.
Nägesa quotes from Yogabhäsya, where Patanjaii is cited as
holding the view that substance is merely a collection of quali-
ties.2 It may sound like a philosophical mystification to assert
that a collection is different from its collected elements, but it is
also undeniable that each element, taken individually, is not
what a collection is meant to be. We should also note that the
Vaisesika notion of substance is^lightly different from that of
Patanjaii. In the Vaisesika system of categories, qualities, actions
and even generic properties (universals) are supposed to be
occurrent in substance, and the substance is said to be the sub-
stratum of all these properties. Not only that, but these proper-
ties are also tied to the substance by a real relation which is
called the relation of inherence (samaväya). Thus, what is called
'substance5 in the Vaisesika school is nota 'bundle' or collection
buta systematic whole having a theoretical structure of its own.
It is a real entity, and ex hypothesi not dependent upon qualities
etc. Rather, qualities etc. are said to be dependent upon sub-
stance for their existence. But, according to Patanjali's theory,
one may say that the notion of quality is primary. A collection
depends upon the collected elements, and hence a substance,
being a bundle of qualities, depends upon the latter. As far as
I can see, Kaiyata in his commentary has missed this subtle
distinction between the Vaisesika concept of substance and Patan-
*Cf. "anvartham khalvapi nirvacanam—gunasanrdrâvo dravyam ///"—Patan-

jaii under Pänini-sütra 5.1.119. Also : "guna-samudâyo dravyam"—Patafi-
jal i under Päninisutra 4.I.3.
2There is a striking similarity between Patanjali's definition of substance
and the Sämkhya-Yoga idea of substance. See Mallavädi's remark in his
Dvädasära-nayacakram (ed : Muni Jambuvijaya, Bhavnagar, India)
1966, p. 303. This point is also relevant to the problem of identity of the
two Patanjalis, the author of Mahäbhäsya and the author of Yoga-sütra.
See Surendranath Dasgupta : A History of Indian Philosophy (Cambridge)
1963, vol. I, p. 231-233. For a brief survey of the controversy over the
identity of two Patanjalis, see P.V. Kane : History of Dharmasastra
(Poona, India) 1962, vol. V, Part II, pp. 1395-99.
jali's notion of substance, and consequently, he has explained

Patanjali as expounding theVaisesika notion of substance. It may
also be noted that Patanjali wrote his Mahâbhâsya at a time when
the Vaiseska. theories were as yet on the threshold of strict syste-
matization. For this reason, care should be taken not to ascribe
too much of Vaisesika thought to Patanjali's writing.
Bhartrhari conceived substance as something dependent upon
the notion of quality. Substance is what is purported to be dis-
tinguished, and quality is the distinguishes On the authority of
Heläräja, we may say that this exposition of the notion of sub-
stance was in accordance with the philosophical view of Vyädi
(the grammarian first mentioned by Kätyäyana). Vyädi held that
words primarily refer to dra%a (substance or individual essence)
whereas Vâjapyâyana (another grammarian also first mentioned
by Kätyäyana) held that words refer to universal concepts. These
two apparently opposing views were much discussed by the
ancient Indian grammarians and philosophers.1 Patanjali elabora-
tely discussed this controversy under Pänini-sütra 1.2.64. In his
conciliatory remarks, Patanjali notes that the meaning of a
word has both aspects; it has a universal aspect, a descrip-
tive feature, an adjectival or attributive2 aspect, and it has a
particular aspect, a substantial aspect or a location. Thus, while
Väjapyäyana tends to emphasize the first aspect, Vyädi tends to
emphasize the second.3
The concept of substance, as expounded by Bhartrhari, is
much wider. In fact, anything that is expressible by a noun or a
substantive becomes a substance under his theory. Thus, if we
1
See M. Hiriyanna, "Vyâdi and Vâjapyâyana", Indian Historical Quarterly
(Calcutta) 1938, vol. 14, pp. 261ff. See also Gaurinath Sastri : The Philo-
sophy of word and Meaning (Calcutta) 1959, chap. VII, pp. 136-71.
*Guna 'attribute' or 'quality' includes, in a broader sense, äkrti 'universal'.
Compare Kaiyata's remark on Mahâbhâsya under Pänini-sütra 2.1.1:
"kecid äcäryä dravyant padärfham pratipannäh, kecid äkrtimj guna-sabden-
äpy äkrtir ucyate / yïathâ yasya gunasya bhäväd iti"—vol. II, p. 540. See
also Heläräja's remark: "jâterapipäratantryäddravye samaväyädgunatvam
uktanT—Comm. on Väkyapadiya, Book III, chap. V (Benares) 1905,
p. 145.
3Compare Mahâbhâsya under Pänini-sütra 1.2.64: "ubhayor ubhayam padär-
thah kasyacit tu kirncit pradhänabhütam kimcid gunabhûtam / äkrti-padär-
thikasya âkrtih pradhânabhûtâ, dravyant gunabhûtam j dravya-padârthikasya
dravyant pradhänabhütam äkrtir gunabhütä\"— (Gurukul Jhajjar, Haryana,
India) 1963, vol. II, p. 156.
use a substantive word to express a universal concept or a

generic property (jäti)> we should regard such a universal concept
as a substance (dravya) inasmuch as it is intended to be distingui-
shed by our*use of the word.1 Even the verbs may be supposed
to express some universal or other, and by nominalization they
can easily be shown to refer to a substance, in this extended
sense of the term. Thus, 'cooking' would refer to the fact of
cooking and 'walking' to the fact of walking as long as the spea-
ker intends to distinguish the act of cooking from the act of
walking. And in this sense, the fact of cooking becomes a dravya
'substance' since it is purported to be distinguished (bhedya).
In verse 1 of the fifth chapter of the third book (kända),
Bhartrhari proceeds to define guna 'quality' as a concept which
is entirely dependent upon the concept of substance that we
have just discussed. A quality is that which being intimately
related {saipsargi) to a substance differentiates that substance
by delimiting its scope. That a quality serves to differentiate is
reflected in such expressions as 'this is blue' and 'that is red'.
The universal gotva (cow-ness or cow universal) is considered
to be a quality, since it distinguishes cow-individuals from other
individuals. In fact, Bhartrhari rejects the rather technical
Vaisesika notion of quality and accepts the popular idea of a
quality. In popular parlance, we conceive of a quality as some-
thing dependent on another thing.2 A quality becomes a quality
only insofar asjt differentiates a substance; it does not become
a quality as far as we think of it as qualified by another quality,
nor does it become a quality when it is expressed as such by
hypostasis. This is the significance of Bhartrhari's expression
savyäpäram as explained by Heläräja. (Compare : 'savyäpärani
avacchedakarüpena pratiyamänam gunavyapadetyam na rüpän-
tarenetyarthah, p. 145).
The above verse of Bhartrhari is quoted by Kaiyata in his
Mahäbhäsya-Pradlpa (a commentary on Patanjali's Mahäbhäsyä)
under sûtra 2 l.l.Nâgesa, in his sub-commentary called Uddyota,
explains this verse in a slightly different way. He reads the first
two words of the verse as one compound word ^samsargi-
bhedakam" and explains that a quality is that which differ-
iSee Heläräja : "jätyädir api visesyatvena ced vivaksitah tadä dravyam"

Comm. on Vâkyapadïya, Book III, chap. IV (Benares) 1905, p. 142.
2See Heläräja : "loke hi paratantro guna iiy ucyate iti pratitir iyam" p. 192.
entiates its locus or substratum.1 'Savyäpäram' is explained by

Nägesa as follows : A quality is such that it invariably represents
its substratum and by presenting its substratum it enters into the
domain of meaning of the word which expresses that quality.
Alternatively, a quality is such that it invariably presents all the
individuals delimited by that quality.2 There is, in fact, no
fundamental difference between these two interpretations of
quality. However Heläräja's interpretation seems to be a little
more illuminating and useful.
The grammarians appear to have a commonsense approach
to the notion of quality. The nature of quality is to qualify other
things. A problem arises when we try to explain Pänini-sütra
5.1.119 tasya bhävas tva-talau. When the word sukla 'white5
means a white thing, the abstract suffix iva (comparable to
English "-ness") is added to this word to express the white colour,
a quality. And when 'white' {'sukla') means white colour,
a particular quality, the abstract suffix will be added to
express white-universal, a generic property, whiteness. But when
'white' ('sukla') means white-universal, the addition of the
abstract suffix tva is grammatically permissible, though it would
only refer back to the same meaning, viz., white-universal.
Similarly, when a proper name, say, 'Ditthd1 (a proper name),
is learnt for the first time, the addition of the abstract suffix is
grammatically permissible. But, semantically it will behave as a
personal name and will not express any quality, because no
quality öf the individual called 'Dittha' has been learnt.3 How-
ever, when one becomes familiar with the character and actions
of the individual called Dittha, one singles out his 'essential'
quality. And then, the addition of the abstract suffix will express
that essential quality. Furthermore, if somebody is described as
vaktr 'speaker', the implied sense of the word will be the quality
of speaking, and the addition of the abstract suffix Hva? to it
will express this implied sense. Sometimes, a certain relation is
expressed by the abstract suffix and hence becomes a quality
under this theory. Thus, the implied sense of the expression
iSee Nâgesa's Uddyota under Mahäbhäsya under sütra 2.1.1.

2See Nägesa, ibid.
»See Helârâja : "evam cädyäyäm sabda-pravrttau pitthädinäm api svarüpa
eva bhävapratyah." p. 193.
râja-purusa 'royal officer' is the property of being related to the

king and this property is what is expressed by the corresponding
abstract term nlja-purusa-tva 'royal-officer-ness'. We may change
the example as follows : If something is called 'the morning
star', the implied sense of this expression would be, as far as the
Sanskrit grammarians are concerned, the star's property of being
related to morning time—a quality which will be expressed by
the corresponding abstract expression.
It is obvious that the Vaisesika notion of substance and quality
is different from that of the grammarians. The Vaisesika notions
are more or less system-bound, whereas the grammarians' ap-
proach is very empirical. Elsewhere I have indicated how the
notion of substance and quality can be fruitfully usçd to formu-
late Sanskrit philosophical semantics. It is probable, therefore,
that while the grammarians were aiming at defining 'semantic'
categories, the Vaisesikas were concerned with 'ontology1. I shall
discuss this point further in the last section, § 5.5.
§ 5.3 : GRAMMATICAL CATEGORIES : A NAVYA-NYÂYA

APPRAISAL
The philosophical insight that can be drawn from the writings of

the Sanskrit grammarians seems to be considerable. It behoves
us, moderners, to critically examine and reconstruct their philo-
sophy of language and meaning, or what 1 have called their
philosophical semantics. Some detailed works have been done by
Dr. Radhika Herzberger (in her Toronto dissertation) and A.
Chakrabarti (in his Oxford dissertation). It is significant-to note
that the theories of the grammarians were not unnoticed by the
Nyâya writers, although the latter disagreed with the former on
many issues. In this section, I shall present briefly how a Navya-
nyâya writer, through an analysis of the Sanskrit language, con-
structs an alternative system of grammatical categories.
Jagadîsa belonged to the sixteenth century. He wrote several
commentaries, well-known Navya-nyâya texts and a short manual
of logic called Tarkämrta.1 His sub-commentary on Raghunâtha's
Tattvacintàmani'didhiti was the most important of them all. This
1
D,C. Bhattacharya is doubtful whether this book was originally written by
Jagadîsa. See his Bange Navya-nyäya-carcä, Calcutta, 1952, p. 168.
work is usually referred to by the scholars as Jägadlsi.1 Jagadïsa

wrote another important book, an original work called Sabda-
saktiprakäsikä.2 This book is also very popular among the tradi-
tional, orthodox scholars of India. The content of this book is
of great interest to both the Naiyäyikas (the logicians) and the
Vaiyäkaranas (the grammarians). Jn traditional seminaries (tols
or päthasäläs) of India, this book is often included in the syl-
labuses of the degree in Nyâya and the degree in Vyâkarana. I
shall give here an outline of the system of grammatical
categories noted in this important book.
In an Indian background, the close relationship between
grammar and logic (vyâkarana and nyàya) has already been
emphasized by me.
A considerable portion of what is known as Navya-nyâya in
India is concerned with what we may call the problems of
'logical grammar'. The task of this 'logical grammar' is to classify
the significant parts of sentences and to discover rules which
determine the notion of a 'meaningful sentence'. Sabdasaktipra-
käsikä is a book dealing with 'logical grammar' in this sense.
There may be some obvious limitations in our using 'logical
grammar' as a method in philosophy. There is the further limita-
tion of this book, since it is concerned only with the peculiarities
of the Sanskrit language. For the Sanskritists, however, Nyâya
offers an alternative approach to Sanskrit grammar and semantics.
Besides, Jagadîsa's account contains much material which deals
with what may be called the 'universals' of language.
Word (sabda) or language is another means through which we
obtain knowledge. It is different from both perception and in-
ference, according to Nyâya. It is calledsäbdabodha. We may
call it the'linguistic' or 'verbal' way of knowing. What we
know from linguistic utterances must be, in some sense, connected
with what is meant by such utterances.
Jagadïsa offers a strong defence of the Nyäya theory that
säbdabodha or our (hearers') knowledge of the 'meaning'of utter-
ed sentences is derived from a source different from that of
perception or inference.
It is derived from our knowledge of the syntactically connected
iChowkhamba Sanskrit Series 29, Benaras, 1906-1908.
2
Edited by Dhundhiraj Sastri with two commentaries, Kashi Sanskrit Series
109, Benares, 1934.
words (säkähksa-sabda). Usually, perception and inference are

regarded as the two main sources of our knowledge.
The knowfedge based upon authority can be reduced to some
form of a disguised inference. The Navya-nyäya school, however,
contends that when we listen to the utterance of a sentence and
immediately understand what is meant (at least literally), we do
not make any inference, nor do we perceive (through any 'mental5
eye) what is expressed by the sentence. Our knowledge about
the 'meaning5 of a sentence is qualificative or propositional in
the sense that one entity appears here as qualified by another.
The meanings of individual woids or word-components are
distinct, and hence stand in an isolated glory. In a sentence-
meaning on the other hand, they are modified in such a way as
to appear as connected and organized in a determinate structure.
Nyäya is, therefore, reluctant to accept the knowledge of this
connected sentence-meaning as arising on the basis of the cogni-
tion of some inferential mark {linga).
What takes the place of the inferential mark {linga) here is simply
an uttered sentence : "The hill has fire." The inferential know-
ledge, to be sure, follows from a premiss called paräinarsa,
which may be of either forms, "the hill has smoke which is
pervaded by fire," or "there is smoke, which is pervaded by fire,
on the hill," giving rise to either forms of inferential knowledge-
episode : "the hill has fire." or "there is fire on the hill." The
inferential situation itself cannot determine or regulate which
form the resulting knowledge would take. These are two different
knowledge-episodes, for, they have two different structures, al-
though their propositional content may mutually imply each
other (and hence be equivalent). -The uttered sentence, "the hill
has fire" would, however, yield a knowledge-episode with a
determinate structure. Hence, it is arguably distinct from an
inferential situation.
Since our knowledge of the sentence-meaning is derived from
what Nyäya calls meaningful words or significant word-compo-
nents, Jagadïsa tries to define the concept of 'significant word'
{sarthaka-sabda). A word is to be considered significant if it can
generate in the hearer the cognition of what is meant only in
collaboration with another syntactically related word. Jagadlsa's
problem here is to distinguish the concept of 'significant word*
from the concept of 'significant sentence'. A significant sentence
is, according to Nyäya, a cluster of syntactically connected words

that have a combined meaning of which a part must be an in-
flectional meaning.1 Thus, even an expression like "ghatam" is
considered to be a sentence (väkya) because it may generate
cognition of its meaning with the help of an inflectional
meaning. But the stem "ghata-i*pot)2" is a word because
it cannot generate cognition of its meaning without being in
collaboration with some inflection like "-anC (the accusative).
Although the meaning of a sentence like "ghatam" is not lingu-
istically complete, it can be said, from the Nyâya point of view,
to be logically complete. Logically, a propositional cognition
needs only a qualifier to qualify a qualificand. The said
sentence, even when it is unaided by other words, may
, generate a cognition of the form 'karmatvam ghatiyam' (the
accusative function belongs to the pot), in other words, the
generated cognition will construe the pot (meaning of the stem)
as the qualifier of the accusative relation (meaning of the inflec-
tion). And thus, we have an 'atomic' but logically complete 'pro-
positional' cognition having a qualifier-qualilicand structure.
Jagadisa further contends that even an incompatible compound-
word like "sasa-visäna-(*rabbit-horn)" is a significant word, inas-
much as it can generate a false cognition of the horns being
related to some rabbit. (See also § 5.5.) Since, only with the help
of some inflection, will such a compound be usable in language,
we cannot, however, call this compound-base a sentence. For
the combined meaning will not have any inflectional meaning as
its part.
Jagadisa classifies the significant word-components into three
types : the inflectable word-base.3 or the lexical items (prakrti),
inflections (pratyaya), and the invariant words (nipäta). The
prakrti or the inflectable word-bases like "ghata (*pot)" and
"<?#/-( *move)" enjoy some sort of primacy over the inflections
{pratyaya) like 4 t -s" (the nominative) and "-*/" (the present
singular indicative) in the sense that, in order to merge the
iOn the notion of vâkya, see following section § 5.4.

2Since the English equivalent of a Sanskrit word-base will be indistinguish-
able from that of Sanskrit inflected word, 1 shall mark the English equi-
valents of Sanskrit word-bases with asterisks.
3To note one terminological quirk : a word-base is, in fact, a kind of word
or Sarthaka-sabda.
'meaning'of a word-base (viz., the object 'pot' or the action

moving) Into the sentence-meaning, we need to have a decision
(niscayci) of the actual word-base itself ("ghata-" or "cal-")
being associated with some inflection. A word-base contributes
its meaning to the sentence-meaning in an independent way, while
an inflection contributes its meaning in a dependent way. An
invariant word like "ca (*and)" or "va (*or)" also enjoys some
independence in this matter. But there is another syntactic feature
which distinguishes an invariant word from a word-base.
Word-bases are of two types : nominal base or stem (nâma)
and verbal base or root (dhätu). It is contended that a nominal
base like "ghata-(* pot)" or a verbal base like "«//-(* move)" can
have an identity of reference with another word-base like
"nila (*b!ue)" or "/nrrf«-(*slowly)" respectively. The implicit
semantic principle is that the adjective and the adverb present
or designate the same entity as is presented or designated by the
noun and the verb they modify respectively. But an invariant
word lacks this privilege. It is said that an invariant word can-
not relate its meaning by identity-relation to the meaning of any
other word-component in the sentence.
A nominal base is also called präiipadika. Jagadîsa resorts to
some complicated device in order to mark off the difference
between the concept of a nominal base and that of a verbal
base. A nominal base needs the assistance of the nominative
terminal inflection in order to generate a cognitive meaning with
its own meaning as the chief qualificand (mukhya-visesya).1 A
verbal base needs the assistance of an 'agent' inflection "-trc"
(resembling the English- "er") in order to generate a cognitive
meaning with its own meaning as the qualifier.
Nominal bases are divided into four groups from the point
of view of their semantic function. One is called rücfha, viz.,
the class of words that express only conventional meaning in a
sentence. Word-bases like Cigau- (*cow)"5 in the sentence "gaur
asti (there is a cow)", fall under this class. The second group
is called laksaka. Words of this class convey metaphorical or
secondary meaning in a sentence. Nominal bases like "gahga-
(*Ganges)" in the sentence '"gahgäyäm ghosab" (a village on the
Ganges ; compare in English "New Castie-upon-Ty ne") fall in this
class, because "ganga* conventionally means a particular river,
i$ee The Navya-nyâya doctrine of Negation, p. 15. r
while in this sentence the same word has to be interpreted as

meaning the riverside. The third group is called yogarüdha.
Words of this class agree with the derivative sense, i.e., with the
combined meaning of the sub-components, but at the same
time express, by convention, only a special object which may
agree with the derivative sense. A word like "pahkaja- (*born-
in-mud)" should have expressed anything that is born in mud,
but, conventionally, its meaning is so fixed that it expresses
only the lotus which happens to be born in mud. The reason
for narrowing down of the scope of reference in this way may
be found in the forgotten history of the language development.
The fourth group is called yaugika. Words of this class retain
their derivative sense, i.e., agree fully with the combined mean-
ing of their sub-components. Thus, "päthaka-^ reader)" means
anybody who reads. It may be noted that the composite word-
base "päthaka- (*reader)" is not to be regarded as a sentence,
but when an inflection, viz., a case-termination, is added it
becomes a sentence in the Nyäya sense of the term. (The
inflectional meaning must form part of the combined meaning
of a sentence.)
À word of the first group, i.e., a conventionally meaningful
nominal base (rüdha-näma)> is also called a samjhâ. These
nominal bases have denotative power (sakti) in the primary
sense. They can refer to the individuals through some attributes
or universals. The grammarians like Patanjali classified these
words into four types. There are some that signify class-charac-
ters or generic properties (jäti) as attributes. The word c 'gau-
(*cow)," for example, refers to cow-individuals through the
generic property cow-ness (gotva). The grammarians say that
"gau-(*cow)" is gotva-väcaka, i.e., it expresses, (designates,
signifies or represents) cow-ness. There are other words which
signify dravya, i.e., (material) substance. For example, the word
"ädihya- (^wealthy)" signifies wealth or money, which is a
substance. The third class of words signifies qualities, e.g.,
"nlla- (*blue)" which signifies blue colour. The fourth class of
words signifies action (kriya), e.g., "cala- (*moving)" which
signifies motion. To put it in another way, we may say that
u
gau- (*cow)" picks out an individual which has cow-ness (a
jäti) while "ädhya (^wealthy)" picks out an individual which
possesses wealth or money (a dravya, substance) and "nïla-•
(*bîue)" picks out an individual which has blue colour (a quality)

and "cala- (*moving)" picks out an individual which possesses
movement (kriyä). Jagadïsa points out that this fourfold classi-
fication was also endorsed by Dandin, the Älankärika.
This fourfold classification of the grammarians is rejected by
Jagadïsa, because, as he says, it fails to make room for such
words as "müka- (*mute)," which signifies neither a generic
character, nor a quality, nor an action, nor a substance. This
word picks out an individual which is characterized by the
absence of speech. Thus, we may say that it signifies an absence
as the distinguishing attribute of the individual it picks out. We
have already noted that the Nyäya school regards the absence of
anything as an attribute much in the same sense as the so-
called positive attributes, such as, cow-ness and blue colour
(§§2.3,2.6).
Jagadïéa favours a threefold classification from the point of
view of a different principle of division: i. naimittikl sarhjnä,
i.e., words or nominal bases which pick out objects by attri-
buting to them some generic character or other; 2. pärihhä§iki
saritjnâ, i.e.., words which pick out objects by attributing some
singular property to them; 3. aupädhikl samjna, i.e., words
which pick out objects by attributing some imposed condition
or property (upädhi). Since Nyäya believes that some uni versais
like cow-ness constitute real generic properties apart from the
individuals where they are instantiated, words like "gau-
(*cow)" are said to attribute cow-ness etc., to the individuals
they describe. A singular name like "gagana- (*the sky) " be-
longs to the second group, because it attributes a singular
property to the individual. Other common names like "düta-
(*messenger)"belong to the third group, because individuals
denoted by them are grouped together by some imposed condition
such as, the property of carrying message. Incidentally, this shows
that Navya-nyäya does not regard all class-characters as real
universals. This, in fact, corresponds to the division of terms
into singular and general, with the proviso that the meanings of
some general terms (natural kinds) are to be construed as
(real) universals, while those of others are only provisional and
constructible properties (cf., upâdhi=* imposed property).
Under yaugika words, Jagadïsa elaborately discusses various
types of compounds (samäsa) and derivative words (krdanta and
taddhitänta). He also observes that there is another way of divid-

ing the nominal bases {näma). From the point of view of
grammatical gender, we may divide them into three : the mas-
culine, the feminine and the neuter. Here he refers to the inter-
esting grammatical theory that gender-suffixes form an integral
part of the nominal bases.
Turning to verbal base, Jagadisa classifies them into three:
1. ganokta or atomic roots, i.e., those that are enumerated in
the list of roots given by the grammarians1. 2. yaugika derivative
roots, i.e., those that are formed from atomic roots and inflec-
tions; 3. the ''sautra' roots or those that are either mentioned
in the grammatical rules of Pânini or current in language as
roots but not listed. This third group evidently accounts for those
roots which are new roots (like "hindol-** and "ändola-") added
to the lexicon as the language grows through history. Incident-
ally, Jagadisa seems to be indifferent to the fact that even a
verbal base can have a contextual meaning, i.e., metaphorical or
secondary meaning depending upon the context or linguistic
environment, which is not the same as its primary meaning.
Probably he accepts the maxim that any root can have a mul-
titude of meanings (cf., anekürthä hi dhâtavah), one or the
other of which becomes relevant to the context. Or, perhaps,
he thinks that it is more suitable to talk about the contextual
meaning of the nominal bases formed from roots rather than
talking about the contextual meaning of roots themselves
Inflections (pratyaya) are of four types: vibhakti (terminal
inflection or ending), dhätvamsa (inflection that forms derivative
roots from atomic roots, such as '"-san-" or the desiderative
suffix), taddhita (inflections forming new nominal bases from
atomic nominal bases), and krt (inflections forming word-
bases from atomic roots). The meanings of inflections are, in
general, related to the meanings of the word-bases, to which they
are added, by the subject-predicate (iiddesya-vidheya) structure.
In other words, the meaning of the word-base (such as, a
pot) qualifies the meaning of the terminal inflection (such as,
the accusative relation). A vibhakti or terminal inflection can
be either a sup (that which is added to a nominal base) or a
tin (that which is added to a verbal base). In this connection,
iCf., Dhätupätha; see O. Bohtliugk, Pâninis Grammatik, Leipzig, pp. 61-84.
Jagadisa defines and describes in detail different case-relations

as well a» the semantic significance of each tense-suffix and
mood-suffix. Jagadisa's discussion may very well be compared
with that of Bhartrhari in the third Kârida of his Vâkyapadiya.
Jagadisa has sometimes suggested some improvement of the
prevalent Nyâya theories. For example, he is critical of Gan-
gesa's definition of Laksanä (metaphorical meaning-function)1
and gives his own improved version of it.2 As a Naiyâyika, he
frequently criticizes the Prâbhàkani views. He refers to Pat-
aîïjàli, the author of the Mahàbhâsya as an authority, in some
places, although he rejects Patanjali's fourfold classification of
the conventionally meaningful nominal bases. He often quotes
from Bhartrhari. There is one particular verse which he quotes
and ascribes to Bhartrhari, but, which is, in fact of doubtful
authenticity.3 It may be that Jagadisa did not have direct
acquaintance with the complete work of Bhartrhari.
Jagadisa's classification is very systematic and quite helpful
for the description of the Sanskrit language. He carefully avoids
cross-division or wrong classification. He tries to scrupulously
define each concept so that each sub-class will be distinct and
well-defined. Since he tries to logically define the basic cate-
gories which we generally grasp through intuitive notion, his
definitions tend to be highly complicated. But this does not
detract from the merit of his contribution to "logical grammar".
Ï like to present here Jagadisa's taxonomy of grammatical
categories as understood by the following tree:
1
See Tattvadntamani, Edited by Pt. Kamakhyanath Tarkavagisa, Calcutta,
1884-1901, vol. IV, pp. 66 off.
2See Sabdasaktiprakâsikâ, pp. 136-39.
3lbid.,p. 389.
Sâbdabodha from the Sentence (väkva)
Sarthaka-kabda
(significant word)
i 1 I
prakrti pratyaya nipäta
word base (affix) invariant word
i
i Krt taddhita dhâtvatpéa vibhakti
i (infix)
I
1 I
sup tin
i dhätu
nâma (verbal base)
(nominal base) i
1
ganokta yaugika sautra
rudha laksaka yogarüdha yaugika

I
sarhjnä
I _
1 I I
naimittika pâribhâsika aupâdhika
(general terms) (singular terms (general terms)
or
proper names)
§5. 4 : THE NOTION OF THE SENTENCE (VÄKYA)
What is a sentence (väkya) ? The question is relevant to linguists

but also to philosophers, especially to those philosophers who
work on the borderlines of logic and language. The Sanskrit
grammarians propounded theories about the nature of the sen-
tence (väkya) and disputed with the Naiyäyikas (and philosophers
sympathetic to the Nyâya system), who held opposite views. I
shall discuss those divergent views and explain some important
concepts used in their analysis.
In general, the philosophers of ancient and mediaeval India
chose Sanskrit as their vehicle of expression. Thus, their views,
especially those about logic and the philosophy of language,

were, to a considerable extent, influenced by the peculiarities of
the Sanskrit language. Some of their views can be better under-
stood and explained by relating them to the structure and nature
of Sanskrit. This should not be regarded as a drawback because
one may very well try to reconstruct a general theory of language
from their specific remarks. Moreover, it may be interesting and
sometimes important for philosophers of language to notice how
difference in structure, idiom and vocabulary may eventually
cause a difference in the corresponding philosophical theory. It
goes without saying that the existence of many natural languages
with radically different structural and logical characteristics offers
opportunities for logical exploration of 'ways of thinking' far
more diverse than those found in any one of them singly. We
have seen an illustration of this general point in the previous
section. The present section will supply another illustration of
the same point.
Kätyäyana defined sentence ( väkya) as follows : a sentence is
that [cluster of words] which possesses a finite verb [as an ele-
ment]1 The expression used in this definition (viz., eka-tin) is a
Sanskrit compound consisting of two elements (eka = one, tin =
[word with] verbal ending). This expression suffers from some
ambiguity which is shared by most Sanskrit compounds. It might
be interpreted as a tatpurusa (i.e., karmadhâraya) compound, in
which case the meaning would be that a sentence is what con-
sists of one finite verb. Such an interpretation will, however, be
inappropriate, and hence, should be rejected, since it would only
include expressions like paeati ("...cooks" or "...is cooking")
under the class of sentence. On the other hand, if the expression
eka-tin is interpreted as a bahuvrihi compound, the meaning
would be that a sentence is that (cluster of words) which contains
one finite verb as an element. The second interpretation is appro-
leka-tin vâkyam; this is a vârttika of Kätyäyana. See Mahàbhàsya under

Sütra 2.1.1. A close study shows that there is a difference between the views
of Panini and Kätyäyana regarding the definition of sentence; the former
could think of mo re than one tin in a sentence (cf. tinn atiiiah, 8.1.28)
whereas Kätyäyana took eka-tin as a sentence. The distinction between
this and the Mimämsä definition based on äkänksä (cf. arthaikyâd ekam
väkyant sakänksani cedvibhäge syät : Jaimini-sütra 2.1.46) was discussed
by Bhartrhari and his commentators. See Vâkyapadîya II, 3-4.
priate, and hence, has been accepted by the commentators. But

this also might lead to difficulties because one might ask whether
such expressions as pasya mrgo dhövati ("look, a deer runs")
etc. which contain more than one finite verb should be regarded
as one sentence or several sentences put together, or no sentence
at all.
This problem, however, is connected with the notion of single-
ness of sentence or sentence-unity (ekavâkyaîâ)—a concept fre-
quently used and discussed by Indian philosophers. 1 shall return
to this point later.
Kâtyâyana's definition was, perhaps, intended to emphasize
the importance of the function of finite verbs in each sentence.
Thus, Patanjali remarked that there is no sentence which lacks a
finite verb. According to this interpretation, containing a finite
verb is a necessary condition for being a sentence.
Our point here is that the early Indian grammarians attempted
to define the notion of the sentence empirically, using what may
be called a formal criterion and without ostensibly referring to
the meaning-content of the sentence. It may be remarked,
however, that, perhaps in order to combat different philosophi-
cal theories of language, some later grammarians, even of the
Pâiiinian school, not only brought in semantic notions but also
used metaphysical concepts in explaining and defining grammati-
cal categories. Thus, in India, grammarians and philosophers
created a common ground for discussing issues of common
interest.
It is clear, however, that the Indian grammarians used the
notion of word (pada) in defining the notion of sentence. One
might say that, in order to define the notion of word it would be
necessary to take semantic criteria into account. But, in a highly
inflected language like Sanskrit, this need not be so. Panini
(c\ fourth century B.C.) defined pada or word as that which has
either a verbal inflection or a nominal inflection2, and, thus
apparently, used a formal criterion. Gautama, the propounder
ofthe Nyäya school of philosophy, also supported this definition
l
na hi kriyâ-vininnuktaiti vcïkyain asti : Mahäbhäsycu ibid.
^suptinantam padam : Pänini-sütra 1.4.14. It may be noted here that Pänini
also gave several definitions of pada in subsequent sûtras, e.g. svädisv
asarvanâmasthânesu : 1.4.17. But these were required for specific purposes,
e.g. for contrasting pada with another technical term.
and asserted that a pada or word is any sound-sequence which

has inflection (yibhakti)} As is clear from the previous section,
in Sanskrit each constituent element of a sentence takes either a
verbal, i.e., conjugational inflection, in which case it becomes
what we call a finite verb, or a nominal, i.e., declensional
inflection, in which case it is said to belong to the non-verbal,
i.e., nâma, category, which includes substantives, adjectives,
verbal adjectives, participles, and even adverbs. Although
prepositions, e.g., upasarga, nipâta etc., do not seem to have
declensional inflections, theoretically they are supposed to
possess zero-occurrences of nominal inflection. The Sanskrit
name for them is avyaya, which is interpreted as that which
undergoes no modification regardless of gender, case and
number.2 I have called them invariant words in the previous
section. Hence, a formal definition of pada or word can some-
how be obtained within Pânini's theory of grammar.
The later Naiyâyikas suggested the useof oniy semantic criteria
in order to determine what is pada. Contrary to what even
Gautama said, they defined pada or word as the meaning-bear-
ing element of a sentence.3 An "atomic" pada, according to them
is the smallest meaning-bearing phoneme sequence. A "complex"
pada is formed out of several "atomic" words or padas. Accord-
ing to this theory, even an affix or suffix should be called a pada
or word, provided one can assign some significance to it. Thus,
pacati ( = "cooks" or "is cooking") is, according to them actually
formed out of two constituent padas, a root, pac-a, meaning the
operation of cooking (päka) and a suffix,-*/, signifying prayatna,
i.e., the mental (us opposed to physical) effort of the person
concerned. We may note that pacati was regarded by them as a
sentence (väkya), rather than a pada.
Amarasiipha (c. fourth-fifth century A.D.), a well-known
lexicographer > gave the following lexical definition of sentence :
Ue vibhaktyantäh padam : Nyùyasûtra 2.2.60.
2The verse which is often quoted to describe avyaya is
sadràaty trhii linge fit sarväsu ca vibhaktisuj
vacanesu ca sarvesu yan na vyeti tad avyayamlj
"that which is similar in three genders and undergoes no change in any
case-terminations and in any number-suffixes, is called avyaya."
Zsaktarfi padam : Annarnbhatta, Tarkasamgraha (A. Foucher. Le Com-
pendium des topiques, Paris, 1949, p. 152).
a sentence is a cluster of verbal and non-verbal words.1 One may

wonder whether a cluster of only finite verbs or nominal words,
e.g.,gaur asvah puruso hasti ("acow a horse a man an elephant"),
constitutes a sentence. This objection can be met, if one wishes,
by a careful formulation of the definition. It differs from Kätyä-
yana's definition only in that it does not stress that a finite verb
must be present in a sentence to make it a sentence. By a slight
shift of emphasis one might as well say that any cluster of words
(no matter which grammatical category they belong to) may
constitute a sentence. Some of the later Naiyäyikas did, in fact,
accept such a definition of sentence.2 In such a definition of
sentence, apparently, neither semantic nor syntactic considera-
tions have any part to play. We have already noted it in § 5.3.
As I have remarked before, the notion of word, according to
the later Naiyäyikas, depends on semantic criteria. This was not
the case in Amarasimha's definition.
The important point, then, in which the grammarians differed
from the Naiyäyikas, is this :The latter did not, while the former
did, think that the verbal element, i.e., the finite verb, is essen-
tial for constituting a proper sentence. This requires clarification,
particularly in a language like Sanskrit, where the verb 'to be*
is seldom used in normal categorical sentences, words in
which can be purely nominal. If we place nouns (i.e.,
substantives) and adjectives side by side, a Sanskrit sentence
(grammatically correct as well as idiomatic) will result, e.g.,
ghato nllah "the pot (is) blue"
naro 'yaiji na sundarah "this man (is) not handsome".
Nominalizing transformations by using convenient verbal adjec-
tives, adjectival phrases, word-compounds, etc., which are formed
directly or indirectly from some verb or other, are so common
in good Sanskrit as to make the use of a finite verb redundant
in many contexts. Thus, we have
rämo vaktä "Räma (is) the speaker" = "Rama speaks"
devadattah kr$na-sritah "Devadatta (is) the one who has resor-
ted to Krsna" = "Devadatta has resorted to Krsna"
sa bhäravähi "he (is) a weight-carrier" = "he carries weight".
isup-tin-ajtta~cayo väkyam : Amarakow, s.v.
tväkyaifi pada-samühah : Annaipbhatta, op. cit.
Such prevailing features of the Sanskrit language may have led

the Nalyâyikas in India to think that a finite verb is not essential
to a normal sentence.1
There are numerous examples of Sanskrit sentences where a
finite verb does not appear at ail. To explain such cases, the
grammarians appealed to their notion of zero-occurrence of
grammatical elements. According to them, a finite verb ('to be',
for instance) is not only to be understood in such contexts but is
to be regarded as constituting the chief element when the mean-
ing of the whole sentence is considered. The Naiyäyikas offered
a different method of analysis. According to them, to imagine
a finite verb (viz. 'to be') to be always understood in a good
many sentences is an unrewardingly complex procedure, springing
from a mistaken idea about the function of the finite verb in a
sentence. If the juxtaposition of even several non-verbal words
is sufficient to make an assertion and constitutes what is called
a grammatical sentence, it would be wrong, so the Naiyäyikas
argued, to drag in a finite verb to perform a task which is no
longer required. They also cited some peculiar examples where
it would apparently be impossible to add any particular finite
verb to the sentence. For example,
(1) trayah käläh ''three the time-stages viz., past, present and
future)".
Here, according to the Naiyäyikas, it would be incorrect in Sans-
krit to postulate the present plural form of the verb as ( = "to
be"), because this can be correctly applied to one time-stage
only, viz. the present, and not to the other two.2 Similarly, for
the future plural form or the past plural form. To drag in three
singular finite verbs, one in the past, one in the present, and
iCf. J.F. Staal, 'Reification, Quotation and Nominalization.' in : Logic and

Philosophy : Essays in honour of IM. Bochenski, Amsterdam 1965, pp.
151-87
2
cf. Nyâyakoêa, Bhimacharya Jhalkikar, Poona, Bhandarkar Oriental
Research Institute, 1928, p. 876 line 16-23. Although according to the
Naiyäyikas example (1). is perfectly in order, and not just an elliptical
sentence where we have to supply the omitted words for the completion
of the sense. The present tense is also used to denote the non-temporal;
see e.g. J.F. Staal, -Philosophy and Language', in Essays in Philosophy
presented to Dr. Mahadevan, Madras 1962, p. 10-25. It could, therefore,
be argued that only the present tense form need be supplied.
one in the future (including one ca "and"), could only be

prompted by an obsession for finite verbs. Some grammarians
have defended their theory by saying that in example (1) one
should supply the finite verb jnäyante ("are known"), i.e., the
present plural passive form of the root jhä ("to know"). But,
according to the Naiyâyikas, this defence was based on a mis-
conception of the significance of counter-example (1) Counter-
example (1) derives its force in the context, because it h used as
an answer to the question kati käläh{"how many time-stages [are
there] ?"). Hence, the rejoinder misses the point. (It is only for
the sake of English that I am adding "are there" in brackets.)
The grammarians seem to have thought of their definiendum
as the set of sentences that are grammatically acceptable.
Whether this included semantically unacceptable sentences is not
clear from the definitions we have so far considered. When,
however, proper attention was given to this problem, the gram-
marians needed other criteria. Judging from the remarks of some
Naiyâyikas, on the other hand, their notion of sentence seems to
include not only the set of grammatical sentences, whether seman-
tically acceptable or not, but also the set of sentences that are
both ungrammatical and meaningless. This is very lavish theory,
and its usefulness may be questioned. This was the view of some
Naiyâyikas. They also divided sentences into two groups, viz.,
pramciria vakya or 'acceptable' sentences and a-pramana väkya
or 'non-acceptable' szntçncçs/ThQ pramcina sentences are mean-
ingful as well as grammatically correct; the a-pramäna sentences
cover the rest, which consists of sentences either ungrammtical',
or semantically unacceptable, or both. 1
The Naiyâyikas used specific criteria to decide whether a
cluster of words or a sentence will be both grammatically and
semantically acceptable, or, to use their own terminology, to
decide whether a sentence will generate a cognitive meaning or
cognition (cf. säbdabodha) in an "ideal" hearer. For a sentence
^Compare : vâkyafft dvividhani pramäna-väkyam a-pramäna-väkyani cetij

tatrapramänaväkyam akänksa-yogyatä-s'ariimdhiniatätfi padänäni samühahj
...a-pramuna-vakyani tu âkcmksâdi-rahitaiti vâkyamj: Nyâyakosa p. 730.
Here a pramana sentence is slid to \fct a duster of words which possesses
äkät'ik>ä,yogyutä,nml äsattl, whereas an a-prœnâna sentence is said to lack
either any one or any two or all three of these, lu the next section,
however I shall suggest a different classification.
to generate a cognitive meaning, it is necessary that the ideal

hearer should have (a) a cognition of what is called yogyatâ or
semantical competency, (b) a cognition of what is called äkänkfä
or expectancy, (c) a cognition of äsatii or contiguity (in space
and time), and perhaps also (d) a cognition of tätparya or
speaker's intention. The fourth condition is particularly needed
in order to disambiguate an otherwise ambiguous expression.
All these concepts have a long history of development. They
were redefined and modified by many authors at different times.
There was a theory, favoured by some grammarians and crfcun-
kärikas1 as well as by some Naiyäyikas, which maintained that
a sentence should be defined as that cluster of words which
possess the three properties yogyatä (semantical competency),
äkähk?ä (syntactic expectancy) and äsatti (contiguity in space
and time).2 A comparatively late view among the Naiyäyikas
was that these properties, viz., yogyatä, etc., were themselves
directly responsible for generating säbdabodha, i.e., for the know-
ledge of the meaning of a sentence. 3 1 shall not expound each of
these concepts here in detail4 but briefly refer to them in connec-
tion with the notion of meaningful and grammatical sentences.
The concept of yogyatä has sometimes been defined as the com-
patibility of one object (or, rather the absence of the absurdity or
incompatibility in one object) with another object in accordance
with the syntactical (grammatical) connection of the respective
words denoting those objects. What was meant by this may
become clear from examples. It was argued that the sentence
(2) vahnind sincati "(he) sprinkles (the field) with fire"
U.e. rhetoricians, such as, ViSvanätha and Mammatha.

zCf.väkyarjitv äkänksä-yogyata-sajjinidhimatärn padänäin samühah : Kesa-
vamiéra, Tarkabhâsâ, Poona, Oriental Book Agency, 1953, p. 16.
Ûsually two different views are mentioned in this connection. The older
view is : we know the meaning of a sentence, when we have already
known such properties like expectancy (äkänksä) etc. possessed by the
sentence. The later view is : we know the meaning of a sentence if
that sentence possesses such properties, viz., expectancy etc. Compare :
itah sarve svarüpasantah säbdabodhe hetavah na tu jnätä iti jneyam ; Jäna-
kinätha, Nyäyasiddhäntumanjari, IV, Benares, 1885. CvIt is to be acknow-
ledged that all these are causes of sabdabodha (directly) by themselves,
(and) not (that they become causes only when they) are known.")
4
They have been discussed in detail by K. Kunjunni Raja in his Indian
Theories of Meaning, Madras, Adyar Library Series 91, 1963, pp. 151-87.
406 Logic, Language and Reaiity
lacks yogyatä, while the sentence

(3) jalena sincati "(he) sprinkles (the field) with water'',
, possesses yogyatä. It may be noted that both (2) and (3) are
grammatically acceptable, but while (3) is also semantically
acceptable, (2) is not. So yogyatä, semantical competency, is what
differentiates a grammatically acceptable but semantically non-
acceptable sentence from a semantically acceptable one. In (2),
since fire cannot be said to be a. fit object to sprinkle with, fire
is said to be 'incompatible' with the activity of sprinkling. That
sprinkling fs to be so related to fire is what is intended by the
instrumental construction. Thus, although grammar allows (2),
säbdabodha or knowledge of the meaning of the sentence will
not be generated by it because it lacks the property yogyatä.
Expectancy or äkähksä may originally have meant the desire
or expectation on the part of the listener roused by the incom-
pleteness of an utterance. But in grammar and philosophy it
gradually came to be identified with the syntactic property
which a sentence lacks when it is not 'grammatical.' GangeSa
defined it as the accompaniment of one string x with another
string y i n such a way that x would not generate congnition of
the meaning {säbdabodha or anvayabodha), unless accompanied
by y1. The Sanskrit example.
(4) ghatam änaya "bring a pitcher"
is said to possess the property expectancy, because the verb
änaya (''bring") is accompanied by an accusative or karma viz.,
ghatam (a pitcher"), the agent (kartr) or vocative being under-
stood as usual (i.e., because the utterance is in the imperative
mood). Furthermore, the string ghata (the nominal stem of
'pitcher') is 'grammatically' acceptable in (4), only because it is
associated with the accusative ending -am. If this ending was
not used and a word expressing the relation of accusative or
karma was used instead, other things remaining the same, the
whole string would be regarded as disconnected and ungram-
matical. The counter-example which is said to lack expectancy
is given as :
iCf. yasya yena vinä svärthanvayänanubhavakatvarn tasya tatpada-sarnni-

dhänam : Tattvacintämani IV, Asiatic Society, Calcutta.
(5) ghatah karmatvam änayanam krtih "a pitcher, to be an

accusative, bringing, an effort35.
This is un grammatical, since it violates syntactic rules of the
language. But is it a meaningful expression? It is contended
that knowledge of a ''connected" meaning of the whole is
not possible here, because the discrete elements do not have
any syntactic relation to each other.
The property äsatti or contiguity refers to absence of any
unnecessary intervention or interval {temporal when the string
is uttered, and spatial when it is written) between word-elements
of a sentence. As a counter-example which lacks this property
we sometimes find:
(6) girir bhukto vahnimän Devadattah "the hill has eaten
has fire Devadatta".
This sounds nonsensical even in Sanskrit where there do not
appear to be strict rules about word order. One may construe
this string as two sentences, viz., girir vahnimän ( = "the hill has
fire") and bhukto Devadattah ( — ''Devadatta has eaten"). But
as a single sentence it fails to generate any knowledge of its
meaning.
It is evident from the brief exposition of these three notions,
viz., semantical competency, expectancy and contiguity, that
they are among the important properties of a grammatical and
meaningful sentence. In other words, they turn an ungrammati-
cal and nonsensical string into a grammatical sentence in such
a way that we may know what it means. But, if sentence or
väkya is taken in the general sense of any word-complex, as
some Naiyâyikas obviously intended, these properties do not form
part of the definition of the sentence. (For problems of such a
definition, see § 2.8.)
A sentence is significant or meaningful if it can generate know-
ledge in a hearer who is a native language-user, whenever he listens
to it. We can conceive of an 'ideal' hearer who knows the
language and also knows 'how to do things with' language, and
who reacts 'rationally' and 'mechanically' when he hears a
grammatically correct and semanticaliy coherent sentence. This
'rational' and 'mechanical' reaction, according to the Indian
theorists, is produced by the utterance generating a particular
direct cognition (jnäna) in the hearer, very much as blowing a
horn produces a particular vibration in the air. This theory also

assumes that the hearer is reasonably attentive, i.e., not pre-
occupied, and that no other negative (non-linguistic) condition
is present so as to stop the required cognition from being gener-
ated. As said before, the Indian theorists called this cognition
a säbdabodha of the sentence concerned. That a hearer knows what
an utterance means signals that the utterance has generated such a
säbdabodha. This cognition or knowledge (i.e., säbdabodha) Is
a result of the utterance, and hence, should not be confused with
the speaker's cognition which is the speaker's private property
and which might have prompted the utterance originally.
To describe the content of a säbdabodha may be said to be
equivalent to describing the "meaning" of the utterance. A
säbdabodha, i.e., knowledge of some sentence-meaning, belongs
to the type which the Naiyâyikas call savikalpajnäna or quali-
ficative cognition1. A qualificative cognition is always expressible
in language in some way or the other. Thus, the content of a
säbdabodha can always be expressed in the language. We may
call the expression of the content of a particular säbdabodha a
'paraphrase5 of the utterance concerned. But note that a 'para-
phrase', in the sense we are here concerned with, is not exactly
equivalent to supplying synonyms for each constituent word
and preserving the grammatical structure of the original as far
as practicable. Roughly, we can characterize the original
utterance as an utterance or sentence in the object-language,
and its 'paraphrase', i.e., the expression of its säbdabodha^ as
the description of the same utterance in a suitably chosen,
corresponding metalanguage. To provide a structural descrip-
tion of the sentences of the object-language, certain metaling-
uistic concepts were developed by the Indian theorists. A rough
characterization of these concepts may be in order.
The content of the basic type of qualificative cognition (savi-
kalpajnäna) is analysed chiefly under two categories, the quali-
ficand (visesya) and the qualifier (visesana or prakära). A
cognition of this type can be roughly described as knowing
iThe expression savikalpajnäna has sometimes been translated as "deter-
minate cognition" which I do not consider very illuminating. I have tried
to explain the notion in some detail, translating it as "qualificative
cognition." See my The Navya-nyäya Doctrine of Negation, Cambridge
(Mass.'), 1968, p. 15.
something (i.e., the qualificand) as something (i.e., the qualifier).

Thus, in the cognition
(7) raktam puspam "the red flower" or "the flower [is] red"
a flower is the qualificand and the colour red is its qualifier.
Usually (as in the present case) the qualificand is a thing and
the qualifier is one of its properties which the cognition happens
to mention; but this is not always the case. For example, in
the congnition
(8) puspe raktah "red-colour (occurs) in the {lower"
the colour red is the qualificand and a property which may be
expressed as "occurrence-in-the-flower" (puspa-vrttitva) is the
qualifier. For, in an extended sense of 'property', to,occur in a
particular flower may be treated as a property of the red colour.
Consider the following cognition:
(9) rakta-puspavatï latä "the creeper possesses (a) red
flower"
Here, broadly, the qualificand is a particular creeper, and the
qualifier is a particular red flower. But 4a red flower' also
denotes a composite concept analysable into flower as the
qualificand and red-colour as the qualifier. The Indian logicians
went much further. They analysed the concept denoted by 'the
flower' (or 'the creeper') into a flower-individual (or, a creeper-
individual) as the qualificand, and the generic property flower-
ness (or, creeper-ness), which is roughly equivalent to flower-
universal (or, creeper-universal), as the qualifier. Using the
expression "Q (xy)99 for "x which is qualified by y"1 the first
variable x standing for the qualificand and the second, yt for
the qualifier, the structural analysis of the meaning of the
sentence (9) can be given as:
(10) Q (Q (ab) Q (Q (cd)
iNote that Q (xy) denotes a complex of terms and not a proposition.
Alternatively, use might be made of restricted-variables, writing Q (xy) for
the relation "x is qualified by y" and a * Q(*, y) for "anjc such that
Q(xyYf {cf. J.F. Staal, 'Correlations Between Language and Logic in
Indian Thought*, Bulletin of the School of Oriental and African Studies 23
(1960) pp. 109-22.
21 offer this symbolic translation instead of a necessarily ghastly English
translation of the Sanskrit expression : raktatva-vieitfab yo raktali tad-
where 'a\ '£,' *c\ 'd9, (e\ and «/' stand respectively for a creeper,
creeperness, a flower, flower-ness, red-colour, and redness i.e.,
red-universal. By manipulating symbols and using the principle
of substiution of elementary logic, (10) can be obtained from
the initial formula:
(il) Qixyf
It is to be noted that, of the elements 6a\ <b\ *c\ *d\ <e9 and </'
some represent qualificands, some represent qualifier and some
both, i.e., a quaiificand in the immediate context and qualifier in
the broader context and vice versa. But the creeper, represented by
*a\ holds a special position. It is simply a quaiificand with respect
to others and never a qualifier. Therefore, it is described as the
chief quaiificand (mukhya-visesya) of cognition (9).
Equivalently, (10) can be represented by constructing the
following diagram2.
To increase the power of this symbolic language we may use the

following two rules of combination :
(12) Q(mn) & Q(no)->Q(mQ(no)).
(13) Q(mn) & (mo)-+Q((mn)o).
"Q(mQ(m>))" can be read as "m which is qualified by n which
is, in turn, qualified by o". Similarly, "O((WAI)O)" can be read
viêUfam yat puspatva-visisfapuspaw tat-prakärakatfl latätva-va§i$ta-latä-

mukhya-visesyakarnjnänam\ "it is a cognition,-whose chief quaiificand is
a creeper qualified-by-creeperness, which creeper is qualified (also) by
a flower-qualified-by-flowerness, which flower is (again) qualified by
red-colour, which red-colour is qualified by red-universal."
iFirst obtain <Q (Q {cd) Q (efyf from 'Q(xy) \ by substituting 'Q(a/)' for
*x' and 4 Q(e/)' for y\ Then substitute 'Q(aby for 'if and 4Q (Q (cd)Q
(ef)Y for V in 'Q (uv)\ and obtain (11)
21 owe this alternative suggestion to Professor J,F. Staal.
as "m which is qualified by both n and o". Further, since the

connective <&' is symmetric, *Q((/M/I)O)' is equivalent to *Q((mo)ny
Thus, if in a certain discourse it is not necessary to talk about
redness, i.e., the universal red, cognition (7) can be represented
as:
(14) Q(cd)e).
Similarly, cognition (9) can be represented, without any reference
to entities like creeper-universal, flower-universal or the univer-
sal red, as follows :
(15) Q(aQ(ce)).
(15) can be read as "a creeper-individual which is qualified by
a flower which is qualified by red-colour". Similarly, (14) can
be read as "a flower-individual which is qualified by both flower-
ness and red colour5'. This, incidentally, shows that the quali-
ficand-qualifier distinction does not always concide with the
subject-predicate ( uddesya-vidheya) distinction of propositions,
because in the corresponding proposition "the flower is red" we
regard 'red colour' as the predicate (yidheya) and not 'flower-ness'.
We can now proceed to explain why the Indian theorists
claimed that 'paraphrases' of ail declarative sentences present a
structural similarity, i.e., are analysable into the quaiificand-
qualifier model. The 'paraphrase' or the säbdabodha of a sen-
tence should reveal that one element is the chief quaiificand and
that the rest play the part of either qualifier or quaiificand or
both. The Naiyâyikas accepted the convention that it is the
nominative or agent (or, more precisely, the word with the first
case-termination) which expresses the chief quaiificand of the
corresponding paraphrase^ while the grammarians .accepted the
convention that it is the finite verb which expresses the chief
quaiificand.1
Thus, the sentence:
(16) harir vihagatfi pasyati "Hari sees a bird"
will, according to Nyäya, generate as the paraphrase expressing
its säbdabodha'.
iThis is, I think, a reasonable interpretation of the two rival theories about
the analysis of the structure of a êàbdabodha, viz., prathamäntärtha-mukhya-
viseçyaka êâbdabodha and dhatvartha-mukhhya~viàe$yaka êâbdabodha. To
avoid complications I have not referred to the Mimärpsä theory in this
connection.
(17) vthaga-karmaka-darsanänuküla-krti-män harih "Hari is

qualified by effort generating the activity of seeing which
has a bird as object".
According to the grammarians, it will generate as the para-
phrase expressing its säbdabodha:
(18) vihaga-kannaka - darsanänuküla-vyäpäro hari-karttrkah
"the operation generating the activity of seeing which has
a bird as object is qualified by Hari as its doer (i.e., is
done by Had)".
It should be noted that all the elements of the orginal
sentence (16) can be correlated to one part or the other of each
paraphrase given above. Each paraphrase also reveals the quali-
ficand-qualifier structure of the original sentence. The philoso-
phical bias of the Nyâya school prompted the Naiyâyikas to
make the noun-substantive, Had, the chief qualificand. The gram-
marians, on the other hand, wanted to emphasize the function
of the finite verb, and hence, described its meaning as the chief
qualificand.
Using our notation, ( 17) can be represented as
(19) Q(AQ(*Q (**)))
where <h\ '£,' 's9 and '£' are abbreviations5 for 'Han', 'krtV ( «
'mental effort"), 'seeing' and 'bird' respectively. Note that '/i'
corresponds to the agent (nominative) of (16), '&' corresponds
to the verbal suffix "-//" (according to the Nyâya theory of
language, which, however, differs from that of some grammarians)
's9 corresponds to the root "drs'* and 'b9 to the karma or accus-
ative vihagam. Note also that seeing is said to be qualified by a
bird through the accusative or patient (karmatva) relation (i.e.,
the relation of having it as its object), effort is said to be qualified
by such seeing through the causal relation (janakatä, i.e., in
respect of being its generator), and so on. Furthermore, one
should note that the karma or accusative relation is expressed
by the accusative ending used in the word -base vihaga- ( =
"bird"), but the generator (janakatä) relation is expressed by
what is called sawsarga-maryädä, i.e., it is obtained from syn-
tactical properties of the sentence concerned. Similarly, the
relation (viz. inherence) through which the chief qualificand,
Had, is said to be qualiiied by the property krti ( - mental
effort) is also given by a syntactical connection (i.e., by sawsarga-

maryâdcl).1
Armed with the conceptual apparatus already described, the
Naiyâyikas as well as the grammarians found it easy to define
the scope of a single sentence, i.e., the notion of sentence unity
(eka-väkyatä). If the paraphrase reveals only one chief quali-
ficand, the corresponding utterance is a single sentence (although
possibly a complex one). But if several (i.e., more than one)
chief qualificands are revealed by the paraphrase, the corres-
ponding sentence is a multiple one. Thus, examples like "Look
a deer runs," etc., should not give the grammarians any trou-
ble, for, in spite of there being apparently two finite verbs,
there is only one 'chief ' finite verb whose meaning constitutes
the c/2/é/qualificand in the usual paraphrase or säbdabodha.
Some grammarians supported their theory that the meaning
of the finite verb becomes the chief qualificand in the cognition
that is generated as säbdabodha with the following examples:
(20) caitrena supyate "[It is] slept by Caitra" (Passive cons-
truction), i.e. "Caitra sleeps".
(21) smu megho garjati ''listen, the cloud roars".
The säbdabodha derived from these two sentences is represen-
ted by the grammarians as follows:
(22) caitra-karttrkah sväpah "sleeping is qualified by Caitra
as its doer".
(23) megha-karttrka-garjanarn srnu "listen to the roaring
which is qualified by the cloud as its doer". (To avoid
unnecessary complications, I have given the säbdabodha
i.e., paraphrase of the sub-sentence only.)
Note that in (22) sleeping (the meaning of the verbal root
sup) is the chief qualificand, and in (23) roaring (the meaning
of the root gqrj) is the qualificand which, in turn, may be re-
lated to the meaning of the principal verb "listen". Note also
that (20) has no word with the first case-ending because it
contains an intransitive finite verb in the passive; but it does
have an agent (kartr), marked by the instrumental ending,
iThis concept has been treated as identical with the tâîparya-sakti in the
Later Nyâya school. See also K. Kimjunni Raja, pp. 187, 209, 221.
because the sentence is in the Passive construction. In this case

it will be difficult to maintain with the Naiyàyikas that the
meaning of the word with the first case-ending (prathamä) should
be construed as the chief qualificand of the resulting säbdabodha ;
for (20) has no word with the first case-ending, and (21) has
megho (which is the word with first case-ending), whose mean-
ing, i.e., the cloud, cannot be suitably construed as the object
of listening.
The Naiyäyikas, however, in support of their theory, men-
tioned example (1) given above as well as examples, such as:
(24) ghato na bhavati pafah "a pot is not a cloth".
(25) asvo gacchaty änaya "bring (back), the horse goes away"
These two will give the following 'paraphrases'.
(26) pata-bhedavän ghatah "a pot is qualified by the mutual
absence of cloth".
(27) gamana kartäram asvam änaya "bring (back) the horse
which is qualified by the activity of going55. (Here, as in
(23), only the paraphrase of the sub-sentence has been
given.)
Note that (25) counterbalances the force of the argument put
forward with reference to example (21). With regard to (20),
the Naiyäyikas say that since there is no word with the first
case-ending, the meaning of the verbal root takes its place. (21),
on the other hand, will be interpreted as
(28) garjana-kartaram megham Érnu "listen to the cloud which
is qualified by roaring".
Although it is true that we do not hear the cloud but its roaring,
yet it is not unnatural to say that we hear the roaring cloud.
In such cases, the Naiyäyikas appealed to the following principle.1
\f something is predicated of a qualified entity (sa-visefana) and
if the predicate is not applicable to the qualificand as such, then
the predicate is supposed to be applicable to the qualifier
only.
The dispute between the grammarians and the Naiyäyikas may
lcf. sa-visesane hi vidhi-nisedhau viéeçanam anubhavatah sati viiesye bädhe :
Nyayakoéa, p. 877.
be accounted for by reminding ourselves of the first grand divi-

sion of word-constituents of sentences, which is twofold : into
verbal words and nominal words. Thus, initially, the choice lies
between these two. A verb phrase usually refers to some action
or state, and a nominal phrase often refers to some object or
substance which is in that state or performs (undergoes) that
action. The grammarians were more interested in forming rules
about phonemes and morphemes and about the inflectional
modifications of different words. They may have been struck by
the fact that it is in relation to the meaning of the verb that the
nominal stems take different endings. The same* nominal word
"Hari", for instance, takes different case-endings as its relation
to the verb jnä ("to know") changes, thereby yielding different
sentences, e.g.:
harir jänäti "Hari knows"—nominative relation,
harimjänäti "(One) knows Hari"—accusative relation,
harlnä jänäti "(One) knows by Hari"—instrumental relation,
barer jänäti "(One) knows from Hari"—ablative relation, and
soon.
This may have led some grammarians to think that the verb is
'all-powerful' in a sentence and holds all other elements together
as its attributes or qualifiers.1
The Naiyâyikas, on the other hand, were more interested
in ontological categories, In their ontological scheme, the idea
of substance (dravya), the obvious reference of most nominal
words, is predominant. Substance is regarded as the substratum
in which different properties, viz., qualities, actions, etc., reside.
According to them, the nominative (or the word with the first
case-ending) usually designates the substratum or the chief quail-
ficand, to which other objects are related as properties or quali-
fiers, designated by other elements of the sentence.
lOne might argue that the same root jnä (~"to know")can also take
different conjugational endings with the same noun-subject to indicate
different tenses, moods, etc. But such variation in conjugational suffixes
does not depend upon, and hence, is not determined by, the variation of
the verb's relation with the noun-subject. The point here is this : in
determining case relations the verb can be said to be an important factor
in some sense, but in determining tenses and moods the noun-subject is
not an important factor in the same sense.
416 logic, Language and Reality
In view of the above, we can make the following observations :

(i) Sentence or väkya may be initially deiined as any word-
complex. In order to account for only those sentences which are
accepted intuitively as both grammatical and meaningful, the
definition has to be qualified by introducing such notions as
competency (yogyatä), expectancy (äkähksä) and contiguity
(äsatti). (ii) For a sentence to be meaningful, it is necessary that
it can be understood by the hearer, which in turn means, accord-
ing to the Indian theorists, that it can generate a corresponding
cognition in the hearer. Such a cognition, called säbdabodha,
can be represented by a paraphrase, (iii) This cognition can only
arise from a sentence which is syntactically well-formed (as
already suggested by the notion of äkänksä), i.e., in which the
elements are interrelated in accordance with the syntactic
and inflexional rules, (iv) The paraphrase which represents the
säbdabodha can be analysed in terms of the qualificand-qualifier
structure, (v) The notion of säbdabodha, thus explicated, may be
taken to correspond to the notion of knowing the meaning of a
sentence.
§ 5.5 : GRAMMATICALLY AND MEANINGFULNESS
In the previous section, I have suggested that the notion knowing

of säbdabodha in the Indian context may be taken to correspond
to the notion of the meaning of a sentence. In this concluding
section, I shall further elaborate certain issues connected with
this point, säbdabodha h a piece of knowledge that arises in the
hearer from the utterance of a sentence. Sometimes it is translat-
ed as 'verbal testimony'or rather'knowledge derived from verbal
testimony' (in traditional terminology) as opposed to knowledge
based upon perception and inference. For, short 'Säbdabodha'
may be translated as' verbal knowledge', provided we are careful
so as not to contrast the word 'verbal' with 'real'. For, some-
times it may be claimed that somebody has only "verbal" know-
ledge but no "real" knowledge. Verbal knowledge is indirect
knowledge as much as inferential knowledge is, but this cannot
weaken its claim to knowledge-hood.
Common sense takes verbal knowledge to be less reliable than
perception, and inference. For, in acquiring verbal knowledge
we have to depend essentially upon the reliability of the speaker,
upon his honesty, competence and authority, and these criteria

are proverbially uncertian. Traditionally, a reliable or trust-
worthy speaker should have the following virtues:freedom from
error or illusion, lack of the intent to deceive, and lack of any
defect in his sense-faculties (cf., bhrama-vipralipsä-karanäpätava).
Such a person is called äpta. Vâtsyâyana has said under Nyäya-
sütra 1.1.7 that any person can be an äpta, irrespective of his
caste, creed or religion. Contrary to our common belief, one
should note that neither perception nor inference should be in
any way better than verbal knowledge as far as reliability or
certainty is concerned. For reliability of the sense-organs, ade-
quacy of the evidence or reason (for inference) and trust-
worthiness of the speaker—all sail by the same boat. As far as
Nyäya is concerned, none of them would be infallible. Each of
them can, on occasion, mislead and generate error. But,
nevertheless, knowledge is generated from all these sources.
The hearer's knowledge or the cognitive episode arising in the
hearer from the utterance of a sentence is said to grasp the
'meaning' {artha) of the sentence uttered. Most Indian philos-
ophers of language agree on this point, viz., on such an
interpretation of the term artha (meaning) in the context of a
sentence. What this episode grasps has a 'structured content'
(cf., visayatä) which we can make more intelligible by calling
it the structure of a thought. When we say that a particular
hearer a understands the meaning, we mean thereby that a has
a particular 'structured' thought. It may be said, therefore, that
the Indian philosophers were concerned with the 'hearer's
meaning' rather than the 'speaker's meaning'. Meaning is not
what is (or happens) in 'the head of the speaker, and
arguably, it is also not what happens 'in the head oV the hearer.
For, the hearer may hear and not understand the meaning at
all or misunderstand it. Sometimes one may say, "I have
heard what you said, but I have not comprehended (or fully
comprehended) the meaning." Presumably, we have access
to the 'inner world' or the mind of neither the speaker nor the
hearer. But the hearer is conceived here to be an 'ideal hearer',
who is any competent language-user. The structured thought
that is supposed to arise in such an ideal hearer is something
that is intersubjectively available: it is presumably shared by
any competent language-user who hears the sentence uttered.
In this way, it is claimed that an attempted account or analysis

of the knowledge of such an ideal hearer would be an account of
the meaning of the sentence. To be sure, the knowledge-episodes
of all the individual hearers may be distinct and different as
episodes, but they all share the same structured content. Obvio-
usly, we have to exclude a lot of other variables from this con-
cept of the knowledge-episode of the ideal hearer. Various
implications and presuppositions of the statement made may
be understood by the hearers, and they may be understood
differently at the same time. But these would not be our con-
cern here. The ideal hearer is like a computing machine, where
the input would be the utterance and the output would be a
corresponding uniquely structured thought or knowing episode.
In this style of philosophizing, or in this way of doing
philosophy of language, the structured content of the ideal
hearer's knowledge is regarded as giving the so-called semantic
interpretation of the sentence heard. It is maintained that each
grammatico-syntactic element along with the lexical items
contributes in someway or other to such a semantic inter-
pretation captured in the knowledge-episode of the hearer. Such
a knowledge-episode arises in the ideal hearer as soon as he
understands what is said although there are no observable
behavioral criteria, unless and until the hearer acts accordingly
or does something in reply. We assume that the hearer possesses
the knowledge-episode, provided he has been an attentive hearer
and a competent language-user.
The grammatico-syntactic elements of the uttered sentence
can be conveniently mapped into the analysis of the structured
content of the said knowledge-episode of the hearer. The term
säbdabodha or anvayabodha is sometimes used to mean simply
the description of this mapping. This knowledge of the hearer
is also propositional or qualificative in the sense that its struc-
tured content admits of a qualificand-qualifier analysis. Besides
the qualifier and the qualificand, the simplest structure contains a
third element, which we may call the 'connector' or 'mixer'. There
are generally two types of connector: identity and non-identity.
Non-identity has various sub-categories: owner-owned, locus-
locatable, content-ness, etc. These connectors are, in fact, only
semantic mirror-images of various syntactic and grammatical
elements represented at the surface structure of the language by
various inflections, etc. The usual rule of thumb is that if the

two expressions have appositional suffix or syntactic parallelism,
then the corresponding connector would be an identity (abheda)
between the 'meanings' or the objects they refer to. Otherwise,
the connector would belong to the sub-category of'non-identity'
(bkedd) connectors. At times, we can perceptually anticipate a
connection between two items; for example, we see that a cat is
on the floor. The object-complex, the reality which gives rise to
the perception, has no doubt a structure, but it has a 'neutral'
structure. It is neutral in the sense that it may give rise to
different perceptions with different sorts of structure. We may,
for example, see either that the cat is ON the floor or that the
floor is UNDER the cat. However, the utterance of the sentence
"The cat is ON the floor" would generate a piece of knowledge
in the hearer with a determinate structure. This is one of the
distinctions that Navya-nyâya emphasizes as existing between
a perceptual knowledge and verbal knowledge. In other words,
verbal knowledge cannot be subsumed under what is called a
mental perception generated by the remembering of the mean-
ings of words, etc. (See also § 5.3 for similar distinction bet-
ween verbal knowledge and inference.)
Let us note that although the English translation, "the cat is
white," of a corresponding Sanskrit expression "sveto mârjârah"
is generally acceptable, a more correct translation, according to
Nyäya, would be "The cat (is) something white or a white thing."
Since the two expressions have the same nominative inflection,
and hence the required syntactic parallelism, the knowledge of
the hearer generated by this utterance would have a structured
content: The cat is identical with a white thing. The non-
verbal (perceptual) knowledge-episode may, however, have
alternative structured contents as expressed by such sentence-
constructions: "There is white colour in the cat" and "The cat
has white colour." This is how one can explain the dictum in
Sanskrit: sabdahodhe nämärthayor abhedänvayah. Notice that
that the last two sentences are only verbal expressions of two
different perceptual episodes. The ideal hearer's knowledge
derived from such utterances would again be different.
Navya-nyäya authors suggest that the mappings of the hearer's
knowledge can be done as follows:
"The cat is on the floor (bhütale märjärah)".—UTTERANCE

The hearer's knowledge: The cat is qualified by occurrence on
the floor, i.e., by the floor through the connector of locat-
ability (âdheyatâ).
The syntactico-grammatical analysis of the sentence: bhütala
(floor) locative / märjâra (cat) nominative s.
It is said that the nominative s indicates the chief qualificand
(mukhya vUesya) for the hearer. Therefore, the cat, i.e., the
•meaning' of the object referred to by the associated word
is the chief qualificand. The locative/ signifies the connector
of locatability (âdheyatâ), and this links the associated word
with the other word in such a way that the object referred
to by the associated word ("floor") becomes the locus of the
cat. In this way, the ideal hearer's knowledge is described by
Nyäya as one whose qualificand is the cat corresponding to
the expression "the-cat", the qualifier is the floor (correspond-
ing to "the floor"), and the connector which makes such quali-
fication possible is locatability signified by the suffix "/" (or
"on"). Since this piece of knowledge is generated by the utterance
of the said sentence, we may say that a semantic description of
the uttered sentence has been given in this way from the hearer's
point of view.
One may now ask: Since the above three elements (to be
found in an atomic sentence) are distinct, what is it that com-
bines them into one unity? The Nyàya answer is: äkänksä 'syn-
tactic and inflectional expectancy'. Let us come back to the
discussion of this very important concept. The term certainly
has some psychological connotation, but -iïi Navya-nyâya it does
not stand for any psychological state of either the speaker or
the hearer. It stands for a property of the elements of a sentence
of a language. It is the sequential relation (cf., änupürvi) between
words and their suffixes as well as between the words. To be
precise, it refers to the interdependence of the lexical items
(nominal and verbal stems) and the grammatical elements
(nominal and verbal suffixes) as well as the interdependence of
certain grammatical categories (verbs, agents, objects, instru-
ments, etc.) among themselves. It is believed that all these items
or units are by themselves 'incomplete' or 'unsaturated; and
hence, require or 'expect' others to complete the sentence. The
sentence is thereby rendered grammatically acceptable and it

acquires the minimal meaningfulness. Identification of these
units or elements of the sentence is what gives its grammatico-
syntactic Äkänksä. It is thus a property of the linguistic elements,
and is regarded by Nyâya as a pre-condition for the arising of
the'ideal'hearer's knowledge. This is not at all surprising, for
the hearer cannot obviously interpret the sentence semantically
(or understand its meaning) unless he has some notion of its
grammatico-syntactic structure.
A sentence, as we have seen, is not simply a cluster of words,
but a cluster of words with a syntax which presumably expres-
ses a complete thought. Nyâya takes an atomistic view of
language, and hence, believes that the meanings of words (stems
and suffixes) come together to constitute the complete sentence-
meaning or the structured content of the ideal hearer's knowl-
edge. In this view, therefore, there arises a problem. Suppose,
the atomic elements are given or presented in a manner which
will not obey any syntactic or grammatical rules of the language,
although they are being presented in that language. It would be
an almost impossible feat to give an illustration. But let us try.
A sequence of words will be given, but it will not reveal any
syntax or observe any grammatical rules. Nyäya argues that the
hearer's knowledge in that case would not arise, for the pre-
condition, äkänksä, is lacking. Udayana believes that the hearer
does not have to have a direct knowledge of äkänksä, but simply
the presence of this property in the uttered sentence would be
enough to pave the way for the arising of the hearer's know-
ledge (provided other conditions are also fulfilled).1 One may
interpret this as saying that a sentence must observe the syntac-
tico-grammatical rules, i.e., must be grammatically acceptable,
and this property of grammatical-ness by its mere presence
would be one of the required conditions for generating the ideal
hearer's knowledge.
Consider the sentence :
Rämo ghatam anayati ("Rama brings a pot").
It generates, according to Nyâya, a knowledge of the form :
(The qualificand) Rama is qualified by the action which is con-
ducive to the fact of bringing whose object is a pot.
^NyayakusumahjalL (Udayana), Ch. 3. verse 13.
The following table shows the mapping of the linguistic elements

into the constituents of the hearer's knowledge :
Linguistic elements Components of the knowledge
The word "Räma" Rama
The verbal suffix ti The action
The verbal stem "bring" (ä-ni) The fact of bringing
The word "ghata" A pot
The nominal accusative suffix "am" The object of
Now, if all the constituents of knowledge are presented without
a syntactically and grammatically well-formed sentence, it would
not generate the required knowledge in the ideal hearer. One
may utter
("Rämo ghatah karmatvam ânayanam krtih" = )
"Rama the action the fact of bringing a pot the object of".
Here the hearer cannot legitimately understand the required
meaning, and hence, he would not have the required knowledge,
although all the components are given. We say "legitimately",
because he may on occasion make an intelligent guess, but obvi-
ously that is not our concern here. We do make intelligent (and
sometimes not so intelligent) guesses when we hear the so-called
ungrammatical sentences. But this will not fall into the category
of säbdabodha. Although the example is a little bizzare, the point
of the argument is, I think, clear. If a sentence does not observe
certain basic syntactico-grammatical rules of the language con-
cerned, it also lacks äkähksä. I shall presently address myself to
the question whether äkänksä ensures grammaticality of a
language-sentence.
We have seen in section 5. 4 that there are two other pre-con-
ditions that are mentioned in this connection. One is äsatti,
which is a 'physical' property. It is contiguity or proximity of
the linguistic elements in time and space. It pertains to the
physical aspects of the linguistic elements, while the other
yogyotä (compatibility or competency) pertains to the semanti-
cal aspects (meaning-aspects) of such elements. It is obvious that
the relevant linguistic elements must be presented (uttered) with-
out long and unnecessary gaps or interventions. Such gaps would
destroy the sentence-hood of a complex of elements.
Let us take a close look at the notion of yogyatä again. This
is sometimes denned as the lack of any disconnection or impos-
sibility of connection among the atomic meaning-elements deli-

vered by the atomic linguistic elements. It ensures the (conceptual)
possibility of one item being connected with another, so that
both can figure in the structured content of the hearer's know-
ledge. The concept of possibility is a modal concept, and hence,
there may be a problem of modality implicit in the concept of
yogyatä. The other problem that is involved here is that of
determination of the borderline between grammaticality and
semantic acceptability of a combination of linguistic elements.
According to some, the possible states of affairs include the
actual (modern modal logicians say that the actual world is,a
member of the set of possible worlds). But possibility may also
be understood in the sense of unactualized possibilities (cf.,
Quine's fictitious philosopher, Wyman, in his "On what There
Is P").1 A still narrower concept would be to limit ourselves to
future possibilities only. TheNyäya concept of yogyatä seems to
be broad enough to coincide with the first kind of possibility.
But this would raise a lot of problems. The typical examples
which are said to be impossible (i.e., incompatible 'a-yogya')
combinations are as follows :
1. (He) wets with fire.
2. The horn of a rabbit.
3. The (water—) lake has fire.
To facilitate our discussion, I shall add several other examples,
which Udayana listed in his Ätmatattvaviveka.2 According to
Udayana's commentators, Bhagïratha and Raghunätha, all these
illustrate different types of impossibility :
4. This mother is a barren woman.
5. I am mute (speechless).
6. I do not know this.
7. There is an elephant in my ear, and it is roaring, hence,
speak about the medicine.
The first two seem to illustrate some sort of metaphyiscal or
ontological impossibility. They are not only non-actual, but also
impossible in the sense that the first item (fire or rabbit) cannot
be related to the second item (wetting or horns) in the way indi-
iQuine, W. V., Froma Logical Point of view, pp. 1-19.

lÄtmatattvaviveka, (Udayana), ch. 2, p. 533 (Bib. Indica edition).
cated. The rabbits do hot have horns, and fire is not used to wet
but to dry. Could we argue here, just as it has been argued by
some in another connection, that creatures like unicorns are
possible objects, although they are not actual, that it is possible
for rabbits to grow horns, for that would not take the rabbit-
hood out of a rabbit? In other words, it does not follow from
the definition (or the essence) of a rabbit that it cannot have
horns. Similarly, one may even say that there may be liquid fire
in some possible world where we can wet or sprinkle the ground
with it. But this line of argument would be deplored by Nyàya,
for the two combinations then would not illustrate incompatibi-
lity. And if they are not incompatible, the ideal hearer would
have a knowledge of their 'meanings', for he would be aided
by, among other things, an awareness of their possibility. The
general Nyâya position is, however, that these combinations do
not generate the required knowledge in the ideal hearer, for
they lack yogyatä. Other examples are clearly intended to show
patent (or logical ?) impossibility. Water-fire may still be in the
borderline, but others can be said to be logically impossible
without further ado : motherhood and barrenness, speech and
speechlessness, knowing and not knowing and listening where
listening is impossible. (The last example was a bit enigmatic,
and different interpretations were given by different commenta-
tors. But I forbear to enter into them here.)
It may be surmised from such discussion that for Nyâya there
is no strict distinction between logical impossibility and factual
impossibility. Incoherent and incompatible conbinations are in
the domain of the impossibles.
The Nyäya theory of Sabdabodha obviously allows that the
domain of 'possible' combinations excludes the above-mentioned
impossible combinations, but it would include, besides the
actual or 'true' combinations, a large number of combinations
or word-clusters that would pass the test of physical proximity
(äsatti), that of the said syntactico-grammaticality (on our inter-
pretation of äkähksä), and also that of possibility or compatibi-
lity {yogyatä), but would still fail to generate a true cognition or
knowledge in the ideal hearer. Why ? Predominance of false
sentences uttered by deceitful or ignorant or incompetent persons
is a matter of common experience. Hence, the hearer, no matter
whether or not he is aware of the speaker's intention or his
qualifications, would have a false awareness (a false belief) with

a structured content similar to that of a piece of knowledge. In
such cases, we usually say that the uttered sentence is meaning-
ful but false. In view of this problem, further conditions were
imposed upon the situation giving rise to an episode of verbal
knowledge, viz., the speaker must be an äpta, a reliable person
(see before).
What then are these'possible' falsities ? How do they differ
from the impossible ones ? For, to be sure, the tendency in Nyâya
has been to push most of the so-called unactualized possibles,
the flying horse or the rabbity horn, into the domain of the
'impossibles' (i.e., the incompatibles, ayogya). This leads to the
discussion of the intricate problems of modal notions, which I
wish to skip in this connection. There is, however, one easy way
to distinguish the 'possibles' from the 'impossibles' in the Nyàya
theory. If our ideal hearer would have an awareness, true or
false, from the utterance concerned, then the combination
(uttered) would be pushed into the domain of possibility. Truth
or falsity would be determined in this theory by the speaker's
qualifications, etc. Our ideal hearer may be ignorant whether the
awareness that he has is a true one or a false one, unless and
until he makes further investigations : he may check the spea-
ker's qualifications, or the situation in which the supposed com-
bination has been stated to be actual, or use some other means.
So long as the awareness arises in him in the required fashion
without any further investigation, we have to accept the said
combination as belonging to the domain of the possibles. If,
inspite of the utterance passing the test of proximity and syntac-
tico-grammaticality, the required awareness does not arise in the
way it has been described, then the combination would belong
to the domain of the impossibles.
Some modern philosophers (e.g., W.V. Quine) have raised
doubts about the intelligibility of the analytic-synthetic distinc-
tion. It has been argued that the criterion for a clearcut distinc-
tion cannot be formulated without involving one in circularity,
and hence, it is a dogma of empiricism to maintain that the
distinction is intelligible. If this is true, then, a consequence of
it seems to be that a clearcut line of demarcation between the
unactualized possibles and the impossibles would vanish, or
would be ever-elusive. For, to be sure, what would be an
'impossible' combination in the Nyäya theory is a mirror-image of

what would be, roughly, analytically false : "Bachelors are
married." Arguments in the fashion of Quine can be given to
show that the combinations, such as, the rabbity horn, sprinking
with fire, the barren woman's son and the water-lake having
fire, are all, strictly speaking, impossible, for, no satisfactory
criterion can be formulated to show that some of them are
'possible,' while the others are not. The idea behind this is
sometimes expressed by the claim of some modern Naiyâyikas
that if horns started growing on rabbits, the new creatures would
not be rabbits any more ! In fact, it has been reported that some
rabbit-like creatures have been found with horns on them. Pro-
bably Nyâya would say that these are not rabbits but belong to
a (slightly) different species.
In fact, Nyâya would contend that the domain of the unactuali-
zed possibles (presumably referred to by such 'possible' but
unactual combinations of words) should be restricted to the
domain of false possibilities. Being guided by various considera-
tions, we do assume a lot of things and facts to be actual. Most
of them, however, after further investigation and further con-
sideration, turn out to be unactual, false. When even such initial
assumption is not made by us even though we are confronted
with some presumably grammatically combinable word-cluster,
we have passed beyond the domain of the 'possibles' and hover
now over the domain of the impossibles. *A rabbity horn' would
be an item of this kind. For, had we nqt been informed about
rabbits in the way we are actually informed, or had we even
been differently informed, we would have assumed the said com-
bination to be a 'possible' one. And in that case, Nyâya would
concede that an awareness would have arisen in us from the
relevant utterance : the rabbit's horn.
I shall conclude with a few comments on the second problem :
that of determining a fixed borderline between syntactic and
grammatical acceptability and what may be called semantic
acceptability. The issue is connected with the problem of defin-
ing grammaticality. The task of defining grammaticality is viewed
by moderners as involving some intricate problems.1 Indian
philosophers sometimes discussed the issue about where the
iLyons, J., Introduction to Theoretical Linguistics, Cambridge, 1968 p. 152.

domain of äkänksä ends and the notion of yogyatä takes over.

For example, why, it may be asked, cannot the requirement of
a liquid or water as an instrument for sprinkling or wetting the
ground be a matter of äkänksä, instead of yogyatä? Some Navya-
nyâya writers resolved the issue by saying that the concept of
yogyatä 'possibility' should be restricted to the lack of such
verbal contradiction or patent (analytical?) impossibility as is
expressed in "What is without fire has fire" (cf., nirvahnir vahni-
märi). Water as an instrument of sprinkling would be included in
this view in the domain of äkänksä. If this is a step towards the
right direction, it would support my attempt to connect äkänksä
with syntax and grammatically, and yogyatä with semantics.
We can reformulate the previous question and ask : where
lies the line where syntax-and-grammar ends, and semantics takes
over? In other words, why 'a noun must take a verb' or'a
transitive or object-taking (cf., sa-karmakd) verb must take an
object-noun,' would be considered only syntactic requirements,
while 'the verb sprinkling needs a liquid or watery substance as
its instrument' and 'the verb eating needs somewhat solid food
as its object' would be considered semantic requirements? Some-
what in the manner of N. Chomsky (in Aspects of the Theory of
Syntax), one can introduce sub-categorization of nouns and
selection rules of verbs in the syntactic theory, and thereby the
domain of syntax and grammar may be extended to include a lot
of consideration which we ordinarily take to be semantic con-
sideration. It may be noted that the Chomskian view (in Aspects)
was that while syntax can take care of numerous problems which
have been traditionally regarded as semantical, the notion of
analyticity and contradictions are to be regarded as matters of
semantics.1
The classification of grammatical elements and lexical items
and the system of rules dealing with them can be made progres-
sively more detailed and thereby the notion of grammaticality
can be redefined. The general idea here is probably this : gram-
maticality is, in the last analysis, to be defined by reference to a
iChomsky, N. specially pp. 148-63. Chomsky concludes : "...the syntactic
and semantic structure of natural languages evidently offers many
mysteries, both of fact and of principle, and that any attempt to delimit
the boundaries of these domains must be quite tentative." (p. 163).
particular system of rules, although it might not always be possi-

ble to formulate these rules explicitly and exhaustively. A string
of words is grammatical (and säkähksä) if it is generated by such
rules. But this idea may go against Chomsky, who would main-
tain that the notion of grammaticality is intuitively determined
by the native speaker.
The Nyâya theory, according to the interpretation that has
been suggested here, can be presented in the form of the follow-
ing (inverted) tree-diagram :
S 1 - t h e set of word-clusters that generate the verbal know-
ledge passing all the four tests.
S0 = the set of combinations that generate false awareness (the
speaker is not âpta).
5 2 = the set of the possibles, acceptable grammatically and
syntactically.
5 3 = the set of the impossibles, acceptable grammatically and
syntactically.
5 4 = the set of those acceptable grammatically and syntacti-
cally
5 5 = the set of those unacceptable grammatically and syntacti-
cally.
(Input)
S
(Output)
Diagram I
5 6 = the set of those that pass only the 'physical proximity*

test.
5 7 = t h e s e t of those which fail even the "proximity 5 text.
S = t h e set of word-combinations of all forms.
The usual Nyâya discussion, however, does not envision such
a rigid system or model. The traditionally understood model is
more like something given below :
Diagram II
Here the nine regions should be interpreted as follows :
1 TRUE i.e., the set of those that generate knowledge in the
ideal hearer. The speaker is äpta.
2 FALSE i.e., those that generate a (false) awareness. The
speaker is not âpîa.
3 Grammatical (syntactically acceptable) and Possible.
4 Possible and (physically) Proximate.
5 Grammatical (syntactically acceptable) and (physically)
Proximate.
6 Grammatically (syntactically) acceptable.
7 Possible
8 Proximate (physically).
9 Non-proximate combinations, if any.
Circle I is for Proximity, Circle 11 for Grammaticality, and
Circle III for Possibility.
The above way of conceiving grammaticality and meaningful-

ness may face two very common objections. First, we do meet
a lot of grammatically incorrect expressions (e.g., a language-
learner's first exercise in composition), of which we seem to
understand the 'meaning'. These expressions will fail the test of
äkänksä, according to our above model. Do we have a säbda-
bodha in such cases ? Our answer would be that we do not
have a verbal knowledge directly from the utterance in such
cases. The incorrect expression reminds us (due to similarity,
etc.) of the correct one, which then generates the required verbal
knowledge.
Second, there are indeed numerous expressions in our langu-
age forming part of poetry, riddle, fiction and fantasy, where
the so-called incompatible or impossible combinations do fre-
quently occur. And we do seem to understand their meaning or
significance. They do not pass the test of yogyatä, according to
our above model. Do they generate säbdabodhal For, it cannot
be denied that "sprinkling with fire" could be part of a meta-
phor or a poetic expression, and even "green ideas sleep furiously"
may be a line of a so-called nonsense poem, or a riddle. In reply
we must say that the notion of compatibility or possibility here
is relative to the actual world we live in and the non-poetic
everyday language. When we enter into the realm of fiction or
the fantasy-world, such 'possibility' test will not be needed. For,
the unactuals and the impossibles, the improbables, except,
perhaps, the properly impossible one (such as, the same part of
a wall being red and blue all over at the same momemt for the
same observer in the same sense), can form part of the furni-
ture of our fantasy-world, or some suitably chosen 'possible'
world. We can even perceive them in dreams, and communicate
them in language or through some other media, paintings
(illustrations of "puzzle" paintings are nuemrous). However,
the fantasy-world is founded necessarily upon the actual world,
and the non-actuals can shade off gradually from the possibles
to the impossibles; we can understand such expressions only
when we cast sidelong glances towards the actual world. We
understand something to be impossible, because we understand
what could have been or is possible; and we understand the
possibles, because we understand the actual. It cannot be the
other way round.
SELECT BIBLIOGRAPHY
Only those books are listed as are alluded to in the footnotes.

Abhidharmadlpa, with Vibhäsä-prabhä-vrtti. (ed. P.S. Jaini), Patna,
1959.
Akalanka. Akalahka-grantha-traya, ed. M.K. Sastri, Singhi Jain
Ser. 12. Calcutta, 1939.
Annambhatta. Tarkasamgraha. (ed. S. Vangiya), Varanasi, 1969.
(Also edited and trans, by A. Foucher, Paris, 1949).
Anuruddha. Abhidhammatthasahgaho (ed. Revatadharma Sastri),
Varanasi, 1965.
Aristotle. The Basic Works of Aristotle (ed. R. McKeon), New
York, 1941.
Asanga. Abhidharmasamuccaya (ed. P. Pradhan), Santiniketan,
1950.
Aung, S.Z. Compendium of Philosophy, London, 1965.
Austin, J.L. How to do Things with Words, Oxford, 1962.
Sense and Sensibilia, Oxford, 1962.
Ayer, A.J. Foundations of Empirical Knowledge, London, 1951.
Bambrough, J.B. New Essays on Plato and Aristotle, London, 1965.
Basham, A.L. History and Doctrines of the Äßvikas, London,
1951.
Berkeley, G. Works (ed. T\ Jessop and A. Luce), London, 1948.
Bhäsarvajna. Nyäyasära (ed. S.S. Sastri and V.S. Sastri), Madras,
1961.
. Nyäyabhüsana (ed. Swami Yogindrananda), Varanasi,
1968.
Bhartrhari. Väkyapadlya (ed. K.V. Abhyankar and V.P. Limaye),
Poona, 1965.
. Väkyapadlya, kända III, with Heläräja's comm. (ed.
K.S. Iyer), Poona, 1963.
Bhattacharya, D.C. Bange Navya-nyäya Carcä, Calcutta, 1953.
Bhattacharya, V. The Ägamasästra of Gaudlapäda. Calcutta,
1943.
Bhavânanda. Kärakacakra (ed. ß.S. Sastri), Varanasi, 1942.

Bochenski, J.M. History of Formal Logic (trans. I. Thomas),
Freiburg, 1956. Notre Dame, 1961.
BöhtJingk, O. Pähinis Grammatik, Leipzig, 1887.
Broad, CD. Scientific Thought, London, 1923.
Brhadäranyaka Upanisad. An.andasrama edition, Poona, 1953.
Candrakirti. Prasannapadä. See M ädhyamika-kärikä of Nägär-
juna.
Caraka. Caraka-samhitä (ed. Y. Sharma), Bombay, 1933.
Cardona, G. Pänini: A Survey of Research, The Hague, 1976.
Carnap, R. The Logical Syntax of Language, Patterson, N . J .
1959.
—_. Meaning and Necessity, Chicago, 1947.
Chi, R.Y.S. Buddhist Formal Logic, London, 1969.
Chishoim, R. (ed.). Realism and the Background of Phenomenology,
Glencoe, Illinois, 1961.
Chomsky, N. Aspects of the Theory of Syntax, Cambridge, Mass.,
1975.
Collingwood, R.G. An Essay on Metaphysics, Oxford, 1940.
Danto, A.C. Mysticism and Morality, New York, 1972.
Dasgupta, S. A History of Indian Philosophy, 5 vols., Cambridge,
1963.
Descartes, R. Philosophical Works (Tr. and ed. by E.S. Haldane
and G.R.T. Ross), Cambridge, 1931.
Dlghanikäya. (ed. T. Rhys Davids and EJ. Carpenter), 3 vols.,
London, 1890-1911.
Dhammasangani (ed. P.V. Bapat and R.D. Vadekar), Poona, 1942.
Dharmakîrti. Pramänavärttika (ed. Swami Dwarikadas Sastri),
Varanasi, 1968.
-. Santänäntarasiddhi (trans, with notes H. Kitagawa),
Kyoto, 1968.
1972. Vädanyäya (ed. Swami Dwarikadas Sastri), Varanasi,
Dihnäga. Älambanapariksä (ed. with Sanskrit re-trans. A. Sas-
tri), Madras, 1942.
. Pramänasamuccaya. See Dignäga, on Perception by M.
Hattori.
Select Bibliography 433
Ducasse, C.J. Nature, Mind and Death, La Salle, 1951.

Evans, J. D.G.Aristotle's Concept of Dialectic, Cambridge, 1977.
Frauwallner, E. History of Indian Philosophy, 2 vols. (tr. V.M.
Bedekar), Delhi, 1773.
Gangesa. Tattvacintämani. 1. with Rahasya of Mathuränätha,
Äloka of Jayadeva, and Prakäsa of Rucidatta, ed. K. Tarkava-
gisa, 4 vols. Calcutta, 1884-1891.
2. with Dldhiti of Raghunâtha Siromani and Jägadisi of Jaga-
dïsa, éd. S. Nyayopadhyaya, Varanasi, 1906-1908.
3. with Dldhiti and Gädädharl of Gadädhara, ed. ß. Bhatta-
charya, Varanasi, 1913-1927.
Geach, P.T. Reference and Generality, Ithaca, 1962.
Ghosaka. Abhidharmämrta, Sanskrit re-trans. S. Sastri, Visva-
bharati Annals, vol. 5, Santiniketan, 1953.
Geokoop. C. The Logic of Invariable Concomitance in the Tattva-
cintämani, Dordrecht, 1967.
Granofif, P. Philosophy and Argument in Late Vedänta, Dordrecht,
1978.
Guha, D.C. Navya Nyäya System of Logic9 Varanasi, 1968.
Hattori, M. Dignäga, on Perception, Cambridge, Mass., 1968.
Hemacandra. Pramänamlmämsä (ed. S. Sanghavi), Ahmedabad,
1939.
Hume, D. A Treatise on Human Nature (ed. L.A. Selby-Bigge),
Oxford, 1951.
. An Inquiry Concerning Human Understanding (ed. C.W.
Hendel), New York, 1955.
Ingalls, D.H.H. Materials for the Study of Navya-nyäya Logic,
Cambridge, Mass., 1951. -
Jaimini'Sütra with Sahara's Bhäsya (ed. M.N.Nyâyaratna), Bib.
Indica, Calcutta, 1889.
Jayanta. Nyäyamanjarl (éd. S.N. Sukla), Varanasi, 1936.
Jayatïlleke, K.N. Early Buddhist Theory of Knowledge, London,
1963.
Jhalkikar, B. Nyäyakosa, Poona, 1928.
Isvarakrsna. Sämkhyakärikä, with Yuktidipikä (ed. R.S. Tri-
pathi), Varanasi, 1970.
Jagadïsa. Sabdasaktiprakäsikä (ed. D. Sastri), Varanasi, 1934,
Jägadisi, see Gangesa.
Kamalasîla. Tattvasamgraha-panjikä, see Sântaraksita.

Kanada. Vaisesika-sütra (ed. Muni Sri Jambuvijaya), Baroda,
1961.
Kane, P.V. History of Dharmasästra, Vol. IV, Poona, 1953; vol.
V, Poona, 1962.
Kant, I. The Critique of Pure Reason (tr. N. Kemp-Smith), Lon-
don, 1933.
Kathävatthu (ed. A.C. Taylor), London, 1894-1897.
Kaviraj, G. Aspects of Indian Thought, Burdwan, 1966.
Keith, A.B. Indian Logic and Atomism, Oxford, 1921.
Kesava Misra. Tarkabhäsä (ed. N. Kulkarni), Poona, 1953.
Kleene, S.C. Introduction to Metamathematics, New York, 1962.
Kneale, W and Kneale, M. The Development of Logic, Oxford,
1964.
Kumärila. Mimämsä-sloka-värttika (ed. R.S. Tailanga Mana-
valli), Varanasi, 1898.
Kundakunda. Pravacanasära (ed. A.N. Upadhye), Agas, 1964.
Locke, J. An Essay Concerning Human Understanding (ed. A.C.
Fraser), Oxford, 1894.
Lyons, J. Introduction to Theoretical Linguistics, Cambridge,
1968.
Mahäyäna-süträlahkära (ed. S. Levi), Paris, 1907-1911.
Maitreya/Asanga. Madhyäntavibhäga, with Vasubandhu's Bhä-
sya (ed. G. Nagao), Tokyo, 1964.
Mallavädin. Dvädasäranayacakra (ed. Muni Sri Jambuvijaya),
Bhavnagar, vol. 1, 1966, vol. 2, 1976.
Manikantha Misra. Nyäyaratna (ed. V.S. Sastri and V. Krishna-
macarya), Madras, 1953.
Matilal, B.K. The Navya-nyäya Doctrine of Negation, Cambridge,
Mass., 1968.
. Epistemology, Logic and Grammar in Indian Philosophical
Analysis, The Hague/Paris, 1971.
. The Logical Illumination of Indian Mysticism, Oxford,
1977.
. Logical and Ethical Issues of Religious Belief, Calcutta,
1982.
• (ed) Nyäyasiddlianta-dipa Ahmedabad: L.D. Indological
Series (Text): 1976.
. and R. D. Evans (eds): Buddhist Logic and Epistemology,
D. Reidel (Studies of Classical India Series), Dordrecht,

1986A.
Perception: An Essay in Classical Indian Theories of
Knowledge, Oxford : Clarendon Press, 1986.
Mill, J.S. A System of Logic, New York, 1881.
Mohanty, J.N. Gangesd's Theory of Truth, Santiniketan, 1966.
Murti, T.R.V. The Central Philosophy of Buddhism, London,
1953.
Nâgârjuna. Mädhyamikakärikä, with Prasannapadä of Candra-
kirti (ed. P.L. Vaidya), Darbhanga, 1960.
. Vigrahavyävartani, (Printed as 5th Appendix to the
above).
Nägesa. Paramalaghumanjüsä (ed. K. Sastri), Kurukshetra, 1975.
Nieuwenhuijze, C.A.O. van Aspects of Islam in Post-Colonial
Indonesia, The Hague, 1958.
Nyäya-darsana. Nyäyasütra of Aksapâda Gotama, Adhyâya I,
with Nyäya-Bhäsya of Vätsyäyana, Nyäyavärttika of Uddyo-
takara, Tâtparyatïkâ of Vâcaspati Misra, and Tätparya-pari-
suddhi of Udayana (ed. A. Thakur), Darbhanga, 1967.
— . Nyäya-sütra (ed. G. Jha), Poona. 1939.
. Nyäya-darsana, (ed. and tr. with notes by M. M.
Phanibhusana Tarkavagisa, Calcutta, 1939 (1346 B.S.).
Pancöstikäya. (ed. A Chakravartinayara), Arrah, 1920.
Patanjali. Mahähhäsya (ed. Vedavrata Varnin), 5 vols., Gurukul
Jhajjhar, 1962-1963.
Plato. Collected Dialogues (eds. E. Hamilton and H. Cairns),
New York, 1963.
Potter, K.H. The Padärthatattvanirüpana of Raghunâtha Siro-
mani, Cambridge, Mass., 1957.
Prabhäcandra. Prameyakamalamärtanda, (ed. M.K. Sastri), Bom-
bay, 1941.
Prasastapäda. Padärthadharmasamgraha; 1. with Sûkti, Setuaxià.
Vyomavatl (ed. G. Kaviraj), Varanasi. 1930.
2. with Nyäyakandall of Sridhara (ed. D. Jha), Varanasi, 1963.
3. with Kiranävali, of Udayana (ed. J.S. Jetly), Baroda, 1971.
Price, H.H. Perception, London, 1951.
Püjyapäda. Sarvärthasiddhi (ed. J. Sastri), Sodasapura, 1939.
Quine, W.V.O. From a Logical Point of View, Cambridge, Mass.,
1961.
. Philosophy of Logic, Englecliffs, N.J.., 1970.
. Methods of Logic, New York, 1961.

- . Word and Object, Cambridge, 1960.
. The Ways of Paradox and Other Essays, Cambridge, 1976.
Radhakrishnan, S. Indian Philosophy, vol. 2, London, 1929-31.
Raghunätha Siromani. Padärthatattvanirüpana, The Pandit Re-
prints, see Potter, K.H.
Didhiti, see Gangesa.
Rahula, W. What The Buddha Taught, Bedford, 1967.
Raja, K. Kunjunni, Indian Theories of Meaning, Madras, 1963.
Rändle, H.N. Fragments from Dinnäga, London, 1926.
—. Indian Logic in Early Schools, London, 1930.
Ratnagotravibhäga Mahäyänottaratantra-sästra, (ed. E.H. Johns-
ton), Patna, 1950.
Ratnakîrti. Ratnakirti-nibandhävali (ed. A. Thakur), Patna, 1957.
Robinson, R. Plato's Earlier Dialectic, Oxford, 1953.
. Definition, Oxford, 1954.
Robinson, R. The Buddhist Religion, ßelmont, Calif., 1970.
Russell, B. Mysticism and Logic, London, 1917.
. Introduction to Mathematical Philosophy, London, 1919.
— . Human Knowledge: Its Scope and Limits, New York,
1948.
. Problems of Philosophy, New York, 1958.
and Whitehead, A.N. Principia Mathematica, Cambridge,
1962.
Sabara : See Jaimini-sütra.
Sälikanätha. Prakaranapancikä (ed. A. Subrahmanya Sastri),
Varanasi, 1961.
Samantabhadra. Äptamimämsä (ed. G.L. Jain), Varanasi, 1914.
Samyuttanikäya. Part III (ed. L. Feer), London, 1890.
Sandhinirmocanasütra (tr. from Tibetan into Sanskrit and French,
E. Lamotte), Louvain, 1935.
Sanghavi, S. Advanced Studies in Indian Logic and Metaphysics,
Calcutta, 1961.
Sântaraksita. Tattvasamgraha, with Pahjikä of Kamalasîla, (ed.
E. Krishnamacharya), Baroda, 1926.
Sasadhara. Nyäyasiddhäntadipa (ed. B.K. Matilal), Ahmedabad,
1976.
Sastri, G. The Philosophy of Word and Meaning, Calcutta, 1959.
Säyana-Mädhava. Sarvadarsanasamgraha (ed. V.S. Abhyankar),
Poona, 1951.
Siddhasena Divâkara. Sanmati-tarka-prakarana, (eds. S. San-

ghavi and V. Doshi), 5 parts, Ahmedabad, 1924-1931.
Searle, J. Speech-Acts: An Essay in the Philosophy of Language,
Cambridge, 1969.
Solomon, E. Avidyä: A Problem in Truth and Reality, Ahmedabad,
1969.
. Indian Dialectics, Ahmedabad, 1976.
Sridhara. See Prasastapâda.
Srîharsa. Khandanakhandakhädya (N.K. Jha), Varanasi, 1970.
Snvailabha. Nyäyalüävatl (ed. D. Sastri), Varanasi, 1934.
Stcherbatsky, T. Buddhist Logic, 2 vols., Leningrad, 1930.
. The Central Conception of Buddhism (Reprint), Delhi,
1974.
Sthiramati. See Vasubandhu.
Stravvson, P.F. Individuals, London, 1959.
. Introduction to Logical Theory, London, 1966.
Swartz, R.J. Perceiving, Sensing and Knowing, New York, 1965.
Takasaki, J. A Study of the Ratnagotravibhaga, Rome, 1966.
The Shorter Oxford Dictionary on Historical Principles, 3rd edn.,
Oxford, 1970.
Trenckner, V. Milindapanho, London, 1962.
Tucci, G. On Some Aspects of the Doctrine of Maitreyanätha and
Asanga, Calcutta, 1930.
. Pre-Dinnäga Buddhist Texts on Logic from Chinese Sour-
ces, Baroda, 1929.
Uddyotakara. Nyäyavärttika (V.P. Dvivedin), Varanasi, 1915.
Udayana. Ätmatattvaviveka (ed. D. Sastri), Varanasi, 1940.
. Kiranävali (ed. S.C. Sarvabhauma and N.C. Vedanta-
tirtha), Calcutta, 1912, 1972. See also Prasastapâda.
. Tätparya-parisuddhi, with Vardhamäna's Nyäyanibandha-
prakäsa (eds. L.S. Dravid and V.P. Dvivedin), Calcutta, 1911.
See also Nyäyadarsana.
. Nyäyakusumänjali (ed. L.S. Dravid), Varanasi, 1912.
(ed. N.C. Vedantatïrtha), Calcutta, 1954-1964.
Umäsväti. Tattvärthasütra with Bhäsya (ed. K. Siddhantasastri),
Bombay, 1932.
Väcaspati Misra. Tätparya-tikä (ed. R.S. Dravid), Varanasi,
1925. See also Nyäyadarsana.
Vädideva Süri. Syädvädaratnäkara (ed. M. Ladtiaji, 4 Parts),
Poona, 1927.
Vardhamäna. See Udayana.

Vallabha. See Srivallabha.
Vasubandhu. Abhidharmakosa with Bliäsya (ed. P. Pradhan),
Patna, 1967, (with Spiiutärthä of Yasomitra, ed. Dwarikadas
Sastri), 4. vols., Varanasi, 1970-1972.
. Vijhaptimätratäsiddhi, wich Svavrtti and Sthiramati's
Bhäsya (ed, S. Levi), Paris, 1925. See also Maitreya/Asanga.
Vätsyäyana. See Nyäyadarsana.
Venkataramanam, K. Sammitiya-nikäya-sästra (English trans.
from Chinese), Visvabharati Annals, Vol. V, Santiniketan, 1953.
pp. 153-243.
Vidyabhusana, S.C. A History of Indian Logic, Calcutta, 1920.
Vidyänanda. Astasähasrl (ed. Vamsidhara), Bombay, 1915.
Vyomasiva. See Prasastapäda.
Warder, A.K. Indian Buddhism, Delhi, 1970.
Whitehead, A.N. Adventures of Ideas, Cambridge, 1942.
. Process and Reality: An Essay in Cosmology, New York,
1969.
and Russell, B. See Russell, B.
Yasomitra. See Vasubandhu.
Yuktidlpikä. See Isvarakrsna.
INDEX
abädhitatva 26, 43 antinomies 86

abhâva (category) 161, two types of, anumâna 23, three types 29 ff, pürva-
157 vaty sesavat, and sämänyatodrsta
abhinivesa (Yoga), 331 29-42, 43, 383
abhfttaparikalpa 231 anumeya 59-68
absence 112, 116-7, absence-range of änupürvi (def. äkänksä) 420
properties 116-7, 122-4, 125, 126 anusaya (klesa) 327-9 six types 328,
absenteehood 140n, see also pratiyo- 342, 'defilements'
gjtä anuväda 'repetition' 193 (in definition)
action (kriyâ) 270 anuyogin 170
actuality (vs. existence) 79 anuvyavasâya (reflective awareness),
adhikarana 82 211-2
adrsta 'unseen forces' 368 anvayabodha (see säbdabodhd) 418
Advaita Vedânta 207, 317 anyâpoha 146
advaya 'non-dual' 346 anyaihâkhyàti 97, 138
Agnosticism 321 anyonyäbhäva 147, 157
ahajfikära (Nyäya) 356 apasabda \11
ajâti-vâda 288, doctrine,of non-origi- apasiddhänta 13
nation 345 apperception 245-6
Âjîvaka 304, 352, 365, 367, 37On. aprasiddha 117
Akalanka 244, 250, 309-10, aprasiddha-sädhyaka 190
äkänksä 'expectancy' 391, 405, 406-7, aprasiddhyä-avyäpti 172
416, 420, 421-2, 424, 427-8, see argument from illusion 217, 234
'expectancy' Aristotle 1-5, 7, 20, 22, 32, 46, 48n,
äkära 'form', 'image' 237 'represen- 49, 51, 63, 89, 113, 123-5, 155, 180,
tation' 260, see also pratibhäsa 183n, 195, 269, 271, 278-80, 284,
akriyä-väda 'fatalism' 367 vs. kriyä- _ 286, 297-8.
väda 369n, 370n Ârtabhâga 364
Aksapâda 178, 179, see also Gautama arthakriyä 227-8
400 âryasatyâni 361
cdambana 'objective basis' 99, 218-9, asädhärana {hetväbhäsä) 52
220-25, 227, 237n, 285. asambhava (defect of definition) 172
Cilaya (-vijnänd) 247, 248, 333-37 Asanga (Maitreya) 10, 216, 228-31,
339-40 341-6, 359n 335n, 337
Al-George 177 asat-kâryavâda 270ff, 288, 318
älocanamätra 211 asat-khyäti 97-9
Amarasimha 401, 402 asaî'praîipaksitva 26, 43
Amratacandrasüri 299 äsatti 'contiguity' 4C5, 407, 416, 422
anaïkântika (=sa-vybhicärd) 54-5, 56 âsrayasiddhi (defect of hetu) 182
ananyaîhâsiddhi 290, 292-3, 294, vs. aîivyâpti 'over-extension' 172, 185-6,
unconditionality of Mill, 292-3, atomism (Buddhist), 221-7, 235-6,
anâtman (no-soul) 270 238; (Vaisesika) 275-6
anekäntatä (many-sidededness, non- atvantâbhâva Ï47
onesidedness) 270, 281, 301, 305, Aimg, S.Z. 336
309-14, 318, 319 Austin, J.L. 176, 234n
Angutîaranikàya 282 avacchedaka (delimitor) 83-4, 127,
Annambhatta 187n, 40In, 402n 135
antar-vyäpti 189n avayava (part) 48
avayavin (whole) 39, 275 Carnap, R., 94, 101, 114, 131n, 166n,
avidyä 232, 238, 252, 319-32, 356-7, 169n, 171n
361 Cartesian doubt 183, 218, 233
avinäbhäva (vyâpti) 144 Cärväka 14
a-vyabhicaritatva (nondeviation) 130, Causal Theorists 223-4, 228
143, 188 causality 277, 284-94 (Buddhist no-
avyäkrta 'indeterminate' 'neither good tion) 284-5, (Nyäya) 285-90 (mod-
nor bad' 336 ern) 290-94, universality of cau-
avyâpti (under-extension) 172, 186 sation 291, causal chain of Bud-
avyäpyavrtti (non-pervasive property) dhism 340-2, 357-61)
119-122, 135, 144, 150 also called catuskoti (Tetralemma) 146
'partial location' 119 Celèrent (syllogism) 6, 7, 32
avyaya ('invariant' words) 393, 401 n certainty (logical vs. psychological) 73
Ayer, A.J. 154n, 291 Chakrabarti, A. 140n, 389
chala 'equivocation' 13, 15
B change and permanence, paradox of,
270,275, substratum of change 275,
bädha (=kälätita) 26 (defect of hetu), see also 'causality'
204 Chi, R.S.Y. 124, 125
Bambrough, J.B. 8 Chisholm, R. 88n
Bankimcandra 201 Chomsky, N. 427-8
Barbara (syllogism) 6, 7, 125 Church, A. 165n
Barnes, W.H.F. 220 citta-viprayukta-sairiskära 334-5, 336
Basham, A.L. 304n class-abstraction 167,
being vs. non-being 282-3 class-identity 159, 167
Berkeley, G. 234 class-logic 169n,
Bhadrabâhu 5 class-paradox I33n.
Bhagavad-Gitä 368 co-extensiveness 186, 187, see
Bhagiratha 423 samaniyata 188, also
Bhartrhari 215, 3O6n, 378-80, 381-2, equivalence 191, also
386-7, 397, 399n co-location 120, 122, 123
Bhâsarvajrïa 30, 33, 37, 43, 143, 183, cognitive event, analysis of the struc-
184-7, 192,3, 243, 246-7 ture of, 156-7
Bhätta, see also Kumärila, 207 Collingwood, R.G. 29
Bhattacharya, D.C. 165n, 389n concomitance 140-45, see also 'perva-
Bhattacharya, K. 45, 57-8 sion' and vyäpti
Bhaftoji Diksita 305, 323 contradiction, Law of 305, 310-11
bhâva (category) 161, 162 self-contradiction 312
bhävatva 162 co-occurrence 74, 144
Bhavänanda, 374n, 375n cosmogomy (Sat vs. Asat) 270
bhavänga-vijnäna 336-7 co\miQïû.çmonsiv&ï\<m(pratisthapanâ)
Biardeau, M. 165 2, 3, 4
bija 'seed' 338-40 see also vipäka Cratylus 271
Blue Annals, The. 67n
Bechenski, I.M. 165n D
Boolean Algebra 172n
Brentanp, F. 350 Dandin 395
Broad, C D . 217 Danto, A.C. 347, 348
Brough, J. 378n Dasgupta, S. 385n
buddhi (cognition) 207 dausthulya, two types of 'blemishes'
344
deductive reasoning 89-91
'defeat' situation 13, see nigrahasthâna
caitanya (consciousness, pure) 207 definition (laksana) 164-174, 176-202,
Candrakîrti 282n, 316 nominal vs. real def. 1804, elimi-
Caudramati 272 native def. 181n, contextual def.
Candränanda 37n, 141 n 181 n, legislative def. 181, 197,
Cantor 174 definitiod 166
Caraka 2-4, 10-12, 29, 38, 40 delimitors (avacchedaka) 121, 127
Cardona, G. 177 demonstration (sthâpanâ), 2-4
Index 441
Descartes, R. 201, 233-4 expectancy (äkanksä) 406-7

Description, theory of, 80-81 extension 166, 169, The thesis of ex-
determinism (fatalism) 352, 365, 368, tensionality 169n
see niyati
deviation (yyabhicärä), 122-4
dharma (in Dharmasästra) 263, (in
Buddhism) 270, 272, 285, 287 (sam- fallacy (see hetväbhäsa) 42-6, 51, 'for-
skrta), 327, (sâsrava), 328, 330, 334, mai' vs. 'non-formal' 46
dharmatä 229 dharma-dharmin 115, family-resemblance 197
dharmin as 'logical subject' 274, flux doctrine 271-2, 278-9, see also
299 'momentariness'; argument for
Dharmakirti, 15, 19, 86, 142, 144, 'flux' doctrine 276-7, 282-4
228, 232, 237-8, 243-6, 252, 276, Frauwallner, E. 5, 30n, 77, 320, 325,
278 326n, 344
Dharmaträta 338 freedom of choice, see 'karma' theory
dialectic 3-4, 8-22 370
Dlgha-nikäya 9n, 302, 325 Frege, G. 114
Dignâga 19, 32, 50, 58-76r#l, 114-5, 'five-membered argument 2-4, see
128-9, 132, 141, 144, 146, 189, 210, . pancävayava
224, 228, 232, 236-7, 243, 245, 274, Foucher, A. 165n, 401 n.
345
dosa (Nyäya) 355-6, 359-60 G
dravya-sat 221, 236
dravya 'substance' 173, 186 Gadädhara 134n, 151,171n, 173n, 175
(Grammarians') 379-89, dravyatva Gangesa 32, 78, 84-5, 119, 121, 123,
173 127,129n, 130,131-40,143-4,165n,
drstänta (example) 55, 89-93, sädhar- 191, 203-5, 211n, 212-5, 290, 294,
mya-d. 75, vaidharmya-à. 91, 92-3 397, 406
Ducasse, C.J. 291, 354 Gangopadhyaya, M.K. 145n
duhkha, 324-5, 346-51, 353-4 garbha (theory) 342-6 see Tathâgata-
garbha
Gau<Japâda 30-1, 288n, 345
Gautama 400-1, see also called Aksa-
elenchus 21 päda
ekasesa (grammar) 380-1 Geach, P.T. 51
emptiness 'sünyata 311, 316-8 Geokoop, C. 78, 91 n, 145n
empty property 9n, see also 'non- God's Eye view 314
existents' Gosäla (Makkhali) 304, 37On
enthymemes 3 grähaka-grähya 237-8, 258
Epimemides 'paradox 214 Granoff, P. 20n
epistemology 72, 73, 89 Guha, D.C. 115n
equi-locatable 148, 149 see satnani- guna, gunatva 173, 186
yata Jaina concept of paryäya and guna
equi-vocation (chaia) 13, 15 295-6;
eristic 20 Vaisesika notion 296-7
essence vs. syabhâva, 165, 176-202 Gymnosophist, 1
vs. meaning 424
vs. existence 197 H
essentialism 197, 198-201
Evans, J.D.G. 4n, 22 hadayavatthu 335
ever-present (property) 134,137 (def.) Hattori, M. 30n, 61n, 67n, 129n,
144, 152: see also 'property' 237n
example, agreeing 9Î-2, 95 Hayes, R. 76
disagreeing 95, 102, 132, 138 hedonism 366
see alsoidrstäntd> Heläräja 379-82, 386n, 387-8
Excluded Middle, Law of 305, 349 Hemacandra 209n, 3O5n
Heracleitus271, 352n
existence 79, 85 ff, 271-3, 280 existent Herzberger, H. 146
vs. non-existent 271-4 Jaina notion Hertzberger, R. 389
of, 293, 298 hetu (Vaisesika) 33;
442 Logic, Language and ReaHty
kevala-vyatirekin 178-80 hetu and Johnson, S. 205-6, 228

pratyaya (Buddhism) 6, 284-5, 337, Johnston, E.H. 342n, 344
sahabhü 337, hetu(trirûpa) 42-3, 48 judgemental perception 210-2
reason, evidence) 2, 11; 23, 26, 50,
51, 64, 68, 71, 90-1, 103-4, 107, K
125, 128-9, 166, 172, 187-8, vyam- Kafka, F. 346
sakay itlsaka, yâpaka and sthäpaka Kaiyata 383n, 384-5, 386n, 387
(Jaina) 11, vyatirekin 189
hetucakra 'wheel of reason' 128-130 Kâlâtîta (mis-timed) 44-5, also called
hetvâbhâsa 26-7, 42-58 kälätyayäpadesa 26
Hiriyanna, M. 386n Kamalasîla 206
Hueng Tsang 342 Kanada 40
Hume, David 234n, 289-92 Kanakavarman 59n, 67, 120a, 132n
Kane, P.V. 362, 364n, 371 n, 385n
1 käraka 372-4
idealism 221-4, 228-40, 243,249-51 kamna (instrumental cause) 372-8
illicit minor 46 kârana (cause) 372-4
illocutionary act 176, 179, 194 karma (Vaisesika) 174. 186
incomplete symbol 170, 181, 420-1 karmatva 173
inference (anumâna), three types of, karmatva (Grammar) 412
karma (doctrine) 260, 261,351-6, 358-
131.
kevalänvayin, kevala-vyatirekin, and 9, 361-71
residual forces of, 326
anvaya-vyatirekin 131 sämänyato- käryasama (jäti), a sophistry 15
drsta 383, three types, pürvavat, ses- kathâ (debate) 12, 20
avat and sämänyatodrsta 383n Kathävastu (Kathövatthu) 10, 333
Ingalls, D.H.H. 38n, if, 119, 130n, Katsura, S. 262
134n, 135n, 136n, 140n, 149,153,
158, I65n, 168n, 170n, 172n,353n, Kätyäyana
Kaundinya 30n
380-1, 386, 399-401
367n Kaviraj, C. 353, 365n, 368n
'insufficient' (hina)—a 'defeat' situ- Keith, A.B. 165n
ation 13 Kesavamisra 405n
intensional abstraction 169-78, kevalänvayin 84-5, 130n, 131-140; see
intensional property 169 fc
everpresent property* 144, 166
khyäti, 257
Kitagawa, H. 238n
Jacobi, H. 303 klesa (ses amtsayà) 331 (in Yoga)
Jagadîsa 389-96 klista-manas, 'ego-shrunk* mind 335,
Jaina 208-9, 239, 244, 262-3, 271, 337, 339
279-82, 294-302, 306-Ï 9 Kîeene, S.C. 174n
Jaini, P.S. 334n, 336n Kneaîe, M. 21
jalpa (debate) 12-13, 15, 20-1, 105 Kneale, W.21n
Jambuvijayaji, Muni Sri 37n Kripke, S. 200
Janaka 364 Kreisel 1
Jânakînâtha 405n Kumärila 25, 59, 63n, 64-5, 85, 141,
jäti (false parity of reasoning) 13-4, 205, 210-11, 231, 239, 243-4, 248,
(also 'sophistry') 56-7 299-300, 309
jäti (sämänya, univresaîs) 131,173-5, Kundakunda 280-1, 294, 296, 299-
197, 387, 394 300, 302
jäti-bädhaka 13 In, 173n
Jayanta 30, 31,33,63, 243-6,248, 375
Jayatilleke, K.N. 303
Jha, Ganganath 24a, 35a, 37, 39n, laksana (definition) 164-175, 176-202;
47n laksya 166, 169, 170-1; laksyatävac-
Jhalkikar, B. 403n chedaka 170 laksana-laksana 184
jivitendriya life' or 'lifeforcc' 335 laksanä (metaphorical meaning) 397
jnàna (cognition) 203, 210, 214, savi- Lankävatära 317
kalpa and nirvika/pa 407-8 quali- Lamotte, E. 343n, 345a
ficative 408-10 Levi, S. 344
Jnânasrimitra 87, 89, 228, 276n Liar paradox 214
jnätatä 'knowness' 208, 215, 243 linga ('indicator') 69-70
Index 443
Locke, J. 199, 234n negation matrix 126-7

logical connectives BX negation, two kinds of, 147
'logical grammar* 390, 397 propositional neg. 17, 116-7
logical subject 299 illocutionary neg. 17-8
Lokäyata 64, 364, 366 double neg. 145-64
negative particle, meanings of, 323-
M 4
negation vs. refutation 17-8,
Mâdhavâcarya 205 term neg. 116-7, 125, 128
Mädhyamika 14, 45, 57-8, 97, 146, two types (again) 157
229, 230, 275, 281, 288, 298. 302, nigamana (conclusion) 3
304, 311, 314, 316-8 nigrahasthäna ('defeat' situation) 12-
Mahäsämghika 339 3, apratibhä 105
Mahäyänasüträlankära 343 nihilism 318
Mahisäsaka 338 nihsvabhävatä 343
Mallavädin 385n nipäta 392-3, 398, 401 n
Mammatha 405n nirvana 229, 250, 270, 284, 331, 344,
Manusmrti 25 346-51, 354, 370
Manikantha 129n, 130, 134n, 139n, nirvikalpa (non-qualificative percep-
143-4" tion) 210-12
manodhätu 334 niscaya 242
Masuda, J. 338-9 nisprakâraka (non-qualificative cogni-
matänujhä 14 tion) 213
materialist 4 nityasamä (a kind ofjclti or sophistry)
Mathuränätha 122, 152-3 15
Matilal, B.K. 20n, 76n, 114n, 118n, niyati 367-8
134n, 135n, 146n, 294n, 366n non-being, 87-8, 95
meaning, speaker's vs. hearer's 417-8, non-deviation (avyahhicâra) 113, 125,
420, 422 143-4
Meinong 87-8, sosein and sein 88, 102 non-existence, non-existent 239-40
memory (s/mti) 208-9, 262-8 non-judgemental perception (see nir-
Milinda 10 vikalpa) 210-2
Mill, J.S. 223, 292, on uncondition- non-theism 352-3
ality 292-3 non-violence 313-4
Mimämsä 203-5, 239, 264, 399n Nyäyasütra 3, 5, 10, 12-3, 15, 29-30,
mithyâdrsti 327-8 34, 40-1, 43-4, 44n, 48n, 49, 54, 55n,
mithyàjnâna (Nyäya) 356-7 57n, 58, 195, 210, 231, 287, 353n,
modal concepts 423-6, see also yog- 354, 357, 383n
yatâ; unactual possible objects 424;
impossible objects 425-6 O
Mohanty, J.N. 78, 98n, 203, 208, 210,
212-5, Oberhammer, G. 14, 143n
momentan ness (flux doctrine) 91-4, Ockham's razor 80, 198
225
Moore, G.E. 215-7, 220
moral causation vs. karma 368
multiple-valued logic 125-8 pada (word) def. 400-1
Murti, T.R.V. 318 padärtha (category) 269, 272, 378-9
mysticism 317 pada-vädin vs vâkya-vâdin 379
Padmarajaiah, Y.j. 301 n
N paksa 27, 50-1, 64, 74, 103, 105, 125,
129n, 166, 1S9; paksatà 129n, 139
Nägärjuna 18-9, 57, 197. 230, 278, paksa-sattva 26-7
282, 298, 304, 311, 316,' 351 panca-skandha 334-5, 341
Nâgasena 10 pancävayava-väkya 2-4
Nägesa 145n, 383n, 385, 387, 388 Pänini 61, 164n, 177, 380-2, 386, 399n
Narain, A.K. 333n Pümni-sütra 305, 380-1, 385n, 386n,
naya (Jainism) 282-3 388n, 400
nàya-vâda 301-10, 315 paramänu-väda 241, see 'atomism'
dur naya 315 para marsa 391
paraphrase 408-13, see sâbdabodha. pratiyogin (negatum, absentee) 82, 84,

parärthänumäna 23-4 189
parïksâ 165 pratiyogitä 83, 133, 136, 170, 190
Parmenides 271 pratyaya (causal condition) 218, 225,
partial location 119-22; 288; hetu and pratyaya 323, 340-1,
partially locatable property 119; 352, 357
see avyäpyavrtti pratyaya (inflection) 392, 396
Patanjali (Grammarian) 322-4, 382-7, pravrtti-nimitta 242
394, 400 Prayoga-sâra 30
Patanjali (Yogasütra) 331, identity of prayojana (purpose) 15, 38
two Patanjali s 385n 'predicate' (jsâdhya) 1
paryäya (modes) 311, see also guna presence-range of properties 122, 116-
(Jainism) 8, 122-6
Peacocke C. 154n presupposition 125
Peano 170n Price, H.H. 223, 234n
perlocutionary act (efFecl) 176, 190 pronouns (two types) 382
pervasion {vyâpti) 113, 122-5, 141-5 proper name 388-9, 398, singular
pervasively (wholly) locatable 120, name 395, 398 also called päri-
petitio principii 45-50, 52-3, 56, 129n, bhäsiki näma 398
189 properties 115-6, 125-8;
Phanibhusan (Tarkavâgisa) 29n, 52n universal or everpresent property
phenomenalism 222 128-40, def. of everpresent p. 134-7
Plato lln, 20-2, 155, 180, 181n, 271 propositional calculus 181 n
Plato's beard 88 potocol language 216
positivism 216 pseudo-inference (nyâyâbhâsa) 28,42-
possible (objects) 79-80, 144 58
Potter, K. 78, 158, 160n psychologism 71-2
Prabhâcandra 310 pudgala (Buddhism) 333-4
Prabhâkara 206, 208-10, 220-1 Püjyapäda 296
pradhäna (unmanifest matter) 287 Putnam, H. 200-1
prajnä 229, 235, 315, 317
Prajnäpäramitä 343
prajnapti-sat 235
prakära (qualifier), 84, 408-10 qualifier-qualincand 114, (prakâra),
prakarana-sama (indecision) 45, 51-6 408-10, chief-qualifier 410-11, 420
prakäsa 204, 206; paratah vs svatah qualifier-qualificand analysis 418
207,215 quality 270, (guna vs paryäya) quality
prakrti (word-base) 392 and modes 279, 295-6;
pramäna 12, 33, 43, 89, 163, 179, 193, Vaisesika notion of, 296-7
203-4, 208-9, 219, 253, 263-4, 372, Grammarians' notions of, 379-89
382; pramä 203, 208-9, 263-4, 266; quantification 167-69
pramätva 203, 206, 214, p-vâkya 404 Quine, W.V. 81, 88, 125, 127, 128,
prameya 207, 253, 254-5 131n, 145, 166, 167n, 170n, I81n,
prasajya-pratisedha 304-5 198-200, 382n, 423, 425-6
prasanga (reductio) 299
Prasastapäda 40, 50, 141, 144, 272n,
273,275, 286n, 331-2, 367 R
pratibandaka (causal antidote) 289
pratibhäsa (appearance) also äkarä Rändle, H.N. 45, 49, 51-3, 55-6, 58-9,
217-8, 220, 224-5, 227 61, 65, 89
pratijpâ 2, 23, 48, 60 Raghudeva 162n
pratijnâ-virodha (contradiction) 14, Raghunätha 122, 127, 134n, 135n,
20-2 137, 149-50, 158, 160-3, I73n, 389,
prätipadika (also called mima) 393, 423
401 Rahula, W. 319n, 320, 324n
pratipaksa 50 Raja, K.K. 405n
pratisthdpanä 2 Rämänujäcärya 208
prailtya-samutpäda (dependent origi- Ramsey, F.P. 164
nation) 287, 298; as twelve 'causal Ratnagoiravibhâga 343-5
chain* 325-6, 340-2, 352 Ratnäkarasänti 228
Index 445
realism 86,100,173-4, 235, 238, naive Samyuktü-nikäya 325, 327, 333

realism 234, 248 Sänätani 16
reductio299 • sarridhäya-sambhäsä (friendly debate)
representative theory 235, 237 11-2
Rg-veda 270, 364 Sandhininnoccmasiitra 343, 345-6
rigid designator 200 sarrigati, prasanga and upodghata 38
redundant (adhika, a 'defeat' situ- Sanjana, J.E. 37On
ation) 13 Sanjaya 17-8,. 302-5
Reid, T. 292 Sântaraksita 206, 225-6, 310, 336
Robinson, R. 20-1, 181, 195n sapaksa 131
Robinson, R (Buddhologist) 32 sapaksa-sattva 26-7
rüdha 393-4 saprakäraka 'qualificatory perception'
Russell, B. 80, 87, 88n, 102, 138,170n, 213
174, 223, 234n, 291 saptabhangl 300-14
Sarvâstivâda 337-8, 340n
Sasadhara 143-4, 289-90
Sasthitantra 141
Sabara 34, 40, 204 Sastri, Haraprasad 34n
säbdabodha (hearer's cognition of the Sastri, Gaurinath 386n
meaning) 157, 390-1, 398, 404-6, sat'käryaväd* 44, 204, 271, 287
408,411-4,416, 418-9,422,424,430 sat-pratipaksa 26, 28
sabda-pramâna (verbal testimony) 23, sattä 173, 174n, 195;
390-1 sattâ'sambandha vs.
sädhya ('predicate', inferable pro- svâtma-sattva 273
perty) 24-5, 27, 44, 48, 50-1, 60, 64, Sauträntika 86, 221-4, 226-7, 229,237,
91-3, 125, 128-9, 160 276, 337-9, 343
sâdhyasama 45, 48-9, 51, 53, 56-7 savikalpa perception 210-12
Saivism 352 sa-vyabhicara 'deviating reason' 43,
sakti (power) 289 53-5
Sälikanätha 206 scepticism 17, 318
sâmagri (cause, Jayanta) 375-6 sceptic 316, sceptic-mystic 17
samaniyata (equi-locatable, co-exten- Stcherbatsky, T. 270n
sive), 167-72, 186 Scholastics 155
sämänya-visesa (see jäti 'universals') Schuster, N. 41 n
272 Searle, J.R. 17n, 18, 176
samanantara-pratyaya 335, 337, 339 self-cognition 237, 246-7
Samantabhadra 282, 294, 299, 306-7 self-consciousness 237
samaväya (inherence) 273 self-contradictory 84-5, 93-4, 96, 105-
samaväyi-kärana, 275, 285, 297, 376, 6
a-samaväyi-kärana 285-6, nimitta sense-data 216-8, 222-4
kärana 285-6 sentence (väkya) def. of 398-416
sameness of reference 166 Siddhasena (Diväkara) 281-3, 294,
sairikara ('intermixture') def. 172-6 300, 315
Sarnkara (Brahmasütra) 310, 370 'sign* (liriga) 70-1
Sâmkhya 44, 141, 144, 205, 271, 275, singular inference 125
287-8, 301 n, 367; Sàmkhya-Yoga singular term 395, 398
207 skandha 338; see also
samkräntiväda (Sauträntika) 338 panca-skandhci ; ekarasa-skandha
Sâmmitïya 333-4 339
samsara (transmigration) 340-2, 349, smrti (memory) 263-7, as a pramäna
351-6 262-7'; as dharmasästra 263
samsargäbhäva 157, 162 Socrates 11, 21, 180, 181n, 182n, 195,
sarrisarga-maryädä 412 196n
sarrisaya (doubt) 49, 140, 242 solipsism 221, 238
sarrisaya-sama 53 Solomon, E. 5n, lln, 3l0n
sarrisaya-vädin (agnosticism) 304 Sondada (Sondala) 83-5, 119
samskära 251, 266, 323, 326-7, 334-6, sophistical refutation (jäti) 56
341 soul vs non-soul 277-8
samvrti-sat vs paramàrtha-sat 221, 225 soundness (of agrument) 90
samyaktva (truth) 208 Sridhara 142n, 143, 210, 273
Srlharsa 173n 95, 96n, 100, 103, 105, 128, 131,

Srilâbha (Srilàta?) 330, 338 136, 138, 143-4, 148-9, 152, 173n,
srnga-grähaka 196n 182-3, 185-91, 193, 265-68, 273,
àr'uti (Vedas) 263 275, 277-9, 288, 289, 374, 421, 423
Staal, J.F. 7, 164n, 165, 167, 172n, uddesa (enumeration) 165, 196; vs.
178n, 403n, 409n, 410n vibhâga 196
sthäpanä 'demonstration' 2 Uddyotakara 5, 12-5, 17, 23-6, 30-1,
Steinkellner, E. 142n 36-9, 50, 58-64, 130-2, 178-9, 189,
Sthaviraväda 335-7 192, 195, 199, 210, 231, 238, 240-3,
Sthiramati 340-5 257
Strawson, P.F. 55, 9in, 102, 123n, Umäsväti 279, 294, 300, 302
124, 298, 299n, 312n 'unexampled' property 94
Subhagupta 225 unconditionality (in def. of cause)
subsistence 97 290, 292-3
substance (drayya) 270, 274-6, 294- unconscious states (murclia, middha,
300; Jaina view of, 279-81, nirodha-samâpatti etc.) 339, 341
Grammarians' view of, 378-89; universal (see/a*/ also) 273 (in causa-
Aristotle's view of, 297; lity) 292
substantialist vs. anti-substantialist uni versais of language 390
270-2, 274 unlocatables (aprasiddha) 112, 117,
Sukhlalji (Sanghavi) 263-4 119
sûnyatâ 231 ; sünya-väda 229, 231, 318 unnegatables (everpresent properties)
'suppressed premise' 91 112, 118, 121,126
svabhäva 298, 317, 384 unsaturated (incompleteness) 420-1
svaprakâsa 206, 215 upädhi 'associate condition' 142-3,145
svarüpäsiddha 48 Upadhye, A.N. 301, 3O2n
sva-samvitti (self-cognition, self-con- upamäna 24-5
sciousness) 230-8, 243 upanaya 3
syäd-väda 301-11 Upqnisad 210, 287;
syllogism (Indian) 2-8, 20, 46 Sveiäsvatara 287;
synonymity 166 Brhadäranyaka 364; Upanisadic
absolutism 345-6
Upäyahrdaya 9-10
uttejaka (antidote to antidote) 289
Takasaki, J. 342n, 343-4
Tarski, A. 214
Tathägatagarbha 334, 342-6
tätparya (speaker's intention) 405 Vâcaspati 23, 26,31, 59, 64-6, 94,98n,
Tattvärthasütra 279-81, 295-6, 300 129n, 138, 141n, 142n, 179, 182,
tautology (as a fault) 191-2 188, 192,195, 208, 210, 265
temporality (kädäcitkatva) 288 vcidâ (debate) 12, 19, 34n, 89, 105
termini (terminus) 168-9, 174-5 Vädideva 310
tetralemma 146 Vâghbata 236
trairüpya (triple character) 68-76 Vaibhäsika 229, 236, 336, 337n, 339,
transmigration 338, 340-6, 351-6 340-1'; (Kasmïra) 235
tricky debates 13 Vaisesika-sütra 33, 37, 40n, 141, 272,
Trilocana 142, 143n, 144, 210 274n, 296, 367
tri-svabhävatä 343, 345 Vaisnavism 352
trsnä (craving) 324 Väjapyäyana 386
truth 203 Vallabha 130, 143-4
truth-functional logic 169n Vardhamäna 59, 65-68
truth-functional tautology 192n väsanä 247
Ts'ing-mu 30-1, 38 Vasubandhu 19, 216, 219-20, 228,
Tucci, G. 9a, 30n, 34, 23In, 237n 230-5, 239, 241-2, 261, 284-5, 293,
two-valued logic 113 324, 328-30, 336n, 337n, 338, 340n,
342n, 343, 357-60
Ü Väsudeva Dîksita 323
Vasudhararaksita 59n, 129n, 132n
udäharana (exemplification) 3, 23, 48 Vasumitra 338
Udayana 14, 17n, 19, 65n, 86, 89, 93, Vätsiputriyas 333, 338
Index 447
Vätsyäyana 5, 12, 15-7, 22-4, 29-30, vyavahära 187, 200, 207, 211;
34-6, 38-9, 43-4, 47-9, 51-4, 56, 64n, vyavahära-siddhi 191
165, 177-8, 192, 210, 218n, 253, vyäpti 125, 130, 141, 167, 188
332,354-5,359,361,417 vyäpya 272
Vedänta 271 Vyäsa 346
Vedânta-sûtra 367 vyävrtti 187, 200
Vedas (sruti) 25, 263, 275, 287, Vyomasiva 273, 286n, 310-11
Vedic 352
Venkataramanam 333 W
Vidyabhusana, S.C. 2, 5n
Vidyânanda 308-10, Warder, AK. 40n, 321, 326n
vigraha, vigrhya sambhäsä 11 Whitehead, A.N. 170n, 176, 180,
Vijfianavâdâ 206-7, 229* 181n, 283
vijnapti 229, 232 'whole' {avayavin) 275
vijnaptimätratä 220, 229, 232, 248 Wittgenstein, L. 197
vikalpa 263
viksepa (evasion) 14
vipâka (maturity of the seed) 248,
337-40 Yâjtiavalkya 10, 364
vipaksa 131-2, 138 Yâska 382
vipaksäsattva 26-7, 71, 75 Yasomitra 329, 33On, 335n, 336n,
viparyaya ('contrapositive') 253 337n, 338, 359n, 360
viruddha-hetu (incoherent reason) 44 yäthärthya (truth) 208
visaya vs. älambana 219; yaukika näma 394-5
visayatä 213-4, 417-8; visesya 408- Yogäcära 103, 226-8, 231, 239-40,
10 248, 334-5, 337-9, 343, 345, 359n
visista-vaisistyävagähin 84 yogarüdha 394
Visvanâtha'405n Yogasütra 346, 348
vitandä 12, 16-7, 21, 105, vâda-v and yogayatä (compatibility) 82, 405-6,
jâlpa-v. 16-7, 19 416, 422-4, 427, 430
von Wright, G.E. 291 Yuktidïpikâ 5
Vrsagana 41
vyabhicära (deviation) 54-5
Vyädi 380, 386 zero occurrences of inflections 401,
vyäpära (operation) 374-5 403

Matilal, Bimal Krishna - Logic, Language, and Reality - Indian Philosophy and Contemporary Issues-Motilal Banarsidass (1990) PDF

Uploaded by

Copyright:

Available Formats

Matilal, Bimal Krishna - Logic, Language, and Reality - Indian Philosophy and Contemporary Issues-Motilal Banarsidass (1990) PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Matilal, Bimal Krishna - Logic, Language, and Reality - Indian Philosophy and Contemporary Issues-Motilal Banarsidass (1990) PDF

Uploaded by

Copyright:

Available Formats

LOGIC, LANGUAGE AND REALITY

Indian Philosophy and Contemporary Issues

BIMAL KRISHNA MATILAL

MOTILAt BANARSIDASS PUBLISHERS

© MOTILAL BANARSIDASS PUBLISHERS PVT. LTD.

ISBN: 81-208-0008-7 (Cloth)

TAMAL AND ANVITA

CHAPTER FOUR: ONÏOLOGICAL PROBLEMS 268

SELECT BIBLIOGRAPHY 431

Ï take this opportunity to note the important issues raised by

in the same way a cognitive state is usually known, i.e., by an

called nyäya-vakyä or Indian syllogism as :

forgotten lest we repeat ourselves and repeat even the mistakes of

4 September, 1989 All Souls College,

This is a book on Indian philosophy. Its aim is to combine

ways of knowing what is the case is to rightly understand

LOGIC IN ANCIENT AND

§1.1 : ANCIENT INDIAN LOGIC AND THE QUESTION

Unlike the Arabic tradition in Philosophy, Indian philosophic

In 1920, SKC. Vidyäbhüsana published his pioneering work A

3. Exemplification (udäharana) or Citation of an example

generally accepted, to the contradictory of a given thesis."

for certain obvious peculiarities it has, viz., the requirement of

such role here at all, is quite different (see F. Staal 1 ). The

F. Staal, "The Concept of Paksa in Indian Logic", Journal of Indian

Indian system, however, the conclusion is given, more often

bambrough, J.B., New Essays on Plato and Aristotle, London (Routledge

§1.2 : RULE OF DIALECTIC AND DEBATE

I have chosen the term 'dialectic' to refer to the art of philo-

debate, types of arguments, tricky devices of debate and the

proposition, and if anyone differs from them, they

about what must be done by a debater before accepting to

One may wonder whether in this type of debate there may

equivocation, and twenty-four varieties of rejoinder based

Udayana, pp. 627-628 (Nyayadariana, A. Thakur's edn.).

This raises two related questions. First, is it possible to

it takes another equally gifted philosopher to uncover that

for the proud people who intend to defeat others.1 Tricky

This is an illocutionary negation and should be distinguished

"I do not say there is no after-life".

alternatives, 1,2 or 3. But they have already been rejected.

This classification of debate into the one for seeking after

The most pervasive form of refutation practised by Socrates

in the thing. It is a search for definitions. In fact, dialectic is

§1.3 : ARGUMENT AND PSEUDO-ARGUMENT

It is, perhaps, Vâtsyâyana who first uses the expression nyäyä-

Pramânair arthaparïksanaifi nyäya h—Nyäya-Bhäsya under Nyäya-Sü tra

One of the best ways to explain definitely what is denoted by

The Buddhist logician, in Nyäyapravesa, gives a different

another inference or by (2) comparison (upamäna) ? Will these

defect is contained in the foregone determination of the absence

the desired inference to follow. So, enquiry into this line to

thesis' is regarded as a logical defect of an argument, because

§ 1.4 : THE EARLY NYÄYA THEORY OF INFERENCE

Nyäyasütra 1.1.5 is very difficult to explain. We find here a

Vâtsyâyana, alternative explanations about them were offered.

UH A. General example of inference given before the

(b) Same asUH(b).

inferring the effect from the cause is pointed out,