J O E L P.

B R E R E T O N

UNSOUNDED SPEECH: PROBLEMS IN T H E

INTERPRETATION OF BU(M) 1.5.10=BU(K) 1.5.3

BU(M) 1.5.2.ff is a commentary on verses quoted in 1.5.1. These verses

describe the seven kinds of food which Prajgtpati created. Of these seven,
the verses say that he created three for himself, and the prose commentary
identifies them as mind, speech, and breath. These three comprise the self:
w10 (etanmryo va ayrm dtm~ valima'yo manomryah, prdn. amryah.). The
passage explains that within the self, the mind (mdnas) comprises all
thoughts, feelings, and mental dispositions and the breath (prdn. 6) comprises
all the vital breaths in the body. 1 In between the explanations of mind and
breath, the text gives an analysis of speech, which is the subject of this
paper. After these discussions of mind, speech, and breath, the text says
that these three principles correspond to the three worlds; the three Vedas;
gods, ancestors and humans; father, mother and child; known, knowable,
and unknowable; earth, heaven, and water; and fire, sun, and moon. In
short, the threefold division of the self corresponds to the threefold division
of every sphere of e.,dstence. Finally, the passage concludes: BU(M) 1.5.21
t6 et~ srtva ev6 sam~h srrve 'nant~h . . . y6 hait~n anantan up~ste 'nantrm
s6 lokrm jayati "So these [three] are all equal, are all endless. He who
respects: these as the endless conquers a world that is endless. "3
In BU(M) 1.5.10, the Upanis.ad sets forth the role of speech: ydh. kdg ca
~dbdo vag evd s~, esa hy 6ntam ayattd, esa hi n6. Radhakrishnan's transla-
tion ~ of this passage follows ~amkara's commentary: "Whatever sound there
is, it is just speech. Verily, it serves to determine an end (object), but is not
itself (determined or revealed). ''5 At the very least, this translation is rather
forced. First, it imports into the passage a view of speech as that which
reveals but is not itself revealed, although neither the context nor the phrase
dntam ayattd readily suggest it. Nor does this interpretation make clear
what such an understanding of speech has to do with the opening identifica-
tion of all sound as speech. According to ~amkara's comments, yf.h k6~ ca
grbdo v~g ev6 s~ describes the nature (svarEtpa) of speech, while the
remainder of the passage concerns its function (kdtya). But he does not
explain how its function derives from its nature, and thus he allows the
passage to fall into two unrelated parts. This separation is reflected in
Radhakrishnan's translation, which has a full stop after the first half and
suggests no relation between the two parts of the passage. Note that in

lndo-lranian Journal 31 (1988), 1 10.

9 1988 by D. Reidel Publishing Company.
2 J O E L P. B R E R E T O N

translating the passage in this way, Radhakrishnan has abandoned the

normal, causal sense of hi, perhaps because he recognized that in this
interpretation, the last two clauses are not explanations for the first.
Other scholars working independently of the Advaitan tradition agree in
not following garg.kara, but in little else. Consider, for example, the transla-
tions of Senart, B6htlingk, and Deussen: 6
Senart: "Tout ce qui est son est parole. Elle est (parole) par la fin
laquelle elle sert; elle n'est pas (par elle-mame une entit6 sp6ciale)."
B6htlingk: "Jeder beliebige Laut ist Stimme, da diese ein Ende findet und
auch nicht (d.i. es gibt immer ein Ger~iusch in der Welt)."
Deussen: "Alles was Ton ist, das ist Rede; selbige gehet zu Ende [als
menschliche Rede], und selbige auch wieder nicht [als die himmlische
V~c ...1."
These scholars agree at least on the translation of the first clause:
"Whatever is sound, that is speech." Note that this statement is not simply
an identification of speech and sound. Rather, the passage says that sound
can be subsumed under the category of speech. This interpretation emerges
from a straightforward translation and is supported by the particle ev6,
which in Vedic prose is characteristically restrictive. 7 While ev6 has a
number of functions within Vedic prose, 8 its restrictive function emerges
fully in identifications. Thus, nominal sentences of the shape x y ev6 or y
ev6 x mean that x is a subset of y, or that all x's are y's, or that x always
implies y. Because the point is an important one for the interpretation of
BU(M) 1.5.10, let me give some examples in which this value of evd is
evident.
First, ChU 7.10 describes the creation of earth out of water: ChU 7.10.1
dpa evemd mhrtd yeyam, prthivi "This earth is just these waters (become)
solidified." That is, the earth arises from nothing else except the waters, but
not all water becomes earth. The waters, which are marked by ev6, are the
set which contains and is greater than the earth. Or again, in BU(M)
4.4.6--7 the particle designates the single principle through which human
destiny is determined: yathdkdrt yathdcdrt tfthd bhavati . . . 6tho khflv dhuh
kdmamdya ev~ydm, pfrusa i t i . . . "As a person does, as a person acts, so
does he become, but as you know, some say that this person is made only
from desires." The evd here designates desire alone as the determining
principle. The other factors mentioned before are excluded. Finally, in the
passage directly preceding BU(M) 1.5.10, the Upanisad lists different kinds
of thought and feeling and then concludes that mind encompasses them all:
BU(M) 1.5.9 kamah, sam. kalp6 vicikits~ ~raddha~raddhd dh~tir ddhrtir hr[r
dhtr bh[r [ty et6t sfrvam mffna ev6 "Desire, conception, doubt, belief and
 UNSOUNDED SPEECH 3

unbelief, steadfastness and lack of steadfastness, shame, insight, fear -- all

this is just mind." That is, all these are part of mind, although mind is not
necessarily only these. It may include other functions as well. In the same
way, 1.5.10 says that all sound is speech, but not all speech need be sound.
This claim that speech encompasses all sound has precedent in the Vedic
tradition. ~fifikhfiyana Aranyaka 7.22 (ed., Keith) makes the same point:
sarvd vdg brahmeti smdha lauhikyah. / ye tu ke ca @bdd vdcam eva tdm
vidydt / tad athaitad t.'sir dha -- aham rudrebhir vasubhig cargtmi (RV
10.125.1) iti/saisd vdk sarva~abdd bhavati "'The brahman is speech as a
whole' -- so Lauhikya used to say. But whatever sounds (there be), one
should understand them to be just speech. And as to this, the .r.si thus said
- 'I move with the Rudras and Vasus.' Thus this speech is that which
-

encompasses the whole of sound." It may appear odd to view sound as the
smaller category subsumed by speech. After all, there are many kinds of
sounds which are not speech, such as the noises of objects or the sounds of
animals. Ordinarily, Vedic thinkers might also have seen speech as a
particular subset of sound, but metaphorically even these sounds of objects
and animals can be understood as speech. Thus, the mortar and pestle
"speak" as they grind the grain for the sacrificial cakes: MS 1.4.10 (58: 13)
utarlf~khaldyor 6dvaditor adhvd~u~ ca ydjamdnaJ ca vacarh yachetdm "And
moreover, until the mortar and pestle raise their voices, the adhvaryu and
the sacrificer should restrain their speech." Likewise, the sounds of the
pressing stones can be powerful utterances: RV 10.36.4a gravd vddann
dpa rdksdrhsi sedhatu "By speaking let the pressing stone keep away
hindrances."9 The sounds of animals are also speech, even if that speech be
unintelligible (rnirukta). According to SB(M) 4.1.3.16, the three quarters of
speech humans do not speak is this unintelligible speech of animals: tdd etdt
trr?yam vrc6 niruktam, ydn manusya vddanty dthaitdt t{triyaq7 vdc6 'niruktam
ydt pa~dvo vddanty dthaitdt t{triyam vdc6 'niruktam ydd vdydmsi vddan O,
dthaitdt tfiriyam, vdc6 'niruktam ydd iddm ksudrd~ saffsrpdm vfdati "So this
quarter of speech which men speak is intelligible; but this quarter of speech
which animals speak is unintelligible; and this quarter of speech which birds
speak is unintelligible; and this quarter of speech which the class of little
crawling things speaks is unintelligible." Thus, when the gfifikhfiyana
Aranyaka and the BU say that speech includes sound, they are implicitly
making the same metaphoric move we find in these passages. 1~
But even if this identification of all sound as speech has precedent, it is
not obvious. So in BU(M) 1.5.10, two hi-clauses follow and give reasons
for this claim. However, not only are these explanations elliptical; they also
appear contradictory. ~ The phrases es~ h i . . . esa hi n6 look as though
4 J O E L P. B R E R E T O N

something is asserted of "this" (e.sa) in the first clause and then denied of it
in the second. Assuming that the text does not simply contradict itself, how
are the two statements to be reconciled? The translations cited earlier
suggest several possibilities:
(1) The predicate of the first hi-clause applies to one aspect of es~ but
not to another. This is the approach of Deussen, for example. Human
speech "comes to an end," but divine speech does not. B6htlingk gives a
variation of this same idea: in one respect speech has an end, but in another
respect it does not, for there is always some kind of noise in the world.
(2) The predicates applied to esa are not the same. This is the approach
of gamkara and of Senart. In Senart's rendering, for example, the first
clause asserts that speech is defined by its purpose and the second that
speech is not an independent entity. Or:
(3) The two esa-s actually have two different antecedents. It is this third
alternative that I consider the best solution.
The choice among these possibilities depends largely on the interpretation
of the first hi-clause. The central problem here is ayatta. The root yat has
been the object of considerable attention in the scholarly literature.
Oldenberg showed that ydt(~yate means "he places himself in position" ["er
stellt sich auf"] and y6tati, ydt6yati means "he assigns the (correct) place to
someone" ["jemandem die (rechte) Stelle anweisen"]. 12 The compound a +
),at carries much the same sense of "place," "fix (in place)," or "arrange (in
place)." Nominal and verbal derivatives of a + yat are fairly frequent in
Vedic, 13 and in meaning, they are closely related to derivatives of pr6ti +
sthd. Compare ~B(M) 9.3.4.13 svayam evdinam etdd diEy ~yattam pr6tis.thi-
tam abhisificati "He anoints him [the yajarndna], placed and firmly
established in his very own region [= the northeast]." Note that in this
passage ayatta governs the locative, not the accusative. Similarly, in all other
Vedic attestations of ~yatta, the complement of ayatta, where it exists, is
locative. Other examples are AiA 3.1.2 (128: 4--5) tasmin hdsminn dkdge
prdn. a dyatto yathdmusminn dkdge v6yur dyattah "In this space [between the
two halves of a person, corresponding to the midspace between heaven and
earth], breath is placed, just as the wind is placed in yonder space"; and
AV(S) 11.7.3 sdnn t~chiste 6sam~ cobhdu, mrtyfir vajah, prajapatih., laukya
6chis.ta ayattdh . . . . "The one which is and which is not, both (are fixed) on
the remnant; (also) Death, Strength, Prajfipati. (The beings) of the world are
fixed on the remnant." This idea that all things in the world are dependent
on the {tchista, the remainder of the oblation, is expressed in the surround-
ing verses through the participles ahita (vss. 1, 2), sdmhita (,vss. 1, 2),
drd.h6 (vs. 4), and grim (vs. 4), which confirm the sense of dyatta as "fixed"
 UNSOUNDED SPEECH 5

or "arranged." Similar also is g~lSkhB 6.11 (25:21--26:1) atra ca bh~yisthd

hotrd dyattd bhavanti "On it [= the RV] are placed the greatest number of
priestly functions." Here, of course, 6tra functions in place of a locative.
This construction continues in post-Vedic literature as well. 14
If ~yatta governs the locative and has the sense of "placed," "fixed" or
the like, then 6ntam cannot be the object of ayatta or the goal of some
motion implied by ayatta. The better alternative is to construe 6ntam
adverbially, in the sense of "to the end" or "to its limit." 15 There is, how-
ever, one other possibility. According to Maue, 16 the accented manuscripts
of the Kanva recension of the BU read 6ntam ~ yatt~. The simplex yatt6 is
attested elsewhere, 17 albeit infrequently, and {mtam ~ should mean "up to
its end" or "to its limit." In this case, the accusative is governed by the
postposition a. 18 Whichever the reading, the sense of the whole phrase
remains much the same: something is "fixed to its limit."
Since 6ntam is not the complement of ayatta in esa hy 6ntam ayattd,
either ayatta is used without a complement or there is an ellipsis of either
vaci or ~6bde. If there is an ellipsis, then which word should be under-
stood? Since esa is feminine, the pronoun seems to refer to yak and
therefore g6bde should be supplied. The translation then would be: "For
this (speech) is fixed to its limit (on sound)." But by making speech
dependent on sound, this translation gives exactly the reverse of the
conclusion we expect. The proposition which this clause explains is that
speech encompasses sound, but this interpretation has sound encompass
speech. Also, recall that the conclusion of the whole passage is that speech,
breath, and thought are all anant6, "endless." Given that, it would be
peculiar to have speech described here as having an end 19 and as depen-
dent on another, more fundamental principle. The phrase ought to be
turned around: "For this (sound) is fixed to its limit (on speech)." But is this
interpretation justifiable?
One possibility would be to correct the text. Perhaps under the influence
of the following esa hi and the preceding sa, an original *es6 hi became esa
hL But I believe that esa is more likely to have been part of the oldest text
and not a textual error. The phrase v~g ev6 sa esa hy 6ntam ~yattd is better
explained in this way: In vag ev6 sa, the antecedent of sa is gfbda, "sound."
The pronoun is feminine according to the rule that in nominal sentences, a
pronoun agrees with its predicate nominative in number and gender. 2~ The
et6 pronoun has the same antecedent as sc~, but because it immediately
follows sa and resumes sa, it takes the gender of s~ rather than that of
g6bda. Because the et6 is in a different clause from s~, we might have
expected it to revert to the gender of ~6bda. But ~6bda is remote, and it is
6 J O E L P. B R E R E T O N

unexpressed in this clause. Moreover, such reversion would have created

the distinctly odd situation of two pronouns in immediate succession, with
the same object but in two different genders. The reversion might be
hypercorrect, but it would sound unnatural.
This construction was also supported by another syntactic pattern. The
collocation td etd is frequent in Vedic. The two pronouns often introduce
clauses, and naturally, if they refer to the same object, they agree with one
another. Here the two are in difference clauses, but their recurrent pairing
would favor the phrase sa esa hi against *s~ esd hL
One advantage of taking g6bda as the subject of the first hi-clause is that
the concluding es~ hi n6 then offers no difficulty. In this clause, esa refers
to speech. "And this is not," that is, "this speech is not fixed to its limit" or
"to any limit." The two hi-clauses, then, imply the following: All sound has
a distinct beginning, an end and an identifiable form; therefore, it is limited
and structured. Speech, on the other hand, is not always articulated; and
therefore it does not always have a limit, a fixed place, or a determined
arrangement. All sound is the "speech" of a being or an object (cf. gB(M)
1.1.4.13), but not all speech is heard, and therefore not all speech is sound.
Other passages in the Veda also describe a speech which is beyond the
limits of sound. The best known example is RV 1.164.45 catvari v~k
pdrimitd padani, t~ni vidur brdhmana y~ man~in, ah., g(thd tr[ni nihitd
ndhgayanti, turlyam, vdc6 manusya vadanti "Speech is divided into four
quarters. These do brahmans who are wise know. The three (parts) hidden
in secret they do not stir up. The fourth (part) of speech humans speak."
The Vedic prose texts interpret the three quarters and one quarter of
speech in various ways, 2~ but Geldner 22 gives the most straightforward
interpretation in identifying the three quarters of speech with divine speech
and the one quarter with audible human speech. Even clearer is AV(g)
5.1.2.cd dhdsylJr y6nim prathamd a rived-, a y6 vacam dnuditdm cikdta
"Seeking his foundation, he first entered his origin -- he who understood
the unuttered speech." Again, unuttered speech, unsounded speech,
constitutes its own distinct realm.
This idea of a speech beyond sound is carried forward in the Brfihmana
tradition. In Brfihmanic symbolism, articulate speech represents the world
and all things which have limits. Normally v~k designates such speech, and
it is often contrasted to thought, which is without definable boundaries, as
in gB(M) 1.4.4.7 vag vdi mfnaso hrdsfyasy dparimitataram ira hi mdnah
pdrimitatareva hi vak "Now, speech is smaller than thought, for thought is
much more unlimited and speech much more limited." But there are also
forms of speech which are not carefully articulated or are not audible, and
these forms of speech are also not bounded and not ordered.
 UNSOUNDED SPEECH 7

Such speech in the ritual tradition is called updmd~ "in an inaudible

voice." This form of speech is the normal method of the recitation of a
yajus (cf. gB(M) 4.6.7.18; KgS 1.3.10). The ritual texts explain updrnd{t as
the recitation of a mantra in which the lips are visibly moving and the
mantra is murmured, but in which no sound is audible at a distance. 23 In
Brfihmanic interpretation, updmd~ belongs to the realm of the "unintelligible"
or "undefined" (dnirukta), and therefore it belongs to Prajfipati, the
undefined god. See, for example, ~B(M) 6.2.2.20--21 tdd va updrhd~
bhavati / etdd dhaitdih, praj@atih, pad~bhih, kdrmeyesa tdd dhatranad-
dhevaivasaniruktam ira tdsmdd updrndti // ydd v evdpdmd~ / prajdpatydm, v~
etdt kdrma prajapatim, hy ~tOna kdrmandrabhatd 'nirukto vdi prajapatih.
"Now, this (mantra) is (recited) in an inaudible voice. By means of these
animals, Prajfipati sought this rite. At this time, that (rite) was not at all
clearly (known), not at all intelligible. Therefore, (the recitation is) in an
inaudible voice. And again as to (the recitation ) in an inaudible voice: (the
reasons are that) this rite belongs to Prajfipati, for by this rite (the sacrificer)
obtains Prajfipati, (and that) Prajfipati is unintelligible. ''24 Elsewhere the
gatapatha Brfihmana also says that indistinct speech is appropriate to
Prajfipati because the inaudible is the unintelligible, the unintelligible
represents the unintelligible whole, and Prajfipati is the whole world: so
~B(M) 1.3.5.10 sdrvam, vdi prajapatih. . . . upd~ggt devdtdm yajaty dniruktam
va updmigt sdrvam va dniruktam. In general, the "unintelligible" ( dnirukta )
comprises all those elements of the world without determined order, such
as breath, wind, thought, and the future. -~5 Since the inaudible is unintel-
ligible, then it too is unstructured and unlimited. Thus, inaudible recitation
is a form of speech which is without limit and without sound. 26 It is an
example of an "unsounded speech" which goes beyond the limits and
structure of sound.
Finally, then, the translation of BU(M) 1.5.10 is: "Whatever is sound,
that is just speech, for this (sound) is fixed to its limit [on speech] and this
(speech) is not [fixed on anything]." Speech encompasses sound, for all
sounds can be understood as the speech of some being or object, but
speech is more than sound, for sound is limited to what is audible and
speech is not. Thus, in BU(M) 1.5.8--10, as the mind (mdnas) includes
within itself all mental processes and as breath (prdn.d) includes all the vital
breaths, so speech contains all the sounds of the body. For these three
powers are not simply located in the head but completely pervade the body.
BU(M) 1.5.9 says that mind is everywhere in the body: tfsmdd dpi prsthfta
@asprsto mdnasd vijdndti "Therefore, if one is touched even from behind, he
perceives (that touch) by the mind." In the same way, the prdn.6 as the five
vital breaths sustains life throughout the body. And likewise again, speech is
8 J O E L P. B R E R E T O N

not limited to audible words; it is everywhere in the body as both its sounds
and as its inaudible murmurings. Furthermore, as the remainder of the
passage makes clear, the body is the image and model of the world as a
whole. Thus the powers of mind, speech and breath comprise not only the
essential self of the person but also the essential nature of the world. They
pervade not only body, but also, through their macrocosmic equivalents,
they pervade all existence. Because these three powers are thus unlimited,
one who knows them possesses an unlimited world.

NOTES

1 BU(M) 1.5.9, 10 kamah, san..zkalp6 vicikitsa graddh~graddhd dhhir 6dhrtir hr~r dh~r bhlr ~ty
etdt sclrvam mdna evd . . . / prdn. d 'pgm6 vydnd ud~nrh saturn6 'n6 #y etrt srrvam, prdnd evr.
z For the sense of 6pa-ds, see H. Falk, ZDMG 136 (1986), 80ft.
3 gamkara explains that they are " e q u a l . . . in that they pervade the living world, including
that of the self and that of creatures" (tulyd ~y6ptimanto ydvatprdnigocaram sddhyrtmddhi-
bhtitam vydpya vyavasthitdh) and that therefore they are endless, "for they exist as long as
the world" (ydvatsamsdrabhdvino hi re).
4 S. Radhakrishnan, trsl., The Principal Upanisads (London: 1953), p. 175.
5 Samkara on BU(K) 1.5.3: athed~nim vdg vaktavyety drambhah -- yah. ka~ ca loke ~abdo
dhvanis tdlvddivya~igyah, prdn. ibhir var'nddilaksana itaro vd vdditrameghddinimittah, sarvo
dhvanir vdg eva sd / idam tdvad vdcah, svargtpam uktam /' atha tasydh kdryam ucyate -- es6
vdg ghi yasmdd antam abhidheygtvasdnam abhidheyanirn, ayam dyattdnugatd / esd punah.
svayam, ndbhidheyavat prakd~ygzbhidheyaprakdgikaiva prakdddumakatvdt pradipddivat.
6 E. Senart, trsl., Brhad-Aranyaka-Upanis.ad (Paris: 1934), p. 18; O. Brhtlingk, trsl.,
Brhad6ran.fakopanishad in der Mddhfarhdina-Recension (St. Petersburg: t889), p. 16; P.
Deussen, Sechzig Upanishad's des Veda, 3 Aufl. (Leipzig: 1921; rpt. Darmstadt: 1963), pp.
401f.
7 Sanskrit commentaries regularly describe ev6 as avadhdrandrthah "having the sense of
restriction." Cf. G. Gren-Eklund, A Stud), of Nominal Sentences in the Oldest Upanisads
(Uppsala: 1978), pp. 107ff., and B. S. Gillon and R. P. Hayes, WZKS 26 (1982), pp. 195ff.
This is the sense of the word according to the Sanskrit grammarians also. See L. Renou,
Terminologie grammaticale du sanskrit (Paris: 1957; reissue of Biblioth6que de l'l~cole
pratique des Hautes l~tudes. 280--282 [1942]), p. 65.
s See the survey in J.-M. Verpoorten, L'Ordre des mots dans l'Aitareya-Brdhman, a (Paris:
1977), pp. 239ff. and especially his summary on p. 256, and A. Minard, Trois dnigmes sur le
Cent Chemins I, II (Paris: 1949, 1956), especially I, pp. 47ff. and II, pp. 225ff.; La sub-
ordination dans la prose vkdique (Paris: 1936), especially pp. 116ft. Both Verpoorten and
Minard see ev6 as fundamentally an emphatic particle, which carries a sense of restriction
only in certain cases.
9 See A. Hillebrandt, Vedische Mythologie, 2 Aufl. (Breslau: 1927; rpt. Hildesheim: 1965) I,
p. 406, especially fn. 3.
~o Note further that in both ~fifikhA 7.22 and BU(M) 1.5.10, the td-pronoun agrees in
number and gender with its predicate nominative or appositive, not with its antecedent. The
Upanisad has rag ev6 sa: sd is feminine with yak, though it refers to masculine gdbda. The
Aranyaka has vdcam eva tg~m vidydt, in which tam agrees with vdcam but refers to gdbda.
Such agreement is regular. Again, because this agreement is occasionally overlooked, let me
give a few illustrations: AiB 3.34.2 y~ni parilcy.@dny dsams te krs.n,dh pa~avo 'bhavan "What
 UNSOUNDED SPEECH 9

were the cinders, they became the black animals." JB 3.326 tasyai pa~cevdksare upodasarpa-
tdm saivaisdnustub abhavat "From her there developed two syllables sometime later. They
became this Anu.stubh." ChU 1.3.3. yo ~ydnah sd v6k. 1.3.4. ya rk tat sdrna . . . yat sdma sa
udg#hah. Cf. I. Ickier, Untersuchungen zur Wortstellung und Syntax der Chdndogyopanisad
(G6ppingen: 1973), pp. 4ft.; L. Renou, Grammaire sanscrite (Paris: 1961), w p. 500; B.
Delbrfick, Altindische Syntax (Halle a.d.S.: 1888; rpt. Darmstadt: 1968), w p. 565; J.
Brereton, ZDIVIG 136 (1986), pp. 99ff.
Jt Double hl-clauses, like the ones here, are of two types. On the one hand, the second
hl-clause may be simply a continuation of the explanation introduced by the first hi-clause.
Cf. ~B(M) 1.3.2.9 6tha y6d ast.a6 kftva upabh(ti g.rhl.z~ti / ch6ndobhyas tdd grhn6ty
anuydjObhyo hi tfd grhn.ati chfnddmsi 173,6nto,djdh "So then, when he takes (butter portions)
eight times in the offering-spoon, he then takes them for the meters, because he then takes
them for the after-offerings and the after-offerings are the meters." In such eases, the second
hi simply resumes the first hL This is obviously the case when one hi-clause is dependent on
another rather than coordinate: SB(M) 1.8.1.28 vatsa u v6i yajfifpatim vardhanti ydsya hy ~t~
bhuyis.t.hd bh6vanti s6 h[yajfidpatir vardhatd " .. and as a rule calves increase the patron of
the sacrifice, for to whome belong these in great abundance, that patron increases." On the
other hand, the second hl-clause may state a reason for the truth of the first hi-clause rather
than for the main clause. In gB(M) 1.5.4.6ff. the Asuras try to find a match for each number
uttered by Indra. When he says dka (m.), the Asuras say dkd (f.). They continue in this way
until at last Indra says, "five" (p6~ca): SB(M) J..5.4.11 t6ta [tare mithunfm n~vindan nO hy
6ta ~rdhv6m. mithunfrn dsti p6tica pfr]cdti hy ~vait6d ubhdyam bhdvati "Then the others
found no pair, for after that (numeral four), there exists no pair, for both (the masculine and
feminine words for "five") are only p6gca." Whichever way the two hi-clauses in BU(M)
1.5.10 are understood, a superficial contradiction remains.
~-~ IF 31 (1912/13), pp. 127--134. In substantial agreement, P. Thieme, Der Fremdling im
Rigveda (Leipzig: 1938), p. 48; pp. 39ff.; B. Schlerath, Das Ko'nigtum im Rig- und Athar-
vaveda (Wiesbaden: 1960), pp. 37ff.; J. Gonda, The Vedic God Mitra (Leiden: 1972), pp.
93ff.; J. Brereton, The R.gvedic Adilyas (New Haven: 1981), p. 23; 8. Jamison, Function and
Form in the -dya-Formations of the Rig Veda and the Atharva Veda (G6ttingen: 1983), p.
131.
r Especially common is the noun dyftana, which Oldenberg defined as "'the place (Ort)
where something belongs or from which it can achieve a particular result." See Vorwissen-
schafiliche Wissenschafi (G6ttingen: 1919), p. 129. In his study of @6tana (ALB 23 [19691,
pp. lff.), Gonda offers a variety of glosses in different contexts, including "homestead,"
"place of rest and safety," "'locus," "the 'seat' of a divinity or power," "place or position,"
"substratum." etc.
~4 Examples are Manu 7.65 amdt3,e dand.a dyattah "The army depends on the official (place
in charge of it)," -- the translation is from G. Bfihler, The Laws of Manu (Oxford: 1886), p.
226 -- and .RgVidh 3.21.2 cittam ca te m6nag ca te m@i dh~ta nlyacchatu m@i te cittam
ayattat.n m6nas te mfyi sfmagnute "Let Dh~tar put your mind and thought on me. Your
thought is fixed on me. Your mind is joined to me."
1~ Cf. gBIM ) 10.5.4.15 s6 'syais6 s6rvasydntfm evatma s6 es6 s6rvdsdm apam mddhye s6 esd
s6rvaih kdmait3 sfrnpannal.t "So this (self) of him [Agni] is the self of all to its very limit: it is
in the midst of all the waters, and it is endowed with all objects of desire." But otherwise, J.
Eggeling, The .~atapatha-Brdhmaqa IV (Oxford: 1897), p. 389.
16 D. Maue, Brhaddral.Tyakopanisad L Versuch einer kritischen Ausgabe nach akzentuierten
Handschrifien der Kdnva-Rezension mit einer Einleitung und Anmerkungen (Giessen: 1976),
p. 30.
~7 E.g., AV(P) 9.12.6; KS 37.11 (62: 7); JB 1.197.
l~ Cf. Delbriick, w p. 452.
10 J O E L P. B R E R E T O N

~9 It is vital that speech not have an end, for if it had an end or were an end, it would be
associated with death. BU(M) 1.3.11 (K 10), for example, explains the link between limit
and death. According to it, the breath in the mouth (~,sany6prdn.d) drove the evil that is
death to the borders of the four directions (~sdm. di~dm 6ntah.). And so, the passage
concludes, t~smdn nd j6nam iydn n~ntam iy~n net papmanam, mr.ty~m anvavayanttl
"therefore, one should not travel to (another) people, one should not travel to the border in
order not to follow the evil that is death."
2o This rule of agreement is discussed in note 10.
21 For example, in gB(M) 4.1.3.16, quoted above, the three parts of speech are the speech
of animals.
22 K. Geldner, Der Rig-Veda, I (Cambridge, MA: 1951), p. 236.
23 Cf. C. Sen, A Dictionary of the Vedic Rituals (Delhi: 1976), p. 55; H. Oldenberg, The
Gr.hya-St~tras, II (Oxford: 1892), p. 318; L. Renou and L. Silburn in Sar~pa-Bhdra6
(Hoshiapur: 1954), pp. 68f.; L. Renou, JAOS 69 (1949), p. 11, n. 1.
24 Cf. ~B(M) 1.4.5.12; 6.2.2.20; 10.3.5.15.
25 Cf. Renou and Silburn, p. 74, and Renou, JAOS 69, p. 14.
26 The Upani.sadic tradition understands sound as the external element corresponding to
hearing. Audibility, therefore, is its defining characteristic, e.g., BU(M) 2.4.11 s~rvesdm
~dbddndm ~r6tram ekdyan~rn "Hearing is the point of union of all sounds"; BU(M) 3.2.6
gr6trena "hi~6bddfic chrn6ti " . . . for by hearing, he hears sounds"; KauU 1.7 kena sabddn iti
grotreneti "'By what (cio you apprehend) sounds?' (He should answer,) 'By hearing.'" Since
mantras spoken updmgft are not audible, then they are not truly sounded.

Department of Religious Studies,

University of Missouri
416 G.B.C.
Columbia, MO 65211
U.S.A.