The Savage Mind PDF

$5.
95
In The Savage Mind, Claude Lévi-
Strauss, modern anthropology's
most revered thinker, takes as his
central theme the very nature of
thought itself. He demonstrates how
each culture has its own system of
concepts and categories derived
from experience and imposed by the
surrounding natural world. The per-
ception that the world is populated
by "species'" of things is a primitive
one, yet it is from a similar classifi-
cation into species and names that
the most sophisticated and abstract
thought evolves. Lévi-Strauss shows,
through the order in the naming of
plants and animais, concepts of
space and time, myths and rituals,
how primitive societies do engage in
a high level of abstract reasoning
different from but not necessarily
inferior to that involved in culti-
vated "systematic thought."
Throughout this book Lévi-
Strauss is concerned with the flux of
history and with the perpetual strug-
gle between established social sys-
tems and their history. He takes
issue with Sartre's Critique de la
raison dialectique, refuting the idea
that there is a dichotomy between
"civilized" and "primitive," non-
historic thought. The role of "his-
tory" in anthropology is constantly
debated, and Lévi-Strauss's intensely
challenging views are a brilliant con-
tribution to that debate.
Lévi-Strauss brings new vision to
the traditional problems of the social
anthropologist and to areas of
human expression such as art, ri tuai,
mythology. He shows that the most
fascinating work is still to be done
in investigating these areas which
have been hitherto neglected by
materialist anthropology.
THE NATURE OF HUMAN SOCIETY
THE SAVAGE MIND

THE NATURE OF HuMAN SOCIETY SERIES
Editors: Julian Pitt-Rivers and Ernest Gellner

THE SAVAGE
MIND
Claude Lévi-Strauss
THE UNIVERSITY OF CHICAGO PRESS

THE UNIVERSITY OF CmcAGO PRF.ss, CHICAGO 60637
WEIDENFELD AND NICOLSON LTD., LONDON W.1
Translated from the French, La Pensée sauvage

© 1962 by Librairie Pion, 8, rue Garancière, Paris-6•
English Translation © 1966 by George Weidenfeld and Nicolson Ltd.
Ali rights reserved
Published 1966
Library of Congress Catalog Card Number: 66-28197
Printed in the United States of America

To the Memory
of
Maurice Merleau-Ponty
CONTENTS
PREFACE Xl
THE SCIENCE OF THE CONCRETE 1
2 THE LOGIC OF TOTEMIC CLASSIFICATIONS 35

3 SYSTEMS OF TRANSFORMATIONS 75
4 TOTEM AND CASTE 109
5 CATEGORIES, ELEMENTS, SPECIES, NUMBERS 1 35

6 UNIVERSALIZATION AND PARTICULARIZATION 161
7 THE INDIVIDUAL AS A SPECIES
8 TIME REGAINED
9 HISTORY AND DIALECTIC
BIBLIOGRAPHY
INDEX
ILLUSTRATIONS
(Betu:een pages I 48 and 149)
Francois Clouet. Portrait of Elizabeth of Austria. (Photo:

~lfuséedu LoU'l:re)
2 Club used for killing fish (Photo: Huillard)
3 The opposite of toternisrn: !\" aturalized Man. Sketch by I ,e
Brun
4 Humanized ~ature. Sketch by Grandville (Bibl. Nationale)
5 Alphabet of Birds (Musée Sational des Arts et Traditions
Populaires. Photo: Huillard)
6 Society of Animais (."fi.fusée National des Arts et Traditions
Populaires. Photo: Huillard)
7 Australian Churinga (Photo: Bandy)
8 Aranda water-colours (Phutos: Australian Information Service)
PREFACE
This book is complete in itself, but the problems it discusses are

closely linked to those which I surveyed more hastily in a recent
work entitled Totemism (trans. Rodney l\eedham, London, 1964).
Without wishing to oblige the reader to refcr to it, it is proper to
draw his attention to the connection between the two: the first
forms a kind of historical and critical introduction to the second. I
have not, therefore, deemed it necessary to return, here, to the
theories, definitions and facts which have already been dealt with
at sufficient length.
?\ evertheless the reader should know what is expected of him
on opening these pages: that he acquiesce in the negative conclu-
sion which the first volume reached in regard to totemism; for,
once it is clearwhy I bclieve that the anthropologists of formertimes
were the prey to an illusion, it is time for me to explore totemism's
positive sicle.
l\;'o one will suppose that, by placing the name of Maurice
:\lerleau-Ponty on the first page of a book whose final chapter is
devoted to a work of Sartre, I have intended to oppose them to one
another. Those who were close to Merleau-Ponty and myself
during recent years know some of the reasons why it was natural
that this book which develops freely certain themes of my lectures
at the Collège de France should be dedicated to him. It would have
been, in any case, had he lived, as the continuation of a dialogue
whose opening goes back to 1930 when, in company with Simone
de Beauvoir, we were brought together by our teaching activities,
on the eve of receiving our final degrees. And, since death has
tom him from us, may this book at least remain devoted to his
memory as a token of good faith, gratitude and affection.
XI
THE SAVAGE J\IIND
If I have felt obliged to give expression to my disagreement with

Sartre regarding the points which bear on the philosophical
fundaments of anthropology, I have only determined to do so
after sevcral readings of the work in question which occupied my
pupils at the Ecole des Hautes Etudes and myself during many
sessions of the year 1960-1. Over and above our inevitable diver-
gences I hope that Sartre will recognize above all that a discussion
to which so much care has been given constitutes on behalf of all
of us a bornage of admiration and respect.
I would Iike to express my warm thanks to my colleague, Jacques
Bertin, professor at the Ecole des Hautes Etudes who was kind
enough to make some of the diagrams for me in his laboratory;
I. Chi va and J. Pouillon whose notes recalled to me some improv-
ised points which were otherwise Iost; ?vlme Edna H. Lemay who
typed the manuscript; :'.\1Ile Nicole Belmont who helped me with
the tasks of assembling the documentation and making the biblio-
graphy and the index; and my wife who aided me in rereading the
text and correcting the proofs.
XII
CHAPTER ONE
THE SCIENCE OF THE CONCRETE
lt has long been the fashion to invoke languages which lack the
terms for expressing su ch a concept as 'tree' or 'animal', even
though they contain ail the words necessary for a detailed inventory
of species and varieties. But, to begin with, while these cases are
cited as evidence of the supposed ineptitude of 'primitive people'
for abstract thought, other cases are at the same time ignored
which make it plain that richness of abstract words is not a
monopoly of civilized languages. In Chinook, a language widely
spoken in the north-west of North America, to take one example,
man y properties and qualities are referred to by means of abstract
words: 'This method', Boas says, 'is applied to a greater extent than
in anyother language 1 know.' The proposition 'The bad mar, killed
the poor child' is rendered in Chinook: 'The man's badness killed
the child's poverty'; and for 'The woman used too small a basket'
they say: 'She put the potentilla-roots into the smallness of a clam
basket' (Boas 2, pp. 657-8).
In every language, moreover, discourse and syntax supply indis-
pensable means of supplementing deficiencies of vocabulary. And
the tendentious character of the argument referred to in the last
paragraph becomes very apparent when one observes that the
opposite state of affairs, that is, where very general terms outweigh
specific names, has also been exploited to prove the intellectual
poverty of Savages:
Among plants and animais he [the lndian] designates by name only
those which are useful or harmful, ail others are included under the
classification of bird, weed, etc. (Krause, p. 104).
A more recent observer seems in the same way to believe that the
THE SAVAGE MIND
native gives names and forms concepts solely in accordance with

his needs:
1 well remember the hilarity of Marquesian friends ... over the (to
them) fatuous interest of the botanist of our expedition in 1921, who was
collecting nameless ('useless') 'weeds' and asking their names (Handy
and Pukui, Part VI, p. 127n).
However, Handy compares this indifference to that which speci-
alists in our civilization show towards phenomena which have no
immediate bearing on their own field. When his native collaborator
stressed the fact that in Hawaii 'every botanical, zoological or
inorganic form that is known to have been named (and person-
alized), was some thing ... used', she is careful to add 'in some
way'. She goes on to say that the reason why 'there was an infinite
variety of living things in forest and sea, of meteorological or
marine phenomena, which were unnamed' was that they were
regarded as being of no 'use or interest' - terms which are not
equivalent, as 'use' concerns practical, and 'interest' theoretical,
matters. '\Vhat follows confirms this by concentrating on the latter
aspect at the expense of the former: 'Living was experience fraught
with exact and definite significance' (id., p. 126-ï)·
In fact, the delimitation of concepts is different in every
language, and, as the author of the article 'nom' in the Encyclopédie
correctly observed in the eighteenth century, the use of more or
less abstract terms is a fonction not of greater or lesser intellectual
capacity, but of differences in the interests - in their intensity and
attention to detail - of particular social groups within the national
society: 'In an observatory a star is not simply a star but~ of Capri-
corn or y of Centaur or ~ of the Great Bear, etc. In stables every
horse has a proper name - Diamond, Sprite, Fiery, etc.' Further,
even if the obsern.tion about so-called primitive languages referred
to at the beginning of the chapter could be accepted as it stands,
one would not be able to conclude from this that such languages are
deficient in general ideas. '\Vords like 'oak', 'beech', 'birch', etc.,
are no less entitled to be considered as abstract words than the
word 'tree'; and a language possessing only the word 'tree'
would be, from this point of view less rich in concepts than one
which lacked this term but contained dozens or hundreds for the
individual species and varieties.
The prolifcration of concepts, as in the case of technical langu-
ages, goes with more constant attention to properties of the world,
2
with an interest that is more alert to possible distinctions which

can be introduced between them. This thirst for objective know-
ledge is one of the most neglected aspects of the thought of people
we call 'primitive'. Even if it is rarely directed towards facts of the
same level as those with which modern science is concerned, it
implies comparable intellectual application and methods of obser-
vation. In both cases the universe is an object of thought at least
as much as it is a means of satisf ying needs.
Every civilization tends to overestimate the objective orientation
of its thought and this tendency is never absent. When we make the
mistake of thinking that the Savage is governed solely by organic
or economic needs, we forget that he levels the same reproach at
us, and that to him his own desires for knowledge seems more
balanced than ours:
These native Hawaiians' utilization of their available natural assets was
well-nigh complete - infinitely more so than that of the present commer-
cial era which ruthlessly exploits the few things that are financially profit-
able for the time being, neglecting and often obliterating the rest (Handy
and Pukui, Part VIII, p. 6z).
Cash-crop agriculture is hardly to be confused with the science of
the botanist. But, in ignoring the latter and taking only the former
into account, the old Hawaiian aristocrat is simply repeating, and
turning to the advantage of a native culture, a mistake of the same
kind that l\1alinowski made when he claimed that primitive peoples'
interest to totemic plants and animais was inspired by nothing but
the rumbling of their stomachs.
Tessman (Vol. 2, p. 192) mentions 'the accuracy with which (the

Fang of the Gabon) identify the slightest differences between
species of the same genus'. The two authors quoted above make a
similar observation about Oceania:
The acute faculties of this native folk noted with exactitude the generic
characteristics of ail species of terrestial and marine life, and the subtlest
variations of natural phenomena such as winds, light and colour, ruffiing
of water and variation in surf, and the currents of water and air (Handy and
Pukui, Part VI, p. 126).
Among the Hanunoo of the Philippines a custom as simple as that

of betel chewing demands a knowledge of four varieties of areca
nut and eight substitutes for them, and of five varieties of betel
and five substitutes (Conklin, 3):
3
THE SAVAGE l\IIND
Almost ail Hanunôo activities require an intimate familiarity with local

plants and a precise knowlcdgc of plant classification. Contrary to the
assumption that subsistcncc level groups never use but a small segment of
the local flora, ninety-three per cent of the total number of native plant
types are recognized by the Hanun6o as culturally significant (Conklin 1,
p. 249).
This is equally true of fauna:
The Hanun<io classify ail forms of the local avifauna into seventy-five
categories ... (they) distinguish about a dozen kinds of snakes ...
sixty-odd types of fish ... more than a dozen ... types of fresh and
sait water crustaceans ... a similar number of ... types of arachnids
and myriapods ... The thousands of insect forms present are grouped
by the Hanun<io into a hundred and eight named categories, including
thirteen for ants and termites ... Salt water molluscs ... of more than
sixty classes are recognized by the Hanun6o, while terrestrial and fresh
water types number more than twenty-five ... Four distinct types of
bloodsucking leeches are distinguishcd ... : altogether 461 animal types
are recorded (id., pp. 67-70).
A biologist writes the following about pygmies of the Philippines:
Another characteristic of .:'\egrito life, a characteristic which strikingly
demarcates them from the surrounding Christian lowlanders, is their
inexhaustiblc knowledge of the plant and animal kingdoms. This lare
includes not only a specific recognition of a phenomenal number of
plants, birds, animais, and insects, but also includes a knowledge of the
habits and behaviour of each . . . .
The ::\'egrito is an intrinsic part of his environment, and what is still
more important, continually studies his surroundings. :\Iany times 1 have
seen a ).°egrito, who, when not being certain of the identification of a
particular plant, will taste the fruit, smell the leaves, break and examine
the stem, comment upon its habitat, and only after ail of this, pronounce
whethcr he did or did not know the plant.
The natiYes are also interested in plants which are of no direct use
to them, bccause of their significant links with the animal and
insect world, and ha\·ing shown this, the same author continues:
The acute obserrntion of the pygmies and their awareness of the inter-
relationships between the plant and animal life ... is strikingly pointed
out by their discussions of the living habits of bats. The tididin lives on
the dry !caves of palms, the dikidik on the underside of the leaves of the
wild banana, the litlit in bamboo clumps, the ko/umboy in holes in trees,
the korzanaba in dark thickets, and so forth. ln this manner, the Pinatubo
Kegritos can distinguish the habits of more than fifteen species of bats.
Of course, the classification of bats, as well as of insects, birds, animais,
fish and plants, is detennined primarily by their actual physical differences
and or similarities.
:\lost ).:egrito men can with case enumerate the specific or descriptive
4
THE SC'IE:\"CE OF THE COI\'CRETE
names of at lcast four hundred and fifty plants, seventy-fivc birds, most
of the snakes, fish, insccts, and animais, and of even twenty species of
ants ... • and the botanical knowlcdgc of the mananambal, the 'medicine
men and women, who use plants wnstantly in their practice, is truly
astounding (R. B. Fox, pp. 187-8).
Of a backward people of the Tyukyu archipclago, we read:
En·n a child can frcqucntly identify the kind of trcc from which a tiny
wood frag'JTlcnt has corne and, furthermore, the sex of that tree, as defined
by Kabiran notions of plant scx, by observing the appearance of its wood
and bark, its smcll, its hardncss, and similar characteristics. Fish and
shellfish by the dozcn arc known by individually diHinctive terms, and
their separatc fcaturcs and habits, as wcll as the scxual <liffcrcnccs within
each type, arc wcll rccognizcd (Smith, p. 150).
Sevcral thousand Coahuila lndians never exhausted the natural
resources of a desert region in South California, in \Vhich today
only a handful of white families manage to subsist. They lived in
a land of plenty, for in this apparently completely harren tcrritory,
they were familiar with no less than sixty kinds of cdible plants and
twenty-cight others of narcotic, stimulant or medicinal properties
(Ilarrows). A single Seminol informant could identify two hundred
and fifty species and varieties of plants (Sturtevant). Three
hundred and fifty plants known to the Hopi Indians and more than
five hundred to the ~avaho have been recorded. The botanical
vocahulary of the Subanun of the Southern Philippines greatly
exceeds a hundred tcrms (Frakc) and that of the l lanunôo
approaches two thousand.t Sillans, working with a single inform-
ant in the Gabon, recently published an ethno-botanical list of
about eight thousand terms, distributed between the languages or
dialects of twelve or thirteen neighbouring tribe& (Walker and
Sillans). The, for the most part unpublished, results of Marcel
Griaule and his co-workers in the Sudan promise tu be equally
. .
1mpress1ve.
Their extreme familiarity with their biological environment, the
passionate attentio:i. which they pay to it and their precise know-
ledge of it has often struck inquirers as an indication of attitudes
and preoccupations which distinguish the natives from their white
visitors. Among the Tewa lndians of New Mexico:
• Also at least forty-five types of edible ground-mushrooms and ear-fungi
(I.e., p. 231) and on the technological plane, more than fifty types of arrows (id.,
pp. 265-8).
t See below, pp. 138, 153.
5
THE SAVAGE MIND
Small differences are noted ... they have a name for every one of
the coniferous trees of the region; in these cases differences are not con-
spicuous. The ordinary individual among the whites does not distinguish
(them) ... lndeed, it would be possible to translate a treatise on botany
into Tewa ... (Robbins, Harrington and Freire-Marreco, pp. 9, 12).
E. Smith Bowen scarcely exaggerates in the amusing description

she gives of hcr confusion when, on her arrivai in an African tribe,
shc wanted to begin by learning the language. Her informants
found it qui te natural, at an elementary stage of their instruction,
to collect a large number of botanical specimens, the names of
which they told her as they showed them to her. She was unable
to identify them, not because of their exotic nature but because she
had never taken an interest in the riches and diversities of the plant
world. The natives on the other hand took such an interest for
granted.
These people are farmers: to them plants are as important and familiar
as people. l'd never been on a farm and am not even sure which are
begonias, dahlias, or petunias. Plants, like algebra, have a habit of looking
alike and being different, or looking different and being alike; conse-
quently mathematics and botany confuse me. For the first time in my life
1 found myself in a community where ten-year-old children weren't my
mathematical superiors. I also found myself in a place where every plant,
wild or cultivated, had a narne and a use, and where every man, woman
and child knew literally hundreds of plants ... (my instructor) simply
could not realize that it was not the words but the plants which baffled
me (Smith Bowen, p. 19).
The reaction of a specialist is quite different. In a monograph in

which he describes nearly three hundred species or varieties of
medicinal or toxic plants used by certain peoples of Northern
Rhodesia, Gilges \\•rites:
It has always been a surprise to me to find with what eagerness the
people in and around Balovale were ready and willing to talk about their
medicines. \Vas it that they found my interest in their methods pleasing?
\Vas it an exchange of information amongst colleagues? Or was it to show
off their knowledge? \Vhatever the reason, information was readily forth-
coming. I remember a wicked old Luchozi who brought bundles of dried
leaves, roots and stems and told me about their uses. How far he was a
herbalist and how far a witch-doctor 1 could never fathom, but 1 regret
that I shall never possess his knowledge of African psychology and his art
in the treatment of his fellow men, that, coupled with my scientific
medical knowledge, might have made a most useful combination (Gilges,
p. 20).
6
Conklin quotes the following extract from his field notes to illus-
trate the intimate contact between man and his environment which
the natiYe is constantly imposing ~m the ethnologist:
At o6oo and in a light rain, Langba and 1 left Parina for Binli ... At
Aresaas, Langba told me to eut off several 10 x 50 cm. strips of bark from
an anapla kilala tree (Albizzia procera (Roxb.) Benth.) for protection
against the leeches. By periodically rubbing the cambium side of the
strips of sapanceous (and poisonous: Quisumbling, 1947, 148) bark over
our ankles and legs - already wet from the rain-soaked vegetation - we
produced a most effective leech-repellent lather of pink suds. At one spot
along the trait near Aypud, Langba stopped suddenly,jabbed his walking
stick sharply into the side of the trait and pulled up a smalt weed, tawag
kugum buladlad (Buchnera urticifolia R. Br.) which he told me he will use
as a lure ... for a spring-spcar boar trap. A few minutes latcr, and we
were going at a good pace, he stopped in a similar manner to dig up a
small terrestrial orchid (hardly noticeable beneath the other foliage)
known as liyamliyam (Epipogum roseum (D. Don.) Lindl.). This herb is
useful in the magical control of insect pests which destroy cultivated
plants. At Binli, Langha was careful not to damage those herbs when
searching through the contents of his palm leaf shoulder basket for apug
'slaked lime' and tabaku (Nicotiana tabacum L.) to offer in exchange for
other betel ingredients with the Binli folk. After an evaluative discussion
about the local forms of betel pepper (Piper betle L.) Langba got per-
mission to eut swect potato (Jpomoea balatas (L.) Poir.) vines of two
vegetatively distinguishable types, kamuti inaswang and kamuti lupaw
... In the camote patch, wc eut twenty-five vine-tip sections (about
75 cm. long) of each variety, and carefully wrapped them in the broad
fresh leaves of the cultivated saging saba (iWusa sapientum compressa (Blco.
Teoforo) so that they would remain moist until we reached Langba's
place. Along the way we munched on a few stems of tubu minuma, a type
of sugar cane (Saccharum officinarum L.), stopped once to gather fallen
bunga area nuts (Areca catecl1u L.), and another time to pick and eat the
wild cherrylike fruits from some bugnay shrubs (Antidesma brunius (L.)
Spreng). 'Ve arrived at the l\lararim by mid-afternoon havingspent much
of our time on the trait discussing changes in the surrounding vegetation
in the last few decades! (Conklin 1, pp. 15-17).
This knowledge and the linguistic means which it has at its dis-
posai also extend to morphology. ln Tewa there are distinct terms
for all or almost all the parts of birds and mammals (Henderson
and Harrington, p. 9). Forty terms are employed in the morpho-
logical description of the leaves of trees or plants, and there are
fifteen distinct terms for the different parts of a maize plant.
The Hanun6o have more than a hundred and fifty terms for the
parts and properties of plants. These provide categories for the
7
THE SAVAGE l\11NI1
identification of plants and for 'discussing the hundreds of charac-

teristics v,·hich diffcrentiate plant types and often indicate signi-
ficant features of medicinal or nutritional value' (Conklin I, p. 97).
Over six hundred named plants have been recorded among the
Pinatuho and 'in addition to having an amazing knowledge of
plants and their uses, ... (they) employ nearly one hundred terms
in describing the parts or characteristics of plants' (R. B. Fox,
p. 1 i9)·
Knowledge as systematically developed as this clearly cannot
relate just to practical purposes. The ethnologist v.,.ho has made
the best study of the lndians of the north-eastern United States
and Canada (the Montagnais, ~askapi, Micmac, Malecite, Penob-
scot) emphasizes the wealth and accuracy of their zoological and
botanical knowledge and then continues:
Such knowledge, of course, is to be expected with respect to the habits
of the larger animais which furnish food and the materials of inélustry to
primitive man. \Ve expect, for instance, that the Penobscot hunter of
Maine will have a somewhat more practical knowledge of the habits and
character of the moose than even the expert zoologist. But when we
realize how the Indians have taken pains to observe and systematize facts
of science in the realm of lower animal life, we may perhaps be pardoned
a little surprise.
The whole class of reptiles ... affords no economic benefit to these
I ndians; they do not eat the flesh of any snakes or batrachians, nor do the y
make use of other parts except in a very few cases where they serve in the
preparation of charms against sickness or sorcery (Speck I, p. 273).
And nevertheless, as Speck has shown, the north-eastern Indians

have developed a positive herpetology, with distinct terms for each
genus of reptile and othcr terms applying to particular species and
varieties.
The precise dcfinition of and the specific uses ascribed to the
natural products which Siberian peoples use for medicinal pur-
poses illustrate the care and ingeniousness, the attention to detail
and concern with distinctions employed by theoretical and practical
workers in societies of this kind. We find, for instance: spiders and
whiteworms swallowed as a cure for sterility among the Helmene
and Iakoute; fat of black beetle (Ossete, hydrophobia); squashed
cockroach, chicken's gal! (Russians of Sourgout, abscesses and
hernias); macerated redworms (Iakoute, rheumatism); pike's gal!
(Bouriate, eye complaints); loach and crayfish swallowed alive
( Russians of Siberia, epilepsy and ail diseases); contact with a
8
THE SCIE:-.ICE OF THE CONCRETE
woodpecker's beak, blood of a woodpecker, nasal insuffiation of

the powder of a mummified woodpecker, gobbled egg of the bird
koukcha (Iakoute, against toothache, scrofula, high fevers and
tuberculosis respectively); partridge's blood, horse's sweat (Oïrote,
hernias and warts); pigeon broth (Bouriate, coughs); powder made
of the crushed feet of the bird tilegous (Kazak, bite of mad dog);
dried bat worn round the neck (Russians of the Altaï, fever); instil-
lation of water from an icicle hanging on the nest of the bird remiz
(Oïrote, eye complaints). Taking just the case of bears among the
Bouriate: the flesh of bears has seven distinct therapeutic uses, the
blood fi ve, the fat ni ne, the brains twel\'e, the bile seventeen, the
fur two. It is also the bear's frozen excretions which the Kalar
collect at the end of the winter season to cure constipation
(Zelenine, pp. 4ï-S9). An equally extensive list for an Africa!l tribe
can be found in a study by Loch.
Examples like these could be drawn from all parts of the world
and one may rcadily conclude that animais and plants are not
known as a rcsult of their usefulness; they are deemed to be uscful
or interesting because they are first of ail kno\'m.
It may be objected that science of this kind can scarcely be of much

practical effect. The answer to this is that its main purpose is nota
practical one. It meets intellectual requiremcnts rather than or
instead of satisfying nceds.
The real question is not whether the touch of a woodpecker's
beak does in fact cure toothache. It is rather whether there is a
point of view from which a woodpecker's beak and a man's tooth
can be seen as 'going together' (the use of this congruity for thera-
peutic purposes being only one of its possible uses), and whether
some initial order can be introduced into the universe by meanSOf
these groupings. Classifying, as opposcd to not classifying, has a
value of its own, whatever form the classification may take. As a
recent theorist of taxonomy writes:
Scientists do tolerate uncertainty and frustration, because they must.
The one thing that they do not and must not tolerate is disorder. The
whole aim of theoretical science is to carry to the highest possible and
conscious degree the perceptual reduction of chaos that began in so
lowly and (in ail probability) unconscious a way with the origin of life. In
specific instances it can well be questioned whether the order so achieved
is an objective characteristic of the phenomena or is an artifact constructed
by the scientist. That question cornes up time after time in animal
9
THE SAVAGE MIND
taxonomy ... Nevertheless, the most basic postulate of science is that

nature itself is orderly.... Ali theoretical science is ordering and if,
systematics is equated with ordering, then systematics is synonymous
with theoretical science (Simpson, p. 5).
The thought we cal! primitive is founded on this demand for or der.

This is equally true of all thought but it is through the proper-
ties common to all thought that we can most easily begin to
understand forms of thought which seem very strange tous.
A native thinker makes the penetrating comment that 'All sacred
things must have their place' (Fletcher 2, p. 34). It could even be
said that being in their place is what makes them sacred for if they
were taken out of their place, even in thought, the entire order of
the universe would be destroyed. Sacred objects therefore contri-
bute to the maintenance of order in the universe by occupying the
places allocated to them. Examined superficially and from the
outside, the refinements of ritual can appear pointless. They are
explicable by a concern for what one might call 'micro-adjust~nt'
- the concern to assign every single creature, object or feature to a
place within a class. The ceremony of the Hako among the Pawnee
is particularly illuminating in this respect, although only because
it has been so well analysed. The invocation which accompanies
the crossing of a stream of water is divided into several parts, which
correspond, respectively, to the moment when the travellers put
their feet in water, the moment when they move them and the
moment when the water completely covers their f eet. The invoca-
tion to the wind separates the moment when only the wet parts of
the body feel cool: 'Now, we are ready to move forward in safety'
(id., pp. 77-8). As the informant explains: 'We must address with
song every object we meet, because Tira'wa (the supreme spirit) is
in all things, everything we corne to as we travel can give us help
... '(id., pp. 73, 81).
This preoccupation with exhaustive observation and the system-
atic cataloguing of relations and connections can sometimes lead to
scientifically valid results. The Blackfoot Indians for instance were
able to prognosticate the approach of spring by the state of develop-
of the foetus of bison which they took from the uterus of females
killed in hunting. These successes cannot of course be isolated
from the numerous other associations of the same kind which
science condemns as illusory. It may however be the case that
magical thought, that 'gigantic variation on the theme of the
IO
principle of Causality' as Hubert and lVIauss called it (2, p. 61), car

be distinguished from science not so much by any ignorance 01
contempt of determinism but by a more imperious and uncom·
promising demand for it which 'ëan at the most be regarded a~
unreasonable and precipitate from the scientific point of view.
As a natural philosophy it (witchcraft) reveals a theory of causation
~lisfortune is due to witchcraft co-operating with natural forces. If i
buffalo gores a man, or the supports of a granary are undermined bJ
termites so that it falls on his head, or he is infected with cerebro-spina
meningitis, Azande say that the buffalo, the granary, and the disease, an
causes which combine with witchcraft to kill a man. \Vitchcraft does nO'
create the buffalo and the granary and the disease for thesc exist in thei1
own right, but it is responsible for the particular situation in which the)
are brought into lethal relations with a particular man. The granary woulc
have fallen in any case, but since there was witchcraft present it fcll at th<
particular moment when a certain man was rcsting beneath it. Of thcsi
causes the only one which permits intervention is witchcraft, for witch
craft cmanatcs from a persan. The buffalo and the granary do not allo\\
of intervention and arc, thcrcforc, whilst recognized as causes, nO'
considcrcd the socially relevant oncs (Evans-Pritchard /,p. 418-19).
Seen in this way, the first ditference bctwecn magic and science ii
therefore that magic postulates a complete and all-embracin!
determinism. Science, on the other hand, is based on a distinctior
between levels: only somc of these admit forms of determinism
on others the same forms of determinism arc hcld not to apply
One can go further and think of the rigorous prccision of magica
thought and ritual practices as an expression of the unconsciow
apprehension of the truth of determinism, the mode in whid
scientific phenomena exist. fn this vicw, the operations of deter·
minism are divined and made use of in an all-cmbracing fashior
before being known and properly applied, and magical rites anc
beliefs appear as so many expressions of an act of faith in a scienct
yet to be born.
The nature of these anticipations is such that they may somé·
times succeed. lVIoreovcr they may anticipate not only science itsel:
but even methods or results which scientific procedure does no1
incorporate until an advanced stage of its development. For i1
seems to be the case that man began by applying himself to tht
most difficult task, that of systematizing what is immediately pre·
scnted to the senses, on which science for a long time turned it~
back and which it is only beginning to bring back into its purvicw
In the history of scientific thought this 'anticipation-etfect', has
11
THE S.'\YAGE !\IIND
incidentally, occurred repeatedly. As Simpson (pp. 84-5) has

shown \\·ith the help of an example drawn from nineteenth-century
biology, it is due to the fact that, since scientific explanation is
always the discm·ery of an 'arrangement', any attempt of this type,
e\·en one inspired by non-scientific principles, can hit on true
arrangements. This is eyen to be foreseen if one grants that the
number of structures is by definition finite: the 'structuring' has
an intrinsic effectiveness of its own whatever the principles and
methods which suggested it.
'.\Iodern chemistry reduces the variety of tastes and smells to
different combinations of five elements: carbon, hydrogen, oxygen,
sulphur and nitrogen. By means of tables of the presence and
absence of the elements and estimates of proportions and minimum
amounts necessary for them to be perceptible, it succeeds in
accounting for ditferences and resemblances which were pre\·iously
excluded from its field on account of their 'secondary' character.
These connections and distinctions are however no surprise to our
aesthetic sense. On the contrary they increase its scope and under-
standing hy supplying a basis for the associations it already divined;
and at the same time one is better able to understand why and in
what conditions it should have been possible to discover such
associations solely by the systematic use of intuitive methods. Thus
to a logic of sensations tohacco smoke might be the intersection of
two groups, o:le also containing broiled meat and brown crusts of
bread (which are like it in being composed of nitrogen) and the
other one to which cheese, beer and honey belong on account of
the presence of diacetyl. \Yild cherries, cinnamon, ,·anilla and
sherry are grouped together by the intellect as well as the senses,
because they ail contain aldehyde, while the closely related smells
of wintergreen, b.Yender and bananas are to be explained by the
presence of ester. On intuiti\·e groundsalone we might group onions,
garlic, cabbage, turnips, radishes and mustard together eYen
th0ugh hotany separates liliaceae and crucifers. In confirmation of
the evidence of the senses, chemistry shows that these different
families are united on another plane: they contain sulphur (\V.K.).
A primitive philosopher or a poet could have etfected these
regroupings on the basis of considerations foreign to chemistry or
any other form of science. Ethnographie literature reveals many of
equal empirical and aesthetic value. And this is not just the result
of some associative madness destined sometimes to succeed simply
12
THE SCIE:"CE OF TIIE CO;-;'CRETE
by the law of chance. Simpson advances this interpretation in the\

passage quoted above; but he displays more insight when he \
shows that the demand for organi,2:ation is a need common to art /
and science and that in consequence 'taxonomy, which is ordering (
par excellence, has eminent aesthetic value' (!oc. cit., p. 4). Given
this, it seems less surprising that the aesthetic sense can by itseif/
open the way to taxonomy and even anticipate some of its resu!~
1 am not ho\\'ever commending a return to the popular belief

(although it has some validity in its own narrow context) according
to which magic is a timid arid stuttering form of science. One
deprives oneself of ail means of understanding magical thought if
one tries to reduce it to a moment or stage in technical and scie -
tific ernlution. Like a sh1dow moving ahead of its owner it is i~ a
sense complete in itself, and as finished and coherent in its imma-
teriality as the substantial being which it preccdes. !\Tagical thought
is not to be regarded as a beginning, a rudiment, a sketch, a part
of a whole which has not yet materialized. 1t forms a well-articu-
lated system, arid is in this respect independent of that other system
which constitutes science, except for the purely formai analogy
which brings them together arid makes the former a sort of meta-
phorical expression of the latter. DJ;; tberdorc b~f
contrasting magic and ~impare them as two para,l_lel.-
1~ of acquiring knowledge. Thêirtnë?rrctical-mtd--pfâêtical
resuJts d1ffcr ln va!ue;-ror If IS true that science is more SUCCessfu!
than magic from this point of view, although magic foreshadows
science in that it is sometimes also successful. Both science and
magic however require the same sort of mental operations and they
differ not so much in kind as in the different types of phenomena to
which they are applied.
These relations are a consequence of the objective conditions in
which magic and scientific knowledge appeared. The history of the
latter is short enough for us to know a good deal about it. But the
fact that modern science dates back only a few centuries raises a
problem which ethnologists have not suffi.ciently pondered. The
Neolithic Paradox would be a suitable name for it.
lt was in neolithic times that man's mastery of the great arts of
civilization - of pottery, weaving, agriculture and the domestica-
tion of a!1imals - became firmly established. No one today would
any longer think of attributing these enormous advances to the
13
THE SAVAGE MIND
fortuitous accumulation of a series of chance discoveries or believe

them to have been revealed by the passive perception of certain
natural phenomena.*
Each of these techniques assumes centuries of active and
methodical observation, of bold hypotheses tested by means of
endlessly repeated experiments. A biologist remarks on the rapidity
with which plants from the New World have been acclimatized in
the Philippines and adopted and named by the natives. In many
cases they seem even to have rediscovered their medicinal uses,
uses identical with those traditional in ~lexico. Fox's interpretation
is this:
... plants with bitter leaves or stems are commonly used in the
Philippines for stomach disorders. If an introduced plant is found to have
this characteristic, it will be quickly utilized. The fact that many Philip-
pine groups, such as the Pinatubo Negritos, constantly experiment with
plants hastens the process of the recognition of the potential usefulness,
as defined by the culture, of the introduced flora (R. B. Fox, pp. 212-13).
To transform a weed into a cultivated plant, a wild beast into a

domestic animal, to produce, in either of these, nutritious or tech-
nologically useful properties which were originally completely
absent or could only be guessed at; to make stout, water-tight
pottery out of clay which is friable and unstable, liable to pulverize
or crack (which, however, is possible only if from a large number of
organic and inorganic materials, the one most suitable for refining
it is selected, and also the appropriate fuel, the temperature and
duration of firing and the effective degree of oxidation); to work
out techniques, often long and complex, which permit cultivation
without soil or alternatively without water; to change toxic roots
or seeds into foodstuffs or again to use their poison for hunting,
war or ritual - there is no doubt that all these achievements
required a genuinely scientific attitude, sustained and watchful
interest and a desire for knowledge for its own sake. For only a
small proportion of observations and experiments (which must be
assumed to have been primarily inspired by a desire for knowledge)
• An attcmpt has been made to discover what would happen if copper ore had
accidentally found its way into a furnace: complex and varied experiments have
shown that nothing happens at ail. The simplest method of obtaining metallic
copper which could be discovered consisted in subjecting finely ground malachite
to intense heat in a pottery dish crowned with an inverted clay pot. This, the
sole result, restricts the play of chance to the confines of the kiln of some patter
specializîng in glazed ware (Coghlan).
could have yielded practical and immediately useful results. There

is no need to dwell on the working of bronze and iron and of
precious metals or even the simple working of copper ore by
hammering which preceded metallurgy by several thousand years,
and even at that stage they all demand a very high level of technical
proficiency.
Neolithic, or early historical, man was therefore the heir of a
long scientific tradition. However, had he, as well as all his
predecessors, been inspired by exactly the same spirit as that of
our own time, it would be impossible to understand how he could
have corne to a hait and how several thousand years of stagnation have
intervened between the neolithic revolution and modern science
like a level plain between ascents. There is only one solution to the
paradox, namely, that there are two distinct modes of scientific
thought. These are certainly not a fonction of different stages of
development of the human mind but rather of two strategic levels
at which nature is accessible to scientific enquiry: one roughly
adapted to that of perception and the imagination: the other at a
remove from it. It is as if the necessary connections which are the
object of all science, neolithic or modern, could be arrived at by
two different routes, one very close to, and the other more remote
from, sensible intuition.
Any classification is superior to chaos and even a classification at
the level of sensible properties is a step towards rational ordering.
It is legitimate, in classifying fruits into relatively heavy and rela-
tively light, to begin by separating the apples from the pears even
though· shape, colour and taste are unconnected with weight and
volume. This is because the larger apples are easier to distinguish
from the smaller if the apples are not still mixed with fruit of
different features. This example already shows that classification
has its advantages even at the level of aesthetic perception.
For the rest, and in spite of the fact there is no necessary con-
nection between sensible qualities and properties, there is very often
at least an empirical connection between them, and the generaliza-
tion of this relation may be rewarding from the theoretical and
practical point of view for a very long time even if it has no founda-
tion in reason. Not all poisonous juices are burning or bitter nor is
everything which is burning and bitter poisonous. N evertheless,
nature is so constituted that it is more advantageous if thought and
action proceed as though this aesthetically satisfying equivalence
THE SAVAGE MIND
also corresponded to objective reality. It seems probable, for

reasons which are not relevant here, that species possessing some
remarkable characteristics, say, of shape, colour or smell give the
observer what might be called a 'right pending disproof' to postu-
la te that these visible characteristics are the sign of equally singular,
but concealed, properties. To treat the relation between the two as
itself sensible (regarding a seed in the form of a tooth as a safeguard
against snake bites, yellow juices as a cure for bilious troubles, etc.)
is ~f more value provisionally than indifference to any connection.
For even a heterogeneous and arbitrary classification preserves the
richness and diversity of the collection of facts it makes. The
decision that everything must be taken account of facilitates the
creation of a 'memory bank'.
---·rt is moreover a fact that particular results, to the achievement
of which methods of this kind were able to lead, were essential to
enable man to assail nature from a different angle. Myths and rites
are far from being, as has often been held, the product of man's
'myth-making faculty', * turning its back on reality. Their principal
value is indeed to preserve until the present time the remains of
methods of observation and reflection which were (and no doubt
still are) precisely adapted to discoveries of a certain type: those
which nature authorised from the starting point of a speculative
organization and exploitation of the sensible world in sensible
terms. This science of the concrete was necessarily restricted by its
essence to results other than those destined to be achieved by the
exact natural sciences but it was no less scientific and its results no
less genuine. They were secured ten thousand years earlier and still
remain at the basis of our own civilization.
There still exists among ourselves an activity which on the technical

plane gives us quite a good understanding of what a science we
prefer to call 'prior' rather dian 'primitive', could have been on
the plane of speculation. This is what is commonly called 'brico-
lage' in French. In its old sense the verb 'bricoler' applied to ball
games and billiards, to hunting, shooting and riding. It was how-
ever always used with reference to some extraneous movement: a
ball rebounding, a dog straying or a horse swerving from its direct
course to avoid an obstacle. And in our own time the 'bricoleur' is
still someone who works with his hands and uses devious means
• The phrase is from Bergson, op. cit., 'fonction fabulatrice' (trans. note).
16
compared to those of a craftsman.* The characteristic feature of

mythical thought is that it expresses itself by means of a hetero~
geneous repertoire which, even if extensive, is nevertheless limited.
It has to u~is repertoire, however. wbJ).t~v_er_tJi~~~ask in hand
because it has nothfgelse at its diseo~.al. l\Iythical thoughtisthere-
fore a kind of intel ectual 'bricolage' - which explains the relation
which can be perceived between the two.
Like 'bricolage' on the technical plane, mythical reflection can
reach brilliant unforeseen results on the intcllectual plane. Con-
versely, attention has often been drawn to the mytho-poetical
nature of 'bricolage' on the plane of so-called 'raw' or 'naïve' art,
in architectural follies like the villa of Cheval the postman or the
stage sets of Georges :\Iéliès, or, again, in the case immortalized by
Dickens in Great Expectations but no doubt originally inspired by
observation, of :\Ir Wemmick's suburban 'castle' with its miniature
drawbridge, its cannon firing at nine o'clock, its bed of salad and
cucumbers, thanks to which its occupants could withstand a siege
if necessary ...
The analogy is worth pursuing since it helps us to see the real
relations between the two types of scientific knowlcdge we have
distinguished. The 'bricoleur' is adept at performing a large
number of diverse tasks; but, unlike the engineer, he docs not sub-
ordinate each of them to the availability of raw materials and tools
concei ved and procured for the pur pose of the project. His uni verse
of instruments is closed and the rules of his game are always to
make do with 'whatever is at hand', that is to say with a set of tools
and materials which is always finite and is also heterogeneous
because what it contains bears no relation to the current project, or
indeed to any particular project, but is the contingent result of all
the occasions there have been to renew or enrich the stock or to
maintain it with the remains of previous constructions or destruc-
tions. The set of the 'bricoleur's' means cannot therefore be defined
in terms of a project (which would presuppose besicles, that, as in
the case of the engineer, there were, at least in theory, as many
sets of tools and materials or 'instrumental sets', as there are
different kinds of projects). It is to be defined only by its potential
• The 'bricoleur' has no precise equivalent in English. He is a man who under-

takes odd jobs and is a Jack of all trades or a kind of profcssional do-it-yourself
man, but, as the text makes clear, he is of a different standing from, for instance,
the English 'odd job man' or handyman (trans. note).
17
THE SAVAGE MIND
use or, putting this another way and in the language of the 'bri-
coleur' himself, because the elements are collected or retained on
the principle that 'they may always corne in handy'. Such elements
are specialized up to a point, sufficiently for the 'bricoleur' not
to need the equipment and knowledge of all trades and profes-
sions, but not enough for each of them to have only one definite
and determinate use. They each represent a set of actual and
possible relations; they are 'operators' but they can be used for any
operations of the same type.
~lements Q(mythicalthought s~!J1_il~}yJi!!half-way between
P~-~_!Vl.Jld-roncep!ê.:_ It would be impossible to separate percepts
from the concrete situations in which they appeared, while recourse
to concepts would require that thought could, at least provision-
ally, put its projects (to use Husserl's expression) 'in brackets'.
~_an _intermediary _betweerL-images and concepts,
~~igns. For signs can always be defined in the way intro-
duced by Saussure in the case of the particular category of
linguistic signs, that is, as a link between images and concepts. In
the union thus brought about, images and concepts play the part
of the signifying and signified respectively.
/Signs resemble images in being concrete entities but they
resemble concepts in their powers of reference. Neither concepts
nor signs relate exclusively to themselves; either may be substi-
tuted for something else. Concepts, however, have an unlimited
,capacity in this respect, while signs have not. The example of the
'Drîèoleur' helps to bring out the differences and similarities. Con-
sider him at work and excited by his project. His first practical step
is retrospective. He has to turn back to an already existent set made
up of tools and materials, to consider or reconsider what it contains
and, finally and above all, to engage in a sort of dialogue with it
and, before choosing between them, to index the possible answers
which the whole set can offer to his problem. He interrogates all the
heterogeneous objects of which his treasury* is composed to dis-
cover what each of them could 'signify' and so contribute to the
definition of a set which has yet to materialize but which will
ultimately differ from the instrumental set only in the interna! dis-
position of its parts. A particular cube of oak could be a wedge to
make up for the inadequte length of a plank of pine or it could be a
pedestal - which would allow the grain and polish of the old wood
• Cf. 'Treasury of ideas' as Hubert and Mauss so aptly describe magic.
to show to advantage. In one case it will serve as extension, in the

other as material. But the possibilities always remain limited by the
particular history of each piece and by those of its f eatures which are
already determined by the use for which it was originally intended
or the modifications it has undergone for other purposes. The
elements which the 'bricoleur' collects and uses are 'pre-con-
strained' like the constitutive units of myth, the possible combina-
tions of which are restricted by the fact that they are drawn from
the language where they already possess a sense which sets a limit
on their freedom of manoeuvre (Lévi-Strauss, 5, p. 35). And the
decision as to what to put in each place also depends on the possi-
bility of putting a different element there instead, so that each
choice which is made will involve a complete reorganization of the
structure, which will never be the same as one vaguely imagined
nor as some other which might have been pref erred to it.
The engineer no doubt also cross-examines his resources. The
existence of an 'interlocutor' is in his case due to the fact that his
means, power and knowledge are never unlimited and that in this
negative form he meets resistance with which he has to corne to
terms. It might be said that the engineer questions the universe,
while the 'bricoleur' addresses himself to a collection of oddments
left over from human endeavours, that is, only a sub-set of the
culture. Again, Information Theory shows that it is possible, and
often useful, to reduce the physicists' approaches to a sort of
dialogue with nature. This would make the distinction we ~e
trying to draw less clearcut. There remains however a diff erence
even if one takes into account the fact that the scientist never
carries on a dialogue with nature pure and simple but rather with a
particular relationship between nature and culture definable in
terms of his particular period and civilization and the material
means at his disposai. He is no more able than the 'bricoleur' to-de-
whatever he wishes when he is presented with a given task. He too
has to begin by making a catalogue of a previously determined set
consisting of theoretical and practical knowledge, of technical
means, which restrict the possible solutions.
The difference is therefore less absolute than it might appear.
It remains a real one, however, in that the engineer is always trying
to make his way out of and go beyond the constraints imposed by
a particular state of civilization while the 'bricoleur' by inclination
or necessity always remains within them. This is another way of
THE SA\'AWo ;1.tlNll
saying that the engineer works hy means of concepts and the

'bricoleur' by means of signs. The sets which each employs are at
different distances from the poles on the axis of opposition between
1îature and culture. One way indeed in which signs can be opposed
to concepts is that whereas concepts airn to be wholly transparent
with respect to reality, signs allow and even require the interposing
and incorporation of a certain amount of human culture into
reality. Signs, in Peirce's vigornus phrase 'address somebody'.
--Hoth the scientist and 'bricoleur' might therefore be said to be
constantly on the look out for 'messages'. Those which the 'bri-
coleur' collects are, howeyer, ones which have to some extent been
transmitted in advancc - like the commercial codes which are
summarics of the past cxpericncc of the tradc and so allow any new
situation to be met cconomically, providcd that it belongs to the
samc class as some eadicr one. The scicntist, on the other hand,
whcther he is an engineer or a physicist, is always on the look out
for that other message which might be wrcsted from an interlocutor
in spite of his reticcncc in pronouncing on questions whose answers
have not been rchearsed. Concepts thus appear like operators open-
ing up the set bcing workcd with and signification like the opcrator
of its reorganizatio11, which ncithcr extends nor rencws it and limits
itse!Lto obtaining the group of its transformations.
/"'Îmagcs cannot be ideas but they can play the part of signs or, to
be more precisc, co-cxist \vith ideas in signs and, if ideas are not
yet present, they can kecp their future place open for them and
make its contours apparent negatively. Images arc fixed, linked in a
si'ilgic way to the mental act which accompanies thcm. Signs, and
images which have acquircd significance, may still lack compre-
hension; unlike concepts, they do not yet possess simultaneous and
thcoretically unlimited relations with other entities of the same
kind. They are howevcr already permutable, that is, capable of
standing in succcssiœ relations with othcr entities - although with
only a limited number and, as we have secn, only on the condition
that they always forma system in which a11 alteration which affects
one clement automatically affects ail the others. On this plane
logicians' 'extension' and 'intension' are not two distinct and com-
plementary aspects hut one and the samc thing. One understands
then how mythical thought can be capable of generalizing and so
be scientific, evcn though it is still entanglcd in imagery. It too
works by analogies and comparisons even tho:.1gh its creations, like
20
THE SCIE1'CE OF THE CONCRETE
those of the 'bricoleur', always really consist of a new arrangement

of elements, the nature of which is unaffected by whether they
figure in the instrumental set or in the final arrangement (these
being the same, apart from the internai disposition of their parts):
'it would seem that mythological wodds haYe been built up, only
to be shattered again, and that new worlds were built from the
fragments' (Boas 1, p. 18). Pe!letrating as this comment is, it ne\'er-
theless fails to take into account that in the continuai reconstruc-
tion from the same materials, it is always earlier ends which arc
called upon to play the part of mea'1s: the signified changes into
the signifying and vice \'ersa.
This formula, "·hich could sen·c as a definitio:i of 'bricolage',
explains how an implicit inventory or conception of the total meanl
a\'ailable must be made in the case of mythical thought also, so tha
a result can be defined which will always be a compromise betwee1
the structure of the i1strumental set and that of the project. Once
it materializes the p:-oject will therefore ine,·itably be at a remm·f
from the initial aim (which was moreovcr a mere sketch), a phcnci-
menon which the surrealists ha,·e felicitously callcd 'objecti\f
hazard'. Further, the 'bricolwr' also, and indecd principall~,
deri,·es his poetry from the fact that hc does not confirie himself to
accomplishrnent and execution: he 'speaks' not only witlz things, ds
we haYe already seen, but also through the medium of thing~:
gi,·ing an account of his personality and life by the choices he
makcs betwccn the limited possibilities. The 'bricoleur' may not
ever complcte his purposc but hc always puts something of himsclf
into it. ~
:\lythical thought appears to be an intellectual form of 'bricolage'
in this scnsc also. Science as a whole is based on the distinction
between the contingent and the necessary, this being also what
di3tinguishes c\'ent and structure. The qualities it claimcd at its
outset as peculiarly scientific were precisely those which formcc!,
no part of li,·ing experiencc and rem-iined outsidc and, as it were,
unrelatcd to e\·cnts. This is the significancc of the notion of primary
qualities . .'.'\m\-, the characteristic f eature of mythical thought, as
of 'bricolage' on the practical plane, is that it builds up stmctured
sets, not directly with other stru:::tured sets* but by using the
• :\lythical thought buil<ls structurc<l sets hy means of a structutul seL,
namcly, languagc. But it is not at the structural levcl that it makcs use of it:
it buil<ls i<lcological castles out of the dchris of what was once a sociai discourse.
21
THE SAVAGE MIND
remains anddebrisof events :in French 'des bribesetdes morceaux',

_c:>r ?dds and ends in English, fossilized evidence of the history
of an individual or a society. The relation between the diachronie
and the synchronie is therefore in a sense reversed. Mythical
thought, that 'bricoleur', builds up structures by fitting together
events, or rather the remains of events, * while science, 'in opera-
tion' simply by virtue of coming into being, creates its means and
results in the form of events, thanks to the structures which it is
~stantly elaborating and which are its hypotheses and theories.
But it is important not to make the mistake of thinking that these
are two stages or phases in the evolution of knowledge._B__o_t_h__
~ ar~~id. Physics and chemistry are already
striving to become qualitative again, that is, to account also for
sccondary qualities which when they have been explained will in
their turn become means of explanation. And biology may perhaps
be marking time waiting for this before it can itself explain life.
l\:Iythical thought for its part is imprisoned in the events and
experiences which it never tires of ordering and re-ordering in its
search to find them a meaning. But it also acts as a liberator by its
protest against the idea that anything can be meaningless with
which science at first resigned itself to a compromise.
~ problem of art has been touched on several times in the fore-

going discussion, and it is worth showing briëfly how, from this
lpoint of view, art lies half-way between scientific knowledge and
m~hical or magical thought. lt is common knowledge that the
artist is both something of a scientist and of a 'bricoleur'. By his
craftsmanship he constructs a material abject which is also an
abject of knowledge. W e have already distinguished the scientist
and the 'bricoleur' by the inverse fonctions which they assign to
events and structures as ends and means, the scientist creating
events (changing the world) by means of structures and the 'bri-
coleur' creating structures by means of events. This is imprecise in
this crude form but our analysis makes it possible for us to refine
it. Let us now look at this portrait of a woman by Clouet and
consider the reason for the very profound aesthetic emotion which
is, apparently inexplicably, aroused by the highly realistic, thread
by thread, reproduction of a lace collar (Plate 1 ).
The choice of this example is not accidentai. Clouet is known to
"'Bricolage' also works with 'secondary' qualities, i.e. 'second hand'.
22
have liked to paint a t less than life-size. His paintings are therefore,
like Japanese gardens, miniature vehicles and ships in bottles, what
in the 'bricoleur's' language are called 'small-scale models' or 'min-
iatures'. Now, the question arises whether the small-scale model or
miniature, which is also the 'masterpiece' of the journeyman may
not in fact be the universal type of the work of art. All miniatures
seem to have intrinsic aesthetic quality - and from what should
they draw this constant virtue if not from the dimensions them-
selves? - and conversely the vast majority of works of art are small-
scale. It might be thought that this characteristic is principally a
matter of economy in materials and means, and one might appeal
in support of this theory to works which are incontestably artistic
but also on a grand scale. \Ve have to be clear about definitions.
The paintings of the Sistine Chapel are a small-scale model in spite
of their imposing dimensions, since the theme which they depict
is the End of Time. The same is true of the cosmic symbolism of
religious monuments. Furthcr, we may ask whether the aesthetic
effect, say, of an equestrian statue which is larger than life derives
from its enlargement of a man to the size of a rock or whether it is
not rather due to the fact that it restores what is at first from a
distance seen as a rock to the proportions of a man. Finally even
'natural sizc' implies a reduction of scale since graphie or plastic
transposition always involves giving up certain dimensions of the
object: volume in painting, colour, smell, tactile impressions in
sculpture and the temporal dimension in both cases since the whole
work represented is apprehendcd at a single moment in time.
What is the virtue of reduction either of scale or in the number
of properties? It seems to result from a sort of reversai in the pro-
cess of understanding. To understand a real object in its totalrty
we always tend to work from its parts. The resistance it off ers us is
overcome by dividing it. Reduction in scale reverses this situation.
Being smaller, the object as a whole seems less formidable. By
being quantitatively diminished, it seems tous qualitatively simpli-
fied. More exactly, this quantitative transposition extends aryd
diversifies our power over a homologue of the thing, and by means
of it the latter can be grasped, assessed and apprehended at a
glance. A child's doll is no longer an enemy, a rival or even an
interlocutor. ln it and through it a person is made into a subject.
In the case of miniatures, in contrast to what happens when we try
to understand an object or living creature of real dimensions,
23
THE SAVAGE l\llND
knowledge of the whole precedes knowledge of the parts. And

even if this is an illusion, the point of the procedure is to create or
sustain the illusion, which gratifies the intelligence and gives rise to
a sense of pleasure which can already be called aesthetic on these
grounds alone.
1 have so far only considered matters of scale which, as we have
just seen, imply a dialectical relation between size (i.e. quantity)
and quality. But miniatures have a further feature. They are 'man
made' and, what is more, made by hand. They are therefore not
just projections or passive homologues of the object: they con-
stitutc a real experiment with it. Now the model being an artefact,
it is possible to understand how it is made and this understanding
of the method of construction adds a supplementary dimension.
As we have already sccn in the case of 'bricolage', and the example
of 'styles' of painters shows that the same is true in art, there are
several solutions to the same problem. The choice of one solution
involves a modification of the result to which another solution
would have led, and the observer is in effect presented with the
general picture of these permutations at the same time as the
particular solution offered. He is thereby transformed into an active
participant without even being aware of it. Merely by contemplat-
ing it he is, as it were, put in possession of other possible forms
of the same work; and in a confused way, he feels himself to be
their creator with mure right than the creator himself because the
latter abandoned them in excluding them from his creation. And
thesc forms are so man y further perspectives opening out on to the
-work which has been rcalized. ln other words, the intrinsic value
of a small-scale mode! is that it compensates for the renunciation
of sensible dimensions by the acquisition of intelligible dimensions.
~et us now return to the lace colhr in Clouet's picture. Every:-
thing that has been said applies in this case, for the procedure
necessary to represent it as a projection, in a particular space, of
properties whose sensible dimensions are fewer and smaller than
that of the object is exactly the reverse of that which science would
have employed had it proposed, in accordance with its function, to
producc (instead of reproducing) not only a new, instead of an
already known, piece of lace but also real lace instead of a picture
of lace. Science would have worked on the real scale but by means
of inventing a loom, while art works on a diminished scale to
produce an image homologous with the object. The former
THE SCIE:\"CE OF THE CONCRETE
approach is of a metonymical order, it replaces one thing by

another thing, an effect by its cause, while the latter is of a meta-
phorical order.
This is not ail. For if it is tr~e that the relation of priority
between structure and event is exactly the opposite in science and
'bricolage', then it is clear that art has an intenncdiate position
from this point of view as well. Even if, as we have snown, tlie
dëpiction of a lace collar m miniature demands an intimate know-
ledge of its morphology and technique of manufacture (and had it
been a question of the representation of people or animais we
should have said: of ana tom y and physical attitudes), it is notjust a
diagram or blueprint. It manages to synthesize these intrinsic
properties with properties which depend on a spatial and temporal
context. The final product is the lace collar exactly as it is but so
that at the samc time its appearance is affected by the particular
perspective. This accentuates some parts and conceals others,
whose existence however still influences the rest through the con-
trast between its whiteness and the colour of the other clothes, the
reflection of the pearly neck it encircles and that of the sky on a
particular day and at a particular time of day. The appearance of
the lace collar is also affected by whether it indicates casual or
formai dress, is worn. either new or previously used, either freshly
ironed or creased, by an ordinary woman or a queen, whose
physiognomy confirms, contradicts or qualifies her status in a
particular social class, society, part of the world and period of
history ... The painter is always mid-way between design and
anecdote, and his genius consists i.n uniting internai a11d external
knowledge, a 'being' and a 'becoming', in producing with his brush
an object which does not existas such and which he is nevertheless
able to create on his canvas. This is a nicely balanced synthesis of
one or more artificial and natural structures and one or more
natural and social events. The aesthetic emotion is the result of tlîis-
union between the structural order and the order of events, which
is brought about within a thing created by man and so also in effect
by the observer who discovers the possibility of such a union
through the work of art. ----
Several points are suggested by this analysis. In the first plac'ê~
the analysis helps us to see why we are inclined to think of myths
both as systems of abstract relations and as objects of aesthetic
contemplation. The creative act which gives rise to myths is in
THE SAVAGE MIND
fact exactly the reverse of that which gives rise to works of art.
In the case of works of art, the starting point is a set of one or more
objects and one or more events which aesthetic creation unifies by
~fing a conunon structure. l\'Iyths travel the same road but
start from the other end. They use a structure to produce what is
itself an object consisting of a set of events (for all myths tell a
story). Art thus proceeds from a set (object + event) to the dis-
co1:ery of its structure. ::'vlyth starts from a structure by means of
which it constructs a set (object + event).
~ -The first point tempts one to generalize the theory. The second
might seem to lead to a restriction of it. For we may ask whether
it is in fact the case that works of art are always an integration of
structure and event. This does not on the face of it seem to be
true for instance of the cedarwood Tlingit club, used to kill fish,
which I have in front of me on my bookshelf (Plate 2 ). The artist
who carved it in the form of a sea monster intended the body of the
implement to be fosed with the body of the animal and the handle
with its tail, and that the anatomical proportions, taken from a
fabulous creature, should be such that the object could be the cruel
animal slaying helpless victims, at the same time as an easily
handled, balanced and efficient fishing utensil. Everything about
this implement - which is also a superb work of art - seems to be a
matter of structure: its mythical symbolism as well as its practical
fonction. 1\Iore accurately, the object, its fonction and its sym-
bolism seem to be inextricably bound up with each other and to
form a closed system in which there is no place for events. The
monster's position, appearance and expression owe nothing to the
historical circumstances in which the artist saw it, in the flesh or in
a dream, or conceived the idea of it. It is rather as if its immutable
being were finally fixed in the \vood whose fine grain allows the
reproduction of all its aspects and in the use for which its empirical
form seems to pre-determine it. And all this applies equally to the
other products of primitive art: an African statue or a Melanesian
mask . . . So it looks as if we have defined only one local and
historical form of aesthetic creation and not its fondamental pro-
perties or those by means of which its intelligible relations with
other forms of creation can be described.
\Ve have only to widen our explanation to overcome this diffi-
culty. \Vhat, with reference to a picture of Clouet's, was provision-
ally defined as an event or set of events now appears under a
26
broader heading: events in this sense are only one mode of the
contingent whose integration (perceived as necessary) into a struc-
ture gives rise to the aesthetic emotion. This is so whatever the type
of art in question. Depending ori- the style, place and period the
contingent plays a part in three different ways or at three distinct
points in artistic creation (or in all of them). It may play a part in
the occasion for the work or in the execution of the work or in the
purpose for which it is intended. It is only in the first case that it
takes the form of an event properly speaking, that is, of conting-
ency exterior and prior to the creative act. The artist perceives it
from without as an attitude, an expression, a light effect or a
situation, whose sensible and intellectual relations to the structure
of the object affected by these modalities he grasps and incorpor-
ates in his work. But the contingent can also play an intrinsic part
in the course of execution itself, in the size or shape of the piece of
wood the sculptor lays hands on, in the direction and quality of its
grain, in the imperfections of his tools, in the resistance which his
materials or project offer to the work in the course of its accomplish-
ment, in the unforeseeable incidents arising during work. Finally,
the contingent can be extrinsic as in the first case but posterior,
instead of anterior, to the act of creation. This is the case whenever
the work is destined for a specific end, since the artist will construct
it with a view to its potential condition and successive uses in the
future and so will put himself, consciously or unconsciously, in the
place of the person for whose use it is intended. _
The rocess of artisti~.rcation therefore consists in trying~o
communicate within the immu~rk-of·a:-müfüâlcon::·
fronfat10n0f structure and accident) either with the mode! or with
the materials or with the future user as the case may be, according
to which of these the artist particularly looks to for his directions
while he is at work. Each case roughly corresponds to a readily
identifiable form of art: the first to the plastic arts of the West, the
second to so-called primitive or early art and the third to the
applied arts. But it would be an oversimplification to take these
identifications very strictly. All forms of art allow all three aspects
and they are only distinguished from one another by the relative
proportion of each. Even the most academic of painters cornes up
against problems of execution, for example. All the so-called primi-
tive arts can be called applied in a double sense: first, because many
of their productions are technical objects and, secondly, because
27
TllE SAVAGE :\llND
cvcn thosc \vhich secm rnostdivorced from practical preoccupations

'.-lave a definite purpose. Finally, as wc know, implcments lend
themsclves to disintcrcstcd contemplation even among ourselves.
\Vith thcse reservations, it is easy to show that the three aspects
arc functionally relatcd and that the prcdomina11ce of any one of
them !caves lcss or no place for the others. So-callcd professional
painting is, or belie,·es itself to be, quite free so far as both execu-
tion and purpose arc concerned. Its bcst examples display a com-
pletc mastery of technical difficultics - which, indeed, can be
considcred to ha,·e bee11 completcly oYercome since Van der
\Veyden; the problcms which painters have set thcmselves since
then amount to little more than a gamc of technical refinement. In
the ex treme case it is as though, given his can vas, paints and brushes,
the painter were able to do cxactly what he pleascd. On the other
hand, hc also tries to makc his work into an ohject indcpcndent of
anything contingent, of value in itself and for itself. This is indeed
what the formula of the 'easel picture' implies. Freed from the
contingent both with regard to execution and purpose professional
painting can, thcn, bring it to bcar upon the occasion of the work,
and indccd if this account is correct it is bound to do so. Pro-
fessional painting can thcrefore be defined as 'genre' painting if the
sense of this expression is considcrably widcned. For, from the
very general viewpoint we are taking, the attempt of a portrait
pain ter - cvcn of a Rembrandt -- to rccaptur e on his cam·as his
model's most renaling expressi,m or secret thoughts belongs to
the samc genre as that of a painter like Detaillc, whose compositions
reproduce the hour and order of battle and the number and dis-
position of the buttons distinguishing the uniforms of each Arm.
To use a disrespcctful analogy, 'opportunity makes the thief'* in
either case. The relative proportions of the three aspects are
reversed in the applied arts. In thest, first place is given to purpose
and execution, contingent factors playing an approximately equal
part in each, in the examples wc ..:onsider the most 'pure', at the
same time the occasion of the work plays no part. This can be seen
from the fact that a winc cup or goblet, a piece of basket work or a
fa bric seems to u;; perfect when its practical value m:mifestly trans-
cends time and corresponds wholly to its fonctions for men of
difîerent periods and civilizations. If the difficulties of executiùn
are entirely mastered, as is the case when it is entrusted to machines,
• ln the original: 'l'Dc:casion fait le larron' (trnns. note).
28
THE SCIENCE OF THE COl\"CHETE
the purposc can becomc more and more prccisc and specific and
applied art is transformed into industrial art. Wc call it peasan~
folk art if the reverse is the case. Finally, primitive art is the
opposite of profcssional or academic art. Professional or acadcmic
art internalizcs execution (which it has, or hclievcs itself to have,
mastercd) and purp~>se ('art for art's sake' being an end in itself).
As a result, it is impellcd to cxternalize the occasion (which it
requires the mode! to provide) and the latter thus becomes a part
of the signified. PrimitiYc art, on the othcr hand, internalizcs the
occasion (sincc the supcrnat ural hcings which it dclights in rcprc-
senting ha\'c a rcality which is timclcss and indcpcndcnt of circum-
stanccs) and it cxtcrnalizcs cxccution a11d purposc which thus
bccome a part of the signifying. _.,,-
On a ditfcrcnt plane wc thcrcforc find once more this dialogue
with the matcrials arid mcans of cxccution hy which we dcfincd
'bricolage'. The esscntial prohkm for the philosophy <.fart is to
know wltcthcr the artist regards th cm as intcrlocutors or not. No
doubt thcy arc always rcgardcd as such, although lcast of all in art
which is too profcssional and most of all in the raw or riaive art
which \'Crgcs on 'bricolage', to the <letrimcnt of structure in hoth
cases. :\o form of art is, howc\·er, worthy of the namc if it allows
itself to corne cntirely undcr the sway of extrancous contingencics,
whcther of occasion or purposc. If it did so it would rate as an icon
(supplementary to the modcl) or as an implement (complemcntary
with the matcrial workcd). Even the most profession;.! art succecds
in mo\·ing us only if it arrcsts in timc this dissipation of the con-
tingent in fa\'our of the prctcxt and incorporatcs it in the work,
thereby imesting it with the dig•1Ïty of hcing a'l ohject in its own
right. In so far as carly art, primitive art a•1d the 'primitive' pcriods
of profcssional painting are the only ones which do not date, thcy
owe it to this dedication of the accidentai to the service of execution
and so to the use, which thcy try to make completc, of the raw
datum as the empirical matcrial of something mcaningful. *
• Pursuing this analysis, one might dt:fine non-rt:prt:st:ntational painting by
two foaturt:s. One, which it has in common with 'case!' painting, consists in a
total rcjection of the contingency of purposc: the picture is not made for a
particular ust:. The other fcature characteristic of non-representational painting
is its methotiical exploitation of the contingcncy of cxecution, which is claimed
to afforti the extl'rnal prctext or occasion of the pictu re. ::\'on-rcprcscntational
painting adopts 'styles' as 'subjccts'. lt claims to gi\'c a concrete rt;prcscntation
of the formai conditions of ail painting. Paradoxically the result is that non-
represcntational painting docs not, as it thinks, crl·atc works which arc as rl'al as,
THE SAVAGE MIND
It is necessary to add that the balance between structure and

event, necessity and contingency, the internai and external is a
precarious one. It is constantly threatened by forces which act in
one direction or the othcr according to fluctuations in fashion, style
or general social conditions. From this point of view, it would seem
that imprcssionism and cubism are not so much two successive
stages in the development of painting as partners in the same enter-
prise, which, although not exact contemporaries, nevertheless
collaborated by complementary distortions to pro]or.g a mode of
expression whose very existence, as we are better able to appreciate
today, was seriously threatened. The intermittent fashion for
'collages', originating when craftsmanship was dying, could not
for its part be anything but the transposition of 'bricolage' into the
realms of contemplation. Finally, the stress on the event can a]so
break away at certain times through greater emphasis either on
transient social phenomena (as in the case of Greuze at the end of
the eighteenth century or with socialist realism) or on transient
natural, or even meteorological, phenomena (impressionism) at the
expense of structure, 'structure' here being understood as 'struc-
ture of the same level', for the possibility of the structural aspect
being re-established elsewhere on a new plane is not ruled out.
\Ve have seen that there are analogies between mythical thought on
the theoretical, and 'bricolage' on the practical plane and that
artistic creation lies mid-way between science and thesetwo forms of
activity. There are relations of the same type between games and
rites.
~<\U-games are defined by a set of rules which in practice allow the
playing of any numher of matches. Ritual, which is also 'played',
is on the other hand, like a favoured instance of agame, remern-
bered from among the possible ones because it is the only one
which results in a particular type of equilibrium between the two
_s~. The transposition is readily seen in the case of the Gahuku-
Gama of New Guinea who have learnt football but who will play,
several days running, as many matches as are necessary for both
if not more real than, the objects of the physical world, but rather realistic
imitations of non-existent models. It is a school of academic painting in which
each artist strives to represent the manner in which he would execu te his pictures
if by chance he were to paint any.
sicles to reach the same score (Read, p. 429). This is treating agame
as a ritual.
The same can be said of the games which took place among the
Fox Indians during adoption ceremonies. Their purpose was to
replace a dead relative by a living one and so to allow the final
departure of the soul of the deceased. * The main aim of funeral rites
among the Fox seems indeed to be to get rid of the dead and to
prevent them from avenging on the living their bitterness and their
regret that they are no longer among them. For native philosophy
resolutely sicles with the living: 'Death is a hard thing. Sorrow is
especially hard'.
Death originated in the destruction by supernatural powers of
the younger of two mythical brothers who are cultural heroes
among all the Algonkin. But it was not yet final. It was made so by
the elder brother \vhen, in spite of his sorrow, he rejected the
ghost's request to be allowed to return to his place among the
living. l\'len must follow this example and be firm with the dead.
The living must make them understand that they have lost nothing
by dying since they regularly receive off erings of tobacco and food.
ln return they are expected to compensa te the living for the reality
of death which they recall to them and for the sorrow their demise
causes them by guaranteeing them long life, clothes and something
to eat. 'lt is the dead who make food increase', a native informant
explains. 'They (the Indians) must coax them that way' (Michelson
1, pp. 369, 407).
Now, the adoption rites which are necessary to make the soul of
the deceased finally decide to go where it will take on the role of a
protecting spirit are normally accompanied by competitive sports,
games of skill or chance between teams which are constituted on
the basis of an ad hoc division into two sicles, Tokan and Kicko.
It is said explicitly over and over again that it is the living and the
dead who are playing against each other. lt is as if the living off ered
the dead the consolation of a last match before finally being rid of
them. But, since the two teams are asymmetrical in what they
stand for, the outcome is inevitably determined in advance:
This is how it is when they play ball. When the man for whom the
adoption-feast is held is a Tokana, the Tokanagi win the game. The
Kickoagi cannot win. And if it is a Kicko woman for whom the adoption-
• See below, p. 199 n.
31
THE SAVAGE MIND
feast is given, the Kickoagi win, as in turn the Tokanagi <lo not wm
(:\llichelson 1, p. 385).
And what is in fact the case? It is clear that it is only the living who
·win in the great biological and social game which is constantly
taking place between the living and the dead. But, as all the North
American mythology confirms, to win a game is symbolically to
'kill' one's opponent; this is depicted as really happening in in-
numerable myths. By ruling that they should always win, the dead
are given_ the illusion that it is they who are really alive, and that
their opponents, having been 'killed' by them, are dead. Under the
guise of playing with the dead, one plays them false and commits
th,em. The formai structure of what might at first sight be taken
for a competitive game is in fact identical with that of a typical
ritual such as the Mitawit or Midewinin of these same Algonkin
peoples in which the initiates get symbolically killed by the dead
whose part is played by the initiated; they feign death in order to
obtain a further lease of life. In both cases, death is brought in but
~beduped.
~hus appear to have a disjunctive effect: they end in the
establishment of a difference between individual players or teams
where originally there was no indication of inequality. And at the
end of the game they are distinguished into winners and losers.
~!, on the other hand, is the exact inverse; it conjoins, for it
brings about a union (one might even say communion in this con-
text) or in any case an organic relation between two initially separ-
ate groups, one ideally merging with the person of the officiant and
the other with the collectivity of the faithful. In the case of games
the symmetry is thercfore preordained and it is of a structural kind
since it follows from the principle that the rules are the same for
both sicles. Asyanmetry is engendered: it follows inevitably from
the contingent nature of events, themselves due to intention, chance
or t::i.Ient. The reverse is true ofrituaL There is an asymmetrywhich
is postulated in advance between profane and sacred, faithful and
officiating, dead and living, initiated and uninitiated, etc., and the
'game' consists in making ail the participants pass to the winning
_s~-~y means of evcnts, the nature and ordering of which is
genuinely structural. Like science (though here again on both the
theoretical and the practical plane) the game produces events by
mcans of a structure; and wc can therefore understand why com-
petitive games should flourish in our industrial societies. Rites
and myths, on the other hand, like 'bricolage' (which these sam
societies only tolerate as a hobby or pastime), take to pieces an
reconstruct sets of events (on a psychical, socio-historical or tech
nical plane) and use them as so many indestructible pieces fo
structural patterns in which they serve alternatively as ends c
means.
33
CHAPTER TWO
THE LOGIC OF TOTEMIC

CLASSIFICATIONS
There is certainly something paradoxical about the idea of a logic

whose terms consist of odds and ends left over from psychological
or historical processes and are, like these, devoid of necessity. Logic
consists in the establishment of nccessary connections and how, we
may ask, could such relations be established between terms in no
way designed to fulfil this fonction? Propositions cannot be rigor-
ously connected unless the terms they contain have first been
unequivocally defined. 1t might seem as if in the preceding pages
we had undertaken the impossible task of discovering the
conditions of an a posteriori necessity.
Against this it may be said that, in the first place, these odds and
ends appear as such only in relation to the history which produced
them and not from the point of view of the logic for which they are
used. It is with respect to content alone that they can be regarded
as heterogeneous for, so far as form is concerned, there is an
analogy between them, which the example of bricolage made it
possible to define. The analogy consists in the incorporation in
their form itself of a certain amount of content, which is roughly
the same for all. The significant images of myth, the materials of
the bricoleur, are elements which can be defined by two criteria:
they have had a use, as words in a piece of discourse which mythical
thought 'detaches' in the same way as a bricoleur, in the course of
repairing them, detaches the cogwheels of an old alarm dock; and
they can be used again either for the same purpose or for a different
one if they are at all diverted from their previous fonction.
Secondly, neither the images of myth nor the materials of the
35
THE SAVAGE MIND
bricoleur are products of 'becoming' pure and simple. Previously,

then they were part of other coherent sets, they possessed the
rigour which they seem to lack as soon as we observe them in their
new use. \Vhat is more, they still possess this precision in so far as
they are not raw materials but wrought products: terms in language
or, in the case of the bricoleur, in a technological system. They are
therefore condensed expressions of neccssary relations which
impose constraints with various repercussions at each stage ol their
employment. Their necessity is not simple and univocal. But it is
there neverthcless as the invariance of a semantic or aesthetic order,
which characterizes the group of transformations to which they
lend themselves, which, as we have seen, are not unlimited.
This logic works rather like a kaleidoscope, an instrument which
also contains bits and pieces by means of which structural patterns
are realized. The fragments are products of a process of breaking
up and destroying, in itself a contingent matter, but they have to
be homologous in various respects, such as size, brightness 1 of
colouring, transparency. They can no longer be considered entities
in their own right in relation to the manufactured objects of whose
'discourse' they have become the i!ldefinable dcbris, but they must
be so considered from a different point of view if they are to partici-
pate usefully in the formation of a new type of entity: one con-
sisting of patterns in which, through the play of mirrors, reftections
are equivalent to real objects, that is, in which signs assume the
status of things signified. Th:::se patterns actualize possibiliti:::s
whose number, though it may be very great, is not unlimited, for
it is a fonction of the possible lay-out and balances which may be
effected between bodies whose number itself is finite. Finally,
and most important, these patterns produced by the conjunction
of contingent events (the turning of the instrument by the person
looking through it) and a law (namely that governing the construc-
tion of the kaleidescope, which corresponds to the invariant
element of the constraints just mentioned) project models of
intelligibility which are in a way provisional, since each pattern
can be expressed in terms of strict relations between its parts and
since these relations haœ no content apart from the pattern itself,
to which no object in the observer's experience corresponds - even
though, by such a manoeuvre, particular objective structure3, such
as those of snow crystals or certain types of radiolaria and
THE LOGIC OF TOTE:\IIC CLASSIFICATI0"1S
diatomaceae might be re\·ealed while their empirical basis werc

yet unknown, to the observer who had not yet seen them.
One can conceive that such a concrete logic may be possible. 1t

now remains to define its features and the way in which thcy can
be obserYed in the course of ethnographie enquiry. Both an
atfecti\'C and an intellectual aspect become apparent.
The beings which native thought endows with significance are
seen as exhibiting a certain affinity with man. The Ojibwa belie\'C
in a uni verse of supernatural beings:
Y et to call thcse hcings supcrnatural slightly misintcrprcts the lndians'
conception. They arc a part of the natural ordcr of the universc no Jess
than man himsdf, whom thcy rcscmblc in tht· possession of intelligence
and cmotions. Likc man, too, they arc male or fcrnale, and in somc cases
at kast may e\·en ha\'l' farnilics of their own. Sorne arc ticd down to
definite localitics, some mon· from place to place at will; senne arc fricndly
to lndians, othcr hostile (Jcnncss z, p. z9).
Other obserYations emphasize that the feeling of indcntification is
stronger than the sense of ditfcrcncc:
A llawaiian's oncncss with the living aspect of native phcnomcna, that
is, with spirit and Cod and with othcr pcrsons as souls, is not corrcctly
described by the word rapport, and certainly not hy such words as sym·
pathy, empathy, ahnormal, supernormal or neurotic, mystical or magical.
lt is not 'cxtra·scnsory', for it is partly of-the·scnsc anJ not·of·thc-scnscs.
It is just a part of natural consciousncss for the normal Hawaiian ...
(Handy and Pukui, Part \'I, vol. 6z, p. 124).
Th~ natiYes themseh-es arc somctimes acutely awarc of the
'concrete' nature of thcir science and contrast it sharply with that
of the whites:
\Ve know what the animais do, what are the nec<ls of the beaver, the
bear, the salmon, and othcr creatures, because long ago men marricd them
and acquire<l this knowlcdge from their animal wivcs. To<lay the priests
say we lie, but wc know better. The white man has bcen only a short timc
in this country and kr.ows very little about the animais; wc have livc<l here
thousands of years and were taught long ago by the animais thcmselves.
The white man writcs everything down in a book so that it will not be
forgotten; but our anccstors married the animais, learned ail their ways,
and passetl on the knowlcdge from one generation to anothcr (Jenness 3,
p. 540).
This disinterested, attentive, fond and affectionate lore acquired
and transmitted through the attachments of marriage and upbring-
ing is here described with such noble simplicity that it seems
37
THE SAVAGE MIND
superfluous to conjure u p the bizarre hypotheses suggested to

philosophers by too theoretical a view of the development of
human knowledge. Nothing here calls for the intervention of a
so-called 'principle of participation' or even for a mysticism
embedded in mctaphysics which we now perceive only through
the distorting lens of the established religions.
The way in which this concrete knowledge works, its means
and mcthods, the affective values with which it is imbued are to
be found and can be observed very close to us, among those of
our contemporaries whose tastes and profession put them in a
situation in relation to animais which, mutatis mutandis, cornes as
close as our civilization allows to that which is usual among all
hunting peoples, namely circus people and people working in
zoos. l\othing is more instructive in this respect, after the native
evidence just quoted, then the account given by the director of
the Zürich zoo of his first tête-à-tête - if one may so call it - with
a dolphin. He notes 'its exaggerated human eyes, its strange
breathing hole, the torpedo shape and colour of its body, the
completely smooth and waxy texture of its skin and not least its
four impressivc rows of equally sharp teeth in its beak-like mouth',
but describes his feelings thus:
Flippy was no fish, and when he looked at you with twinkling eyes from
a distance of less than two feet, you had to stifie the question as to whether
it was in fact an animal. So new, strange and extremely weird was this
creature, that one was tempted to consider it as some kind of bewitched
being. Dut the zoologist's brain kept on associating it with the cold fact,
painful in this connection, that it was known to science by the dull name,
Tursiops trnncatus (Hediger, p. 138).
Comment like this from the pen of a man of science is enough to

show if indeed it is necessary, that theoretical knowledge is not
incompatible with sentiment and that knowledge can be both
objective and subjective at the same time. It also shows that the
concrete relations between man and other living creatures some-
times, especially in civilizations in which science means 'natural
science', colour the entire universe of scientific knowledge with
their own emotional tone, which is itself the result of this primitive
identification and, as Rousseau saw with his profound insight,
responsible for all thought and society. But if a zoologist can
combine taxonomy and the warmest affection, there is no reason
THE LOGIC OF TOTEMIC CLASSIFICATfONS
to invoke distinct principles to explain the conjWlction of these

two attitudes in the thought of so-called primitive peoples.
F ollowing Griaule, Dieterlen and Zahan have established the

extensi,·eness and the systematic nature of native classification in
the Sudan. The Dogon <livide plants into twenty-two mainfamilies,
some of which are further divided into eleven sub-groups. The
twenty-two families, listed in the appropriate order, are divided
into two series, one of which is composed of the families of odd
numbers and the other of those of even ones. ln the former, which
symbolizes single births, the plants called male and female are
associated with the rainy and the dry seasons respectively. In the
latter, \vhich symbolizes twin births, there is the same relation but
in reverse. Each family is also allocated to one of three categories:
tree, bush, grass; '*' finally, each family corresponds to a part of the
body, a technique, a social class and an institution (Dieterlen I, 2).
Facts of this kind caused surprise when they were first brought
back from Africa. Very sirnilar modes of classification had, how-
ever, been described considerably carlier in America, and it was
these which inspircd Durkhcim's and Mauss's famous essay.
The reader is referred to it, but it is worth adding a f ew further
examples.
The ~avaho Indians, who regard themselves as 'great classificrs',
di vide living creatures into two categories on the basis of whether
they are or are not cndowed with speech. The category of creatures
without speech consists of animais and plants. Animais are divided
into three groups, 'running', 'flying' and 'crawling'. Each of these
groups is further divided in two ways: into 'travellers by land' and
'tra\·ellers by water' and into 'travellers by day' and 'travellers by
night'. The division into species obtained by this means is not
always the same as that of zoology. Thus birds grouped in pairs
on the basis of a classification into male and female are in fact
sometimes of the same sex but of different kinds. For the associa-
tion is based on the one hand on their relative size and, on the
other, on their place in the classification of colours and the fonction
• Among the Fulani: plants with \·ertical trunks, climbing plants, creeping
plants, respectively subdivided into plants with and without thorns, with and
without bark and with or without fruit (Hampaté Ba and Dieterlen, p. 23). Cf.
Conklin I, pp. 92-4 for a tripartite classification of the same type in the Philip-
pines ('tree', 'creeper', 'grass') and Colbacchini, p. 202, for one in Brazil among
the Bororo ('trees' = land; 'creepers' = air and 'marsh-plants' = water).
39
THE SAVAGE MIND
assigned to them in magic and ritual (Reichard, 1, 2).* But native

taxonomy is often precise and unambiguous enough to allow
certain identifications, for instance, the one made only a few years
ago of the 'Great Fly', mentioned in myths, with a tachinid
llystricia pollinosa.
Plants are named on the basis of three sorts of characteristics:
their supposed sex, their medicinal properties and their visual or
tactual appcarance (prickly, sticky, etc.). Each of these three
groups is subdivided into three further groups according to size
(large, medium, small). The classification is of the same kind
throughout the reserve of some twenty-five thousand square miles,
in spite of the dispersion of its sixty thousand occupants over this
rather extensive territory (Reichard, Wyman and Harris, Vesta!,
Elmore).
Each animal or plant corresponds to a natural element, itself
dependent on rites whose extreme complexity among the Navaho
is well known. The following correspondences are found in the
'Flint-Chant': crane - sky, red songbird - sun; eagle - mountain;
hawk - rock; bluebird - tree; hummingbird - plant; cornbeetle -
earth; heron - water (Haile, pp. 120-1).
The Hopi, like the Zuni who particularly engaged Durkheim's
and Mauss's attention, classify living creatures and natural
phenomena by means of a vast system of correspondences. The
facing table is based on the information scattered in several
authors. It is undoubtedly only a modest fragment of an entire
system, many of whose elements are missing.
Similar correspondences are also known among people whose
social structure is very much looser than that of the Pueblo: the
Eskimo sculptor of salmon uses the wood whose colour is closest
to that of their flesh for carving each species: 'Ali woods, they say,
are salmon' (Rasmussen, p. 198).
These are only a few of the examples which might be given.
There would be even more examples than there are, had ethnolo-
gists not often been prevented from trying to find out about the
complex and consistent conscious systems of societies they were
studying by the assumptions they made about the simpleness and
coarseness of 'primitives'. It did not occur to them that there
could be such systems in societies of so low an economic and
• In contrast to the Cancla of Brazil, who 'In ail known cases, showed them-
sclvcs to be informcd about sexuHI dimorphism' (Vanzolini, p. 170).
1 NORTH-WEST SOUTH-WEST SOUTH-EAST NORTH-EAST ZENITH NADIR
1
COLOURS yellow blue, green red white black multicoloured
1
A:-;'IMALS puma bear wild cat wolf vulture snake

1 1 1 1 1 1 1
BIR US oriole blue-bird par rot magpie swallow warbler

(Sialia)
1 1
TREES Douglas-fir white pine red willow aspen
1 1
Bt:SHES green rabbit- sage-brush cliff-rose grey rabbit-

brush (Artemisia) (Cozcania brush
(chrysothamnus) stansburiana) (ChrysothammtS)
FLOWERS mariposa lily larkspur (Castilleja) (Anogra) ~
(calochortus) (delphinium)
1 1 1 1 1 1 1
CORN yellow blue red white purple sweet
1 1 1 1 1 1 1
French bean Butter-bean dwarf bean lima bean vanous
...,
BEANS .....
(Î
(P haseolus (Phas. rnlg.) (Phaseolus >
ml{!.) 1 lunatus --!
.....
0
Beans were also subdivided into: z
light light white white blue en
black yellow black grey red
red brown spotted yellow pink
red etc.
black
THE SAV AGE MIND
technical level since they made the unwarranted assumption that

their intellectual level must be equally low. And it is only just
beginning to be realized that the older accounts which we owe to
the insight of such rare inquirers as Cushing do not describe
exceptional cases but rather forms of science and thought which
are extremelywidespread in so-called primitive societies. We must
therefore alter our traditional picture of this primitiveness. The
'savage' has certainly never borne any resemblance either to that
creature barely emerged from an animal condition and still a prey to
his needs and instincts who has so often been imagined nor to that
consciousness governed by emotions and lost in a maze of confusion
and participation. The examples given, and others which could be
added, are evidence of thought which is experienced in all the
exercises of speculation and resembles that of the naturalists and
alchemists of antiquity and the middle ages: Galen, Pliny, Hermes
Trismegistus, Albertus Magnus ... From this point of view
'totemic' classifications are probably doser than they look to the
plant emblem systems of the Greeks and Romans, which were
expressed by means of wreaths of olive, oak, laurel, wild celery,
etc., or again to that practised by the medieval church where the
choir was strewn with hay, rushes, ivy or sand according to the
festival.
Astrological herbalism distinguished seven planetary plants,
twelve herbs associated with signs of the zodiac and thirty-six
plants assigned to decantates and to horoscopes. Sorne had to be
gathercd on a particular day and at a particular time in order to
be effective. These were specified for each: Sunday for hazel nut
and olive; Monday for rue, clover, peonies and endives;. Tuesday
for vervain; \Vednesday for periwinkle; Thursday for vervain,
periwinkle, peonies, laburnum and cinquefoil if these were
intended for medicinal use; Friday for endives, mandrake and
vervain used for incantations; Saturday for crosswort and plantain.
In Theophrastus one even finds a system of correspondence
between plants and birds, in which the peony is associated with
the woodpecker, centaury with the triorchis and falcon and black
hellebore with the cagle (Delatte).
All this, which we are perfectly willing to attribute to a natural
philosophy worked out over the years by specialists with a tradition
of centuries behind them, is very closely n.·~roduced in exotic
societies. The Omaha Indians consider one of the main differences
42
THE LOGIC OF TOTEMIC CLASSIFICATIONS
between themselves and the whites to be that 'Indians never pick

flowers', that is, never picked them for pleasure. For 'plants have
sacred uses known (only) to the ~_ecret owners'. Even soapweed
which is commonly used for steam baths to cure toothache,
earache or rheumatism is lifted as though it were a sacred root:
In the hole from which the root was withdrawn a pinch of tobacco was
left. Often a knife or some money was left there also, and the taker of the
root uttered a brief prayer, 'I have taken what you have given, and I am
leaving this here for you. I want to lead a long life and to have no harm
strike me or my family' (Fortune I, p. 175).
\Vhen a medicine man of eastern Canada gathers roots or leaves

or bark for medicines, he is careful to propitiate the soul of each
plant by placing a tiny offering at its base. For he believes that
without the co-operation of the soul the mere 'body' of the plant
can work no cures (J enness I, p. 60 ).
The Fulani of the Sudan class plants in series, each being
related to a day of the week and to one of eight directions:
The plant ... must be gathered in accordance with thesc various
classifications ... Bark, root, !caves or fruits must be collected on the day
of the lunar month to which the plant corresponds, the lare, the 'guardian
spirit' of the flocks which is related to the sequence of the months, must be
invoked and the sun must be in the right position. Thus in giving direc-
tions the silatigi will say for instance: 'To do such and such, you must
take the leaf of a thorn y climber which is without bark on such and such a
day when the sun is in such and such a position, facing in such and such a
direction and invoking such and such a lare' (Hampaté Ba and Dieterlen,
p. 23).
Native· classifications are not only methodical and based on care-
fully built up theoretical knowledge. They are also at times
comparable from a formai point of view, to those still in use in
zoology and botany.
The Aymara Indians of the Bolivian plateau (who may be
descended from the legendary Colla to whom the great civilization
of Tiahuanaco is attributed) are able experimenters in the preser-
vation of foodstuffs. I t was by direct imitation of their technique
of dehydration that the American army was able during the last
war to reduce the rations of powdered potatoes sufficient for a
hundred meals to the volume of a shoe box. They were also
agriculturalists and botanists, and developed the cultivation and
taxonomy of the genus Solanum (of great importance to the
Indians since they live at an altitude of more than twelve thousand
43
THE SAVAGE .i\11.'.':D
fcct whcrc maizc will not ripcn) further than it has pcrhaps evcr
bccn donc.
ÜYcr two hundrcd and fifty varicties arc still distinguished in
nati\·c rncabuLtry and the figure was ccrtainl y higher in the past.
This taxonomy opcratcs by using a tcrm to dcsignate the variety
and adding a qualifying adjccti\'e for cach sub\'ariety. Thus the
Yaricty imilla 'girl' is subdividcd cither according to colour (black,
blue, white, rcd, blood-coloured) or according to othcr characteris-
tics such as grassy, insipid, cgg-shaped and so on. Thcre are about
twenty-two main \'arietics which arc subdiYided in this way. In
addition, thcre is a gcncral dichotomy bctwccn thosc which may
be caten after simple cooking and thosc which can only be eatcn
after bcing altcrnatcly frozen and fermcnted. A binomial taxonomy
also always uses critcria such as form (flat, thick, spiral, like cact11s
lcaf, lumpy, cgg-shapcd, in the shapc of an ox tonguc, etc.),
texture (mcaly, clastic, sticky, etc.), or 'sex' (boy or girl) (La
Barre).
It was a professional biologist who pointcd out how many
crrors and misunderst:mdings, some of which haYe only recently
bccn rcctificd, could have bcen avoided, had the older travellers
been content to rcly on nati\·c taxonomies instead of improvising
cntircly ncw <mes. The rcsult was that clevcn diffcrcnt authors
hetween thcm applicd the scicntific namc Canis azarae to thrce
distinct gcncra, cight spccies and nine differcnt sub-spccics, or
again that a single Yariety of the samc specics was rcfcrrcd to by
seYeral different names. The Guarani of Argentine and Paraguay,
on the other hand, work methodically with namcs composcd of
one, two or thrce tcrms. By this mcans they distinguish for
instance between large, small and medium felincs: the dyagua eté
is the suprcme example of the large fcline, the mharakadya eté of
the srnall wild cat. The mini (small) among the dyague (large)
correspond to the guasu (large) among the chin', that is, the
medium-sized frlines:
ln general, nati\·e tcrms can hc sai<l to constitutt' a wcll-concci\'c<l
system, and, with a pinch of salt, they can hc sai<l to hear sonH.· resem-
bbnce to our scientific nomenclature. Thcsc primitiYc lmlians <li<l not
lca\·c the naming of natural phcnomcna to chance. They asscmblc<l tribal
councils to <leci<le which tcrms best correspomle<l to the nature of spccics,
classifying groups and suh-groups with grcat precision. The preservation
of th(' indigL·nous tcm1s for tht· local fatma is not just a matter of piety and
int('grity; it is a duty to science (lknnler, pp. 2J-J. and Z-J.4).
THE LOGJC OF TOTE.MIC CLASSIFICATIONS
In a large part of the Cape York Peninsula in North Australia

foods are distinguished as 'plant' or 'animal' hy means of two
special morphemes. The \Yik :\Iu!l\;.an, a tribe living in the valley
a!ld estuary of the Archer River on the west coast, refine this
di\·ision. They prefix the name of every plant or food derived from
it with the term mai and every animal, as well as flesh or animal
food, with the term min. Similarly, yukk is used as a prefix for ail
names of trees or terms refcrring to a stick, a piece of wood or
wooden ohject, koi for kinds of string and fibre, wakk for grasses,
tukk for snakes, l~iimpan and wank for straw and string baskets
respectively. And the same sort of construction of names with the
prefix ark allows types of scenery and their association to this or
that characteristic Aora or fauna to he distinguished: ark tomp
sandheach proper, ark tomp nintiin dune country hehind the
beaches, ark pint'l coastal plains with hrackish water, etc.
The nati\·es arc acutely aware of the characteristic trces, underscrub
and grasses of each distinct 'association area', using this tt:rm in its
ecological sense. They are able to list in detail and without any hcsitation,
the characterü~tic trees in each, and also to record the string, resin, grasses,
and other products used in material culture, which they obtain from each
association, as well as the mammals and birds characteristic of each
habitat. lndeed, so detailcd and so accurate is their knowlcdge of these
arcas that they note the graduai changes in marginal areas ... My
informants werc able to relate without hcsitation the changes in fauna and
in food supply in each association in relation to the scasonal changes.
The natiw taxonomy allows the following Zoological and hotanical
genera, species and varieties to he distinguished: mai' watti'yi
(Dioscorea satii·a 'Var. rotunda, Bail.); yukk putt a (Eucalyptus
papuana) - yukk pont (E. tetrodonta); tukk pol (p_-rthon spilotes -
tukk oingorpiin (l'. amethystinus); min piink (Alacropus agilis) -
min ko'impia (M. rufus) - min lo'along (1H. giganteus), etc. It is
therefore not t'.Jo much to say, as the writcr in question does, that
the arrangement of animais a!1d plants, and the foods or tech-
nological materials derived from them, hears some resemblance
to a simple Linnaean classificatio'.1 (Thomso-1 pp. 165-7).
In the face of such accuracy and care one begins to wish that every
ethnologist were also a minerdogist, a botanist, a zoologist and
even an astronomer ... For Reichard's comment about the Navaho
applies not only to the Australians and Sudanese but to ail or
almost all native peoples:
45
THE SAVAGE MIND
Since the Navajo regard ail parts of the universe as essential to well-
being, a major problem of religious study is the classification of natural
objects, a subject that demands careful taxonomical attention. We need a
list, with English, scientific (Latin) and r'\avajo names of ail plants,
animais - especially birds, rodents, insects and worms - minerais and
rocks, shells and stars (Reichard, p. 7).
Indeed, it becomes increasingly apparent as time goes on that it is

not possible to interpret myths and rites correctly, even if the
interpretation is a structural one (not to be confused with just a
formai analysis) without an exact identification of the plants and
animais which are referred to or of such of their remains as are
directly used. Let us quickly look at two examples, one from
botany and the other from zoology.
In the whole, or nearly the whole, of North America, the plant
called 'sage' or 'sage-brush' plays a major part in the most diverse
rituals, either by itself or associated with and at the same time, as
the opposite of other plants: Solidago, Chrysothamnus, Gutierrezia.
This all remains anecdotal and unsystematic until one enqufres
into the exact nature of American 'sage', which is nota labiate but
a composite plant. The vernacular term in fact covers several
varieties of wormwood (Artemisia) (carefully distinguished in
native nomenclatures and each given a different ritual fonction).
This identification, which was completed in an enquiry into
popular pharmacopoeia all over the world, shows that in North
America, as in the Ancient World, artemisia is a plant with a
connotation of feminine, lunar, nocturnal, mainly used in the
treatment of dysmenorrhoea and difficult child births. * A similar
investigation of the other group of plants makes it clear that it
consists of the same species under different names or of species
assimilated in native thought on account of their yellow flowers
and their use as dyes or medicine (for curing disorders of the
urinary tract, that is, of the male genitals). We thus have a set,
symetrically opposed to the first, with a connotation of male,
sun, day, and the quality of sacredness belongs to the pair as a
whole rather than to each plant or kind of plant separately. This
system has been made explicit in the analysis of some rituals, such
as that of eagle hunting among the Hidatsa (though only thanks
" Artemisia seems to have had a female connotation in ancient Mexico also
since women were adorned with it to dance in the June f estivais in honour of the
Goddess Huixtociuatl (Reko, pp. J9, 75; Anderson and Dibble, pp. 88--<J). Cf.
Paso and Troncoso for everything relating to Nahuatl ethnobotany.
Fig 1 Artemisia frigida (from C. Ledebour, Icones Plantarum).
47
THE SAVAGE MIND
to the unusual insight of an observer, G. L. Wilson, pp. 150-1),

and it ean also be extended to other cases where it has not been
made clear. The Hopi Indians, for instance, make 'prayer-sticks'
by typing sprigs of Gutierrezia euthamiae and Artemisia f rigida to
the feathers which form their main element, or, again, this same
people descrihe the four cardinal points in terms of different
associations of Artemisia and Chrysothamnus (cf e.g. Voth l passim;
2, p. 75 sq; 5, p. 130).
Ali this hegins to suggest a method of formulating, and some-
times even solving, various problems which have so far been
neglected. For example, the dichotomy of the 'feminine' pole
which the Navaho make in terms of Chrysothamnus (which is
however male in the main pair of opposites) and J>enstemon, a
scrophulariacea (Vesta!), can be represented schematically in the
following way:
,,
Cfemale> cm~le)
,Artemisia
,,
,/ Chrysothamnus
, ,,
/
/ , ,/
,
,
/
,,
,, ,,
,,/ ,/I
(to assist Penstemon Chrysothamnus
,,
birthJ= (female birthJ /Cmale birthJ
,
,, ,,
o' .c:,.'
The meaning of certain ritual features found in peoples who

are geographically distant and differ in language and culture, also
hecomes apparent. A rough system is in evidence throughout the
continent. Finally, for comparative study the analogy between the
position of Artemisia in the Old and the New \Vorld opens a new
field of enquiry and consideration, not least on account of the part
played in the New \Vorld hy Solidago virga aurea, in other words,
a 'Golden Bough'.
The second example concerns rites already mentioned above:
those of eagle hunting among the Hidatsa who, like many other
American peoples, regard this as an eminently sacred pursuit.
According to the Hidatsa, the hunting of eagles was taught to
48
THE LOGIC OF TOTEl\IIC CLASSIFICATIONS
Fig 2 Soli<lago virga aurea (from Torrey Botanical Club).
49
THE SAVAGE MIND
men by the supernatural animais who originally invented its tech-

niques and procedures. The myths refer to them vaguely as 'bears'.
lnformants seem to hesitate between the small black bear and
the animal termcd glutton, wolverine or carcajou (Cula luscus).
The specialists on the Hidatsa, such as \Vilson, Densmore, Bowers
and Beckwith, did not altogether ignore the problem but they did
not attach great importance to it, since it was after all a mythical
animal which was in question, and the identification of a mythical
animal might seem pointless, if not indeed impossible. ln fact,
however, the whole interpretation of the ritual depends on this
identification. So far as interpretation of eagle hunting ritual is
conccrned, bears are no help. But wolverines or carcajous - a
Canadian adaptation of an lndian word meaning 'bad-temper' are
a different matter, for they have a very special place in folklore.
ln the mythology of the Algonkin of the north-east the wolverine
is the animal of craft and cunning. It is hated and feared by the
Eskimos of Hudson Bay as well as by the western Athapaskan and
coastal tribes of Alaska and British Columbia. If one pieces together
what is known about all these peoples one arrives at the same
explanation as that obtained independently from trappers by a
contemporary geographer: 'Gluttons are almost the only members
of the weasel family which cannot be trapped. They amuse them-
selves in stealing not only trapped animais but also the hunter's
traps. The onlyway to get rid ofthem is to shoot them' (Brouillette,
p. 55). Now, the Hidatsa hunt eagles by hiding in pits. The eagle
is attracted by a bait placed on top and the hunter catches it with
his bare hands as it perches to take the bait. And so the technique
presents a kind of paradox. Man is the trap but to play this part he
has to go clown into the pit, that is, to adopt the position of a
trapped animal. He is both hunter and hunted at the same time.
The wolverine is the only animal which knows how to deal with
this contradictory situation: not only has it not the slightest f ear
of the traps set for it; it actually competes with the trapper by
stealing his prey and sometimes even his traps.
It follows, if the interpretation which 1 have begun to give is
correct, that the ritual importance of eagle hunting among the
Hidatsa is at least partly due to the use of pits, to the assumption
by the hunter of a particular lorv position (literally, and, as we have
just seen, figuratively as well) for capturing a quarry which is in
the very highest position in an objective sense ( eagles fly high) and
50
THE LOGIC OF TOTE:\1IC CLASSIFICATIONS
also from a mythical point of view (the eagle being a t the top of
the mythical hierarchy of birds).
Analysis of the ritual shows that it accords in every detail with
the hypothesis that there is a dualis~ between a celestial prey and
a subterranean hunter, which at the same time evokes the strongest
possible contrast between high and low in the sphere of hunting.
The extreme complexity of the rites which precede, accompany and
conclude an eagle hunt is the counterpart of the exceptional
position which eagle hunting occupies within a mythical typology
which makes it the concrcte expression of the widest possible
distance between a hunter and his game.
Sorne obscure features of the ritual become clear at the same
time, in particular the significance and meaning of the myths
which are told du ring hunting expeditions. They ref er to cultural
heroes, capable of being transformed into arrows and masters of
the art of hunting with bows and arrows; and thercforc, in their
guise of wild cats and racoons, doubly inappropriatc for the role
of hait in eagle hunting. Hunting with bows and arrows involves
the region or space immediatdy above the earth, that is, the
atmospheric or middle sky: the hunter and his game meet in the
intermediate space. Eagle hunting, on the other hand, separates
them by giving them opposite positions: the hunter below the
ground and the game close to the cmpyrean sky.
Another striking feature of eagle hunting is that women have a
beneficial effect during their periods. This is contrary to the
belief held almost universally by hunting peoples, including the
Hidatsa themselves in the case of ail except eagle hunting. What
has just been said explains this detail also, when it is remembered
that in eagle hunting, conceived as the narrowing of a wide gulf
between hunter and game, mediation is effected, from the technical
point of view, by means of the hait, a piece of meat or small piece
of game, the bloodstained carcass of which is destined to rapid
decay. A first hunt to procure the bait is necessary in order for
the second hunt to take place. One hunt involves the shedding of
blood (by means of bows and arrows), the other does not (eagles
are strangled without any effusion of blood). The one hunt, which
consists in a close union of hunter and game, fumishes the means
of effecting a union between what is so distant that it looks
at first as if there is a gulf which cannot be bridged - except,
precisely, by means of blood.
SI
THE SAVAGE :\11:\D
:\lenstruation acquires a positiYe signific:rnce from three points

of view in a system of this kind. From a strictly formai point of
view, since one hunt is the re\·erse of the other, the role attributed
to menstruation is accordingly reversed. It is harmful in one case
because its similarity is too great, but it becomes beneficial in the
other where it has not only a metaphorical but also a metonymical
sen se: it ernkes the hait as blood and organic deca y and the hait
is a part of the system. From the technical point of view the
hloodstained carcass, soon to be carrion, which is close to the
living hunter for hours or even days is the means of effecting the
capture, and it is significant that the same native term is used for
the embrace of lovers and the grasping of the hait by the hird.
Finally, at a semantic level, pollution, at least in the thought of
the :\orth American Indians, co:i.sist:> in too close a conjunction
between two things each meant to remain in a state of 'purity'.
In the hunt at close quarters menstrual periods always risk
introducing excessiYC union which would lead to a saturation of
the original relation and a neutralization of its dynamic force by
redunda•1cy. In the hunt at a dista'1ce it is the re\·erse. The
conjunction is inadequate and the only means to remedy its
deficiency is to allow pollution to enter. Pollution appears as
periodicity on the axis of successions or as putrefaction on that of
simultaneities.
One of these axes corresponds to the mythology of agriculture
and the other to that of hunting. This analysis therefore makes it
possible to arrive at a general system of rcference allowing the
detection of homologies hetween themes, whose elaborate forms
do not at first sight seem related in any way. This result is of great
importance in the case of eaglc hunting sincc it is to be found in
ditfercnt forms (though always strongly imbued with ritual) almost
all over the American continent and among people of different
cultures, some hunters and some agriculturalists. The relatively
minor but positi,·e rolc attributcd to pollution by the Hidatsa, the
:\Iandan and the Pawnee (the variations of which can be interpreted
as a fonction of the social organization of each trihe) can therefore
be regarded as a particular case of something more general.
Another particular case is illustrated by the Puehlo myth of a
man hetrothed to an eagle-girl. The Puchlo connect this myth
with another, that of the 'corpse-girl' and 'ghost-wife'. In this
case pollution has a fonction which is strong (inrnh-ing the death,
instead of the success, of the hunter). For among the Pueblo, as

the myths explain, the blood of rab bits who are the prime object of
the ritual hunt must not be shed. Among the Hidatsa, on the other
hand, it must be shed so that they cân be used as the means of the
ritual hunt par excellence: the hunt of eagles, whose blood may
not be shed. The Pueblo indeed capture and rear eagles but they
do not kill them and some groups even refrain altogether from
keeping them for fear that they should forget to feed them and
let them die of hunger.
To return brieflytothe Hidatsa :other problems arise in connection
with the mythical role of the wolverine further north, in a region at
the edge of the major area of the diffusion of this animal species. *
1 raise this point to emphasize the fact that historical and geo-
graphical problems, as well as semantic and structural ones, are
ail related to the exact identification of an animal which fulfils a
mythical fonction: Gulo luscus. This identification has an impor-
tant bearing on the interpretation of myths among peoples as far
from the region of the wokerine as the Pueblo or even, in the
heart of tropical America, the Sherente of Central Brazil who also
have the myth of the ghost wife. The suggestion is not however
that ail these myths were borrowed from a northern culture in
spite of the great distance. Any question of this kind could arise
only in the case of the Hidatsa in whose myths the wolverine
figures explicitly. The most that can be said in the other cases is
that analogous logical structures can be constructed by means of
different lexical resources. It is not the elements themselves but
only the relations between them which are constant.
This last remark leads us to another difficulty which must be

considered. The accurate identification of every animal, plant,
• The Hidatsa seem to have lived at various points in the state of N"orth
Dakota for as long as their traditions go back.
As for the wolverine: it 'is a circumpolar species belonging to the northern
forested areas of both continents. In North America it fonnerly ranged from the
northern limit of trees south to New England and New York, and down the
Rocky :\fountains to Colorado, and down the Sierra Nevada to near Mount
\Vhitney, California' (Nelson, p. 427). The Common Wolverine is found 'from
the Arctic Ocean and Baffin Bay southward and from the Pacifie to the Atlantic,
reaching the extreme north-eastern United States, Wisconsin, Michigan, Minne-
sota, North Dakota, and down the Rocky Mountains into Utah and Colorado'
(Anthony, p. III). Species which appear to be synonymous have been reported
in the mountains of California and at Fort Union, North Dakota (id.).
53
THE SAVAGE ::\11ND
stone, heavenly body or natural phenomenon mcntioned in myths

and rituals is a complcx task for which the ethnographcr is rarely
equipped. Even this is not however eno\lgh. It is also necessary
to know the role which each culture gives them within its own
system of significances. It is of course useful to illustrate the
wcalth and precision of native observation and to describe its
methods: long and constant attention, painstaking use of all the
senses, ingenuity which does not despise the methodical analysis
of the droppings of animais to discover their eating habits, etc.
Of all these minute details, patiently accumulated over the cen-
turies and faithf ully transmitted from generation to generation,
only a few are however actually employed for giving animais or
plants a significant fonction in the system. And it is necessary to
know which, sin ce the y are not constant from one society to another
so far as the same species is concerned.
The Iban or Sea Dayaks of South Borneo derive omens by
interpreting the song and flight of a number of species of birds.
The rapid cry of the Crested Jay (Platylophus galericulatus Cuvrer)
is said to resemble the crackling of burning wood and so presages
the successful firing of a family's swiddens. The alarm cry of a
Trogon (Harpactes diardi Temminck) is likened to the death
rattle of an animal being slain and augurs good hunting. Again,
the alarm cry of the Sasia abnormis Temminck is supposed to get
rid of the cYil spirits which haunt the crops by scraping them off
sin ce it resembles the sound of a scraping knif e. The 'laugh' of
another Trogon (Harpactes dm.:auceli Temminck) is a good omen
for trading expeditions and because of its brilliant red breast it is
also associated with the renown attending successful war and
distant voyages.
It is obvious that the same characteristics could have been given
a diff erent meaning and that different characteristics of the same
birds could have been chosen instead. The system of divination
selects only some distinctive features, gives them an arbitrary
meaning and restricts itself to seven birds, the selection of which
is surprising in view of their insignificance. Arbitrary as it seems
when only its individual terms are considercd, the system becomes
coherent when it is seen as a whole set: the only birds used in it
are ones whose habits readily lend themselves to anthropomorphic
symbolism and which are easy to distinguish from each other by
means off eatures that can be combined to fabricate more complex
54
messages (Freeman). Ne,·ertheless when one takes account of the

wealth and diversity of the raw material, only a f ew of the innumer-
able possible elements of which are made use of in the system,
there can be no doubt that a considerable number of other systems
of the same type would have been equally coherent and that no
one of them is predestined to be chosen by all societies and all
civilizations. The terms never have any intrinsic significance.
Their meaning is one of 'position' - a fonction of the history and
cultural context on the one hand and of the structural system in
which they arc called upon to appear on the othcr.
Vocabulary already shows this selectiveness. In Navaho the
wild turkey is the bird which 'pecks' while the woodpecker is the
bird which 'hammers'. \Vorms, maggots and insects are grouped
under a generic term, meaning swarming, eruption, boiling,
effen·escence. Insects are thus thought of in their larval state
rather than in their chrysalis or adult form. The name of the lark
refers to its extendcd hind claw whilc the English tcrm 'horncd
lark' derives from the protubcrant feathcrs of its hcad (Reichard 1,
pp. IO-II).
When he began his study of the classification of colours among
the Hanun6o of the Philippines, Conklin was at first baffied by
the apparent confusions and inconsistencies. These, however,
disappeared when informants were asked to relate and contrast
specimens instead of being asked to define isolated ones. There
was a cohcrent system but this could not be understood in terms
of our own system which is founded on two axes: that of brightness
(value) and that of intensity (chroma). All the obscuritics dis-
appeared when it became clear that the Hanun6o system also has
two axes but different ones. They distinguish colours into relatively
light and relatively dark and into those usual in fresh or succulent
plants and those usual in dry or desiccated plants. Thus the natives
treat the shiny, brown colour of newly eut bamboo as relatively
green while we should regard it as nearer red if we had to classify
it in terms of the simple opposition of red and green which is
found in Hanun6o (Conklin 2).
In the same way, very closely related animals may often appear
in folklore but with a different significance in different instances.
The woodpecker and other birds of the same genus are a good
example. As Radcliffe-Brown has shown (z), the Australians'
interest in the tree-creeper is due to the fact that it inhabits the
55
THE SAVAGE MIND
hollows of trees; but the Indians of the North American prairies

pay attention to quite a different feature; the red-headed wood-
pecker is believed to be safe from birds of prey since its remains
are never found (Schoolcraft). The Pawnee of the Upper Missouri,
a little further south (in common apparently with the ancient
Romans) associate the woodpecker with tempests and storms
(Fletcher 2), while the Osage associate it with the sun and stars
(La Flesche). The lban of Borneo mentioned above give a sym-
bolic role to one variety of woodpecker ( Blythipicus rubiginosw;
Swainson) because of its 'triumphal' song and the solemn warning
character attributed to its alarm cry. It is not of course exactly the
same bird which is in question in all these cases but the example
helps us to understand how different peoples can use the same
animal in their symbolism, employing unrelated characteristics,
habitat, meteorological association, cry, etc., the live or the dead
animal. Again, each feature can be interpreted in different ways.
The Indians of the south-west United States, who are agricult
uralists, regard the crow primarily as a garden pest, while the
lndians of the north-west Pacifie coast, who live entirely by fishing
and hunting regard it as a devourer of carrion, and consequently,
of excrement. The semantic load of Corvus is different in the two
cases: plant in one, animal in the other; man's rival when its
behaviour is like his own, his enemy when it is the reverse.
Bees are a totemic animal in Africa as well as in Australia. But
among the Nuer they are a secondary totem associated with
pythons, because the two species have similarly marked bodies. A
man who has pythons as a totem also refrains from killing bees or
eating their honey. There is an association of the same type
between red ants and cobras, because the literai meaning of the
latter is 'the brown one' (Evans-Pritchard 2, p. 68).
The semantic position of bees among the Australian tribes of
Kimberley is very much more complex. Their languages have
noun classes. Thus the Ngarinyin recognize three successive
dichotomies: first: into animate and inanimate things, then oi the
animate class into rational and irrationâl, and finally, of the former
into male and female. ln the languages with six classes, the class
of manufactured articles included honey as well as canoes on the
grounds that honey is 'manufactured' by bees just as canoes are
manufactured by men. It is thus understandable that in the
languages which have lost some classes animais and manufacturcd

goods have corne to be grouped together (Capell).
There are cases in which one can make hypotheses with
regard to the logical nature of classification, which appear true
or can be seen to eut across the natives' interpretation. The
Iroquois peoples were organized into clans whose number and
names varied considerably from one to another. It is not however
unduly difficult to detect a 'masterplan' which is based on a fonda-
mental distinction between three kinds of clan: water clans (turtle,
beaver, eel, snipe, heron), land clans (wolf, deer, bear) and air
clans (sparrowhawk, ?ball). But even so the case of aquatic birds
is determined arbitrarily for being birds they could belong to the
air rather than to the water, and it is not certain that research into
the economic life, techniques, mythical imagery and ritual
practices would supply an ethnographie context rich enough to
decide this point.
The ethnographie data on the central Algonquain and the neigh-
bouring Winnebago suggests a classification into five categories
which correspond, respectively, to land, water, the subaquatic
world, the sky and the empyream. '*' The difficulties begin when
one tries to classify each clan. The :\1enomini have fifty clans
which seem to be divisible into quadrupeds on dry land (wolf, dog,
deer), quadrupeds inhabiting swampy places (elk, moose, marten,
beaver, pekan), 'terrestrial' birds (eagle, hawk, raven, crow), aquatic
birds (crane, heron, duck, coot) and finally subterranean animais
(Hoffman, pp. 41-2). But this last category is particularly recalci-
trant as man y of the animais included in it (bear, turtle, porcupine)
could also be included in other classes. The difficulties would be
evert greater in the remaining classes.
Australia presents similar problems. Durkheim and Mauss,
following Frazer, discussed the global classifications of tribes
like the Wotjobaluk, who bury their dead facing in a particular
direction which depends on the clan. (See illustration on
p. 58.)
'*' 'Among the \Vinnebago, a number of other Siouan, and Central Algonquain
tribes, there was a fivefold classification; earth animais, sky animais, empyrean
animais, aquatic animais, and subaquatic animais. Among the Winnebago the
thunderbird belong to the empyrean; the eagle, hawk, and pigeon, to the sky;
the bear and wolf, to the earth; the fish, to the water and the water spirit, bclow
the water' (Radin I, p. 186).
57
THE SAVAGE MIND
Petican
Carpetsnake White Cockatoo
A Tuber Deaf Adder
BlackCockatoo The Sea

Pelican
(Frazer, vol. i, p. 455)
The available data are certainly fragmentary, but one cannot fail
to :10tice a rough system, which, indeed, only looks rough to the
onlooker because the ethnographie background necessary for its
interpretation is almost entirely missing. The white cockatoo,
'diurnal', is next to the sun, and the black cockatoo, which is
almost directly opposite, is itself next to the tubers, 'chthoni<in'
plants, and on the same axis as the cave, which is also 'chthonian'.
Snakes are on one axis, and the 'sea' creatures, pelican, sea, hot
wind, also seem to be grouped according to axes. But does this
wind belong to land or sea? W e do not know and, as so often, we
must go to a geographer or meteorologist or to a botanist, zoologist
or geologist, for the solution to this ethnographie problem.
The truth of the matter is that the principle underlying a classifi-
cation can never be postulated in advance. It can only be discovered
a posteriori by ethnographie investigation, that is, by experience. *
• 1 hcrc includc scvcral pages of a paper intended for the Mélanges Alexandre
Koyré.
58
THE LOGIC OF TOTEJ\1IC CLASSIFICATIONS
The case of the Osage, who are southern Sioux, is illuminating

because their classifications appear to be systematic. The Osage
divide animate and inanimate things into three categories. These
are associated respectively with the sky (sun, star, crane, heavenly
bodies, night, the constellation of the Pleiad, etc.), with water
mussel, turtle, Typha lat1jolia (a type of rush), mist, fishes, etc.)
and with dry land (black bear and white bear, porcu-
pine, deer, eagle, etc.). The position of the eagle would be
incomprehensible •sere it not known that in Osage thought caglcs
are associated with lightning, lightning with fire, fire with coal and
coal with the earth. The eagle is thus one of the 'masters of coal',
that is, a land animal. Similarly, and with nothing to suggest this
in adYance, the pelican has a symbolic role on account of the great
age it reaches, metal on account of its hardncss. The turtle with a
serrated tail, an animal of no practical use, is oftcn involved in
rites. The reason for its importance could not be understood with-
out the further knowledge that the number thirteen has a mystic
value for the Osage. The rising sun emits thirteen rays, which arc
di,·ided into a group of six and a group of seven corresponding
respectiœly to right and left, land and sky, summer and winter.
The tail of this spccics of turtle is said to have sometimes six and
sometimes seven scrraturcs. Its chcst thcrcforc rcprescnts the
nuit of the sky and the grey line across it the l\Iilky Way. I t would
be equally impossible to predict the pan-symbolic fonction
according to the elk, whose body is a veritable imago mundi: its
coat represents grass, its hams hills, its flanks plains, its backbone
the skyline, its neck valleys and its antlers the whole hydrographie
network.
A few Osage interpretations can thus be reconstructed, since an
enormous amount of documentation is available. The data were
collected by La Flesche \vho was himself the son of an Omaha
chief and particularly aware of all the intricacies of native thought.
But in the case of an almost extinct tribe like the Creek the
difficulties are unsurmountable. The Creek were at one time
divided into more than fifty matrilineal totemic clans. These were
named mainly after animais but also in some cases after plants or
meteorological, geological and anatomical phenomena ( dew and
wind, sait, pubic hair, respectively). The clans were grouped into
phratries, the villages were also divided into two groups possibly
corresponding to land animais and sky animais, although there is
59
THE SAVAGE MIND
nothing to suggest this in their designation as 'people of a different

sprech' and 'white people' or as 'reds' and 'whites'. But why, one
may ask, are totems distinguished into 'uncles' and 'nephews' (in
the same way that the Hopi distinguish totems into 'mother's
brothers' on the one hand and 'father', 'mother' or 'grandmother'
on the other)• and why, more particularly, given this division, is it
sometimes the less 'important' animal which has the higher
position, wolves, for instance, being 'uncles' of bears and wildcats
those of the large felines called 'panthers' in the southern United
States? Again, why is the clafl of the alligator associated with that
of the turkey (unless perhaps because both lay eggs) and that of
the racoon with the clan of the potato? In Creek thought the sicle
of the 'whites' is that of peace but the fieldworker was given a
hopelessly vague explanation: the wind (the name of a 'white' clan)
brings good, that is, peaceful, weather; the bear and the wolf are
exceedingly watchful animais and therefore useful in the interests
of peace, etc. (Swan ton 1, p. 108 ff.).
The difficulties which these examples illustrate are of two types,
extrinsic and intrinsic. The extrinsic difficulties arise from our
lack of knowledge of the (real or imaginary) observations and the
facts or principles on which classifications are based. The Tlingit
Indians say that woodworms are 'clever' and 'neat' and that land
otters 'hate the smell of human excretion' (Laguna, pp. 177, 188).
The Hopi believe that owls have a favourable influence on peach
trees (Stephen, pp. 78, 91, 109). Ifthese attributes were taken into
account in classifying these animais we might hunt for the key
indefinitely did we not happen to possess these small but precious
dues. The Ojibwa Indians of Parry Island number the eagle and
squirre! among their 'totems'. Fortunately a native text explains
that these animals are included as symbols of the trees they each
inhabit: hemlock trees (Tsuga canadensis) and cedar trees (Thuja
occidentalis) respectively (Jenness 2). The interest which the
Ojibwa have in squirrels is therefore really an interest in a kind of
tree. lt has no connection with the great interest which the Asmat
of New Guinea also take in squirrels, for different reasons:
• An interpretation has been suggested in the case of an analogous distinction
in an African tribe: 'God is the father of the greater spirits of the air, and the
lesser of them are said to be children of his sons, of his lineage. The totemic
spirits are ofren said to be children of his daughters, that is, they are not of his
lineage, which is the Nuer way of placing them yet lower in the spiritual scale'
(E\'3ns-Pritchard 2, p. 119).
60
Parrots and squirrels are famous fruiteaters ... and men about to go
headhunting feel a relationship to these beings and call themselves their
brothers ... (because of the) parallelism between the human body and a
tree, the human head and its fruit (~egwaard, p. 1034).
A different kind of consideration leads to squirrels being forbidden

to pregnant women among the Fang of the Gabon. Squirrels
shelter in the holes of trees and a future mother who ate their flesh
would run the risk of the foetus copying a squirrel and refusing to
leave the uterus.* The same reasoning could equally well apply to
weasels and badgers, who live in burrows, but the Hopi follow a
different line of thought: they hold that the meat of these animais
is favourable to child-bearing because of their habit of working
their way through the ground and 'getting out' at some other
place when they are chased into a hole. They therefore help the
baby to 'corne out quickly' - as a result of which they can also be
called on to make rain fall (Voth 1, p. 34 n.).
An Osage ritual incantation makes a puzzling association
between a flower - Blazing Star (Lacinaria pycnostachya), a plant
used for food - maize, and a mammal - the bison. It would be
impossible to understand why they associate these things if an
independent source did not bring to light the fact that the Omaha,
who are closely related to the Osage, hunted bison during the
summer until blazing stars were in flower in the plains; they then
knew that the maize was ripe and returned to their villages for the
harvest (Fortune 1, pp. 18--<)).
The intrinsic difficulties are of a different kind. They are due
not to our lack of knowledge of the objective characteristics used to
establish a connection between two or more terms in native
thought but to the polyvalent nature of logics which appeal to
several, formally distinct types of connection at the same time.
The Luapala of Northern Rhodesia provide a good illustration of
this feature. Their clans have the names of animais, plants or
manufactured articles and they are not 'totemic' in the usual sense
•And not only squirrels: 'The worst danger threatening pregnant women is
from animais who live or are caught in any sort of hole (in the ground, in trees).
One can positively speak of a horror vacui. If a pregnant woman eats an animal
of this kind, the child might also want to stay in its hole, "in the belly'', and a
difficult birth is to be expected. Similarly the parents must not, during this
period, take out any birds' nests which have been built into the hollows of trees,
and one of my employees, who had made a woman pregnant, absolutely refused
to make me a mode! of a loaf of cassava, on the grounds that it would have to be
hollow' (Tessman, vol. l, p. 193).
61
THE SAVAGE MIND
of the term, but, as among the Bemba and the Ambo, joking
relationships link them in pairs. The reasoning behind this has,
from our point of view, the same interest. As 1 showed in an
earlier book and am continuing to establish here, so-called
totemism is in fact only a particular case of the general problem of
classification and one of many examples of the part which specific
terms often play in the working out of a social classification.
The following clans stand in a joking relationship to each other
among the Luapula: the Leopard and Goat clans bccause the
leopards eat goats, the Mushroom and Anthill clans because
mushrooms grow on anthills, the :\Iush and Goat clans because
men like meat in their mush, the Elephant and Clay clans because
women in the old days used to carve elephants' footprints in the
ground and use these natural shapes as receptacles instead of
fashioning pots. The Anthill clan is linked with the Snake clan
and also with the Grass clan because grass grows tall on anthills
and snakes hide there. The Iron clan jokes with ail clans with
animal names because animais are killed by metal spears and
bullets. Reasoning of this kind allows the definition of a hierarchy
of clans: the Leopard clan is superior to the Goat clan, the Iron
clan to the animal clans and the Rain clan to the 1ron clan because
rain rusts iron. l\foreover the rain clan is superior to ail the other
clans because animais would die without it, one cannot make mush
(a clan name) without it, clay (a clan name) cannot be worked
without it, and so on (Cunnison, pp. 62-5).
The Navaho give man y different justifications of the virtues they
ascribe to medicinal plants and their modes of employing them:
the plant grows near a more important medicinal plant; one of the
plant's organs looks like a part of the body; the odour (or feel or
taste) of the plant is 'right'; the plant makes water the 'right' colour;
the plant is associated with an animal (as food, or in habitat, or by
contact); the knowledge was revealed by the gods; its uses were
learnt from someone else without any explanation: the plant is
found near a tree struck by lightning; it is good for a certain ailment
in that part of the body or an ailment with similar effects, etc.
(Vesta!, p. 58). The terms used to differentiate plant names among
the Hanunôo belong to the following categories: leaf shape, colour,
habitat, size/dimension, sex, habit of growth, plant host, growing
time, taste, smell (Conklin 1, p. 131 ).
These examples, together with those given earlier, make it clear
62
that such systems of logic work on several axes at the same time.
The relations which the y set u p between the terms are most
commonly based on contiguity (snake and anthill among the
Luapula and also the Toreya of South India)* or on resemblance
(red ants and cobras which, according to the Nuer, resemble each
other in 'colour'). In this they are not formally distinct from other
taxonomies, e\·en modern ones, in which contiguity and resem-
blance also play a fondamental part: contiguity for discovering
things which 'belong both structurally and fonctionally ... to a
single system' and resemblance, which does not require member-
ship of the same system and is based simply on the possession by
abjects of one or more common characteristics, such as ail being
'yellow or ail smooth, or ail with wings or ail ten feet high (Simpson,
pp. 3-+).
But other types of relation intervene in the examples we have
just examined. Relations may be established, in effect, on either
the sensible Ie,·el (the bodily markings of the bee and the pythons)
or on the intelligible level (the fonction of construction common
to the bce and the carpentcr): the samc animal, the bec, fonctions,
as it were, at diff erent levcls of abstraction in two cultures. Again,
the connection can be close or distant, synchronie or diachronie
(the relation between squirrels and cedars for instance on the one
hand, and that between pottery and elephants' footprints on the
other), static (mush and goat) or dynamic (fire kills animals, rain
'kills' fire; the flowering of a plant indicates that it is time to
return to the village), etc.
It is probable that the number, nature and 'quality' of these
logical axes is not the same in every culture, and that cultures could
be classified into richer and poorer on the basis of the formai
properties of the systems of reference to which they appeal in the
construction of their classifications. However, even those which
are the least well endowed in this respect employ logics of several
dimensions, the listing, analysis and interpretation of which would
require a wealth of ethnographie and general information which
is all too often lacking.
•'The members of the sept, at times of marriage, worship anthills, which are
the homes of snakes' (Thurston, vol. VII, p. 176). Similarly in l\"ew Guinea:
'certain types of plants, as well as the animais and plants parasitic on them, are
thought of as belonging to the same mythical and totemic unit' (Wirz, vol. II,
p. 21).
THE SAVAGE MIND
Two types of difficulties characteristic of 'totemic' logics have so

far been mentioned. The first is that we do not usually know
exactly which plants or animais are in question. We have seen that
an approximate identification is not adequate since native obser-
vations are so precise and finally shaded that the place of each
element in the system often depends on a morphological detail or
mode of behaviour definable only at the level of varieties or sub-
varieties. The Eskimos of Dorset carved effigies of animais out of
pieces of ivory no bigger than the head of a match with such
precision that, when they look at them under microscopes,
zoologists can distinguish varieties of the same species such as, for
example, the Common Loon and the Red Throated Loon
(Carpenter ).
The second difficulty mentioned is that each species, variety
or subvariety could suitably fill a considerable number of different
ro!es in symbolic systems in which only certain roles are effectively
ascribed to them. The range of these possibilities is unknown to
us and for an understanding of the choices which are made wc
need not only ethnographie data but also data from other source:> -
zoological, botanical, geographical, etc. The rare cases where the
data are adequate make it clear that even neighbouring cultures
may construct totally different systems out of superficially identical
or very similar elements. If North American peoples can regard
the sun in some cases as a 'father' and benefactor and in others as
a cannibalistic monster thirsting for human flesh and blood, there
seems to be no limit to the variety of interpretations to be expected
when it is something as specific as the subvariety of a plant or
bird which is in question.
A comparison of colour symbolism among the Luvale of
Rhodesia and some Australian tribes of the north-east of the state
of South Australia provides an example of a case in which the
same very simple structure of opposites recurs but the semantic
loads are reversed. ln the Australian tribes in question, the mem-
bers of the matrilineal moiety of the deceased paint themselves
with red ochre and approach the body, while members of the other
moiety paint themselves with white clay and remain at a distance
from it. The Luvale also use red and white soi! but white clay and
white meal are used by them as offerings to ancestral spirits and
red clay is substituted on the occasions of puberty rites because
redis the colour of life and fertility (C. M. N. White 1, pp. 46-7).*'
White represents the 'unstressed' situation in both cases, while
red - the chromatic pole of th~_ opposition - is associated with
death in one case and with life in the other. ln the Forrest River
district of Australia, members of a deceased person's own genera-
tion paint themselves black and white and keep away from the
corpse while those of other generations do not paint themselves
and approach the corpse. The opposition white/red is thus re-
placed, without any change of semantic load, by an opposition
black+ white/O. Insteadof the values given to white and red being
reversed, as in the previous case, the value of white (here associated
with black, a non-:hromatic colour) remains constant, and it is the
content of the opposite pole which is reversed, changing from the
'super-colour' red to the total absence of colour. Finally, another
Australian tribe, the Bard, constructs its symbolism by means of
the opposition black/red. Black is the colour of mourning for even
generations (grandfather, Ego, grandson) and red the colour of
mourning for uneven generations (father, son), that is, those which
are not assimilated with the generation of the deceased (Elkin 4,
pp. 298-9). The opposition between two differently stressed terms
- life and deathamong the Luvale, 'someone else's deathand ''my''
death' in Australia - is thus expressed by pairs of elements taken
from the same symbolic chain; absence of colour, black, white,
black + white, red (as the supreme presence of colour), etc.
The same fundamental opposition is found among the Fox, but
transposed from colour to sound. While the burial ceremony is in
progress 'those burying (the dead) talked to each other. But the
others did not say a thing to each other' (Michelson 1, p. 411). The
opposition between speech and silence, between noise and the
absence of noise, corresponds to that between colour and the
absence of colour or between two chromatisms of different
degrees. These observations seem to make it possible to dispose of
theories making use of the concepts of 'archetypes' or a 'collective
unconscious'. It is only forms and not contents which can be
common. If there are common contents the reason must be sought
either in the objective properties of particular nature or artificial
entities or in diffusion and borrowing, in either case, that is,
outside the mind.
• As in China, where white is the colour of mouming and red the colour of
marriage.
65
THE SAVAGE MIND
Another difficulty is due to the natural complexity of concrete

logics for which the existence of some connection is more essential
than the exact nature of the connections. On the formai plane one
might say they will make use of anything which cornes to hand.
Conscquently, we cannot even postulate the formai nature of a
connection when we are merely given the two connected terms.
Relations between terms, like the terms themselves, must be
approached indirectly and, as it were, off the cushion. The same
difficulty is found in structural linguistics at the present time,
although in a different field, because this also is based on a qualita-
tive logic. It identifies pairs of oppositions made up of phonemes
but the spirit of each opposition remains largely hypothetical; it is
difficult to avoid a certain impressionism in defining them in the
preliminary stages, and several possible solutions to the same
problem remain open for a long time. One of the major difficulties
of structural linguistics, and one which has not yet been full y over-
come, derives from the fact that the reduction which it effects by
means of the concept of binary opposition has to be paid for by the
diversity of kind which is insidiously reconstituted to the benefit of
each opposition. The number of dimensions is reduced on one
plane only to be increased on another. It may howcver be the case
that what we have here is not a difficulty of method but a limitation
inhcrent in the nature of particular intellectual operations, whose
wcakncss as well as whose strength is their capacity to be logical
while remaining firmly rooted in the qualitative.
A final kind of difficulty must be considered separately. It more

particularly concerns the classifications which are called 'totemic'
in the fullcst sense, that is those which are not only thought but
lived. Whenever bocial groups are named, the conceptual system
formed by these names is, as it were, a prey to the whims of
demographic change which follows its own laws but is related to it
only contingently. The system is given, synchronically, while
demographic changes take place diachronically, in other words,
there are two determinisms, each operating on its own account
and indcpcndently of the other.
This conflict betwecn synchrony and diachrony is also found on
the linguistic plane. The structural features of a language will
probably change if the population using it, which was once
numerous, becomes progressively smaller; and it is obvious that a
66
language disappears with the men who speak it. Nevertheless the
connection between synchrony and diachrony is not rigid. For, in
the first place, all the speakers are, taken as a whole, equivalent
(though this would cease to be s~ if one were to consider specific
cases), and secondly, and more important, the structure of the
language is to some extent protected by its practical purpose, which
is to ensure communication. Language is therefore sensitiYe to the
influence of demographic ernlution only up to a point and in so
far as its fonction is not impaired. But the conceptual systems
we are studying here are not, or are not primarily, means of
communication. They are means of thinking, an activity which is
governed by Yery much less stringent conditions. One either
succeeds or not in making oneself understood, but one can think
more or less well. Thinking admits of degrecs and a way of thinking
can degenerate imperceptibly into a way of remembering. This
explains why the synchronie structures of so-called totemic sys-
tems should be so extremely Yulnerable to the effects of diachrony:
a mnemotechnic procedure is less trouble to operate than a
speculati\·e one which in turn is less exacting than a device for
communication.
This point can be illustrated by an only slightly imaginary
example. Suppose that a tribe was once divided into three clans,
each of which had the name of an animal symbolizing a natural
element:
1 1 1
bear eagle turtle
(land) (sky) (water)
Suppose further that demographic changes led to the extinction of
the bear clan and an increase in the population of the turtle clan,
and that as a result the turtle clan split into two sub-clans, each of
which subsequently gained the status of clans. The old structure
will disappear completely and be replaced by a structure of this
type:
1
eagle
1 1
yellow turtle grey turtle
Without further data it would be useless to tty to detect the

original scheme behind this new structure; it is even possible that
THE SAVAGF. MI:->D
any such scheme, whether perceived consciously or not, should

have completely disappeared from native thought, and that after
this upheaval the three clan names might survive only as tradition-
ally accepted titles with no cosmological significance. This outcome
is probably very common and explains why an underlying system
can sometimes be posited in theory even though it would be
impossible to reconstruct in practice. But it is also frequently the
case that things turn out otherwise.
On a first supposition the initial system will be able to survive
in the mutilated form of a binary opposition between sky and
water. Another solution could be derived from the fact that there
were originally three terms and that the number of the terms is
still the same at the end. The original three terms expressed an
irreducible trichotomy while the final three terms are the result of
two successive dichotomies; between sky and water and then
between yellow and grey. If this opposition of colour were given a
symbolic sense, for instance with reference to day and night, then
there would be not one but two binary oppositions: sky /water and
day/night, that is a system of four terms.
It can be seen therefore that demographic evolution can shatter
the structure but that if the structural orientation survives the
shock it has, after each upheaval, several means of re-establishing a
system, which may not be identical with the earlier one but is at
least formally of the same type. ~or is this ail. W e have so far
considered only one dimension of the system and there are always
in fact several, not ail cf which are equally vulnerable to demog-
raphic changes. Let us return to the beginning of the example. When
our imaginary society was at the stage of the three elements, the
division into three classes did not operate only on the plane of clan
names. The system rested on myths of creation and origin and
permeated the entire ritual. Even if its demographic basis collapses,
this upheaval will not have immediate repercussions on ail planes.
The myths and rites will change, but only with a time-lag and as
if they possessed a certain residual vigour which for a time pre-
served ail or part of their original orientation. This latter will
therefore continue through the myths and rites, so as to maintain
new structural solutions along approximately the same lines as the
previous structure. If, for the sake of argument, we suppose an
initial point at which the set of systems was precisely adjusted, then
this network of systems will react to any change affecting one of
68
its parts like a motor with a feed-back device: governed (in both
senses of the word) by its previous harmony, it will direct the dis-
cordant mechanism towards an .equilibrium which will be at any
rate a compromise between the old state of affairs and the confusion
brought in from outside.
Whether they are historically correct or not, the traditional
legends of the Osage show that native thought itself may well
envisage this sort of interpretation, based on the hypothesis of a
structural adjustment of the historical process. When the ancestors
emerged from the bowds of the earth they were, according to
Osage tradition, divided into two groups, one peace-loving, vege-
tarian and associated with the left side and the other warlike,
carnivorous and associated with the right. The two groups resolved
to ally themseh·es and to exchange their respective foods. In the
course of their wanderings, they met a third group, which was
ferocious and lived entirely on carrion, with whom they eventually
united. Each of the three groups was originally composed of seven
clans, making a total of twenty-one clans. In spite of the symmetry
of this three-clan division, the system was in disequilibrium since
the newcomers belonged to the side of war and there were fourteen
clans on one side and seven on the other. In order to remedy this
defect and to preserve the balance between the side of war and the
side of peace, the number of clans in one of the groups of warriors
was reduced to five and that in the other to two. Since then the
Osage camps, which are circular in shape with the entrance facing
east, consist of seven clans of peace occupying the northern half on
the left of the entrance and seven clans of war occupying the
southern half on the right of the entrance (J.O. Dorsey /, 2 ). The
legend suggests twin processes. One is purely structural, passing
from a dual to a three-fold system and then returning to the earlier
dualism; the other, both structural and historical at the same time,
consists in undoing the effects of an overthrow of the primitive
structure::, resulting from historical events, or events thought of as
such: migrations, war, alliance. Now, the social organization of the
Osage, as it was to be seen in the nineteenth century, in fact
integrated both aspects. Although they were each composed of the
same number of clans, the side of peace and the side of war were
in disequilibrium since one was simply 'sky' while the other, also
referred to as 'of the earth', consisted of two groups of clans
associated respectively with dry land and water. The system was
THE SAVAGE MIND
thus a t once historical and structural, composed of two and of

three classes, symmetrical and asymmetrical, stable and top
heavy ...
Nowadays the reaction to a difficulty of the same kind is quite
different. Consider for instance this agreement to diff er with which
a recent conference ended:
lVI. Bertrand de JOL'VENEL - 1\1. Priourct, would you like to say a few
words in conclusion?
1\1. Roger PRIOL'RET - It sccms to me that we arc faced with nvo
diametrically opposed thcorics. Raymond Aron follows the vicws of André
Siegfried, according to which there are two basic political attitudes in
France. This country is somctimcs Bonapartist and sometimes Orleanist.
Bonapartism consists in the acceptance and even desire for persona!
power, Orleanism in leaving the administration of public affairs to
rcpresentatives. ln the face of crises like the defcat of 1871 or a protracted
war like the one in Algiers, France changes in attitude, that is, turns from
Bonapartism to Orleanism as in 1871 or from Orleanism to Bonapartism,
as on lJ May 1958.
ln my own vicw, on the other hand, the actual change, although not
entirely indcpcndcnt of thcsc constants in French political temperament,
is connectcd with the uphcavals which industrialization brings into
socicty. A <liffercnt political analogy occurs to me. The coup <l'etat of
2 Deccmber l 85 l corresponds to the first industrial revoltttion and
the coup d'ctat of lJ '.\1ay 1958 to the second. In other words, history
shO'ws that upheaval in the conditions of production and consumption
seems incompatible with parliamentary government and lea<ls this
country to the fonn of authoritarian power which suits its temperament,
namcly, persona! power' (Scdcis, p. 20).
The Osage would probably have used these two types of opposi-
tion, one synchronie and the other diachronie, as a point of
departure. Instead of expecting to be able to choose between them
they would have accepted both on the same footing and would
have tried to work out a single scheme which allowed them to
combine the standpoint of structure with that of event.
Considerations of the same kind could undoubtedly provide an
intellectually satisfying explanation of the curious mixture of
divergencies and parallcls characteristic of the social structure of
the five Iroquois nations and, on an historically and geographically
larger scale, of the similarities and differences exhibited by the
Algonkin of the east of the United States. In societies with unilin-
ear exogamous clans, the system of clan names is almost always
mid-way between order and disorder. And this, it seems. can onlv
be explained by the combined action of two forces, one of demo-

graphic origin which pushes it towards disorganization and the
other of speculative inspiration wfiich pushes it tO\vards a reorgani-
zation as closely as possible in line with the earlier state of
affairs.
The Pueblo Indians provide a good example of this pheno-
menon. Their villages present so many sociological variations on a
theme which looks as if it may be common to ail of them. In
collecting data on the Hopi, Zuni, Keres and Tanoan, Kroeber at
one time belieœd he could show that 'a single, precise scheme
rervades the clan organization of ail the Pueblos', even though each
village affords only a partial and dist0rted picture. This scheme
was supposed to consist of a structure with twelve pairs of clans:
rattlesnake - panther; deer - antclope; squash - crane; cloud -
corn; lizard - earth; rabbit - tobacco; Tansey-mustard - Chaparral
cock; Katchina (raven - macaw; pine - cotton wood); firewood -
coyote; a group of four clans (arrow - sun; eagle - turkey); badger
- bear; turquoise - shell or coral (Kroeber I, pp. 137-40).
This ingenious attempt to reconstruct a 'master plan' was
criticized by Eggan on the basis of fuller and less ambiguous
material than Kroeber had at his disposai in 1915-16when he made
his study. Another damaging argument could howe\·er also be
brought forward, namcly that it does not seem possible that a
master plan should ha\·e survived the different demographic
changes in each village. Using Kroeber's own material, let us
compare the distribution of clans at Zuni (one thousand, six
h undred and fif ty inhabitants in 191 5 and in two Hopi villages of
the first mesa, the number of whose inhabitants Kroeber multi-
plied by five (result: 1,610) to facilitate the comparison: see p. 72.
If one traces a curve showing the distribution of Zuni clans in
declining order of population and superimposes a curve of the
distribution of Hopi clans of the first mesa, it is clear that the
demographic de\·clopment is different in the two cases and that the
comparison does not in principle allow the reconstruction of a
common plan (figure 3, p. 72).
In these circumstances and even if we admit that Kroeber's
reconstruction may have clone violence to the facts at some points,
it is none the less remarkable that so man y common elements and
systematic relations continue to exist in different local organiza-
tions. This suggests a rigour, tenacity and fidelity to distinctions
71
THE SAVAGE MIND
ZUNI HOPI
(Walpi and Sichurnovi)
Sun, Eagle, Turkey 520 90

Dogwood 430 55
Corn, Frog 195 225
Badger, Bear 195 160
Crane 100 nil
Coyote 75 80
Tansey-Mustard, Chaparral Cock 60 255
Tobacco 45 185
Deer, Antelope 20 295
Rattlesnake nil 120
Lizard, Earth nil 145
(Of unknown affiliation) IO nil
-----~ -- - --- -
Total 1,650 1,610
(Kroeber, /, p. 149)
500 Zuni
450 Hopi (lirst mesa)
400
450
300
,/\
\
250 1\ I \
1' \ I \
, \ I \
\ I \
c: 200 I ' I \
"' V \
,,
I
ü \
t;; \
,,
a. 150
V>
c:
0
"....
\ -~
\
\
V>
t;; 100
a. .. \
\
\
ë \
t;; 50
.0
\
E \
:::J \
z 0
2 3 5 6 7 8 9 10 12
Clan "
Figure 3. Distribution of the population at Zuni and among the Hopi of
the first mesa according to clan.
72
and differences on the theoretical plane of which a botanist has

collected equally convincing proofs on the practical one:
1n l\:Iexico 1 worked almost exclüsively with farmers of European or
partly European ancestry. Even those who had strikingly lndian features
were mostly Spanish-speaking and did not consider themselves lndians.
ln Guatemala 1 ·worked with such people but also with lndians who had
retained their old languages and their own cultures. 1 found, to my sur-
prise, that their cornfields had been more rigidly selected for type than
those of their Latin-speaking neighbours. Their fields were quite as true
to type as had been prize-winning American cornfields in the great corn-
show era when the American farmer was paying exquisite attention to
such fancy show points as unifornùty. This fact was amazing, considering
the great variability of Guatemalan maize as a whole, and the fact that
corn crosses so easily. A little pollen blown from one field to another will
introduce mongrel germ plasm. Only the most finicky selection of seed
ears and the pulling out of plants which are off type could keep a variety
pure under such conditions. Yet for l.Ylexico and Guatemala and our own
South-west the evidence is clear: wherever the old lndian cultures have
survived most completely the corn is least variable within the variety.
::\luch later 1 grew a collection of corn made among an even more
primitive people, the :'\aga of Assam, whom some ethnologists describe
as still living in the Stone Age in so far as their daily life is concerned.
Each tribe had several different varieties which were sharply different
from one another, yet within the variety there were almost no differences
from plant to plant. Furthermore, some of the most distinctive of these
varieties were grown not only by different families but by different tribes,
in different areas. Only a fanatical adherence to an ideal type could have
kept these varieties so pure when they were being traded from family to
family and from tribe to tribe. lt is apparently not true, as has so fre-
quently been stated, that the most primitive people have the most variable
varieties. Quite the opposite. lt is rather those natives most frequently
seen by travellers, the ones who live along modern highways and near big
cities, the ones whose ancient cultures have most' completely broken
down, who have given rise to the impression that primitive peoples are
careless plant breeders (Anderson, E., pp. I 86-7).
This is a striking example of the concern with diff erentiating

features which pervades the practical as well as the theoretical
activities of the people we call primitive. Its formai nature and the
'hold' it has over every kind of content explain how it is that
native institutions, though borne along on the flux of time, manage
to steer a course between the contingencies of history and the
immutability of design and remain, as it were, within the stream
of intelligibility. They are always at a safe distance from the Scylla
and CharybG.is of diachrony and synchrony, event and structure,
73
THE SAVAGE MIND
the aesthetic and the logical, and those who have tried to define
them in terms of only one or the other aspect have therefore
necessarily failed to understand their nature. Between the basic
absurdity Frazer attributed to primitive practices and beliefs and
the specious validation of them in terms of a supposed common
sense invoked by Malinowski, there is scope for a whole science
and a whole philosophy.
74
CHAPTER THREE
SYSTEMS OF TRANSFORMATIONS
As we have just seen, the practico-theoretical logics governing the

life and thought of so-called primitive societies are shaped by the
insistence on differentiation. The latter is already evident in the
myths underlying totemic institutions (Lévi-Strauss 6, pp. z7-8
and 36-7) and it is also found on the plane of technical activity,
which aspires to results bearing the hallmark of permanence and
discontinuity . .!\ ow, on the theoretical as well as the practical plane,
the existence of differentiating features is of much greater impor-
tance than their content. Once in evidence, they form a system
\vhich can be employed as a grid is used to decipher a text, whose
original unintelligibility givcs it the appcarancc of an uninterrupted
flow. The grid makes it possible to introducc divisions and con-
trasts, in other words the formai conditions necessary for a
significant message to be conveyed. The imaginary example
di~cussed in the last chapter showed how any system of differen-
tiating features, provided that it is a system, permits the organiza-
tion of a sociological field which historical and demographic
evolution are transforming and which is hence composed of a
theoretically unlimited series of different contents.
The logical principle is always to be able to oppose terms which
previous impoverishment of the empirical totality, provided it has
been impoverished allows one to conceive as distinct. How to
oppose is an important but secondary consideration in relation to
this first requirement. In other words, the operative value of the
systems of naming and classifying commonly called totemic derives
from their formai character: they are codes suitable for conveying
messages which can be transposed into other codes, and for
75
THE SAVAGE .MIND
expressing messages rcceived by means of different codes in terms

of their own system. The mistake of classical ethnologists was to
try to reify this form and to tie it to a detcrminate content when in
fact what it providcs is a method for assimilating any kind of
content. Far from bcing an autonomous institution definable by
its intrinsic characteristic, totemism, or ~hat is referred to as such,
corresponds to certain modalities arbitrarily isolated from a formal
system, the fonction of which is to guarantee the convertibility of
ideas between different levels of social reality. As Durkheim seems
sometimes to have realized, the basis of sociology is what may be
called 'socio-logic' (Lévi-Strauss 4, p. 36; p. 137).
ln the second volume of Totemism and Exogamy, Frazer was
particularly concerned with the simple forms of totemic beliefs
observed in Melanesia by Codrington and by Rivers. He believed
that in them he had discovered the primitive forms which were the
origin of the conceptual system of totemism in Australia, whence,
in his view, all other types of totemism derived. ln the New
Hebrides (Aurora) and in the Banks' Islands (Mota) certain peof)le
think that their life is associated with that of a plant, animal or
inanimate object called atai or tamaniu in the Banks' Islands and
nunu at Aurora. The sense of nunu and perhaps also of atai is
roughly that of soul (figure 4).
According to Codrington, a Mota native discovers his tamaniu
through a vision or with the aid of divining techniques. In Aurora, on
the other hand, it is the future mother who believes that a coconut,
a bread-fruit tree or some other object is mysteriously connected
with the child, who will be a sort of echo of it. Rivers found the
same beliefs in Mota where many people observe food prohibitions
because they believe themselves to be an animal or fruit which their
mother found or noticed while she was pregnant. ln such a case,
a woman takes the plant, fruit or animal back to her village and asks
about the significance of the event. It is explained to her that she
will give birth to a child who will resemble, or actually be, the
object. She then returns it to the place where she found it and, if it
was an animal, builds it a shelter out of stones. She visits and f eeds
it every day. When the animal disappears, it is because it has
entered her body, from which it will reappear in the form of a
child.
The child may not eat the plant or animal with which it is
identified under pain of illness or death. If it is an inedible fruit
76
•Salomon Islands
'•' Nukumanu
r::. ·~ Ongtong Java 1~
.yaumako
0 .•N1filole
en•c?s!• Cruz I~
'•Vanikoro
Tikopia•
. ..,0
Banks I~ !~ Motla11 ..,
Ureparapara49 ~Mota :Il
Vanua Lava Merig >
S~• Marial> •,Merlav z
cri
C\J\ &Aurora ..,.,
Espiritu Santo~ ca!-Aoba 0
New Hebrides ('.... .1Wentecost :Il
MalekulauQ Ambrym
\$:Epi
::::
>
Tongoa::emae
0 Efate
...0
-!
Eromangao
z
cri
AUSTRALIA .Aniwa
Tanna Il •Futuna
20·
0 250miles
150• 160'
Fig 4 Partial map of Melanesia (Centre documentaire sur l'Oceanie de !'Ecole Pratique des Hautes Etudes).
THF. SAVAGE MIND
which was found, then the child may not even touch the tree on
which it grows. Ingestion or contact are regarded as a sort of
auto-cannibalism. The relation between the person and the object
is so close that the person possesses the characteristics of the object
with which he is identified. If, for example, it was an eel or sea-
snake which was found, the child will, like these, be weak and
indolent, if a hermit crab, it will be hot-tempered; or again, it will
be gentle and sweet-natured like the lizard, thoughtless, hasty and
intemperate like the rat or, if it was a wild apple which was found,
it will have a big belly the shape of an apple. These identifications
are also to be found at Motlav (the name of a part of Saddle Island)
(Rivers, p. 462). The connection between an individual on the one
hand and a plant, animal or object on the other is not general; it
only affects some people. It is not hereditary and it does not in volve
exogamous prohibitions between the men and women who happen
to be associated with creatures of the same species (Frazer vol. II,
pp. 81-3, pp. 89-91 (quoting Rivers), and vol. IV, pp. 286-7).
Frazer regards these beliefs as the origin and explanation of those
found at Lifu in the Loyalty Islands and at Ulawa and Malaita in
the Solomon Islands. At Lifu a man before he dies sometimes
indicates the animal (or bird or butterfly) in whose form he will be
reincarnated. All his descendants are then forbidden to eat or kill
this animal. 'It is our ancestor', they say, and offerings are made to
it. Similarly Codrington observed that in the Solomon Islands
(Ulawa) the inhabitants refused to plant banana trees or eat
bananas because an important person, so he could be reincarnated
in them, had once forbidden it before his death.* In Central
Mclanesia the origin of food taboos must therefore be sought in the
fanciful imaginations of particular ancestors. Frazer believed that
they were the indirect result and distant repercussions of the
cravings and sickly imaginings common among pregnant women.
He held that this psychological trait, which he elevated to the status
of a natural and universal phenomenon, was the ultimate origin of
all totemic bcliefs and practices (Frazer, vol. II, pp. 106-7 et
passim).
The fact that the women of his period and circle of society
• This fact is confirmcd by Ivens, pp. 269-70, who puts forward a somewhat
different interprctation. However, he cites other prohibitions originating in
reincarnation of an ancestor. Cf. pp. 272, 468 and passim. Cf. also C.E. Fox for
heliefs of the same type at San Cri,;to,·al.
experienced cravings when they were pregnant and that the savage
Australian and :\Ielanesian women also did so was enough to
convince Frazer of the universality and natural origin of the
phenomenon. He would have had otherwise to dissociate it from
nature and attribute it to culture, thus admitting that there could
in some way be direct, and so alarming, resemblances between late
nineteenth century European societies and those of the cannibals.
~ow, apart from the fact that there is no evidence that pregnant
women the world over have cravings, their incidence has dimin-
ished considerably in Europe in the last fifty years and they may
even have disappeared altogether in some sections of society. They
certainly occurred in Australia and :.Vlelanesia but in the form of an
institutional means of defining in advance certain aspects of the
status of persons or groups. And in Europe itself, it is unlikcly that
the cravings of pregnant women will survive the disappearance of
a similar type of belief which fosters them - on the pretext of
referring to them - in order to diagnose (not predict) certain
physical or psychological peculiarities noticed after (not before) a
child's birth. Even if it were the case that the cravings of pregnant
women had a natural basis, this latter could not account for beliefs
and practices which, as \Ve have seen, arc far from being general
and which can take different forms in diffcrcnt societies.
Further, it is not clear what made Frazer give the 'sick fancies' of
pregnant women priority over those of old men at death's door,
except perhaps the fact that people must be born before they can
die. But by this reasoning ail social institutions should have corne
into existence in the course of a single generation. Finally, had the
system of Ulawa, l\Ialaita and Lifu been derived from that of
:\Iotlav, Mota and Aurora, the remains or survivais of the latter
should be found in the former. What is striking, however, is thatthe
two systems are exact counterparts of each other. There is nothing
to suggest that one is chronologically prior to the other. Their
relation is not that of an original to a derivative form. I t is rather
that between forms symmetrically the reverse of each other, as if
system represented a transformation of the same group.
Instead of trying to discern the priorities, let us think in terms of
groups and attempt to define their properties. W e can summarize
these properties as a triple opposition: between birth and death on
the one hand, and between the individual or collective nature
either of a diagnosis or of a prohibition on the other. It is worth
79
THE SAVAGE MIND
obscrving that the prohibition follows from a prognostication:

anyone eating the forbidden fruit or animal will die.
On the :\fotlav-Mota-Aurora system, the relevant term in the
first opposition is birth. In the Lifu-Ulawa-Malaita system it is
death. All the terms in the other oppositions are similarly reversed.
In the case where birth is the relevant event, the diagnosis is made
collectively and the prohibition (or prognostication) applies to
individuals; a woman who is pregnant or about to become so and
who finds an animal or fruit on the ground or in her loin cloth,
returns to the village where she questions her friends and relatives.
The social group collectively (or through the mouthpiece of
q ualified representatives) gives a diagnosis of the distinctive status
of the person about to be born, who will be subject to an individual
prohibition.
The whole system is reversed at Lifu, Uwala and Malaita. The
relevant event is death, and accordingly the diagnosis is individual
since it is made by the dying man and the prohibition is collective,
binding as it does ail the descendants of the same ancestor and
sometimes, as at Ulawa, the whole population.
The two systems are therefore in the position of exact opposites
within the same group. This can be seen in the following table
where + and - correspond to the first and second member of
each pair of opposites respectively:
Motlav- Lifu-
Significant oppositions: l\1ota-Aurora Ulawa-:\1alaita
Birth/death +
diagnosis: +
1ndividual/ collective
1 prohibition : +
These facts make it possible to extract a feature which is common
to each group and serves to distinguish it as a group from all others
of the same set, that is, the set of systems of classification which
postulate a homology between natural distinctions and cultural dis-
tinctions (an expression to be preferred to 'totemic institutions').
The common feature of the two systems just discussed is their
statistical and non-universal nature. Neither applies indiscrimin-
ately to all the mcmbers of the society: an animal or plant plays a
part in the conception of only some children; only some of the
dying are reincarnated in a natural species. The domain governed
by each system therefore consists of a sample, the selection of which
80
is, at least in theory, left to chance. For this double reason, these
systems belong immediately beside the Australian systems of the
Aranda type, as Frazer saw althm,igh he misunderstood the nature
of the relationship - logical and not genetic - which unîtes them
while at the same time recognizing the distinctiveness of each. The
Aranda systems are also of a statistical nature but their rule of
application is universal since the domain they govern is coextensive
with the whole society.
ln the course of their passage across Australia, Spencer and Gillen

were already struck by the fact that the institutions of the people
lying on a south-north axis from the Great Australian Bight to the
Gulf of Carpentaria looked as if they formed a coherent system.
In the Arunta and the \Varramunga [socio-religious] conditions are
exactly revcrsed, but, as in othcr matters, sn we find an Intcrrnediate
state in the Kaitish (Spencer and Gillen, p. 164).
The Arabanna in the sou th recognize two exogamous moieties and

a number of exogamous totemic clans, all matrilineal. Marriage,
shown by Spencer and Gillen to be preferential with the daughter
of the mother's clder brothcr or the father's eider sister, bclongs,
according to Elkin, to the same type as the Aranda, but complicated
by totemic restrictions, which, as we know, the Aranda do not
have.
ln mythical times (ularaka) the totemic ancestors place .. "--
spirit-children (maiaurli) in the totemic places. The Aranda have
an equivalent belief. But~ whereas among the Aranda the spirits
regularly return to their place of origin to await a new incarnation,
Arabanna spirits change their sex, moiety and totem at each
successive incarnation and every spirit therefore regularly passes
through a complete cycle of biological and socio-religious statuses
(Spencer and Gillen, pp. 146 sq).
If this were a description of the course of events in the real
world, the picture presented would be of a system symmetrically
the reverse of that of the Aranda. Among the Aranda descent is
patrilineal and not matrilineal. Totemic affiliation is determined
not by a rule of descent but by the place at which a woman happens
to be when she becomes conscious of her pregnancy. In other
words, totems are allocated according to a rule among the Ara-
banna statistically and on the basis of chance among the
Sr
THE SAVAGE MIND
Aranda. Totemic groups are strictly exogamous among the Ara-

banna but play no part in the regulation of marriage among the
Aranda; marriage among the Aranda is regulated by a system of
eight subsections (and no longer just two moieties), unrelated to
totemic affiliations and operating as a cycle which can be repre-
sented in the following way (figure 5). *
Simplifying a great deal and confining ourselves for the moment
to the older sources, it is tempting to say that the course of events
for human beings among the Aranda is the same as the course of
events for spirits among the Arabanna. In each generation spirits
change their sex and moiety. (I am leaving aside the change of
totemic group since totemic affiliation is not relevant in the Aranda
system and I am replacing it by a change of subsection, which is
the relevant f eature.) These two requirements, translated into the
terms of the Aranda system, pro duce the cycle:
Ai
1
Di = az
1
B1 = dz
1
Ci = bz
1
A1 = cz
where capital letters and small letters represent men and women
respectively. This cycle corresponds not to the actual structure of
Aranda society in which exclusively male and exclusively female
cycles are distinguished, but to the procedure, implicit in the terms
of the system, by means of which these bits are, as it were, sewn
together.
Elkin's criticisms of his predecessors' description should how-
ever be taken into account. Elkin believed that Spencer and Gillen
had noticed only one form of totemism among the Aranda (Elkin +,
pp. 138-9) when in fact there were two, such as he had himself
discovered among the Arabanna: one patrilineal and a matter of
cuit and the other matrilineal and social and so also exogamous:
l\lembers of a patrilineal totemic cult-group do perfonn increase cere-
monies, assisted by their 'sisters' sons', and they do ritually band this
cuit-totem ovcr to the latter, and through them to others, to eat, but they
• I am indcbtcd to my collcague G. Th. Guilbaud for this mode of diagram-

matic representation in the form of annulus.
Matnlineal moiety 1
\
\
\ Female cycles
'
1
}' A man Al
(!rom matrilineal moiety 1)
marries
(te::::)~)
-,c2 a woman C2
...
The children are:
Cl= B2
'~{
Boys A2 (continuous line)
Dl=A2 who marry women 01 (broken
line. outer circle)
A2
Social structure and
marriage rules of
the Aranda type
A2! Girls A2 (broken line. inner
circle) married by men
.~ Dl (cont1nuous line)
Fig 5 Social structure and marriage rules of the Aranda type (Laboratoire
de cartographie de !'Ecole Pratique des Hautes Etudes).
THE SA V AGE MIN D
do not henceforth observe a taboo on this totem themselves. 0 n the other

hand ... they do observe a rigid prohibition on eating their madu, social
totem, which however is not the object of a cuit (Elkin 2a, p. 180).
Elkin's objection to Spencer and Gillen's account is therefore that

the supposition that the totemic spirits pass through a complete
cycle is inconsistent because it would involve a mixture of two
forms of totemism which, as he himself holds, cannot be assimi-
lated. The most one could admit is that the cuit totems which are
patrilineal, alternate between the two moieties within any given
male line.
Without claiming to decide, 1 shall restrict myself to referring
the reader to objections of principle which 1 have elsewhere put
forward against the analyses of Elkin with regard to particulars. It
is also worth pointing out that Spencer and Gillen were acquainted
with Arabanna culture when it was still intact while, on his own
admission, Elkin saw it only in an already advanced state of decay.
Even if Elkin's qualifications had to be accepted it would still be
the case that it is the living who pass through a cycle among the
Aranda while among their southern neighbours it is the dead. ln
other words, what looks like a system among the Aranda is dupli-
cated among the Arabanna into a prescription on the one hand and
a theory on the other. For the regulation of marriage by the assess-
ment of totemic incompatibilities described by Elkin is a purely
empirai procedure, while the cycle of spirits obviously depends
only on theoretical considerations. There are also other differences
between the two groups which correspond to genuine inversions
and can be seen on ail planes: matrilineal/patrilineal; two moieties/
eight sub-sections; systematic totemism/statistical totemism; and,
if Spencer and Gillen's analyses are exhaustive: exogamous
totemism/non-exogamous totemism. 1t will also be observed that
the Aranda sub-sections are of great functional yield, since they are
transitive: the children of a marriage X= y are Z, z, that is, of a
different (social) group from their parents. The Arabanna (totemic)
groups (the sociological fonction of which is also to regulate
marriage) have on the other hand a weak functional yield because
they are intransitive: the children of a marriage X = y are Y, y,
reproducing only their mother's group. Transitivity (whole or
partial according to whether Spencer and Gillen's or Elkin's
account is adopted) occurs only in the Arabanna spirit world where
84
a society similar to that of the living among the Aranda is

reproduced.
Finally, the same inversion can,_be seen in the role which terri-
toriality plays in each tribe. The Aranda treat it as having a real
and absolute value. It is the only clearly meaningful content in
their system for from the beginning of time each place has always
belonged to one and only one totemic species. Among the Ara-
banna, due to the spirits' ability to move through a cycle, locality
loses much of its significance and the value accorded to it is
relative and formai. The totemic places are ports of call rather than
ancestral homes.
Let us now compare the social structure of the Aranda with that
of the \Varramunga, a people further north who are also patrilineal.
Among the \Varramunga totems are connected with moieties, that
is, their function is the reverse of the one which they have among
the Aranda and analogous (although in a different way) to that
which they have among the Arabanna, and their geographical
position is also the reverse in relation to the Aranda, being their
northern, while the Arabanna are their sourthern, neighbours.
Like the Arabanna, the \Varramunga have paternal and maternai
totems, but, in contrast to what occurs among the Arabanna, it is
the paternal totems which are prohibited absolutcly. The maternai
totems are permitted through the agency of the opposite moiety.
Among the Arabanna, on the other hand, it is the paternal totems
which are allowed to the opposite moiety through the offices of
the cult groups belonging to the same moiety.
The role given to the opposite moiety indeed lends itsclf to an
analysis by transformation. There is no reciprocity between
moieties in the rites of increase among the Aranda: each cult group
performs its rites as it pleases for the benefit of other groups who
are themselves free to eat the food made more abundant only
through the auspices of the officiating group. Among the Warra-
munga, on the other hand, the moiety which eats the food actively
intervenes to make the other moiety perform the ceremonies from
which it benefits.
This diff erence leads to other related ones. ln one case the rites
of increase are an individual, in the other a group, affair. Among
the Aranda the performance of the rites of increase, which is left
to the initiative of the man whose property they arc, is of a statistical
85
THE SAVAGE MIND
nature: a person officiates when he wishes without co-ordinating

his efforts with thosc of othcrs. Among the \Varramunga on the
other hand there is a calendar of ritual, and ceremonies succeed
each other in a prescribed order. \Ve thus find again, on the ritual
plane, an opposition between pcriodic and non-periodic structures
which we noticed earlier (in the case of the Aranda and Arabanna)
and which appcared to be characteristic of the community of the
living and the community of the dead. The same formai opposition
exists between the Aranda on the one sicle and the Warramunga
and Arabanna on the other but this time it manifests itself on a
different plane. Simplifying a good deal, we could say that in both
these respects the position among the \Varramunga is symmetrical
with that prevailing among the Arabanna, \Vith the difference that
descent is patrilineal in one case and matrilineal in the other. The
Aranda, on the other hand, who are patrilineal like the \Varra-
munga, can be opposed to their northern and southern neighbours
by the statistical nature of their performance of rituals which
contrasts with the pcriodic celebration of rituals among the
latter.*
Furthcrmore, the Arabanna and \Varramunga think of their
totemic ancestors as single individuals who are half-human half-
animal and have an air of completeness. The Aranda on the other
hand favour the idea of a multiplicity of ancestors (for each totemic
group), who are, however, incomplete human beings. As Spencer
and Gillen ha\·e shown, the groups lying between the Aranda and
\Yarramunga, the Kaitish and Unmatjera, provide an intermediate
case in this respect, for their ancestors are represented in ·myths as
a mixture of incomplete human beings and fully fl.edged men. In
general, the distribution of beliefs and customs on a north-south-
axis shows sometimes a graduai change from one extreme type to
its reverse form and sometimes the recurrence of the same forms
at the two poles but in that case expressed in a reverse context:
patrilineal or matrilineal; the structural inversion then occurs in
the centre, that is, among the Aranda:
• Among the Aranda 'there was no set order ... each separate ceremony is the
property of some special individual' but among the \Varramunga 'the cere-
monies connected with a given totem are perfonned in a regular sequence: A,
B, C, D' (Spencer and Gillen, p. 192-3).
86
SOUTH NORTH
Ara banna Aranda Kaitish, \Varramunga

Unmatjera
--------
Totemic complete incomplete incomplete complete
ancestors beings, half- human human beings, half-
human half- beings beings human half-
animal + animal
fully de-
single multiple veloped single
men
multiple
---- -- - ----
Social totems and totems and

organiza- moietics do moieties
tion not coïncide coïncide
exogamous non-exo- exogamous
totemism gamous totcmism
totemism
Ritual reciprocal initiative reciprocity of

exclusivism of totemic moieties:
of moieties group + initiative of
assistance opposite
moiety moiety
Totemic individual collective

cere- property property
momes
--
Perform- non-periodic periodic
an ce
---- -- - - --
\Ve see therefore that going from the Aranda to the \Varramunga,
one passes from a system with a collective mythology (multiplicity
of ancestors) but an individualized ritual to the reverse system of
an individualized mythology but a collective ritual. Similarly the
properties ascribed to the earth are religious (associated with
totems) among the Aranda and social among the \Varramunga
(territory is divided between the moietics). Finally, the churinga
progressively disappears as one goes north. \Vhat has already been
said might almost have led one to expect this, since in the Aranda
context the churinga serves to bring unity into multiplicity: it
represents the physical body of an ancestor and is held by a series
of successive individuals as proof of their genealogical descent, and
THE SAVAGE MIND
i t bears witness to the continuity of the individual over time, the

possibility of which might seem to be excluded by the Aranda
conception of mythical times. •
Ali these transformations could be systematically set out. The
Karadjera among whom a man dreams the totemic affiliation of his
future child, furnish an instance of a case the reverse of the Aranda
where it is the woman who experiences it. The increasingly exacting
nature of totemic prohibitions in North Australia provide a sort of
'culinary' equivalent of the restraints on marriage imposed by a
system with eight exogamous sub-sections. Thus some peoples
forbid a man, conditionally or absolutely, to eat not only his own
totem but also those of his father, his mother, his father's father
(or mother's fath~r). Among the K-auralaig of the islands north of
Cape York Peninsula a persan recognizes as a totem not only his
own but also those of his father's mother, mother's father and
mother's mother; and marriage is forbidden in the four corres-
ponding clans (Sharp, p. 66 ). The eatingprohibitions resulting from
the belief that an ancestor is reincarnated in a species of animal or
plant were discussed above. The same type of structure is found in
the Melville and Bathurst islands but this time on the linguistic
plane. Al! the homophones of the name of the deceased are avoided
by his descendants even if they are terms in current use with only a
remote phonetic resemblance to it.t It is words rather than
bananas which are forbidden. The same ideas appear and disappear
in different societies either identical or transposed from one level
of consumption to another, sometimes applying to the treatment
of women, sometimes to that of foods, sometimes to the words
used in speech.
It is perhaps because Spencer and Gillen's material related to a
comparatively limited number of Australian tribes and was at the
same time remarkably full for each, that they were more acutely
aware than their successors of the systematic relations between the
different types. Later specialists' horizons came to be confined to
the small area which they themselves were studying; and the sheer
quantity of data, together with their own caution, deterred even
those who did not abandon the idea of synthesis from looking for
•Cf. below, p. 238.
t As in various lndian tribes where the prohibition on uttering the names of
parcnts-in-law extends to al! the words of which these names were composed.
Cf. below, p. 176.
88
laws. The greater our knowledge, the more obscure the overall
scheme. The dimensions multiply, and the growth of axes of
reference beyond a certain point- paralyses intuitive methods: it
becomes impossible to visualize a system when its representation
requires a continuum of more than three or four dimensions. But
the day may corne when all the available documentation on Austra-
lian tribes is transferred to punched cards and with the help of a
computer their entire techno-economic, social and religious
structures can be shown to be like a vast group of transformations.
The idea is the more attractive since it is at least possible to
see why Australia should be a particularly favourable field for such
an experiment, more so than any other continent. In spite of the
contact and inter-change with the outside world which has also
taken place in Australia, Australian societies have probably
developed in isolation more than appears to have been the case
elsewhere. :\foreover, this development was not undergone
passively. It was desired and conceptualized, for few civilizations
seem to equal the Australians in their taste for erudition and
speculation and what sometimes looks like intellectual dandyism,
odd as this expression may appear when it is applied to people
with so rudimentary a level of material life. But lest there be any
mistake about it: these shaggy and corpulent savages whose
physical resemblance to adipose bureaucrats or veterans of the
Empire makes their nudity yet more incongruous, these meticulous
adepts in practices which seem to us to display an infantile perver-
sity - manipulation and handling of the genitals, tortures, the
industrious use of their own blood and their own excretions and
secretions (like our own more discreet and unreflecting habit of
moistening postage stamps with saliva)- were, in various respects,
real snobs. They have indeed been referred to as such by a special-
ist, born and brought up among them and speaking their language
(T. G. H. Strehlow, p. 82). When one considers them in this light,
it seems less surprising that as soon as they were taught accom-
plishments of leisure, they prided themsdves on painting the dull
and studied water-colours one might expect of an old maid (Plate 8).
Granting that Australia has been turned in on itself for hundreds
of thousands of years, • that theorizing and discussion was all the
• With the undoubted exception of the northern regions and these were not
without contact with the rest of the continent. This is therefore only an approxi-
mation.
THE SAVAGE MIND
rage in this closed world and the influence of fashion often para-
mount, it is easy to understand the emergence of a sort of common
philosophical and sociological style along with methodically
studied variations on it, even the most minor of which were
pointed out for favourahle or adverse comment. Each group was
no doubt actuated by the only apparently contradictory incentives
of being like others, as good as others, better than others and
different from others, that is, of constantly elaborating themes only
the general outlines of which were fixed by tradition and custom.
In short, in the field of social organization and religious thought,
the Australian communities behaved like the peasant societies of
Europe in their manner of dressing in the la te eighteenth and early
nineteenth centuries. That each community had its own dress and
that this was composed of roughly the same elements for men and
women respectively was never callcd in question. It was in wealth
or ingenuity of detail alone that people tried to distinguish them-
selves from, and to outdo, the neighbouring village. Ali women
wore coifs but they were different in different regions. ln France
marriage rules of an endogamous kind were expressed in terms of
coifs (':\Iarry within the coif') just as Australian rules (of an
exogamous kind) were expressed in terms of sections or totemism.
Here, as elsewhere, among the Australian aborigines as in our own
peasant societies the combination of general conformity (which is
a feature of a closed world) with the particularism of the parish
results in culture being treated like thcmes and variations in music.
It is therefore conceivable that the favourable historical and geo-
graphical conditions outlined have led to Australian cultures stand-
ing in a relation of transformation with each other, possibly more
completely and systematically than those of other regions of the
world. But this external relation must not make us neglect the
same relation, this time internai, which exists, in a very much more
general fashion, between the different levels of a single culture.
As 1 have already suggestcd, ideas and beliefs of the 'totemic' type
particularly merit attention because, for the societies which have
constructed or adopted them, they constitute codes making it
possible to ensure, in the form of conceptual systems, the con-
vertibility of messages appertaining to each level, even of those
which are so remotc from each other that they apparently relate
solely to culture or solely to society, that is, to men's relations with
each other, on the one hand, or, on the other, to phenomena of a
90
technical or economic order which might rather seem to concern

man's relations with nature. This mediation between nature and
culture, which is one of the distinctive fonctions of the totemic
operator, enables us to sift out what may be true from what is
partial and distorted in Durkheim's and Malinowski's accounts.
They each attempted to immure totemism in one or other of these
two domains. ln fact however it is pre-eminently the means (or
hope) of transcending the opposition between them.
This has been brought out very clearly by Lloyd Warner in the
case of the :\Iurngin of Arnhem Land. These North Australians
explain the origin of things by a myth which is also the basis of an
important part of thcir ri tuai. At the beginning of time the Wawilak
sisters set off on foot towards the sea, naming places, animais and
plants as they went. One of them was pregnant and the other
carried her child. Before their dcpa1ture they had both indeed had
incestuous relations with men of their own moiety.
After the birth of the younger sister's child, they continued their
journey and one day stopped ncar a water hole \vhere the great
snake Yurlunggur lived who was the totem of the Dua moiety to
which the sisters belonged. The older sister polluted the water
with menstrual blood. The outraged python came out, caused a
deluge of rain and a general flood and then swallowed the women
and their children. When the snake raised himself the waters
covered the entire earth and its vegctation. \Vhen he lay clown
again the flood receded.
As \Varner explains, the ~forngin consciously associate the snake
with the rainy season \vhich causes the annual inundation. In this
area the various seasonal changes arc so regular that, as a geo-
grapher points out, they can be predicted almost to the day.
Rainfall is often as high as fifty or sixty inches in two or three
months. It increases from just two inches in October to some ten
inches in December and fifteen in January. The dry season cornes
equally rapidly. The graph of the rainfall recorded at Port Darwin
over a period of forty-six years might be a picture of the snake
Yurlunggur raised above his \Vater hole with his head in the sky
and flooding the earth (figure 6, p. 92).
This division of the year into two contrasting seasons, one of
seven months and extremely dry and the other of five months with
heavy rainfall and great tides which flood the coastal plains for
scores of miles, leaves its mark on native thought and activities.
THE SAVAGE MIND
::1 1 1 1
1.4
1 11H 1111
13 1 1 1 1 1 1 t-]-f1. 1
1
1
1
1
1
1
I~
1 1 1
~
1 1
12
[
1
::r_l
1
::rJ_
1
I
1
::r
1
I
1111
:I I 1 :.I:
1 :I
11
10
L
1
1
1
1
1
1
1
1
1
J_
1
1
1
1
1
1
J_
1
1
1
1 J_
II
I
7.1
1
1
1
1
1
1
,
.11 1
1
1
1
1
1
1
1
1
J_ .1
9
1
8 -1:I
7
1
1
1
1
1
1
1
1
1
1
1 J_I
1 I..1
1
1
1
1
t _li11, .1
1
1
1
1
1 1 1 1 1 1 1 1 1
1
1
1 1 1 1 1111 1 1 1 1\ 1
1
1
1
1
1
1
1
1
1
1
1
11t 1
1
1
1
1
1
1 +! 1
1
1
+--1 I~
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1
L J_ J_ J_ J_
±1
1Z
J_ \
1
L
1
J_
1 1
J_
1
J_
1 1
1
1
1
1
1
1
1 "
1\
1
1
~
L J_ J_ / J_ 1 1 1 1
I ::r I ::r
June July Aug. Sept. Oct. Nov. Dec. Jan. Feb. March April May
Fig 6 Rain chart for Port Darwin (forty-six-year record) (from Warner,
Chart XI, p. 380).
The rainy season forces the :\1urngin to disperse and take refuge in
small groups in the areas which have not been submerged. Here
they carry on a precarious existence, threatened by famine and
inundation. A few days after the floods have receded the vegetation
is lush again and animais reappear. Collective life begins once
more and abundance reigns. None ofthis would however have been
possible had the floods not swamped and fertilized the plains.
Just as the seasons and winds are divided between the two
moieties (the rainy season together with the west and northwest
winds are Dua while the dry season and southeast wind are Yiritja),
so the protagonists of the great mythical drama, the snake and the
Wawilak sisters, are associated with the rainy and the dry season
respectively. The former represents the male and initiated element,
the latter the female and uninitiated. They must collaborate if
there is to be life. As the myth explains: had the Wawilak sisters
92
SYSTE:\tS OF TRANSFORMATIONS
not committed incest and polluted the water hole of Yurlunggur

there would have been neither life nor death, neither copulation
nor reproduction on the earth, ançl there would have been no cycle
of seasons.
The mythical system and the modes of representation it employs
serve to establish homologies between natural and social conditions
or, more accurately, it makes it possible to equate significant
contrasts found on different planes: the geographical, meteorologi-
cal, zoological, botanical, technical, economic, social, ritual,
religious and philosophical. The equivalences are roughly the
following:
Pure, sacred: male superior fertilizing had season

(rains)
Impure, profane: female inferior fertilized good season

(land)
One can see at once that there is contradiction embedded in this

table which sets out the canon of native logic. In fact, men are
superior to women, the initiated to the uninitiated and the sacred
to the profane. On the other hand all the terms on the top line are
treated as homologous with the rainy scason, which is the season
of famine, isolation and danger, whilc the tcrms on the lower line
are homologous with the dry season during which abundance
reigns and sacred rites are performed:
The men's age grade is a snake and purifying element, and the socio-
logical women's group is the unclean group. The male snake-group in the
act of swallowing the unclean group 'swallows' the initiates [and so makes
them pass] into the ritually pure masculine age grade, and at the same
time the whole ritual purifies the whole group or tribe.
The snake is the fertilizing principle in nature according to Murngin
symbolism; this explains why it is identified with the men's group rather
than with the women; otherwise one would suppose that the male
principle, being identified with the positive higher social values, would
be associated by the Mumgin with the dry season - the time of the year
of high social value (Warner, p. 387).
The primacy of the intrastructure is th us in a sense confirmed, for
the geography and climate, together with their repercussions on
the biological plane, present native thought with a contradictory
93
THE SAVAGE MIND
state of affairs. There are two seasons just as there are two sexes,
t\vo societies, two degrees of culture (the 'high' - that of the
initiated - and the 'low'; cf. Stanner 1, p. 77 for this distinction).
On the natural plane, however, the good season is subordinate to
the bad, while on the social plane the relation between the corres-
ponding terms is reversed. I t is therefore necessary to decide how
to interpret the contradiction. If the good season is said to be male
on the grounds that it is superior to the bad season and that men
and the initiated are superior to women and the uninitiated (a
category to which women also belong) then not only power and
efficacy but sterility as well would have to be attributed to the
profane and female element. This would be doubly absurd since
social power belongs to men and natural fertility to women. The
other alternative is equally contradictory but its inconsistency can
at least be disguised by the double division of the whole society
into the t\rn classes of men and women (now ritually as well as
naturally differentiated) and the group of men into the two classes
of old and young, initiated and uninitiated, according to tfi.e
principle that the uninitiated stand in the same relation to the inia-
ted in the society of men as women do to men within the society as
a whole. But in consequence men forego embodying the happy sicle
of existence for they cannot both rule and personify it. Irrevocably
committed to the role of gloomy owners of a happiness accessible
only through an intermcdiary, they fashion an image of themselves
on the mode! of their sages and old men; and it is striking that two
types of people, women on the one hand and old men on the other,
constitute, as the mcans to and the masters of happiness respec-
tively, the two poles of Australian society and that to attain full
masculinity young men must temporarily renounce the former and
lastingly subm1t to the latter.
The sexual privileges which old men enjoy, the contrai they
exercise over an esoteric culture and sinister and mysterious
initiation rites are undoubtedly general features of Australian
societies, and examples could also be found in other parts of the
world. 1 do not wish to daim that these phenomena are attributable
to what are obviously local natural conditions. ln order to avoid
misunderstandings and in particular the charge of reviving an old
geographical determinism, this needs to be explained.
The first point is that natural conditions are not just passively
accepted. \Vhat is more they do not exist in their own right for
94
they are a fonction of the techniques and way of life of the people
who define and give them a meaning by developing them in a
particular direction. Nature is n«>t: in itself contradictory. It can
become so only in terms of some specific human activity which
takes part in it; and the characteristics of the environment take on
a different meaning according to the particular historical and
technical form assumed in it by this or that type of activity. On
the other hand, even when raised to that human level which alone
can make them intelligible, man's relations with his natural
environment remain objects of thought: man never perccives them
passively; having reduced them to concepts, he compounds them
in order to arrive at a system which is never determined in
advance: the same situation can always be systematized in various
ways. The mistake of Mannhardt and the Naturalist School was to
think that natural phenomena are what myths seek to explain, when
they are rather the medium through which myths try to explain
facts which are themseh-cs not of a natural but a logical order.
The sense in which infrastructures are primary is this: first, man
is like a player who, as he takes his place at the table, picks up cards
which he has not invented, for the cardgame is a datum of history
and civilization. Second, each deal is the result of a contingent
distribution of the cards, unknown to the players at the time. One
must accept the cards which one is given, but each society, like
each player, makes its intcrprctations in terms of several systems.
These may be common to them ail or individual: rules of the game
or rules of tactics. And we are well aware that different players will
not play the same game with the same hand even though the rules
set limits on the games that can be played with any given one.
To explain the noticeable frequency of certain sociological
solutions, not attributable to particular objective conditions, appeal
must be made to form and not content. The substance of contra-
dictions is much less important than the fact that they exist, and
it would be a remarkable coïncidence if a harmonious synthesis of
the social and natural order were to be achieved at once. Now, the
form contradictions take varies very much less than their empirical
content. The poverty of religious thought can never be over-
estimated. It accounts for the fact that men have so often had
recourse to the same means for solving problems whose concrete
elements may be very different but which share the feature of ail
belonging to 'structures of contradiction'.
95
THE SAVAGE MIND
To return to the Murngin: one can see clearly how the system
of totemic symbols permits the unification of heterogeneous
semantic fields, at the cost of contradictions which it is the fonction
of ritual to surmount by 'acting' them: the rainy season literally
engulfs the dry season as men 'possess' women, and the initiated
'swallow up' the uninitiated, as famine destroys plenty, etc. The
example of the Murngin is not however unique. There are sugges-
tive indications of a 'coding' of a natural situation in totemic terms
in other parts of the world also. A specialist on the Ojibwa, reflect-
ing on the symbolization of thunder as a bird which is so common
in North America, makes the following remark:
According to meteorological observations, the average number of days
with thunder begins with one in April, increases to a total of five in mid-
summer (July) and then declines to one in October. And if a bird calendar
is consulted, the facts show that species wintering in the South begin to
appear in April and disappear for the most part not later than October ...
The avian character of the Thunder Birds can be rationalized to some
degree with reference to natural facts and their observation (Hallowell, p.
32).
One must refer to meteorological data, as Warner did in the case
of Australia, to give a correct account of the personification of
natural phenomena, common in the Hawaiian pantheon. The gods
Kane-hekili (Male-in-the-form-of-gentle-rain), Ka-poha-Ka-a
(Male [= sky]-the-rock-roller) who is the same as Ka'uila-
nuimakeha (Male- [ = sky]-lightning-flash-great-streaking), etc.,
cannot be distinguished and accurately placed without certain
relevant information:
The downpours, which corne in late January, February and March
predominantly and with greatest frequency and violence, develop in the
following sequences of meteorological phenomena; lowering dark cumu-
lus over sea and uplands, with atmospheric stillness inducing an increasing
sense of ominous oppression; 'dry' thunder, sharp and threatening if near,
like distant cannon if far away: followed very soon by slow, gentle pre-
cipitation, which increases rapidly to a downpour; with continuing heavy
thunder, then, in the cloud and rain shrouded uplands, resounding and
thudding, slowly passing along the ridges or round the mountain's fiank
and often then out to sea, where it resounds in dull thuds, and may return
in a direction opposite to that along the ridge, a phenomenon produced
by the miniature cyclonic action of winds, and by convection (Handy and
Pukui, part IV, p. 125 n).
If totemic representations amount to a code which makes it possible

to pass from one system to another regardless of whether it is
formulated in natural or cultural terms, then it may perhaps be

asked why it is that they are accompanied by rules of conduct. At
first sight at least totemism, or what is claimed as such, is some-
thing more than a mere language. It does not just set up rules of
compatibility and incompatibility between signs. It is the basis of
an ethic which prescribes or prohibits modes of behaviour. Or at
least this consequence seems to follow from the very common
association of totemic modes of representation with eating
prohibitions on the one hand and rules of exogamy on the other.
The first answer is that this supposed association is the result of
a petitio principi. If totemism is defined as the joint presence of
animal and plant names, prohibitions apply to the corresponding
species and the forbidding of marriage between people of the same
name and subject to the same prohibition, then clearly a problem
arises about the connection of these customs. It has however long
been known that any one of these features can be found without the
others and any two of them without the third.
That this is so is particularly clear in the case of eating prohibi-
tions, which forma vast and complex set of which the prohibitions
referred to as totemic (that is those resulting from a group's
affinity with a natural species or class of phenomena or objects) are
only a particular case. The Ndembu sorcerer, who is primarily a
diviner, must not eat the flesh of bush-buck because of its irregu·
larly spotted hide, for if he did so his divination would stray from
the main point. There is a prohibition on the zcbra for the same
reason, on animais with dark coats (which would cast a shadow
over his clairvoyance), on a species of fish with sharp bones because
they might prick the diviner's organ of divination, the liver, and
on several sorts of spinach with slippery leaves because they might
cause his powers of divination to slip away from him (V. W.
Turner 2, pp. 47-8).
A Luvale boy ma y urinate only against the following trees during
the time of his initiation: Pseudo achnostylus deckendti, Hymeno-
cardia mollis, Afrormosia angolensis, Vangueriopsis lancifiora,
Swartzia madagascarinensis. These are hard woods symbolizing
the penis in erection, their fruits being associated with fertility and
life. He is also forbidden to eat the flesh of various animais: Tilapia
melanopleura, a fish with a red belly, the colour of blood; Sarco-
daces sp. and Hydrocyon sp. which have sharp teeth symbolizing
the painful after-effects of circumcision; Cl arias sp. whose slimy
97
THE SAVAGE MIND
skin symbolizes difficulty in the healing of the scar; the genet whose
spotted skin symbolizes leprosy; the hare with his sharp teeth and
'hot' chillies which symbolize painful healing, etc. Female
initiates are subject to analogous prohibitions (C. M. N. White, 1,
2).
These prohibitions have been mentioned because they are
specialized, well-defined and rationalized. with precision. They can
readily be distinguished from totemic prohibitions within the
gcneral category of eating prohibitions and contrasted with them.
But Tessman's list of the very large number of prohibitions of the
Fang of the Gabon includes examples of intermediate as well as
extreme types - which explains why the question of whether the
Fang are totemic has been so hotly contested even by those
believing in the concept of totemism.
The prohibitions which the Fang call by the gcneral term beki
may apply to men or to women, to the initiated to the uninitiated,
to adolescents or adults, to households in which a child is expected
or to those in which it is not. The reasons given for them are m6re-
over of very different kinds. The inside of elephants' tusks may not
be eaten because it is a soft and bitter substance. The trunk of the
elephant may not be eaten for fear of softening limbs, sheep and
goats lest they communicate their panting. Squirrels are forbidden
to pregnant women because they make childbirth difficult (cf.
above [p. 6r]). Mice are specially forbidden to girls because they
shamelessly steal manioc while it is being washed and the girls
would run the risk of being 'stolen' in the same way. Mice are
however also more generally forbidden because they live near
homesteads and are regarded as members of the family. Sorne
birds are avoidecl for such reasons as their ugly cry or their physical
appearance. Children may not eat the larvae of dragonfly because
it might make them unable to hold thcir urine.
The idea of a dietetic experiment envisaged by Tessman has
recently been taken up by Fischer with respect to the natives of
Ponapy who believe that the violation of food taboos results in
physiological disorders very like allergies. Fischer shows that
allergie disorders often have a psychosomatic origin even among
ourselves: in man y patients they result from the violation of a
taboo of a psychological or moral kind. This apparently natural
symptom derives, then, from a cultural diagnosis.
Among the Fang, of whose prohibitions I have mentioned only
a few taken at random from Tessman's imposing list, it is rather a

matter of religious analogies: horned animais associated with the
moon; chimpanzee, pig, python, ~te. because of their symbolic role
in particular cuits. That the prohibitions result not from intrinsic
properties of the species to which they apply but from the place
they are gi\'en in one or more systems of significance is clearly
shown by the fact that the guinea fowl is forbidden to initiates of
the female cuit Nkang while in male cuits the reverse rule holds:
the cuit animal is permitted to initiates but forbidden to novices
(Tessman vol. 2, pp. 184--<)3).
Food prohibitions may therefore be organized into systems
which are extra- or para-totemic. Conversdy, many of the systems
traditionally regarded as totemic include prohibitions which do not
concern food. The only prohibition on food recorded among the
Bororo of Brazil attaches to the meat of deer, a non-totemic species;
but the animais or plants after which clans and sub-clans are
named appear not to be objects of particular prohibitions. The
privileges and prohibitions associated with clans are found on a
different plane: that of techniques, raw materials and ornaments.
Each clan distinguishes itself from others, especially at festivals,
by the use of certain f eathers, mother-of-pearl and other substances
whose form and fashioning as well as nature are strictly laid clown
for each clan (Lévi-Strauss 2, ch. 19).
The northern Tlingit who inhabit the coast of Alaska also have
jealously guarded clan crests and emblems. But the animais
depicted or e\•oked are only prohibited in a mock fashion: the
Wolf clan may not rear wolves nor the Crow clan crows and mem-
bers of the Frog clan are said to be afraid of frogs (McClellan).
The central Algonkin have no food prohibitions applying to
their clans' eponymous animais. Clans are primarily differentiated
by the way they paint their bodies, their special clothes and the
custom of a special ceremonial food for each. Clan prohibitions are
never or hardly ever eating ones among the Fox and they are of
the most diverse kinds. The Thunder clan is not allowed to draw
on the west sicle of tree trunks nor to take their clothes off to wash.
The Fish clan is forbidden to build dams and the Bear clan to
climb trees. The Buffalo clan may not skin any cloven-footed
animal nor look at these animais while they are dying. Members of
the Wolf clan must not bury anyone belonging to their clan and
must not strike dogs of any kind. The Bird clan must not do any
99
THE SAVAGE MIND
harm to birds. The members of the Eagle clan are forbidden to put
feathers on their heads. The members of the 'Chief' clan should
never say anything against a human being, those of the Beaver clan
should not swim across rivers and those of the White Wolf should
not shout (:Vlichelson 2, pp. 64-5).
Even in the areas where there is the best evidence of eating
prohibitions it is striking that they are rarely distributed evenly.
A dozen neighbouring cultures (consisting of a hundred tribes) have
been described and analysed in the well-circumscribed region of
the Cape York Peninsula. They ail have one or more forms of
totemism, of moiety, section, clan or cuit group, but only some of
them have eating prohibitions. Clan totemism involves prohibitions
among the patrilineal Kauralaig. The Yathaikeno, on the contrary,
who are also patrilineal, only forbid initiation totems which are
transmitted in the maternai line. The Koko Yao have moiety
clan totems which are patrilineally transmitted and permitted, and
initiation totems which are matrilineally transmitted and forbidden
initiation totems which are matrilineally transmitted and for-
bidden. The Tjongandji have only patrilineal clan totems which
are not subject to any prohibition. The Okerkila are divided into
two groups, east and west, one of which has prohibitions while the
other does not. The Maithakudi abstain from eating the clan
totems, which in their case are matrilineal. The Laierdila have
the same rule although they are patrilineal (Sharp) (figure 7).
As Sharp says:
Eating and killing tabus on edible group totems are invariably associ-
ated with maternai cuit and matrilineal social totemism. Tabus are more
variable in connection with patrilineal cult totemism, being present more
commonly for moiety than clan totems (Sharp, p. 70).
The findings within a particular region confirm the general relation
between eating prohibitions and matrilineal institutions which
Elkin held to exist throughout the continent. As social institutions
are the work of men - as a general rule and especially in Australia -
what this amounts to is that there is a connection between the male
and the consumer and between the f emale and the thing consumed.
1 shall return to this later.
Finally there are cases where the idea of eating prohibitions is
as it were turned inside out. The prohibition is turned into an
obligation and applies not to Ego himself but to someone else; and
it no longer relates to the totemic animal thought of as food but to
100
,,
I
VI .
1
' ..,
''
' ...\
, I
1
1
-.
VII
' \
'\.
\
\
'\
I
,,
IX I Vil 1
1
'
1
l
'\.
'
',
', ,'
/
,
l
,' --
..... ....
.. ,
, ... ____ ___ .,
' , '..,~,' ; ' Vlll '
, ,
,,
,,
Fig 7 Types of totemic organization in the Cape York Peninsula (following
Sharp). 1. Kauralaig Type; Il. Yaithaikeno Type; Ill. Koko Yao Type;
IV. Tjongandji Type; V. Yir Yoront Type; VI. Oikol Type; VII.
Okerkila Type; VIII. Maithakudi Type; IX. Laierdila Type.
IOI
THE SAVAGE :>'IlND
the food of this food. This remarkable transformation has been

found among some Chippewa lndian groups where the totem may
be killed and eaten but not insulted. If a native laughs at or insults
the eponymous animal of some other native, the latter informs his
clan. The clan prepares a feast preferably consisting of the food of
the totemic animal, for instance berries and wild nuts if the animal
in question is a bear. The insulter is formally invited and made to
gorge himself until he is, as the informants say, 'ready to burst' and
is prepared to recognize the power of the totem (Ritzenthaler).
Two conclusions can be drawn from these facts. In the first place
the f act that some species are forbidden and others permitted is not
attributable to the belief that the former have some intrinsic
physical or mystic property which makes them harmful but to the
concern to introduce a distinction between 'stressed' and 'unstres-
sed' species (in the sense linguists give to these terms). Prohibiting
some species is just one of several ways of singling them out as
significant and the practical rules in question can thus be thought
of as operators employed by a logic which, being of a qualitative
kind, can work in terms of modes of behaviour as well as of images.
From this point of view some of the older reports may appear more
worthy of attention than has generally been supposed. The social
organization of the \Vakelbura of Queensland, in east Australia,
was said to consist of four classes which were strictly exogamous
but, as one might put it, 'endo-culinary'. Durkheim already had
doubts about this last f eature and Elkin points out that the evidence
for it depends on only a single, not very reliable, source. Elkin also
however mentions that a similar situation is depicted in Aranda
mythology: the totemic ancestors lived entirely on their own
particular food, while today the opposite is truc: each totemic
group lives on the other totems and forbids its own.
This comment of Elkin's is important because it shows that the
suggested organization of the \Vakelbura can be transformed into
Aranda institutions simply by inverting ail its terms. Among the
Aranda totems are not relevant to marriage but they are relevant
to eating: totemic endogamy can take place while endo-cuisine
cannot. Among the \Vakelburo where endo-cuisine is obligatory,
totemic endogamy seems to have been subject to a particularly
strict prohibition. The tribe in question has of course long been
extinct and the data on it are contradictory. (Compare the account
given by Frazer vol. 1, p. 423 and that given by Durkheim, p. 215,
102
SYSTEMS OF TRA.'.'ISFOR::\lA TIONS
n. 2). It is however striking that the symmetry with Aranda

institutions remains whichever account is accepted. The difference
is only that the relation between ·marriage rules and eating rules
would be supplementary in one case and complementary in the
other. i\ow, the examples of the feminine or masculine Fang cuits
made it clear that 'the same thing' can be said by means of rules
which are formally identical but whose content alone is the
reverse. And in the case of Australian societies, when the 'stressed'
foods are few or even, as often, consist of a single species, their
prohibition provides the most effective method of differentiation.
But, if the number of stressed foods should increase (a common
phenomenon, as we have seen [p. 88] in the northern tribes who
respect their mother's, father's and mother's mother's totems as
well as their own) it is easy to see that, without any change in the
spirit of the institutions on that account the distinguishing stresses
would be reversed and as in photography, the 'positive' would be
easier to read than the 'negative' although both convey the same
information.
Eating prohibitions and obligations thus seem to be theoretically
equivalent means of' denoting significance' in a logicalsystcm some
orall of whose elements are ediblc specics. These systems can how-
ever themselves be of different types, and this leads to a second
conclusion. There is nothing to suggest the existence of totemism
among the Bushmen of South Africa in spite of their complicated
and exacting prohibitions on food. This is because their system
operates on a diff erent plane.
All game killed by means of bows and arrows is forbiddcn, soxa,
until the chief has eaten a piece of it. This prohibition does not
apply to the liver which the hunters eat on the spot, but which
remains soxa to women in all circumstances. In addition to these
general rules, there are permanent soxa for certain functional or
social categories. For example, the wife of the man who killed the
animal can only eat the superficial covering of meat and fat of the
hindquarters, the entrails and the trotters. These pieces constitute
the portion reserved for women and children. The adolt>scent boys
have a right to the fiesh of the abdominal wall, kidneys, genital
organs and udders and the person who killed the animal to the
ribs and shoulder blade from one side of it. The chief's part con-
sists of a thick steak from each quarter and each side of the back
and a cutlet taken from cach sicle (Fourie, p. 55).
103
THE SAVAGE MIND
1t might seem at first sight as if no system could be less like one

of 'totemic' prohibitions. Nevertheless a very simple transforma-
tion allows one to pass from one to the other. One need only
substitute ethno-anatomy for ethno-zoology. Totemism postulates
a logical equivalence between a society of natural species and a
world of social groups. The Bushmen postulate the same formai
equivalence but in their case it is between the parts making up an
individual organism and the functional classes making up the
society, that is to say, the society itself is thought of as an organism.
Natural and social groupings are homologous in both cases and the
selection of a grouping in one order involves the adoption of the
corresponding grou ping in the other, at least as a predominant form. *
The next chapter will be devoted entirely to an analysis of the
same kind (that is, one of a transformation within a group) of the
empirical relations which are found between endogamy and
exogamy. All that 1 want to do now is to indicate the connection
between this problem and the one which has just been discussed.
ln the first place there is an empirical connection between
marriage rules and eating prohibitions. Among both the Tikopia of
Oceania and the Nuer of Africa a husband abstains from eating
animais or plants which his wife may not eat. The reason for this
is that ingested food contributes to the formation of the sperm and
he would otherwise introduce the forbidden foods into his wife's
body during intercourse (Firth 1, pp. 3 r9-20, Evans-Pritchard 2,
p. 86). ln the light of the preceding discussion, it is interesting to
observe that the Fang use the reverse argument. One of the many
reasons given in support of the prohibition applying to the inside
of elephants' tusks is that the penis might become as limp as the
gums of the pachyderm (which are apparently particularly so).
Out of consideration for her husband, a wife also obeys this
prohibition lest she should enfeeble him during intercourse
(Tessman, vol. 2, p. r92).
• So-called totemic societies also practise anatomical division but they use it to
make secondary distinctions: those between sub-groups of groups or individuals
belonging to the group. There is therefore no incompatibility between the two
sorts of grouping. lt is rather their respective positions in the logical hierarchy
which is to be treated as significant. 1 shall return to this later, cf. p. 174.
If, as G. Dieterlin points out, the Dogon associate totems and parts of the
body of a sacrificed ancestor, it is by applying a classificatory system with an
inter-tribal application. Consequently the totemic groupings within each tribe,
which are made to correspond to parts of the body, are in fact already secondary
units.
104
Now, these comparisons are only particular instances of the very

profound analogy which people throughout the world seem to find
between copulation and eating. ln a very large number of languages
they are even called by the same term. • ln Yoruba 'to eat' and 'to
marry' are expressed by a single verb the general sense of which is
'to win, to acquire', a usage which has its parallel in French, where
the verb 'consommer' applies both to marriage and to meals. ln the
language of the Koko Yao of Cape York Peninsula the word kuta
kuta means both incest and cannibalism, which are the most
exaggerated forms of sexual union and the consumption of food.
For the same reason the eating of the totem and incest are expressed
in the same way at Ponapy; among the Mashona and Matabele of
Africa the word 'totem' also means 'sister's vulva', which provides
indirect confirmation of the equivalence between eating and
copulation.
If eating the totem is a kind of cannibalism then one understands
that the punishment of real or symbolic cannibalism should be
reserved for those who violatc the prohibition - voluntarily or
otherwise. The symbolic roasting of the guilty in an oven in Samoa
is an example. And the equivalence is confirmed again by the
similar custom found among the Wotjobaluk of Australia where a
man who has committed the crime of absconding with a woman
forbidden by the law of exogamy is really eaten by the totemic
group. Without going so far afield or invoking other exotic rites,
one can quote Tertullian: 'Through love of eating, love of impurity
finds passage' (De Jejune, 1) and St. John Chrysostom: 'The
beginning of chastity is fasting' (Homilia I in Epistolam II ad
Thessalonicenses).
These associations could be multiplied indefinitely. Those 1 have
given as examples show how hopeless it is to attempt to establish a
relation of priority between nutritional prohibitions and rules of
exogamy. The connection between themisnotcausal butmetapho-
rical. Sexual and nutritional relations are at once associated even
today. Consider for instance slang expressions such as 'faire frire',
'passer à la casserole' etc. t But how is this fact and its universality
to be explained? Here again the logical level is reached by semantic
impoverishment: the lowest common denominator of the union of
•Cf. Henry, p. 146 for a particularly telling South American example.
t i.e. 'fry' and 'put in the pot', terms used to refer to seduction as well as
cooking (trans. note).
105
THE SAVAGE MIND
the sexes and the union of eater and eaten is that they both effect a
conjuncti on by corn plementarity:
\Vhat is destitute of motion is the food of those endowed with locomo-
tion; (animais) without fangs (are the food) of those with fangs, those
without hands of those who possess hands, and the timid of the bold (The
Laws of 1\fanu, v, 29).
The equation of male with devourer and female with devoured is
the more familiar to us and certainly also the more prevalent in the
world but one must not forget that the inverse equivalence is often
found at a mythological level in the theme of the vagina dentata.
Significantly enough this is 'coded' in terms of eating, that is,
directly (thus confirming the law of mythical thought that the
transformation of a metaphor is achieved in a metonymy). It may
also be the case that the thème of the vagina d enta ta corresponds to
a point of view, no longer inverted but direct, in the sexual
philosophy of the Far East where, as the works of Van Gulik show,
for a man the art of love-making consists essentially in avoiding
having his vital force absorbed by the women and in turning this
risk to his ad\·antage.
This logical subordination of resemblance to contrast is clearly
illustrated in the complex attitude of some so-called totemic
peoples toward the parts of body of eponymous animais. The
Tikuna of the High Solimoes, who practise 'hypertotemic'
exogamy (the members of the Toucan clan cannot inter-marry
either among themselves or with the members of any clan called by
a bird name, etc.), freely eat the eponymous animal but they respect
and preserve a sacred part of it and use others as distinctive
ornaments (Alviano). The totemic animal is thus resolved into an
edible part, a part to be respected and an emblematic part. The
Elema of Southern New Guinea observe a very strict prohibition
on eating their totems. Each clan however retains the exclusive
prerogatiœ of using the beak, feathers, tail, etc. of its totem animal
for ornamental purposes (Frazer, vol. II, p. 41). There is thus in
both cases an opposition between edible and non-edible parts,
homologous to that between the categories of food and emblem. In
the case of the Elema it is expressed through a two-fold exclusivism
which may be either negative or positive: in relation to the totemic
species, each clan abstains /rom its meat but retains those parts of it
which display the characteristics of its species. The Tikuna are
equally exclusive with respect to the distinctive parts of the
106
animais but they adopt a communal attitude towards meat (in

terms of which animais distinct in species but edible resemble each
other as food). The group of attitudes can be represented in the
following way:
\ \
Tikuna: non-exèlusive exclusive
\ \
\ \
\ \
\ \
\ \
\ \
\ \
\ \
\ \
\
\ \
\ \
\ \
\ \
'\ \
Elema: \ prohibition(-) \ prerogative (+)

' \
(inedible
\edible parts
and common parts and
properticsJ distinctive
properties)
The fur, feathers, beak and teeth can be mine because they are that
in which the eponymous animal and I differ from each other: this
difference is assumed by man as an emblcm and to assert his
symbolic relation with the animal. The parts which are edible and
so can be assimilated are the sign of genuine consubstantiality but
which, contrary to what one might suppose, it is the real aim of
the prohibition on eating them to deny. Ethnologists have made
the mistake of taking only this second aspect into account and this
has led them to conceive the relation between man and animal to
be of a single kind: identity, affinity or participation. l\Iatters are in
fact infinitely more complex: there is an exchange of similarities
and differences between culture and nature, sometimes as amongst
animais on the one sicle and man on the other and sometimes as
between animais and men.
The differences between animais, which man can extract from
nature and transfer to culture ( ei ther by describing them in terms
of opposites and contrasts and thus conceptualizing them or by
taking over concrete, non-perishable parts: feathers, beaks, teeth -
which equally constitutes an 'abstraction') are adopted as emblems
by groups of men in order to do away with their own resemblances.
And the same animais are rejected as food by the same group of
men, in other words the resemblance between man and animal
resulting from the fact that the former can assimilate the flesh of
107
THE SAVAGE MIND
the latter is denied, but only in so far as it is perceived that the

reverse course would imply recognition on the part of men of their
common nature. The meat of any animal species must therefore not
be assimilated by any group of men.
It is clear that the second approach derives from the first, as a
possible but not necessary consequence: nutritional prohibitions
do not always accompany totemic classifications and they are
logically subordinate to them. They do not therefore raise a
separate problem. If, by means of nutritional prohibitions, men
refuse to attribute a real animal nature to their humanity, it is
because they have to assume the symbolic characteristics by which
they distinguish different animais (and which furnish them with a
natural mode! of differentiation) to create differences among
themselves.
108
CHAPTER FOUR
TOTEM AND CASTE
Both the exchange of women and the ex change of food are means
of securing or of displaying the interlocking of social groups with
one another. This being so, we can see why they may be found
either together or separately. They are procedures of the same type
and are indeed generally thought of as two aspects of the same
procedure. They may reinforce each other, both performing the
actual function, or one performing it and the other representing it
symbolically. Or they may be alternatives, a single one fulfilling the
whole function or if that is otherwise discharged, as it can be even
in the absence of both procedures, then the symbolic representation
of it:
If ... a people combines exogamy with totemism, this is because it has
chosen to reinforce the social cohesion already established by totemism
by superimposing on it yet another system which is connected with the
first by its reference to physical and social kioship and is distinguished
from, though not opposed toit, by its Jack of reference to cosmic kinship.
Exogamy can play this same part in types of society which are built on
foundations other than totemism; and the geographical distribution of
the two institutions coïncides only at certain points in the world.
(Van Gennep, pp. 351).
However exogamy, as we know, is never entirely absent. This is due
to the fact that the perpetuation of the group can only be effected
by means of women, and although varying degrees cf symbolic
content can be introduced by the particular way in which a society
organizes them or thinks of their operation, marriage exchanges
always have real substance, and they are alone in this. The exchange
of food is a different matter. Aranda women really bear children.
But Aranda men confine themselves to imagining that their rites
THE SAVAGE MIND
result in the increase of totcmic species. ln the former, although it

may be described in conventional terms which impose their own
limits on it, what is in question is primarily a way of doing some-
thing. ln the latter it is only a way of saying something.
Examples of 'accumulation' of some kind have attracted particu-
lar attention, no doubt because the repetition of the same scheme
on two different planes made them look simpler and more consis-
tent. It is mainly this which has led to the definition of totemism by
the parallelism between eating prohibitions and rules of exogamy
and to making this supplementary set of customs a special object
of study. There are, however, cases in which the relation between
marriage and eating customs is one of complementarity and not
supplementarity and where they are therefore dialectically related
to each other. This form clearly also belongs to the same group.
And it is groups in this sense, and not arbitrarily isolated trans-
formations, which are the proper subject of the sciences of man.
In an earlier chapter I quoted a botanist's testimony with regard
to the extreme purity of types of seed in the agriculture of so-called
primitive peoples, in particular among the Indians of Guatemala.
\Ve also know that there is intense fear of agricultural exchanges
in this area: a transplanted seedling may take the spirit of the plant
\Vith it, \Vith the result that it disappears from its original locality.
One may, then, exchange women but refuse to exchange seeds.
This is common in Melancsia.
The inhabitants of Dobu, an island to the south-east of New
Guinea, are divided into matrilineal lineages called susu. Husband
and wife, who necessarily corne from different susu, each bring
their own seed yams and cultivate them in separate gardens without
ever mixing them. :\'"o hope for a person who has not his own seed:
a woman who has none will not succeed in marrying and will be
reduced to the state of a fisherwoman, thief or beggar. Seed which
docs not corne from the susu will not grow, for agriculture is poss-
ible only by the use of magic inherited from the maternai uncle: it
is ritual which makes the yams swell.
These precautions and semples rest on the belief that yams are
persons: 'Like women, they give birth to children .. .' They go
abroad at night and people wait for their return before harvesting.
This is the source of the rule that yams may not be dug too early
in the morning: they might not yet have returned. It is also the
source of the conviction that the fortunate cultivator is a magician
110
TOTEM AND CASTE
who has known how to persuade his neighbours' yams to move and
establish themselves in his garden. A man who has a good harvest
is reckoned a lucky thief (Fortune 2).
Beliefs of the same type were to be found even in France until
recently. In the middle ages there was a penalty of death for 'the
sorceress who defiled and injured crops; who, by reciting the
psalm Super aspidem ambulabis, emptied the fields of their corn to
fill her own granary with this goodly produce'. Not so long ago at
Cubjac in the Perigord a magical invocation was supposed to
assure the person using it of a good crop of turnips: 'May our
neighbours' be as big as millet seed, our relations' as big as grains
of corn and our own as big as the head of Fauve the ox !' (Rocal,
pp. 164-5).
Apart from the modicum of exogamy resulting from the pro-
hibited degrees, European peasant societies practised strict local
endogamy. And it is significant that at Do bu extreme endo-agricul-
ture can act as the symbolic compensation for lineage and village
exogamy which is practised with repugnance and even f ear. In
spite of the fact that endogamy within the locality - which consists
of between four and twenty villages - is generally assured, marriage
even into the next village is looked on as putting a man at the mercy
of assassins and sorcerers and he himself always regards his wife
as a powerful magician, ready to decei\'e him with her childhood
friends and to destroy him and his (Fortune 2). In a case like this,
endo-agriculture reinforces a latent tendency towards endogamy,
if indeed it does not express symbolically the hostility to the
unwillingly practised rules of a precarious exogamy. The situation
is symmetrically the reverse of that prevailing in Australia where
food prohibitions and rules of exogamy reinforce one another, as
we ha\·e seen in a more symbolic and clearly conceptual way in the
patrilineal societies (where the food prohibitions are flexible and
tend to be formulated in terms of moieties, that is, at a level which
is already abstract and lends itself to a binary coding by pairs of
oppositions) and in a more literai and concrete fashion in the
matrilineal societies (where the prohibitions are rigid anrl stated in
terms of clans which one might often be hesitant to regard as
members of systematic sets, given the determining part of demo-
graphic and historie factors in their genesis ).
Apart from these cases of positive or negative parallelism, there
are others in which reciprocity between social groups is expressed
III
THE SAVAGE MIND
only on one plane. Omaha rules of marriage are formulated very

differently from those of the Aranda. Instead of the class of the
spouse being precisely specified, as it is among the Aranda, any
clan not expressly forbidden is permitted. On the plane of food,
however, the Omaha have rites very similar to the intichiuma :•the
sacred maize is entrusted to particular clans who annually distri-
bute it to the others to vitalize their seeds (Fletcher and La
Flesche). The totemic clans of the Nandi of Uganda are not
exogamous; but a remarkable development of clan prohibitions,
not only on the plane of food but also on those of technical and
economic activities, dress and impediments to marriage based on
some detail or other of the persona! history of the forbidden spouse,
compensates for this 'non-functionality' in the sphere of marriage
exchanges (Hollis). No system can be constructed from these differ-
ences: the distinctions recognized between the groups seem rather
to spring from a propensity to accept ail statistical fluctuations.Jn
a different form and on a different plane, this is also the method
employed by the systems termed 'Crow-Omaha' and by contemp-
orary Western societies to ensure the overall equilibrium of
matrimonial exchanges. t
This emergence of methods of articulation more complex than
those resulting just from rules of exogamy or food prohibitions,
or even of both at once, is particularly striking in the case of the
Baganda (who are near the Nandi) because they seem to have
accumulated ail the forms. The Baganda were divided into forty
clans, kika, each of which had a common totem, miziro, the con-
sumption of which was forbidden by virtue of a rule of food
rationing: by depriving itself of the totemic food, each clan leaves
more of it available to other clans. This is the modest counterpart
of the Australian daim that by refraining from ccnsuming its
totem each clan retains the power to increase it.
As in Australia, each clan is characterized by its links with a
territory, among the Baganda generally a hill. There is a secondary
totem, kabiro, as well as the principal totem. Each Baganda clan
is thus defined by two totems, food prohibitions and a territory.
There are also prerogatives such as eligibility of its members for
• See below, p. 226.
t Rightly or wrongly, Radcliffe-Brown (3, pp. 32-3) treats the Nandi kinship
system as an Omaha system.
112
TOTEM AND CASTE
the kingship or other honours, the right to provide royal wives,

making and caring for the royal emblems or utensils, ritual
obligations to provide other clans with certain kinds of food and
technical specializations: the Mushroom clan, for instance, makes
all the bark cloth, and all blacksmiths corne from the clan of the
Tail-less Cow, etc. Finally, we find some prohibitions, such as that
the women of a particular clan cannot be the mothers of male
children of the blood royal, and restrictions with regard to the
bearing of proper names (Roscoe).
In cases like this it is no longer very clear what type of society is in

question. There can, for instance, be no doubt that the totemic
clans of the Baganda also fonction as castes. And yet at first sight
it seems that nothing could be more different than these two fonr,s
of institution. \Ve have become used to associating totemic groups
with the most 'primitive' civilizations and thinking of castes as a
feature of highly developed, sometimes even literate, societies.
l\loreover, a strong tradition connects totemic institutions with the
strictest exogamy while an anthropologist asked to define the con-
cept of a caste would almost certainly begin by mentioning the
rule of endogamy.
It may therefore seem surprising that between about 1830 and
1850, the first investigators of Australian societies often referred to
their marriage classes as 'castes' even though they had some idea
of their fonction (Thomas, pp. 34-5). These intuitions which have
the freshness and vivacity of a perception of societies which were
still intact, and a vision undistorted by theoretical speculation, are
not to be despised. Without going to the root of the problem now,
it is clear that there are at least superficial analogies between
Australian tribes and societies with castes. Each group has a
specialized fonction indispensable to the collectivity as a whole and
complementary to the fonctions assigned to other groups.
This is particularly clear in the case of tribes whose cians or
moieties are bound together by a rule of reciprocity. Among the
Kaitish and the Unmatjera, northern neighbours of the Aranda,
anyone who gathers wild seeds in the territory of a totemic group
named after these seeds must ask the headman's permission before
eating them. It is the duty of each totemic group to provide the
other groups with the plant or animal for whose 'production' it is
specially responsible. Thus a man of the Emu clan out hunting on
113
THE SAVAGE Ml1';D
his own may not touch an cmu. But if, on the othcr hand, he is in
company he is permitted and even supposed to kill it and offer it
to hunters of other clans. Conversely, when he is alone a man of
the Water clan may drink if hc is thirsty but when he is with others
he must receive the water from a member of the other moiety,
prefcrably from a brother-in-law (Spencer and Gillen, pp.
159-60). Among the \Varramunga each totemic group is respons-
ible for the increase and a vailability to other groups of a particular
plant or animal species: 'The members of one moiety ... take
charge ... of the ceremonies of the other moiety which are des-
tined to secure the increase of their own food suppl y'. Among the
\Valpari as well as the \Varramunga the secondary totemic prohi-
bitions (applying to the maternai totem) are waived if the food in
question is obtained through the agency of a man of the other
moiety. l\1ore generally and for any totem, therc is a distinction
between the groups which never eat it (because it is their own
totem), those which eat it only if it is procured through the agency
of another group (as in the case of the maternai totems), and those
which eat it freely in any circumstances. Similarly in the case of
the sacred water-holes, women may never approach them, uniniti-
ated men may approach but not drink from them, while some
groups drink from them on the condition that the water is given
to them by members of other groups who can themselves drink
freely from them (Spencer and Gillen, pp. 164, 167). This mutual
interdependence is already to be seen in marriage which, as
Radcliffe-Brown has shown in the case of Australia (but the same
could equally well be said of other clan societies such as the
Iroquois), was based on reciprocal gifts of vegetable food (fem-
inine) and animal food (masculine): the conjugal family in these
cases was like a miniature society with two castes.
There is thus Jess difference than would appear between
societies which, like some Australian tribes, assign a distinctive
magico-cconomic fonction to totemic groups and, for instance,
the Bororo of Central Brazil, among whom specialists are in charge
of the same fonction of 'liberating' the food production - whether
animal or vegctable - for the whole group (Colbacchini). This
leads one to doubt whether the opposition between endogamous
castes and exogamous totcmic groups is really radical. There seem
to be connections between these two extreme types, whose nature
114
TOTEM AND CASTE
would appear more clearly if we could show that intermediate

forms exist.
I have drawn attention elsewhereto a feature of so-called totemic
institutions which in my own ,·iew is fundamental to them. The
homology they ernke is not between social groups and natural
species but between the differences which manifest themselves on
the level of groups on the one hand and on that of species on the
other. They are thus based on the postula te of a homology between
two systems of differt'nces, one of which occurs in nature and the
other in culture. Indicating relations of homology by vertical lines,
a 'pure totemic structure' could thus be represented in the follow-
mg way:
);ATL'RE: species I + species 2 + species 3 + ..... species n
1 1 1
Cl'LTl'RE: group I + group 2 + group 3 + ..... group n
This structure would be fundamentally impaired if homologies
between the terms themselves were added to those between their
relations or if, going one step further, the entire system of homo-
logies were transferred from relations to terms:
NATt:RE: species I species 2 -1- species 3 + ..... speciesn
1 1 l
Cl'LTL'RE: group 1 + group 2 + group 3 + ..... group 11
In this case the implicit content of the structure would no longer

be that clan 1 differs from clan 2 as for instance the eagle differs
from the bear but rather that clan 1 is like the eagle and clan 2 like
the bear. In other words, the nature of clan 1 and the nature of
clan 2 would each be involved separately instead Df the formal
relation between them.
Now, the transformation whose theoretical possibility has just
been considt:red can sometimes be directly observed. The islanders
of the Torres Straits have totemic clans, numbering about thirty
at l\fabuiag. These exogamous patrilineal clans were grouped into
two moieties, one comprising terrestrial and the other marine
animais. At Tutu and Saibai this division seems to have corres-
ponded to a territorial division within the village. The structure
was already in an advanced state of decay at the time of Haddon's
expedition. Nevertheless, the natives had a very strong sense of the
physical and psychological affinity between men and their totems
and of the corresponding obligation of each group to pursue the
THE SAVAGE MIND
appropriate type of behaviour. Thus the Cassowary, Crocodile,

Snake, Shark and Hammer-headed Shark clans were said to love
fighting and the Shovel-nosed Skate, Ray and Sucker-Fish clans
to be peace loving. The Dog clan was held to be unpredictable,
dogs being of a changeable disposition. The members of the
Crocodile clan were thought to be strong and ruthless and those
vf the Cassowary clan to have lùng legs and to run fast (Frazer,
vol. Il, pp. 3-<), quoting Haddon and Rivers). lt would be inter-
esting to know whether these beliefs are survivais from the old
organization or whether they developed as the exogamous rules
decayed.
The fact is that similar, though not equally developed, beliefs
have been observed among the Menomini of the Great Lakes and
among the Chippewa further north. Among the latter, people of
the Fish clan were reputed to be long-lived, frequently to go bald
or to have thin haïr, and al! bald people were assumed to corne from
this clan. Peoples of the Bear clan, on the other hand, had long,
thick, coarse haïr which never went white and they were said to be
ill-tempered and fond of fighting. People of the Crane clan had
loud ringing voices and provided the tribe with its orators (Kinietz,
pp. 76--7).
Let us pause for a moment to consider the theoretical implica-
tions of views like these. When nature and culture are thought of
as two systems of differences between which there is a formai
analogy, it is the systematic character of each domain which is
brought to the fore. Social groups are distinguished from one
another but they retain their solidarity as parts of the same whole,
and the rule of exogamy furnishes the means of resolving this
opposition balanced between diversity and unity. But if social
groups are considered not so much from the point of view of their
reciprocal relations in social life as each on their own account, in
relation to something other than sociological reality, then the idea
of diversity is likely to prevail over that of unity. Each social group
will tend to form a system no longer with other social groups but
with particular differentiating properties regarded as hereditary,
and these characteristics exclusive to each group will weaken the
framework of their solidarity within the society. The more each
group tries to define itself by the image which it draws from a
natural mode!, the more difficult will it become for it to maintain
its links with other social groups and, in particular to exchange its
u6
TOTEM AND CASTE
sisters and daughters with them since it will tend to think of them
as being of a particular 'species'. Two images, one social and the
other natural, and each articulated ~eparately, will be replaced by a
socio-natural image, single but fragmented :*
::-;ATl"RE:
species r species 2 species 3 species n
Cl"LTL'RE:
group group 2 group 3 group n
It is of course only for purposes of exposition and because they

form the subject of this book that I am apparently giving a sort of
priority to ideology and superstructures. I do not at ail mean to
suggest that ideological transformations give rise to social ones.
Only the reYersc is in fact truc. l\'Ien's conception of the relations
betwccn nature and culture is a function of modifications of their
own social relations. But, since my aim here is to outline a theory
of superstructures, reasons of method require that they should be
singled out for attention and that major phenomena which have no
place in this programme should seem to be left in brackets or given
second place. \Ve are however mcrely studying the shadows on th~
wall of the Cave without forgetting that it is only the attention wc
give them which lcnds them a semblance of reality.
This said, to avoid misunderstanding we can summarize what has

gone before as an account of the conceptual transformations mark-
ing the passage from exogamy to endogamy (or vice versa). Sorne,
at any rate, of the Algonkin tribes, who furnished the last examples,
had a hierarchical clan structure which one might suspect would
lead to some difficulty in the functioning of exogamous rules form-
ulated in egalitarian terms. But it is in the south-east of the United
States, in the tribes of the Muskogi linguistic group, that hybrid
institutional forms, half-way between totemic groups and castes,
can be seen most clearly; and this also explains the existing unccr-
tainty as to whether they are endogamous or exogamous.
The Chickasaw may perhaps have been exogamous at the clan
• It will pcrhaps be objcctcd that in the above mcntioned work (6), I dcnicd
that totcmism can be interprded on the basis of a direct analogy betwccn human
groups and natural species. But this criticism was directed against a theory put
forward by ethnologists and what is in question hcre is an - implicit or explicit -
native theory which indeed corresponds to institutions that cthnologists would
refuse to class as totemic.
I 17
THE SAVAGE MIND
level and endogamous so far as their moieties were concerned. The

latter in any case had the feature, highly unusual for structures of
this type, of displaying an exclusivism verging on mutual hostility.
Illness and death were of ten attributed to the sorcery of people of
the opposite moiety. Each moiety performed its rites in jealous
isolation; members of the other moiety who witnessed them were
punishable by death. The same attitude existed among the Creek:
with regard to moieties it is strikingly reminiscent of that prevailing
towards totemic groups among the Aranda. Each performed its
rites 'among themselves' although only 'the others' were to benefit
from them. And this, it is worth remarking in passing, shows that
endo-praxis and exo-praxis are never definable separately and in
absolute terms. As Morgan demonstrated against McLennon, they
can only be defined as complementary aspects of an ambiguous
relation to self and to others.
The moieties, which probably formed opposite sicles in com-
petitive gamcs, were considered to differ in temperament and
habitat: one was warlike and preferred open country while the
other was pacifie and lived in forests. They may also have been
ordered hierarchically as is suggested by the terms sometimes
applied to them : '[people of the] hickory-choppings' [substantial
habitations] and '[people of the] worn-out place' [hovels]. How-
ever, thesc hierarchical, psychical and functional differences were
primarily manifested at the level of clans or their subdivisions
into hamlets. In native accounts of the past, comments like these
about each clan or hamlet constantly occur, like a leit-motiv:
'They were very peculiar people ... they were not like others ...
they had customs and traditions of their own .. .' These peculiari-
ties were of ail sorts of different kinds: place of residence,
economic activity, dress, food, talents and tastes.
The Raccoon people were said to live on fish and wild fruit,
thosc of the Puma lived in the mountains, avoided water of which
they were very frightened and lived principally on game. The
Wild Cat clan slept in the daytime and hunted at night, for they
had keen cyes; they were indifferent to women. Members of the
Ilird clan were up before da y-break: 'They were like real birds
in that they would not bother anybody ... The people of this
clan have different sorts of minds, just as there are different
species of birds'. They were said to live well, to be polygamous,
disinclined to work, and prolific.
II8
TOTEM AND CASTE
The people of the Red Fox clan were professional thieves,

loving independence and living in the heart of forests. The
'wandering Iksa' were nomadic and improvident but nevertheless
enjoyed robust health 'for they did not do anything to run them-
selves down'. They moved slowly, thinking that they were going
to live for ever. The men and women paid little attention to their
dress or appearance. They were beggars and lazy. The inhabitants
of 'the bending-post-oak' house group lived in the woods. They
were of a changeable disposition, not very encrgetic, given to
dancing, always anxious and full of care. Theywere early risers and
clumsy. The High Corncrib house group people were respected in
spi te of their arrogance: they were good gardeners, very indus-
trious but poor hunters, they bartered their maize for game. They
were said to be truthful and stubborn, and skilled at forecasting
the weather. As for the Redskunk house group: they lived in
dugouts underground (Swanton 2, pp. 190-213).
This material was collected at a time when the traditional
institutions no longer existed except in old informants' memories
and it is plain that it is partly made up of old wives' tales. No
society could allow itself to 'act nature' to this extent or it would
split up into a whole lot of independent, hostile bands, cach
denying that the others were human. The data which Swanton
collected consist of sociological myths as well as or rather than
ethnographie facts. Nevertheless, their wealth, the resemblances
they have among themselves, the unity of the underlying scheme,
the existence of similar data for neighbouring groups ail suggest
that even if the real institutions were different, we have here at
least a sort of conceptual model of Chickasaw society which has the
extremely interesting f eature of recalling a socièty with castes,
even though the attributes of the castes and their relations to each
other are coded in terms of natural species, that is, after the
manner of totemic groups. Further, the relations held to exist
between clans and their eponyms are like those found in classical
'totemic' societies: the clan is descended from the animal or
alternatively a human ancestor of the clan contracted a marriage
with one in mythical times. Now, these societies, which are at the
very least conceived of as if they were composed of 'natural'
castes - or, in other words, in which culture is thought of as a
projection or reflection of nature - are the link between the societies
classical authors used to illustrate their conception of totemism
THE SAVAGE MIND
(the tribes of the plains and of the south west) and societies such
as the Natchez which afford one of the rare examples of genuine
castes known in ~ orth America.
W e have thus established that in the two classical territories of
so-called totemism, the institutions defined with ref erence to this
misleading notion can either also be characterized from the point
of view of their function, as in Australia or, as in America, make
way for forms which are still conceived on the model of totemic
groups although they operate more like castes.
Let us now turn to India, also classical territory but of castes.
I shall try to show that through their influence institutions tradi-
tionally thought of as totemic undergo a transformation exactly the
reverse of that in America: instead of castes conceived in terms of
a natural model we have here totemic groups conceived in terms
of a cultural mode!.
l\Iost of the totemic names found among certain tribes in Bengal
derive from animals or plants. This is the case with some sixty-
seven totems recorded among the Oraon of Chota Nagput with
the exception of iron which, as there is little point in proscribing
its consumption, is forbidden to corne into contact with lips or
tongue. This prohibition is thus still formulated in terms that
make it approximate to an eating prohibition. Among the Munda
of the same region, the majority of the three hundred and forty
exogamous clans recorded have animal or plant totems, the
consumption of \vhich is forbidden. Totems of a different kind are
however already noticeable: full moon, moonlight, rainbow,
months of the year, days of the week, copper bracelet, verandah,
umbrella, professions or castes such as that of basket-maker and
torch-bearer (Risley, vol. II and Appendix). Further west, the
forty-three names of the Bhil clans are divided into nineteen plant
and seventeen animal names and seven relate to objects: dagger,
broken pot, village, thorny stick, bracelet, ankle ring, piece of bread
(Koppers, pp. 118-19).
It is towards the south that the reversa! in the relation of natural
specits and objects or manufactured goods becomes particularly
conspicuous. Few plants and scarcely any animais figure in the
names of the clans of the Devanga, a caste of weavers in the
:\Iadras area. On the other hand, the following names are found:
buttermilk, cattle-pen, money, dam, house, collyrium, knife,
scissors, boat, lamp, cloths, female clothing, ropes for hanging pots,
120
TOTE:\I A:-ID CASTE
old plough, monastery, funeral pyre, tile. Sixty-seven exogamous

clans are recorded among the Kuruba of Mysore. They have
animal or plant names or names. like the following: cart, cup,
silYer, flint, roll of woollen thread, bangle, gold, gold ring, pickaxe,
coloured border of cloth, stick, blanket, measure, moustache, loom,
bamboo tube, etc. (Thurston, vol. II, p. 160 ff., vol. IV, pp. 141-2).
It is possible that this phenomenon is a peripheral rather than a
southern one for one is inclined here to recall the mythical role
which some south-east Asian tribes ascribe to manufactured
objects such as sabre, knife, lance, needle, post, rope, and so on.
Howevcr this may be, in India the manufactured objects from
which clans take their names receive special respect, like totemic
plants and animais. Either they constitute the object of a cuit at
marriages or alternatively the respect paid to thcm takes a bizarre
and specific form. Among the Bhil for example, the clan of the
broken pot must collect pieces of a particular kind of pottery and
give them burial. At times a certain freedom of invention is per-
ceptible: the Arisa na gotram of the Karnba bears the name of
turmeric, but as it is held to be inconvenient to hc deprived of so
essential a condiment, the Korra grain replaces it as the forbidden
food.
Heterogeneous lists of clan names are known in other parts of the

world. Perhaps significantly, they are particularly found in the
north of Australia, the part of the continent most subjcct to outside
influences. Individual totems such as a razor blade or money haYe
been noted in Africa :
\Vhen I asked [the Dinka] what 1 myself should invoke as my clan-
divinity, it was half-jokîngly suggested that 1 should invoke Typewriter,
Paper, and Lorry, for were these not the things which had always helped
my people and which were passed on to Europeans by their ancestors?
(Lienhardt, p. 1 ro).
But this heterogeneity is most apparent in lndia where a high

proportion of totemic names are names of manufactured objects,
that is, of products or symbols of functional activities which -
because they are clearly differentiated in a caste system - can serve
to express distinction between social groups within the tribe or the
caste itself. It is as if in America the rudiments of castes had been
contaminated by totemic classifications, while in lndia the vestiges
121
THE SAVAGE MIND
of totemic groups had allowed themselves to be won over by sym-

bolism of technological or occupational origin. These 'chassés-
croisés' seem less surprising when one realizes that there is a
neater and more direct way of translating Australian institutions
into the language of castes than that used above.
I have sug~ested that since each totemic group makes itself
rcsponsible for the control of a species of plant or animal for the
benefit of the other groups, these specializations of fonction are,
from one point of view, similar to those assumed by occupational
castes sincc the latter also practise a distinctive activity, indispen-
sable to the life and well-being of the whole group. However, in the
first place, a caste of potters really makes pots, a caste of launderers
really washes clothes, a caste of barbers really shaves people, while
the magical powers of Australian totemic groups are of an imagi-
nary kind. And there is a distinction here even in belief in the
efficacy of magical powers is shared by their supposed beneficiaries
and by those who, in all good faith, daim to possess them.
Secondly, the connection between a sorcerer and the natural
species he daims to control cannot be conceived in terms of the
same logical model as that between a craftsman and his product.
For it was only in mythical times that totemic animais were really
begotten by the ancestor. Nowadays it is kangaroos which produce
kangaroos, and the sorcerer contents himself with assisting them.
Now, if Australian (and other) institutions are considered from
a wider point of view, it becomes possible to distinguish a field in
which the parallel with a caste system is very much clearer. \Ve
need only turn our attention to social organization instead of
religious belicfs and practices. For the early observers of Australian
societies were in a sense right to speak of marriage classes as
'castes': an Australian section produces its women for other
sections in the same way as an occupational caste produces goods
and services which other castes cannot obtain otherwise than
through its agency ... It would therefore be superficial to regard
them as opposites simply because one is exogamous and the other
endogamous. Occupational castes and totemic groups are really
both 'exo-practising', the former in the exchange of goods and
services and the latter so far as marriage is concerned.
A coefficient of 'endo-praxis' is however always discernible in
either case. Castes are ostensibly endogamous apart from the
restrictions on marriage which, as I have shown elsewhere (I, ch.
122
TOTEM AND CASTE
25), tend to multiply in compensation. The Australian groups are

exogamous but their exogamy most commonly takes the form of
restricted exchange which is an imitation of endogamy within
exogamy itself, for in restricted exchange groups consider them-
selves as closed to the outside and their internai exchanges double
up on each other. It can therefore be contrasted with generalized
exchange which is more open to the outside and allows the incor-
poration of new groups without upsetting the structure. These
relations can be shown by means of a diagram:
1--- -----~1
endogamy exogamy
1 1
restricted generalized
ex change ex change
It will be seen that restricted exchange, the 'closed' form of

exogamy is logically doser to endogamy than the 'open' form,
generalized exchange.
There is a further point. A fondamental difference exists
between the women who are exchanged and the goods and services
which are also exchanged. Women are biological individuals, that
is, natural products naturally procreated by other biological
individuals. Goods and services on the other hand are manufact-
ured objects (or opcrations performed by means of techniques
and manufactured objects), that is, social products culturally
manufactured by technical agents. The symmetry between
occupational castes and totemic groups is an inverted symmetry.
The principle on which they are differentiated is taken from
culture in one case and from nature in the other.
Nevertheless, this symmetry is present only on an ideological
plane. It has no concrete basis. So far as culture is concemed
prof essional specialities are truly different and complementary. The
same could not be said, with respect to nature, of the specialization
of exogamous groupings in the production of women of different
species. For even granting that occupations do constitute distinct
'social species', this does not alter the fact that the women of differ-
ent sections or sub-sections all belong to the same natural species.
This is the trap reality sets for the imagination and men try to
escape it by seeking real diversity in the natural order, which is
(if they pay no attention to the division of labour and occupational
specialization) the only objective model on which they can draw
123
THE SA VAGE MI);D
for establishing relations of complcmentarity and co-operation

among themselves. ln other words, men conceive these relations
on the model of their conception of the relations between natural
species (and at the same time of their own social relations). There
are in fact only two true models of concrete diversity: one on the
plane of nature, namely that of the di\·ersity of species, and the
other on the cultural plane provided by the diversity of fonctions.
The mode! illustrated by marriage exchanges lying between these
two true models, has an ambiguous and equivocal character.
\Vomen are alike so far as nature is concerned and can be regarded
as different only from the cultural angle. But if the first point of vie\v
is predominant (as is the case when it is the natural mode! \vhich is
chosen as the mode[ of diversity) resemblance out\veighs differ-
ence. \Vomen certainly have to be exchanged since they have been
decreed to be different. But this exchange presupposes that
basically they are held to be alike. Conversely, when the other
viewpoint is taken and a cultural mode! of diversity adopted,
difference, which corresponds to the cultural aspect, outweighs
resemblance. \Vomen are only recognized as alike within the limits
of their respective social groups and consequently cannot be
exchanged between one caste and another. Castes decree women
to be naturally heterogeneous; totemic groups decree them to be
culturally heterogeneous. And the final reason for this ditference
between the t\VO systems is that castes exploit cultural hetero-
geneity in earnest while totemic groups only create the illusion of
exploiting natural heterogeneity.
Ali this can be expressed in a different way. Castes, which are
defined on the basis of a cultural mode!, really exchange cultural
abjects. But they have to pay a price for the symmetry they
postula te bet\veen nature and culture: in that the castes are them-
selves composed of biological beings, they are constrained to
conceive their natural product according to a natural world, since
this product consists of women \vhom they bath produce and are
produced by. lt follows that women are made di\·erse on the mode!
of natural species and cannot be exchanged any more than species
can cross with one another. Totemic groupings make the reverse
sacrifice. They are defined on the basis of a natural mode! and
exchange natural abjects - the women they produce and are pro-
duced by naturally. The symmetry postulated between nature ;:nd
culture inrnlves in that case the assimulation of natural species to
124
TOTE:\1 AND CASTE
the cultural plane. In the same way that women who are homo-
geneousso far as nature is concernedare declared to be heterogeneous
from the point of \'Îew of culture, so natural species, which are
heterogeneous so far as nature is concerned are proclaimed to be
homogeneous from the point of Yiew of culture: culture asserts
them ail to be subject to the same type of beliefs and practices since
in the eyes of culture, they han the comrnon feature that man has
the power to contrai and increase them. Consequently, men by
cultural means exchange women who perpetuate these same men
by natural means and they claim to perpetuate species by cultural
means and exchange them sub specie naturae: in the form of food-
stuffs which are substitutable for each othcr since they ail provide
nourishment and since, as with women also, a man can satisfy
himself by means of some foods and go without others in so far as
any women or any foods are equally suitable to achieYe the ends
of procreation or subsistence.
\Ve thus arrive at the common properties of which occupational

castes and totemic groups prO\·ide contrary illustrations. Castes
are heterogeneous in fonction and can therefore be homogeneous
in structure: since the di\·crsity of fonction is real, complement-
arity is already established on the lenl of reality and the operation
of marri age exchanges - between the same social units - would be
a case of 'accumulation' of fonctions (why this is of no practical
value has been shown above [cf. p. 109]). ConYersely, totemic
groups are homogeneous so far as their fonction is concerned, for it
makes no real yield and àmounts to no more than a repetition of
the same illusior. for ail the groups. They therefore haYe to be
heterogeneous in structure, each being destined for the production
of women of a diff erent social species.
In totemism, consequently, purported reciprocity is constructed
out of modes of behaviour homogeneous with each other and
simply jlLxtaposed. Each group similarly imagines itself to have
magical power over a species, but as this illusion has no foundation
it is in fact no more than an empty form and as such identical to
the other forms. The true reciprocity results from the articulation
of two processes: the natural one which cornes about by means of
women, who procreate bath men and women, and the cultural one
which men bring about by characterizing these women socially
when nature has brought them into existence.
THE SAVAGE MIND
In the caste system reciprocity is manifested by specialization

of fonction and it is practised on the the cultural plane. The
valencies of homogeneity are therefore freed; from being formai,
the analogy postulated between human groups and natural species
bccomes substantial (as the example of the Chickasaw and the
quotation from the Laws of Manu [cf. p. 106], shows) and true
reciprocity being otherwise, secured, endogamy is made available.
There are, however, limits to this symmetry. Totemic groups
certainly give an imitation of gift-giving which has a fonction. But,
apart from the fact that it remains imaginary, it is not cultural
cither since it must be classed, not among the arts of civilization,
but as a fake usurpation of natural capacities which man as a
biological species Jacks. Certainly also the equivalent of food
prohibitions are found in caste systems but, significantly enough,
they are principally expressed in the reverse form of 'endo-cuisine'
and moreover they occur on the level of the preparation, rather
than the production of food, in other words, on the cultural plane.
They are precise and detailed but mainly with respect to culinary
operations and utensils.
Finally, women are naturally interchangeable (from the point of
view of their anatomical structure and the physiological fonctions)
and in their case culture finds the field open for the great game of
differentiation (whether this is thought of in a positive or a negative
way and used therefore as a basis for exogamy on the one hand or
endogamy on the other). Foods however are not altogether able to
be substituted for each other. The game reaches its limits more
quickly in this second domain for one is much Jess indined to class
ail foods as totemic since, as we have seen, it is harder to do without
turmeric than knrra. i\ow, this applies even more to occupational
fonctions. Because they really are different and complementary,
they allow the establishment of reciprocity in its truest form. On
the other hand, they exclude negative reciprocity and so set bounds
to the logical harmony of caste systems. Each caste remains partly
'endo-fonctional'; it cannot forbid rendering also to itself the
differentiating services it is called on first of ail to provide for the
other castes, since these have been ruled to be irreplaceable. Or
who would shave the barber?
Introducing (socially) instituted diversity into a single natural
species, the human species, is not therefore the same as projecting
the diversity (naturally) existing between animal and plant
126
TOTEM AND CASTE
species on to the social plane. Societies with totemic groups and

exogamous sections in vain believe that they manage to play the
same game with species \vhich are different, and women who are
identical. They do not notice that since women are identical, it
falls to the social will to make them different, while species being
different, no one can make them identical, in the sense of all sub-
jects in the same way to human will. Men produce other men, they
do not produce ostriches.
~evertheless it remains true that we canon a very general plane
perceive an equivalence between the two main systems of diff-
erences to which men ha\·e had recourse for conceptualizing their
social relations. Simplifying a great deal, it may be said that castes
picture themselves as natural species while totemic groups picture
natural species as castes. And this must be refined: castes natura-
lize a true culture falsely, totemic groups culturalize a false nature
truly.
ln both views it must be granted that the system of social
fonctions corresponds to the system of natural species, the world of
living creatures to the world of objects, and we must therefore
recognize the system of natural species and that of manufactured
objects as two mediating sets which man employs to overcome the
opposition between nature and culture and think of them as a
whole. But there is also another means.
Several hunting tribes in North American say that at the begin-
ning of time buffaloes were ferocious beasts and 'all bone'. They
were not only inedible to man but also cannibal. Men were thus
once the food of the animal which later came to be their prime
food but which was at that time the reverse of a food since it was
animal food in its inedible form: bone. How is so complete a
change to be explained?
It came to pass, according to the myth, that a buffalo fell in love
with a girl and wanted to marry her. This girl was the only member
of her sex in a community of men, for a man had conceived her after
being pricked by a thorny plant. The woman thus appears to be
the product of a negative union, between nature hostile to man (the
bush of thorns) and human antinature (the pregnant man). ln spite
of their affection for their daughter and their fear of the buffalo,
men thought it wise to agree to the marriage and they collected
together presents, each of which was to stand for a part of the
buffalo's body : a war-bonnet was to become the backbone, a
127
THE SAV AGE MIND
quiver of otter-skin the skin on its chest, a woven blanket the

paunch, a pointed quiver the stomach, moccasins the kidneys, a
bow the ribs, etc. Nearly forty correspondences are enumerated in
this way (cf. Dorsey and Kroeber, no. 8 r, for a version of this myth).
The marriage exchange thus fonctions as a mechanism serving
to mediate between nature and culture, which were originally
regarded as separate. By substituting a cultural architectonie for a
supernatural primitive one, the alliance creates a second nature
over which man has a hold, that is a mediatized nature. After these
occurrences buffaloes became 'ail flesh' instead of 'ail bone' and
edible instead of cannibal.
The same sequence is sometimes reversed as in the Navaho
myth which ends in the transformation of a woman into a cannibal
she-bear: exactly the converse of a cannibal buffalo being trans-
formed into a husband. The metamorphosis is extended in a
scattered pattern which follows the mode! of the differences
between wild species: the vagina of the ogress turned into a
hedgehog, her breasts into pinyon nuts and acorns, her paunch
into other seeds ('alkali': sporobolus cryptandrus, airoides, Torr.),
the trachea turned into a medicinal plant and the kidneys into
mushro0ms, etc. (Haile-Wheelwright, p. 83).
These myths are an admirable expression of the way in which
marriage exchanges can furnish a model directly applicable to the
mediation between nature and culture among peoples where to-
temic classifications and functional specialization, if present at ail,
have only a very liP1ited yield. This confirms what I suggested
above, namely, first that the 'system of women' is, as it were, a
middle term between the system of (natural) living creatures and
the system of (manufactured) objects and secondly that each
system is apprehended as a transformation within a single group.
The system of living creatures is the only one of the three
systems which has an objective existence outside man and that of
fonctions the only one which has a completely social existence,
that is, within man. But the completeness of each on one plane
explains why neither is readily handled on the other: a food in
general use cannot be wholly 'totemized' at least not without
a kind of cheating* and, equally, castes cannot avoid being
• \Ve read the following about the 'clan divinitics' of the Dinka-which the
oldcr authors would ha\'c had no hesitation in calling totems:' ... few are of any
dietetic importance, and whcrc thcy are the respect paid to them may yet permit
128
TOTE~ AND CASTE
endofunctional while they serve to construct a grandiose scheme

ofreciprocity. Reciprocityis not therefore absolute in eithercase. It
is, as it were, blurred and distorted at the periphery. Logically
speaking, the reciprocity of marriage exchanges represents an
equally impure form since it lies mid-way between a natural and a
cultural mode!. But it is this hybrid character which allows it to
fonction perfectly. Associated with one or the other form, or with
both, or present on its own, as the case may be, the reciprocity of
marriage exchanges alone can daim universality.
The first conclusion \\·hich emerges from this analysis is that

totemism, which has been rendcred amply formai in 'primitive
language', could at the cost of a very simple transformation equally
well be expressed in the language of the regime of castes which is
quite the reverse of primitive. This is already sufficient to show
that we are here dealing not with an autonomous institution, which
can be defined by its distinctive properties and is typical of certain
regions of the world and certain forms of civilization but with a
mfJdus operandi which can hc discerned cven behind social struc-
tures traditionally defined in a way diametrically opposed to
totemism.
Secondly, we are in a better position to resolve the difficulty
arising from the fact that so-called totemic institutions include not
only the conceptual systems we have chosen to consider but also
rules of action. For I have tricd to show that food prohibitions
are not a distinctive feature of totemism ; they are also found
associated with other systems which they similarly serve to 'stress'
and conversely systems 'of names deriving from the natural king-
doms are not always accompanied by food prohibitions: they can
be 'stressed' in diverse fashions.
Further, exogamy and food prohibitions are not objects distinct
from the nature of society, which should be studied separately or
between which causal relations could be discovered. As language
shows almost anywhere, they are two aspects or two modes serving
to give concrete expression to a praxis which as a social activity can
be turned outwards or inwards and which always has these two
orientations even although they manifest themselves on different
them to be eaten'. Thus the clan of the Giraffe consider that they can eat the
meat of this animal provided only that they do not shed its blood (Lienhardt,
pp. I 14-15).
129
THE SAVAGE MIND
planes and by means of diffcrent codes. If the relation between

totemic institutions and castes can be regarded as superficially
identical to one between exogamy and endogamy (we have seen
that things are in fact more complex), between species and fonc-
tion, and finally between a natural and cultural mode!, it is
because a similar scheme emerges in ail the empirically observ-
able and apparently heterogeneous cases and it is this which
furnishes scientific investigation with its true subject of study.
There is an analogy between sexual relations and eating in ail
societies, but either the man or the woman may occupy the
position of eater or eaten according to the case and the level
of thought. This can but indicate the common requirement that
terms should be differentiated from each other and each identified
unequivocally.
Here again I do not mean to suggest that social !if e, the relations
between man and nature, are a projection or even result, of a con-
ceptual game taking place in the mind. 'Ideas', Balzac wrote, 'form
a complete system within us, comparable to one of the natural
kingdoms, a sort of bloom whose iconography will be traced by a
man of genius who will pass perhaps as mad.'* But more madness
than genius would be required for such an enterprise. If, as I have
said, the conceptual scheme governs and defines practices, it is
because these, which the ethnologist studies as discrete realities
placed in time and space and distinctive in their particular modes
of life and forms of civilization, are not to be confused with praxis
which- and here at least I agree with Sartre (p. r 8 r )- constitutes the
fondamental totality for the sciences of man. l\Iarxism, if not Marx
himself, has too commonly reasoned as though practices followed
directly from praxis. \Vithout questioning the undoubted primacy
of infrastructures, I believe that there is always a mediator between
praxis and practices, namely the conceptual scheme by the opera-
tion of which matter and form, neither with any independent
existence, are realized as structures, that is as entities which are
both empirical and intelligible. It is to this theory of superstruc-
tures, scarcely touched on by ~Iarx, that I hope to make a
contribution. The development of the study of infrastructures
proper is a task which must be left to history - with the aid of
demography, technology, historical geography and ethnography.
•II. De Balzac, 'Louis Lambert' in: Oeunes complètes, Bibl. <le la Pleiadc,
Vol X, p. 396.
130
TOTEM AND CASTE
It is not principally the ethno!ogist's concem, for ethnology is first

of all psychology.
All that I claim to have shown so far is, therefore, that the
dialectic of superstructures, like that of language, consists in set-
ting up constitutive units (which, for this purpose, have to be
defined unequirncally, that is by contrasting them in pairs) so as
to be able by means of them to elaborate a system which plays the
part of a synthesizing operator between ideas and facts, thereby
turning the latter into signs. The mind thus passes from empirical
diversity to conceptual simplicity and then from conceptual
simplicity to meaningful synthesis.
The most appropria te conclusion to this chapter is an illustration
of this idea by a native theory. The Yoruba myth, a veritable
Totem and Taboo before the fact, takes the complex edifice of
denomination and prohibition to pieces bit by bit.
What is in question is the explanation of the following rules. On
the third day after a child is born, a pricst is called in to give the
child 'its Orisha and its ewaws'. The Orisha is the creature or thing
which it worships, and the child may not marry anyone who has
the same Orisha. This creature or thing becomes the principal
ewaw of the person in question who passes it on to his descendants
for four generations. His son takes as his second ewaw his father's
wife's animal ewaw. The son of this son in turn takes his father's
wife's third or vegetable ewaw. And the son of the son of this son
takes the same relative's fourth ewaw, i.e. a rat, bird or snake.
ln native thought these complicated rules are based on an
original division of people into six groups: that of the fisherman;
that of 'omens': fish, snake and bird; that of the hunter; that
of quadmpeds; that of the farmer; that of plants. Each group
comprises both men and women, giving twelve categories
altogether.
To begin with unions were incestuous within each group,
brother marrying sis ter. The same Yoruba term is used for marr-
iage, meal, ownership, merit, gain and eamings or winnings.
Marrying and eating are one and the same thing. Using A and B
to represent a brother and sister of the first group, C and D a
brother and sister of the second group and so on, the initially
incestuous position can be summarized in the table:
1 2 3 4 5 6
AB CD EF GH IJ KL
THE SAV1\GE .MJNIJ
II uma n bcings however soon got tired of this monotonous 'di et',
so the son of the couple AB took the female product of CD and
so on for EF and G Il etc.:
ABD CDB EFII GIIF IJL KLJ
En:n thcn they were not satisfied and so the fisherman made war
<'n the hunter, the hunter on the farmer and the farmer on the
fisherman, and each appropriated the other's product. The result
was that from then on the fisherman ate flesh, the hunter the
products of the soil and the farmer fish:
ABDF CDBH EFHJ GHFL IJLB KLJD
By way of reprisai, the fisherman demanded the products of the

soil, the farmer flesh and the hunter fish:
ABDFJ CDRIIL EFHJB GHFLD LJLBF KLJDH
Things could not go on like this, so a great palaver was called

and the families agreed that they would give their daughters in
marriage to one another; and charged the priests to prevent con-
fusion and disorder by the rule that a \vife continues to worship her
own Orisha after marriage but her children do not inherit it. The
Orisha represented by the second letter in each series (viz. BD
FHJL) thus drop out in the next generation and the systems of
ewau:s becomes:
ADFJ CBHL EHJB GFLD ILBF KJDH
In future each person's ewriws were to consist of one Orisha, one

'omen', one animal and one plant. Each ewaw would continue in
the family line for four generations, after which the priest would
renew it. So A C E G I K now drop out and a male Orisha is
needed to complete the ewaws. A person whose index is ADF J
(group 1) can marry a child of group 2 whose ewaws are ail
different. A and C are therefore permutable and similarly E and G,
I and K:
DFJC BIILA llJBG FLDE LBFK JDHI
ln the next generation the letters D B H F L J drop out. Group

1 needs fish and takes B; group 2 also needs fish and takes D;
group 3 needs meat and takes F; group 4 also needs meat and
132
TOTE:\I AND CASTE
takes H; group 5 needs a plant and takes J; group 6 needs a

plant and takes 1:
F]CB HLAD JBGF '"LDEH BFKJ DHIL
The letters F H J L B D now drop out in turn. Being short of
meat groups I and 2 marry H and F respectively; short of plants,
groups 3 and 4 marry L and J; short of fish, groups 5 and 6 marry
D and B:
JCBII LADF BGFL DEIIJ FK]D HILB
J L B D F l [ drop out and the male Orishas corne to the fore
agam:
CBIIL ADF] GFLD EIIJB KJDII ILBF
As there are said to be two hundred and one Orishas of which
about half are male, and also a considcrablc number of omens,
animais and plants which arc uscd for dcsignating impediments to
maniage, the numher of possible combinations is very high
(Dennett, pp. Ij6-8o).
~o doubt, what we have here is just a theory in the form of a
fable. The author who recorded it mentions various facts which
seem, if not to contradict it, at least to suggest that things did not
fonction with this perfect regularity in his day. But, as theories go,
the Yoruba secm to have becn able to throw more light than
ethnologists on the spirit of institutions and rules which in their
society, as in many others, are of an intellecutal and delibcrate
character.* Sensible images undoubtedly corne in, but they do so
as symbols: they are counters in a game of combinations which
consists in permuting them according to rules without ever !osing
sight of the empirical significants for which, provisionally, they
stand.
•The example of the Ashanti among whom a boy inherits his father's, and a
girl her mothcr's, food prohibitions, equally suggests that the spirit of such
systems is 'logical' rather than 'genealogical'.
1 33
CHAPTER FIVE
CATEGORIES, ELEMENTS, SPECIES,

NUMBERS
In inquiring into the nature of mythical thought Boas came to the

conclusion in 191+ that the 'essential problem' was to know why
'human tales are preferably attached to animais, celestial bodies,
and other personified phenomena of nature'. (Boas 5, p. 490).
This problem is in fact the last remnant of the speculations about
totemism but it seems possible to solve it.
I have already tried to show that the heterogeneous beliefs and
customs arbitrarily collected together un der the heading of totem-
ism do not rest on the idea of a relationship of substance between
one or more social groups and one or more natural domains. They
are allied to other beliefs and practices, directly or indirectly linked
to classificatory schemes which allow the natural and social universe
to be grasped as an organized whole. The only distinctions which
could be introduced between ail these schemes derive from pref-
erences, w1'ich are never exclusive, for this or th<:lt level of
classification.
Ali the levels of classification in fact have a common characteris-
tic: whichever, in the society under consideration, is put first it
must authorize - or even imply- possible recourse to other levels,
formally analogous to the favoured one and differing from it only
in their relative position within a whole system of reference which
operates by means of a pair of contrasts: between general and
particular on the one hand, and nature and culture on the other.
The mistake which the upholders of totemism made was arbi-
trarily to isolate one level of classification, namely that constituted
by reference to natural species, and to give it the status of an
135
THE SAVAGE MIND
institution, when like all levels of classification it i sin fact only one
among others and there is no reason to regard it as more important
than, say, the level operating by means of abstract categories or
that using nominal classes. What is significant is not so much the
prescnce - or absence - of this or that level of classification as the
existence of a classification with, as it were, an adjustable thread
which givcs the group adopting it the means of 'focusing' on ail
planes, from the most abstract to the most concrete, the most
cultural to the most natural, without changing its intellectual
instrument.
In the work referred to above, Boas was doubtful whether the
predilection which is so frequent for classification based on a
natural mode! could be adequately explained by the 'distinctness
and individualization of species of animais ... [which] set them
off more clearly as characters of a tale than the undifferentiated
members of mankind' (!oc. cit.). Boas did, however, touch on an
important truth here. To have recognized it he need only, contrary
to a commonly held position, have been prepared, instead of
reducing the story or myth to a mere narrative, to try to discover
the scheme of discontinuous oppositions governing its organization
behind the mythical 'discourse'. Furthermore, the natural 'dis-
tinctiveness' of biological species does not furnish thought with a
means of access to other distinctive systems which have their own
repercussions on it. Ali things considered, if it is the case that zoo-
logical and botanical typologies are employed more often and more
readily than other typologies, this can only be by reason of their
intermediate position as logically equidistant from the extreme
forms of classification: categorical and singular. There is a balance
bctween the point of view of extension and that of comprehension
in the notion of a species: considered in isolation, a species is a
collection of individuals; in relation to other species, however, it is
a system of definitions. Moreover each of these individuals, the
theoretically unlimited collection of which makes up the species, is
indefinable in extension since it forms an organism which is a
system of functions. The notion of species thus possesses an in-
ternai dynamic: bcing a collection poised between two systems, the
species is the operator which allows (and even makes obligatory)
the passage from the unity of a multiplicity to the diversity of a
unity.
CATEGORIES, ELEMENTS, SPECIES, NUMBERS
As I have tried to show elsewhere (6, p. 133 ff.), Bergson dis-

cerned the importance of the part which, in view of its logical struc-
(ure, the notion of species was to.-play in the critique of totemism.
But there is every reason to fear that, had he been forced to make
his analysis more specific, he would have restricted it to the sub-
jective and practical aspect of the relation between man and the
natural world, of the kind exemplified by someone asking what
there is for lunch today and being wholly satisfied with the answer
'veal'. ln fact the importance of the notion of species is to be
explained not so much by a propensity on the part of the practising
agent to dissolve it into a genus for biological and utilitarian reasons
(which would amount to extending to man the famous dictum that
it is grass in general which attracts the herbivore)* as by its pre-
sumptive objectivity: the diversity of species furnishes man with
the most intuitive picture at his disposai and constitutes the most
direct manifestation he can perceive of the ultimate discontinuity
of reality. It is the sensible expression of an objective coding.
I t is a striking fact that modern biology turns to schemata
resembling those used in Communication Theory to explain the
diversity of species. \Ve cannot here embark on problems outside
the ethnologist's province. But if it were true, as biologists hold,
that the anatomical, physiological, and ethnological diversity of
some two million living species may be analysed in terms of varia-
tions of the chromosomes which are reducible to a periodicity in
the distribution of four distinct groups on the molecular chain,
then we could perhaps grasp the deeper reason for the special
significance man has seen in the notion of species. W e should
understand how this idea can furnish a mode of sensory appre-
hension of a combination objectively given in nature, and that the
:ictivity of the mind, and social life itself, do no more than borrow
it to apply it to the creation of new taxonomies. This would serve
to resolve the apparent mystery of the fascination which the notion
of species has always possessed for men, of which the obscure
• \Vhich indeed is as false in the case of animais as of man. The attempts to

establish national parks in Africa to preserve species threatened with extinction
cornes up against the difficulty that even if the pasturage is of sufficient area, the
animais onl)· use it as a home-base and go far beyond the limits of the reserve in
search of grasses richer in protein than the pastures allotted to them for the
simplistic reason that they are sufficiently extensive (Grzimek, p. 20). It is thus
not grass but the difference between species of grass i.vhich interests the
herbivore ...
1 37
THE SAVAGE MIND
fascination which totemism has had for ethnologists would just

constitute a particular case.
The natural sciences for a long time regarded themselves as con-
cerned with 'kingdoms', that is, independent and sovereign
domains each definable by its own characteristics and peopled by
creatures or objects standing in special relations to one another. This
view, which is now out-moded but still the 'commonsense' one,
could not but obliterate the logical power and the dynamism of the
notion of species, since in this light species appeared as inert and
separate classes, confined within the limits of their respective
'kingdoms'. The societies which we call primitive do not have any
conception of a sharp division between the various levels of classi-
fication. They represent them as stages or moments in a continuous
transition.
The Hanunoo of the Southern Philippines divide the universe
into what can and what cannot be named. \Vhat can be named is
distinguished into things on the one hand and persons or animais
on the other. When a Hanunoo utters the '.vord 'plant' he excludes
the possibility that the thing of which he is speaking should be a
rock or a manufactured object. The class of 'herbaceous plant' in
turn excludes other classes of plants such as 'woody plant', etc.
Among herbaceous plants, the term 'pepper plant' is differential
in relation to 'rice plant', etc. 'Houseyard pepper' excludes 'wild
pepper', and 'houseyard chi le pepper' excludes 'houseyard green
pepper'; finally, 'cat-penis' specifies the individual in question as
not being a member of the five other varieties or taxa which the
native culture distinguishes within the group of houseyard peppers
(Conklin, 4, p. 131).
This mode of operation, which can be represented by a series of
dichotomies, has been described in the following way:
\Vhile not to be confused, nor categorically equated, with the botanical
concept of species, the Hanun6o plant type does exhibit one similar
feature: barring mistaken identification, plant type narnes are mutually
exclusive ... Each of the l,625 Hanun6o plant types• has a full specific
narne diff ering in at least one component from al! others ...• Full plant
narnes are made up of from one to five, free, full-word lexical units. The
most cornmon form is a binomial combination ... Similarities between
• Of which 500 to 600 are only edible (I.e., p. 14 l) and 406 of purely medicinal
use (l.c., p. r88). These l ,625 types, grouped into 890 categories by indigenous
thought, correspond to 650 genera and about l,100 species in scientific botany
... (I.e., pp. 122-3).
Hanunoo and botanical classification decrease rapidly as we approach the

higher-level, more inclusive categories (Conklin I, pp. l 15-17 and
p. 162).
In fact the classes which correspond to Linnaean categories (pepper

plant: Capsicum sp., houseyard pepper: Capsicum annum L., wild
pepper: Capsicum frutescens L.) do not belong either to the same
level or to the same side of the system of dichotomies. And above
all the domain of scientific botany is eut off neither from the
popular botany practised by the gardener or housewife nor from
philosophers' and logicians' categories. It is mid-way between them
and makes it possible to pass from one to another and to concep-
tualize each level with the help of a code borrowed from another
(cf. diagram [p. 141 ]).
The Subanun, another Philippine tribe, classify diseases on the
same principle. They begin by distinguishing skin diseases from
wounds. They subdivide skin discases into 'inflammations', 'sores'
and 'ringworm'. Each of these three forms is further specified by
means of several binary oppositions: simple/multiple, hidden/
exposed, severe / mild, shallow / deep, distal/ proximal (Frake,
p. 129).
The documentation adduced in Chapters 1 and 2 in conjunction
with these examples shows how commonly zoological and botanical
classifications do not constitute scparatc domains but form an
integral part of an all-embracing dynamic taxonomy the unity of
which is assured by the pcrfcct homogcncity of its structure, con-
sisting as it does of successive dichotomies. One consequence of
this feature is that it is always possible to pass from species to cate-
gory. Again there is no inconsistency between the system (in evidence
at the top) and the lexicon whose role becomes progressively more
dominant as one descends the ladder of dichotomies. The problem
of the relation between continuous and discontinuous receives a solu-
tion in terms of origin since the universe is represented as a
continuum made up of successive oppositions.
This continuity is already apparent in the schemewhich governs
the liturgy of seasonal rites among the Pawnee lndians: the posts of
the lodge where the celebration takes place are chosen, according to
their orientation, from four species of trees painted in different
colours, which themselves correspond to directions symbolizing the
seasons which together make up the year:
139
THE SAVAGE MIND
poplar white south-west

1 south summer
i.i box-eider red south-east i.i
(.)
year ~
'!- ;::
;;; elm black north-east
1 north winter
willow ycllow north-wcst
The same explicit passage from species or group of species to a

system of propertics or categories can be illustrated by Melanesian
examples. \Ve have already seen that at .Mawatta, an island
of the Torres Straits, clans with animal names are grouped accord-
ing to species into land or sea, warlike or pacifie. Among the Kiwai
an opposition between the sago people and the yam people is
expressed by means of two emblems: that of the nude woman and
that of the bull-roarer, also called 'the mother of yams', and it also
corresponds to the alternation of the seasons and of the prevailing
winds. In the Trobriand Islands there is a correspondence, in the
case of each clan, between a bird, an animal, a fish, and a plant. The
binary system of the Solomons appeals either to two birds (wild cock
and hornbill) or to two insects (pharma and mantis) or to two div-
inities who are, however, the incarnation of antithetical modes of
beha,·iour (\'Ir Wise and :\fr Clumsy) (Frazer, vol. II, passim).
We can thus see that, depending on the code chosen, the logical
rigour of the oppositions can be unequally manifested without
thereby implying any difference of kind. The classificatory systems
of the Sioux provide a good example, for they constitute so many
variations on a common theme. Only the semantic level adopted to
signify the system changes.
Ali the tribes have circular camps which an imaginary diameter
divides into two moieties. But in several of them this apparent
dualism conceals a principle of tripartition, the symbolic substance
of which varies from one tribe to another. Among the \Vinnebago
there are twice as man y clans in one moiety as in the other ( eight and
four respectively). The ten Omaha clans are distributed equally
between the two moieties, but one has two chiefs and the other only
one. There are seven clans in each Osage moiety but one moiety
is divided into two sub-moieties white the other is homogeneous.
ln ail three cases and in whichever way the opposition is realized it
q.o
ENTITY (somcthing which can be
1
name<l)
1
THING (not a person or animal)
1
1
PLANT (not a rock, etc.)
1
1
HERBACEOUS PLANT (not a woody plant, etc.)
1
1
PEPPER PLANT (not a rice plant, etc.)
(Capsicum sp.)
1
1
HOUSEYARD (not a ·wild pepper plant, ~
Capsicum f rutescens L., ~
PEPPER PLANT
(Capsicum annum L.) etc.) -
tTl
r:n
1
z
1 c::
HOUSEYARD CHI LLI (not a houseyard green ::
PEPPER PLANT pepper plant, etc.) l:t!
tTl
1
:>:!
r:n
1
'CAT-PENIS' (not one of the five other
terminal taxa)
(Conklin 4, p. 13 1)
THE SAVAGE MIND
is the simple form which exemplifies the higher or sky moiety

and the complex form which exemplifies the lower or land
moiety.
On the other hand, and to stick to the system of moieties, the
opposition high/low, although implicit in ail the groups, is not
always the one explicitly formulated. One finds it indicated in
diverse ways which may be exclusively present or juxtaposed: sky/
land, thunder /land, day /night, summer /winter, right/left, west/east,
male/female, peace/war, peace-war/policing-hunting, religious
activities/political activities, creation/conservation, stability/move-
ment, sacred/profane ... Finally, according to the group (or within
the same group according to the circumstances) it is sometimes the
binary and sometimes the ternary aspect which is put to the fore.
Sorne, like the Winnebago, compound them into a quinary system,
while the Ponca decompose the dualist structure into a system of
two pairs: land and sea, fire and wind.
Similarly, among the Algonkin the apparently non-significant
multiplicity of the forty or fifty Ojibwa clans, which do, however,
allow themselves to be grouped into animal, fish, and bird clans, can
be traced back to the more explicit Mohican scheme (whose clans
are divided into three phratries, one consisting of the Wolf, Bear,
Dog, and Oppossum clans, another of the Little Turtle, the Great
Turtle, the M ud Turtle, and the Yellow Eel clans, and third of the
Turkey, the Crane, and the Chicken clans) and then to the Delaware
scheme, which is simplified in the extreme and whose logic is
immediately apparent, since there are only three groups, the Wolf,
T!utle, and Turkey, c\early corresponding to land, sea, and air
respectively (Frazer, vol. III, p. 44 ff.).
The vast corpus of rites of the Osage, assembled ând publishecJ
by La Flesche, to which I referred earlier (pp. 59-60), provides a
wealth of illustrations, which are sometimes demonstrations, of the
mutual convertibility of 'concrete classifiers': animais and plants,
and 'abstract classifiers' such as numbers, directions, and the
cardinal points. Thus, for example, bows and arrows figure in the
list of clan names but it is not merely manufactured objects which
are here in question. It is apparent from texts of prayers and invoca-
tions that one arrow is painted black and the other red and that this
opposition of colours corresponds to that of day and night. The same
symbolism recurs in the colours of the bow, red on the inside and
black on the outside: shooting with the red and black bow, using
142
alternatively a red and a black arrow, is an expression of Time, itself

measured by the alternation of day and night (cf. La Flesche, 2, p. 9,
and 3, pp. 207, 233, 364-5).
The concrete classifiers not only serve to convey ideas; they can
also, in their sensory form, show that a logical problem has been
solved or a contradiction surmounted. Among the Osage, when a
pair of moccasins are made, their owner must perform a complex
rite. This special attention accorded an article of clothing might
seem surprising did the texts not reveal in the moccasin something
other than its utilitarian fonction: the moccasin, a cultural object,
is opposed to 'evil' grass which the walker tramples and crushes. lt
th us corresponds to a warrior who tramp les his enemies underfoot.
Now, in the sociocosmological scheme of the Osage, the martial
fonction connotes the land moiety, to which grass also belongs. The
particular symbolism of the moccasin is thus inconsistent with the
general symbolism, since the moccasin is 'anti-land' in the former
and congruent with land in the latter. The minutia of the ritual is
clarified by the evidence of what one would like to call the logical
instability of a manufactured article: an instability which a highly
ritualized manufacturing technique precisely serves to palliate (cf.
I.e. 3, pp. 6r-7).
In Osage thought, the most important opposition, which is also
the simplest and has the greatest logical power, is that between the
two moieties: Tsi'-zhu: sky, and Hon'-ga ;subdivided intoHon'-ga,
properly speaking: dry land, and Wa-zha'-zhe: water. Starting from
here a complex grammar is developed by means of a system of
correspondences with more concrete or more abstract domains but
within which the original scheme, acting as a catalyst, initiates the
crystallization of other schemes of two, three, four, or more vari-
ables. First the cardinal points, since, in the initiating hut, sky and
land are opposed as north and south, and dry land and water as east
and west respectively.
Secondly, the opposition of odd and even gives rise to a mystic
numerology. As 1 indicated in Chapter 4. the number six belongs
to the sky moiety, the number seven to the land moiety, and their
sum, thirteen, corresponds, on the cosmological plane, to the
number of rays of the rising sun (which is a demi-sun) and, on the
social plane, to the notable actions which may be counted to his
credit by an accomplished warrior (who is a demi-man, since the
1 43
THE SAVAGE MIND
functicn of war is the prerogative of one of the two moieties which

together constitute the tribe).'*'
Thus the quality and the unity of the two great divisions of the tribe
might be symbolized as a man or an animal, but the Hon'-ga great
division must always represent the right sicle of the man or animal and the
Tsi' -zhu the great division of the left. This idea of the duality and unity
of nature was not only reflected in the tribal organization but, in former
times, instilled in the minds of the people by certain persona! habits, as
for instance members of the Hon' -ga great division when putting on their
moccasins put the moccasin on the right foot first, while members of the
Tsi' -zhu great division put the moccasin on the left foot first (La Flesche
3, p. I I 5),
I wanthere to digress fora moment to point out that this meticulous
rigour in the practical application of a logical system is not an
ir.olated phenomenon. At Hawaii the death of a chief was marked
by violent manifestations of mourning. The participants wore their
loin cloths around their neck instead of loins. This vestimentary
inversion of high and low was accompanied by (and no doubt also
signified) sexual licence. The importance of the opposition between
high and low was expressed in a large number of prohibitions. A
receptacle containing food must not be covered by any object which
may have been walked on or sat on; it was forbidden to sit on a
pillow or to use it as a footrest, to lay your head on a seat cushion, to
sit above anything containing food, or, for menstruating women,
to use as pads anything but material from old skirts worn below the
waist:
The old time Hawaiians used to talk often, when I was a small child, of
the terrible custom the Whites had of using a sheet sometimes to lie upon
and sometimes to lie under - they (the Whites) did not seem to know that
what belonged above (ma luna) should remain above and what belonged
below (ma lalo) should stay below ...
ln a hula school conducted by my cousin Ilala-ole-o-Ka'ahumanu,
one of the pupils thoughtlessly draped her skirt over her shoulder. The
hula master spoke sharply to her, saying 'What belongs above should stay
above, and what belongs below should stay below' (Ko luna, no luna no
ia; Ko lalo no lalo no ia). (Handy and Pukui, Part VI, p. 165, and Part
VII, pp. 316, 317).
Recent studies (Needham 3, Beidelman) show the refinement with
which the African tribes of Kenya and Tanganyika exploit the, to
them, fondamental opposition between right and left (usually
• The responsibility for this analysis, which is not given in the texts, is my
own.
apparently the right and left hands rather than feet, but we have
already noticed the particular attention which the Osage pays to
the lower extremities). A Kaguru man uses his lef t hand for making
love and a woman her right hand (that is, the hands which are
impure for each sex respectively). The first payment which has to
be made to a healer, before treatment can begin, is made with the
right hand, the last with the left, etc. The nomadic Fulani of the
Sahel zone of the Niger, the Bororo of Africa, seem, like the Kaguru,
to associate the right side with men and - in the temporal order -
with what cornes first, the lef t si de with women and what cornes
after.• Symmetrically, the masculine hierarchy goes from south to
north and the feminine from north to south, so that, in the camp,
the woman places her calabashes in order of decreasing sizc, with
the largest to the south, while a man fastens his calves in the reverse
or der (Du pire).
To return to the Osage: the number thirteen, as we have seen, is
first of ail the sum of the two social groups, right and left, north and
south, winter and summer; thereafter it is specified concretely and
developed logically. In the image of the rising sun, in which the
beholder venerates the source of ail lifc (thus facing east, which
means that the south is on his right and the north on his left),t the
number thirteen can symbolize the union of two terms: six and
seven, sky and land, etc. But when it relates to a star the solar sym-
bolism is particularly attached to the sky moiety. Hence there corne
to be other concrete specifications of the number thirteen, in this
case reserved to sub-groups of the other moiety: thirteen foot-
prints of the black bear to represent the notable actions of the land
clans and thirteen willow trees to represent those of the water clans
(La Flesche 3, p. 147).
Thirteen is thus the expression of a double human totality: collec-
tive, since the tribe is made up of two asymmetrical moieties (quanti-
tatively: one is single, the other divided; and qualitatively: one in
• Cf. Diamond for an analogous spatio-temporal system in the same region.
t The officiant is painted red 'to express his craving that through the sun his
life may be made fruitful and that he may be blessed with a long line of descend-
ants'. \Vhen his whole body has been painted red, 'a dark line is drawn on his
face running upward from one cheek to the forehead, then across to the opposite
side and downward to the middle of the other cheek. This line represents the
dark horizon line of the earth and is called a "snare" or an enclosure into which
all life is drawn and held captive'. (La Flesche 3, p. 73).
145
THE SA VAGE MIND
charge of peace, the other of war) and individual but equally asym-
metrical (the right and the left). As a totality, this union of even and
odd, of collective and individual, social and organic, is geared to the
ternary cosmological schcme: there is a 'thirteen' of sky, a 'thirteen'
of land, a 'thirteen' of water. Finally, added to this coding by
elements, there is a coding by species where the two groups com-
posed respectively of seven and six 'animais' are duplicated by the
appearance of antagonists, thus (as might be foreseen) bringing the
number of units of the system at the most concrete level to twenty-
six. The seven animais and their antagonists are shown in the
following table :
ANIMALS ANT AGONISTS
lynx young male deer, with curved horns
grey wold young male deer, with grey horns
male puma full grown male deer, with dark homs
male black bear hummock full of little bugs (insects?)
buffalo bull high bank
elk a plant whose blossoms always look up to the sun
(Silphium laciniatum)
deer• no antagonist: his strength lies in fiight
The system of six animais is Jess neat. It includes two varieties of

owl, each opposed to a male raccoon, one young and one adult, the
golden eagle opposed to the turkey, finally, apparently, the river
mussel (the shell of which is used to make the mother of pearl pen-
dants which symbolize the sun), buffalo hair (?), and a little pipe
(?)(La Flesche 3, pp. 54-61).
A logical structure - initially a simple opposition - thus fans out
in two directions: one abstract, in the form of a numerology, and
the other concrete, first of elements and then of species. On each
level semantic short-circuits permit direct connections with the
level furthest away. The level of species, which is also the most par-
ticularized of those we have considered, does not, however, consti-
tute a sort of limit or stopping point of the system: the latter does
• Timidity of the deer is due toits having no gall-bladder. Its raie is twofold:
alimentary, its meat being regarded as the most regular source of animal food,
comparable from this point of view to the vegetable food provided by four
essential plants: Nelumbo lutea, Api os apios, Sagitarria latifolia, Falcata comosa.
The deer and these four plants are necessary to the very existence of the tribe
and the primary raie of the warrior is to defend that land where they are found
(I.e., pp. 129-30). The deer also has a cultural raie: its body is the source of the
sinews which women use for sewing and men for fastening feathers to their
arraws (I.e., p. 322).
not fall into inertia but continues to progress through new detotal-
izations and retotalizations which can take place on several planes.
Each clan possesses a 'symbol of life' - a totem or divinity-whose
name it adopts: puma, black bear, golden eagle, young deer, etc.
The clans are thus defined, in relation to each other, by means of
differentiating features. Nevertheless, the ritual texts found each
distinctive choice on a system of invariant characteristics, assumed
to be common to ail species: each affirms of itself what the puma, for
instance, declares on its own account:
'Behold the soles of my feet, that are black in colour.
I have made the skin of the soles of my feet to be as my charcoal.
\Vhen the little ones [men] also make the skin of the soles of my feet
to be as their charcoal,
They shall always have charcoal that will easily sink into their skin as
they travel the path of life.
Behold the tip of my nose, that is black in color, etc.
Behold the tips of my ears, that are black in color, etc.
Behold the tip of my tail, that is black in color, etc.'
(La Flesche 2, pp. 106-'7)
Each animal is thus decomposed into parts, according to a law of
correspondence (muzzle = beak, etc.) and the equivalent parts are
regrouped among themseh·es and then all together in terms of the
same relevant characteristic: the presence of 'charcoaled' parts, on
account of the protective raie which the Osage attribute to fire and
its product, charcoal, and finally, and by way of consequence, to
the colour black. The 'black thing', charcoal, is the abject of a
special rite which warriors have to perform before going into battle.
If they neglect to blacken their faces, they will Jose the right to
recount their notable actions and to claim military honours (La
Flesche 3, p. 327 seq.). We therefore already have a system on two
axes, one devoted to diversities and the other to similarities:
CHARCOAL ANIMAL
1 1
black paws black muzzle black tail etc.
~ s -puma-___,l,__~~~+-~~~-lc--~~~+1 ~~~~
a
u
:
s
-eagle -1
p -bear----+-------+------+------+-----
ec - deer---------1-----
~ -swan - -
--etc.---~------'-----~-----'-------
FI -----
1 47
TllF. SAVAGE MIND
The analytic procedure which makes i t possible to pass from cate-

gories to elements and from elements to species is thus extended by
a sort of imaginary dismembering of each species, which progress-
ively re-establishes the totality on another plane.
This double movement of detotalization and retotalization also
takes place on a diachronie plane, as is shown by the admirable
chants of the Bear and the Beaver (representing land and water
respectively). They are meditating on the coming winter and pre-
paring themselves for it in accordance with their particular habits
(here endowed with symbolic significance ), so that the coming of
spring and their restored strength can appear as the pledge of the
long life promised to men: When 'six moons had passed ... [the
bear] made a close examination of his body, looking carefully over
all its parts'. He enumerated the signs of his emaciation (that is, a
diminished body which, however, as he has remained alive, testifies
the more to the power of life: shrunken flesh, shrivelled toes,
wrinkled ankles, flabby abdomen, protruding ribs, flaccid arms,
sagging chin, lines on the corner of the eyes, bald forehead, scant
hair). He next made footprints, symbols of warlike deeds, six on
one sicle, seven on the other, and 'then the bear went forth with
quickenedfootsteps, and came to a land upon which theair quivered
with the warmth of the sun' (La Flesche 3, pp. 148-64).
The synchronie structure of the tribe, expressed in the division
into three elemental groups, themselves divided into clans bearing
totemic names, is indeed, as we have seen, * no more than a projec-
tion into the order of simultaneity of a temporal process which the
myths describe in terms of succession: when the first men appeared
on the earth (according to this version, from the sky; in another
version (Dorsey l) they came from the subterranean world), they
began to march in the order of their arrivai: first the water people,
then the land people, and last the sky people (La Flesche 2, pp.
59-6::>); but when they found the earth covered with water they
appealed first to the water-spider, then to the whirligig, next to the
white leech, and finally to the black letch to guide them to habitable
spots (id., pp. 162-5).
It may be seen, therefore, that in none of these cases can tht•
animal, the 'totem' or its species be grasped as a biological entity:
through its double character of organism - that is, of system - and
of emanation from a species-which is a termina system -the animal
• Cf. ahove, pp. 68-9.
CATEGORIES, ELEMENTS, SPEC!ES, NUMBERS
appears as a conceptual tool with multiple possibilities for detotal-

izing or retotalizing any domain, synchronie or diachronie, concrete
or abstract, natural or cultural.
Thus it is never, properly speaking, the eagle which the Osage
invoke. For, according to the time and circumstances, it is eagles of
differentspecies which are in question: the golden eagle, the spotted
eagle, the bald eagle, etc., or eagles of different colours: red, white,
spotted, etc.; or finally eagles at different stages of their life: young,
old, etc. This three-dimensional s,1trix, a genuine system by means
of a creature, and not the creature itself, constitutes the object of
thought and furnishes the conceptual tool. *' Were the image not so
trivial one would be tempted to compare this tool to a utensil with
crossed metal blades which is used for cutting potatoes into slices
or chips: a 'preconceived' grid is applied to ail the empirical situa-
tions with which it has sufficient affinities for the elements obtained
always to preserve certain general properties. The number of pieces
is not always the same nor is the form of each absolutely identical
but those which corne from the centre remain in the centre, those
which corne from the periphery on the periphery ...
As media! classifier (and therefore the one with the greatest yield
and the most frequently employed) the species level can widen its
net upwards, that is, in the direction of elements, categories, and
numbers, or contract downwards, in the direction of proper names.
This last aspect will be considered in detail in the next chapter. The
network to which this twofold movement gives rise is itself cross-eut
at every level, for there are a great many different manners in which
these levels and their ramifications can be signified: nomenclature,
differences of clothing, bodily paintings or tattoos, ways of being or
behaviour, privileges and prohibitions. Each system is therefore
defined with reference to two axes, one horizontal and one vertical,
which correspond up to a point with Saussure's distinction between
syntagmatic and associative relations. But 'totemic' thought, unlike
speech, has this in common with mythical and poetical thought
that, as Jakobson has established for the latter, the principle of
equivalence acts on both planes. The social group can code the
message without any alteration in its content by means of different
• 'We do not believe', as an Osage explained 'that our ancestors were really
animais, birds, etc., as told in traditions. These things are only wa-u:i-k11-ska'-ye
(symbols) of something higher' (J.O. Dorsey, 1, p. 396).
1 49
THE SAVAGE MIND
lexical elements: as a categoric opposition: high/low, or as an

elemental one: sky/earth, or again as a specific one: eagle/bear. And
equally it has the choice of several syntacticprocedures to assure the
transmission of the message : nomenclature, emblems, modes of
behaviour, prohibitions, etc., used either alone or together.•
\Vere the task not so immense, a classification of these classifica-
tions could be undertaken. Systems would be distinguished accord-
ing to the number of categories they employ - ranging from two to
several dozen - and according to the number and choice of elements
and dimensions. They would then be distinguished into macro- and
micro-classifications, the former being characterized by the
admission of a large number of animal and plant species to the
status of totems (the Aranda recognize more than four hundred) and
the latter by having totems all, as it were, inscribed within the limits
of the same species. The Banyoro and Bahima in Africa provide an
• Considered separately, in their component part& and their respective rela-

tions with their surroundings, a suburban villa and a stronghold are syntagmatic
sets: their elements are related to each other by contiguity: container and con-
tent, cause and effect, end and mean, etc. What, as a bricoleur, Mr Wemmick of
Great Expectations undertook and realized (cf. above, p. 17) was the establish-
ment of paradigmatic relations between the elements of these two chains: he can
choose between villa and castle to signify his abode, between pond and moat to
signify the piece of water, between flight of steps and drawbridge to signify the
entrance, between salad and food reserves to signify his lettuces. How has this
corne about?
lt is clear that to begin with his castle is a small-scale model, not of a real
castle but of a castle signified by camouflages and fittings which have the fonc-
tion of symbols. He has not indeed acquired a real castle through these trans-
formations, but he has well and truly lost a real villa, since his fantasy binds him
to a whole number of servitudes. lnstead of living as a bourgeois, his domestic
life becomes a succession of ritual actions, the minute repetition of which serves
to promote, as the sole reality, the paradigmatic relations between two equally
unreal syntagmatic chains: that of the castle which has never existed and that
of the villa which has been sacrificed. The first aspect of bricolage is thus to
construct a system of paradigms with the fragments of syntagmatic chains.
But the reverse is equally true. His old father's deafness !ends Mr \Vemmick's
castle a real value: a stronghold is normally providcd with cannons; and his
fathcr is so deaf that only the noise of a cannon can penetrate to him. The initial
syntagmatic chain, that of the suburban villa, is objectively broken by the
patemal infinnity. Father and son, its sole inhabitants, lived juxtaposed without
it being possible to establish any relation whatever between them. The villa need
only become a castle for the cannon, fircd daily at nine o'clock, to institute an
effectuai fonn of communication. A new syntagmatic chain thus results from
the system of paradigmatic relations. A practical problem is solved: that of com-
munication between the inhabitants of the villa, but thanks to a total reorganiza-
tion of the real and the imaginary, whereby metaphors take over the mission of
metonyms, and vice versa.
150
instance of this. Clans are named after particular types or parts of

cows: striped cow, brown cow, cow in full milk, etc., cow's tongue,
tripe, heart, kidneys, etc. The syst~ms may equally be distinguished
by the number of their dimensions. Sorne are purely animal, some
purely plant, others appeal to manufactured articles, and yet others
juxtapose a variable number of dimensions. They can be simple
(one name or one totem per clan) or multiple as in the case of the
Melanesian tribes who define each clan by a plurality of totems: a
bird, a tree, a mammal, and a fish. Finally, the systems can be
homogeneous like that of the Kavirondowhere the totemic lists are
composed of elements of the same type: crocodile, hyena, leopard,
baboon, vulture, crow, python, mongoose, frog, etc. Or they can be
heterogeneous, as the totemic lists of the Bateso illustra te: sheep,
sugar cane, boiled bon es of meat, mushroom, antelope ( common to
several clans), sight of the forbidden antelope, shaved skull, or
again, those of certain north-east Australian tribes: sexual passion,
adolescence, various diseases, named places, swimming, copulation,
the making of a spear, vomiting, various colours, various physical
states, heat, cold, corpse, ghost, various accessories of ri tuai, various
manufactured objects, sleep, diarrhoea, dysentery, etc.*
Such a classification of classifications is perfectly conceivable but
so many documents would have to be gone through and dimensions
of such variety taken into account to realize it, that even if we
confined ourselves to the societies for which the data are sufficiently
full, accurate, and comparable among themselves, it could not be
done without the aid of machines. Let us therefore content ourselves
with mcntioning this programme, reserved for the ethnology of a
future century, and return to the most simple properties of what 1
am, for convenience, calling the totemic operator. A description
with the help of a diagram, and considering only a small portion of
the ce!! {for we are making it begin at the species level and arbi-
trarily restricting the number of species, and also the number of
parts of the body, to three each), is sufficient to give an indication
of its complexity (figure 8).
It can be seen that the species admits first empirical realizations:
Sea! species, Bear species, Eagle species. Each includes a series of
• 'It appears that a totem may be any enduring element of the physical or
mental environment, either unique conceptual entities, or, more frequently,
classes or species of things, activities, states, or qualities which are constantly
recurring and are thus considered to be perdurable' (Sharp, p. 69).
THE SAVAGE MIND
SPECIES
0
INDIVIDUAL
Fig 8 The totemic operator.

individuals (also reduced to three in the diagram): seals, bears,

eagles. Each animal can be analysed into parts: head, neck, f eet, etc.
These can be regrouped first withi.n each species (seals' heads, seals'
necks, seals' feet) and then together by types of parts: ail heads, ail
necks ... A final regrouping restores the mode! of the individual in
his regained entirety.
The whole set thus constitutes a sort of conceptual apparatus
which filters unity through multiplicity, multiplicity through unity,
diversity through identity, and identity through diversity. Endowed
with a theoretically unlimited extension on its median level it
con tracts (or expands) into pure comprehension at its two extreme
vertices, but in symmetically reverse forms, and not without under-
going a sort of torsion.
The mode! which 1 have used here as an illustration clearly only
represents a minute fraction of the ideal mode!; since the number of
naturalspecies is in theorder of two million, the conceivable number
of individuals is unlimited, and the organs or parts of the body which
are distinguished and named rises to almost four hundred in some
indigenous vocabularies (Marsh and Laughlin). And there are prob-
ably no human societies which have not made a very extensive
inventory of their zoological and botanical environment and des-
cribed it in specific terms. Is it possible to estin:ate an order of
magnitude, or limits? ln going through ethnozoological and ethno-
botanical works, one notices that, with rare exception, the species
and varieties recorded seem to be in the order of several hundred,
around three hundred to six hundred. No work of this kind is
exhaustive, however, being limited by the time spent in collecting
the material, the number and competence of the informants, and
finally the fieldworker's own competence, the extent of his know-
ledge, and the variety of his preoccupations. One can therefore
hardly go wrong in putting the real figure considerably higher, and
the best works confirm this:
The Hanun6o classify their local plant world, at the lowest (terminal)
level of contrast, into more than 1800 mutually exclusive folk taxa, while
botanists divide the same ftora - in tenns of species - into less than 1300
scientific taxa (Conklin 4, p. 129).
This quotation from an ethographer specializing in taxonomy

curiously echoes a comment of Tylor's on the subject of Rabbinical
philosophy:
153
THE SAVAGE MIND
... which apportions to each of the 2100 species, of plants for in-
stance, a presiding angel in heaven, and assigns this as the motive of the
Levitical prohibition of mixtures among animals and plants (Tylor, ,-ol.
II, p. 246).
In the present state of knowledge, the figure of two thousand
appears to correspond well, in order of magnitude, to a sort of
threshold corresponding roughly to the capacity of memory and
power of definition of ethnozoologies or ethnobotanies reliant on
oral tradition. 1t would be interesting to know if this threshold has
any significant properties from the point of view of Information
Theory.
A recent student of initiation rites among the Senufo has brought

to light the role of fifty-eight figurines which are shown to novices
in a determinate order, and form, as it were, the canvas of the
instruction imparted to them. These figurines represent animais
or persons, or symbolize types of activities. Each therefore
corresponds to a species or to a class:
The elders present a certain number of objects to the neophytes ...
This inventory, sometimes very lengthy, constitutes a sort of lexicon of
symbols and the different possible modes of using them are indicated. In
the most developed poro the men leam in this way to handle the ideo-
graphical supports of a way of thought which manages to assume a truly
philosophical character (Bochet, p. 76).
In other words, in systems of this type there is a constant passage

in both directions between ideas and images, grammar and lexicon.
This phenomenon, to which I have drawn attention several times
already, raises a difficulty, namely, whether it is legitimate to
postulate, as it might be objected I have implicitly done, that these
systems are motivated at ail levels. Or, more precisely: are what
we have here genuine systems, where images and ideas, lexicon
and grammar are always united by strict and invariable relations,
or must we acknowledge that there is a certain measure of con-
tingence and arbitrariness at the most concrete level - that of
images and lexicon - which casts doubt on the systematic nature
of the whole? This problem arises wl:enever one claims to have
discovered a logic of clan names. As 1 showed in an earlier chapter,
one almost always cornes up against a difficulty which may at
first seem insurmountable. The societies which claim to form a
coherent and articulated system (whether the 'stress' of the
system is on names, modes of behaviour, or prohibitions) are also

collectivities of living beings. Even if, consciously or unconsciously
they apply rules of marriage whpse effect is to keep the social
structure and rate of reproduction constant, these mechanisms
never fonction perfectly; and they are also endangered by wars,
epidemics, and famines. It is thus plain that history and demo-
graphic development always upset the plans conceived by the
wise. In such societies there is a constantly repeated battle between
synchrony and diachrony from which it seems that diachrony must
emerge victorious every timc.
What this means in relation to the problem just raised is that
the nearer we get to concrete groups the more we must expect to
find arbitrary distinctions and denominations which are explicable
primarily in terms of occurrences and events and defy any logical
arrangement. 'Everything is a potential totem', it is said of tribes
of North-west Australia which already number such things as
'white fellow' and 'sailor' among their totems in spite of the recent
date of their first contact with civilization (Hernandez).
Certain tribes of Groote Eylandt, in the east of Arnhem Land,
are divided into two moieties comprising six clans. Each clan
possesses one or more miscellaneous totems: winds, ship, water,
animais and plant species, stones. The 'wind' totems are probably
connected with the annual visits of the Makassans and the same is
true of the 'ship' totem, as a myth referring to the manufacture of
ships by the Makassans in the island of Bickerton demonstrates.
Other totems are borrowings from natives of the interior. Sorne
are in the process of beiiig abandoned while others are recent
introductions.
Consequently, the author of these remarks concludes, it would
be imprudent to regard the choice and distribution of totems as an
attempt to fit various f eatures of the environment into the dualist
scheme: 'the list ... is in fact a product of historical accretion
rather than any attempt to systematize the natural environment
according to some philosophical scheme.' There are totemic songs
inspired by two known ships, the Cora and the Wanderer, and
even, since there was an air-base established on one clan's territory
during the War, bythe Catalina flying boats. Facts ofthis kind lead
us to think that some totems may have had their origin in historical
events, all the more so because in the language of the tribes in
question the same word is used to refer to totems, myths, and any
155
THE SAVAGE MIND
species of beautiful, unusual, or curious object, such as a particu-

larly attractive beauty-spot or a pretty little phial of medicine. The
aesthetic inspiration and individual invention, no Jess than the
events, speak in favour of contingence (Worsley).
I referred to the role of the aesthetic imagination in the elabora-
tion of classificatory systems several times in the first chapter of
this book. It is arole already recognized by theorists on taxonomy,
which, in Simpson's words, 'is also an art' (p. 227). There is
nothing disquieting about this aspect of the problem; quite the
contrary. But what of historical factors?
Linguists have been well aware of this problem for a long time
and Saussure resolved it very clearly. Saussure, who laid down the
principle (for which the evidence seems to us today much Jess
conclusive) of the arbitrary character of linguistic signs, himself
concedes that this arbitrariness admits of degrees and that the sign
may be relatively motivated. So much is this the case that languages
can be classified in terms of the relative motivation of their signs:
the Latin inimicus is more strongly motivated than the French
ennemi (which is Jess readily identified as the reverse of ami); and
signs are also differently motivated within a single language: the
French dix-neuf is motivated, while the French vingt is not. For
the word dix-neuj'suggests it own terms and other terms associated
with it'. If the irrational principle of the arbitrariness of the sign
were applied without restriction, it 'would lead to the worst sort
of complication ... But the mind contrives to introduce a principle
of order and regularity into certain parts of the mass of signs, and
this is the role of relative motivation'. In this sense we might say
that some languages are more lexicological and others more
grammatical:
Not because 'lexical' and 'arbitrary' on the one hand and 'granunar'
and 'relative motivation' on the other, are always synonymous, but
because they have a common principle. The two extremes are like poles
between which the whole system moves, two opposing currents which
share the movemen t of language: the tendency to use the lexicological
instrument (the unmotivated sign) and the preference given to the gram-
matical instrument (structural rules) (Saussure, pp. 1 JJ-4).
For Saussure, therefore, language moves from arbitrariness to

motivation. The systems we have been considering so far on the
other hand go from motivation to arbitrariness: conceptual
schemes (at the limit, simple binary opposition) are constantly
156
broken open to introduce elements taken from elsewhere; and

there is no doubt that these additions often entai! modification of
the system. ~Ioreover, they do, not always succeed in getting
incorporated in it and the systematic appearance is then disturbed
or temporarily put in abeyance.
The example of the nine hundred survivors of some thirty
Australian tribes which were haphazardly regrouped in a Govern-
ment Settlement provides a tragic illustration of this constant
struggle between history and system. The Settlement contained
(in 1934) about forty wooden cottages, supervised dormitories
segregated by sex, a school, a hospital, a prison, ships; and the
missionaries (unlike the natives) could enjoy themselves to their
hearts' content: in the course of a few months, visits were made
by ?\on-Ccnformists, Presbyterians, the Salvation Army, Church
of England, and Roman Catholics ...
I mention these facts not for polemical reasons but because
they make the maintenance of traditional bcliefs and customs seem
so highly improbable. The natives' first response to the regrouping
was, however, the adoption of a common terminology and of ru les
of correspondence for harmonizing the tri bal structures, which
were basically oncs of moieties and sections in the whole of the
relevant region. A native, asked what his section was, could there-
fore give the reply: '1 am so-and-so in my own lingo - that is
Wungo here.'
The distribution of totemic species between moieties does not
seem to have been uniform, which is hardly surprising. But the
regularities and the systematic way in which the informants
solved each problem is ail the more striking. Except in one region,
opossum belongs to the \Vuturu moiety. Fresh water is Yanguru
along the coast but in the west it belongs to the Wuturu moiety.
The natives say: 'nearly always cold skin with Wuturu and feathers
with Yanguru', so that Wuturu had water, !izard, frog, etc., and
Yanguru emu, duck, and other birds. But where the frog was in
the opposite moiety to the opossum another principle of opposition
was brought to the rescue: both kinds of animal hop and this
resemblance is ascribed to the fact that the frog is the opossum's
'father', and in a matrilineal society father and son belong to
opposite moieties :
\Vhen informants worked out lists of totem contained in each moiety,
1 57
THE SAVAGE MIND
they invariably reasoned along these lines: trees and the birds which made
their nests in them were in the same moiety; trees which grew alongside
creeks or in water-holes and swamps were in the same moiety as water,
fishes, water-fowl, lily-roots. 'Eaglehawk, plain turkey, everything that
flies ail work together. Carpet snake [Python variegatus] and ground
goanna [Varanus Gould?] ail work together - they travel together in olden
time' (Kelly, p. 465).
The same species sometimes figures in both moieties. This is so in

the case of the Python variegatus ( carpet-snake); but the natives
distinguish four varieties according to their skin markings and
these varieties are di vided between the moieties in pairs. The sa me
holds for the varieties of turtle. The grey kangaroo is Wuturu and
the red Yanguru but they would not fight with each other in battle.
Another native group say that certain natural species belong to
fire and others to water: opossum, bee, and sand goanna (Varanus
eremius ?) are said to 'own fire', Python variegatus ( carpet-snake ),
Leipoa ocellata (scrub turkey), lizard, and porcupine 'own water'.
It was claimed that long ago the ancestors of the group in question
had fire, and the people living in the scrub had water. They joined
up and shared water and fire. Finally, each totem was specially
related to a species of tree, and a branch of the relevant clan tree is
placed in the grave of a deceased member of the clan. The emu is
said to own box tree (Bursaria sp.?), porcupine and eagle-hawk
own certain varieties of acacia (brigalow), opossum another acacia
(Kidji), the carpet-snake owns sandal wood, and the sand goanna
bottle-tree ( Sterculia ?). In the western groups the dead were
buried facing east or west according to the moiety (I.e., pp.
+61-6).
So, although the social organization is reduced to chaos by the
new conditions of life imposed on the natives and the Jay and
religious pressures to which they have been subjected, the theoret-
if'al attitude continues to flourish. When it is no longer possible
to retain the traditional interpretations, others are worked out
which, like the first, are inspired by motivations (in Saussure's
sense) and by schemes. Social structures previously simply juxta-
posed in space are made to correspond at the same time as the
animal and plant classifications of each tribe. According to their
tribal origin, the informants conceived the dual scheme on the
model of opposition of resemblance, and they formalized it in
158
terms of kinship (father and son) or directions (east and west) or

elements (land and sea, water and fire, air and land), or again in
terms of the differences or resem\:>lances between natural species.
They took these various procedures into account also and sought
to formulate rules of equivalence. If the process of deterioration
were halted, there is no doubt that this syncretism could serve as
the starting point of a new society, for working out an entire
system with ail its aspects adjusted.
This example shows how the logical dynamism, which is a
property of the system, succeeds in overcoming what, even for
Saussure, does not constitute an antinomy. Apart from the fact
that systems of classification, like languages, may differ with
respect to arbitrariness and motivation without the latter ceasing to
be operative,* the dichotomizing character which we have found in
them explains how the arbitrary aspects (or those which appear to
us arbitrary, for one can never be sure that a choice which is
arbitrary for the observer may not be motivated from the point of
view of indigenous thought) corne to be grafted on to the rational
aspects without altering their nature. 1 have represented systems
of classification as 'trees'; and the growth of a tree is a good illus-
tration of the transformation just mentioned. A tree is, as it were,
strongly motivated so far as its lower parts are concerned: it must
have a trunk and the trunk must be nearly vertical. The lower
branches already allow more arbitrariness: their number, although
it may be expected to be limited, is never fixed in advance, nor is
the orientation of each and its angle in relation to the trunk. But
these aspects nevertheless remain bound by reciprocal relations,
since the larger branches, given their own weight and the foliage-
laden branches they hold up, must balance the pressures which
they apply at the common point of support. The part played by
motivation, however, diminishes, and that of arbitrariness increases
progressively as we turn our attention higher: the terminal
branches can no longer compromise the tree's stability nor alter its
characteristic shape. Their multiplicity and insignificance has
freed them from the initial constraints and their general distribu-
tion can be explained either as a series of repetitions, on an ever-
diminishing scale, of a plan which is also written into the genes in
• & the Lovedu of South A.frica say: 'The ideal is to retum home, for "The
only place one never retums to is the womb" '(Krige, p. 323).
1 59
THE SAVAGE MIND
their cells or as the result of statistical fluctuations. The structure,

intelligible at the start, in branching out reaches a sort of inertia or
logical indifference. Without contradicting its primary nature, it
can thereafter undergo the effect of multiple and varied incidents
which occur too late to prevent an attentive observer from
identifying it and classing it in a genus.
160
CHAPTER SIX
UNIVERSALIZATION AND
PARTICULARIZATION
The antinomy which some believe they have detected between

history and system* would seem to be present in the cases con-
sidered only if we were not aware of the dynamic relation between
the two aspects. There is room between them for a diachronie and
non-arbitrary construction providing a transition from one to the
other. Starting from a binary opposition, which affords the simplest
possible example of a system, this construction proceeds by the
aggregation, at each of the two poles, of new terms, chosen because
they stand in relations of opposition, correlation, or analogy to it.
It does not, however, follow from this that the relations in question
have to be homogeneous. Each 'local' logic exists in its own right.
It consists in the intelligibility of the relation between two
immediately associated terms and this is not necessarily of the same
type for every link in the semantic chain. The position is some-
what comparable to that of inexperienced players in a game of
dominoes who consider only the value of the adjacent halves in
joining the pieces but manage to continue the game none the Jess
for their Jack of previous knowledge of its composition.
The logic of the system need not, therefore, coïncide at every
• But this disillusioned reflection on the part of one of the champions of pu rel y
historical ethnology is enough to convince one that these two notions are of value
only as limiting cases: 'The present state of Zande clans and that of their totemic
affiliations can only be understood in the light of the political development of
Zande society, even though it can be for us only a glimmering light. Hundreds
of thousands of people of different ethnie origin ail jumbled up - the ethnologist
in Africa may sc.metimes sigh for some neat little Polynesian or Melanesian
island community!' (Evans-Pritchard, 2, p. 121).
161
THE SAVAGE MIND
point with the set of local logics inserted in it. This general logic
can be of a different order. lt is then definable by the number and
nature of the axes employed, by the ru les of transformation making
it possible to pass from one to another, and finally by the relative
inertia of the system, that is, its greater or Jess receptiveness to
unmotivated factors.
The so-called totemic classifications, the beliefs and practices
connected with them, are only one aspect or mode of this general
systematic activity. From this point of view 1 have so far done
little more than develop and deepen some comments of Van
Gennep's:
Every ordered society necessarily classes not only its human members,
but also the objects and creatures of nature, sometimes according to their
extemal form, sometimes according to their dominant psychic character-
istic, sometimes according to their utility as food, in agriculture or in
industry, or for the producer or consumer ... Nothing entitles us to
regard any one system of classification, say the zoological system of
toternism or the cosmographie system or the occupational system (castes)
as prior to the others (Van Gennep, pp. 345--6).
A footnote makes it clear that Van Gennep was fully conscious of
the boldness and novelty of this passage:
It will be seen that 1 do not accept Durkheim's view (Formes, p. 318)
that the cosmic classification of living creatures (including man) and
material objects is a consequence of totemism. In my own view, the
special form of cosmic classification found in totemism is not even a
nuance of it but one of its primitive and essential components. For
peoples without totemism also possess a system of classification which in
this case too is one of the primordial elements of their general system of
social organization and as such reacts on their Jay and magico-religious
institutions. Examples are the oriental systems, Chinese and Persian
dualism, Assyro-Babylonian cosmographism, the so-called magical
system of syrnpathetic correspondences, etc.
Van Gennep's demonstration is inadequate, however, in spite of
these sound views, for he continued to believe in totemism as an
institutional reality. He no longer tried to make it into a classi-
ficatory system from which ail others derived, but he still attempted
to preserve a distinctiveness for it comparable to that of a species
objectively identifiable within a genus:
The notion of totemic kinship is th us composed of three elements:
physiological kinship ... social kinship ... and cosmic, classificatory
kinship which links al! the men of a single group to creatures or objects
theoretically belonging to the group. What characterizes totemism •.•
UNIVERSALIZATION AND PARTICULARIZATION
is ... the particular combination of these three elements, just as a certain

combination of copper, sulphur, and oxygen makes copper sulphate
(I.e.).
Although so near the truth, Van Gennep thus remained a prisoner
of the traditional classification in whose framework he was content
to confine his argument. ln neither his own nor his predecessors'
writings can one find any foundation for the rash comparison he
brings in support of his thesis. That copper sulphate is a chemical
substance, in spite of the fact that none of its constituent elements
are exclusive to it, is due to the fact that their combination results
in a set of differential properties: form, colour, flavour, effect on
other substances and on biological beings, all properties which are
not found together in anything else. Dut the same is not true of
totemism, however it may be defined. It is not a substance in an
ethnological kingdom, but consists rather of rough qualities of
varying elements, whose thresholds are arbitrarily selected by each
theorist and whose presence, absence, or degree involves no
specific effects. The most that can be discerned in the cases tradi-
tionally diagnoscd as 'totemic' is a relative inflation of the classi-
ficatory scheme at the species level, without any real change in its
nature or structure. l\tloreover, we are never sure that this inflation
is an objective property of the scheme and not just the result of the
particular conditions under which the observation was conducted.
The work of the late Marcel Griaule, of G. Dieterlein, G. Calame-
Griaule, and D. Zahan among the Dogon and the Bambara has
shown in the course of its development over twenty years how the
observers have gradually had to connect the 'totemic' categories,
originally isolated in conformity with thefiatof traditional ethnology,
with facts of a different order, so that they have now corne to look
like no more than one of the points of view from which a system of
several dimensions may be comprehended.
Ali that can therefore be conceded to the upholders of totemism
is the special position devolving on the notion of a species considered
as a logical operator. But this discovery considerably antedates the
earliest speculations about totemism, having been formulated first
by Rousseau (Lévi-Strauss 6, pp. 142-6) and then, with reference
to issues considered in this work, by Comte. Comte sometimes
employed the notion of taboo but, although he could have been
acquainted with Long's book, the idea of totemism seems to have
remained unknown to him. lt is all the more significant that in his
THE SAVAGE MIND
discussion of the passage from fetishism to polytheisrn (where he

would probably have put toternisrn), Comte makes it a consequence
of the ernergence of the notion of species:
\Vhen, for example, the similar vegetation of the different trees in a
forest of oaks had finally to lead to the representation, in theological con -
ceptions, of what their phenomena presented in common, this abstract
being was no longer the fetish belonging to any tree; it became the God
of the forest. So the intellectual passage from fetishism to polytheism is
essentially reducible to the inevitable preponderance of specific over
general ideas (5ze leçon, vol. V, p. 54).
Tylor, the founder of modern ethnology, appreciated the use that

could be made of Cornte's idea, which, he remarks, is even more
applicable to this special category of deities, the deified species:
The uniformity of each kind not only suggested a common parentage,
but also the notion that creatures so wanting in individuality, with quali-
ties so measured out as it were by line and rule, might not be independent
arbitrary agents, but mere copies from a common mode], or mere instru-
ments used by controlling deities (Tylor, vol. Il, p. 243).
The logical power of the specific operator can also be illustrated in

other ways. For this allows dornains very different frorn each other
to be integrated into the classificatory scherne, th us affording classi-
fications a rneans of going beyond their Iimits: either extending to
dornains outside the initial set - by universalization, or alternatively
by particularization - taking the classificatory approach past its
natural bounds, that is, to individuation.
The first point can be dealt with briefly. A few exarnples are suffi-
cient. So Iittle is the species 'grid' confined to sociological categories
that (notably in America) it sornetimes serves to order a dornain as
restricted as that of diseases and remedies. The Indians of the south-
east United States attribute pathological phenomena to a conttict
between men, animais, and plants. Vexed with men, animais sent
them diseases. Plants, who were friends of men, retorted by supply-
ing rernedies. The important point is that each species possesses a
specific disease or remedy. Thus, according to the Chickasaw,
stornach disorders and pains in the legs corne frorn snakes, vomiting
from dogs, aches in the jaw frorn deer, pains in the abdomen from
bears, dysentery frorn skunks, nose-bleeding from squirrels, jaun-
dice frorn otters, disturbances of the lower part of the abdomen
and the bladder frorn moles, cramp from eagles, eye diseases
164
U:-\IVERSALIZATION AND PARTICULARIZATION
and somnolence from owls, pain in the joints from rattle-snakes, etc.
(Swanton, 2).
Similar beliefs are found among the Pima of Arizona. They attri-
bute throat diseases to the badgei,- swellings, headaches and fever to
the bear, diseases of the throat and lungs to the deer, children's
diseases to the dog and coyote, stomach trouble to the gopher or
prairie-rat, ulcers to the jack-rabbit, constipation to the mouse,
nose-bleed to the ground-squirrel, haemorrhages to the hawk and
eagle, syphilitic sores to the vulture, children's fevers to the Gila
monster, rheumatism to the horned toad,* 'white fever to the !izard,
kidney and stomach troubles to the rattle-snake, ulcers and paralysis
to the turtle, internai pains to the butterfly, etc. (Russcll).t Among
the Hopi, who are a day's march from the Pima, an analogous classi-
fication is based on the organization into religious orders, each of
which can inflict a punishment in the form of a particular discase:
abdominal swelling, sore cars, horn-like swelling on the top of the
head, deafness, cczcma on the upper part of the body, twisting and
twitching of the face and neck, soreness in the bronchial tubes, bad
knee (Voth 2, p. io9 n.). The problem of classifications could un-
doubtedly be tackled from this angle and some curious resem-
blances, symptomatic of logical connections which could be of
considerable importance, might be found between distant groups
(the association of the squirre! and nose-blceding, for instance,
seems to recur in a large number of North Amcrican peoples).
The spccific categories and the myths connccted with them can
also serve to organize space, and the classificatory system is thcn
extended on a territorial and geographical basis. A classic example
is the totemic geography of the Aranda, but there are other people
equally exacting and subtle in this respect. ln Aluridja territory a
rocky site measuring five miles round the base was recently dis-
covered and described in which every accident of relief corresponds
to a phase of ritual in such a way that this natural rock illustrates the
structure of their myths and the order of the ceremonies for the
natives. Its north sicle is the sicle of the moiety of the sun and of the
• It may be noted, in support of the considerations adduced earlier (pp. 64-5)
that it is probably the same behaviour on the part of this animal which suggests
entirely different associations to the American Indians and the Chinese. The
Chinese ascribe aphrodisiac virtues to the flesh of the phrynosoma or to the wine
in which it has been macerated because the male holds on to the female so hard
during copulation that he does not let go even when caught.
t For closely related ideas among the Papago, cf. Densmore.
7-TSM
THE SAVAGE MIND
ritual cycle Kerungera, its sou th side that of the moiety of the shade
and the ritual Arangulta. Thirty-eight points on the base üf the
plateau are named and annotated (Harney).
North America also furnishes examples of mythical geography
and totemic topography, from Alaska to California as well as in the
south-west and north-east of the continent. ln this respect the
Pcnobscot of ~faine exemplify a general tendency on the part of the
northern Algonkin to interpret all the physiographic aspects of the
tribal territory in terms of the peregrinations of the civilizing hero
Gluskabe and other mythical personage or incidents. An elongated
rock is the hero's canoe, a streak of white rock represents the entrails
of the moose he killed, Kineo mountain is the overturned cooking
pot in which he cooked the meat, etc. (Speck 2, p. 7).
Again, in the Sudan, a mytho-geographical system has been dis-
covered which covers the entire Niger valley, and thus extends over
more than the territory of any single group, and in which is trans-
lated down to the minutest detail a conception of the relations
between diff erent cultural and linguistic groups which is at once
diachronie and synchronie.
This last example shows that the systems of classification not
only permit the 'furnishing', as it were, of social time (by means of
myths) and of tribal space (with the help of a conceptualized topo-
graphy). The filling in of the territorial framework is accompanied
by enlargement. In the same way that, on the logical plane, the
specific operator effects the transition to the concrete and individual
on the one hand and the abstract and systems of categories on the
other, so, on the sociological plane, totemic classifications make it
possible both to define the status of persons within a group and to
expand the group beyond its traditional confines.
Primitive socit:Lic:-. have, and not without justification, been said
to treat the limits of their tribal group as the f rontiers of humanity
and to regard everyone outside them as foreigners, that is, as dirty,
coarse sub-mcn or even non-men: dangerous beasts or ghosts. This
is often true, but what is overlooked when this is said is that one of
the essential fonctions of totemic classifications is to break down
this closing in of the group into itself and to promote an idea some-
thing like that of a humanity without frontiers. There is evidence of
this phenomenon in all the classical areas of so-called totemic organ-
ization. ln a region in \Vest Australia the clans and theirtotems 'are
grouped together into a number of what may be called inter-tribal
166
totemic divisions' (Radcliffe-Brown l, p. 214). This is equally true

cf other parts of the same continent:
1 have ascertained that in 167 casés (s6o/o) out of 300 names of conunon
totemic animais, the \Vestem Aranda and the Loritja use the same or
similar terms; and comparison of totemic plant names used by the
\Vestem Aranda and Loritja shows that the same terms are found in both
languages to refer to 147 of the 220 species of plants which 1 recorded
(6i/'0 ) (C. Strehlow, pp. 66-'7).
Analogous observations have been made in America arnong the

Sioux and the Algonkin. The :VIenomini Indians, who belong to the
latter, entertain a:
... general belief in the common relationship of not only the indi-
viduals of a certain totem with!n the tribe, but of ail persons of a similarly
named totem of another tribe belonging to the same linguistic family;
and in the belief of the Menomini this extended also to tribes other than
those of the same linguistic family (Hoffman, p. 43).
Similarly, among the Chippewa:

Ali members with the same totem regarded themsclves as rclated even
though of different villages or different tribes ... \Vhen two strangers met
and found themselves to be of the same totem they immediately began to
trace their genealogy, ... and the one became the cousin, the uncle, or
the grandfather of the other, although the grandfather might often be
the younger of the two. The ties of the totem were considered so strong
that if a quarre! should happen between a person with the same totem as a
bystander and a cousin or other near relative of the latter but with a
different mark, the bystander would sicle with the person with the same
totem whom perhaps he had never seen before (Kinietz, pp. 69-70).
This totemic universalization not only breaks clown tribal frontiers

and creates the rudiments of an international society. It also some-
times goes beyond the limits of humanity in a biological, and no
longer merely sociological, sense, when totemic names are applic-
able to totemic animais. This occurs among the Australian tribes of
the Cape York Peninsula in the case of dogs* - also referred to as
'brothers' or 'sons' according to the group (Sharp, p. 70; Thomson)
- and among the Iowa y and Winnebago Indians in the case of dogs
and horses (Skinner 3, p. 198).
• Among the Wik Munkan a dog is called Yatut 'extracting bones of •.. the
totem fish' if his master is of the Bone-fish clan, Owun 'illicit or stolen meeting'
if his master is of the Ghost clan (Thomson, pp. 161-2).
THE SAVAGE MIND
1 have given a brief indication of how the mcshes of the net can
stretch indefinitely in accordance with the dimensions and gener-
ality of the field. 1t remains to be shown how they can also shrink
to filter and imprison reality but this time at the lower limit of the
system by extending its action beyond the threshold which one
would be inclined to assign to ail classification, that beyond which
it is no longer possible to class, but only fo name. These extreme
operations are in fact Jess widely separated than they might appear
and, when seen in the perspective of the systems we are studying,
they may even be superposed. Space is a society of particular places
as people are landmarks within the group. Places and individuals
alike are designated by proper names, which can be substituted for
each other in many circumstances common to many societies. The
Yurok of Ca!ifornia provide one example among others of this
personified geography, where trails are conceived of as animated
beings, each house is named, and the names of places replace
persona! names in current usage (\Vaterman).
An Aranda myth well expresses this feeling of correspondence
between geographical and biological individuation: the earliest
divine beings are shapeless, without limbs, and welded together
until the coming of the god Mangarkunjerkunja (the fi.y-catcher
!izard), who proceeded to separate them and fashion them indivi-
dually. At the same time (and is this not indeed the same thing ?) he
taught them the arts of civilization and the system of sections anJ
sub-sections. The eight sub-sections were originally divided into
two main groups: four land ones and four water ones. lt was the god
who 'territorialized' them by allotting each site to a pair of sub-
sections. Now, this individuation of territory corresponded to bio-
logical individuation in another way as well, in that the totemic
mode of fertilizing the mother explains the anatomie differences
observable among children. Those with fine features were con-
ceived by the operation of a ratapa, embryo-spirit; those with large
features by magical projection of a rhomb into a woman's body;
children with fair haïr were direct reincarnations of totemic
ancestors (C. Strehlow). The Australian tribes of the Drysdale
River, in Northern Kimberley, divide ail kinship relations, which
together compose the social 'body', into five categories named after
a part of the body or a muscle. Since a stranger must not be ques-
tioned, he announces his kinship by moving the relevant muscle
(Hernandez, pp. 228-9). In this case, too, therefore, the total system
168
u:-;rvERSALIZATION AND PARTICULARIZATION
of social relations, itself bound up with a system of the universe,

can be projected on to the anatomical plane. In Toradja there are
fifteen terms to name the cardinal points which correspond to the
parts of body of a cosmic divinity (Woensdregt ). Other examples
could be cited, from ancient Germanie kinship terminology as
well as frorr. the cosmological and anatomical correspondences
of the Pueblo and Navajo Indians and the Sudanese Negroes.
It would certainly be instructive to make a detailed study of the
mcchanics of this homological particularization in a sufficient
number of cases. Its general relation with the forms of classification
we have encountered so far can be seen clearly in the following
derivation:
If
(group a) : (group b) : : (bear species) : (eagle species)
then
(member x of a) : (member y of b) : : (member l of bear):
(member m of eagle)
These formulae have the advantage of throwing into relief a

problem traditionally discussed in Western philosophy, though the
question of whcthcr or not, or in what form, it arises in exotic
societies has received little attention: I mean the problem of organic-
ism. The equations above would be inconceivable were it not for
the postulate of a fairly general correspondence between the 'mem-
bers' of society and if not the members themselves, then at least the
attributes of natural species: parts of the body, characteristics
details, ways of being or behaviour. The available evidence on this
subject suggests that a great many languages equate parts of the
body, regardless of the diversity of orders and families, or sometimes
even of kingdoms, and that this system of equivalence is susceptible
of very considerable extensions (Harrington).'* Morphological
classifiers therefore fonction in addition to and alongside the
specific classifiers. The theoretical account of these remains to be
given, but we have seen that they operate on two planes: that of
anatomical detotalization and that of organic retotalization.
These are equally interdependent, as we have shown in the case
• ln America the following equivalences are found: horns (quadrupeds) =
eyestalks (Molluscs) = antennae (arthropoda); penis (vertebrates) = siphon
(molluscs); blood (animais) = juice (plants); slobber (babies' 7= saliva of adults)
= excretion; byssus of musse! = tie, string, etc. (Harrington).
THE SAVAGE MIND
of the other levels. We showed a moment ago that the Aranda attri-
bute empirically established morphological differences to hypothetic
differences in modes of totemic conception. But the examples of the
Omaha and Osage furnish evidence of a correlative tendency which
consists in introducing symbolically expressed specific differences
into empirical and individual morphology. The children of each
clan wear their haïr eut in a characteristic Style evoking a distinctive
feature or aspect of the animal or natural phenomenon which serves
as an eponym (La Flesche 4, pp. 87-9).
This modelling of the appearance according to specific, elemental,
or categorical schemes has psychological as well as physical conse-
quences. A society which defines its segments in terms of hi gh and
low, sky and land, day and night, can incorporate social or moral
attitudes, such as conciliation and aggression, peace and war,justice
and policing, good and bad, order and disorder, etc., into the same
structure of opposition. In consequence, it does not confine itself
to abstract contemplation of a system of corresponder.ce but rather
furnishes the individual members of these segments with a pretext
and sometimes even a provocation to distinguish themselves by their
behaviour. Radin (I, p. 187), referring to the Winnebago, very
rightly insists on the reciprocal influence of religious and mythical
conceptions of animais on the one hand and the political fonctions
of social units on the other.
The Sauk Indians provide a particularly instructive example by
reason of their individuating rule for determining membership of a
moiety. The moieties werc not exogamous and their role, which was
pureiy ceremonial, was principally manifested at feasts. It is
important, from our point of view, to notice that these were con-
nected with rites for the giving of names. Membership of moieties
was determined by a rule of alternation: the first born was affiliated
to the moiety of which his father was nota membcr, the next child to
themoietyof which hisfatherwas amember, andsoon. Now, at least
in theory, these affiliations determined behaviour which one might
call 'dispositional': members of the Oskûsh moiety ('the Blacks')
had to complete all their enterprises, those of the Kishko moiety
('the Whites') might give up or turn back. In theory, if not in prac-
tice, an opposition in terms of categories thus directly influenced
everyone's temperament and vocation, and the institutional scheme
which made it possible for this to happen testifies to the link between
11 12
Fig 9 Cut of Osage and Omaha boys' hair according to clan. I. Head and
tail of elk; 2. Head, tail and horns of buffalo; w, Homs of buffalo;
3. Line of buffalo's back as seen against the sky; 4b. Head of bear; 4c.
Head, tail and body of small birds; 4d. Shell of the turtle with the head,
feet and tail of the animal; 4e. Head, wings and tail of the eagle; 5. Four
points of the compass; 6. Shaggy side of the wolf; 7. Homs and tail of the
buffalo; 8. Head and tail of the deer; 9. Head, tail and knobs of the grow-
ing horn of the buffalo calf; xo. Reptile teeth; xx. Petals of the cone ftower;
12. Rock with algae floating round (following La Flesche 4, pp. 87, 89).
171
THE SAVAGE MIND
the psychological aspect of persona! destiny and its social aspect,

which is the result of the giving of a name to each individual.
Thus we reach the final level of classification: that of individua-
tion, for in the systems we are considering here individuals are not
only ranged in classes; their common membership of the class does
not exclude but rather implies that each has a distinct position in it,
and that there is a homology bet\veen the system of individuals
within the class and the system of classes within the superior cate-
gories. Consequently, the same type of logical operation links not
only all the domains internai to the system of classification but also
peripheral domains which might, by their nature, have been sup-
posed to escape it: atone extreme (owing to their virtually unlimited
extension and their disregard of principle) the physico-geographical
substratum of social life and even social life itself, overflowing the
mould in which it was cast; and at the other extreme (by reason of
their concrete nature which is equally given) the ultimate diversity
of individual and collective beings, which, soit has been claimed,
were named only because they could not be signified (Gardiner).
Proper names do not therefore constitute a mere practical modal-
ity of classificatory systems which cou Id be mentioned af ter the
other modalities. They raise a problem for ethnologists, even more
than for linguists. The problem for linguists is the nature of proper
names and their place in the system of the language. \Ve are con-
cerned with this problem but also with another, forwe are faced with
a twofold paradox. \Ve need to establish that proper names are an
integral part of systems we have been treating as codes: as means of
fixing significations by transposing them into terms of other signi-
fications. \Vould this be possible if it \vere true, as logicians and some
linguists have maintained, that proper names are, in Mill's phrase,
'meaningless', lacking in signification? Again, and this is the most
important point, we have conceived of the forms of thought with
which we have been concerned as totalizing thoughts, which exhaust
reality by means of a finite number of given classes, and have the
fondamental property of being transformable into each other. How
could this quantified thought, to which, on the practical plane, we
attributed the great disconries of the neolithic revolution, have been
both theoretically satisfying and effectively applied to the concrete,
if the latter contained a residue of unintelligibility - to which, in the
last analysis, concreteness itself is reducible - which is essentially
recalcitrant to signification? The principle of ail or nothing not only
U::\'IVERSALIZATIO!\' AND PARTICULARIZATION
has heuristic value to thought founded on the operation of dicho-

tomies, but is also an expression of a property of what exists: either
everything, or nothing, makes sel}.se. *
Let us now return to the ethnographie facts at the point at which
we left them. Almost all the societies 1 have mentioned construct
their proper names from clan appellations. The proper names of
the Sauk, who furnished our last example, are said always to be
related to the clan animal: either because they mention the name of
the animal itself, or because they suggest one of its habits, attributes,
or characteristic qualities (real or mythical), or because they refer
to some animal or object with which it is associated. Sixty-six names
cf the Bear clan, eleven of the Bison clan, thirty-three of the Wolf
clan, twenty-three of the Turkey clan, forty-two of the Fish clan,
thirty-seven of the Great Sea clan, forty-eight of the Thunder clan,
fourteen of the Fox clan, and thirty-four of the Deer clan have been
recorded (Skinner 2).
The list of Osage proper names which belong to clans or sub-
clans, although fragmentary, is long enough to occupy forty-two
quarto pages in La FI esche 4 (pp. I 22-64). The rule of formation is
the same as among the Sauk. Thus for the Black Bear clan we have:
Flashing-eyes (of the bear), Tracks-on-the-prairies, Ground-
cleared-of-grass, Black-bear-woman, Fat-on-the-skin of the black
bear, etc. Among the Tlongit of Alaska 'all names ... are felt to
belong to a particular sib, and some have even been designated as
belonging to a particular "house" or "lineage" ' (Laguna, p. 185).
These examples can be multiplied, for similar ones are found in
almost all the Algonkin and Sioux tribes and in those of the north-
west coast, that is to say, in the three dassical domains of 'totemism'
in North America.
South America provides illustrations of the same phenomenon,
notably among the Tupi Kawahib, whose clans possess proper
names derived from the eponym (Lévi-Strauss 3). The proper
names of the Bororo also seem to belong to particular clans or even
to powerful lineages. Those who have to depend on the good will of
other clans for having a name are said to be 'poor' (Cruz).
The connection between proper names and totemic appellations
is found in l\1elanesia:
• Everything, except the existence of what exists, which is not one of its
properties (cf. below, p. 255).
1 73
THE SAVAGE MIND
The totcmic system [of the Iatmul] is enormously elaborated into a

scries of persona! names, so that every individual bears names of totemic
ancestors - spirits, birds, stars, pots, adzes, etc., etc. - of his or her clan,
and one individual may have thirty or more such names. Every clan has
hundreds of these polysyllabic ancestral names which refer in their
Etymology to secret myths (Bateson, p. 127).
The position seems to have been the sa me throughout Australia. 'By

knowing a native's name, if one knew the Aranda language well
enough, it would seem one would also know his or her totem by
deduction' (Pink, p. 176). Another observation, relating to the
Murngin of Arnhem Land, echoes this one: 'The names of the
living are all taken from some part of the totemic complex and refer
directly or indirectly to the totem' (Warner, p. 390). The proper
names of the \Vik M unkan also derive from their respective totems.
For instance, for men whose totem is the fish barramundi (Osteo-
glossum), which they spear: the-barramundi-swims-in-water-and-
sees-a-man, the barramundi-moves-its-tail-as-it-swims-around-its-
eggs, the-b.-breathes, the-b.-has-its-eyes-open, the-b.-breaks-a-
spear, the-b.-eats-mullet, etc. And for women whose totem is the
crab: the-crab-has-eggs, tide-takes-crabs-out-to-sea, the-crab-
stops-down-hole-and-is-dug-out, etc. (McConnel, p. 184). The
tribes of the Drysdale River region have proper names derived from
clan appellations, as the author quoted above underlines: 'Proper
names ... always bear relation to one's totem' (Hernandez).
It is clear that these individual appellations belong to the same
system as the collective appellations discussed above and that,
through the intermediary of the latter, one can, with the help of
transformations, pass from the horizon of individuation to that of
more general categories. Everyclan orsub-clan possesses a quota of
names which only its own members can bear and, just as an indivi-
dual is part of the group, so an individual name is 'part' of the
collective appellation, either in the sense that it may cover the whole
animal, and the individual names correspond to its parts or limbs,
or that the collective appellation may depend on an idea of the
animal conceived at the highest level of generality and the individual
appellations correspond to one of its predications in time or space:
Tiarking-dog, Angry-bison, or there may be a combination of both
procedures: Flashing-eyes-of-the-bear. The animal may be sub ject
or predicate in the relation so expressed: The-.fish-moves-its-tail,
Tide-takes-crabs, etc. Whichever procedure is employed (and they
are most often foundjuxtaposed) the proper name refers to a partial

aspect of the animal or plant entity just as it corresponds to a partial
aspect of theindividual being-in general, and in particular, in those
societies in which an individual receives a new name at every impor-
tant point of his life. Neighbouring societies, moreover, employ the
same constructions for forming in some cases persona! names
(borne by individual members of a clan group) and in others collec-
tive names (borne by bands, lineages, or groups of lineages, that is,
sub-groups of a single clan).
Two parallel detotalizations are thus involved: of species into
parts of the body and attitudes, and of social segments into indivi-
duals and roles. But just as the detotalization of the concept of a
species into particular species, of each species into its individual
members, and of each of these individuals into organs and parts of
the body could, as I tried to show with the help of a figurative mode!,
issue into a retotalization of the concrete parts into abstract parts
and of the abstract parts into a conceptualized individual, so here
the detotalization takes place in the form of retctalization. Kroeber's
observations about proper names among the Miwok of California
serve to provide a final example and at the same time open up a new
perspective:
There are no subdivisions of any sort within the moieties. Associated
with each, however, is a long list of animais, plants and objects; in fact,
the native concept is that everything in the world belongs to one or the
other side. Each member of a moiety stands in relation to one of the
objects characteristic of his moiety - a relation that must be considered
totemic - in one way only: through his name. This name given him
in infancy by a grandfather or other relative, and retained through
life, refers to one of the totem animais or objects characteristic of his
moiety.
Nor is this ail: in the great majority of cases the totem is not mentioned
in the name, which is formed from some verbal or adjectival stem, and
describes an action or condition that might apply equally well to other
totems. Thus on the verb hausu-us are based the names Hausu and
Hauchu, which connote, respectively, the yawning of an awakening bear
and the gaping of a salmon drawn out of the water. There is nothing in
either name that indicates the animais in question - which even belong
to opposite moieties. The old men who bestowed them no doubt an-
nounced the totemic reference of the names; the bearers, and their family,
kin, and more intimate associates, knew the implication; but a Miwok
from another district would have been uncertain whether a bear, a salmon,
or one of a dozen other animais was meant (Kroeber 2, pp. 453-4).
1 75
THE SAVAGE MIND
This feature does not appear to be peculiar to the Miwok. Lists of

clan names arnong the Sioux tribes contain plenty of analogous
exarnples and what Kroeber says also coïncides with a characteristic
of the Hopi Indians' system of narning. Thus the narne Cakwyam-
tiwa, the literai rneaning of which is 'Blue (or green )-having-corne-
out' can refer to the rnatured flower of the tobacco plant or to that
of the Delphinium scaposum or again to the germination of plants in
general, depending on the clan of the don or of the narne. Sirnilarly,
the narne Lornahongiorna, 'Stand up' or 'rise gracefully' can refer
to the stem of the reed or to the erect wings of the butterfly, etc.
(Voth 3, pp. 68-<)).
The prevalence of the phenornenon raises a psychological prob-
lern, relevant to the theory of proper narnes, which 1 shall corne back
to la ter. I want now just to ernphasize that the relative indeterrninacy
of the system corresponds, virtually at least, to the phase of retotal-
ization: proper narnes are forrned by detotalizing species and by
deducting a partial aspect of thern. But by stressing exclusively the
fact cf the deduction and leaving the species which is the subject of
it indeterrninate, there is a suggestion that all the deductions (and
so all the acts of narning) have sornething in cornrnon. A unity
divined at the heart of diversity is clairned in advance. Frorn this
point of view, too, the dynarnics of individual narnes derives frorn
thr. classificatory systems we have been analysing. It consists of
approaches of the sarne type, and sirnilarly oriented.
Further, it is striking that systems of prohibitions with the sarne
characteristics are to befound on the plane of individual appellations
as well as on that of collective narnes. In sorne cases the plant or
animal serving as a social group's eponyrn rnay not be used as food
by that group; in other cases, however, it is the linguistic use of the
plant or animal serving as an individual's eponyrn which is pro-
hibited. It is possible to sorne extent to pass frorn one plane to the
other. Proper narnes of the type to which we have so far confined
ourselves are usually forrned by an irnaginary dissection of the Lody
of the animal, inspired by the gestures of the huntsrnan or the
cook; but they can also be forrned by linguistic dissection. In the
tribes of the Drysdale River region in North Australia, the wornan's
narne Poenben is forrned frorn the English 'spoon', an irnplernent
associated, as one rnight expect, with the totem 'white-fellow'
(Hernandez, p. 218).
In Australia, as well as in America, we find prohibitions on the
176
use of the names of the dead which 'contaminate' any words with a
phonetic resemblance to these names. The Tiwi of the Melville and
Bathurst islands taboo not only the proper name 'IV1ulankina' but
also the word 'mulikina' which means: full, filled, enough (Hart).
This usage finds a parallel in that of the Yurok of California: 'When
Tegis died, the common word tsis, "woodpecker scalps", was not
uttered in the hearingofhis relatives or by them' (Kroeber 2, p. 48). •
The Dobu islanders forbid the use of proper names between
individuals temporarily or permanently connected by a 'species' tie
through being companions on a voyage, eating together, or sharing
the favours of the same woman, as the case may be (Bateson).
Such facts have a double daim on our interest. In the first place,
they :tfford an indisputable analogy with food prohibitions, which
have been wrongly associated with totemism alone. In the same
way that at Mota a woman is 'contaminated' by a plant or animal, as
a result of which she gives birth to a child subject to the correspond-
ing eating prohibition, and at Ulawa it is the dying man who 'con-
taminates' by his incarnation in an animal or plant species which his
descendants are then forbidden to consume, so, by homophony, a
name 'contaminates' other words, the use of which then cornes to be
forbidden. Secondly, this homophony defines a class of words, to
which the prohibition applies because they belong ta the same
'species', and which th us acquires an ad hoc reality comparable ta
that of animal or plant species. These 'species' of words 'stressed' by
the same prohibition bring together both proper names and common
names. And this is a further reason for suspecting that the diff erence
between the two types of words is not as great as we were near to
admitting at first.
The customs and procedures which 1 have just mentioned are no!
found in all exotic societies nor even in all those which designate
their segments by animal and plant names. The Iroquois, who are
an instance of the latter, seem to have a system of proper names
entirely distinct from the system of clan appellations. Their names
most commonly consist of a verb with an incorporated noun or a
noun followed by an adjective: In-the-Centre-of-the-Sky, He-
raises-the-Sky, Beyond-the-Sky, etc.; Hanging-Flower, Beautiful-
Flower, Beyond-the-Flowers; He-announces-Defeat (or Victory),
• There are other examples in Elmendorf and Kroeber 1960, which was not
available tome at the tÏine of writing this section.
177
THE SAVAGE MIND
He carries-the-News, etc.; She-works-in-the-House, She-has-two-

Husbands, etc.; The-Place-where-Two-Rivers-meet, The-Cross-
ing-of-the-Roads, etc. There is thus no reference at ail to the
eponymous animal but only, and whatever the clan, to technical or
economic activities, to peace and war, to natural phenomena and
celestial bodies. The example of the Mohawk of the Grand River,
where the clan organization decayed more rapidly than in the other
groups, suggests how ail these names could originally have been
arbitrarily created. Thus we find: Ice-floating-down-the-River for
a child born when ice was thawing on the river, or She-is-in-want
for the son of a poor woman (Goldenweiser, pp. 366-8)."
The situation is not, however, fundamentally different from that
among the Miwok and Hopi, described above, whose names,
although in theory evocative of the clan plant or animal, do not make
explicit reference to either and contain a hidden meaning. Even if
this meaning is not indispensable, it is no less true that among the
Iroquois also, proper names, of which there are several hundred or
thousand, are jealously guarded clan possessions. It is indeed this
which made it possible for Goldenweiser to show that the clans of
the Little and Great Turtle, of the Little and Great Snipe, etc., were
formed by duplication : they hold the same na mes in common. The
names instanced by Goldenweiser are no doubt not the result of a
detotalization of the clan animal. But they do suggest a detotalization
of those aspects of social life and the physical world which the system
of clan appellations has not already caught in the meshes of its net.
The principal difference between the Iroquois system of proper
names and the systems of the Miwok, Hopi, Omaha, and Osage (to
mention only a few examples) may therefore be that the latter tribes
extend an analysis already begun in the sphere of clan appellations
to the plane of proper names while the Iroquois employ proper
names for undertaking an analysis devoted to new abjects but of the
same formai type as the other.
The case of various African tribes is more troublesome. The
Baganda have names (more than two thousand of which have been
collected) each of which belongs to a clan. As among the Bororo of
Brazil, some clans are rich in names and others poor. These names
are not confined to human beings. They are also given to hills, rivers,
rocks, forests, wells, landing places, bushes, and single trees. But,
•An analytic classification of about 1 ,500 Iroquois proper names can be fowid
in Cooke.
in contrast to the cases considered above, these names form only

one category among others (Nsimbi, pp. 212-13), and a very differ-
ent procedure for the formation- of names becomes even more
evident in other tribes of the same region:
1\lost usually ~yoro persona! names may be regarded as expressing
what may perhaps best be described as the 'state of mind' of the parent or
parents who give the name (Beattie, p. 100).
The phenomenon has been studied closely in another Uganda tribe,

the Lugbara, where the child is givcn its name by its mother, some-
times assisted by her mother-in-law (her husband's mother). Three-
quarters of the 850 names collected from a single sub-clan refer to
the behaviour or character of one or other of the parents: 'In-
laziness' because the parents were idle, 'In-the-beer-pot' because
the father was a drunkard, 'Give-not' because the mother fed her
husband badly, etc. The other first names refer to recent or immin-
ent death (of other children of the same parents, the parents them-
selves, or other members of the samc group) or to attributes of the
child. It has been pointed out that most of these names are uncom-
plimentary to the child's father or even mother, although it is she
who invents the name. They allude to the negligence, immorality,
or social or material destitution of one or both of the parents. How
can a woman describe herself, in choosing a name for her child, as
an evil sorceress, an unfaithful wife, or as kinless, poverty-stricken,
and starving?
The Lugbara say that names of this type are not generally chosen
by the mother but by the grandmother (f ather' s mother ). The latent
antagonism between lineages allied through marriage, which ex-
plains that the mother avenges herself for the hostility of which she
is a victim among her family-in-law by giving their son a name
humiliating to his father, also explains that the grandmother, linked
to her grandchildren by a very strong emotional tie, symmetrically
expresses her antagonism towards the wife of her son (Middleton).
This explanation, however, does not seem very satisfactory, since,
as the author who records it remarks, the grandmother also cornes
from an outside lineage and was herself in the same position as her
daughter-in-law in the past. Beattie's explanation of a similar
custom among the Banyoro seems sounder and more coherent.
Among the Banyoro also 'persona! names are concerned with the
themes of death, sorrow, poverty, neighbourly spite.' But 'the
179
THE SAVAGE MIND
person giving the name is almost always thought of as being acted

upon, not as acting; the victim of the envy and hatred of others'.
This moral passivity, which pro jects upon the child an image of the
self created by others, finds expression on the linguistic plane:
' ... the two verbs "to lose" and' 'to forget" ... are used in Lunyoro
with the thing forgotten as the subject, the forgetter as the object ...
The loser or forgetter does not act upon things, they act upon him
.. .' (Beattie, p. 104 and n. 5).
Different as this method of forming persona! names is from the
one considered above, the two are found sicle by sicle among the
Banyoro and the Lugbara. Special names are reserved for children
whose birth was marked by some notable circumstance.
Among the Lugbara we find, for instance: Ejua for a male twin
and Ejurua for a female twin; Ondia for a boy and Ondirua for a girl,
if their mother was previously thought to be barren; 'Bilene' ('for
the grave') for a child who is the first of several children to survive
(l\-Iiddleton, pp. 34-5). These names exist before the individuals
who bear them and they are assigned to them on account of the
position which is objectively theirs but in which other individuals
may equally find themselves, and which the group regards as
charged with significance. They differ in every respect from names
which are freely invented by a determinate individual for another
equally determinate individual and express a transitory state of
mind. Are we to say that the former denote classes and the latter
individuals? Both are, however, equally proper names and the cul-
tures in question are so well aware of this that they consider them
substitutable for each other: should the occasion arise, a Lugbara
mother can choose between the two methods of naming.
There are, moreover, intermediate types. A feature in which they
resemble the second kind was provisionally disregarded in classi-
f ying Hopi names in the first category. Although they have to relate
to an objective order (in this case that of clan appellations) the rela-
tion is not with the clan of the bearer of the name (as it is, for
example, among the Yuma) but with that of its donor. •The name I
• This rule is reminiscent of that of the Australian tribes of Cherburg, Queens-
land. Every individual has three names. The first is associated with the bearer's
totemic place and the other two with the father's totem, although totemic
affiliations are transmitted in the maternai line. Thus a woman whose persona!
totem is opossum has the name Butilbaru, which designates a particular dried-
up-creek-bed, and two names derived from her father's totem - in this case,
emu -which means 'emu moves his head this way and that' and 'old emu walking
180
C);I\"ERS.-\LIZATIO.N A~D P.-\RTICCLARIZATIO.N
bear refers to an aspect not of the plant or animal which sen·es as

my clan eponym but of that which sen·es as my sponsor's clan
èponym. This objecti,·ity subjecti,·ized by another, of which I am
the ,·ehicle, is no doubt concealed by the indeterminate nature of
the names which, as we have seen, do not refer explicitly to eponyms.
But it is also reinforced in two ways: by the fact that one needs to
refer back to the concrete social circumstances in which the name
was concei,·ed and attributed in order to understand it; and by the
donor's relatiYe freedom to follow his own inclinations in creating
the name pro,·ided only that he respects the initial restriction that it
must be possible to interpret the name in tcrms of his own clan
appellation. Jlutatis mutandis, this was also the position among the
:!.liwok where the (equirncal and im·ented) name has to relate to
beings or things attached to the moiety of the person named.
\\'hat we ha\·e here are thus two extreme types of proper name
between which there are a whole series of intermediate cases. At
one extreme, the name is an identifying mark which, by the applica-
tion of a rule, establishes that the indi,·idual who is named is a
member of a preordained class (a social group in a system of groups,
a status by birth in a system of statuses). At the other extreme, the
name is a free creation on the part of the individual who gi1.:es the
name and expresses a transitory and subjecti,·e state of his own by
means of the person he names. But can one be said to be really
naming in either case? The choice seems only to be between identi-
fying someone else by assigning him to a class or, under cover of
giving him a name, identifying oneself through him. One thereforc
ne\"er names: one classes someone else if the name is given to him
in virtue of his characteristics and one classes oneself if, in the
belief that one need not follow a rule, one names someone else
'freely', that is, in virtue of characteristics of one's own. And most
commonly one does both at once.
Suppose 1 buy a pedigree dog. If 1 determine to preserve his value
and prestige and to transmit them to his descendants, then 1 shall
have scrupulously to observe certain rules in the choice of his name,
these rules being mandatory in the society of pedigree dog owners
to which I aspire to belong. Usually the dog will in any case already
up and down'. The son of an opossum father is called 'Karingo' (the name of a
little spring), :\1yndibambu: 'the opossum when his chest is slit up' and
:\1ynwhagala: 'opossum up the tree, now down', etc. (Kelly, p. 468).
181
THE SAVAGE MIND
have been given a namc on the initiative and responsibility of the

kennels wherc he was born, and when I acquire him he \vill already
have been registered with the authorized dog-breeders' association.
The name will begin with an initial which by convention corres-
ponds to the animal's year of birth and it will sometimes be com-
pletcd by a prefix or affix connoting its breeding, much in the
manner of a patronymic name. I am, of course, at liberty to address
my dog differently. It still remains the case that this miniature
poodle to which, for the purpose of calling him, his master has given
the name 'Ilow-wow', bears the name of 'Top-Hill Silver Spray' in
the registers of the British Kennel Club, this name being composed
of two terms, the first connoting a particular kennel and the second
representing an available name. Only the choice of a term of address
is thcrefore left to the owner's initiative. The term of reference is
stereotyped and since it connotes both a date of birth and member-
ship of a group it is, as we shall see later, identical with the product
of the combination of what ethnologists call a clan name and an
ordinal name.
I may, on the other hand, regard rr..yself as free to name my dog
according to my own tastes. But if I select 'Médor' I shall be classed
as commonplace; if I select 'Monsieur' or' Lucien', as eccentric and
provocatiœ; and if I select 'Pelléas' as an aesthete.
1\1oreover, the name selected must also be a member of the class
of dog-names which is conceivable in the civilization to which I
belong and one which is available, if not absolutely, then at least
rclatively, that is, not chosen :ilready by my neighbour. My dog's
name will therefore be the product of the intersection of three
domains: it is a member of a class, a member of the sub-class of the
names vacant within the class, and a member of the dass formed by
my own wishes and tastes.
It is clear that the problem of the relation between proper names
and common names is not that of the relation of naming and signi-
fying. One always signifies, either oneself or someone else. It is only
here that there is a choice, rather like that open to a painter between
representational and non-representation:il art, which amounts to no
more than a choice between assigning a class to an identifiable
object or, by putting i t outside a class, making the objecta means of
classing himself by expressing himself through it.
From this point of view, systems of appellations also have their
'abstracts'. Thus the Seminole Indians form adults' names by
182
U:\'IVERSALIZATIO":\' AND PARTICULARIZATION
employingseveral seriesof a fe,velements, combined without regard

for their meaning. There is a 'moral' serics: wise, crazy, cautious,
malicious, etc.; a 'morphological' series: square, round, spherical,
elongated, etc.; a 'zoological' seri~~: wolf, eagle, beaver, puma, etc.
By means of these, by taking a term from eachseries and juxtaposing
them, they form the name 'crazy-spherical-puma' (Sturtevant,
p. 508).
The ethnographie study of persona! names has constantly encoun-

tered difficulties of a kind wcll analysed by Thomson in the Aus-
tralian example of the Wik .:\Iunkan, who inhabit the western sicle
of the Cape York Peninsula. On the one hand, proper names are
derived from totems and depend on sacred and esoteric knowledge;
but, on the other, they are connected with social personality and
are the occasion of customs, rites, and prohibitions. On both
accounts they are inextricably associated with a more complex
system of appellations, which includes the kinship terms normally
used as terms of address, and therefore profanely, and sacred names
which themselves include proper names and totemic appellations.
But even granting this distinction betv.:een sacred and profane, it
rcmains the case that proper names ( sacred) and kinship terms (pro-
fane) used as terms of address arc individual terms, while totemic
appellations (sacred) and kinship tcrms (profane) used as terms of
reference are group terms. For this reason, the sacred and profane
aspects are bound up together.
Another difficulty arises from the multiplicity of prohibitions
affecting the use of proper names. The \Vik Munkan forbid any
mention of a name or namcs for three years after the death of their
bearer, that is, until his mummified body has been burnt. There are
certain names, such as those of his sistcr or his wife's brother, which
a man may never mention. An inquirer committing the blunder of
asking them is supplied with a substitute name which means literally
'no name', 'without name' or 'the second born'.
There is a final difficulty, due to the large number of categories of
names. Among the Wik Munkan, the following must be dis-
tinguished: kinship terms, niimp kiimpan; names indicating status
or condition; nicknames, niimp yann, literally; 'name nothing'. such
as 'crawler' (for someone partly paralysed) or 'the left-handed'; and
finally true proper names, niimp. Only kinship terms are normally
used as terms of address except in periods of mouming when names
183
THE SAVAGE MIND
corresponding to the nature of the mourning are employed, the

meaning of which is: widower or widow, or alternatively, 'affiicted
by the loss of a relative', the relative being specified as a (eider or
younger) brother or sister, child, nephew or niece (parallel or cross),
or grandparent. Later on we shall encounter a parallel usage among
the tribes of the interior of Borneo.
The procedure for forming proper names is of particular interest.
Each individual has three persona! names: an 'umbilical' name,
niimp kort'n, a big name, niimp pi'in, and a little name, niimp miiny.
Big and little names are ail derived from totems or attributes of
totems and they are therefore owned by clans. Big names relate to
the head or top half of the totemic animal, little names to its legs,
tail, or the lower half of its body. Thus a man of the Bonefish clan
will have Piimpikiin, 'the man strikes' (the head), as his big name
and Yiink 'leg' (= the narrow base of the tail) as his little name,
while a woman of the same clan will be callcd Pamkotjiitta and
Tippunt (fat) 'of the belly'.
The umbilical names are the only ones which can corne from a
clan, or even sex, other than the bearer's. As soon as a child is born,
before the placenta is delivered, an authorized person shakes the
umbilical cord and at the same time a number of names are called,
first names of men of the paternal lineage, then women's names, and
finally names of men of the maternai lineage. The name spoken at
the moment the placenta is delivered becomes the child's name. The
r.ord is no doubt often manipulated to produce the desired name
(Thomson). We thus have here, as in the cases instanced above, a
method of forming names which reconciles requirements of an
objec:tive order with the play (which is partly free within the limits
of this order) of inter-persona! relations.
This apparc:utly (but falsely) 'probabilistic' technique with
respect to birth corresponds to ones recorded in other Australian
tribes on the occasion of death to determine, not the name of the
new born, but that of the supposed murderer. The Bard, Ungarin-
yin, and \Varramunga put the corpse between the branches of a tree
or on a raised platform. They lay out a circle of stones or row of
sticks on the ground directly beneath. Each unit represents a
member of the group and the guilty man will be indicated by the
exudations of the corpse ftowing towards his stick or stone. In North
Western Australia the body is interred and as many pebbles put on
the tomb as the group contains members or suspects. The pebble
184
found to be bloodstained indicates the murderer. Or again the

dead man's hairs are tugged one by one and each time the name of a
suspect is mentioned: the first haïr to corne out identifies the
murderer (Elkin 4, pp. 305-6).
It is clear that ail these procedures are formally of the same type
and that they have a striking characteristic which they share with
the other systems of proper names of societies with finite classes.
It was shown above that in such systems (which no doubt illustrate
a general state of affairs) names always signify membership of an
actual or virtual class, which must be either that of the persan named
or of the persan giving the name, and that ail the differences between
names attributed in accordance with a rule and invented names can
be reduced to this slight distinction. It is worth adding that this
distinction does not correspond, except perhaps superficially, to
that drawn by Gardiner between 'disembodied' and 'embodied'
names, the former being chosen from a compulsory and limited list
(like that of the Saints of the Calendar) and so borne simultaneously
and successively by a large number of individuals, and the latter
given to a single individual like Vercingetorix and Jugurtha. The
former seem in fact to be of too complcx a nature to be dcfinablc by
the only characteristic retained by Gardiner. They class the parents
who selected their children's names in a milieu, in a period, and in a
style; and they class their bearers in various wa ys : in the first place,
because a John is a member of the class of Johns, and secondly,
because every christian namehas a consciousor unconscious cultural
connotation which parades the image others form of its bearer, and
may have a subtle influence in shaping his personality in a positive
or negative way.*
Now, ail this could also be shown in the case of 'embodied' names
did we not lack the ethnographie background. The name Vercinge-
torix only appears to be confined to the victor at Gergovia because
we know so little about the Gauls. Gardiner's distinction does not,
therefore, concern two types of names but two situations in which
the observer faces either the system of names of his own society or
that of a society foreign to him.
• ' ... children's names were chosen by their parents ... and scholars have
frequently honoured colleagues by naming discoveries after them. But the choice
involved has often not been an entirely arbitrary one. Parents are guided by
social and religious traditions, scholars by a rule of priority. Everyone reveals the
nature of his preoccupations and the limits of his horizon by his choice' (Briindal,
p. 230).
THE SAVAGE MIND
This said, it is easier to extract the principle underlying the Wik

i\fonkan system of n<!.mes. They construct people's names in a way
analogous to that in which we ourselves construct species names. To
identify an individual, they first combine two class indicators, a
major and a minor (the 'big' and 'little' names respectively). This
set already produces the dual effect of attesting the bearer's member-
snip of a totemic group, by evoking meanings well-known to be its
exclusive property, and of circumscribing the position of the indivi-
dual within the group. The combination of the big and little names
is not itself individuating. It delimits a sub-set of which the bearer
of the name, along with others provisionally defined by the same
combination, is a member. I t is therefore the 'umbilical' name which
completes individuation, but the principle on which it is based is
quite different. For one thing it can be either a 'big' or a 'little' name
(of the same or another clan), either a masculine or a feminine name
(irrespective of the bearer's sex). Then again, its attribution is a
function not of a system but of an event: the coïncidence of a
physiological effect (theoretically independent of man's will) and a
point in an enumeration.
Let us now compare this trinome with those of scientific botany
and zoology. Consider in botany: Pscilocybe mexicana Heim, or in
zoology: Lutrogale perspicilata maxwelli. The first two terms of each
trinome assigns what is under consideration to a class and sub-class
belonging to a preordained set. But the third term, which is the
discoverer's name, completes the system by recalling an event. It is
a series, not a group, term.
There is no doubt a difference: in scientific trinomcs the dis-
coverer's name adds nothing to the identification, which is already
completed by the first two terms; it is merely a tribu te to their
author. This is not, however, quite accurate. The statistical term
has a logical, as well as a moral, fonction. It brings into evidence the
system of distinctions employed by the author in question or one of
his colleagues and so enables the specialist to effect the transforma-
tions necessary to resolve problems of synonymy: to know, for
instance, that Juni'perus occidentalis Hook and Juniperus utahensis
Engelm., which might be thought to be different without the dis-
coverer's or sponsor's name, are in fact the same thing. The fonction
of the statistical term in scientific taxonomies is consequently
exactly the reverse of that which it fulfills among the \Vik Munkan.
It makes it possible to associa te, not dissocia te; instead of attesting
186
the perfection of a unique way of making distinctions, i t brings into

evidence a plurality of possible ones.
The case of the \Vik l\Iunkan i~.only of particular demonstrative
value because of the strangeness of the technique conceived by the
natives, which throws a harsh light on the structure of the system.
But this structure is readily found in the societies from which ail
our examples were taken. Among the Algonkin, for instance, a
complete persona! name consists of three terms :* a name derived
from the clan appellation, an ordinal name (expressing the order of
birth in the family), and a military title, that is, in this case, one
'mechanical' and two 'statistical' terms of different power. There
are more military titles than ordinal names and the probability of
the same combination being repeated in the case of two distinct
people is ail the smaller because, although the former derive from a
group which is obligatory as a group, the choice exercised among
ail the possible names by the giver of the name will be dictated,
among other considerations, by the need to avoid duplications. This
is a good opportunity to make the point that the 'mechanical' or
'statistical' character is not intrinsic: it is defined in relation to the
pe1 !'-on of the donor and that of the bearer. The name derived from
the clan appellation identifies the bearer without question as a
mèmber of a given clan but the way it is chosen from a list is governed
by complex historical conditions: names currently vacant, the
donor's personality and intentions. Conversely, 'statistical' terms
unequivocally define an individual position in the system of
natal statuses or in the military hierarchy; but actual occupation of
these positions is an outcome of demographical, psychological, and
historical circumstances, that is to say, the future bearer is not
determined objectively.
This impossibility of defining proper names otherwise than as a
means of allotting positions in a system admitting of several dimen-
sions is also brought out by another example, drawn from contem-
porary societies. For the social group taken as a whole, the second
term, in names like John Smith and John Robinson, denotes the
class and the first the individual. John Smith belongs, first, to the
class Smith and in this class he occupies an unequivocal position as
John. Within the class Smith, he is John Smith as distinct from
" Two terms among the Lacandons of l\Iexico, who spcak the l\Iayan languagc.
They form names by the employment of a binome composcd of an animal name
and an ordinal name (Tozzer, pp. 42-3 and 46-7).
THE SAVAGE MIND
Peter Smith, Andrew Smith, etc. So little is i t a case of a 'proper'

name that, within a more restricted group, the logical relation
hetween the terms is reversed. Suppose there is a family whose
members ail address each other by their christian names in the usual
way and where a brother and brother-in-law both happen to have
the same christian name, John. The ambiguity will be dispelled by
the discriminative apposition of the patronym to the christian name.
Thus when one member of the family sa ys to another, 'John Smith
phoned', he no longer in fact refers to the same binome: the pat-
ronym has become a nickname. For the members of the family in
question there is, first, a class of Johns, within which 'Smith' and
'Robinson' secure individuation. The fonctions of the binomial
terms are reversed according to whether they are seen in the context
of the administration or of a particular social grou p.
Given that the same term can, depending only on its position in a
context, play the part either of a class indicator or of an individual
determinant, it is fruitless to enquire, as many ethnologists have
clone, whether the appellations in use in this or that society really
constitute proper names. Skinner admits this in the case of the Sauk
but doubts it in the case of their neighbours the '.\1enomini, whose
names seem rather to be honorific titles, limited in number and to
which an individual accedes for his lifetime without power to trans-
mit them to his descendants (Skinner 2, p. 17). Similarly, among the
Iroquois:
Clearly the individual name ... is only to a very limited extent com-
parable to our persona! name. It must rather be conceived of as a sort of
ceremonial designation, and also as a more intimate expression of one's
membership in a clan than is involved in the association with a clan
name (Goldenweiser, p. 367).
As for Wik :\Iunkan proper names:
Although I have called them 'persona! names' they are really group
names and signify membership of, and solidarity with, a totemic group
(Thomson, p. r 59).
These resen·ations are due to the fact that the list of names which
are the property and prerogatiœ of each clan is often restricted and
that two people cannot bear the same name simultaneously. The
Iroquois have 'guardians' to whom they entrust the task of remem-
bering the repertoire of clan names and who always know which
i88
u:-:1vERSALIZATION AND PARTICULARIZATION'
names are anilable. \Vhen a child is born the 'guardiar..' is sum-

moned to say which names are 'free'. Among the Yurok of Cali-
fornia, a child can remain with01Jt a name for six or seven years,
until a relative's name becomes vacant through the death of its
bearer. On the other hand, the taboo on the dead man's name cornes
to an end after a year if a young member of the lineage puts it back
into circulation.
Sorne names seem still more perplexing: these, like the ones given
in Africa to twins or to the first child to survive after a series of still-
births, assign some individuals a place in a rigid and restricted taxo-
nomie system. The Nuer reserve the names of birds which fly low,
such as guinea fowl, francolin, etc., for twins. In effect, they regard
twins as creatures of supernatural origin like birds (Evans-Pritchard
2, discussion in Lé\'i-Strauss 6 ). The Kwakiutl of British Columbia
express an analogous belief in their association of twins with fish.
Thus the names Salmon-Head and Salmon-Tai! are reserved for
children whose birth immediately precedes or succeeds that of
twins. Twins themselves are believed to be descended cither from
ulachan (if they have small hands) or from Oncorhynchus kisutch
(Silver-Salmon) or from Oncorhynchus nerka (Sockeye-Salmon).
The diagnosis is made by an old man who is himself a twin. In the
first case he names the male twin 1\faking-Satiated and the female
twin Making-Satiated-Woman; in the second case the names are
Only-One and Abalone-Woman respectively; and in the third
Head-Worker and Head-Dancer (Boas 4, part I, pp. 684-93).
Among the Dogon of the Sudan there is a very strict method of
allotting proper names: each individual's position is plotted on the
basis of a genealogical and mythical mode] where each name is
linked with a sex, a lineage, and an order of birth and with the
qualitative structure of the group of siblings to which the individual
belongs: himself a twin, first- or second-born before or after twins,
boy born after one or two girls, or vice versa, boy born between two
girls or vice versa, etc. (Dieterlen 3).
Finally, one is often hesitant to regard as proper names the ordinal
names found among most of the Algonkin and Sioux, among the
Mixe (Radin 2), the Maya (Tozzer) and in the south of Asia (Bene-
dict), etc. Let us confine ourselves to one example, that of the
Dakota, among whom the system is particularly developed, with the
following names corresponding to the order of birth of the first
seVèn girls and the first six boys :
THE SAVAGE MIND
GIRLS BOYS
!. \Vino'ne Tcaske'
2. Ha'pe Hepo'
3. Ha'psti Hepi'
4. \Viha'ki \Vatca'to
5. Hapo'nA Hake'
6. HapstinA 'l 'atco'
7. Wihakc'da
(\\'allis, p. 39).
Terms which replace propcr namcs at different stages of initiation
can be put in the same category. The Australian tribes in the north
Dampier Land have a series of ninc names gi\'en to novices hefore
the extraction of tceth, thcn bcfore circumcision, before the ritual
bleeding, etc. The Ti,vi of the ~Ielville and Bathurst islands off
.i\ orth Australia give novices special namcs according to thcir grade.
Thcre arc scven mcn's names covering the pcriod from the fifteenth
to the twenty-sixth year and seven women's names going from the
tenth to the t\venty-first ycar (Hart, pp. 286-7 ).
However, the problcms which arise in thesc cases are no different
from that raised by the custom with which wc arc acquainted in our
society whcreby an eldest son is giYCn his patcrnal grandfather's
christian namc. 'Grandfathcr's name' can also be regarded as a title
which is both obligatory and exclusive. There is an imperceptible
transition from namcs to titles, which is connected not with any
intrinsic property of the terms in question but with their structural
rolc in a classificatory system from which it would be vain to claim
to separate them.
CHAPTE.R SEVEN
THE I~DIYIDUAL AS A SPECIES
The naming system of the Penan, who are nomads of the interior of
Borneo, enablcs us to givc a more precisc account of the relation
betwcen the terms to which wc should be inclined to reserve the
title of propcr namc and othcrs which scem at first sight to be of a
differcnt kind. Dcpcnding on his age and family situation, a Penan
may be dcsignatcd by thrcc sorts of tcrms: a persona! name, a
tcknonym ('father of so-and-so', 'mothcr of so-and-so') and, finally,
what one feels like calling a necronym, which expresses the kinship
relation of a deceascd relative to the subjcct: 'father dead', 'niece
de::ad', etc. The western Penan have no Jess than twcnty-six distinct
necronyms, corresponding to the degrcc of kinship, relative age
of the deceased, sex and the order of birth of children up to the ninth.
The ru les gO\·erning the use of thesc names are of surprising com-
plexity. Simplifying a great deal, we can say that a child is known by
its propername until one of his ascendants dies. If it is a grandfather
who dies, the child is then called Tupou. If his father's brother dies
he becomes Ilun and remains so until another relative dies. He then
receives a new name. A Penan may th us pass through a series of six
or se,·en or more necronyms before he marries and has children.
At the birth of their first child a father and mother adopt a
Teknonym expressing their relation to the child whose name forms
part of it. Thus: Tama Awing, Tinen Awing, 'father (or mother) of
Awing'. Should the child die, the teknonym is replaced by a nec-
r0nym: 'Eldest child dead.' When the next child is born a new
teknonym takes the place of the necronym, and so on.
The position is further complicated by the special rules relating
to s!blings. A child is called by his O\Vn name if ail his brothers and
191
THE SAVAGE MIND
sisters are alive. If one of them dies he takes on a necronym 'eider

(or younger) sibling dead', but the necronym is abandoned at the
birth of a new brother or sister and hc uses his own name again
(::\eedham !, 4).
There are man y obscurities in this description and it is difficult to
see how the various rules affect each other in spi te of the fact that
they appear to be functionally related. The system as a whole is
definable by three types of periodicity: in relation to his ascendants,
an individual passes from necronym to necronym; in relation to his
siblings he passes from autonym (the term it is convenient to use for
proper narnes in a system of this kind) to necronym; and in relation
to his children he passes from teknonym to necronym. But what is
the logical relation between thcse three types of terms and again
between these three types of periodicity? Teknonyms and nec-
ronyms refer to a kinship tie and are therefore 'relational' terms.
Autonyms do not have this characteristic and can be opposed to
them from this point of view: an autonym only determines a 'self'
by contrast to other 'selves'. This opposition between 'self' and
'other', implicit in the autonym allows in return a distinction
between the teknonym and the necronym. The former contains a
proper name ( which is not that of the subject and it can be defined as
expressing a relation to an 'other' self. The necronym contains no
proper name at ail and consists in the statement of a kinship relation,
which is that of an unnamed 'other' with a 'self', equally unnamed.
It can therefore be defined as an 'other' relation. And finally, this
relation is negative since the necronym mentions it only to declare
it extinct.
The relation between autonym and necronym emerges clearly
from this analysis. 1t is one of inverted symmetry:
AUTONY'.\1 NECROl'YM
_ _ _ _ __I
1
relation
present ( +) or 1
absent(-): +
opposition
between self
( +) and other
(-): +
THE INDIVIDUAL AS A SPEC!ES
A first conc.:lusion becomes apparent at the same time: the autonym,

which we have no hesitation in considering as a proper name, and
the necronym, which has the features of a simple class indicator,
belong in fact to the same group. One passes from one to the other
by means of a transformation.
Let us now turn to the teknonym. \Vhat is its relation with the
other two types of term, in particular with the necronym? It is
tempting to say that the teknonym connotes the entry into life on an
other self, while the necronym refers to the departure of an other
self into death, but this is too simple for it does not explain why the
teknonym mentions the self of an other (autonym being incorpor-
ated in it) while the necronym may be reduced to a negation of the
other relation without reference to a self. There is, therefore, no
formai symmetry between the two types.
In the study which I have used as the point of departure for this
analysis, ~eedham makes the interesting comment:
Something very slightly similar to death-names is seen in the older
English usage of 'widow' as a title .. ., the contemporary French and
Belgian use of 'veuve', and similar usages in many other countries of
Europe. But these are so far from death names in nearly cvcry respect
that they can provide no pointers to understanding (Needham I, p. 426).
This is giving up too soon. To have perceived the implications ofhis
observation, ~eedham nced only have noticed, in the cxamplcs he
quotes, the connection between the right to a necronym and the
prior possession of a name in every way comparable to a teknonym.
In traditional French usage 'veuve' (widow) incorporates a proper
name but the masculine 'veuf' (widower) or 'orphelin' (orphan) does
not. Wh y this exclusivism? The patronym belongs to the children in
their own right; it is, we say, a classifier of lineages in our societies.
Children's relation to a patronym is therefore not affected by the
death of parents. This is ail the more true in the case of men, whose
relation to their patronym remains immutable whether they be
unmarried or widowed.
lt is otherwise with women. If a woman's husband dies she
becomes 'widow so-and-so'; but this is only because during her
husband's lifetime, she was 'the wife of so-and-so', in other words,
had already relinquished her autonym in favour of a term expressing
her relation to an other self. And this is precisely how we defined the
teknonym. In this case, of course, 'teknonym' would not be the
appropriate word. We could coin the term andronym (Greek ixuµ.p,
1 93
THE SAVAGE MIND
husband) to presen·e the par aile! but there i s little point in this as the
identity of structure is immediately apparent without recourse to
neologism. In French usage the right to a necronym thus depends
on the possession of a name analogous to a teknonym: because my
self is ddined by my relation to an other self my identity can be
presen·ed at this other's death oniy by retaining this relation, un-
changed in form but governed from now on by a negative sign.
'\Vidow Smith' is the \Vife of a Smith who is not extinct but who
now exists only in his relation to that other who defines herself
through him.
I twill be objected that in this example bath terms are constructed
in the samc way, by adding a kinship relation to a patronymic deter-
minant, while, as I have pointed out, the necronym among the
Penan contains no propcr name. Before resolving this difficulty, let
us consider the series of siblings where autonym and necronym
alternate. \Vhy autonym and not a term analogous to t'!knonym,
say a 'fratronym' of the type 'brother (or sister) of so-and-so'? The
answer is easy: the persona! name of the child who has just been
barn (whose brothers and sisters are thereby reliend of their nec-
ronyms) is brought into play elsewhere. It is used to form the
parents' teknonym and they have in some way appropriated it for
inclusion in the particular system by which they are defined. The
name of the last-born is thus scparated off from the series of siblings
anc!, as the othcr siblings can be defined neither through it nor
through that of their dead brother or sister (since one is no longer
in the 'key of death', as it were, but in the 'key of life'), they fall back
on the only course left open to them: the resumption of their own
name which is also their proper name. But, it must be emphasized,
this is only in default of 'other' relations which haœ either become
unavailable because they are turned to adifferent use ornotrelevant
becausc the sign governing the system has changed.
The clarification of this point !caves us with only two problcms:
the use of teknonyms by parents, and the absence of proper names
in necronyms, a problem wc came across just now. At first sight the
former seems to raise a question of content and the second one of
form. But they are in fact a single problem amenable to one and the
same solution. T lie names of the dead are never mentioned, and this
sufficcs to cxplain the structure of the necronym. So far as the
teknonym is concerned, the inference is clear: the reason why
parents may no longer be called by their namc when a child is barn
1 94
THE INDIVIDUAL AS A SPECIES
is that they are 'dead' and that procreation is conccivcd not as the
addition of a ncw being to those who exist already but as the substi-
tution of the one for the others. ,.
The Tiwi custom of forbidding the use of proper names during
initiation and child-birth is also to be understood in this way:
The hirth of a child is, to a native, a most mysterious affair and the
woman is regar<led as being intimately in touch with the spirit world.
Hence her name as part of hersclf is investe<l with a ghostly character and
this is expressed hy the tribe in treating her husband as if she did not
cxist, as if she were dead in fact and for the time being no longer his wife.
Shc is in touch with the spirits and the result will be a child for her
husband (Hart, pp. 288-9).
A remark of::-\eedham's suggcsts a similar intcrprctation in the case
of the Penan. The teknonym, hc says, is not honorific and not to have
a child is nota matter of shame.
If you have no child, infonnants say, ... it is not your fault. You arc
sorry because thcrc is no one to replace you, no one to remember your
namc ... Ilut you arc not ashamcd. \\"hy should you be? (l.c., p. 417).
The couvade can be explained in the samc way for it 'vould be a
mistakc to suppose that a man is taking the place of the woman in
labour. The husband and wifc somctimcs have to takc the samc
precautions because they arc idcntificd with the child who is subject
to great dangers during the first wecks or months of its life. Some-
times, frcqucntly for instance in South America, the husband has to
take cven greater prccautions than his wife because, according to
native thcorics of conception and gestation it is particularly his
person which is identificd with that of the child. In ncithcr cvcnt
docs the father play the part of the mother. He plays the part of the
child. Anthropologists arc rarely mistaken on the first point; but
they yct more rarcly grasp the second.
Thrcc conclusions can be drawn from this analysis. In the first
place, proper names, far from constituting a catcgory apart, form a
group with othcr tcrms which differ from thcm evcn though thcy
arc unitcd with thcm by structural relations. The Penan themselvcs
think of these terms as class indicators: they speak of 'cntering into'
a necron ym, not of taking or receiving it.
Sccondly, propcr namcs occupy a subordinate position in this
complcx system. It is really only children who overtly bear their
names, either because thcy are too young to be structurally qualified
by the family and social system or becausc the mcans of qualification
1 95
THE SAVAGE l\llND
ha \·e, for the time hcing, been suspended in favour of their parents.
P;-oper namcs thus undergo a truly logical devaluation. They are
the mark of being 'unclassed' as candidates for a class, of being
temporarily obliged to definc themseh-es either as unclassed (as in
the case of siblings resuming the use of their autonyms) or alter-
nati,-cly by their relation to an unclassed person (like parents when
they assume a teknonym). \Vhen, however, death causes a breach in
the social fabric, the individual is, as it ,.,,·ere drawn in. Thanks to
the nccronym, which has an absolute logical priority over other
forms, his proper name, a mere place in the queue, is replaced by a
position in the system, which from the most general point of view,
can thcreforc be considered as formed of discrete quantified classes.
The proper name is the reverse of d.e necronym, of which the
teknonym in turn presents an inverted image. The case of the Penan
is to ail appearanccs the opposite of that of the Algonkin, froquois
and Yurok. Among the former a person must await a relative's
death to be rid of the na me he bears, white often among the latter
he must await it to succeed to his name. But in fact there is as great a
logical devaluation of names in the latter case as there is in the
former:
The individual name is never used in either direct address or indirect
reference to relatives, the relationship term doing service in all such cases.
Even when addressing a non-relative, the individual name is very seldom
used, the form of address consisting in a relationship term, according to
the relative age of the speaker and the person addressed. Only when non·
relatives are referred to in conversation is it customary to use the indi-
vidual name, which even then will not be used if the context plainly
indicates the person referred to ((ioldenweiser, p. 367).
So, among the Iroquois too, and in spite of the differences pointed
out, a person is left uncla5sed only when there is no alternative.•
•Ta avoid using propcr names, the Yurok of California have conccived a
system or appellations composed of a root corresponding to a place of residence -
a village or house - and a suffix, which differs for men and women, describing
their marital status. The men's names refer to the wife's place of birth and the
women's to the husband 's place of hirth. The suffix of the name indicates whether
it is a case of a patrilocal marriage hy purchase, a matrilocal one or a f ree union,
whcther the marriage has hcen dissoked through the dcath of hushand or wife
or hy divorce, etc. Other affixes, entering into the names of children and the
unmarried, refer to the place of hirth of their Ji,·ing or deceased mother, or their
deceascd father. Thus the only names employe<l arc of one of the following types:
:\1arried to a woman of---; ::\Iarried to a man of--; Has a 'simi' husband in
the house of hirth o f - - ; Is 'half' married to a woman o f - - ; Widower
belonging t a - - ; Divorccd from a woman (or man) o f - - ; \Voman o f - -
THE 1:-;DIVIDUAL AS A SPECIES
Ali sorts of beliefs have been invoked to explain the very common
prohibition on pronouncing the names of the dead. These beliefs
are real and well authenticated 1?-ut the question is whether they
should be regarded as the origin of the custom, as one of the factors
which have contributed to reinforce it or perhaps even as one of its
consequences. If the explanations 1 have given are correct, the pro-
hibition on the names of the dead is a structural property of certain
systems of naming. Proper names are either already class operators
of alternatively they provide a temporary solution for those awaiting
classification. They always represent classes at their most modest.
In the limiting case, as among the Penan, they are no more than the
means, temporarily unclassed, of forming classes or again, as it were,
bills drawn on the logical solvency of the system, that is on i1s
rliscounted capacity to suppl y the crcditor with a class in due course.
Only newcomers, that is, the children who are born, raise a problem:
there they are. Any system which treats individuation as classifica-
tion (and 1 have tried to show that this is always so) risks having its
structure called in question every time a new member is admitted.
There arc two types of solution to this problem, and intermediate
forms between them. If the system in question consists of classes of
positions, it has only to command a reserve of unoccupied p9sitions
sufficient to accommodate ail the children born. The available
positions bcing always more numerous than the population, syn-
chrony is protected against the vagaries of diachrony, at least in
theory. This is the Iroquois solution. The Yurok are Jess farsighted.
Among them, children have to wait their turn; but as thcy are
nevertheless assured of being classified after a few years, they can
remain temporarily undifferentiated while awaiting a position in a
class, which the structure of the system guarantees them.
Everything is different when the system consists of classes of
relations. It is then no longer the case that one individual ceases to
exist and another replaces him in a position labelled by means of a
proper name which outlasts any particular person. For the relation
itself to become a class term, proper names, which present the terms
related as so many distinct entitles, have to be eliminated. The ulti-
mate units of the system are no longer single member classes with a
train of successive living occupants, but classed relations between
who allows a man to li\e with her, has a lover or illegitimate children; His father
was o f - - ; Their late mother was - - ; Unmarried person o f - - ; etc.
(\\'aterman, pp. 214-18; Kroeber in: Elmcndorf and Kroeber, pp. 372-4, n. 1)
1 97
S-TSM
THE SA\'AGE ;-.11:-;n
the dead or even those who are soin effect (parents are described as
deaJ by contrast with the life they have created) and the living,
whether really li\·ing or in etfect so (new born children who have a
proper name so that their parents may be defined in relation to them
until the real death of an ascendant allows them in turn to be
defined in relation to him). In these systems, the classes arc thus
composed of different types of dynamic relations associating entries
and departures, while among the Iroquois and other societies of the
same sort, they are founded on a collection of static positions which
may be vacant or occupied.*
The prohibition on the names of the dead does not therefore raise
a separate problcm for anthropology. A dead persan !oses his name
for the same reason that - among the Penan - a living persan !oses
his; \\·hen the living Penan enters the system he assumes a nec-
ronym, that is te say, he becomes one of the terms in a relationship,
of which the other - since he is dead - no longer exists save in that
relation which defines a li\·ing persan with reference to him. Finally,
for the same reason as the dead !ose their names a mother and father
Jose theirs also whe:n they assume a teknonym, resolving in this way
(until the dcath of one oftheir children) the difficulty created for the
system by the procreation of a supernumerary member. The latter
rnust wait 'outside the door' as a named persan until someone's
departure allows him to make his entrance. Then two beings of
whom rme did not previous!y belong within the system and the
" Consequently, systems of relations, unlikc systems of positions whose dis-

continuous nature is e'·ident, tend rather to be continuous. 1\nother Penan usage
shows this clearly, although ::\cedham (2), who also records it, rejects an explan-
ation which in fact ,;eems vcry plausible. The reciprocal terms 'grandparent'
and 'i:rrandchild' replace the usual, doser terms for members of the same nuclear
family when one of the two pcrsons im·oh·ed is in mourning. Is not the point
that the pcrson in mourning is looked upon as ha,·ing moved O\'Cr somewhat
towards death and so furthcr away than he was frnm his closes! relatives? Death
loosens the tics of kinship network. ::\.,edham is unwilling to allow this because
he regards v.hat is in fact one prohlem as se,·eral: the mourner does not call a
son, daughter, ncphcw or niecc or their spousc 'grandchild' hecausc the same
mouming directly or indirectly affects them hut quitc simply by way of recipro-
city. Ali the exampks ::\cedham mentions confirm this except that of a young
child who suffers a slight mishap (has fallcn clown, rccei,·ed a blow or whose food
has heen stolcn hy a dod and is called on that occasion by the necronym
normally rcscn·ed for thosc who ha,·c lost a grandparent. But my account covers
this case also since the chil<l is placed metaphorically in mourning by the damage
donc to him and, due to his tender ycars, a slight injury to his well-being (actual
in the case of a fall or pot~ntial in that of a loss of food) is enough to dri,·e him
back towar<ls dcath.
THE I:-<DIVIDUAL AS A SPECIES
other ceases t o de s o are merged in one of the classes of relations of

which the system is col!1posed.
Sorne societies jealously watch.over their names and make them
last practically for e\·er, others squander them and destroy them at
the end of every individual existence. They then get rid of them by
forbidding them and coin other names in their place. But these
apparently contradictory attitudes in fact merely express two aspects
of a constant property of systems of classification: they are fini te
and inflexible in form. By its mies and customs, each society, to
impose a structure on the continuous flux of generation, does no
more than apply to a rigid and discontinuous grid, and a slight shift
in its logic is enough to secure this in one position or the other: either
the system of propcr names fonns the finest mesh of the filter of
which it is, consequcntly, an esscntial part; or alternatively, it is left
üutsidc while retaining ail the same the function of individuating
the continuous and th us setting up in a formai manner d iscontinuity,
which can then be seen as a prcliminary condition of classification.
In bothcascs, too, the dcad as their distance from the grid increases,
Jose their names which are either taken by the living as symbols of
positions which must always be fi lied* or are clone away with under
the impact of the same movement which, at the other end of the
grid, extinguishes the names of the living.
The Tiwi system of naming, to which I have referred several
times, is intermediate between these two forms. In the first place,
proper names are meticulously confined to a single bearer:
It is impossible for any two people to have the same name ... Although
the Tiwi number ncarly cleven hundrcd people at the present time, and
each one of these has on an average threc names, a careful study of these
three thousand three hundred namcs failcd to reveal any two as being
identical (Hart, p. 281 ).
This proliferation of na mes is further increased by the number and

variety of prohibitions relating to them. These prohibitions extend
• ln the Fox myth of the origin of de a th, the persan in mouming is told: Now
this is what you are to do. You, i.e. you and the deceased, must always rclease
each other (i.e. hold an adoption feast). Then the soul of the dead will safely and
speedily go yondcr. Y ou must adopt somL"One. And you must think exactly the
same of that toward them. And you will be related to him exactly (as you were to
the dead). That is the only way the soul of your relative may depart safely and
speedily (!Vlichelson, I, p. 411 ). The text is eloquent about the fact that, in this
case too, the quick drive out the dead.
THE SAVAGE MIND
in two directions. As f mentioned in an example,* they attach in the

first instance to all the words in current use which sound like the
names of the deceascd; and they apply not only to the latter, but
also to all the names which the deceased gave to others, whether they
were his own or someone else's children. A young child with only
one name, given hy his father, would become nameless if his father
dicd and would remain so until another name came to him from
elsewhere (I.e., p. 282). Every time a woman remarries, her husband
gives new names not just to his predecessor's children but to all the
children his wife has borne throughout her life whoever was their
father. As the Tiwi practise a form of polygyny chiefly favouring old
men, a man has little hope of marriage before the age of thirty-five
and women pass from husband to husband. This is due to the
difference in age between husband and wife which makes it very
likely that a husband will die before his wife. No one can therefore
boast a definitive name until his mother's death (id., p. 283).
A system as strange as this would remain incomprehensible were
it not for an hypothesis which suggests an explanation of it, namely,
that relations and positions are here put on the same footing. In
addition, the abolition of a relation involves in each case that of the
proper names which were a fonction of it either socially (names
bestowed by the deceased) or linguistically (words which resemble
the deceased's names). And every creation of a new relation starts a
process of renaming within the domain of the relation.
Sorne ethnographers have approached the problem of proper names

from the angle of kinship terms :
Logically, tenns of relationship may be regarded as ranking between
proper names anr\ pronouns. They occupy a position between both and
might be called individualizcd pronouns or generalized proper names
(Thumwald, p. 357).
But if this transition is likewise possible it is because from the

ethnological point of view, proper names always appear as terms
which are generalized or have a generalizing fonction. In this respect
they do not differ fundamentally from the names of species, as is
shown by the popular tendency to attribute names ofhuman beings
to birds in accordance with their respective species. In French, the
sparrow is 'Pierrot', the parrot 'Jacquot', the magpie 'IVIargot', the
200
THE IXDIVIDUAL AS A SPECIES
finch 'Guillaume', the wren 'Bertrand' or 'Robert', the water-rail

'Gerardine', the sparrow-owl 'Claude', the eagle-owl 'Hubert', the
crow 'Colas', the swan 'Godard' ... This last name also refers to a
significant social condition for iÎ:. was applied to husbands whose
wives were in labour (Witkowski, pp. 501-2).* Names of species
for their part seem to have some of the features of proper names.
Following Brondal, t Gardiner admits this in the case of the scien-
tific terminology of zoology and botany:
The name Brassica rapa easily evokes the thought of a botanist classi-
fying a number of specimens which to the lay mind arc much alike, and
to one of which he gives the name Brassica rapa, just as a parent names
his baby. \Ve have no such thought about the word turnip, and Brassica
rapa is simply the scientific name for the ordinary turnip. We may find
confirmatory support for regarding Brassica rapa as a proper name, or
at least as much more of a µroper name than turnip, in the fact that we
do not say This is a Brassica rapa, or Tlzese are fine specimens of Brassica
rapa. ln so saying we appeal to the name of any single example of the
type, whcreas in speaking of a certain vegetable as a turnip, we appeal to
the similarity of that vegetable to others of its kind. The difference of
linguistic attitude is a mere nuance, but it is a real one. In the one instance
the sound of the name, what we usually describe as 'the name itself', is
more in the forcground than in the other instance (Gardiner, p. 5z).
This account illustrates the author's central thesis that: 'proper

names are identificatory marks recognizable, not by the intellect,
but by the senses' (Le., p. 41 ). Now, wc justified the assimilation of
botanical and zoological terms to proper names by showing that in
a very large number of societies proper names are constructed in
the same way as species named in the natural sciences. A conclusion
diametrically opposed to Gardiner's follows from this. Proper names
seemed tous to have close affinities with species names, in particular
when their role was clearly that of class indicators, that is, when
• It is very significant that even so limited and simple a serics includes tcrms
from different logical sphere. 'Pierrot' can be a class indicator since one can say
'There are three pierrots on the balcony'. But 'Godard' is a term of address. As
the author of the article on this word in the Dictionary de Trevoux ( 1 732 ed.) so
excellently puts it: 'Godard is the name given to swans. When one calls them,
when one wants them to corne, one says to thcm: "Godard, Godard, corne
Godard, corne. Here Godard". Jacquot and perhaps J\largot seem to have an
intermediate role.' (Cf. Rolland, Faune, Vol. II on the human proper names
given to birds.)
t From the point of view of eternity, particular species of plants and animais
and simple substances are unique things of the same king as, for example, Sirius
or Napoleon (Brondal, p. 230).
201
THE SAVAGE MINU
they belonged to a mcaningful system. Gardiner, on the other hand,

attcmpts to explain this same analogy by the meaningless nature of
scientific terms, which he reduces, like proper names, to mere dis-
tinctive sounds. Ifhcwcrcright, we should arrive at an odd paradox.
Brassica rapa is certainly reducible to a distinctive sound so far as
the layman, unversed in Latin and botany, is concerned, but hc does
Fig 10 Brassica rapa (from Ed. Lambert,

Traite pratique de Botanique, Paris, 1883).
not know what it expresses. \Vithout extraneous information, he

could not therefore recognize this term as a proper name but only
as a word of unknown sensc, if not indecd a jlatus 'Docis. This is in
fact the situation in some Australian tribes in which totemic species
are given names from sacred language which ha,·e no animal or
plant associations for the uninitiated. If, therefore, Brassica rapa
does look like a proper name, it can do so only to the botanist, who
alone can say '\Vhat good specimens of Brassica rapa'. For the
hotanist, hmvever, what is in question is something quite different
from a distinctive sound, for he knows both the meaning of the Latin
words and the ru les of the taxonomy.
So it appears the Gardiner's account must be confined to the case
of the semi-initiated who recognize Brassica rapa as the name of a
botanical species without knowing \vhich plant is in question. In
202
THF. I:\DI\'IDUAL AS A SPECIES
spi te of the author's protestations (p. 51 ), this is a return to Ven-

dryes' bizarre idea (p. 222) that a bird's name becomes a proper
name when the obsen·er is unable to distinguish the species towhich
it belongs. Howeyer, eYerything sâld so far suggests that the connec-
tion between proper names and species names is not a contingent
one. It rests on the fact that a phrase like Brassica rapa is 'outside
discourse' in two ways: because it derh-es from scientific language
and because it is composed of Latin words. lt is therefore only \Vith
difficulty that it is included in the syntagmati..: cr.ain. Its paradig-
matic nature thus cornes to the fore. Similarly, it is because of the
paradigmatic role of proper names in a system of signs external to
the system of the language that their insertion in the syntagmatic
chain makes a perceptible break in the latter's continuity: in French,
by the absence of an article before them and by writing them with a
capital letter.
The :\anho seem to have gained a fairly clear conception of the
problems just discussed. One of their myths sets aside Gardiner's
account in ad\·ance:
One day '.\Iouse met llear and asked him if his name was Cac. This
made Bear angry and he trie<l to hit '.\louse, who ran bchin<l Bcar and set
fire to his fur. Bear coul<ln't put it out, so he gave four songs as ofîerings
to :\Iouse, begging him to put out the fire; nowadays if you carry mouse
hair with you, no bear will touch you (a<lapted from Haile-\Vheelwright,
p. 46).•
The myth humorously underlines the difference between species

na mes and distinctiw sounds. One of the reasons for this difference
so far as the :\ayaho are concerned, lies in the fact that names of
species are, at least in part, proper names. In thestory quoted above
"\Iouse offends Bear because he addresses him incorrcctly and using
a ridiculous word. );ow, the botanical terms of the Navaho (whose
Zoological rncabulary has not been so well studied) generally consist
0f a trinome, the first element of which is the name proper, the
5econd describes the plant's use and the third its appearance . .!\fost
oeople apparently know only the descriptiYe term. The 'real name'
is a term of addrcss which priests use when they speak to the plant:
a proper name therefore and one which it is essential to be well
,. [Ed. note.] The usage of capitals and articles in the tcxt is quite haphazard
and corresponds neither to any fcaturc in the :\avaho languagc nor to any prin-
ciplc of English usage. ln this regard English is similar to French usage and the
text has thereforc bccn adaptcd for the sakc of the clarity of the illustration.
203
THE SAVAGE l\IlND
acquainted with and to pronounce correctly (Wyman and Harris;

Leighton).
\Ve do not ourselves use scientific nomenclature for conversing
with plants and animais. \Ve do however readily give animais, and
borrow from plants, some of the names serving as terms of address
between people: girls are sometimes called Rose or Violet and con-
versely sevcral animal species are allowed to share christian names
commonly borne by men or women. But why should it, as we have
already seen, particularly be birds who profit from this liberal atti-
tude? They are further removed than dogs from men in their ana-
tomical structure, theirphysical structure and their mode of life, and
human christian names cannot be given to dogs without causing
uneasiness or even mild offence. The explanation is already con-
tained in what has just been said.
Birds are given human christian names in accordance with the
species to which they belong more easily then are other zoological
classes, because they can be permitted to resemble men for the very
reason that they are so different. They are feathered, winged, ovi-
parous and they are also physically separated from human society
by the element in which it is their privilege to move. As a result of
this fact, they form a community which is independent of our own
but, precisely because of this independence, appears te us like
another society, homologous to that in which we live: birds love
freedom; thcy build themsclves homes in which they live a family
life and nurture their young; they often engage in social relations
with other members of their species; and they communicate with
them by acoustic means recalling articulated language.
Consequently everything objective conspires to make us think of
the bird world as a metaphorical human society: is it not after ail
literally parallei to ir on another level? There are countless examples
in mythology and folklore to indicate the frequency of this mode of
representation. Consider for instance the comparison mentioned
above which the Chickasaw Indians make between the society of
birds and a human community. *
•Cf. above, p. 118.
::\ly account is inversely confirmed by the cases of those animais which are
likewise given human christian namcs, although they are not birds: 'Jean Lapin',
'Robin Mouton', 'Bernard (or :'\1artin) !'Ane', 'Pierre (or Alain) le Renard',
'Martin !'Ours', etc. (Sebillot, II, p. 97, III, pp. 19-20). These animais do not
form a natural series: some are domestic, some wild; some are herbivorous,
othe1,; carnivorous; some are lo\'ed (or scorned) and others feared. \Vhat we have
THE l1'DIVIDL1AL AS A SPECIES
Now, this metaphorical relation which is imagined between the

society of birds and the society of men, is accompanied by a pro-
cedure of naming, itself of a metonymical order (we need not in this
work regard oursekes as bound by grammarians' refinements, and
I shall not treat synecdoche - a species of metonymy according to
Littré - as a distinct figure of speech): when species of birds are
christened 'Pierrot', ':\fargot' or 'Jacquot', these names are drawn
from a portion which is the preserve of human beings and the
relation of bird names to human names is th us that of part to whole.
The position is exactly the reverse in the case of dogs. Kot only
do they not form an independent society; as 'domestic' animais they
are part of hum an society, although with so low a place in it that we
should not dream of following the example of some of the Austra-
lians and Amerindians in designating them in the same way as
human beings - whether what is in question are proper names or
kinship term.* On the contrary, we allot them a special series:
'Azor', ':\Iedor', 'Sultan', 'Fido', 'Diane' (the last of these is of
course a human christian name but in the first instance conceived
as mythological). Kearly all these are like stage names, forming a
series parallel to the names people bear in ordinary life or, in other
words, metaphorical namcs. Consequently whcn the relation
between (human and animal) species is socially conceived as meta-
phorical, the relation betwecn the respective systems of naming
takes on a metonymical character; and when the relation between
species is conceived as metonymical, the system of naming assumes
a metaphorical character.
Let us now consider another case, that of cattle, the social position
of which is metonymical (they form part of our technical and
economic system) but different from that of dogs in that cattle are
more overtly treated as 'abjects' and dogs as 'subjects' (this is sug-
gested, first, by the collective name we use to designate the former
and, secondly, by the taboo on the eating of dogs in our culture; the
situation is different among the African pastoral peoples who treat
cattle as we treat dogs). Now, the names given to cattle belong to a
different series from birds' or dogs'. They are generally descriptive
here is therefore an artificial system, based on characteristic oppositions between
temperaments and ways of life and tcnding to rcconstruct a small-scalc mode) of
human Society in the animal kingdom. The Roman de Renart offers a typical
illustration of this.
• Cf. above, pp. 167-8; and even Jess named human beings in terms of them -
as the Dayak do: faLhcr {or mother) of this or that dog ... (Geddes).
205
THE SAVAGE :\IIND
terms, refcrring to the colour of their coats, their bearing or tempera-

ment: 'Rustaud', 'Rousset', 'BI anchettc', 'Douce', etc. (cf. Lévi-
Stranss 2, p. 280 ). These names often ha\·e a metaphorical character
hut they ditfer from the names gi\·en to dogs in that they arc
epithets coming from the syntagmatic chain, while the latter corne
from a paradigmatic scries; the former th us tend to derive from
speech, the latter from language.
Finally, let us consider the names gi\'en to horses - not ordinary
horses whose place approxima tes more or Jess closely to that of cattle
or that of dogs according to the class and occupation of their owner,
and is made eYen more uncertain hy the rapid technological changes
of recent times, hut racehorscs, whose sociological position is clearly
distinguishable from the cases already examined. The first question
is how to define their position. They cannot be said to constitute an
independent soci~ty after the manner of birds, for, they are products
of human industry and thcy are born and li\'e as isolated individuals
juxtaposed in stud farms deYised for their sake. On the other hand,
they do not form part of human society either as subjects or as
objects. Rather, they constitute the desocialized condition of exist-
ence of a pri Yate society: that which lives off race-courses or fre-
quents them. Another differencc, in the system of naming, corres-
ponds to these, although twc rcsen·ations must be made in drawing
this comparison: the names given to racehorses are chosen in
accordance with particular ru les which ditfer for thoroughbreds and
half-breds and they display an eclecticism which draws on learned
Iiterature rather than or<!l tradition. This said, there is no doubt that
there is a significant contrast between the names of racehorses and
those of birds, dogs or cattle. They are rigorously indi\'idualized
since, as among the Tiwi, two indi\'iduals cannot ha\'e the same
name; and, although they share with the names ginn to cattle the
feature of being formed by drawing upon the syntagmatic chain:
'Ocean', 'Azimuth', 'Opera', 'Belle-de-i\uit', 'Telegraphe', 'Luci-
ole', 'Orvietan', '\Yeekend', 'Lapis-Lazuli', etc., they are dis-
tinguished from them by the absence of descripti\·e connotation.
Their creation is entirely unrestricted so long as they satisfy the
requirement of unambiguous indi\'iduation and adhere to the par-
ticular mies referred to ahove. Thus, whilc cattle are gi\'en descrip-
ti\'e names formed out of words of discourse, the names assigned to
racehorses are words from discourse which rarely, if e\·er, describe
them. The former type of name perhaps resembles a nickname and
206
THE I:'.\:DIVIDUAL AS A SPECIES
these latter perhaps merit the title of sub-names* as 1t 1s m this

second domain that the most extreme arbitrariness reigns.
To sum up: birds and dogs are relernnt in connection with human
society either because they suggest it by their own social life (which
men look on as an imitation of theirs), or alternatively because,
having no social life of their own, they form part of ours.
Cattle, like dogs, form part of human society, but as it were,
asocially, since they verge on objects. Finally racehorses, like birds,
form a series disjoined from human society, but like cattle, lacking
in intrinsic sociability.
If, therefore, birds are metaphorical human beings and dogs, metony-
mical human beings, cattle may be thought of as metonymical inhuman
beings and racehorses as metaphorical inhuman beings. Cattle are
contiguous only for want of similarity, racehorses similar only for
want of contiguity. Each of these two categories offers the com·erse
imaee of one of the two other categories, which themselves stand
in the relation of im·erted symmetry.
The linguistic equivalent of this system of psycho-sociological
differences is to be found on the plane of appellations. Bird and dog
names are derived from language. But, although they have the same
paradigmaticcharacter, they differ in that the former arc rcal, while
the latter arc conventional christi:rn names. The names of birds are
taken from the preserve of ordinary human christian names, of
which they constitute a part; while the names of dogs in effect
reproduce in its entiretya portion of names formallysimilar to human
christian names although rarely borne by ordinary human beings.
The names of cattle and horses derive rather from speech, since
they are in both cases taken from the syntagmatic chain. But the
names of cattle remain closest to it for, as descriptive terms, they
are scarcely proper names. A cow habitually said to be gentle
('douce') is called 'Douce'. The names given to cattle thus survive
as testimony of a bygone discourse, and they can at any moment
resume their function of epithets in discourse: even when one talks
to cattle, their character as objects never allows them to be anything
but v.:hat is spoken about. The names of racehorses are 'within dis-
course' in a different fashion: not 'still in discourse' but 'made of
discourse'. To find horses' names it is necessary to take the syntag-
myatic chain to pieces and to transform its discrete units into proper
•The opposition here cornes out more clearly in the French wherc nicknamc
'surnom' is apposed to sub-name 'sous-nom' [trans. note].
THE SAVAGE MIND
names which will not be able t~ figure in discourse in any other

capacity unless the ambiguity is dispelled by the context. The differ-
ence here is due to the f act that cattle are ranked in the inhuman part
of human society, while racehorses (who, objectively speaking,
belong to the same category) primarily present the image of an anti-
society to a restricted society which owes its existence entirely to
them. The system of naming applied to them is the m.o:;t frankly
inhuman of ail and the technique of linguistic demolition employed
to construct it likewise the most barbarous.
\Vhen ail is said and done, we arrive at a three-dimensional
system:
1 1
_bird~----------------------_h9rses
--... --- ---
--------- __ ... ---
--- ---
------ ____ ... -~::::. -:.:: ----
___ ...
--- ... ____ _
--- ---
_dogs.;;...·_--_-_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ cattle_ ------
1 1
The upper line on the horizontal plane corresponds to the positive

or negative metaphorical relation between human and animal society
(birds) or between the society of men and the anti-society of horses,
and the lower to the metonymical relation between the society of
men on the one hand and dogs and cattle on the other which are
members of the former in the capacity either of subjects or of abjects.
The left-hand column on the vertical plane associates birds and
dogs which have a metaphorical or a metonymical relation with
social life. The right-hand column associates horses and cattle,
which have no relation with social life, although cattle form part of
it (metonymy) and racehorses bear a negative resemblance to it
(metaphor).
Finally, two oblique axes have to be added, for the names given
to birds and cattle are formed by metonymical appropriation ( either
from a paradigmatic set or from a syntagmatic chain), while the
names given to dogs and horses are formed by metaphorical repro-
duction (either of a paradigmatic set or of a syntagmatic chain).
\Vhat we have here is thcrefore a coherent system.
208
The systematic relations between these usages are not their only
interesting feature from our point of view.* Although borrowed
from our civilization where their place is a minor one, they off er
ready access to other usages which-are given extreme importance by
the societies which employ them. The attention we have paid to
certain aspects of our customs which some may consider futile is
justified for two reasons. ln the first place, we hope thereby to gain
a clearer and more general idea of the nature of proper names.
Secondly and above ail, we are led to inquire into the hidden
motives underlying ethnographie curiosity: the fascination exer-
cised over us by customs apparently far removcd from ours, the
contradictory feeling of proximity and strangeness with which they
affect us, stem perhaps from the fact these customs are very much
closer to our own than they appear and present us with an enigmatic
image which needs deciphering. ln any case this is what seems to be
shown by a comparison of the facts just analysed with some features
of the Tiwi naming systems which were provisionally left on one
side.
1twill be remembered that the Tiwi have an inordinate consump-
tion of na mes: first, because each person has several names;
secondly, because ail the names have to be distinct; thirdly, because
every re-marriage (and we have seen that they are frequent) in volves
giving new namcs to ail the children a woman already has; and
finally, because a pen;on's death results in a prohibition not only
of the names he has borne but also of ail those which he may have
been led to bestow in the course of this life.t How, one may ask, do
the Tiwi constantly manage to fabricate new names in these
circumstances?
Several cases have to be distinguished. A proper name can be put
back into circulation by the son of the deceased if he decides to take
it after the period during which its use is forbidden. Many names
are thus put in reserve and constitute a sort of onomastic fond which
can be drawn on at will. N evertheless, if one assumes a constant
rate ofbirths and deaths, the pool is likely to diminish on account of
the prolonged duration of the taboo, unless a sudden demographic
•This book was already completed when M.M. Houis kindly drew my atten-
tion to the work of V. Larock. Although I have not made use of it bccause its
context is so different from my own, l feel it right to pay tribute to it as the first
tentative account of names of people from an ethnographie point of view.
t Cf. abo\·e, p. 199.
THE SAVAGE :\IIND
disequilibrium should act as a compensating factor. The system

must therefore ha\·e other procedures at its disposai.
The chief of these, and there are in fact several, stems from
extending to common nouns the prohibition on proper names,
when it is noticed that the two are phonetically similar. These
common nouns, which are withdrawn from circulation in daily life,
are not however entirely destroyed: they pass into the sacred langu-
age, reseryed for ritual, where they progressively !ose their meaning,
the sacred language being by definition incomprehensible to the
uninitiated and partly freed of any meaningful function so far as the
initiated themseh-es are concerned. :'\ow, sacred words, which have
lost their sense, can be used to construct proper names by the
addition of a suffix. Thus the word matirandzjingli from sacred
language, the meaning of which is obscure, turns into the proper
name JlaterandJingimirli. This procedure is systematically em-
ployed and it has been possible to say that the sacred language is
mainly composed of wmds which have become taboo, J>ukimani, on
account of the contamination of ordinary language by the prohibi-
tion attaching to the names of the dead. Sacred language is itself
exempt from this contamination (Hart).
These facts are important from two points of view. In the first
place, it is clear that this complicated system is perfectly coherent:
proper names contamina te common nouns; common nouns, ban-
ished from ordinary language, pass into sacred language, which in
turn allows the formation of proper names. This cyclical movement
is, as it were, sustained by a double pulsation: proper names origin-
ally lacking in meaning acquire one by being attached to common
nouns and the latter relinquish their meaning on passing into the
sacred language, which allows them to become proper names again.
The system th us functions by pumping semantic charge alternately
from common no uns to proper na mes and from the profane language
to the sacred language. In the last analysis, the energy consumed
derives from ordinary language, which coins new words for the
needs of communication in proportion as the old words are taken
from it. This example is an admirable demonstration of the subsi-
diary nature of the interpretations proposed, whether by ethnolo-
gists or by the native, to explain the prohibition on the names of the
dead. For a system as well adjusted as this could not have been born
of a fear of ghosts. Rather, the latter has corne to be grafted on
toit.
210
THE 1:-;nn·rnl·AL AS :\ Sl'ECIES
That this is so seems eyen more certain when one notices that the
Tiwi system has striking analogies, on the human plane, \\·ith the
system in our own society, to which I dre\\. attention in analysing
the various ways of naming animals·, where, it is hardly necessary to
say, fear of the dead plays no part. Among the Tiwi also, the system
rests on a sort of arbitrage, exercised by means of proper names,
between a syntagmatic chain (that of ordinary language) and a para-
digmatic set (the sacred language, which is esscntially of this nature
since the words there become progressiYely unfitted to forma syn-
tagmatic chain as they Jose their meaning). In addition, proper
names are metaphorically connected with common nouns through
a positive phonetic resemblance, white sacred \rnrds are metonym-
ically connected with proper names (as means or ends) through a
negative resemblance, based on the absence or pm·erty of semantic
content.
Even if one defines it, on the most general level, as consisting in
an exchange of words between the profane and the sacred language,
through the medium of proper names, the Tiwi system clarifies
phenomena which minor aspects of our culture allowed us only to
do so muchas broach. \\"e arc better able to un<lerstand how terms
likc Brassica rapa which belong to a language 'sacred' in two respects
(being Latin and scientific) can have the charactcr of proper names,
not, as Gardiner suggests and Hart seems prepared to admit,
because they are deYoid of mcaning but because, in spi te of appear-
ances, they arc part of a whole system in which mcaning is neYer
entirely lost. \Vere this not so, the sacred language of the Tiwi
would not be a language but a conglomeration of oral gestures.
Therecan, however, be no doubt that even an obscure sacred langu-
age retains potential meaning. I shall corne back to this aspect of the
question later.
For the moment we must distinguish another type of 'sacrcd'
language which we employ, in the manncr of the Tiwi, to introduce
proper names into ordinary language, even transforming the
common nouns from the domain in question into proper names.
Thus, as we have already seen, we use flo\ver names as proper n amcs
for girls, but we do not stop at this, since the inventivc11ess of horti-
culturalists provides newly introduced flowers \vith proper names
taken over from h uman beings. This chassé-croisé has some notable
peculiarities. The names we take from flowers and give (mainly to
persons of the female sex) are common nouns belonging to ordinary
211
THE SAVA<;E :\HNIJ
language (a girl may perhaps be called Rosa but definitely not Rosa
centrfolia); those we give them in return, however, corne from a
'sacred' language: a title is added to the patronym or christian name
and )ends it a mysterious dignity. A new flower is not usually called
'Elizabeth', 'Doumer', or 'Brigitte' but 'Queen-Elizabeth', 'Presi-
dent-Paul-Doumer', ''.\Iadame-Brigitte-Bardot'. '*' Moreover, no
account is taken of the bearer's sex (in this case the grammatical
gender of the name of the flower) in naming it: a rose or a gladiolus
(fcminine and masculine respectively in French) can be given either
a man's or a woman's name indifferently. This recalls one of the
rules for the attribution of 'umbilical' names among the Wik
'.\Iunkan.t
.:\ow, these usages, whether also taken from our mvn culture or
from that of Australian islanders, clearly derive from the same group
as ail those we have been considering; we observe the same equiva-
lence between metonymical and metaphorical relations which has
seemed to play the part of common denominator between them
from the start. The names we take over from flowers to make into
proper names have the force of metaphors: fair as a rose, modest
as a violet, etc. But the names drawn from 'sacred' language which
flowers receive in exchange have the force of metonymy, in two
ways: Brassica rapa removes the self-sufficiency of cabbage-turnip
to make it a species of a genus, part of a whole. The name 'Impéra-
trice-Eugénie', given to a new variety of flower, performs a converse
transformation, since it takes place at the levcl of the signifying
instead of at that of the signified: this time the flower is designated
by means of part of a whole; not any Eugénie but a particular
Eugénie; Eugénie de .i\Tontijo not before her marriage but after it;
nota biological indiYidual but a person in a determined social raie,
etc.:j: One type of'sacred' name is thus 'metonymizing' and the other
•This tendcncy i~ already apparent in popular tradition which, when it

attributes human christian names to certain ftowers, generally inserts them in a
phrase:' Beau :'\icolas' for the rose campion, '!\tarie Cancale' for the cornftower,
'Joseph Foireux' fur the cowslip, etc. (Rolland, Flore, rnl. Il). Similarly in
English, the ftower names: 'Jack in the Pulpit', 'Jack behind the Garden Gate',
etc.
t Cf. above, p. 184.
t :\:ote the im·crsion of the cycle as compar~'Cl to the Tiwi system. Among
ourselves, the cycle gocs from ordinal)' language to proper name, from proper
namc t::> the 'sacred' languagc, to return finally to ordinary language. Ordinary
212
THE I.'."DIVIDl"AL AS A 5PECIE5
'metonymized', and this opposition holds in the case ex.amined

abon. lt will be remembered that, as well as taking names from
fl.owers, human beings gin certain of their names to birds. These
names are also 'metonymizing' for they most commonly consist of
diminuti,·es drawn from the popular language and treat the com-
munir:-· of birds (im·ersely to that of fl.owers) as equinlent, as a
whole, to a humble and well-beha,·ed sub-group of human society.
Similarly, it will readily be agreed that the metaphorical names gi,·en
to dogs and cattle place the role of the figure of speech at the lenl
of the signifying and the signified respectiYely.
Systematic as the set of naming procedures we have reviewed
may seem, they nevertheless raise a problem. These equinlent pro-
cedures, connected with each other by relations of transformation,
opera te at different lenls of generality. The human christian names
gi,·en to birds apply to any member of a determinate species: any
magpie is called ':'.\largot'. But the names gi\'en to flowers: Queen-
Elizabeth, Impératrice-Eugénie, etc., co,·er no more than the
,·ariety or sub-variety. The field of application of the names ginn
to dogs and cattle is e\·en more restricted. The animal's owner
intends them to refer to a single individual, e\·en though in
fact each can be borne by several: there is not just one dog called
':'.\lédor'. Only the names of racehorses and other pedigree animais
are absolutely individualized. :\o other trotting horse than the one
alreadv . so christened can be called On·ietan II 1 in the twent\·-six
.
years of the alphabetical cycle.•
This is, however, in my ,·iew, the clearest proof one could wish for
that proper names and species names form part of the same group
so that there is no fundamental difference between the two types of
name. Or rather the reason for the difference lies not in their lingu-
istic nature but in the way in which each culture <livides up reality
and in the nriable limits it assigns to the enterprise of classification,
depending on the problems it raises (which can diner for each
particular group within the total society). It is thus by ,·irtue of an
lanll'uage fumishes the common name rose which first becomes Rose, a woman's
christian name and then retums to ordinary language through the imermediary
of the sacred language in the form: Princess ~1argarer-Rose, naming a variery
of rose of which (if the f\ower is a success) it will ra;iidly become the common
na me.
•The alphabetical cycle refers to the system "'hereby the initial lener of a
pedigree animal should correspond to the year of its bi:rth, and th.at the le!!ers
succeed one another in alphabetical order [trans. note).
213
extrinsic detennination that a certain Jeye) of classification requires
appellations, which can be common nouns or proper namcs accord-
ing to the circumstances. But this does not, howevcr, bring us back
to the Durkheimian thesis of the social origin of logical thought.
Although there is undoubtedly a dialectical relation between the
social structure and systems of categories, the latter are not an effect
or result of the former: each, at the cost of laborious mutual adjust-
ments, translates certain historical and local modalities of the
relations between man and the world, which form their common
substratum.
These clarifications were necessary to enable us to emphasize,
without risk of misunderstanding, the sociological and at the same
time relative nature of the notion of a species as well as of an indiù-
dual. From the biological point of view, men who belong to the same
race (assuming that a precise sense can be given to this term) are
comparable to the indiYidual flowers which blossom, open and
wither on the same tree: they are so many specimens of a variety or
sub-nriety. Similarly, ail the members of the species Homo sapiens
are logically comparable to the members of any other animal or
plant species. Howeyer, social life effects a strange transformation
in this system, for it encourages each biological individual to
deYelop a personality; and this is a notion no longer recalling speci-
mens within a nriety but rather types of varieties or of species,
probably not found in nature (although there is a suggestion of it
now and again in the tropical environment) and which could be
termed 'mono-indn·idual'. '\'hat disappears with the death of a
personality is a synthesis of ideas and modes of behaviour as exclu-
sive and irreplaceable as the one a iloral species develops out of the
simple chemical suhstances common to ail species. "'hen the Joss
of someone dear to us or of some public personage such as a poli-
tician or writer or artist moves us, we suffer much the same sense of
irreparable printion that wc should expericnce were Rosa centzfolia
to bccomc extinct and itsscent to disappear for ever. From this point
of view it seems not untrue to say that some modes of classing,
arbitrarily isolated under the title of totemism, are universally
employed: among ourselves this 'totemism' has merely been
humanized. EYerything takes place as if in our civilization every
individual's own personality were his totem: it is the signifier of
his signified being.
214
In so far as they derive from a paradigmatic set,• proper names

th us form the fringe of a general system of classification: they are
both its extension and its limit. \Yhen thev corne on to the stage the
curtain rises for the last act of th~ logi~al performance. But the
length of the play and the number of acts are a matter of the civiliza-
tion, not of the language. The more or Jess 'proper' nature of names
is not intrinsically determinable nor can it be discovered just by
comparing them with the other words in the language. It depends
on the point at which a society declares its work of classifying to be
complete. To say that a name is perceived as a proper name is to say
that it is assigned to a level beyond which no classification is
requisite, not absolutely but within a determinate cultural system.
Proper names always remain on the margin of classification.
In e,·ery system, therefore, proper names represent the quanta of
signification below ,,·hich one no longer does anything but point.
This brings us to the root of the parallel mistakes committed by
Peirce and by Russell, the former in defining proper names as
'indices' and the latter in believing that he had discovered the logical
mode! of proper names in demonstrative pronouns. This amounts
in effect to allowing that the act of naming belongs to a continuum
in which there is an imperceptible passage from the act of signifying
to that of pointing. I hope that I have succeeded in showing that this
passage is in fact discontinuous although each culture fixes its
thresholds differently. The natural sciences put theirs on the level
of species, varieties or subvarieties as the case may be. So terms of
different degrees of generality will be regarded each time as proper
names. The same mental operation is involved when the native sage
- and sometimes scientist - also practising these modes of classifica-
tion, extends them to individual members of the social group or,
more precisely, to the single positions which individuals - each con-
stituting a sub-class - can occupy at the same time or successively.
From a formai point of view there is th us no fondamental difference
• E,·en Vercingetorix which Gardiner regards as a pcrfcct cxample of an

'embodied' name. Without speculating on the place of Vercingetorix in the
system of Gallic names, it is clear that for us it refers to the warrior of ancient
times who enjoyed an exclusi,·e name of curious consonance and is not Attila
nor Genseric nor Jugurtha nor Ghengiz Khan. As for Popocatepetl, another
example dear to Gardiner, e\·ery schoolboy, even if ignorant of geography,
knows that it must belong to the same class as Titicaca. One classes as best one
can.
215
THE SAVAGE MIND
between the zoologist or the botanist who allots a recently discovered

plant the position Elephantopus spicatus Aubl., arranged for it by
the system (if it has not indeed already been written into it), and the
Omah priest who defines the social paradigms of a new member of
the group by conferring the available name Old-bison's-used-hoof on
him. They know what they are doing in both cases.
216
CHAPTER,EIGHT
TIME REGAINED
Taking an overall view of the devices and procedures which 1 have

so far been primarily concerned to list, one is first of ail struck by
the systematic nature of the relations between them. Two aspects of
this system are also immediately apparent: its internai coherence
and its practically unlimited capacity for extension.
As the examples given shO\v, the structure is in ail cases supported
by an axis (which it is convenient to picture as vertical). This con-
nects the general with the particular, the abstract with the concrete;
but the classificatory intention can always reach its limits whichever
direction is in question. These are defined in terms of an implicit
axiomatic according to which ail classification proceeds by pairs of
contrasts: classification only ceases when it is no longer possible to
establish oppositions. Strictly speaking, therefore, the system knows
no checks. Its internai dynamism is progressively weakened as it
proceeds along its axis in either direction. And when the system
cornes to a hait, this is not because of any unforeseen obstacles pre-
sented by empirical properties of beings or things nor through any
jamming of its mechanism but because it has completed its course
and wholly fulfilled its fonction.
\Vhen the classificatory intention ascends, as it were, towards
the greatest generality and most extreme abstraction, no diversity
prevents it from applying a scheme through the operation of
which reality undergoes a series of progressive purifications, whose
final term will be provided, as intended, in the form of a simple
binary opposition (high and low, right and left, peace and war,
etc.), and beyond which it is, for intrinsic reasons, useless as well
as impossible to go. The 5ame operation can be repeated on other
217
THE SA Y AGE MIND
planes: on that of the internai organization of the social group,

which the so-called totemic classifications allow to grow to the
dimensions of an international society by application of a similar
scheme of organization to an ever-greater number of groups; or
again on the spatio-temporal plane, thanks to a mythical geo-
graphy which, as an Aranda myth mentioned above* shows,
permits the organization of an inexhaustible variety of landscapes
by successive reductions which once again terminate in a binary
opposition (in this case between directions and elements, since the
contrast here is between land and water).
At the lower end there is no external limit to the system either,
since it succeeds in treating the qualitative diversity of natural
species as the symbolic material of an order, and its progress
towards the concrete, particular and individual is not even arrested
by the obstacle of persona) appellations: even proper names can
serve as terms for a classification.
What is in question is thus a total system, which ethnologists
in vain tried to pull to pieces in order to fashion them into distinct
institutions, of which totemism continues to be the most famous
example. But in this way one is led only into paradoxes bordering
on the absurd. Thus, Elkin (4, pp. 153-4), in an otherwise admir-
able work of synthesis, taking totemism as the point of departure
for his analysis of the religious thought and organization of
Australian natives, when confronted by its theoretical wealth,
evades the difficulty by introducing a special heading, 'classifica-
tory totemism'. He thus treats classification as a special form of
totemism when in fact, as I hope I have established, it is totemism
or so-called totemism, which constitutes not even a mode of
classification, but an aspect or moment of it. Comte knew nothing
of totemism (ar.d it was no doubt this which saved him from being
deceived by a phantom), and though he lacked the evidence which
would have confirmed his thesis, yet he had roughly gauged the
importance in the history of thought of a classificatory system,
whose organization and tenor he understood better than
ethnologists of the present day:
.i'iever sin cc that epoch have human conceptions been able to recover to a
degree at all comparable, that great unity of method and homogeneity of
doctrine which constitutes the fully normal state of our intelligence, and
which it had then acquired spontaneously ... (Comte,53e leçon, p. 58).
•Cf. ahoYe, pp. 168-9.
2r8
TI:\lE REGAINED
::\ o doubt Comte assigns this 'savage mind' to a period of history -

to the ages of fetishism and polytheism - white in this bcok it is
neither the mind of sa\·ages nor that of primitin or archaic
humanity, but rathcr mind in ffs untamed state as distinct from
mind cultirnted or domesticated for the purposc of yiclding a
rcturn. This latter has appearcd at certain points of the globe and
at certain moments in history, and it is natural that Comte, lacking
ethnographie data (and that ethnographie scnsc which can be
acquired only by the collection and handling of data of this type)
should ha\·e apprchcndcd the former in its rctrospcctivc form, as a
mode of mental acti\·itv antcrior in timc to the latter. \Ve are better
able to undcrstand today that it is possible for the two to co-exist
and intcrpcnctrate in the same way that (in theory at lcast) it is
possible for natural species, of which somc arc in thcir sa vagc statc
and others transformed by agriculture and domestication, to co-
exist and cross, although - from the vcry fact of thcir dcvclopmcnt
and the gcncral conditions it rcquircs - the existence of the latter
thrcatcns the former with extinction. But, whethcr one dcplorcs or
rcjoiccs in the fact, thcrc arc still zones in which savage thought,
likc sarngc spccics, is rclativcly protcctcd. This is the case of art,
to which our ci\·ilization accords the status of a national park, with
ail the ad\·antages :md inconveniences attending so artificial a
formula; and it is particularly the case of so man y as yet 'un-
clcarcd' sectors of social life, whcre, through indifferencc or
inability, and most often without our knowing why, primitive
thought continues to Aourish.
The exccptional fcatures of this mind which we cal! savage and
which Comte described as spontaneous, relate principally to the
extensive nature of the ends it assigns itself. lt daims at once to
analyse and to synthcsize, to go to its furthest limits in both direc-
tions, while at the same time remaining capable of mediating
between the nrn poles. Comte noted the analytical orientation vcry
\\'C Il :
The very superstitions which seern rnost absu1d to us today ...

originally had a truly progressive philosophical character, in that they
provided, as a rule, a forcible stimulation to persistent observation of
phcnornena, the exploration of which could not, a t that tirne, dircctly
have inspired any sustained interest (id., p. 70).
The error of judgment of which Comte was guilty in the last pro-
position explains why he was so entirely mistaken about the
219
THE SAVAGE MIND
synthetical aspect. Contemporary savages were slaves of 'the infi-

nitc variety of phenomena' and were, as he thought 'discerning
exploration' of them confirmed, unacquainted with any 'nebulous
symbolization' (p. 63). Now, 'the discerning exploration of con-
temporary savages', which is precisely what ethnography practises,
invalidates the positivist preconception on both these points. If
savage thought is definable both by a consuming symbolic ambition
such as humanity has never again seen rivalled, and by scrupulous
attention directed entirely towards the concrete, and finally by
the implicit conviction that these two attitudes arc but one, it is
surcly precisely because it rests, from a theoretical as well as a
practical point of view, on this 'sustained interest' of which Comte
denies it to be capable. When man observes, experiments, classi-
fies, and thcorizes, he is not impelled by arbitrary superstitions
any more than by the vagaries of chance, to which it would, as
we saw at the beginning of this book, be naïve to attribute a part in
the discovery of the arts of civilization.*
Given only these two explanations to choose from, one would
prefer Comte's, provided it were first rid of the paralogism on
which it rests. For Comte, ail intellectual evolution really proceeds
from 'the incvitable primitive dominance of theological philo-
sophy', that is to say, from the fact that originally man found it
impossible to interpret natural phenomena without assimilating
them 'to his own actions, the only ones whose essential mode of
production he could ever believe himself to understand' (id., 51e
leçon, IV, p. 347). But how could he have clone this without
simultaneously making the opposite move of attributing a power
and efficacy comparable to that of natural phenomena to his own
actions ? This man, externalized by man, can serve to shape a god
only if the forces of nature have already been internalized in him·.
The mistake made by Comte and the majority of his successors
was to believe that man could at ail plausibly have peopled nature
with wills comparable to his own without ascribing some of the
attributcs of this nature, in which he detected himself, to his
desires; for the mere sense of his own impotence, had he started
with this alone, would never have furnished him with a principle
of explanation.
Indeed, the difference between practical action, which produces
returns, and magical or ritual action, which Jacks efficiency, is not
•Cf. pp. 13-14.
220
Tll\IE REGAINED
that which one thinks one perceives in defining them in terms of

their objective or subjective orientation respectively. This might
seem true when one considers things from the outside, but the
relation is the opposite from th~· point of view of the agent: he
conceives practical actions as subjective in their principle and
centrifugai in their orientation, since they result from his inter-
ference in the physical world. '.\Iagical operations, on the other
hand, appear to him as additions to the objective order of the
universe: they present the same necessity to those performing them
as the sequence of natural causes, in which the agent believes
himself simply to be inserting supplementary links through his
rites. He therefore supposes that he observes them from outside
and as if thcy did not emanate f rom himself.
This rectification of traditional perspectives enables us to dispose
of the spurious problem which some see in the 'normal' recourse
to fraud and trickery during magical operations. For, if the syst~m
of magic rests entirely on the belief that man can intervene in
natural determinism to complete or modify its course, then it
hardi y matters whether he does so a little more or a little Jess:
fraud is consuhstantial with magic and, strictly speaking, the
sorcerer never 'cheats'. The difference between his theory and his
practice is one of degree, not of kind.
Secondly, light is thrown on the question of the relations between
magic and religion which is so controversial. For, although it can,
in a sense, be said that religion consists in a humanization of
natural laws and magic in a naturalization of human actions - the
treatment of certain human actions as if they were an integral part
of physical determinism - these are not alternatives or stages in an
evolution. The anthropomorphism of nature (of which religion
consists) and the physiomorphism of man (by which we havé
defined magic) constitutc two components which are always given,
and vary only in proportion. As we noticed earlier, each implies
the other. There is no religion without magic any more than there
is magic without at least a trace of religion. The notion of a super-
nature exists only for a humanity which attributes supernatural
powers to itself and in return ascribes the powers of its super-
humanity to nature.
There is therefore no need to invoke the exercise of vanished
faculties or the employment of some supernumerary sensibility to
221
THE SAVAGE MIND
understand the penetration \vhich so-called pnm1t1ves show in

their observation and intcrprctation of natural phenomena. The
procedure of the American lndian who follows a trail by means of
imperceptible dues or the Australian who unhesitatingly identifies
the footprints lcft hy any member of his group (:\1eggitt) is no
different from our proccdure when \\"e drive a car and assess the
moment to pass or arnid a vehicle at a glance, by a slight turn of
the wheels, a fluctuation in the normal speed of the enginc or even
the supposcd intention of a look. This comparison is highly instruc-
tive, however incongruous it may seem, for, what sharpens our
faculties, stimulates our perception, gives assurance to our judg-
ments is, on the one hand, that the mcans we command and the
risks we run arc immeasurably increased by the mechanical power
of the engine and, on the other, that the tension resulting from the
feeling of this incorporated force exercises itself in a series of
dialogues with other drivers whose intentions, similar to our own,
ue translated into signs which we set about deciphering precisely
because they are signs, and call for intellection.
This reciprocity of perspectives, in which man and the world
mirror each othcr and which seems tous the only possible explana-
tion of the properties and capacities of the savage mind, we thus
find transposed to the plane of mechanized civilization. An exotic
observer would certainly declare the traffic in the centre of a large
town or on a motorway to be beyond the scope of human faculties;
and so in effect it is, in as much as it is neither men nor natural
laws which are brought exactly face to face but systems of natural
forces humanized by drivers' intentions and men transformed into
natural forces hy the physical energy of \vhich they make them-
selves the mediators. It is no longer a case of the operation of an
agent on an incrt object, nor of the return action of an object, pro-
moted to the role of an agent, on a subject dispossessing itself in its
favour without demanding anything of it in return; in other words,
it is no longer situations involving a certain amount of passiveness
on one sicle or the other which are in question. The beings con-
front each other face to face as subjects and objects at the same
time; and, in the code they employ, a simple variation in the
distance separating them has the force of a silent adjuration.
This enables us to understand how it is that attentive, meticulou'l

observation turned entirely on the concrete finds both its principle
222
TIME REGAINED
and its result in symbolism. Savage thought does not distinguish

the moment of observation and that of interpretation any more
than, on observing them, one first registers an interlocutor's signs
and then tries to understand tfiem: when he speaks, the signs
expressed carry with them their meaning. Articulated language
decomposes into elements, each of which is not a sign but rather
the medium of a sign: a distinctive unit which could not be
replaced by another without a change of meaning, which it
expresses by being joined or opposed to other units.
This conception of classificatory systems as systems of meaning
will be thrown into relief even more clearly if two traditional prob-
lems are briefly discussed; the problcm of the relation between
so-called totemism and sacrifice, and the problem raised by the
similarities exhibited throughout the world by the myths used to
explain the origin of clan appellations.
That it should have been possible to regard totemism as the
origin of sacrifice in the history of religion remains, after so long, a
matter of astonishment. Even if, for convenience, one were to
agree to grant totemism a semblance of reality, the two institu-
tions would only look the more contrasting and incompatible, as
:Vlauss, not without hesitation and afterthought, was often led to
affirm.
I am not claiming that segmentary societies in which clans are
called by animal or plant names could not have practised some
forms of sacrifice. The example of the sacrifice of dogs among the
Iroquois is sufficient to prove the contrary. But the dog does not
serve as any clan's eponym among the Iroquois, and the system of
sacrifice is thus independent of that of clan affinities. Above ail,
there is another reason for which the twô systems are mutually
exclusive. Granting that in both cases there is an implicit or
explicit recognition of an affinity between a man or group of men,
and an animal or plant (in the capacity of an eponym of the group
of men or in that of a sacrified object replacing a man, or, again, in
that of a medium to the human sacrificer), it is clear that in the case
of totemism no other species or natural phenomenon is substitut-
able for the eponym: one beast can never be taken for another. If
I am a member of the bear clan, then I cannot belong to the eagle
clan, since, as we have seen, the system's sole reality consists in a
network of differentiation between terms posited as discontinuous.
The opposite is true in the case of sacrifice. Although distinct
223
THE SAVAGE MIND
things are often destined in a preferential manner, for certain

deities or certain types of sacrifice, the fondamental principle is
that of substitution: in default of the prescribed object, any other
can replace it, so long as the intention, the only thing of conse-
quence, persists and although the zeal itsclf can vary. Sacrifice
therefore bclongs to the realms of continuity:
\Vhen a cucumber is uscd as a sacrificial victim :\'uer speak of it as an
ox. In doing so they are asserting something rather more than that it takes
the place of an ox. They do not, of course, say that cucumbers are oxen,
and in speaking of a particular cucumber as an ox in a sacrificial situation
they are only indicating that it may be thought of as an ox in that particu-
lar situation. And they act accordingly by performing the sacrificial rites
as closcly as possible to what happens when the victim is an ox. The
resemblance is conceptual, not perceptual. The 'is' rests on qualitative
analogy. And the expression is as:i-mmetrical, a cucumber is an ox, but an
ox is nota cuc:imber (Evans-Pritchard 2, p. 128).
There are therefore two fondamental differences between the

system of totemism and that of sacrifice. The former is a quantified
system while the latter permits a continuous passage between its
terms: a cucumber is worth an egg as a sacrificial victim, an egg
a fish, a fish a hen, a hen a goat, a goat an ox. And this gradation is
oriented: a cucumber is sacrificed if there is no ox but the sacrifice
of an ox for want of a cucumber would be an absurdity. In tote-
mism, or so-called totemism, on the other hand, relations are
always reversible. In a system of clan appellations in which both
figured, the oxen would be genuinely equivalent to the cucumbers,
in the sense that it would be impossible to confound them and that
they would be equally suitable for manifesting the differentiation
bctween the groups they respectively connote. But they can only
play this part iu so far as totemism (as distinct from sacrifice)
proclaims them to be distinct and not substitutable for each
other.
The underlying reason for these differences is to be found in the
respective roles assigned to natural species in each system. Tote-
mism is based on a postulation of homology between two parallel
series - that of natural species and that of social groups - whose
respecti\·e terms, it must be remembered, do not resemble each
other in pairs. It is only the relation between the series as a whole
which is homomorphie: a formai correlation between two systems
of differences, each constituting a pole of opposition. ln sacrifice,
TL\IE REGAIN'ED
the series of natural species (continuous and no longer discon-

tinuous, oriented and no longer re\·ersible) plays the part of an
intermediary between two polar t~ms, the sacrifi.cer and the deity,
between which there is initially no homology nor even any sort of
relation. For, the object of the sacrifice precisely is to establish a
relation, not of resemhlance but of contiguity, by means of a series
of successive identifications. These can be made in either direction,
depending on whether the sacrifice is expiatory or represents a rite
of communion: thus, either of the person offering the sacrifice
with the sacrifi.cer, of the sacrifi.cer with the victim, of the
sacralized Yictim with the deity; or in the reverse order.
SYSTE:\I OF TOTE:\IIS:\I: SYSTDI OF SACRIFICES:
natural series: -,+---~-_+-.---;:--t---+-

,
:t deity:
relations of
1
1
1
1
1 :.·1: natural series and
homology: 1 1
1
-::_~ ::
relatwns of
1
1
1 :.: :: contiguity:
cultural series:
. :. . ~ l .L
1nan:
~1- .-1-1--1-.·t-
This is not ail. Once the relation between man and the deity is
secured by sacralization of the victim, the sacrifice breaks it by
destroying this same victim. Human action thus brings about a
solution of continuity, and, as it had pre\"Ïously established com-
munication between the human reservoir and the divine reservoir,
the latter will automatically fi.Il the gap by discharging the antici-
pated benefi.t. The scheme of sacrifice consists in an irreversibk
operation (the destruction of the victim) with a view to setting off
an equally irreversible operation on another plane (the granting of
divine grace), which is required by the fact that two 'recipients',
situated a t different levels, have previously been brought into
communication.
So sacrifice is an absolute or extreme operation which relates to
an intermediary object. From this point of view, it resembles,
though it is at the same time opposed to them, the rites termed
'sacrilegious', such as incest, bestiality, etc., which are intermediary
operations relating to extreme objects. This was shown in an earlier
chapter in the case of a minor sacrilege: the appearance of a
menstruating woman while the rites of eagle hunting are in pro-
gress, among the Hidatsa lndians.* Sacrifice seeks to establish a
desired connection between two initially separate domains. As
•Cf. abO\·e, p. 51 seq.
225
THE SAVAGE MIND
language so well expresses it, its object is to bring to pass the

fui.filment of human prayers by a distant deity. I t daims to achieve
this by first bringing together the two domains through a sacralized
victim (an ambiguous object, in effect attaching to both), and then
eliminating this connecting term. The sacrifice thus creates a Jack
of contiguity, and by the purposive nature of the prayer, it induces
(or is supposed to induce) a compensating continuity to arise on
the plane where the initial deficiency experienced by the sacrificer
traced the path which leads to the deity, in advance and, as it were,
by a dotted line.
I t follows that the eating of totemic species, sometimes occurring
in the Australianrites of increase known bythe nameof Intichiuma,
cannot be treated as a primitive, or even an aberrant, form of
sacrifice. The resemblance is as superficial as that which would lead
one to identify whales and fish. Moreover, these rites of increase
are not regularly connected with so-called totemic classifications;
they do not always accompany them, even in Australia, and there
are numerous and widespread examples of rites of increase without
'totemism' and of 'totemism' without rites of increase.
Above ail, the structure of rites of the Intichiuma type and the
implicit ideas on which they rest are very far removed from those
we have discerned in sacrifice. In such societies (magical)
production and (real) consumption of natural species are normally
separate as a result of an identity postulated between each group
of men and a totemic species, and a proclaimed or ascertained dis-
tinction between different social groups on the one hand and
different natural species on the other. The role ot the lntichiuma
is therefore periodically and momentarily to re-establish the
contiguity between production and consumption, as though human
groups and natural species had from time to time to count them-
selves two by two and in pairs of allies, before each taking their
own place in the game: species nourishing the men who do not
'produce' them, and men producing the species which are for-
bidden to eat. In the Intichiuma, consequently, men momentarily
confirm their substantial identity with their respective totemic
species, by the two-fold rulc that each group is to produce what it
consumes and consume what it produces, and these things are the
same for each and different for ail. Thanks to this, there will no
longer be a risk of the normal play of reciprocity creating confu-
sions between fondamental definitions which must be repeated
226
TIME REGAINED
periodically. If the natural series is presentcd by capitals and the

social series by small letters,
B c D E .......... X
a b c d e .......... 11
then the Intichiuma rccalls the affinity between A and a, B and b,

C and c, ~and n, attesting that whe:1, in the normal course of events,
group b incorporates, by eating, species A, C, D, E ... N, groupa
species B, C, D, E ... );, and so on, what is in question is an
exchange between social groups and an arbitrage between resemb-
lance and contiguity, not the replacement of one resemblance by
another resemblance nor one contiguity by another contiguity. *
Sacrifice turns to comparison as a mcans of effacing differences
and in order to establish contiguity; the so-called totemic meals
institute contiguity, but only with a view to making possible a
comparison, the anticipated result of which 1s to confirm
diffcrcnccs.
The two systems arc thcrcforc opposcd by their orientation,
metonymical in one case and metaphorical in the other. But this
anti-symmetry lea\'es them still on the same plane, when in tact
they are on different leYcls from the epistemological point of view.
Totcmic classifications have a doubly objective basi >. There
really arc natural species, and they do indccd form a discontinuous
serics; and social segments for their part also cxist. Totemism, or
so-called totemism, confines itself to conceiving a homology of
structure between the two series, a perfectly legitimate hypothesis,
for social segments are instituted, and it is in the power of each
society to render the hypothesis plausible by shaping its rules and
representations accordingly. The system of sacrifice, on the other
•The lndians of eastern Canada do not eat dcer-mcat whilc thcy arc hunting
the deer, nor trout du ring the trout-fishing season (J enness, 1, p. 60). So they
eat only when thcy arc not killing and kill only whcn thcy arc not eating. The
reciprocity between man and animal specics is of the samc type as that bctwecn
two groups of men in some Australian tribcs, the occasion for which is provided
by a natural specics. On the othcr hand, what is in question in Canada is dia-
chronie, and nota> in Australia synchronie, rcciprocity. The same difference can
also be sccn among the Keresr1n group of the Pueblo: 'Each ycar the ... toraikatsi
chief of the wildcrness would select a fcw wild plant foods and gamc animais
upon which thcy would concentratc their efforts to bring forth an abundance.
They would alter the list of plants and animais somewhat from ycar to ycar'
(L.A. White, p. 306). This is thus an Intichiuma, but situated on the axis of
successiveness instead of the axis of simultaneity.
227
THE SAVAGE l\llND
hand, makes a non-existent term, divinity, intervene; and it

adopts a conception of the natural series which is false from the
objective point of view, for, as we ha\·e seen, it represents it as
continuous. To express this difference in level between totemism
and sacrifice it is not, then, enough to say that the former is a
system of reference and the latter a system of operations, that one
works out a schema of interpretaticn while the other sets up (or
claims to set up) a technique for obtaining certain results: that
one is true, and the other is false. Rather, to put it preci!>ely,
classificatory systems belong to the levels of language: they are
codes which, however well or badly made, aim always to make
sense. The system of sacrifice, on the other hand, represents a
private discourse wanting in good sense for all that it may
frequently be pronounced.
fn another work 1 briefly mentioned the myths of origin of so-

called totemic institutions, and showed that even in regions di">tant
from each other and despite the difference in their stories, these
myths all teach the same lesson, namely: ( I) that so-called totemic
institutions are based on a global correspondence between two
series, not en particular correspondences between their terms;
(2) that the correspondence in question is metaphorical, and not
metonymical; and (3) that it becomes evident only after each
series has first been impo\·erished by a suppression of elements so
that their internai discontinuity emerges clearly (Lévi-Strauss 6,
pp. 19-20 and 26-7).
There is a singular contrast between the precision and wealth of
this lesson (the more striking as the myths analysed are known to
us only in abridged or mutilated \·ersions)* and the insignificance
of myths accounting for the appellations of each clan. The latter
are all very much alike throughout the world, but notably in their
poverty. Certainly Australia possesses complex myths which !end
themselYes to a semantic analysis inspired by that which we have
a pplied to myths of other regions (Stanner 2 ). N evertheless,
specialists on this continent are used to collecting myths in which
the attribution of a half-human, half-animal ancestor to a totemic
group rests on a mere statement: the myth establishes that the
ancestor appeared at such and such a place, followed such and
such a course, performed certain actions in this or that place,
• Firth (.-) has just published more complete Yersions of Tikopia myth.
228
TL\IE REG:\10:ED
which mark him out as the originator of gcographical features

which can still be seen, finally that he stopped or disappeared in a
particular spot. Properly speaki~g, therefore, the myth amounts to
the description of an itinerary and adds little or nothing to the
remarkable facts which it claims to establish: that a particular
course, and the water-points, thickets, or rocks which mark it, are
of sacred rnlue for a human group and that this group proclaims its
affinity with this or that natural species: catcrpillar, os tri ch, or
kangaroo.
:\o doubt, as T. G. H. Strehlow has pointcd out, the exclusive
use of pidgin for a long time forced enquirers to content themselves
with sketchy and ridiculous \·ersions. But, in the first place, wc
now have plenty of tcxts with interlincar translations and of adap-
tations which are the work of compctent specialists; and, secon<lly,
myths of exactly the samc type arc found in other regions of the
world wherc the linguistic difficultics wcrc more quickly oycrcome.
l shall restrict rnysclf to thrcc examplcs, ait frorn Ame: ica. The
first two corne from the i\orthcrn and Southern Unitd States
respectively, and the third from Central Brazil.
The '.\Ienomini explain thcir clan names by saying that whcn the
bear was endowed with human form, he settlcd down with his wife
not far from the mou th of the :\Ienomini river. IIcre thcy fi.shed
the sturgcons which comtitutcd thcir only food (the bcar and
sturgeon clans belong to the samc phratry). One day, threc thundcr
birds perched on the great ledge of rock which projects into Lake
\Vinnebago, ncar the place called Fond du Lac. After changing
into men, they visitcd the bcars and came to an agreement with
them to convoke sevcral animais whose place of birth or residence
is sp~cified by the myth. They ail set off. When they arrived at
Green Bay on Lake Michigan an obliging wavc carricd the wolf,
who could not swim, to the other bank. In token of his gratitude,
ht> adopted the \Va\·e as one of his clan totems. An analogous inci-
dent, placed near i\Iackinaw, also on Lake !\Iichigan, resulted in
the association of the black bear and the bald eagle. The relations
between the other clans (elk, crane, dog, decr, beavcr, etc.) were
likewise established through fortuitous meetings and services
rendered (Hoffman, pp. 39-42 ; Skinner 1, p. 8, ff.).
The reason why the Hopi wild mustard clan bears this name as
well as those of oak, chaparral-cock, and warrior is that, during a
legendary migration, they tried to stop a child crying by offering it
229
9-TS~
THE SAVA<;E :\1!;'-;D
mustard kan:s and a branch of oak, gathered and eut on the way.
After this, they met the cock and then the warrior. The badger and
butterflv clan is so-called because its ancestors took with them a
man-badger whose acquaintance they had made a short time before
they caught a butterfly to amuse a child; but the child was ill and
it was Badger who curcd him with simples. The ancestors of the
rab bit and tobacco clan found the plant and met the animal. Those
0î the Patki clan took the names of lake, cloud, rain, snow, and fog
as a result of various incidents on their journey. Somewhere
h;:tweer~ ihe actual site of Phoenix (Arizona) and the Little Colora-
do, the ancestors of the bear clan came upon a dead bear, whence
their name; but another band founcl the skin, from which small
rodcnts had taken the hair to line their hales. They made hide
straps out of the skin, and sin ce then the hide strap and bear clans
haYe been associated. A third band took the name of the rodents
and \\·as allied to the former clans (Yoth 4, Parsons, pp. 26-30).
Let us now turn to South America. The Bororo say that the sun
and moon belong to the Badedgeba clan of the Cera moiety on
account of a dispute bet\\·een a father and son, who both wanted to
appropriate the names of these hea\·cnly bodies. A compromise
gaYe the father the names of Sun and Bath-of-the-Sun. Tobacco
belongs to the Paiwe clan because an Indian belonging to it hap-
pened by chance to discovcr its leaves in the innards of a fish he was
gutting in order to cook. The chief of the 'black' Badedgeba clan
at one time possessed somc black and some red birds ( Phimosus
infuscatus and Ibis ruhra respecti\·ely), but his colleague 'red'
Babedgeba stole them from him and he had to agree to a division
according to colour (Colbacchini).
Ali thesc myths of origin of clan appellations are so similar that
it is unnecessary to cite examples from other parts of the world,
such as Africa where they also abound. \Vhat, then, are their com-
mon characteristics? In the first place, they ail haYe in common a
breYÏty leaYing room for none of those apparent digressions which
often haYe a wealth of concealcd meaning. A story reduced to its
essential outlines has no surprises in store for the analyst. Secondly,
thcsc myths are falsely acticlogical (supposing that a myth could be
genuinely so) in as muchas the kind of explanation they give Ï5
reducible to a scarcely modified statement of the initial position;
from this point of \·iew they appear redundant. Their raie seems
to be demarcative, rather than aetiological; they do not really
230
TI:\IE REGAINED
explain an origin or indicate a cause; what they do is to invoke an

origin or cause (insignificant in itself), to make the most of some
detail or to 'stress' a species. This detail or that species acquires a
differential nlue not because of the particular origin attributed to
it but just because it is endowed with an origin when other details
or species are not. History is surreptitiously introduced into the
structure in a modest, almost negative way: it does not account for
the present, but it makes a selection between its elements, accord-
ing only some of them the privilege of having a past. The poverty
of totemic myths is therefore due to the fact that the fonction of
each is only to establish a difference as a difference: they are the
constitutive units of a system. The question of significancc does
not arise at the level of each myth taken in isolation but at that of
the system of which they form the elements.
Xow, here we again encounter a paradox discussed earlier :*
the systems with which we are concerned are not easily 'mytholo-
gizable' as systems hecause their virtually synchronie nature is
engaged in a never-ending struggle with diachrony. Ex lzypotlzesi
the elements of the system are on this sicle of myth, but, in terms of
its destination, the set lies always beyond; myth, as it were, runs
after it te catch up with it. It only exceptionally succeeds because
the system is constantly being drawn in by histcry; and when one
thinks it has succeeded a new doubt. makes itself felt: do mythical
representations correspond to an actual structure which models
social and religious practiccs or do they translate only the congealed
image by means of which native philosophers give themselves the
illusion of fixing a reality which escapes them? Important as
1\Iarcel Griaule's discoveries in Africa are, one often wonders to
which of these interpretations they relate.
The earliest theories on totemism are, as it were, contaminated
by this paradox, which they were unable to formulate clearly. The
reason why McLennan, and later Robertson Smith and Frazer (IV,
pp. 73-6, 264-5) were so convinced that totemism was anterior to
exogamy (a proposition which I find meaningless) was that the
former appeared to them simply denotative, whereas they divined
the systematic nature of the latter. N ow, systems can be established
only between elements which are already denoted. But to perceive
totemism too as a system it would have been necessary to place it
in the linguistic, taxonomie, mythical, and ritual context from
"Cf. above, pp. 66-71.
THE SAVAGE MIND
which, in their concern to trace the contours of an arbitrary

institution, they began by isolating it.
ln fact, as l have tried to show, matters are not so simple. The
ambiguity of totemism is real, even if the institution we have
imagined in the hope of getting rid of it is not. Indeed, so-called
totemism may either present or preclude the characteristics of a
system, depending on the point of view adopted: it is a grammar
fated to degenerate into a lexicon. Unlike other systems of classi-
fication, which are primarily conceived (like myths) or acted (like
rites), totemism is always lz"ved, that is to say, it attaches to concrete
groups and concrete individuals because it is an hereditary system
of classification.*
This enables us to understand the appearance of a permanent
conflict between the structural nature of the classification and the
statistical nature of its demographic basis. The classification tends to
be dismantled like a palace !'>Wept away upon the flood, whose parts,
through the effect of currents and stagnant waters, obstacles and
straits, corne to be combined in a manner other than that intended by
the architect. ln totemism, therefore, fonction inevitably triumphs
over structure. The problem it has never ceased presenting to
theorists is that of the relation between structure and event. And
the great tesson of totemism is that the form of the structure can
sometimes survive \vhen the structure itself succumbs to events.
There is thus a sort of fondamental antipathy between history
and systems of classification. This perhaps explains what one is
tempted to cal! the 'totemic void', for in the bounds of the great
civilizations of Europe and Asia there is a rema1 kable absence of
anything which might have reference to totemism, even in the form
of remains. The reason is surely that the latter have elected to
explain themselves by history and that this undertaking is incom-
patible with that of classifying things and beings (natural and social)
by means of finite groups. Totemic classifications no doubt <livide
their groups into an original and a derivative series: the former
contains zoological and botanical species in their supernatural
aspect, the latter human groups in their cultural aspect, and the
former is asserted to have existed before the latter, having in some
sort engendered it. The original series, however, lives on in diach-
rony through animal and plant species, alongside the human series.
• ;-.;o doubt some fnrms of totemism are not, propcrly speaking, hereditary;
but eYen in this case, the system is sustained by concrete men.
232
TIME REGAINED
The two series exist in time but under an atemporal regime, since,
being both real, they sa il through time together, remaining such as
they were at the moment of s~_paration. The original series is
always there, ready to serve as a system of reference for the inter-
pretation and rectification of the changes taking place in the
derivative series. In theory, if not in practice, history is subordi-
nated to system.
\Yhen, however, a society sicles with history, classification into
finite groups becomes impossible because the derivative series,
instead of reproducing the original series, merges with it to form
a single series in which each term is derivative in relation to the
one preceding it and original in relation to the one coming aftcr it.
Jnstead of a once-for-all homology het\\·een two series each fini te
and discontinuous in its own right, a continuous evolution is postu-
lated within a single serics that accepts an unlimited number of
terms.
Sorne Polynesian mythologies are at the critical point where
diachrony irrerncably prevails on:~r synchrony, making it impos-
sible to interpret the human order as a fixed projection of the
natural order hy which it is cngcndcrcd; it is prolongation, rather
than a rcflcction, of the natural order:
Firc and ·water married, and from them sprung the earth, rocks, trees,
and eYerything. The cuttle-fi.sh fought with the fi.re and was beaten. The
fire fought with the rocks, and the rocks conquered. The large stones
fought ..,,·ith the small ones; the small ones conquered. The small stones
fought with the grass, and the grass conquered. The trees fought with the
creepers, the trecs wcre heaten :iml the creeper> conquen·d. Tlw creepers
roCCt"d, ~"·arrnl'd \\'llh 1nc1lLl{nts, <:ltH~ fnnn tn;1~\!<•t!'i they ~•«·"· tu 1u_· 111t·n
(C. Turnrr. pp. (1 7).
This eYolutionism precludes any synthesis of a totemic type, for

things and natural beings do not afford the static modcl of a like-
wise static di \·ersity between human groups: they are ordercd as
the genesis of a humanity whose advent they prepare. But this
incompatibility in turn raises a problem, namely: how, if it cxists,
do classificatory systems succeed in elimin;iting history or, when
that is impossible, integrating it?
I have suggested elsewherc that the clumsy distinction between
'peoples without history' and others could with advantage be
rcplaced by a distinction hetween what for com·enience I called
'cold' and 'hot' societies: the former secking, by the institutions
2 33
THE SAVAGE MIND
they give themselvcs, to annul the possible effects of historical

factors on their equilibrium and continuity in a quasi-automatic
fashion; the latter resolutely internalizing the historical process
and making it the moving power of their development (Charbon-
nier, pp. 35-..J.ï; Lévi-Strauss 4, pp. 41-3). Severa! types of his-
torical sequences will still need to be distinguished. Sorne, while
existing in duration, are of a recurrent nature; the annual cycle of
the seasons, for instance, or that of individual life or that of
exchanges of goods and services within the social group. These
scquences raise no problem because they are periodically repeated
in duration without their structure necessarily undergoing any
change; the object of 'cold' societies is to make it the case that the
order of temporal succession should have as little influence as
possible on their content. No doubt they do not succeed perfectly;
but this is the norm they set themselves. Apart from the fact that
the procedures they employ are more efficacious than some con-
temporary ethnologists (Vogt) are willing to admit, the real
question is not what genuine results they obtain but rather by what
lasting purpose they arc guidcd, for their image of themselves is an
essential part of their reality.
It is tcdious as well as useless, in this connection, to amass
arguments to pro\'e that ail societies are in history and change:
that this is so is patent. But in getting embroiled in a superfluous
demonstration, thcrc is a risk of overlooking the fact that human
societies react to this common condition in very differcnt fashions.
Sorne acccpt it, with good or ill grace, and its consequences (to
themselves and other societies) assume immense proportions
through their attention to it. Others (which we cal! primitive
for this reason) w;int to deny it and try, with a dexterity we under-
cstimatc, to make the states of their development which they
consider 'prior' as permanent as possible. It is not sufficient, in
order that they should succeed, that their institutions should
exercise a rcgulating action on the recurrent sequences by limiting
the incidence of demographic factors, smoothing clown antago-
nisms which manifest themselves within the group or between
groups and perpctuating the framework in which individual and
collective activities take place.'*' It is also necessary that these
•At the beginning of a recent studv, G. Balandier announce<l with much
ccrcmony that it is high timc that the so~ial sciences 'graspcd socicty in its uctual
234
TIME REGAI:\"ED
non-recurrent chains of eyents whose effects accumulate to

produce economic and social uphcavals, should be broken as soon
as they form, or that the socicty should have an effectiYe procedure
to prevent their formation. \\"e aré acquainted with this procedurc,
which consists not in denying the historical proccss but in
admitting it as a form without content. Thcrc is indeed a before
and an aftcr, but their sole significancc lies in reflecting each
other. It is thus that ail the activities of the ~orthern Aranda
reproduce those that their totcmic ancestors are still believed to
perform:
The gurra anccstor hunts, kills, anJ eats banùicoots; and hissons arc
always engageù upon the same quest. The witchetty gruh men of Lukara
spenù every Jay of thcir li,·cs in digging up grubs from the roots of
acacia trees . . . . The ragia (wilù plum trec) <Jncestor lives on the ragia
berries which he is continually collecting into a large wooùen vesse!. The
crayfish anccstor is always building fresh weirs across the course of the
moving flood of water which hc is pursuing; ami he is for e\·er engageù in
spcaring fish . . . if the myths gathercù in the ::'\orthcrn Aranda <Jrca are
trcateù collectin:ly, a full ancl n~ry Jetailcù account will be founJ of '111
the occupations which arc still practisctl in Central Australia. In his
myths wc sec the native at his Jaily task of hunting, fishing, gathcring
vcgctablc food, cooking, and fashioning his implemcnt:;. Ali occupations
originatcù with the totemic an ces tors; anJ hcre, too, the native follows
tradition blinùly: he clings to the primitive \\·capons usc<l by his fore-
fathers, and no thought of improving them ever enter his mind (T.(;. IL
Strehlow, pp. J+-5).
1 select this in prefercnce to ail the other evidence to the samc pur-
pose available from other parts of the world because it emanatcs
from an ethnologist born and brought up among the natives,
speaking their language fluently and remaining deeply attached to
life and development'. After which hc describes, in \'Cry pertinent fashion more-
over, institutions whose object is, to use his own terms, to 'regroup' lineages
threatened with dispersion; to 'allay' their crumbling; to 'recall' their solidarity,
'establish' communication with the ancestors, 'prc,·cnt scparatcd members of the
clan from becoming strangers to each other', furnish 'an instrument of protection
against conAicts', 'control' and 'master' antagonisms and subversions by means
of a 'minutely regulated' ritual which is 'a factor reinforcing social and political
structures'. One is easily in agreement with him (while, however, questioning
whether he is so with his own premises), th:it the institutions he began by deny-
ing to ha,·e been founded on 'logical relations' and 'fixed structures' (p. 23)
demonstrate in fact the 'prevalence of traditional social logic' (p. JJ), and that
'the classical system thus reveals, over a long pcriod, a surprising capacity for
"assimilating" ... ' (p. J4). The only surprising thing in ail this is the author's
own surprise.
235
THE SA\'AGE :'vll:-;'D
them. He can be suspected of neither incomprehension nor ill-will.

I t is no doubt difficult for us (as it is for him, judging by what
follows in the text) not to pass an unfaYourable judgment on an
attitude so flagrantly in contradiction with the aYid need for change
characteristic of our own ciYilization. HoweYer, the obstinate fide-
lity to a past conceived as a timeless mode!, rather than as a stage
in the historical process, betrays no moral or intellectual deficiency
whatsoeYer. It expresses a consciously or unconsciously adopted
attitude, the systematic nature of which is attested all over the
world by that endlessly repeated justification of every technique,
ru le, and custom in the single argument: the ancestors taught it to
us. As for us in other domains until recently, antiquity and con-
tinuance are the foundations of legitimacy. But the antiquity is
conceiYed as absolu te, for it goes back to the origin of the wor Id,
and the continuance admits neither of orientation or of degree.
'\Iythical history thus presents the paradox of being both dis-
joined from and conjoined with the present. It is disjoined from it
bccausc the original ancestors were of a nature different from con-
temporary men: they were creators and these are imita tors. lt is
conj()ined with it because nothing has been going on since the
appearance of the an ces tors except e\'ents whose recurrence period-
ically effaces their particularity. It remains to be shown how the
sa\'age mind succeeds not only in overcoming this twofold contra-
diction, but also in deriYing from it the materials of a coherent
system in which diachrony, in some sort mastered, collaborates
with synchrony without the risk of further conflicts arising
between them.
Thanks to ritual, the 'disjoined' past of myth is expressed, on the
one hand, through biological and seasonal periodicity and, on the
other, through the 'conjoined' past, which unîtes from generation
to generation the liYing and the dead. This synchro-diachronic
system has been well analysed by Sharp (p. ïI ), who di vides the
rites of the Australian tribes of the Cape York Peninsula into three
categories. The rites of wntrol are positive or negative. They aim
to increase or restrict species or totemic phenomena, sometimes
for the benefit of, sometimes to the detriment of the group, by fix-
ing the quantity of spirits or spirit-substance allowed to emanate
from the totemic centres established by the ancestors at various
points in the tribal terri tory. The commemoratiYe or historical rites
recreate the sacred and beneficial atmosphere of mythical times -
Til\IE REGAINED
the 'dream age', as the Australians call it - mirroring its protago-

nists and their great deeds. The mourning rites correspond to an
inverse procedure: instead of cl).arging living men with the per-
sonification of remote ancestors, these rites assure the conversion
cf men who are no longer living men into ancestors. It can thus be
seen that the fonction of the system of ritual is to overcome and
integrate three oppositions: that of diachrony and synchrony;
that of the periodic or non-periodic features which either may
exhibit; and, finally, within diachrony, that of reversible and
irreversible time, for, although present and past are theoretically
distinct, the historical rites bring the past into the present and the
rites of mourning the present into the past, and the two processes
are not equivalent: mythical heroes can truly be said to return, for
their only reality lies in their personification; but human beings
die for good. Hence the schema:
LIFE
permanence and periodicity
rites of control ( ·'- - )
SYNCHRONY
DREA:\I
past- present
/~ DEA TH
present-.. past
historical rites / ""' mourning rites
( _;_) /
DIACHRONY
""' ( )
ln central Australia, this system is completed or reinforced by the

usage of the churinga or tjurunga which has given rise to a great
deal of past and present spcculation. The above considerations help
to explain it. The commemorative and funeral rites postulate that
the passage bctween past and present is possible in both directions.
They do not furnish the proof of it. They pronounce on diachrony
but they still do so in tcrms of synchrony since the vcry fact of
celebrating them is tantamount to changing past into present. lt is
therefore understandable that some groups should have sought to
give tangible confirmation of the diachronie essence of diachrony
at the very heart of synchrony. From this point of view it is signi-
ficant that the importance of the churinga is especially great among
the Western and N orthern Aranda and di minishes un til it d isa ppears
altogether as one progresses northwards. The problem of the rela-
tion between diachrony and synchrony is indeed made even
thornier among the Aranda grcups by the fact that they represent
237
THE SAVAGE MIND
the totemic ancestors not, in the manncr of the Arabanna and

Warramung<i (Spencer and Gillen, pp. r6r-2), as indiv;dualized
herocs from whom the members of the totemic group are suppos.::d
to hc <lircctly descended, hut as an indistinct multitude and this
ought to cxcludc, in principlc, the very notion of gencalogical con-
tinuity. ln fact, as wc saw earlier, * among the Aranda, from one
point of vicw it is as if evcry individual had bcfore his birth drawn
an anonymous ancestor hy lot, of whom he then hecame the rein-
carnation. :\ o douht hccausc of the refinemcnts of their sccial
organization which lavishes on synchrony the benefits of clear-cut
distinctions and prccise definitions, even the relation bctween past
and prcsent appeared to them in terms of synchrony. The role of
the churinga would therefore be te offset the correlative impover-
ishmcnt of the diachronie dimension. They are the pa5t mate1 ially
present and they provide the means of reconciling empirical
individuation and mythical confusion.
It is known that the churinga are stone or wooden objects,
roughly oval in shapc with pointed or rounded ends, cften engraved
with symbolic signs, sometimes just picces of wood or unworked
pcbhles. Whatcver its appcarance, each churinga represents the
physical body of a dcfinite ancestor and generntion after generaticn,
it is formally confcrrcd on the living person bdic\·cd ta be this
ancestor's reincarnation. The churinga are hidden in piles in
natural ca\"Cs, far from frequented ways. Periodically they are taken
out to hc inspected and handled, and on thesc cccasions they are
always polished, grcased and coloured, and prayers and incanta-
tions arc addresscd to them. Their rcle and the treatment accorded
to thcm thus have striking analogies with the documentary archives
which \ve secrete in strcngboxes or cntrust to the safe-keeping of
solicitors and which we inspcct from time to time with the care
duc to sacred things, to repair them if necessary or to commit them
to smarter dossiers. On these occasions we too arc prone to recite
great myths rccallcd to us by the contemplation of the tom and
ycllowcd pages: the decds and achievements of our ancestors, the
history of our homes from the time they were built or first acquired.
It is not thercfore profitable to look as far aficld as Durkheim did
to discover the reason for the sacred charactcr of the churinga.
\Vhen an exotic custom fascinates us in spite of (or on account of)
its apparent singularity, it is generally because it presents us with a
• Cf. abovc, p. 88.
TIME REGAINED
distorted reflection of a familiar image, which we confusedly recog-

nize as such without yet managing to identify it. Durkheim (pp.
120-3) would ha,·e it that the churinga were sacred because they
bore the totemic mark, drawn or engraved on them. But, in the
first place, it is now known that this is not always so. T. G. H
Strehlow draws attention, among the Northern Aranda, to stone
churinga, valued above the rest which he describes as 'insignificant
and rude objects, roughly polished by being rubbed together
during increase ceremonies' (p. 54); and among the Southern
Aranda he saw churinga which arc 'plain picces ... of wood,
devoid of markings and hcavily coated with a thick lumpy mixture
of red ochre and grcase' (p. 73). The churinga can even be a smooth
pebble, a natural rock or a tree (p. 95). Secondly, according to
Durkheim's own argument, his account of the churinga had to
confirm one of his fondamental theses, that of the emhlematic
nature of totemism. The sacredness of the churinga, the most
sacred objects known to the Aranda, had to be explained by an
emblematic figuration of the totem to show that the rcprcsented
totem is more sacrcd than the rcal one. But, as I have alrcady said,
there is no such thing as the real totem;* the individual animal
plays the part of the signifying, and the sacredness attaches neithcr
to it nor to its icon but to the signified, which either can stand for.
After ail, a document does not become sacred by virtue of bearing
a stampwhich has prestige, such as that of the Archives Nationales:
it bears the stamp because it has first been acknowledged to be
sacred, and it would remain so without it.
I\or can it be said, following another explanation which Durk-
heim moreoYer relates to the preceding one', that the churinga is
the ancestor's body. This Aranda formula, quoted by C. Strehlow,
must be taken in its metaphorical sense. The ancestor does not lose
his body because at the moment of conception he surrenders his
•Cf. above, p. 148.
There is no single chief ruling an entire Indian tribe, but a chief in every
band; similarly thcre is no single boss for every species of animal or plant, but a
boss in each locality. The bosses are always 1argcr than other plants and animais
of their kind, and in the case of birds, fish, and animais, always white. Now and
then the lndians see and kill them, but generally they keep out of sight of human
beings. They are like the government in Ottawa, an old Indian rcmarked. An
ordinary Indian can never see the 'government'. He is sent from one office to
another, is introduced to this man and to that, each of whom sometimes daims
to be the 'boss', but he never sees the real government, who keeps himself
hidden (Jenness, I, p. 61).
239
THE SA \'AGE M IND
1
TL\lE REGAINED
churinga (or one of them) to the next incarnation. Rather, the

churinga furnishes the tangible proof that the ancestor and his
living descendant are of one flesh. \Vere this not so, how could it be
that if the original churinga is not discovered on the site where the
woman was mystically fertilized, another is fabricated to take its
place? This probative feature of the churinga is another respect in
which they resemble documentaryarchives, particularlytitle-deeds,
which pass through the hands of ail the successive purchasers (and
can be restored in case of loss or destruction), but that, in their
case, it is a question of a moral and physical personality held by a
usufructurary, and not of the real property of an owner. And
although in our case also archiœs are the most rnluable and sacred
of ail goods, we, like the the Aranda, sometimes corne to cntrust
them to foreign groups. And if \\·e send Louis X IV's will to the
United States or the United States !ends France the Dcclaration of
lndependence or the Liberty Bell, this is proofthat, in the Aranda
informant's own words:
... we are living at peace with our ncighbours: wc cannot engage in
strife or fight with men who are guarding our tjuringa and who have
entrusted thcir tjuringa to our safc-keeping (T. G. H. Strehlow, p. 161).
Rut why do wc set such store hy our archives? The cvents to
which they relate arc indepcndcntly attcsted, in innumerable ways:
they survive in the present and in our books; in themselves they
are devoid of meaning; they acquire it entirely through their his-
torical repercussions and the commentaries which explain them by
relating them to other e\·ents. Paraphrasing an argument of Durk-
heim's, we might say of archives that they are after ail only pieces
of paper.* They need only ail have been published, for our know-
• ... in themselves, the churinga are objects of woods and stone like . , . [so
many] others ... (Durkheim, p. 122).
Fig r r Churinga of an Aranda man of the Frog totem. The large concentric
circles (a) represent three celebrated trees which mark the totemic site
near the Hugh River. The straight lines betwcen them (b) depict their
large roots, and the curved lines (c) their smaller roots. The small con-
centric circles (d) represent Jess important trees with their roots, and the
dotted lines (e) are the tracks of the frogs as they hop about in the sand
of the river bed. The frogs themselves are represented on one side of the
churinga (see the left) by the complicated network of lines (their limbs)
linking small concentric circles (their bodies) (from B. Spencer and
F. J. Gillen, The Native Tribes of Central Australia, new e<l., London,
1938, pp. 145-7).
THE SAVAGE MIND
ledge and condition to be totally unaffected were a cataclysm to

destroy the originals. \Ve should, however, feel this loss as an irrepa-
rable injury that strikes to the core of our being. And not without
reason: if my account of churinga is correct, their sacredness
attaches to the fonction of diachronie significance which they atone
attest in a system which, being classificatory, is displayed in its
entirety in a synchrony that succeeds even in assimilating duration.
The churinga are the palpable proofs of mythical times, the
Alcheringa, which could still be conceived without them but of
which there would no longer be any physical evidence. Similarly,
our past would not disappear if we lost our archives: it would be
deprived cf what one is inclined to call its diachronie flavour. It
would still exist as a past but preserved in nothing but contempo-
rary or recent books, institutions, or even a situation. So it too
would be exhibited in synchrony.
The virtue of archives is to put us in contact with pure histori-
city. As I ha\·e already said about myths conceming the origin of
totemic appellations, their value does not lie in the intrinsic signi-
ficance of the events evoked: these can be insignificant or even
entirely absent, if what is in question is a few lines of autograph or
a signature out of context. But think of the value of Johann Sebas-
tian Bach's signature to one who cannot hear a bar of his music
without a quickening of his pulse. As for events themselves, 1 have
pointed out that they are attested otherwise than by the authentic
documents, and generally better. Archives thus provide something
else: on the one hand they constitute events in their radical con-
tingence ( since only interpretation, which forms no part of them,
can ground them in reason), and, on the other, they give a physical
existence to history, for in them atone is the contradiction of a
completed past anù a present in which it survives, surmounted.
Archives are the embodied essence of the e\·ent.
By this approach we recover, at the \'ery centre of the savage
mind, that pure history to which we were already led by totemic
myths. It is not inconceivable that somc of the events they relate
are genuine, even if the picture they paint of them is symbolic and
distorted (Elkin ·h p. 2 ro ). However, this is not at issue, for ail
historical events arc to a large extent the products of the historian's
choice of categories. Even if mythical history is false, it at least
manifests in a pure and accentuated form (the more so, one might
say, because it i s fa Ise) the characteristic traits of an historical event.
242
TI:\!E REGAINED
These depend on the one hand on its contingent status (the

ancestor appeared in such and such a spot; he went here, then
there; he performed this and that deed ... ) and on the other on
its power of arousing intense and '\·aried feelings:
The ?\orthcrn Aranda clings to his nati,·e soi! with evcry fibre of his
being. He will always speak of his own 'birthplaee' with love and rever-
enee. Today, tears will corne into his eycs when hc mentions an ancestral
home site which has been, sometimes unwittingly dcsecratcd by the
white usurpers of his group territory . . . . Love of home, longing for
home, these arc dominating motives which constantly rc-appear also in
the myths of the totcmic anccstors (T. G. H. Strchlow, pp. 30-1).
:\ow, this passionate love of the soil is primarily accounted for

historically :
:\Iountains and crecks and springs and watcr-holcs arc, to him [the
native] not mcrely intercsting or beautiful scenic featurcs ... ; thcy arc
the handiwork of ancestors from whom hc himself has desccnded. Ile
sees recorded in the surroumling landscapc the ancient story of the lives
and the deeds of the inunortal bcings whom hc rcvercs; beings, who for
a hrief space may take on human shape once more; beings, many of
whom hc has known in his own expericncc as his fathers and grand-
fathers and brothers, and as his mothers and sisters. The whole country-
side is his living, age-old family trec. The story of his own doings at the
bcginning of timc, at the dim dawn of life, when the world as hc knows it
now was bcing shapcd and mouldcd by all-powcrful hands (ibid., pp.
pp. 30-31).
\ Vhen i t i s noted that these events and sites are the samc as th ose
which furnished the materials of the symbolic systems to which the
pre,·ious chapters were devoted, it must be acknowledged that so-
called primitive peoples have managed to evolve not unreasonable
methods for inserting irrationality, in its dual aspect of logical
contingence and emotional turbulence, into rationality. Classifica-
tory systems thus allow the incorporation of history, even and
particularly that which might be thought to defy the system. For
make no mistake: the totemic myths which solemnly relate futile
incidents and sentimentalize over particular places are comparable
only to minor, lesser history: that of the dimmest chroniclers. Those
same societies, whose social organization and marriage rules
require the efforts of mathematicians for their interpretation, and
whose cosmology astonishes philosophers, see no solution which
would maintain continuity between the lofty theorizing to which
they devote themselves in these domains and a history which is
243
THE SAVAGE MINIJ
that not of a Burckhardt or Spengler but of a Lenôtre and a La

Force. Considered in this light, the style of the Aranda water
colourists will perhaps appear less unexpected. And nothing in our
ci vilization more clos el y resembles the period ic pilgrimages made by
the initiated Australians, escorted by their sages, than our con-
ducted tours to Goethe's or Victor Hugo's house, the furniture of
which inspires emotions as strong as they are arbitrary. As in the
case of the churinga, the main thing is not that the bed is the self-
same one on which it is proved V an Gogh slept: ait the visitor asks
is to be shown it.
CllAPTER X I:'\E
HISTORY AND DIALECTIC
In the course of this work I have allowed m ysclf, not without

ulterior motive, to borrow a certain amount of Sartre's vocabulary.
I wanted to lead the reader to face a prohlcm, the discussion of
which will serYe to introducc my conclusion. The problem is to
what extent thought that can and will be both anecdotal and geo-
metrical may yet be called dialectical. The savage mind totalizes.
It daims indeed to go very much further in this direction than
Sartre allows dialectical reason, for, on the one hand, the latter lets
pure seriality escape (and we have just seen how classificatory sys-
tems succeed in incorporating it) and, on the other, it excludes
schematization, in which thcse same systems reach their consum-
mation. In my view, it is in this intransigcnt refusai on the part of
the savage mind to allow anything human (or even living) to remain
a lien toit, that the real principle of dialectical reason is to be found.
But my idea of the latter is very different from Sartre's.
In reading the Critique it is difficult to avoid feeling that Sartre
vacillates between two conceptions of dialectical reason. Sometimes
he opposes dialectical and analytical reason as truth and error, if not
as God and the devil, while at other times these two kinds of reason
are apparently complementary, different routes to the same truths.
The first conception not only discredits scientific knowlcdge and
finally even leads to suggesting the impossibility of a science of
biology, it also in volves a curious paradox; for the work entitled
Critique de la raison dialectique is the result of the author's exercise
of his own analytical reason: he defines, distinguishes, classifies and
opposes. This philosophical treatise is no different in kind from the
works it examines and with which it engages in discussion, if only to
245
THE SAVAGE MIND
condemn them. It is difficult to sec how analytical reason could be

applied to dialectical rcason and daim to establish it, if the two are
dcfined by mutually exclusive characteristics. The second con-
ception is open to a different objection: if dialectical and analytical
reason ultimately arrive at the same results, and if their respective
truths merge into a single truth, then, one may ask in what way they
are opposed and, in particular, on what grounds the former should
be pronounced superior to the latter. Sartre's endeavour seems
contradictory in the one case and superfluous in the other.
How is the paradox to be explained, and avoided? Sartre attributes
a reality sui generis to dialectical reason in both the hypotheses
bctween which he hesitates. It exists independently of analytical
reason, as its antagonist or alternatively its complement. Although
in both our cases lV1arx is the point of departure of our thought, it
secms to me that the Marxist orientation leads to a different view,
namely, that the opposition between the two sorts of reason is rela-
tive, not absolute. It corresponds to a tension within human thought
which may persist indefinitely de f acta, but which has no basis de
jure. In my view dialectical reason is always constitutive: it is the
bridge, forever extended and improved, \Vhich analytical reason
throws out over an abyss; it is unable to see the further shore but it
knows that it is there, even should it be constantly receding. The
term dialectical reason thus covers the perpetual efforts analytical
rcason must make to reform itself if it aspires to account for langu-
age, society and thought; and the distinction between the two forms
of rcason in my view rests only on the temporary gap separating
analytical reason from the understanding of life. Sartre calls analy-
tical reason reason in repose; 1 call the same reason dialectical when
it is roused to action, tensed by its efforts to transcend itself.
In Sartre's té1minology 1 am therefore to be defined as a trans-
ccndental materialist and aesthete. 1 am a transcendental materialist
(p. 124) because 1 do not regard dialectical reason as something other
than analytical reason, upon which the absolute originality of a
human order would be based, but as something additional in analy-
tical reason: the necessary condition for it to venture to undertake
the resolution of the human into the non-human. And I count as an
aesthete since Sartre app lies this term to an yone purporting to study
men as if they were ants (p. 183). But apart from the fact that this
seems to me just the attitude of any scientist who is an agnostic,
there is nothing very compromising about it, for ants with their
246
artificial tunnels, their social life and their chemical messages,

already present a sufficiently tough resistance to the enterprises of
analytical reason ... So r accept the characterization of aesthete in
so far as I believe the ultimate g'oal of the human sciences to be not
to constitute, but to dissolve man. The pre-eminent value of anthro-
pology is that it represents the first step in a procedure which in-
volves others. Ethnographie analysis tries to arrive at invariants
beyond the empirical diversity of hum an societies; and, as the present
work shows, these are sometimes to be found at the most unforeseen
points. Rousseau (2, ch. VIII) foresaw this with his usual acumen:
'One needs to look near at hand if one wants to study men; but to
study man one must learn to look from afar; one must first observe
differences in order to discover attributes'. However, it would not
be enough to reabsorb particular humanities into a general one. This
first enterprise opens the way for others which Rousseau would not
have been so ready to accept and which are incumbent on the exact
natural sciences: the reintegration of culture in nature and finally of
life within the whole of its physico-chemical conditions.*
However, in spite of the intentionally brutal turn given to my
thesis, I am not blind to the fact that the verb 'dissolve' does not in
any way imply (but even excludes) the destruction of the constitu-
ents of the body subjected to the action of another body. The solu-
tion of a solid into a liquid alters the disposition of its molecules. It
also often provides an efficacious method of putting them by so that
they can be recovered in case of need and their properties be better
studied. The reductions I am envisaging are thus legitimate, or
indeed possible, only if two conditions are satisfied. First, the
phenomena subjected to reduction must not be impoverished; one
must be certain that everything contributing to their distinctive
richness and originality has been collected around them. For it is
pointless to pickup a hammer unless to hit the nail on the head.
Secondly, one must be ready to accept, as a consequence of each
reduction, the total overturning of any preconceived idea concern-
ing the level, whichever it may be, one is striving to attain. The idea
of some general humanity to which ethnographie reduction leads,
will bear no relation to any one may have formed in advance. And
when we do finally succeed in understanding life as a fonction of
• The opposition between nature and culture to which I attached much
importance atone time (1, ch. I and 2) now seems to be of primarily methodo-
logical importance.
247
THE SAVAGE MIND
inert matter, it will be to discover that the latter has properties very
different from those previously attributed toit. Levels of reduction
cannot therefore be classed as superior and inferior, for the level
taken as superior must, through the reduction, be expected to com-
municate retroactively some of its richness to the inferior level to
which it will have been assimilated. Scientific explanation consists
not in moving from the complex to the simple but in the replacement
of a less intelligible complexity by one which is more so.
Se en in this light, therefore, my self is no more opposed to others
than man is opposed to the world: the truths learnt through man
are 'of the world', and they are important for this reason.* This
explains why I regard anthropology as the principle of all research,
\vhile for Sartre it raises a problem in the shape of a constraint to
overcome or a resistance to reduce. And indeed what can one make
of peoples 'without history' when one has defined man in terms of
dialectic and dialectic in terms of history? Sometimes Sartre seems
tempted to distinguish two dialectics: the 'true' one which is sup-
posed to be that of historical societies, and a repetitive, short-term
dialectic, which he grants so-called primitive societies whilst at the
same time placing it very near biology. This imperils his whole
system, for the bridge between man and nature which he has taken
such pains to destroy would turn out to be surreptitiously re-estab-
lished through ethnography, which is indisputably a human science
and derntes itself to the study of these societies. Alternatively Sartre
resigns himself to putting a 'stunted and deformed' humanity on
man's sicle (p. 203), but not without implying that its place in
humanity does not belong toit in its own right and is a fonction only
of its adoption by historical humanity: either because it has
begun to internalize the latter's history in the colonial context,
or because, thanks to anthropology itself, historical humanity has
given the blessing of meaning to an original humanity which was
"This enn holds for mathematical truths of which a contemporary logician,
however, says that 'The characteristic of mathematical thought is that it does not
convey truth about the external world' (Heyting, pp. 8-9). But mathematical
thought at any rate reflects the free functioning of the mind, that is, the activity
of the cells of the cerebral cortex, relatively emancipated from any external
constraint and obeying only its own laws. As the mind too is a thing, the func-
tioning of this thing tcaches us something about the nature of things: even pure
reflection is in the last analvsis an interna-lization of the cosmos. Tt illustrates the
structure of what lies ou;side in a symbolic form: 'Logic and logistics are
empirical sciences belonging to ethnography rather than psychology' (Beth,
p. I 51).
without it. Either way the prodigious wealth and diversity of habits,
beliefs and eus toms is allowed to escape; and it is forgotten that
each of the tens or hundreds of t):iousands of societies which have
existed sicle by sicle in the world or succeeded one another since
man's first appearance, has claimed that it contains the essence of
all the meaning and dignity of which human society is capable and,
reduced though it may have been to a small nomad band or a hamlet
lost in the depths of the forest, its daim has in its own eyes rested on
a moral certainty comparable to that which we can invoke in OUT
own case. But whether in their case or OUT own, a good deal of ego-
centricity and naïvety is necessary to believe that man has taken
refuge in a single one of the historical or geographical modes of his
existence, when the truth about man resides in the system of their
differences and common properties.
He who begins by steeping himself in the allegedly self-evident
truths of introspection never emerges from them. Knowledge of
men sometimes seems easier to those who allow themselves to be
caught up in the snare of persona) identity. But they thus shut the
door on knowledge of man: written or unavowed 'confessions' form
the basis of all ethnographie research. Sartre in fact becomes the
prisoner of his Cogito: Descartes made it possible to attain univers-
ality, but conditionally on remaining psychological and individual;
by sociologizing the Cogito, Sartre merely exchanges one prison for
another. Each subject's group and period now take the place of time-
less consciousness. Moreover, Sartre's view of the world and man
has the narrowness which has been traditionally creditcd to closed
societies. His insistence on tracing a distinction hetween the primi-
ti,·e and the civilized with the aid of gratuitons contrasts reflects, in
a scarcely more subtle form, the fondamental opposition he postu-
la tes between myself and others. Y et there is little difference
between the way in which this opposition is formulated in Sartre's
work and the way it would have been formulated by a Melanesian
savage, while the analysis of the practico-inert quite simply revives
the language of animism. *
Descartes, who wanted to found a physics, separated Man from
*lt is precisely because ail these aspects of the savage mind can be discovered
in Sartre's philosophy, that the latter is in my view unqualified to pass judgment
on it: he is prevented from doing so by the very fact of furnishing its equivalent.
To the anthropologist, on the contrary, this philosophy (like ail the others) affords
a first-class ethnographie document, the study of which is essential to an under-
standing of the mythology of our own time.
2 49
THE SAVAGE MIND
Society. Sartre, who daims to found an anthropology, separates his

own society from others. A Cogito - which strives to be ingenuous
and raw - retreats into individualism and empiricism and is lost in
the blind alleys of social psychology. For it is striking that the situa-
tions which Sartre uses as a starting point for extracting the formai
conditions of social reality - strikes, boxing matches, football
matches, bus-stop queues - are all secondary incidentals of life in
society; and they cannot therefore serve to disclose its foundations.
This axiomatic, so far removed from the anthropologist's,
is all the more disappointing when he feels himself very close
to Sartre whenever the latter applies himself, with incomparable
artistry, to grasping, in its dialectical movement, a present or past
social experience within our own culture. Sartre then does what
every anthropologist tries to do in the case of different cultures: to
put himself in the place of the men living there, to understand the
principle and pattern of their intentions, and to perceive a period
or a culture as a significant set. ln this respect we can often learn
from him, but these are lessons of a practical, not a theoretical,
nature. It is possible that the requirement of 'totalization' is a great
novelty to some historians, sociologists and psychologists. It has
been taken for granted by anthropologists ever since they learned it
from Malinowski. But i\falinowski's deficiencies have also taught us
that this is not where explanation ends. It only begins when we have
succeeded in constituting our object. The role of dialectical reason
is to put the human sciences in possession of a reality with which it
alone can furnish them, but the properly scientific work consists in
decomposing and then recomposing on a different plane. With all
due respect to Sartrian phenomenology, we can hope to find in it
only a point of departure, not one of arrivai.
Furthermore, dialectical reason must not let itself be carried away
by its own elan, nor must the procedure leading to the comprehen-
sion of an other reality attribute toit, in addition toits own dialectical
features, those appertaining to the procedure rather than to the
object: it does not follow from the fact that all knowledge of others
is dialectical, that others are wholly dialectical in every respect. By
making analytical reason an anti-comprehension, Sartreoften cornes
to refuse it any reality as an integral p. rt of the object of compre-
hension. This paralogism is alreadyapparent in his mannerof invok-
ing hlstory, for one is hard putto it to see whether it is meant to be
the history men make unconsciously, history of men consciously
made by historians, the philosopher's interpretation of the history

of men or his interpretation of the history of historians. The diffi-
culty becomes even greater, however, when Sartre endeavours to
explain the life and thought of the present or past members not of
his own society but of exotic societies.
He thinks, rightly, that this attempted comprehension stands no
chance of succeeding unless it is dialectical; and he concludes,
wrongly, that the relationship between native thought and his know-
ledge of it, is that of a constitutive to a constituted dialectic, and
thus, by an unforeseen detour, he repeats ail the illusions of theorists
of primiti\·e mentality on his own account. It seems even Jess toler-
able to him than to Levy-Bruhl thatthe savage should possess'com-
plex understanding' and should be capable of analysis and demons-
tration. Of the Ambrym native, made famous by Deacon's work,
who was able to show the field-workcr the functioning of his
marriage rules and kinship system by a diagram in the sand (an
aptitude in no way exceptional as plenty of similar cases are recorded
in ethnographie literature) Sartre says: 'It goes without saying that
this construction is not a thought: it is a piece of manual work
governed by unexpressed synthetical knowledge' (p. 505). Granted:
but then the same must be said of a professorat the Ecole Polytech-
nique demonstrating a proof on the blackboard, for cvcry cthno-
grapher capable of dialectical comprehension is intimatcly per-
suaded that the situation is exactly the same in both cases. So it
would follow that a11 reason is dialectical, which for my part 1 am
prepared to concede, since dialectical reason seems tome like analy-
tical reason in action; hut then the distinction between the two forms
of reason which is the basis of Sartre's entcrprise would become
pointless.
1 must now confess to having myself unintentionally and unwit-
tingly lent support to these erroneous ideas, by having seemed all
too often in Les structures élémentaires de la parenté as if 1 were seek-
ing out an unconscious genesis of matrimonial exchange. I should
have made more distinction between exchange as it is expressed
spontaneously and forcefully in the praxis of groups and the con-
scious and deliberate rules by which these same groups - or their
philosophers - spend their time in codifying and controlling it. If
there is anything to be learnt from the ethnographie enquiries of
the last twenty years, it is that this latter aspect is much more
important than has generally been realized by observers, who
THE SAVAGE MIND
labour under the same delusion as Sartre. Thus we must, as Sartre

adrncates, apply dialcctical reason to the knowledge of our own and
other societics. But we must not Jose sight of the fact that analytical
rcason occupies a considerable place in ail of thcm and that, as it is
present, the approach we adopt must also allow us to rediscover it
there.
But e\·en were it not present, Sartre's position would not be
improved. For in this case exotic societies would merely confront
us, in a more general manner than others, with an unconscious
teleology, which, although historical, completely eludes human
history: that of which certain aspects are revealed by linguistics and
psycho-analysis and which rests on the interplay of biological mech-
anisms (structure of the hrain, lesions, internai secretions) and psy-
chological ones. There, it seems to me, is 'the bone' (to borrow a
phrase from Sartre) which his critique does not manage to break,
and moreover carcs nothing about, which is the most serious charge
one could level at it. For language does not consist in the analy-
tical reason of the old-style grammarians nor in the dialectic consti-
tuted by structural linguistics nor in the constitutive dialectic of
individ ual praxis facing the practico-inert, since ail three presu pp ose
it. Linguistics thus presents us with a dialectical and totalizing
entity but one outside (or beneath) consciousness and will. Langu-
age, an unreftecting totalization, is human reason which has its
reasons and of which man knows nothing. And if it is objected that
it is so only for a subject who internalizes it on the basis of linguistic
theory, my reply is that this way out must be refused, for this subject
is one who speaks: for the same light which reveals the nature of
language to him also reveals to him that it was so when hé did not
know it, for he already made himself understood, and that it will
remain so tomorrow without his being awarc of it, since his discourse
never was and never will be the result of a conscious totalization of
linguistic laws. Rut if, as speaking subject, man can find his apodictic
experience in an other totalization, there scems no longer any reason
why, as li\·ing subject, he should not have access to the same experi-
ence in other, not necessarily human, but living beings.
This method could also Jay daim to the name 'progressive-
regressive'; in fact, what Sartre desc:-ibes as such is the \'ery method
anthropologists ha\·e been practising for many years. But Sartre
restricts it to its preliminary step. For our method is progressive-
regressive not once but twice over. In the first stage, we observe the
datum of experience, analyse it in the present, try to grasp its

historical antecedents as far as we can delve into the past, and bring
ail these facts back to the light of day to incorporate them into a
meaningful totality. The second st~ge, which repeats the first on a
different plane and at a different level, then begins. This intemalized
human thing which we have sought to provide with ail its wealth and
originality, only fixes the distance analytical reason must cover, the
leap it must make, to close the gap hetween the ever unforeseen
complexity of this new object and the intellectual means at its
disposai. It must therefore transform itself as dialetical reason, in
the hope that once flexible, widened and strengthened, by its agency
this unforeseen object will be assimilated to others, this novel
totality will he merged into othcr totalities and that thus little hy
little clambering on to the mass of its conqucsts, dialectical reason
will dcscry othcr horizons and other objects. No doubt the pro-
cedure would go astray if it were not, at cvery stage and, above ail,
when it seemed to have run its course, ready to retrace its steps and
to double hack on itself to preserve the contact with that experi-
enced totality which serves both as its end and means. This return
on itself is in my view a verification, rather than, as Sartre regards
it, a demonstration, for, as I sec it, a conscious being awarc of itself
as such poses a problem to which it provides no solution. The dis-
covery of the dialectic subjects analytical reason to an imperative
requirement: to account also for dialectical reason. This standing
requirement relentlessly forces analytical reason to extend its pro-
gramme and transform its axiomatic. But dialectical reason can
account neitherf or itself nor for analytical reason.
It will be objected that this expansion is illusory since it is always
accompanied by a contraction in meaning, and we should abandon
the substance for the shadow, clarity for obscurity, the man ifest for
the conjectural, truth for science fiction (Sartre, p. 129). Again,
Sartre would have to show that he himself avoids this dilemma, in-
herent in every attempt at explanation. The real question is not
whether our endeavour to understand involves a gain or a loss of
meaning, but whether the meaning we preserve is of more value
than that we have been judicious enough to relinquish. In this
respect Sartre seems to have remembered only half of Marx's and
Freud's combined tesson. They have taught us that man has mean-
ing only on the condition that he view himself as meaningful. So far
I agree with Sartre. But it must be added that this meaning is nei1er
253
THE SAVAGE MIND
the right one: superstructures are faulty acts which have 'made it'
socially. Hence it is vain to go to historical consciousness for the
truest meaning. \Vhat Sartre catis dialectical reason is only a recon-
struction, by what he catis analytical reason, of hypothetical moves
about which it is impossible to know - unless one should perform
them without thinking them -whether they bear any relation at ail to
what he tells us about them and which, if so, would be definable in
terms of analytical reason atone. And so we end up in the paradox
of a system which invokes the criterion of historical consciousness to
distinguish the 'primitive' from the 'civilized' but - contrary toits
claim - is itself ahistorical. I t offers not a concrete image of history
but an abstract schema of men making history of such a kind that
it can manifest itself in the trend of their lives as a synchronie
totality. lts position in relation to history is therefore the same as that
of primitives to the eternal past: in Sartre's system, history plays
exactly the part of a myth.
Indeed, the problem raised by the Critique de la raison dialectique
is reducible to the question: under what conditions is the myth of
the French Revolution possible? And 1 am prepared to grant that
the contemporary Frenchman must believe in this myth in order
fully to play the part of an historical agent and also that Sartre's
analysis admirably extracts the set of formai conditions necessary
if this result is to be secured. But it does not follow that his meaning,
just because it is the richest (and so most suited to inspire practical
action), should be the truest. Here the dialectic turns against itself.
This truth is a matter of context, and if we place ourselves outside
it - as the man of science is bound to do -what appeared as an experi-
enced tru th first becomes confused and finally disappears altogether.
The so-called men of the Left still cling to a period of contemporary
history which btstowed the blessing of a congruence between prac-
tical imperati ves and schemes of interpretation. Perhaps this golden
age of historical consciousness has already passed; and that this
eventuality can at any rate be envisaged proves that what we have
here is only a contingent context like the fortuitous 'focusing' of an
optical instrument when its object-glass and eye-piece move in
relation to each other. \Ve are still 'in focus' so far as the French
Revolution is concerned, but so we should have been in relation to
the Fronde had we lived earlier. The former will rapidly cease to
afford a coherent image on which our action can be modelled, just
as the latter has already clone. \Vhat we leam from reading Retz is
that thought is powerless to extract a scheme of interpretation from

events long past.
At first sight, there seems no doubt: on one sicle the privileged, on
the other the humble and exploited ;'how could we hesitate? We are
Frondeurs. However, the people of Paris were being manoeuvred
by noble houses, whose sole aim was to arrange their own affairs
with the existing powers, and by one half of the royal family which
wanted to oust the other. And now we are already only half Fron-
deurs. As for the Court, which took refuge at Saint-Germain, it
appears at first to have been a faction of good for nothings vegetating
on their privileges and growing fat on exactions and usury at the
expense of the collectivity. But no, it had a fonction all the same since
it retained military power; it conducted the struggle against
foreigners, the Spaniards, whom the Frondeurs invited without
hesitation to invade the country and impose their wills on this same
Court which was defending the fatherland. The scales, however, tilt
the other way again: the Frondeurs and Spaniards together formed
the party of peace. The Prince de Condé and the Court only sought
warlike adventures. \Ve are pacifists and once again bccome Fron-
deurs. But nevertheless did not the militaryexploits of Mazarin and
the Court extend France to its present fronticrs, thus founding the
state and the nation? Without them we should not be what we are
today. So here we are on the other sicle again.
lt suffices therefore for history to move away from us in time or
for us to move away from it in thought, for it to cease to be internal-
izable and to !ose its intelligibility, a spurious intclligibility attaching
to a temporary internality. I am not however suggesting that man
can or should sever himself from this internality. It is not in his
power to do so and wisdom consists for him in seeing himself live
it, while at the same time knowing (but in a different register) that
what he lives so completely and intensely is a myth - which will
appear as such to men of a future century, and perhaps to himself a
few years hence, and will no longer appear at ail to men of a future
millenium. Ali meaning is answerable to a lesser meaning, which
gives it its highest meaning, and if this regression finally ends in
recognizing 'a contingent law of which one can say only: it is thus,
and not otherwise' (Sartre, p. 128), this prospect is not alarming to
those whose thought is not tormented by transcendence even in a
latent form. For man will have gained ail he can reasonably hope
for if, on the sole condition of bowing to this contingent law, he
2 55
THE SAVAGE MIND
succeeds in determining his form of conduct and in placing ail else

in the realm of the intelligible.
Sartre is certainly not the only contemporary philosopher to have

valued history above the other human sciences and formed an
almost mystical conception of it. The anthropologist respects his-
tory, but he does not accord it a special value. He conceives it as a
study complementary to his own: one of them unfurls the range of
human societies in time, the other in space. And the difference is
even Jess great than it might seem, since the historian strives to
reconstruct the picture of vanishcd societies as they were at the
points which for them corresponded to the present, while the ethno-
grapher does his best to reconstruct the historical stages which
temporally preceded their existing form.
This symmetry bt:tween history and anthropology seems to be
rejected by philosophers who implicitly or explicitly deny that dis-
tribution in space and succession intime afford equivalent perspec-
tives. ln their eyes some special prestige seems to attach to the tem-
poral dimension, as if diachrony were to establish a kind of intelligi-
bility not merely superior to that provided by synchrony, but above
ail more specifically human.
1t is easy to expia in, if not to justify, this preference. The diversity
of social forms, which the anthropologist grasps as deployed in
space, present the appearance of a discontinuous system. Now,
thanks to the temporal dimension, history seems to restore tous, not
separate states, but the passage from one state to another in a con-
tin111111,; forrn. Antl as \\'e helic,·c that wc apprchcn<l the trend of 011r
pl·rsll11;1J histor~· as a conrinuous change, hiscorical knoll'le<lgc
appe1rs LO co11licrn the t'\·idelil't of inncr :kfü\e. 1 liswry seem,; co do
müïe than describe beings tous from the outside, or at best giYe us
intermittent flashes of insight into internalities, each of which are
so on tlwir own account while remaining external to t>ach other: it
appears to re-establish our connection, outsi<le oursel\'cs, with the
vcry essence of change.
Therc would be plenty to say about this supposcd totalizing con-
tinuity of the self which seems to me to be an illusion sustained by
the demands of social life - and consequently a reflection of the
external on the internai - rather than the abject of an apodictic
experience. But there is no need to resolve this philosophical proh-
lem in order to perceive that the proposed conception of history
256
JIISTORY AND DIALECTIC
corresponds ton o kind of reality. As historical knowledge i s claimed

to be privileged, 1 feel entitled (as 1 would not otherwise feel) to
make the point that there is a twofold antinomy in the very notion
of an historical fact. For, ex hypotlzesi, a historical fact is what really
took place, but where did anything take place? Each episode in a
revolution or a war resolves itself into a multitude of individual
psychic movements. Each of these movements is the translation of
unconscious development, and these resolve themselves into cere-
bral, hormonal or nervous phenomena, which themselves have
reference to the physical or chemical order. Consequently, historical
facts are no more given than any other. It is the historian, or the
agent of history, who constitutes them by abstraction and as though
under the threat of an infini te regress.
\Vhat is true of the constitution of historical facts is no Jess so of
their selcction. From this point of view, the historian and the agent
of history choose, sever and carve them up, for a truly total history
would confront them with chaos. Every corner of space conceals a
multitude of individuals each of whom totalizes the trend of history
in a manner which cannot be compared to the others; for any one
of these individuals, each moment of timc is incxhaustibly rich in
physical and psychical incidents which all play their part in his
totalization. Even history which daims to be uni versai is still only a
juxtaposition of a few local histories within which (and between
which)very much more is left outthan is put in. And it would be vain
to hope that by increasing the number of collabora tors and making
research more intensive one would obtain a bettcr result. In so far
as history aspires to mcaning, it is doomed to select regions, periods,
groups of men and indiviéluals in these groups and to make them
stand out, as discontinuous figures, against a continuity barely good
enough to be used as a backdrop. A truly total history would cancel
itself out - its product would be nought. What makes history
possible is that a sub-set of events is found, for a given period, to
have approximately the same significance for a contingent of indivi-
duals who have not necessarily experienced the events and may even
consider them at an interval of several centuries. His tory is therefore
ne\·er history, but history-for.* lt is partial in the sense of being
• Quite so, will be the comment of the supporters of Sartre. But the latter's
whole endeavour shows that, though the subjectivity of history-for-me can make
way for the objectivity of history-for-us, the 'I' can still only be converted into
'we' by condemning this 'we' to being no more than an 'I' raised to the power of
two, itself hermetically sealed off from the other 'we's. The price so paid for the
257
THE SAVAGE MIND
biased cven when it daims not to be, for it inevitably remains partial
- that is, incomplete - and this is itself a form of partiality. When one
proposes to write a history of the French Revolution one knows (or
ought to know) that it cannot, simultaneously and under the same
heading, be that of the Jacobin and that of the aristocrat. Ex hypo-
thesi, their respective totalizations (each of which is anti-symmetric
to the other) are equally true. One must therefore choose between
two alternatives. One must select as the principal either one or a
third (for there are an infinite number of them) and give up the
attempt to find in history a totalization of the set of partial totaliza-
tions; or alternatively one must recognize them all as equally real:
but only to discover that the French Revolution as commonly con-
ceived never took place.
History does not therefore escape the common obligation of all
knowledge, to employa code to analyse its object, even (and especi-
ally) if a continuous reality is attributed to that object. «<The distinc-
tive features of historical knowledge are due not to the absence of a
code, which is illusory, but toits particular nature: the code consists
in a chronology. There is no history without dates. To be convinced
of this it is sufficient to consider how a pupil succeeds in learning
history: he reduces it to an emaciated body, the skeleton of which is
formed by dates. Not without reason, there has been a reaction
against this dry method, but one which often runs to the opposite
extreme. Dates may not be the whole of history, nor what is most
interesting about it, but they are its sine qua non, for history's entire
originality and distinctive nature lie in apprehending the relation
between be/ore and a/ter, which would perforce dissolve if its terms
could not, at least in principle, be dated.
Now, this chronological coding conceals a very much more com-
plex nature than one supposes when one thinks of historical dates
illusion of having overcome the insoluble antinomy (in such a system) between
my self and others, consists of the assignation, by historical consciousness, of
the metaphysical fonction of Other to the Papuans. By reducing the latter to the
state of means, barely sufficient for its philosophical appetite, historical reason
abandons itself to a sort of intellectual cannibalism much more revolting to the
anthropologist than real cannibalism.
• In this sense tao, one can speak of an antinomy of historical knowledge: if it
daims to reach the continuous it is impossible, being condemned to an infinite
regress; but to render it possible, events must be quantified and thereafter
temporality ceases to be the privileged dimension ofhistorical knowledge because
as soon as it is quantified each event can, for ail useful purposes, be treated as
if it were the result of a choice between possible pre-existents.
as a simple linear series. In the first place, a date denotes a moment

in a succession: d 2 is after d 1 and before d 3- From this point of
view dates only perform the fonction of ordinal numbers. But each
as
date is also a cardinal number and, such, expresses a distance in re-
lation to the dates nearest to it. \Ve use a large number of dates to
code some periods of history; and fewer for others. This variable
quantity of dates applied to periods of equal duration are a gauge
of what might be called the pressure of history: there are 'hot'
chronologies which are those of periods where in the eyes of the
historian numerous events appear as differential elements; others,
on the contrary, where for him (although not of course forthe men
who lived through them) very little or nothing took place. Thirdly
and most important, a date is a member of a class. These classes of
dates are definable bythe meaningful character each date has within
the class in relation to other dates which also belong to it, and by
the absence of this meaningful character with respect to dates
appertaining to a differcnt class. Thus the date 1685 belongs to a
class of which 1610, 1648 and 1715 are likewise members; but it
means nothing in relation to the class composed of the dates: lst,
2nd, 3rd, 4th millenium, nor ciocs it mean anything in relation to
the class of dates: 23 January, 17 August, 30 September, etc.
On this basis, in what would the historian's code consist? Cer-
tainly not in dates, since these are not recurrent. Changes of temp-
craturc can be coded with the help of figures, because the reading of
a figure on the thermometer evokes the return of an earlier situation:
whenever I read o° C, I know that it is freezing and put on my
warmest coat. But a historical date, taken in itself, would have no
meaning, for it has no reference outside itself: if I know nothing
about modern history, the date 1643 makes me none the wiser. The
code can therefore consist only of classes of dates, where each date
has meaning in as muchas it stands in complex relations of correla-
tion and opposition with other dates. Each class is defined by a
frequency, and derives from what might be called a corpus or a
domain of history. Historical knowledge thus proceeds in the same
way as a wireless with frequency modulation: like a nerve, it codes a
continuous quantity- and as such an asymbolic one - by frequencies
of impulses proportional to its variations. As for history itself, it
cannot be represented as an aperiodic series with only a fragment of
which we are acquainted. History is a discontinuous set composed
259
THE SAVAGE MIND
of domains of history, each of which is defined hy a characteristic

frequency and by a differential coding of before and after. lt is no
more possible to pass between the dates which compost the different
domains than it is to do so between natural and irrational numbers.
Or more precisel y: the dates appropria te to each class are irrational
in relation to ail those of other classes.
It is thus not only fallacious but contradictory to conceive of the
historical process as a continuous development, beginning with
prehistory coded in tens or hundreds of millenia, then adopting the
scale of millenia when it gets to the 4th or 3rd millenium, and con-
tinuing as history in centuries inter lardcd, at the pleasure of each
author, with slices of annual history within the century, day to day
history within the year or even hourly h istory within a day. Ali these
dates do not forma series: they are of diffcrent species. To give just
one example, the coding we use in prehistory is not preliminary to
that we employ for modern and contemporary history. Each code
refers to a system of meaning which is, at least in theory, applicable
to the virtual totality of hum an history. The events which are signi-
ficant for one code are no longer so for another. Coded in the system
of prehistory, the most famous episodes in modern and contem-
porary history ccase to he pertinent; except perhaps (and again we
know nothing about it) certain massive aspects of demographic
evolution viewed on a world-wide scale, the invention of the
steam-engine, the discovery of electricity and of nuclear energy.
Given that the general code consists not in dates which can be
ordcrcd as a linear series but in classes of dates each furnishing an
autonomous system of reference, the discontinuous and classifica-
tory nature of historical knowledge emerges clearly. It operates by
means of a rectangular matrix:
where each line represents classes of dates, which may be called

hourly, daily, annual, secular, millenial for the purposes of
260
schematization and which together make up a discontinuous set. In a

system of this type, alleged historical continuity is secured only by
dint of fraudulent outlines.
Furthermore, although the inter~al gaps in each class cannot be
filled in by recourse to other classes, each class taken as a whole
nevertheless always refers back to another class, which contains the
principle of an intelligibility to which it could not itself aspire. The
history of the 17thcentury is 'annual' but the 17th century, as a do main
ofhistory belongs to another class, which codes itin relation to earlier
and la ter centuries; and this do main of modern times in i ts turn be-
comes an element of a class where it appears correlated with and op-
posed to other 'times': the middle ages, antiquity, the present day,
etc. Now,these various domains correspond to histories of different
power.
Biographical and anecdotal history, right at the bottom of the
scale, is low-powered history, which is not intelligible in itself and
only becomes so when it is transferred en bloc to a form of history
of a higher power than itself; and the latter stands in the samc rela-
tion to a class above it. It would, however, be a mistake tothinkthat
we progressively reconstitute a total history by dint of these dove-
tailings. For any gain on one side is offset by a loss on the other.
Biographical and anecdotal history is the least explanatory; but it is
the richest in point of information, for it considers individuals in
their particularity and details for each of them the shades of charac-
ter, the twists and turns of their motives, the phases of their deliber-
ations. This information is schematized, put in the background and
finally done away with as one passes to histories of progressively
greater 'power'.* Consequently, depending on the level on which
he places himself, the historian loses in information what he gains
in comprehension or vice versa, as if the logic of the concrete wished
• Each domain of history is circumscribed in relation to that immediately

below it, inscribed in relation to that above it. So each low-powered history of an
inscribcd domain is complementary to the powerful history of the circumscribed
domain and contradictory to the low-powered history of this same domain (in
so far as it is itself an inscribed domain). Each history is thus accompanied by
an indeterminate number of anti-histories, each complementary to the others: to
a history of grade I, there corresponds an anti-history of grade 2, etc. The
progress of knowledge and the creation of new sciences take place through the
generation of anti-histories which show that a certain order which is possible
only on one plane ceases to be so on anothcr. The anti-history of the French
Revolution envisaged by Gobineau is contradictory on the plane on which the
Revolution had been thought of before him. lt becomes logically conceivable
261
THE SAVAGE :\1INU
t o remind us of its logical nature b y modelling a confused outline of

Godel's theorem in the clay of 'becoming'. The historian's relative
choice, with respect to each domain of history he gives up, is always
confined to the choice between history which teaches us more and
ex plains Jess, and history w hi ch explains more and te aches Jess. The
only way he can avoid the dilemma is by getting outside history:
either by the bottom, if the pursuit of information leads him from
the consideration of groups to that of individuals and then to their
motivations which depend on their persona! history and tempera-
ment, that is to say to an infra-historical domain in the realms of
psychology and physiology; or by the top, if the need to understand
incites him to put history back into prehistory and the latter into
the general evolution of organized beings, which is itself explicable
only in terms of biology, geology and fin ail y cosmology.
There is, however, another way of avoiding the dilemma without
thereby doing away with history. \Ve need only recognize that his-
tory is a method with no distinct abject corresponding to it to
reject the equivalence between the notion ofhistory and the notion
of humanity which some have tried to foigt on us with the una\'Owed
aim of making historicity the last refuge of a transcendental human-
ism: as if men could regain the illusion of liberty on the plane of the
'we' merely by giving up the' I's that are too obviously wanting in
consistency.
In fact history is tied neither to man nor to any particular object.
It consists wholly in its method, which experience proves to be
indispensable for cataloguing the elements of any structure what-
ever, human or non-human, in their entirety. It is therefore far from
being the case that the search for intelligibility cornes to an end in
history as though this were its terminus. Rather, it is history that
serves as the point of departure in any quest for intelligibility. As
we say of certain careers, history may lead to anything, provided
you get out of it.
This furthcr thing to which history leads for want of a sphere of

refcrence of its own shows that whatever its value (which is
(which docs not mcan that it is truc) if one puts oneself on a new plane, which
incidentally Gobineau chose clumsily; that is to say: if one passes from a
history of 'annual' or 'secular' grade (which is also political, social and
idcological) to a history of 'millcnial' or 'multi-millenial' grade (which is also
cultural and anthropological), a proccdure not im·ented by Gobineau which
might be called: Boulainvillicrs' 'transformation'.
262
indisputable) historicalknowledge hasnoclaimto beopposed toother

forms of knO\vledge as a supremely privileged one. \Ve noted above*
that it is alreadv found rooted in the savage mind, and we can now
see why it does ~ot corne to fruitio~-there. The characteristic feature
of the savage mind is its timelessness; its object is to grasp the world
as both a synchronie and a diachronie totality and the knowledge
which it draws therefrom is like that afforded of a room by mirrors
fixed on opposite walls, which reflect each other (as well as objects.
in the intervening space) although without being strictly parallel. A
multitude of images forms simultaneously, none exactly like any
other, so that no single one furnishes more than a partial knowledge
of the decoration and furniture but the group is characterized by
invariant properties expressing a truth. The savage mind deepens
its knowledge with the help of imagi11es mundi. It builds mental
structures which facilita te an understanding of the world in as much
as they resemble it. In this sense savage thought can be defined as
analogical thought.
But in this sense too it differs from domesticated thought, of
which historical knowledge constitutes one aspect. The concern for
continuity which inspires the latter is indeed a manifestation, in the
temporal order, of knowledge which is interstitial and unifying
rather than discontinuous and analogical: instead of multiplying
objects by schemes promoted to the role of additional objects, it
seems to transcend an original discontinuity by relating objects to
one another. But it is this reason, wholly concerned with closing
gaps and dissolving differences, which can properly be called 'analy-
tical'. By a paradox on which much stress has recently been laid, for
modern thought 'continuity, variability, relativity, determinism go
together' (Auger, p. 4ï5).
This analytic, abstract continuity will doubtless be opposed to
that of the praxis as concrete individuals live it. But this latter con-
tinuity seems no less derivative than the former, for it is only the
conscious mode of apprehending psychological and physiological
processes which are themselves discontinuous. I am not disputing
that reason develops and transforms itself in the practical field:
man's mode of thought reflects his relations to the world and to
men. But in order for praxis to be living thought, it is necessary first
(in a logical and nota historical sense) for thought to exist: that is to
say, its initial conditions must be given in the form of an objective
•Pp. 242-3.
THE SAVAGE MIND
structure of the psyche and brain without which there would be

neither praxis nor thought.
\Vhen therefore I describe savage thought as a system of concepts
embedded in images, I do not corne anywhere near the robinson-
nades* (Sartre, pp. 642-3) of a constitutive constituent dialectic:
all ccnstitutive reason presupposes a constituted reason. But even
if one allowed Sartre the circularity which he invokes to dispel the
'suspect character' attaching to the first stages of his synthesis, what
he proposes really are 'robinsonnades', this time in the guise of
descriptions of phenomena, when he daims to restore the sense of
marriage exchange, the potlatch or the demonstration of his tribe's
marriage mies by a l\!Ielanesian savage. Sartre then refers to a com-
prehension which has its being in the praxis of their organizers, a
bizarre expression to which no reality corresponds, except perhaps
the capacity which any foreign society presents to anyone looking
at it from the outside, and which leads him to project the lacunae in
his own observation on toit in the form of positive attributes. Two
examples will show what I mean.
No anthropologist can fail to be struck by the common manner of
conceptualizing initiation rites employed by the most diverse
societies throughout the world. Whether in Africa, America, Aus-
tralia or Melanesia, the rites follow the same pattern: first, the
novices, taken from their parents, are symbolically 'killed' and kept
hidden in the forest or bush where they are put to the test by the
Beyond; after this they are 'reborn' as members of the society. \Vhen
they are returned to their natural parents, the latter therefore
simulate all the phases of a new delivery, and begin a re-education
even in the elementary actions of feeding or dressing. It would be
tempting to interpret this set ot phenomena as a proof that at this
stage thought is wholly embedded in praxis. But this would be
seeing things back to front, for it is on the contrary scientific praxis
which, among ourselves, has emptied the notions of death and birth
of everything not corresponding to mere physiological processes
and rendered them unsuitable to convey other meanings. In societies
with initiation rites, birth and death provide the material for a rich
and varied conceptualization, provided that these notions (like so
many others) have not been stripped by any form of scientific
• This term alludes ta Robinson Crusoe and the Swiss family Robinson
whose 'creation' of civilization was not a genuine invention but merely an
application of their pre-existing knowledge. [Trans. note.]
knowledgeoriented towards practical returns-which they Jack-of the

major part of a meaning which transcends the distinction between
the real and the imaginary: a complete meaning of which we can
now hardly do more than evoke the ghost in the reduced setting of
figurative language. What looks tous like being embedded in praxis
is the mark of thought which qui te genuinely takes the words it uses
seriously, whereas in comparable circumstances we only 'play' on
words.
The taboos on parents-in-law furnish the matter for a cautionary
tale leading to the same conclusion by a different route. Anthro-
pologists have found the frequent prohibition of any physical or
verbal contact between close affines so strange that thcy have exer-
cised their ingenuity in multiplying explanatory hypotheses, with-
out always making sure that the hypotheses are not rendereJ re-
dundant by one another. Elkin for instance explains the rarity of
marriagewith the patrilateral cousin inAustralia by the rule that as a
r.ian has to avoid any contact with his mother-in-law, he will be
wise to choose the latter among women entirely outside his own
local group (to which his father's sisters belong). The aim of the
rule itself is supposed to he to prevent a mother and daughter from
being rivais for the affections of the same man; finally, the taboo is
supposed to be extended by contamination to the wifc's maternai
grandmother and her husband. There are thus four concurrent
interpretations of a single phenomenon: as a fonction of a type of
marriage, as the result of a psychological calculation, as protection
against instinctive tendencies and as the product of association by
contiguity. This, however,still does not satisfy Elkin, for in hisview
the taboo on the father-in-law rests on a fifth explanation: the father-
in-law is the creditor of the man to whom he has given his daughter,
and the son-in-law feels himself to be in a position of inferiority in
relation to him (Elkin 4, pp. 66-7, I 17-20).
I shall content myself with the last explanation which perfectly
covers ail the cases considered and renders the others worthless by
bringing out their naïvety. But why is it so difficult to put these
usages into their proper place? The reason is, I think, that the usages
of our own society which could be compared with them and might
furnish a landmark to identify them by, are in a dissociated form
among ourselves, while in these exotic societies they appear in an
associated one which makes them unrecognizable tous.
265
THE SAVAGE MIND
\Ve are acquainted with the taboo on parents-in-law or at least

with its approxima te eq uivalent. By the same token we are forbidden
to address the great of this world and obliged to keep out of their
way. All protocol asserts it: one does not speak first to the Queen of
England or the President of the French Republic; and we adopt the
samc reserve when unforeseen circumstances create conditions of
doser proximity between a superior and ourselves than the social
distance between us warrants. Now, in most societies the position
of wife-giver is accompanied by social (and sometimes also eco-
nomic) su periority, tha t of wife-taker by inferiority and dependence.
This inequality between affines may be expressed objectively in
institutions as a fluid or stable hierarchy, or itmay be expressed sub-
jectively in the system of interpersonal relations by means of
privileges and prohibitions.
There is therefore nothing mysterious about these usages which
our own experience enables us to see from the inside. We are discon-
certed only by their constitutive conditions, different in each case.
Among ourselves, they are sharply separated from other usages and
tied to an unambiguous context. In exotic societies, the same usages
and the same context are as it were embedded in other usages and a
different context: that of family ties, with which they seem to us
incompatible. W e find it hard to imagine that, in priva te, the son-in-
law of the President of the French Republic should regard him as
the head of the state rather than as his father-in-law. And although
the Queen of England's husband may behave as the first of her
subjects in public, there are good reasons for supposing that he is
just a husband when they are alone together. It is either one or the
other. The superficial strangeness of the taboo on parents-in-law
arises from its being both at the same time.
Consequently, and as we have already found in the case of oper:i-
tions of understanding, the system of ideas and attitudes is here
presented only as embodied. Considered in itself, this system has
nothing about it to baffle the anthropologist. My relation to the
President of the Republic is made up entirely of negative ùbser-
vances, since, in the absence of other ties, any relations we may have
are wholly defined by the rule that I should not speak unless he
invites me to do so and that I should remain a respectful distance
from him. But this abstract relation need only be clothed in a
concrete relation and the attitudes appropriate to each to acc~1mu
late, for me to find myself as embarrassed by my family as an
266
lllSTORY AND DIALECTIC
Australian aborigine. \Vhat appears tous as greater social ease and

greater intellectual mobility is th us due to the fact that we prefer to
operate with detached pieces, if not indeed with 'small change',
while the native is a logical hoardê~: he is forever tying the threads,
unceasingly turning over all the aspects of reality, whether physical,
social or mental. \Ve traffic in our ideas; he hoards them up. The
savage mind puts a philosophy of the finite into practice.
This is also the source of the renewed interest in it. This language
with its limited vocabulary able to express any message by combina-
tions of oppositions between its constitutive units, this logic of com-
prehension for which contents are indissociable from form, this
systematic of finite classes, this universe made up of meanings, no
longer appears tous as retrospective witnesses of a time when: ' ... le
ciel sur la terre l\larchait et respirait dans un peuple de dieux',* and
which the poet only evokes for the purpose of asking whether or not
it is to be regretted. This time is now restored to us, thanks to the
discovery of a universe of information where the laws of savage
thought reign once more: 'heaven' too, 'walking on earth' among a
population of transmitters and receivers whose messages, while in
transmission, constitute objects of the physical world and can be
grasped both from without and from within.
The idea that the universe 1lf primitives (or supposedly such) con-
sists principally in messages is not new. But until recently a negative
value was attributed to what was wrongly taken to be a distinctive
characteristic, as though this difference between the universe of the
primitives and our own contained the explanation of their mental
and technological inferiority, when what it does is rather to put
them on a par with modern theorists of documentation.t Physical
science had to discover that a semantic universe possesses ail the
characteristics of an object in its own right for it to be recognized
that the manner in which primitive peoples concep~ualize their
" i.e. 'when hea\'en walked and breathed on earth among a population of
Gods'. From A.de \lusset 'Rolla' 1833 reprinted in Poesies Nouvelles. [Trans.
note.]
t The documentalist neither disallows nor disputes the substance of the
works he analyses in order to derive the constitutive units of his code or to adapt
them, either by combining them among themselves or, if necessary, decomposing
them into liner units. He thus treats the authors as gods whose revelations are
written down, instead of bcing inscribed into bcings and things, but which have
the same sacred value, which attaches to the supremely meaningful character
that, for methodological or ontological reasons, it is ex h:ypothesi necessary to
recognize in them in both cases.
THE SAVAGE .MIND
world is not merely coherent but the very one demanded in the
case of an object whose elementary structure presents the picture of
a discontinuous complexity.
The false antinomy between logical and prelogical mentality was
surmounted at the same time. The savage mind is logical in the
samc sense and the same fashion as ours, though as our own is only
when it is applied to knowledge of a universe in which it recognizes
physical and semantic properties simultancously. This misunder-
standing once dispelled, it remains no less true that, contrary to
Levy-Bruhl's opinion, its thought proceeds through understanding,
not affectivity, with the aid of distinctions and oppositions, not by
confusion and participation. Although the term had not yet corne
into use, numerous texts of Durkheim and Mauss show that they
understood that so-called primitive thought is a quantified form of
thought.
It will be objected that there remains a major difference between
the thought of primitives and our own: Information Theory is con-
cerned with genuine messages whereas primitives mistake mere
manifestations of physical determinism for messages. Two con-
siderations, however, deprive this argument of any weight. In the
first place, Information Theory has been generalized, and it extends
to phenomena not intrinsically possessing the character of messages,
notably to those of biology; the illusions of totemism have had at
least the merit of illuminating the fondamental place belonging to
phenomena of this order, in the internai economy of systems of
classification. In treating the sensible properties of the animal and
plant kingdoms as if they were the elements of a message, and m
discovering 'signatures' - and so signs - in them, men have made
mistakes of identification: the meaningful element was not always
the one they supposed. But, without perfected instruments which
would have permitted them to place it where it most often is -
namely, at the microscopie level-theyalready discerned 'as through
a glass darkly' principles of interpretation whose heuristic value and
accordance with reality have been revealed tous only through very
recent inventions: telecommunications, computers and electron
m1croscopes.
Above all, during the period of their transmission, when they
have an objective existence outside the consciousness oftransmitters
and receivers, messages display properties which they have in
common with the physical world. Hence, despite their mistakes
268
with regard to physical phenomena (which were not absolute but

relative to the level where they grasped them) and even though they
interpreted them as if they were messages, men were nevertheless
able to arrive at some of their properties. For a theory of information
to be able to be evolved it was undoubtedly essential to have dis-
covered that the universe of information is part of an aspect of the
natural world. But the validity of the passage from the laws of nature
to those of information once demonstrated, implies the validity of
the reverse passage - that which for millenia has allowed men to
approach the laws of nature by way of information.
Certainly the properties to which the savage mind has access are
not the same as those which have commanded the attention of
scientists. The physical world is approached from opposite ends in
the two cases: one is supremely concrete, the other supremely
abstract; one proceeds from the angle of sensible qualities and the
other from that of formai properties. But if, theoretically at least
and on condition no abrupt changes in perspective occurred, these
two courses might have been destined to meet, this explains that
they should have both, independently of each other in time and
space, !cd to two distinct though equally positive sciences: one
which flowered in the neolithic period, whose theory of the sensible
order provided the basis of the arts of civilization (agriculture,
animal husbandry, pottery, weaving, conservation and preparation
of food, etc.) and which continues to provide for our basic needs by
these means; and the other, which places itself from the start at the
level ofintelligibility, and ofwhich contemporary science is the fruit.
\Ve have had to wait until the rniddle of this century for the
crossing of long separated paths: that which arrives at the physical
world by the detour of communication, and 1:hat which as we have
recently corne to know, arrives at the world of communication by
the detour of the physical. The entire process of human knowledge
thus assumes the characti!r of a closed system. And we therefore
remain faithful to the inspiration of the savage mind when we
recognize that the scientific spirit in its most modern form will, by
an encounter it alone could have foreseen, have contributed to
legitimize the principles of savage thought and to re-establish it in
its rightful place.
12 June-16 October 1961
BIBLIOÇRAPHY
ALVIA;>;O, F.DE: ':\otas etnograficas sobre os Ticunas do Alto Solimôes',

Revista do lnstituto Historico e Geografico Brasileiro, vol. 180, 1943.
ANoERSO:'\, A.J.O. and DIBHLE, Ch. E.: Florentine Codex, Book 2, Santa
Fé, X.1\1., 1951.
A!'DERSON, E.: Plants, Man and J,ife, London, 1954.
A::-;THO::-;Y, H. G.: Field Book of North American Mammals, New York,
1928.
AUGER, P.: 'Structures et complexités dans l'univers de l'antiquité à nos
jours', Cahiers d'histoire mondiale, vol. 6, no. 3, Neuchâtel, 1960.
BALANDIER, G.: 'Phénomènes sociaux totaux et dynamique sociale',
Cahiers internationaux de sociologie, vol. 30, Paris, 1961.
BALZAC, H.DE: La Comédie humaine, 10 vol, Bibl. de la Pléiade, Paris,
1940-50.
BARRETT, S. A.: 'Totemism among the Miwok', Journal of American Folk-
lore, vol. 21, Boston-New York, 1908.
BARROWS, D.P.: The Ethno-Botany of the Coahuila lndians of Southern
Cahfomia, Chicago, 1900.
BATESO~. G.: Naven, Cambridge, 1936.
BEATTIE, J.H.M.: 'Nyoro Persona! Names', The Uganda Journal, vol.
21, no. 1, Kampala, 1957.
BECKWITH, M. \V.: 'Mandan-Hidatsa Myths and Ceremonies', M emoirs
of the American Folklore Society, vol. 32, New York, 1938.
BENEDICT, P. K.: 'Chinese and Thai Kin numeratives', Journal of the
American Oriental Society, vol. 65, 1945.
BEIDELMAN, T. O.: 'Right and Left Hand among the Kaguru: A note on
Symbolic Classification', Africa, vol. 31, no. 3, London, 1961.
BERGSON, H.: The two sources of morality and religion ( translated from the
French by R.Ashley Audra and Cloudesley Brereton), New York,
1956.
271
THE SAVAGE MIND
BETH, E. W.: The Fou11datio11s of Jl!athematics, Amsterdam, 1959.

BOAS, F.: ( 1) Introduction to: James Teit, 'Traditions of the Thompson
River Indians of British Columbia', Memoirs of the American Folklore
Society, rnl. 6, 1898.
(2) 'Handbook of American Indian Languages', Part 1, Bulletin 40,
Bureau of American Ethnology, 'Vashington, D.C., 1911.
(3) 'The Origin of Totemism', American Anthropologist, vol. 18, 1916.
(4) 'Ethnology of the Kwakiutl', 35th Annual Report, Bureau of
American Ethnology, 2 vol. (1913-14), Washington, D.C., 1921.
(5) 'Mythology and Folk-Tales of the .Korth American Indians',
reprinted in: Race, Language and Culture, -:'\ew York, 1940.
BocHET, G.: 'Le Poro des Dieli', Bulletin de l'Institut Français d'Afn.que
noire, vol. 21, nos. 1-2, Dakar, 1959.
BowERS, A. ,V.: 1'vlandan Social and Ceremonial Organization, Chicago,
1950.
BRoNDAL, V.: Les Parties du discours, Copenhagen, 1928.
BROUILLETTE, B.: La Chasse des animaux à fourrure au Canada, Paris,
1934.
CAPELL, A.: 'Language and World View in the Northern Kimberley,
,V. Australia', Southwestern Journal of Anthropology, vol. 16, no. 1,
Albuquerque, 1960.
CARPENTER, E.: (Persona! communication, 26.10.61.)
CHARBONNIER, G.: 'Entretiens avec Claude Lévi-Strauss', Les Lettres
1\!ouve//es 10, Paris, I96I.
CoGHLAX, H. H.: 'Prehistoric Copper and some Experiments in Smelt-
ing', Transactions of the Nm:comen Society, vol. 20, London, I939-40·
CoLBACCllINI, P.A. et ALBISETTI, P.C.: Os Bororos Orientais, Sâo Paulo-
Rio de Janeiro, 1942.
Co:\ITE, A.: Cours de philosophie positi·ve, 6 vol., Paris, n. ed., I908.
CoxKLIN, H. C.: (I) The Relation of Hanunoo Culture to the Plant World,
Doctoral Dissert., Yale, 1954 (microfilm).
(2) 'Hanunoo Color Categories', Southu:estern Journal of Anthropo-
logy, vol. II, no. ·h Albuquerque, 1955.
(3) 'Betel Chewing among the Hanunüo', Proceedings of the 4th Far-
eastern Prehistoric Congress, Paper no. 56, Quezon City (Nat. Res.
Council of the Philippines), 1958.
(4) 'Lexicographical Treatment of Folk Taxonomies', International
Journal of American Linguistics, vol. 28, -:'\ew York, 1962.
CoOKE, Ch.A.:'Iroquois Persona! Names - Their Classification', Pro-
ceedings of the American Philosophica/ Society, vol. 96, fasc. 4, Phila-
delphia, 1952.
CRUZ, 1\1.: 'Dos nomes entre os Bororos', Revista do lnstituto Historico e
Geografico Brasileiro, vol. 175 (I940), 1941.
272
BIBLIOGRAPHY
Cuxx1sox, 1. G. :The Luapala Peoplesof Northern Rhodesia, Manchester,

1959·
DELATTE, A.: 'Herbarius: Recherches sur le cérémonial usité chez les
anciens pour la cueillette des simpi-es et des plantes magiques', Bibl. de
la Fac. de Phil. et Let. Univ. de Liège, fasc. LXXXI, Liège-Paris, 1938.
DE:-\XETT, R. E.: Xigerian Studies, London, 1910.
DE::->XLER, J.G.: 'Los nombres indigenas en guarani', Physis, no. 16,
Buenos Aires, 1939·
DEXS'.\!ORE, F.: (1) 'Pap1go i\1uûc', Bulletin 90, Bureau of American
Ethnology, Washington, D.C., 1929.
(2) 'i\Iandan and Hidatsa i\lusic', Bulletin 80, Bureau of American
Ethnology, Washington, D.C., 1923.
D1.u10:-.:o, S.: 'Anaguta Cosmography: The Linguistic and Behavioral
Implications', Anthropological Linguistics, vol. 2, no. 2, 1960.
DICKENS, Ch.: Great Expectations, Complete \Vorks, 30 vol., 7'\ew York
and London, n.d.
DIETERLEN, G.: (1) 'Les Correspondances cosmo-biologiqucs chez les
Soudanais', Journal de Psychologie normale et pathologique, +3rd year,
no. 3, Paris, 1950.
(2) 'Classification des végétaux chez les Dogon', Journal de la Société
des Africanistes, tome XXII, Paris, 1952.
(3) 'Parenté et i\lariage chez les Dogon (Soudan français)', Africa, vol.
26, no. 2, London, April 1956.
(+) 'Mythe et organisation sociale au Soudan français', Journal de la
Société des Africanistes, tome XXV, fasc. 1 et II, 1955·
(5) 'Mythe et organisation sociale en Afrique occidentale', Journal de
la Société des Africanistes, tome XXIX, fasc. 1, Paris, 1959·
(6) ':0-fote sur le totémisme Dogon', L'Homme, II, l, Paris, 1962.
DORSEY, G. A. and KROEBER, A. L.: 'Traditions of the Arapaho', Field
Columbian il1useum, Pub!. 8I, Anthropological Series, vol. 5, Chicago,
1903.
DORSEY, J.O.: (1) 'Osage Traditions', 6th Annual Report, Bureau of
American Ethnology (1884-5), Washington, D.C., 1888.
(2) 'Siouan Sociology', r5th Annual Report, Bureau of American
Ethnology (1893-+), Washington, D.C., 1897.
DUPIRE, 1\1.: 'Situation de la femme dans une société pastorale (Peul
nomades du ~iger)', in: D. Paulme ed., Femmes d'Afrique Noire,
Paris-La Haye, 1960.
DL'RKHEIM, E.: The Elementary Forms of the Religious Life (translated
from the French by J.W.Swain), London, 1915.
DuRKHEI:>.1, E. and MAUSS, M.: Primitive Classification (translated from
the French by R. Needham), London, 1963.
273
TITE SAVAGE MI'.'iD
ErnrN, A. P.: ( 1) 'Studies in Australian Totemism. Sub-Section, Section

and l\Ioiety Totemism', Oceania, vol. 4, no. 1, 1933-4.
(2) 'Studies in Australian Totemism. The Nature of Australian
Totemism', Oceania, vol. 4, no. 2, 1933-4.
(za) 'Cult Totemism and l\'lythology in Northern South Australia',
Oceania, vol. 5, no. 2, 1934.
(3) 'Kinship in South Australia', Oceania, vol. 8, 9, 10, 1937-40.
(4) The Australian Aborigines, Sydney-London, 3rd ed., 1961.
ELMENDORF, \V. \V. and KROEBER, A. L.: 'The Structure ofTwana Cul-
ture', Research Studies, Jorlonographic Supplement no. 2, Pullman,
\Vashington, 1960.
Eu.mRE, F. H.: 'Ethnobotany of the Navajo', The Uni·i:ersity of New
1\lexico Bulletin, Jl,Ionograph Series, vol. 1, no. 7, Albuquerque, 1943.
EvA!"S-PRITCHARD, E. E.: ( 1) '\Vitchcraft', Africa, vol. 8, no. 4, London,
1955·
(2) Nuer Religion, Oxford, 1956.
(3) 'Zande Clans and Totems', Man, vol. 61, art. no. 147, London,
1961.
FIRTH, R.: (1) 'Totemism in Polynesia', Oceania, vol. 1, nos. 3 and 4,
193o, 1931.
(2) History and Traditions of Tikopia, \Vellington, 1961.
FISCHER, J.L., FISCHER, A., and MAHONY, F.: 'Totemism and Allergy',
The International Journal of Social Psychiatry, vol. 5, no. 1, 1959.
FLETCHER, A.C.: (1) 'A Pawnee Ritual used when changing a Man's
name', American Anthropologist, vol. 1, 1899.
(2) 'The Hako: A Pawnee Ceremony', 22nd Annual Report, Bureau of
FLETCHER, A. C. and LA FLESCHE, F.: 'The Omaha Tribe', 27th Annual
Report, Bureau of American Ethnology (1905-6), Washington, D.C.,
191 I.
FoRTl'NE, R.F.: ( 1) 'Omaha Secret Societies', Columbia University Con-
tributions to A11thropology, vol. 14, ~ew York, 1932.
(2) Sorcerers of Dobu, :-:cw York, 1932.
Fol'RIE, L.: 'Preliminary Notes on Certain Customs of the Hei-/om
Bushmen', Journal of the South1ast Africa Scientific Society, vol. 1,
Windhoek, S.W.A., 1927.
Fox, C.E.: The Threshold of the Pacifie, London, 1924.
Fox, R. B.: 'The Pinatubo Negritos: their useful plants and material
culture', The Philippine Journal of Science, vol. 81 ( 1952 ), nos. 3-4,
Manila, 1953.
FRAKE, Ch. O.: 'The Diagnosis of Disease among the Subanun of Min-
danao', American Anthropologist, vol. 63, no. 1, 1961.
FRAZER, J. G.: Totemism and Exogamy, 4 vol., London, 1910.
274
BIBLIOGRAPHY
FREE:\!AX, J. D.: '!ban Augury', Bijdragen toi de Taal-, Land-en Volken-

kunde, Deel l 17, 1st Afl., 'S-Gravenhage, l96r.
FREl'D, S.: Totem and Taboo (translatcd from the German by J. Strachcy),
London, 1950.
GARDINER, A. H.: The Theory of Proper Names. A Controversial E'ssay,
London, 2nd ed., 1954·
GEDDES, W. R.: The Land Dayaks of Sarmt·ak, Colonial Office, London,
195+·
GILGES, ,Y.: 'Sorne African Poison Plants and Mcdicines of Northern
Rhodcsia', Occasional Papers, Rhodes-Livingstone Museum, no. l l,
1955·
GüLDE:-1\VEISER, A.A.: 'On Iroquois '\'ork', Summary Report of the Geo-
logical Survey of Canada, Ottawa, Dcpartmcnt of J\'lines, 1913.
GRZI:\IEK, B.: 'The Last Great Hcrds of Africa', Natural History, Yol. 70,
no. l, Xew York, l96r.
HAILE, Father B.: Orzgin Legend of the Nm:aho Flintway, Chicago, l94J·
HAILE, Fathcr B. and \YIIEELWEIGllT, 2\1. C.: Emergence Myth according
to the J/anelthnayhe Upicard-Reaching Rite, Navajo Religion Series,
rnl. 3, Santa Fe, 1949·
HALLOWELL, A. I.: 'Ojibwa Ontology, Behavior and \Vorld Vicw', in:
S. Diamond, cd., Culture in Jlistory. Rssays in J/onor of Paul Radin,
.:\ew York, l 960.
HA:O.IPATÉ BA, A. and DIETERLEN, G.: 'Koumcn. Texte initiatique des
Pasteurs Peul', Cahiers de l'Jlomme, nouvelle série I, Paris-La Haye,
1961.
HANDY, E.S. Craighill and PUKL'I, M.Kawena: 'ThcPolyncsian Family
System in Ka-'u, Hawai'i', parts VI and VII; and VIII,Journal of the
Polynesian Society, vol. 62; and vol. 64, W cllington, ~.z., 1953 and
1955·
HARXEY, \\'.E.: 'Ritual and Bchaviour at Aycrs Rock', Oceania, vol. 3 l,
no. l, Sydney, 1960.
HARRINGTOX, J. P.: 'Mollusca among the American lndians', Acta
Americana, vol. 3, no. 4, 1945.
HART, C. ,Y. l\T.: 'Persona! Names among the Tiwi', Oceania, vol. l, no,
3, 1930.
HEDI GER, H.: Studies of the Psychology and Behaviour of Captive Animais
in Zoos and Circus (translated from German), London, 1955·
HENDERSON, J. and HARRINGTON, J. P.: 'Ethnozoology of the Tewa In-
dians', Bulletin no. 56, Bureau of American Ethnology, Washington,
D.C., 1914.
HENRY, J.: Jungle People. A Kaingdng Tribe of the Highlands of Brazil1
New York, l94L
275
THE SAVAGE MIND
I iER!'IANDEZ, Th.: 'Social Organization of the Drysdale River Tribes,

~orth-\Vest Australia,' Oceania, vol. Il, no. 3, 1940-1.
IIEYTI1'G, A.: Les Fondements des ,t/athématiques, Paris, 1955·
HoFFMAX, \V.J.: 'The Menomini Indians', I4th Annual Report. Bureau
of American Ethnology, part l ( 1892-3), Washington, D.C., 1896.
Hor.Lis, A. C.: The Nandi, their Language and Folklore, Oxford, 1909.
HUBERT, R. and Mauss, M.: (1) Mélanges d'histoire des religions, 2nd ed.,
1929.
(2) 'Esquisse d'une théorie générale de la magie', L'Année Sociolo-
gique, tome VII, 1902-3, in: Mauss, M. Sociologie et Anthropologie,
Paris, 1950.
IVENS, W. G.: Melanesians of the South-East Solomon Islands, London,
1927.
JAKOBSO:\, R.: 'Concluding Statement: Linguistics and Poctics', in:
Thomas A. Scbeok, ed. Style in Language, New York-London, 1960.
JAKOBSON, R. and HALLE, M.: Fundamentals of Language, 'S-Graven-
hage, 1956.
}E:\NESS, D.: (1) 'The lndian's Interpretation of Man and Nature', Pro-
ceedings and Transactions, Royal Society of Canada, 3rd series, vol.
XXIV, Section Il, 1930.
(2) 'The Ojibwa Indians of Parry Island. Their Social and Religious
Life', Bulletins of the Canada Department of 1l1ines, National Museum
of Canada, no. 78, Ottawa, 1935·
(3) 'The Carrier lndians of the Bulkley River', Bulletin no. IJJ,
Bureau of American Ethnology, Washington, D.C., 1943·
JENSEN, B.: 'Folkways of Greenland Dog-Keeping', Folk, vol. 3, Copen-
hagen, 1961.
K., \V.: 'How Foods Derive their Flavor', (reportofa communication by
E. C. Crocker to the Eastern New York Section of the American Chem-
ical Society). The New York Times, May 2, 1948.
KELLY, C. Tennant: 'Tribes on Cherburg Seulement, Queensland',
Oceania, vol. 5, no. 4. 1935·
KINIETZ, W. V.: 'Chippewa Village. The Story of Katikitegon', Bulletin
no. 25, Cranbrook lnstitute of Science, Detroit, 1947·
KoPPERS, W.: Die Bhil.in Zentralindien, \Vien, 1948.
KRAt:SE, A.: The Tlingit lndians, translated by E. Gunther, Seattle, 1956.
KRtGE, E. J. and J. D.: The Realm of a Rain Queen, Oxford, 1943·
KROEBER, A. L.: ( 1) 'Zufli Kin and Clan', Anthropological Papers of the
AmericanMuseumof Natural History, vol. 18, part 11, New York, 1917.
(2) 'Handbook of the Indians of California', Bulletin 78, Bureau of
American Ethnology, Washington, D.C., 1925.
KROTT, P.: '\Vaysof the \Volverine', Natural History, vol. 69, no. 2, New
York, 1960.
BIBLIOGRAPHY
LA BARRE, \V.: 'Potato Taxonomy amongtheAymara lndiansofBolivia',

Acta Amen.cana, vol. 5, )l°os. 1-2, 1947.
LA FLESCHE, F.: (1) 'Right and Left in Osage Ceremonies', Ho/mes Anni-
versary Volume, \Vashington, D.C., 1916.
(2) 'The Osage Tribe. Rites of the Chiefs: Sayings of the Ancient
Men', 36th Annual Report, Bureau of American Ethnology (1914-15),
Washington, D.C., 192i.
(3) 'The Osage Tribe. The RiteofVigil', 39th Annual Report, Bureau
of American Ethnology ( I9I7-I8), Washington, D.C., I925.
(4) 'The Osage Tribe. Child Naming Rite', 43rd Annual Report,
Bureau of American Ethnology ( 1925-6), Washington, D.C., 1928.
(5) 'The Osage Tribe. Rite of the Wa-Xo'-Be', 45th Annual Report,
Bureau of American Ethnology ( 1927-8), Washington, D.C., 1930.
LAGU.:\A, F. DE: 'Tlingit ldeas about the lndividual', Southwestern
Journal of Anthropology, vol. IO, no. 2, Albuquerque, 195+
LAROCK, V.: Essai sur la valeur sacrée et la valeur sociale des noms de
personnes dans les sociétés inférieures, Paris, 1932.
LÉVI-STRAUSS, C.: ( I) Les Structures élémentaires de la parenté, Paris, 1949.
(2) A World on the JVane, (translated from the French by John
Russell), London, 196i.
(3) 'Documeflts Tupi-Kawahib', in: Miscellanea Paul Rivet, Octo-
genario Dicata, l\Icxico, I958.
(4) Collège de France, chaire d'Anthropologie sociale. Leçon inaugurale
faite le mardi 5 janvier 1960. Paris, I 960.
(5) 'La Structure et la forme, réflexions sur un ouvrage de Vladimir
Propp', Cahiers de l'Institut de Science économique appliquée (Re-
cherches et dialogues philosophiques et économiques, 7), no. 99,
Paris, 1960.
(6) Totemism (translated from the French by R. Needham), Boston,
I963.
(7) 'The Bear and the Barber', Journal of the Royal Anthropological
Society, vol. 93, London, 1963.
LIENHARDT, G.: Divinity and Experience. The Religion of the Dinka,
London, 1961.
LOEB, E. M.: 'Kuanyama Ambo Magic', Journal of American Folklore,
vol. 69, 1956.
LONG, J.K.: Voyages and Travels of an lndian lnterpreter and Trader
(1791), Chicago, 1922.
MANU (The Laws of): The Sacred Books of the East, ed. by F.Max
Müller, vol. 25, Oxford, 1886.
McCLELLAN, C.: 'The Interrelations of Social Structure with Northcrn
Tlingit Ceremonialism', SouthwesternJournal of Anthropology, vol. IO,
no. 1, Albuquerque, 1954·
277
THE SAVACJE MIND
McCONNEL, U.: 'The Wik-Munkan Tribe of Cape York Peninsula',

Oceania, vol. 1, nos. 1 and 2, 1930-1.
MARSll, G. H. and LAUGllLIN, W. S.: 'Human Anatomical Knowledge
among the Alcutian Islanders', Southzcestern Journal of Anthropology,
vol. 12, no. 1, Albuquerque, 1956.
MAUSS, l\1.: (cf. also HUBERT and MAUSS, DURKHEIM and MAuss).
'L'âme et le prénom', Bulletin de la Société française de Philosophie,
Séance du 1er juin 1929 (29e année).
:\1EGGITT, M.J.: 'The Bindibu and Others', ~Man, vol. 61, art. no. 172,
London, 1961.
M1cHELSON, T.: (1) 'Notes on Fox Mortuary Customs and Bcliefs', 4oth
Annual Report, Bureau of American Ethnology ( 1918-19), Washington,
D.C., 1925.
(2) 'Fox :\1iscellany', Bulletin II4, Bureau of American Ethnology,
Washington, D.C., 1937.
MIDDLETON, J.: 'The Social Significance of Lugbara Persona! Names',
The UgandaJournal, vol. 25, no. 1, Kampala, 1961.
Moo:ŒY, J.: 'The Sacred Formulas of the Cherokee', 7th Annual Report,
Bureau of American Ethnology, Washington, D.C., 1886.
NEEDHA:>.1, R.: (1) 'The System ofTeknonyms and Death-Names of the
Penan', Southu·estern Journal of Anthropology, vol. 10, no. 4, Albu-
querque, 1954.
(2) 'A Penan Mourning-Usage', Bijdragen toi de Taal-, Land- en
Volkenkunde, Deel IIO, 3rd Aft., 'S-Gravenhage, 1954.
(3) 'The Left Hand of the Mugwe: An Analytical Note on the Struc-
ture of l\'leru Symbolism', Africa, vol. 30, no. 1, London, 1960.
(4) 'l\fourning Terms', Bzjdragen tot de Taal-, Land- en Volkenkunde,
Decl u5, Ist Afl., 'S-Gravenhage, 1959.
NELSO:\', E. W.: Wild Animais of North America, Washington, D.C.,
1918.
NsIMBI, N.B.: 'Baganda Traditional Persona) Names', The Uganda
Journal, vol. q., no. 2, Kampala, 1950.
PARSONS, E. C.: 'Hopi and Zui'ii Ceremonialism', Memoirs of the Ameri-
can Anthropological Association, no, 39, Menasha, 1933.
PASO Y TRoNcoso, F. del: 'La Botanica entre los N ahuas', A na/es Mus.
Nac. IHexic., tome III, Mexico, 1886.
PEIRCE, Ch. S.: 'Logic as Semiotic: the Theory of Signs', in: J. Buchler,
ed. The Philosophy of Peirce: Selected Writings, London,3rd ed., 1956.
PINK, O.: 'Spirit Ancestors in a Northern Aranda Horde Country',
Oceania, vol. 4, no. 2, Sydney, 1933-4.
RADCLIFFE-BROWN, A. R.: (1) 'The Social Organization of Australian
Tribes', Oceania, vol. 1, no. 2, 1930-1.
(2) 'The Comparative Method in Social Anthropology', Huxley
278
BIBLIOGRAPHY
l\Iemorial Lecture for 1951. Journal of the Royal Anthropological

lnstitute, vol. 81., parts 1 and II, 1951 (Published 1952). Republished
in: Jlethod in Soda/ Anthropology, Chicago, 1958, ch. v.
(3) 'Introduction' in: A. R. Radcliffe-Brown and Daryll Forde, eds.
African Systems of Kinship and Marn'age, Oxford, 1950.
RADI:-:, P.: ( 1) 'The Winnebago Tribe', 37th An nuai Report, Bureau of
A1.11.erican Ethnology (1915-16), Washington, D.C., 1923.
(2) '.:\Iexican Kinship Terms', University of California Publications in
AmericanArchaelogy and Ethnology, vol. 31, Berkeley, 1931.
R.\S'.\IUSSE:-:, K. : 'Intellectual Culture of the Copper Eskimos', Report of
the Fifth Thule E.,pedition, vol. 9, Copenhagen, 1932.
READ, K. E.: 'Leadership and Consensus in a New Guinea Society',
American Anthropologist, vol. 61, no. 3, 1959.
REICHARD, G.A.: (1) ':\avajo Classification of Natural Objects', Plateau,
vol. 2 l, Flagstaff, 1948.
(2) ,\/œvaho Religion. A Study of Symbolism, 2 vol. Bollingen Series
X\ïll, New York, 1950.
REKO, B. P.: 11litobotanica Zapoteca, Tacubaya, 1945·
RETz, Cardinal DE: llfémoires, Bibliothèque de la Pléiade, Paris, 1949·
RISLEY, H. H.: Tribes and Castes of Ben gal, 4 vol., Calcutta, 1891.
RnzE:o-;THALER, R.: 'Totemic Insult among the \Visconsin Chippewa',
American Anthropologist, vol. 47, l9f5·
RIVERS, \V. H. R.: 'Island-Names in Melancsia', The Geographical
Journal, London, l\lay, 1912.
RoBBIJ\"S, \Y.\V., HARRINGTON, J.P., and FREIRE-MARRECO, B.:'Ethno-
botanyof the Tewa Indians', Bulletin no. 55, Bureau ofAmerican Ethno-
logy, Washington, D.C., 1916.
RoCAL, G.: Le Vieux Périgord, 3rd ed., Paris, 1928.
ROLLAND, E.: (1) Faune populaire de la France, Tome Il, 'Les Oiseaux
sauvages', Paris, 1879. ·.
(2) Florepopulairedela France, Tome II, Paris, 1899.
RoscoE, J.: The Baganda: An Account of their Native Customs and
Beliefs, London, l 911.
RoussEAU, J.J.: (1) Discours sur l'origine et les fondements de l'inégalité
parmi les hommes, Œu vresmelées, Tome II, ~ouvellc ed., London, 1776.
(2) Essai sur l'origine des langues, Œuvres posthumes, Tome Il,
London, 1783.
RUSSELL, B.: 'The Philosophy of Logical Atomism', The Monist, 1918.
RussELL, F.: 'The Pima Indians', 26th Annual Report, Bureau of
SARTRE, J.P.: Critique de la raison dialectique, Paris, 1960.
SAUSSURE, F. DE: Course in General Linguistics, (translated from the
French by W.Baskin), London, 1960.
279
THE SAVAGE MIND
Sc1100LCHAFT, H. R.: Cf. WILLIA:\1S.

SEDEIS (Société d'l~tudes et de Documentation l~conomiqucs, Industri-
elles et Sociales): Bulletin no. 796, supplcment 'Futuribles', no. 2,
Paris, 1961.
SHARP, R. Lauriston: 'Notes on ::-.!ortheast Australian Totemism', in:
Studies in the Anthropology of Oceania and Asia, Papers of the Peabody
JHuseum, Harvard University, vol. 20, Cambridge, Mass., 1943.
SEBILLOT, P.: Le Folklore de France, Tome III, 'La Faune et la flore',
Paris, 1906.
S1MPSO:>:, G. G.: Princip/es of Animal Ta.wnomy, New York, 1961.
SKINNER, A.: ( 1) 'Social Life and Ccremonial Bundles of the Menomini
lndians', Anthropological Papers of the American Museum of Natural
l!istory, vol. 13, part 1, New York, 1913.
(2) 'Observations on the Ethnology of the Sauk Indians', Bulletins of
the Public kluseum of the City of Milwaukee, \'OI. 5, no. 1, 1923-5.
(3) 'Ethnology of the Iowa y Indians', Bulletins of the Public Museum
of the City of I'.lilwaukee, vol. 5, no. 4, 1926.
S:'>!ITII, A. H.: 'The Culture of Kabira, Southern Ryûkyü Islands', Pro-
ceedings of the American Philosophical Society, vol. 104, no. 2, Phila-
delphia, 1960.
SMITH BowE:>:, E.: Return to Laughter, London, 1954.
SrECK, F.G.: (1) 'Reptile Lore of the Northern lndians', Journal of
American Folklore, vol. 36, no. 141, Boston-New York, 1923.
(2) 'Penobscot Tales and Religious Belicfs' ,Journal of American Folk-
lore, vol. +8, no. 187, Boston-New York, 1935.
SPENCER, B. and G1LLEN, F. J.: The Northern Tribes of Central Australia,
London, 1904.
STA:->:->ER, W.E.H.: (1) 'Durmugam, A ::-.!angiomcri (Australia)', in:
J. B. Casagrandc, cd. ln the Company of Alan, Xcw York, 1960.
(2) 'On Aboriginal Religion, IV, The Design-Plan of a Ritelcss Myth',
Oceania, vol. 31, no. 4, 1961.
STEPHEN, A.:\1.: 'Hopi Journal', cd. by E.C.Parsons, 2 vol., Columbia
Unh:ersity Contributions ta Anthropology, vol. 23, New York, 1936.
STREllLOW, C.: Die Aranda und Loritja-Stiimme in Zentral Australien,
+vol., Frankfort am l\Iain, 1907-13.
STREJILOW, T. G. H.: Aranda Traditions, Melbourne, 1947.
STURTEVANT, \\'. C.: 'A Semi noie Medicine Maker', in: J. B. Casagrande,
ed. ln the Company of Man, New York, 1960.
SWANTON, J. R.: (1) 'Social Organization and Social Usages of the
Indians of the Creek Confederacy', .pnd Annual Report, Bureau of
(2) 'Social and Religious Beliefs and Usages of the Chickasaw Indians',
44th Annua/ Report, Bureau of American Ethnology ( 1926-7), Wash-
ington, 1928.
280
BIBLIOGRAPHY
TESS~IA:\:\, G.: Die Pang11:e, Vblkerkundliche Jl1onograplu'e eines u·est-

afrikanischen Negerstammes, 2 vol., Berlin, 1913.
TttœtAS, ~.\Y.: Kinship Organisations and Croup 1Uarriage in Australia,
Cambridge, 1906.
Tttœrsox, D. F.: ·~ ames and Naming in the \Vik Monkan Tribes',
Journal of the Royal Anthropological lnstitute, vol. 76, part II, London,
19+6.
THt.:RCÇWALD, R.: 'Banaro Society. Social Organization and Kinship
System of a Tribe in the lnterior of 1'\ew Guinea', Jllenwirs of the
American Anthropological Association, rnl. 3, no.+, 1916.
THt.:RSTO:::->, E.: Castes and Tribes of Southern lndia, 7 vol., l\ladras,
1909.
TozZER, A. ::\1.: 'A Comparative Study of the Mayas and the Lacan-
dones', Archaeological Institute of America, Report of the fellcw in
American Archaeology (1902-5), New York, 1907.
Tl'R:\"ER, G.: Samoa a Hundred Years agoand Long Before .. .. London,
I88+.
Tl"RCÇER, Y.\\'.: ( I) 'Lunda Rites and Ceremonies', Occaswnal Papers,
Rhodes-Livingstone ]i/useum, no. IO, Manchester, I953·
(2) 'Ndembu Divination. Its Symbolism and Techniques', The
Rhodes-Li·vingstone Papers, no. 31, l\fanchester, I96I.
TYLOR, E.B.: Primitive Culture, 2 vol., London, 1871.
VA::-> GEN:::->EP, A.: L'Etat actuel du problème totémique, Paris, 1920.
VAN GuLIK, R. H.: ( I) Erotic Colour Prints of the I\ling Period, 3 ml.,
Tokyo, 195 r.
(2) Sexual Lije in Ancient China, Lciden, I96I.
VANZOLI:::->I, P. E.: 'Notas sôbre a zoologia dos lndios Canela', Revista do
I\luseu Paulista, N .S., voj. IO, Sao Paulo, I956-8.
VE:-;DRYES, J.: Le Langage.. Introduction linguistique à l'histoire, Paris,
1921.
V ESTAL, P. A.: 'Ethnobotany of the Ramah Navaho', Papers of the Pea-
body 1l1useum, Harvard University, vol. +o, no. 4, Cambridge, Mass.,
I952·
VoGT, E. Z.: 'On the Concepts of Structure and Process in Cultural
Anthropology', American Anthropologist, vol. 62, no. I, 1960.
V üTH, H. R.: ( 1) 'The Oraibi Soya! Ceremony', Field Columbian illuseum,
Publication 55. Anthropological Series, vol. 3, no. 1, Chicago, 190 r.
(2) 'The Oraibi Powamu Ceremony', Field Columbian l'.1useum,
Anthropological Seri es, vol. 3, no. 2, Chicago, 190 r.
(3) 'Hopi Proper Names', Field Columbian A1useum, Publication IOO,
Anthropological Series, vol. 6, no. 3, Chicago, I905.
(4) 'The Traditions of the Hopi', Field Columbian Museum Publica-
tion 96, Anthropological Series, vol. 8, Chicago, 1905.
281
THE SAVAGE MIND
(5) 'Brief Miscellaneous Hopi Papers', Field Museum of Natural
History, Publication r57, Anthropological Series, vol. XI, no. 2,
Chicago, 1912.
\VALKER, A. Raponda, and SILLANS, R.: Les Plantes utiles du Gabon,
Paris, 1961.
WALLIS, W.D.: 'The Canadian Dakota', Anthropological Papers of the
American 1\!Juseum of Natural History, vol. 41, part 1, New York, 1947.
\VARNER, \V.Lloyd: A Black Civilization, revised edition, New York,
1958.
WATERMAN, T. T.: 'Yurok Geography', University of CaliforniaPublica-
tions in American Archaeology and Ethnology, vol. 16, no. 5, Berkeley,
1920.
WHITE, C.M.N.: (1) 'Elements in Luvale Beliefs and Rituals', The
Rhodes-Livingstone Papers, no. 32, Manchester, 1961.
(2) (J.CHINJAVATA and L.E.MuKWATo.) 'Comparative Aspects of
Luvale Puberty Ritual', African Studies, Johannesburg, 1958.
\VmTE, L.A.: 'New Material from Acoma', in: Bulletin r36, Bureau of
American Ethnology, Washington, D.C., 1943.
WmTING, A. F.: 'Ethnobotany of the Hopi', Bulletin no. r4, Museum of
N orthern Arizona, Flagstaff, 19 50.
WILLIAMS, M. L. W.: Schoolcraft's lndian Legends, Michigan U.P., 1956.
\V IL50N, G. L.: 'Hidatsa Eagle Trapping', Anthropological Papers of the
American Museum of Natural History, vol. 30, part IV, New York,
1928.
WIRz, P.: Die Marind-Anim von Hollandisch-Süd-Neu-Guinea, 1 Band,
Teil Il, 1922.
WITKOWSKI, G. J.: Histoire des accouchements chez tous les peuples, Paris,
1887.
WoENSDREGT, J.: 'Mythen en Sagen der Berg-Toradja's van Midden-
Celebes', Verhandelingen van het Bataviaasch Genootschap van Kunsten
en Wetenschappen, vol. 65, no. 3, Batavia, 1925.
WoRSLEY, P.: 'Totemism in a Changing Society', American Anthro-
pologist, vol. 57, no. 4, 1955.
WY:0.1AN, L.C. and HARRIS, S. K.: 'Navaho Ethnobotany', University of
New Mexico, Bulletin no. 366, Anthropological Series, vol. 3, no. 4,
Albuquerque, 1941.
ZAHAN, D.: Sociétés d'initiation Bambara, Paris-La Haye, 1960.
ZEGWAARD, G. A.: 'Headhunting Practices of the Asmat of Netherlands
New Guinea', American Anthropologist, vol. 61, no. 6, 1959.
ZELENINE, D.: Le culte des idoles en Sibérie, French translation, Paris,
1952.
INDEX
Abstraction, l-5, 107, 135-60, 164, Ashanti, 133 n.

166, 175. 182, 217-20, 257, 266, Asmat, 60
269 Astrology, 42
Affectivencss, 37-<J Athapaskan, 50
Agriculture, 52, 73-4, l l0-3 An:ER, P., 263
ALBERTVS l\IAG:-;vs, 42 Aurora, 76-80
ALBISETTI, P.C. See CoLBACCHINI Aymara, 43
Algonkin, 31-2, 50, 57, 70, 99, 117, Azandc, l l
142, 166-7, 173, 187, 189, 196
Allcrgy, 98
Aluri<lja, 165, 167 BACH, J.S., 242
AL\'IANO, F. DE, 106 Baganda, 1l2-3, 178
Ambo, 62 Bahima, 150
Analytical rcason, 245-69 BALANDIER, G., 234-5 n.
A:-;DERSO:>;, A.J.O., 46 n. Balovale, 6
AXDERSOX, E., 73 BALZAC, H. DE, l 30
Animais, names given to, 204-8, Bambara, 163
213 Hanyoro, 150, 179-80
Animism, 249 Bard, 6 5, l 84
AxTHOXY, H.G., 53 n. BARROWS, D.P., 5
Arabanna, Arabana, 81-2, 84-7, Bateso, 151
238 BATESOX, G., 174, 177
Aranda, 81-8, 102-3, 109, 112-3, BEATTIE, J.H.M., 179-80
n8, 150, 165, 167-8, 170, 174, BECKWITH, 1\1.W., 50
218, 235,237-41, 243-4 BEIDELMAN, T.O., 144
Arbitrary. See Motivation Bemba, 62
Archetypes, 65 IlE:>;EDICT, P. K., 189
Archives, 239-44 BERGSON, H., 137
AROK, R., 70 BETH, E. W., 248 n.
Art, 22-36, 156, 182 Bhil, l 20-1
Artemisia, 46-8 Bir<ls, 54-61, 96, 147, 149, 189
Arunta. See Aranda Bir<ls, names given to, 203, 213
THE SAVAGE MIND
Birth, 46-8, 61, 76-81, 131, 168, Comprehension, 20, 137, 15 3,

177, 184, 189-98, 264 264-5
Blackfoot, 10 Computer, 89, 151
BOAS, P., I, 21, 135-6, 189 COMTE, A., 163-4, 218-20
BOCHET, G., 154 Concept, conceptual, 1-5, 18-21,
Bororo (Africa), 145 90, 107, 117, 119, 129-31,
Bororo (Brazil), 39 n., 99, 114, 173, 135-60, 223-4, 26..:-5 and passim
178, 230 CoNKLIN H. C., 3, 7-8, 39 n., 55,
Bot:LAINYILLIERS (Boulainviller), 62, 138-9, 141, 153
Comte H. DE, 261-2 n. Continuous and discontinuous,
Bouriate, 8-9 137-9, 172, 195-200, 215-6,
BoWERS, A. \V., 50 224-8, 232-4, 256-64, 268
Brassica rapa, 201-3, 211-2 CooKE, CH.A., 178 n.
Bricolaf(e, 16-36, 150 n. Couvade, 195-6, 201
BRiiNDAL, V., 185 n., 201 n. Creek, 59-60, 118
IlROUILLETTE, B., 50 Crow, 112
BURCKHARDT, J., 244 CRUZ, M., 173
Bushmen, 103-4 CUNNISON, 1. G., 62
CUSHING, F. H., 42
CALAME-GRIAULE, G., 163
Canela, 40 n. Dakota, 189
Cannibalism, 78, 105, 257-8 n. Dates, 258-62
CAPELL, A., 57 Dayak, 54, 56, 205 n.
Carcajou (Gulo luscus), 50-3 DEACON, B., 251
CARPENTER, E., 64 Death, 31-2, 65, 76-81, 84, 144,
Caste, 113-33, 162 158, 177, 184-5, 189, 191-200,
Chance, 14, 184-5, 187, 220 210, 236-44, 264
CHARBOKNIER, G., 234 DELATTE, A., 42
Cherburg, 180-1 n. Delaware, 142
CHEVAL (the postman), 17 Demography, 66-74, 111-2, 155,
Chickasaw, 117, 119, 126, 164, 204 187, 209-10, 232, 234, 260
Chinese, 65 n., 106, 165 n. DENNETT, R. E., 133
Chinook, I DENNLER, J.G., 44 n.
Chippewa, 102, 116, 167 DENSMORE, F., 50, 165 n.
Churinf(a, 87-8, 237-44 DESCARTES, R., 249
Classification, 39-74, 135-221, DETAILLE, E., 28
229-44, 259-62, 267-8 Determinism, geographical, 94-6
CLOUET, F., 22, 24, 26-7 Devanga, 120
Coahuila, 5 Diachrony, 52-3, 63, 66-74, 87-8,
Code, 90-8, 106, 111, 119, 137-40, 148-9, 155, 166, 197, 227 n.,
149-50, 172, 222, 228, 258-62 231-44, 256, 263
CooRINGTON, R. H., 76, 78 Dialectical reason, 245-69
CoGHLAN, H. H., 14 n. DIAMOND, S., 145 n.
Cogito, 250 DIBBLE, CH. E., 46 n.
Coifs, 90 DICKENS, CH., 17, 150
CoLBACCHINI, P.A., 39 n., 114, 230 DIETERLEN, G., 39, 43, 104, 163,
Colla, 43 166, 189
Colours, 40, 55, 64-5 Dinka, 121, 128-9 n.
INDEX
Diseases, 139, 164-5 FREUD, S., 253

Dobu, 110-1, 177 Fronde, 254-5
Dogon, 39, 104 n., 163, 189 Fulani, 39 n., 43, I45
Dogs, names given to, I67, 18I-2,
204-8, 2I3 Gahuku-Gama, 30
DORSEY, C.A., 128 GALEN, 42
ÜORSEY, J.O., 69, I48-9 Garnes, 30-2, 126
Drysdale river, 168, I74, I76 GARDl!';ER, A.H., 172, 185, 20I-3,
DuPIRE, :\1., 145 2II,2I5n.
DURKHEIM, E., 39, 40, 57, 76, 91, GEODES, W.R., 205
102, I62, 2I4, 238-9, 241, 268 GILGES, W., 6
GILLEN, F.J., 8I-2, 84, 86, 88, I 14,
Eagle hunting, 46-5I, 225 238, 240-I
EGGAN, F., 7I GOBINEAU, Comte J.A.DE, 26I-2 n.
Elema, I06-7 GÜDEL, K., 262
ELKIN, A.P., 65, 8I-2, 84, IOO, I02, GOETHE, \V., 244
185, 218, 242, 265 GOGH, V. VAN, 244
EL!\1ENDORF, \V. \V., I77n., I96-7 n. 'Golden Bough', 48
EL:\IORE, F. H., 40 Golden Rod (Solidago ), 46-8
Endogamy, I 1 I-33 GOLDENWEISER, A.A., I78, I88, I96
Eskimo, 40, 50, 64 Grammar. See Vocabulary
EVAKS-PRITCHARD, E. E., II, 56, GREUZE, J. -B., 30
60 n., I04, I61 n., I 89, 224 GRIAULE, M., 5, 39, I63, 23I
Exogamy, 77-I33, 23I Groote Eylandt, I55
Experimenting, I3-I5, 73-4, 220-3 GRZI:\IEK, B., I 37 n.
Extension, 20, I36, I53 Guarani, 44
Gu1LBAUD, G. TH., 82 n.
Fang, 3, 6I, 98-<J, 103-4
FIRTH, R., 228 n. HADDON, A.C., II5-6
FISCHER, A., 98 HAILE, Fa. B., 40, 128, 203
FISCHER, J. L., 98 HALLE, M. See JAKOBSON
FLETCHER, A. c., IO, 56, 112 HALLOWELL, A. I., 96
Flowers, names given to, 211-6 HAMPATÉ BA, A., 39 n., 43
Forrest river, 65 HANDY, E.S.CRAIGHILL, 2, 3, 37,
FORTl:NE, R.F., 43, 61, 1I1 96, I44
FOURIE, L., I03 Hanunoo, 3-5, 7, 55, 62, I38, I53
Fox, C.E., 78 n. HARNEY, W.E., 166
Fox, R.B., 5, 8, I4 HARRINGTON, J. p., 6, 7, I 69
Fox tribe, 3I, 65, 99, 199 n. HARRIS, S. K.' 40, 204
FRAKE, CH. o., 5, I 39 HART, C.W.M., I77, 190, 195,
France, 70, 90, Io5, III, I93-4, I99-200, 2IO-I
254-5, 258, 26I-2 n. Hawaii, 3, 37, 96, I44
FRAZER, J.G., 57-8, 74, 76, 78-<), REDIGER, H., 38
81, 102, I06, 116, I40, I42, 23I HEIJ\I, R., I86
FREEMAN, J.O., 55 Helmene, 8
FREIRE-MARRECO, B., 6 HENDERSON, J., 7
French Revolution, 254-5, 258, HENRY, J., I05 n.
26I-2 n. HERMES TRISMEGISTUS, 42
285
TIIE SA\'AGE l\II;-.;D
IIER:'\AN!JEZ, TH., 155, 168, 174, KRIGE, J. D., 159 n.

176 KROEBER, A.L., 71, 128, 175-7,
llEYTI1'G, A., 248 n. 196-7 n.
1 lidasta, 46, 48, 52-3, 225 Kuruha, 121
llistory, 66-74, 155-60, 186-7, Kwakiutl, 189
231-44, 248-9, 250-1, 252-63
lloFFl\IA:'-.', \V.J., 57, 167, 229 LA BARRE, W., 44
Bor.us, A. C., 1 1 2 Lacandon, 187 n.
Hopi, 5, 40, 48, 60-1, 71-2, 165, LA FLESCHE, F., 56, 59, 112, 142-3,
176, 178, 180, 229 144-8, 170-1, 173
1[orses, names given to, 206-8, 213 LA FoRCE, Duc DE, 244
llOl'IS,:\J., 209 n. LAGCXA, F.DE, 60, 173
Ht:BERT, R., 11, 18 n. Laierdila, 100-1
llt:GO, V., 244 Language, 1-3, 56-7, 66-7, 105-6,
129, 131' 155-60, 172-216, 228,
Iakoute, 8, 9 252, 264-5, 267
Iatmul, 174 LAROCK, V., 209 n.
Ihan, 54, 56 LACGHLIN, \V. S., 153
Information thcory, 154, 268-9 LEIGHTON, A. H., 62, 204
Infrastructures, 90-6, 130-1, 213-4 LEIGHTON, D. c., 62, 204
lntichiuma, 226-8 LEXOTRE, G., 244
Ioway, 167 LEVY-IlRCHL, L., 251. 268
Iroquois, 57, 70, 114, 177-8, 188, LIENHARDT, G., 121, 128-9 n.
196-8, 223 Lifu, 78-80
IVE:'-.'S, \V.G., 78 n. Linguistics, 66-7, 102, 155-60,
172-222, 231, 252
}AKOBSOX, R., 149 LITTRÉ, E., 205
}ENXESS,0., 37, 43, 60, 227, 239 n. LüEB, E. '.\1., 9
JocvExEL, B. DE, 70 Logic, logical thought, 35-76,
102-3, 105, 126, 129, 135-216,
K. (W.), 12 261 , 267 and passim
Kaguru, 145 LONG, J.K., 163
Kaitish, 81, 86-7, 113 Loritja. See Aluridja
Kalar, 9 Lovedu, 159 n.
Kaleidoscope, 36 Luapula, 61 -3
Karadjera, 88 Luchozi, 6
Karuba, 121 Lugbara, 179-80
Kauralaig, 88, 100-1 Luvale, 64-5, 97
Kavirondo, 151
Kazak, 9 '.\1abuiag, 1 15
KELLY, C. TEXXAXT, 158, 180-1 n. '.\1cCLELLA;-;', C., 99
Keres, 71, 227 n. '.\1cCON:'i'EL, U., 174
KIXIETZ, \V. V., 116, 167 :\1cLENNAN, J.F., I 18, 231
Kiwai, 140 Magic and religion, 220-8
Koko Yao, 100-1, 105 i\IAHONY, F. See FISCHER
KoPPERS, \V., 120 i\1aithakudi, 100-1
KRAcSE, A., 1 l\faize, 61, 73, 112
KRIGE, E. J., 159 n. Malaita, 78-80
286
INDEX
Malecite, 8 Motlav, 78-80

l\IALI:-OOWSKI, B., 3, 74, 91, 250 Munda, 120
l\Iandan, 52 l\lurngin, 91-3, 96, 174
l\1ANNHARDT, \\'., 95 Muskogi, 117
::\IA::-.;t:, Laws of, 106, 126 Myths, mythical thought, 16-22,
l\Ianufactured articles, 121-9, 25-6, 30-3, 51-2, 68-70, 90-3,
143-4, 151, 174-5 127-8, 131-6, 148-50, 165-6,
l\Iarriage rules, 81-4, 88, 103-6, 168-9, 218, 223, 228-44, 254-5
109-33, 155, 243, 264-6 and passim
:\IARSH, G.H., 153
Naga, 73
:\lARX, K., 130, 246, 253
Nandi, 114
l\1ashona, 105
Namcs, propcr, 112-3, 149-50,
l\1atabele, 105
167-8, 172-216, 218
Mathcmatics, 248 n. Naskapi, 8
:V1Auss, :\1., 11, 18 n., 39, 40, 57, Natchez, 120
223, 268 Navaho, 5, 39,40,45-6,48,55,62,
Maya, 187 n., 189 128, 169, 203
:\1EGGITT, :\1.J., 222 Ndcmbu, 97
MELIES, G., 17 NEEDHAM, R., 144, 192-3, 195,
l\Ienomini, 57, 116, 167, 188, 229 198 n.
l\1enstruation, 46-8, 51-2, 91, 144, Negrito, 4
225 NELSON, E. \V., 53 n.
Mentality, primitive, 37-8, 251, Neolithic, 13-15, 172, 269
267-8 Ngarinyin. See Ungarinyin
l\letaphor, 51-2, 105-6, 150 n., NsrMBI, N.B., 179
204-8, 212-3, 224-8, 239 Nuer, 56, 60 n., 63, 104, 189, 224
Meteorology, 91-7 Numerology, 142-7
l\1etonymy, 52, 106, 150 n., 204-8,
212-3, 224-8 Observation, 3-6, 219-23
NIICHELSON, T., 31-2, 65, 100, Oïrote, 9
199 n. Ojibwa, 37, 60, 96, 142
Micmac, 8 Okerkila, 100-1
MIDDLETON, J., 179-80 Omaha, 42, 59, 61, 112, 140, 170-1,
MILL, J.S., 172 178, 21-6
Mind, savage, 219-20, 222-3, 245, Omens, 54-6, 131-3
249 n., 263, 267-9 Oraon, 120
Miwok, 175, 178, 181 Ordinal names, 182, 187, 189-97
Mixe, 189 Organicism, 169
Mocassin, 143 Organism, 103-4, 136, 148, 151-4,
Models, small-scale, 23-5, 150 n. 169-73, 175-6
Mohawk, 178 Osage, 56, 59, 61, 69-70, 140,
Mohican, 142 142-3, 145, 147, 1+9, 170-1, 173,
Montagnais, 8 178
MORGAN, L., 1 18 Ossete, 8
Morphology, 7-<) Painting, 22-30
Mota, 76, 79-80, 177 Papago, 165 n.
Motivation, 154-60 Papuans, 257-8 n.
THE SAVAGE l\lIND
Paradigmatic. See Syntagmatic RIVERS, W. H. R., 76, 78, 116

Parents-in-law, taboo of, 265-6 RoRB1:-.:s, \V. \V., 6
PARSO:\S, E. C., 230 RoRERTSox SMITH, \V., 23 1
Particularization, 168-216 RocAL, G., 111
PASO Y TRoNcoso, F.DEL, 46 n. RoLLAXD, E., 201 n., 212 n.
Pawnee, 10, 52, 56, 139 Roman de Renart, 205 n.
Peasant societies, 90, 111 RoscoE, ]., 113
PEIRCE, Cu.S., 20, 215 RoussEAt:, ].-]., 38, 163, 247
Penan, 191, 194-8 RUSSELL, B., 215
Penobscot, 8, 166 RUSSELL, F., 165
Personality, 214-5 Russians (Altai), 9
Pima, 165 Russians (Siberia), 8
Pinatubo, 4, 8 Russians (Sourgout), 8
P1:-.:K, 0., 174 Ryukyu islands, 5
PLI:-.:Y, 42
Poetry, 148 Sacrifice, 223-8
Ponapy, 98, 105 Saibai, 115
Ponca, 142 St John Chrysostom, 105
Praxis, 118, 122, 129-30, 251-2, Samoa, 105
263-5 SARTRE, J. -P., 130, 245-6, 248-56,
PRIOL'RET, R., 70 257-8 n., 264
Prohibited food, 61-2, 76-80, 88, Sauk, 170, 173, 188
97-133, 177, 205, 226-8 SAUSSURE, F. DE, 18, 1,49. 156,
Prohibitions, linguistic, 176-7, 15 8--ç
183-4, 194-200, 209-10, 264-5 ScHOOLCRAFT, R. H. See \VILLIAMS,
Psychoanalysis, 252-4 :\-1.L. \V.
Pueblo, 40, 52-3, 71, 169, 227 n. SEBILLOT, P., 204 n.
Pt1KVI, :\LKAWE:\'A, 2, 3, 37, 96, SEDEIS (Société d'études et de
144 dornmentation économiques, indus-
Pygmies, 4 trielles et sociales), 70
Seminole, 5, 182
RADCLIFFE-BRow:-.:,A. R.,55, 112n., Sense, 172, 210, 249, 253-4, 255
114, 167 Senufo, 154
RADIN, P., 57 n., 170, 189 SHARP, R.LAURISTOl', 88, 100-1,
RAS'.\tl'SSE:\', K., 40 151 n., 167, 236
READ, K. E., 3 I Sherente, 53
REICHARD, G.A., 40, 45, 55 SIEGFRIED, A., 70
REKO, B.P., 46 n. Sign, Signification, 18-20, 36,
Religion. See ::\Iagic and religion 130-3, 140, 156-60, 172-3,
RDIBRA:\DT, 28 181-3, 185-6, 203-15, 219-23,
RETZ, Cardinal DE, 254-5 239, 250, 257, 259, 266--ç
Right and left, 144-5 Silence, 65
RISLEY, H.H., 120 SILLANS, R., 5
Rites, ritual, 30-3, 42-3, 45-53, S1MPSOX, G.G., 12, 13, 63, 156
75-133, 143-7, 151, 154, 209-11, Sioux, 57 n., 59, 140, 167, 173, 176,
220-8, 231, 235-44, 264-5 and 189
passim SKINNER, A., 167, 173, 188, 229
RITZENTHALER, R., 102 Smells, 12-3
288
INDEX
S:\llTH. A. H., 5 Tlingit, 26, 60, 99, 173

S:\!ITH IlOWE'.'I, E., 6 Toradja, 169
Solomon Islands, 140 Toreya, 63
Species, notions of, 136-72, 17 5, Totalization, detotalization, 146-9,
177. 185-7, 200-16, 224-8 151-4, 169, 175-6, 178, 250-62
SPECK, F. G., 8, 166 Totem, totemism, passim
SPENCER, B., 81-2, 84, 86, 88, l 14, TozzER, A.l\II., 187 n., 189
238, 240-1 Tree, 158-9
SPE:-.:GLER, o., 2+4 TREVOUX (dictionary), 201 n.
STA);NER, \V. E. H., 94, 228 Trobriand islands, 140
Statistics (distribution), 72, 81-2, Tupi Kawahib, 173
85-6, 112, 159-60, 186-7, 232 TURNER, G., 233
STEPHEX, A. l\1., 60 TUR'.';ER, V. W., 97
STREHLOW, C., 167-8, 239 Tutu, 115
STREHLOW, T. G. H., 89, 229, 235, Twins, l 80, 189-90
239, 241, 243 TYLOR, E.Il., 153-4, 164
STt:RTE\"AXT, \V.C., 5, 183
Subanum, 5, 139 Ulawa, 78-80, 177
Superstructures, l l 7, 130, 2 13-4, Unconscious, collective, 65
254 Ungarinyin, 56, 184
Surrealists, 21 U niversalization, 164-7
SWA);TON, J.R., 60, Il9, 165 Unmatjera, 86-7, 113
Synchrony,52-3,63,66-74, 148-9,
155, 166, 197, 227 n., 231-44, Vagina dentata, 106
254, 256, 263 VAN GENNEP, A., 109, 162-3
Syntagmatic, paradigmatic, l 50 n., VAN GuLIK, R.H., 106, 165 n.
203, 206-8, 211, 215-6 VANZOLINI, P. E., 40 n.
VENDRYES, J., 203
Tanoan, 71 VESTAL, P.A., 40, 48
Tastes, 12-3 Vocabulary, 138-9, 143, 154-60,
Taxonomy, 9-13, l 5-6, 38-45, 231-2
137-9, 153-4, 156, 186-9, 203, VOGT, E.Z., 234
231 VoTH, H.R., 48, 61, 165, 176, 230
Teleology, 252
TERTULLIAN, 105 Wakelbura, 102
TESSMANN, G., 3, 61 n., 98-9, 104 WALKER, A. RAPONDA, 5
Tewa, 5-7 WALLIS, W.D., 190
THOMAS, N.W., 113 Walpari, 114
THO:\ISO!\T D.F., 45. 167, 183-4, WARNER, W.LLOYD, 91-3, 96, 174
188 Warramunga, 81, 85-7, l 14, 184,
Thunder Birds, 96, 229 238
THUR!'."WALD, R., 200 Water-colours (Aranda), 89, 244
THURSTON, E., 63 n., 121 WATERMAN, T.T., 168, 196-7 n.
Tikopia, l 39, 228 n. Wawilak sisters, 91-3
Tikuna, 106-7 WEYDEN, R. VAN DER, 28
Tiwi, 177, 190, 195, 199-200, 206, WHEELWRIGHT, 1\1.C., 128, 203
209, 2l1, 21 2-3 n. WHITE, C.M.N., 65, 98
Tjongandji, 100-1 WHITE, L.A., 227 n.
289
THE SAVAGE MIND
Wik Munkan, 45, 167 n., 174, 183, WYMAN, L.C., 40, 204
186-8, 212
WILLIAMS, M.L.W., 56 Yathaikeno, 100-1
WILSON, G.L., 48, 50 Yoruba, 105, 131-3
Winnebago, 57, 140, 142, 167, 170 Yuma, 180
WIRz, P., 63 n. Yurlunggur (snake), 91-3
WITKOWSKI, G.J., 201 Yurok, 168, 177, 189, 196-7
W OENSDREGT, J., 169
Wolverine, 50-3 ZAHAN, D., 39, 163
Wormwood. See Artemisia ZEGWAARD, G.A., 61
WORSLEY, P., 156 ZELENINE, D., 9
Wotjobaluk, 57, 105 Zuni, 40, 71-2

The Savage Mind PDF

Uploaded by

Copyright:

Available Formats

The Savage Mind 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

The Savage Mind PDF

Uploaded by

Copyright:

Available Formats

$5.

THE SAVAGE MIND

Editors: Julian Pitt-Rivers and Ernest Gellner

THE UNIVERSITY OF CHICAGO PRESS

Translated from the French, La Pensée sauvage

Library of Congress Catalog Card Number: 66-28197

Printed in the United States of America

THE SCIENCE OF THE CONCRETE 1

2 THE LOGIC OF TOTEMIC CLASSIFICATIONS 35

5 CATEGORIES, ELEMENTS, SPECIES, NUMBERS 1 35

7 THE INDIVIDUAL AS A SPECIES

9 HISTORY AND DIALECTIC

(Betu:een pages I 48 and 149)

Francois Clouet. Portrait of Elizabeth of Austria. (Photo:

This book is complete in itself, but the problems it discusses are

If I have felt obliged to give expression to my disagreement with

THE SCIENCE OF THE CONCRETE

native gives names and forms concepts solely in accordance with

with an interest that is more alert to possible distinctions which

Tessman (Vol. 2, p. 192) mentions 'the accuracy with which (the

Among the Hanunoo of the Philippines a custom as simple as that

Almost ail Hanunôo activities require an intimate familiarity with local

E. Smith Bowen scarcely exaggerates in the amusing description

The reaction of a specialist is quite different. In a monograph in

identification of plants and for 'discussing the hundreds of charac-

And nevertheless, as Speck has shown, the north-eastern Indians

woodpecker's beak, blood of a woodpecker, nasal insuffiation of

It may be objected that science of this kind can scarcely be of much

taxonomy ... Nevertheless, the most basic postulate of science is that

The thought we cal! primitive is founded on this demand for or der.

principle of Causality' as Hubert and lVIauss called it (2, p. 61), car

incidentally, occurred repeatedly. As Simpson (pp. 84-5) has

by the law of chance. Simpson advances this interpretation in the\

1 am not ho\\'ever commending a return to the popular belief

fortuitous accumulation of a series of chance discoveries or believe

To transform a weed into a cultivated plant, a wild beast into a

could have yielded practical and immediately useful results. There

also corresponded to objective reality. It seems probable, for

There still exists among ourselves an activity which on the technical

compared to those of a craftsman.* The characteristic feature of

• The 'bricoleur' has no precise equivalent in English. He is a man who under-

to show to advantage. In one case it will serve as extension, in the

saying that the engineer works hy means of concepts and the

those of the 'bricoleur', always really consist of a new arrangement

remains anddebrisof events :in French 'des bribesetdes morceaux',

~ problem of art has been touched on several times in the fore-

knowledge of the whole precedes knowledge of the parts. And

approach is of a metonymical order, it replaces one thing by

cvcn thosc \vhich secm rnostdivorced from practical preoccupations

It is necessary to add that the balance between structure and

• See below, p. 199 n.

THE LOGIC OF TOTEMIC

There is certainly something paradoxical about the idea of a logic

bricoleur are products of 'becoming' pure and simple. Previously,

diatomaceae might be re\·ealed while their empirical basis werc

One can conceive that such a concrete logic may be possible. 1t

superfluous to conjure u p the bizarre hypotheses suggested to

Comment like this from the pen of a man of science is enough to

to invoke distinct principles to explain the conjWlction of these