Orthoepic Diaskeuasis Of Rgveda And Date Of Panini

Author(s): Johannes Bronkhorst
Publisher: Johannes Bronkhorst
Catalog link: https://jainqq.org/explore/269708/1

Page #1

JOHANNES BRONKHORST
THE ORTHOEPIC DIASKEUASIS OF THE RGVEDA AND THE DATE OF PANINI*
P. RPr
ABBREVIATIONS
Paninian sutra
Rgveda-Prätisakhya. The sūtra numbers both according to Mangal Deva Shastri's and Max Müller's editions are given.
Rgveda
RV
1.1. The Rgveda is known to us in a form which is fixed down to the minutest details. It obtained this form as the result of a process which, in as far as it concerns details of sandhi etc., is known by the name 'orthoepic diaskeuasis'.
The main hypothesis to be defended in this article is that the orthoepic diaskeuasis of the Rgveda was not yet completed in the time of the RgvedaPrātisakhya, and ended when but one version of the Rgveda remained, i.e., probably with the disappearance of the Başkala Samhita. (I do not take here into consideration the Kashmir Rgveda; see Bronkhorst, forthcoming.) The hypothesis contrasts with the currently held belief that the Śākhās of the Ṛgveda, as well as the Rgveda-Prätisakhya, presuppose, and therefore postdate, the final redaction of the Rgveda (Renou, 1947: 21, 35; cf. 1960: 1-2, 10).
A decision procedure, on the basis of which we can choose between these two opinions, is provided by the following. We have some idea of the original form of the hymns of the Ṛgveda, since the present Rgveda often deviates from the metre in a way that can easily be restored by undoing the sandhi or other minor changes. If the Ṛgveda-Prātisakhya stands somewhere in the process which began with the original form of the Rgvedic hymns, we may expect that at least some of the authorities who preceded the Prātisakhya but took part in the same process, came out in defence of a form of those hymns which, at least in some cases, deviates from their present, and is closer to their original one. If, on the other hand, the Prätisakhya belongs to a period which came after the orthoepic diaskeuasis, we may not expect such opinions on the part of those who took part in the development in which the Prätisakhya participates.
The Rgveda-Prätisakhya mentions the following authorities: Anyatareya1 (3.22(208)), Gärgya (1.15(16); 6.36(412); 11.17 (629); 11.26(638); 13.31(739)), Pancala (2.33(137); 2.81(185)), Pracya (2.33(137); 2.81(185)); Mākṣavya (Intr. v. 2); Mándükeya (Intr. v. 2; 3.14(200)), Yáska (17.42(993)), Vedamitra
Indo-Iranian Journal 23 (1981) 83-95. 0019-7246/81/0232-0083 $01.30. Copyright © 1981 by D. Reidel Publishing Co., Dordrecht, Holland, and Boston, U.S.A.

Page #2

84
JOHANNES BRONKHORST
(1.51(52)), Vyāli (3.23(209); 3.28(214); 6.43(419); 13.31(739); 13.37(745)), Sākațāyana (1.16(17); 13.39(747)), Śākala (1.64(65); 1.75(76); 6.14(390); 6.20 (396); 6.24(400); 6.27(403); 11.19(631); 11.21(633); 11.61(673)), Śākalya (3.13 (199); 3.22(208); 4.13(232); 13.31(739)), Śākalya (sthavira) (2.81(185)), sākalyapit? (4.4(223)), Śūravīra (Intr. v. 3), Śūravīra-suta (Intr. v. 3). Unfortunately, none of the opinions ascribed to these authorities in the Prātiśākhya has an effect on the metre of the hymns, be it positively or negatively. However, many of these authorities are mentioned elsewhere in the ancient, and not so ancient, literature, and opinions are ascribed to them which are not found in the Rgveda-Prātiśākhya. Many of these other opinions, also, do not affect the metre, but there are some which do in a way that deserves our attention:
(i) Pāņini's Astādhyāyi contains the following rule: P.6.1.127: iko'savarne śākalyasya hrasvaś ca (samhitāyām (72), ekah pūrvaparayoh (84), na (115), aci (125)] “[In the opinion) of Śākalya, in connected speech (samhitā), no single (substitute] of what precedes and what follows (comes) in the place of the vowels) i, i, u, ū, ?, ?, !, when a dissimilar vowel follows; and (if the earlier vowel is long,] a short (vowel comes in its place]."
The translation here given follows the interpretation of the Kašikā (except in so far as this is not possible in view of footnote 12). The interpretation may, however, be improved upon by understanding the word chandasi "in Sacred Literature" (Thieme, 1935: 68) in this rule, from the preceding one. Both the mention of the name "Śākalya" and the unusual kind of sandhi described support this. We may. expect that this rule was (also) valid for the Rgveda.
The Rgveda in its present form is not in agreement with Sakālya's rule. The earlier form of the Rgveda, on the other hand, agrees with it. E. Vernon Arnold (1905) makes the following statements about the original Rgveda. First: “Before dissimilar vowels final -i -i -u -ū are regularly used with hiatus" (p. 76). Second: “The vowels -i, -ū are regularly shortened when followed by dissimilar vowels, but there are many exceptions" (p. 135). Third: “Final -a, -ă are regularly combined with an initial vowel or diphthong following: and final -2 -1 -u -ū are regularly combined with similar vowels, that is-i or -1 with either - or -7, and -u or ū with either -y or ū” (p. 72). These three statements are so close to the opinion ascribed to Śākalya in P. 6.1.127 that they are almost a translation of that rule.
(ii) Puruşottamadeva's Bhāṣāvrtti on P.6.1.77 contains the following line (quoted in Mishra, 1972: 30n, 32n; Mīmāmsaka, 1973:1:26): ikām yanbhir vyavadhānam vyādigālavayor iti vaktavyam/dadhiyatra dadhy atra madhuvatra madhv atra/ "It must be stated that in the opinion) of Vyādi and Gālava there is separation of [the vowels] i, u, ?, ? by [the consonants) y, v, r, 7 [respectively. Examples are] dadhi-y-atra [for dadhi atra, where we normally find) dadhy atra, madhu-v-atra [for madhu atra, where we normally find) madhv atra." The kind of

Page #3

RGVEDA AND PĀŅINI
sandhi here ascribed to Vyādi and Gálava is not found in our Rgveda. (It is found in a few places elsewhere in the Vedic literature; see Mīmāņsaka, 1973: I: 27f.) It would, however, make good the metre of the hymns of the Rgveda in innumerable instances (Whitney, 1888: 39, 8113).
(iii) The third case rests upon a somewhat unorthodox interpretation of some rules of the Astādhyāyi, an interpretation which, however, has rather strong arguments to support it. They will be discussed in § 1.2.3.
Pāņini's grammar contains the following three rules: P. 8.3.17: bhobhagoaghoapūrvasya yo'ói [roh (16), raḥ (14)] "In the place of r of rU, which is preceded by bho, bhago, agho, a ora, (comes) y, when a vowel or voiced consonant follows."
P. 8.3.18: vyor laghuprayatnataraḥ śākatāyanasya (asi (17)] “According to sākatāyana, in the place of v and y [comes a substitute] of which the [articulatory] effort is lighter, when a vowel or voiced consonant follows."
P. 8.3.19: lopah sākalyasya [vyoh (18), aśi (17)] “According to Śākalya, there is elision of v and y when a vowel or voiced consonant follows."
When these rules are applied to a word en ding in as that is followed by a-, this sandhi evolves: asta-> -a-rU+a- (8.2.66) > -ay+a-(8.3.17) or aỹta-(8.3.17&18) or ata-(8.3.17&19). None of these three forms is ever found in our Rgveda, which invariably haso- or ota-. The metre requires two distinct syllables in the vast majority of cases and that the first syllable be metrically short (Wackernagel, 1896: 324, § 272b; Ghatage, 1948: 14). Oldenberg (1888: 458) has argued that the original reading was ata-4 We note that this is the opinion of Sākalya expressed in P. 8.3.19. Oldenberg (1888: 457-58) further shows that av for as occurs in the Vedic literature, and does not exclude the possibility that uy+a- for asta- was the original form in the Rgveda. This would correspond to the opinions of Śākațāyana (p. 8.3.18) and Pāņini (if P. 8.3.17 gives indeed Pāṇini's opinion).
All these three passages need some further comments.
1.2.1. There is no reason to doubt that the sākalya mentioned in the Aştādhyāyi is identical with the sākalya mentioned in the Rgveda-Prātišākhya. On one occasion we find in the Astādyāyi an opinion ascribed to sākalya which the Rgveda-Prātiśākhya ascribes to the followers of Sākalya (Bronkhorst, forthcoming). P. 1.1.16, moreover, seems to bring Śākalya in connection with a Padapātha. We know from Nirukta 6.28 that the author of the Padapātha of the Rgveda was called thus. The connection of the Sākalya mentioned in the Aştādhyāyi with the Rgveda seems therefore established.
1.2.2. Of the two, Vyāļi and Gālava, only the first one is mentioned in the Rgveda-Prātiśākhya." It is unlikely that Purusottamadeva derived his knowledge directly or indirectly from the Samgraha, a work reputedly written by someone

Page #4

86
JOHANNES BRONKHORST
called 'Vyādi'. All we know about this work (see Mīmāmsaka, 1973: 1: 282-90) shows that the Samgraha dealt with philosophical questions, and was not just a grammar. We are therefore justified in neglecting the claim of the commentator Abhayanandin on the Jainendra grammar to the extent that this rule derives from the Samgraha and is there ascribed to "some" (Jainendra Mahāvștti 1.2.1: ikām yanbhir vyavadhānam ekeșām iti samgrahaḥ; quoted in Mīmāmsaka, 1973: 1: 26n). We further do not have to decide whether the two Vyādis are one and the same or not. 1.2.3. The example asta- would yield o- according to the orthodox interpretation of Pāņini's grammar, in the following manner: asta- > -a-rU+a- (8.2.66) > -a-u+a- (6.1.113) > -ota-(6.1.87) > -o- (6.1.109). There can be no doubt that this form of sandhi was also accepted by Pāņini, for his own grammar makes an . abundant use of it, e.g., in P. 8.3.17 (see above) which has yośi for yastasi. The question is if only this form was accepted. Some circumstances indicate that such is not the case.
The fact is that a strict application of the principles of Pāṇini's grammar can not lead to o-, only to -ay+a-, -aỹta-, and a+a-! To understand why, we must recall that the substitute rU for s is introduced in P. 8.2.66, a rule which is part of the last three sections of the Astādhyāyi, the so-called “Tripādi", which has a linear rule ordering (Bronkhorst, 1980: 72f.). Use of P. 8.2.66 can therefore only be followed by application of a rule which comes after P. 8.2.66, certainly not by application of P. 6.1.113, which would be necessary to obtain o-.
The location of P. 6.1.113 is the most flagrant violation of the principle of linear rule ordering of the Tripādi which there is in the Astādhyāyi (cf. Buiskool, 1939: 83,99). P. 6.1.113 reads: ato ror aplutād aplute (ati (109), ut (111)] "In the place of rU which follows a that is not prolated, (comes] u, when a nonprolated a follows." This rule presupposes the presence of the substitute rU. But rU is not introduced except in the Tripādī. Strictly speaking P. 6.1.113 should never apply, and be superfluous. Why was P. 6.1.113 not located in the Tripādi, somewhere after P. 8.2.66 and before P. 8.3.17?
I think there are two answers to this question, which are simultaneously valid. The first is that P. 6.1.113 has to "feed" P. 6.1.87 in the derivation of o- out of asta- (see above). This answer alone is not fully satisfying, for if the linear ordering of the Tripādi was to be broken, then why not after the application of P. 6.1.113?? The second answer is that if P. 6.1.113 were located in the Tripādi, it would make the derivation of ay+a-/-ay+a-/-a+a- out of asta-impossible. That this second answer leads to a result which agrees so well with the original Rgveda, only confirms that it is most probably correct. 1.3. The above shows that Śākalya was not the final redactor of the Rgveda, as

Page #5

RGVEDA AND PĀNINI
87
Patanjali's Mahābhāsya seems to say he was (on P. 1.4.84, vol. I, p. 347, 1.3: śākalyena sukrtām samhitām anunišamya devah prāvarsat). Patañjali's opinion illustrates the process of apotheosis which Śākalya underwent,8 as I observed elsewhere (Bronkhorst, forthcoming).
I shall now show that other data we possess about Sākalya and his Padapātha agree, or at any rate do not disagree, with the view that Śākalya preceded the final redaction of the Rgveda.
1.3.1. Aitareya Āranyaka 3.2.6 lays down two rules: where there is doubt whether or not ņ is to be used, there ņ must indeed be used;' where there is a similar doubt regarding ș, there s must be used (p. 139: sa yadi vicikitset saņakāram bravāniṁ anakārā3ṁ iti sanakāram eva brūyāt sasakāram bravāņi3ṁ aşakāra 3ṁ iti sasakāram eva brūyāt). The same chapter of the Aitareya Aranyaka (3.1.2) mentions the opinion of Sākalya regarding the mystical significance of union (samhitā). Doubts regarding the correct form of the Rgveda were apparently still alive in the time after Sākalya.
1.3.2. Six verses of the Rgveda have no Padapātha. They are RV 7.59.12; 10.20.1; 121.10; 190.1–2–3 (Kashikar, 1951: 44). This is most easily explained by the assumption that these verses were not considered part of the Rgveda by Śākalya. It further shows that the final redactors did not hesitate to deviate from the composer of the Padapātha in deciding what did, and what did not, belong to the Rgveda. (It is interesting to note that at least one hymn of the Rgveda (10.95) is known to have had fewer verses than at present at as late a date as that of the Śatapatha Brāhmana. See Oldenberg, 1912: 303.)
1.3.3. Oldenberg (1888: 384-85) points out that the Samhitā text contains several nom. sing. fem. words ending in a which are not joined with a following vowel. Oldenberg, following Lanman, explains this by assuming that the final redactors of the Rgveda considered these words as really ending in āh. The Padapātha, on the other hand, presents all these forms as actually ending in a. This indicates that the maker of the Padapātha and the final redactors of the Samhitā were different persons. Since the final redactors did not consider the Padapātha authoritative (see above, further fn. 8), this fact does not conflict with Sākalya's temporal priority to these redactors.10 .
2.1. In what phase of the development of the Rgveda does Pāṇini fit? There is no doubt that Pāṇini came after Śākalya, for he mentions the latter four times (P. 1.1.16; 6.1.127; 8.3.19; 4.51; see above pp. 84 and 85. The question is: Had the Rgveda known to Pāņini already obtained the form which it had in the time of the

Page #6

88
JOHANNES BRONKHORST
Rgveda-Prātiśākhya, and which was to remain virtually unchanged ever since? Three places of the Astādhyāyī seem to indicate that this was not the case.
(i) P. 6.1.134: so'ci lope cet pādapuranam (sulopaḥ (132)] “There is elision of the nom. sing. case-affix] SU of sa 'he' before a vowel, if, in case of elision, there is completion of the Pāda." This rule is obeyed in our Rgveda where sas is followed by a vowel different from a; e.g., in RV 1.32.15: sed u rājā kşayati carşaņīnām for sah/ it/ etc., and in RV 8.43.9: sausadhir anu rudhyase for sah/ oşadhih/ etc. (cf. Oldenberg, 1888: 464; Arnold, 1905: 74). Where, however, sas is followed by a- and the metre requires contraction, "ist in einer Reihe von Fällen să-überliefert ..., in einigen andern so a- oder so mit dem Abhinihita Sandhi" (Oldenberg, 1888: 464; cf. Arnold, 1897: 292). Oldenberg is of the opinion that all these cases originally had sā- 11 Apparently Pāṇini defends here quite generally an older reading which survived but in a number of cases. Moreover, Pāņini's concern for metre contrasts with the unconcern for the same found in the RgvedaPrātiśākhya; see Oldenberg, 1888: 372-73n; Müller, 1891: lxxixf.
(ü) P. 6.1.115: nāntaḥpādam avyapare 12 (samhitāyām (72), ekaḥ pūrvaparayoḥ (84), pūrvaḥ (107), enah padāntād ati (109)] "In a Samhitā [text], when e or o which are final in a word precede, [and] when a which is not [itself] followed by v or y, follows, (then) the preceding (sound is not the single (substitute) of both the preceding and the following (sound], when these sounds occur] in the interior of a Pada."
P. 6.1.116: avyādavadyādavakramuravratayamavantvavasyuşu ca [samhitāyām (72), ekaḥ purvaparayoḥ (84), pūrvaḥ (107), enah padāntād ati (109), nāntahpādam (115)] "In a Samhita [text], when e or o which are final in a word, precede, [and] when a follows which is the initial sound) in [one of the following words:) avyāt, avadyāt, avakramuḥ, avrata, ayam, avantu, avasyu, (then) the preceding (sound is) not the single (substitute] of both the preceding and the following (sound), when [these sounds occur] in the interior of a Pāda.”
P. 6.1.116 is not in agreement with the facts of our Rgveda. There are at least two places where ayam has been joined with a preceding e or 0, viz. RV 1.108.6 vrnāno'yam and RV 5.30.3 vahate'yam. Nowhere does ayam behave in the prescribed manner. avasyu is joined with a preceding o in RV 8.21.1 bharanto'vas yavah. And avantu is always joined with a preceding e oro (RPT 2.40(144); Böhtlingk, 1887: 298). The precise prescription contained in P. 6.1.116 makes it very difficult to believe, with Thieme (1935: 51), that this rule does not imply strict application”. Indeed, there is reason to believe that sutras 6.1.115 and 116 were forerunners of certain sūtras from the Rgveda-Prātiśākhya, and, like those, did imply strict application; see below, $ 2.2.
(iii) Pāņini seems to consider the sandhi form ayta- for asta-correct, which agrees with the original Rgveda, but not with the Rgveda known to us. This has been explained in § 1.2, above.

Page #7

RGVEDA AND PĀŅINI
89
2.1.1. It must still be shown that the sūtras 6.1.134 and 6.1.115-116 really are about the Veda. In the case of P. 6.1.134 there can be no doubt. The preceding rule contains the word chandasi "in Sacred Literature". The Kāšikā illustrates the rule with the help of the two examples from the Rgveda which were reproduced above (and adds that some think that the rule is not confined to Vedic verse alone (pädagrahaṇenātra slokapādasyapi grahaṇam kecid icchanti; this would justify a verse subsequently quoted in the Kāšikā)). It seems that wherever in the Aṣṭadhyayi the word pada is used to specify a context, it refers to feet of Vedic verse. The remaining places are: P. 3.2.66: here chandasi is understood from rule 63; P. 8.3.9: ṛksu is understood from the preceding rule; P. 6.1.115 and 8.3.103: here yajusi "in a sacrificial formula in prose" occurs in a following rule (P. 6.1.117 and 8.3.104 respectively), suggesting that the verse-feet (pāda) talked about in the earlier rules likewise belong to sacrificial formulas, and therefore to Vedic verse; P. 8.1.6, finally, deals with a phenomenon which is only found in Vedic verse (see the Käsikä on this rule).
2.1.2.
P. 8.3.17, which justifies the sandhi form ay+a- for asta-, occurs in the company of P. 8.3.18 and 19, which mention Šakaṭāyana and Sakalya respectively (see p. 85, above). These two authorities are mentioned in the Ṛgveda-Prātisakhya, and their opinions may be considered to apply also to the Rgveda, if not primarily to that work. It is therefore safe to say the same of P. 8.3.17.
2.2. The above seems to show that Panini worked with a version of the Ṛgveda which is earlier than the versions described in the Rgveda-Prätiśäkhya. The only serious objection which one might raise, as far as I can see, is that Pāņini's version is not earlier, but quite simply different from the ones of the Prātisakhya. And indeed, we have no guarantee that the Rgveda-Prätiśākhya describes all the versions of the Ṛgveda which existed in its time. The fact that we obtain opinions of the authorities mentioned in the Prātisakhya from sources other than the Prätisakhya shows that the information provided by the Prätišäkhya is in no way complete.
There is, nonetheless, reason to think that Panini did not draw upon an altogether different version of the Rgveda. To begin with, Panini mentions Šakalya on four occasions (p. 87, above) and also knows of the Sakalas, or so it seems (P. 4.3.128). Perhaps more important is that his rules 6.1.115-116 (which we discussed in § 2.1, above) seem to be an earlier version of some rules of the RgvedaPrātiśākhya, 13 This I shall show now.
P. 6.1.115-116 specify the circumstances in which e and o retain their original form before a. The Reveda-Prätisakhya adopts the opposite procedure: it specifies the circumstances when e and o merge with a. In spite of this difference, there is a remarkable similarity.

Page #8

90
JOHANNES BRONKHORST
RPr 2.35(139) reads: antaḥpadam akarac cet samhitāyām laghor laghu yakarādy akṣaram param vakārādy api vā bhavet "In the interior of a Pada, if, in the Samhita [text], a light syllable beginning with y or even v follows a light vowel a, [this a becomes one with the preceding e or o]". This means the same as P. 6.1.115, and more. In addition it contains a restriction on that rule. According to P. 6.1.115, e and o merge with a following a, when that a is followed by v or y. According to RPr 2.35(139), e and o merge with a following a, when that a is followed by v or y, and is a light vowel, and when moreover the syllable beginning with v or y is light.
The advantage of the formulation in the Prātisakhya is clear. Of the seven exceptions which Panini had to enumerate in rule P. 6.1.116, six are excluded by the added restriction of the Prätisakhya. But a price had to be paid. Twenty exceptions are enumerated in the immediately following sutras of the Rgveda-. 14 Prätisakhya. This means that the complicated qualification which we find in RPr 2.35(139) does not in any way simplify the description of the subject-matter. The formulation of the Prātisakhya can most easily be accounted for by taking it as an improvement upon an earlier formulation, the one found in the Aṣṭadhyayi or one closely similar to it.
2.3. I shall now enumerate a few more circumstances which seem to fit our conclusion that Panini preceded the Rgveda-Prātiśākhya and made use of an earlier version of the Rgveda.
2.3.1. Pāṇini's grammar does not know the retroflex consonant !. Our Ṛgveda contains this sound, but we know that not all versions had it (Bronkhorst, forthcoming). The introduction of was "doubtless a dialectical anticipation of the more general identical process in MidIA" (Allen, 1962: 54) and may have taken place rather late. This is supported by the fact that / occupies the place of d where our Rgveda would otherwise have had d between two vowels, not where the original Rgveda would otherwise have had d between two vowels (Wackernagel, 1896: 255-56). E.g., vidv-anga was originally pronounced viduv-anga, but contains nonetheless no 1. One way of explaining the absence of in the Aṣṭadhyay is that Pānini lived before this sound made its appearance in the Veda, and therefore before the Rgveda-Prätiśäkhya 15 (If Panini lived after the sound / had found entrance into the Sakala version of the Ṛgveda, it would be hard to account for the absence of from the Aṣṭadhyayi by saying that this sound was not used in the language of the region where Panini lived (Lüders, 1923: 301-02). Panini knew the Sakalas (above, p. 89) and therefore probably also the peculiarities of their version of the Rgveda. If these peculiarities included in Panini's time, this sound would, and should, have been mentioned in the Aṣṭadhyayi, irrespective of the presence or absence of the sound in Pāṇini's own dialect.)

Page #9

RGVEDA AND PĀŅINI
91
2.3.2. Vowels with circumflex accent are described as follows in the Astādhyāyi: P.1.2.31: samāhāraḥ svaritaḥ [ac (27)] "A vowel which is a mixture (of an udātta and an anudātta vowel] is svarita". P.1.2.32: tasyādita udāttam ardhahrasvam "Of that (svarita vowel] half (the length of] a short (vowel, starting) from the beginning, is udātta."
There has been some discussion why. this description is included in the Astādhyāyi (Thieme, 1957; Cardona, 1968), which does not concern us here. We note the difference from the Rgveda-Prātiśākhya, 16 which has the following sūtras: RP 3.4(189-90): tasyodāttatarodāttād ardhamātrārdham eva vā "Of that (svarita accent 17] half a mātrā or even half of the svarita accent) is higher than the udātta (accent)”. RPr 3.5(191): anudattah parah sesaḥ sa udāttaśrutih "The following remainder (of the svarita accent] is anudatta; it sounds like udātta."
RPT 3.6(192) further specifies that this description is not valid when a syllable follows which has an udātta or svarita accent. The commentator Uvaţa explains that in such cases the latter part of the svarita accent becomes really udātta (p. 114: yadi tūdāttam svaritam vā param syāt tadanudāttah parah sesah syāt). The description of the Rgveda-Prātiśākhya makes the impression of being more sophisticated than the description of the Astādhyāyi. This may be due to the fact that the former is of later date than the latter.18
3.1. We see that there are good reasons to think that our hypothesis is correct. The orthoepic diaskeuasis of the Rgveda took place over a rather long period of time, and was not yet fully completed when the Rgveda-Prātiśākhya was composed (better perhaps: reached its present form).
The investigation has further provided us with some chronological information, most important among which is, no doubt, that Pāṇini's Astādhyāyi is older than the Rgveda-Prātiśākhya. We also saw that Sakalya, who in Yaska's Nirukta was no more than the composer of the Padapātha, had become the redactor of the Samhita in Patanjali's Mahābhāsya. Since in the Anuvākānukramani he is said to have seen the Veda (Bronkhorst, forthcoming) and apparently has reached his apex, it is reasonable to think that these three works have this chronological order: the Nirukta preceded the Mahabhasya, which in its turn preceded the Anuvākānukramani.
There have been attempts to discover the rules employed in the orthoepic diaskeuasis of the Rgveda (Hejib-Sharma, 1979; Ghatage, 1948: 18). Such rules may be discoverable in some cases, but the complicated history of the process, in which many people participated while representing different views, makes it unlikely that all phonetic peculiarities of our Rgveda fall under rules.
It must, finally, be pointed out that the lack of agreement between the Astādhyāyi and our Rgveda may henceforth have to be looked at through different

Page #10

92
JOHANNES BRONKHORST
eyes. Certainly where phonetic questions are concerned, Pāṇini may describe an earlier form of the Rgveda, and may not deserve to be blamed for being lacunary, as he is, e.g., by Renou (1960: 27).
3.2. It remains to say a few words about the difference that may have existed between schools that were concerned primarily with the Rgveda Samhita and those that were concerned primarily with the ritual. Karl Hoffmann (1974) has argued - on the basis of P. 7.2.69: sanim sasanivāmsam, which is found in Mänava Srauta Sūtra 1.3.4.2 and Vārāha Srauta Sūtra 1.3.5.16 - that Pāṇinì lived in the older Sūtra period, 19 i.e., after the Mänava Srauta Sūtra and the Väräha Srauta Sutra: I am not sure if Hoffmann's arguments are compelling, for (as Hoffmann himself observes, pp. 75–76) the words sanim sasanivāmsam occur in a cited mantra, which may be older than these two Sūtras. Be this as it may, Hoffmann's hypothesis places Pāṇini in a time when differences of opinion regarding the ritual had given rise to different schools (Renou, 1947: 25-26). This means that we may have to distinguish between simultaneously existing schools connected with the supposedly correct form of the Rgveda Samhitā, and such as owe their existence to particular views on the ritual. Schools belonging to these two groups may, but by no means have to, coincide.
We know the names of at least three schools that were concerned with the form of the Rgveda: Sakala, Saisirīya (see Bronkhorst, forthcoming), Bāşkala. Schools of the second type, which were primarily concerned with the ritual, may have been the Āśvalāyana and Sankhāyana schools (Renou, 1947: 25f). R. G. Bhandarkar (1893). has argued that these two schools belonged to both the Śākala and the Başkala Śākhā. This point of view is confirmed by the commentator Gārgya Nārāyaṇa on Asvalāyana Śrauta Sūtra 1.1.1 (p. 1); see also his comments on Aśvalāyana Grhya Sūtra 3.5.9 (pp. 167-68). Some other evidence tends to ascribe both the Āśvalāyana and the Sānkhāyana school to the Bāskala sākhā (Renou, 1947: 25, and esp. Aithal, 1969: 187-89).
It is interesting to observe that the unification of Śakhās which we noticed with respect to the form of the Samhita, has its counterpart in the tendency to rejoin which is found in the ritual schools of the Rgveda (Renou, 1947: 46; cf. Surya Kanta, 1933: 9-11, 66).
Kern Institute, Leiden
NOTES
* This article came into existence as a result of discussions which I had with Prof. S. D. Joshi. At a later stage I could avail of valuable suggestions made by Prof. M. Witzel. I wish to express my gratitude to both these scholars.

Page #11

RGVEDA AND PANINI
1 The Ṛgveda-Prātisākhya does not enable us to decide whether "Anyatareya" or "Anyatareya" is the correct name. The commentator on Caturädhyāyikä 3.74, however, cites the opinion of one Anyatareya. See Whitney, 1862: 174.
Many such passages are given in Mīmāmsaka, 1973: I: 69-71, and elsewhere in the same book, to be found with the help of the index (Mīmāmsaka, 1973: III: 111-50). 3 See note 12, below.
4
93
Ghatage's (1948) attempts to prove that the passages concerned must be read -o+a-, with short Ŏ, show at best that this was "an intermediate stage of abhinihita sandhi", as he himself seems to admit (p. 18).
5 "Vyali", which is found in the Rgveda-Pratiśäkhya, is a sakalization of "Vyaḍi". See Bronkhorst, forthcoming.
6 Explicitly said by Bhartṛhari, Mahābhāṣyadīpikā p. 23, 1. 19. Vyaḍi and the Samgraha are both mentioned in Patanjali's Mahābhāṣya, possibly with the understanding that the former was the author of the latter; see Scharfe, 1977: 125.
7 As far as I can see, no difficulties would arise if P. 6.1.113 and 6.1.87 but then also P. 6.1.109 and 6.1.78 would be taken into the Tripādī, in this order (after 8.2.66 and before 8.3.19, of course). If this is correct, the riddles surrounding P. 6.1.113 intensify and depend for their solution exclusively on the second answer.
Interestingly, Patanjali has no respect for the makers of Padapathas (padakāra), for he says that they must follow grammar (lakṣana), rather than vice versa: na lakṣanena padakārā anuvartyaḥ/padakarair nama lakṣaṇam anuvartyam/ yathalakṣanam padam kartavyam// (vol. II, p. 85, 11. 4-5; vol. III, p. 117, 11. 18-19; p. 398, 11. 8-10). We may recall that also Yaska did not hesitate to disagree with Sakalya's Padapatha (Nirukta 6.28).
9 This advice has been followed by the Taittiriyas with regard to borrowed mantras (Renou, 1947: 33n). According to Bhartṛhari (Mahabhāṣyadīpikā p. 1, 1. 7) the Taittiriyas read even the word agni with n. This probably refers to Taittiriya Brahmana 3.5.6 (borrowed from RV 6.16.34): agnir vṛtrani janghanat. This line has no n in agnir in our version of that text, but Jayantabhatta records that it sometimes has (Nyāyamañjarī vol. I, p. 685).
10 Oldenberg (1888: 386) thinks that these redactors preceded the Padapatha. Since he gives no real arguments, we can ignore his opinion.
11 Oldenberg later (1907: 834-35) changed his view, on the basis of the later language. This, of course, is a weak argument. Pāṇini's rule is evidence that Oldenberg's earlier opinion was the correct one.
12 This is the reading found in Patanjali's Mahābhāṣya. The Käsika has: prakṛtyantaḥpādam avyapare. The Bhāṣya-reading seems to be older, for, although Patanjali is acquainted with the reading prakṛtyā, Katyāyana's vārttikas show no sign of such an acquaintance. See Thieme, 1935: 47-48. The word prakṛtya may have been borrowed from RPr 2.51 (155), which defines the meaning of pragṛhya.
13
Already Renou (1957: 120, n. 580) pointed at the similarity between P. 6.1.115f and RPr 2.35(139) f.
14 Sandhi with preceding e or o takes place in avartraḥ, avyatyai, ayopäṣṭihḥ, avantu, aviratā, avatvacaḥ, avirate, avamsi, avaḥ (RPr 2.40(144)). Further exceptions: agne 'yam (RPr 2.42 (146)); yavase 'visyan, vṛtrahatye'viḥ (RPr 2.43 (147)); tavase 'vaci, vahate'yam, januṣo'ya (RPr 2.44 (148)); viso 'yanta, santo 'vadyani, bharanto 'vasyavaḥ (RPr 2.45 (149)); te 'vardhanta (RPr 2.46(150)); te'vindan (RPr 2.47(151)).
15
That the Padapatha contains 1, may be explained by the process of sakalization, which also affected the Ṛgveda-Pratiśäkhya (Bronkhorst, forthcoming).
16 The Aṣṭadhyayi differs in this respect from the other Prätisakhyas as well. See Whitney's (1862: 164-69) description of the svarita in the Prātisakhyas.
17
The terms udatta, anudatta and svarita apply to vowels in the Aṣṭādhyāyi, to accents in the Rgveda-Prätisakhya (Cardona, 1968: 455).

Page #12

94
JOHANNES BRONKHORST
18 Cardona (1968: 459) thinks that the description of svarita in the Astādhyayi was only meant for svarita vowels occurring in the Astādhyāyi. This seems unlikely. 19 Of course, we must be careful not to revert to the belief that there was a clear Brāhmaṇa period followed by a clear Sūtra period; see Renou, 1947: 36, Gonda, 1975: 22.
REFERENCES
Aitareya Aranyaka. Edited by Arthur Berriedale Keith. Oxford: Clarendon Press. 1909. Aithal, K. Parameswara: 1969, “RV-Khilas and the Sūtra-s of Āśvalāyana." ALB 33, 182-94. Allen, W. Sidney: 1962, Sandhi. 's-Gravenhage: Mouton. Arnold, Edward Vernon: 1897, "Sketch of the Historical Grammar of the Rig and Atharva
Vedas." JAOS 18, 203–353. Arnold, Edward Vernon: 1905, Vedic Metre. Cambridge: Cambridge University Press. Āśvalāyana Grhya Sūtra. Edited, with the commentary of Gärgya Nārāyana, by Rāmanārāyana
Vidyaratna and Anandachandra Vedāntavāgisa. Calcutta: Baptist Mission Press. 1869. Āśvalāyana Srauta Sūtra. Edited, with the commentary of Gārgya Nārāyana, by K. V. S. R.
Gokhale, Poona: Ānandasrama. 1917. Bhandarkar, R. G.: 1893, "The Relations between the Sütras of Āśvalāyana and Sankhāyana and
the Sākala and Bashkala Sākhas of the Riksam hita." Transactions of the Ninth International
Congress of Orientalists. Vol. I, pp. 411-20. Kraus Reprint. Nendeln/Liechtenstein. 1968. Bharthari: Mahâbhâsyadīpika. Edited by K. V. Abhyankar and V. P. Limaye. Poona:
Bhandarkar Oriental Research Institute. 1970. Böhtlingk. Otto: 1887, Panini's Grammatik. Hildesheim: Georg Olms Verlagsbuchhandlung.
1964. Bronkhorst, Johannes: 1980, "Asiddha in the Astādhyāyi: A Misunderstanding among the
traditional Commentators?" JIP 8, 69-85. Bronkhorst, Johannes: forthcoming. "The Rgveda-Prātiśākhya and its Sākhā." To appear in SII. Buiskool, H. E.: 1939, The Tripādi. Leiden: E. J. Brill. Cardona, George: 1968, "Panini's Definition, Description and Use of svarita." Pratidanam.
Indian, Iranian and Indo-European studies presented to Franciscus Bernardus Jacobus Kuiper on his sixtieth birthday. Edited by J. C. Heesterman, G. H. Schokker, V.I.
Subramoniam. The Hague-Paris: Mouton, pp. 448-61. Gärgya Nārāyana. See Aśvalāyana Grhya Sūtra and Āśvalāyana Srauta Sutra. Ghatage, A. M.: 1948, "Traces of Short and in Rgveda." ABORI 29, 1-20. Gonda, Jan: 1965, Vedic Literature (Samhitās and Brahmanas). Vol. I, Fasc. I of A History of
Indian Literature. Wiesbaden: Otto Harrassowitz. Hejib, Alaka, and Sharma, Arvind: 1979, "The Formulation of a Rule concerning the
Cerebralization of a dentals in external Sandhi in the Rgveda." IL 40, 49-50. Hoffmann, Karl: 1974, “Pāṇini VII 2, 69 sanim sasanivāmsam.” Münchener Studien zur
Sprachwissenschaft 32, 73-80. Jayantabhatta. Nyäyamanjari. Vol. I. Edited by K. S. Varadacharya. Mysore: Oriental Research
Institute. 1969. Kanta, Surya: Editor. 1933, Rktantram. Delhi: Meherchand Lachhmandas. 1970. Kashikar, C. G.: 1951, "The Problem of the Galantas in the Rgveda-Padapatha." PAIOC 13
(1946), 39-46. Lüders, Heinrich: 1923. "Zur Geschichte des l im Altindischen." Festschrift Wackernagel.
Göttingen: Vandenhoeck & Ruprecht, pp. 294-308. Mimāmsaka, Yudhisthira: 1973, Samskrta Vyakarana-Sastra ka Itihasa. Parts I-III. Sonipat:
Rāma Lāl Kapür Trust. Samvat 2030. Mishra, Vidhata: 1972, A Critical Study of Sanskrit Phonetics. Varanasi: Chowkhamba Sanskrit
Series Office.

Page #13
RGVEDA AND PANINI 95 Muller, F. Max: 1891, Vedic Hymns. Part I. Oxford: Clarendon Press. Oldenberg, Hermann: 1888, Die Hymnen des Rigveda. Band I. Metrische und textgeschichtliche Prolegomena. Berlin: Wilhelm Hertz. Oldenberg, Hermann: 1907, "Vedische Untersuchungen." ZDMG 61, 803-36. Oldenberg, Hermann: 1912, Rgveda. Textkritische und exegetische Noten. Siebentes bis Zehntes Buch. Berlin: Weidmannsche Buchhandlung. Patanjali: Vyakarana-Mahabhasya. Edited by F. Kielhorn. Third Edition by K. V. Abhyankar. 3 volumes. Poona: Bhandarkar Oriental Research Institute. 1962-72.. Renou, Louis: 1947, Les Ecoles Vediques et la Formation du Veda. Paris: Imprimerie Nationale. Renou, Louis: 1957, Introduction generale. Nouvelle edition du texte paru en 1896. au tome I. In: Jakob Wackernagel. Altindische Grammatik. Gottingen: Vandenhoeck & Ruprecht. Renou, Louis: 1960, "La Forme et l'Arrangement interne des Pratisakhya." JA 248,1-40. Rgveda-Pratisakhya. 1. Edited, with Uvata's commentary, by Mangal Deva Shastri. Vol. II. Allahabad: The Indian Press. 1931. 2. Edited and translated by Max Muller. Leipzig: F. A. Brockhaus. 1869. Scharfe, Hartmut: 1977, Grammatical Literature. Vol. V, Fasc. 2 (pp. 77-216) of A History of Indian Literature, edited by Jan Gonda. Wiesbaden: Otto Harrassowitz. Thieme, Paul: 1935, Panini and the Veda. Allahabad: Globe Press. Thieme, Paul: 1957, "Panini and the Pronunciation of Sanskrit." Studies presented to Joshua Whatmough on his sixtieth Birthday, pp. 263-70. Reprinted: Kleine Schriften. Teil 2. * Wiesbaden: Franz Steiner, pp. 612-19. Uvata: See Rgveda-Pratisakhya. Vamana-Jayaditya: Kasika. Edited by Aryendra Sharma, Khanderao Deshpande and D. G. Padhye. 2 parts. Hyderabad: Sanskrit Academy, Osmania University. 1969-70. Wackernagel, Jakob: 1896, Altindische Grammatik. I. Lautlehre. Gottingen: Vandenhoeck und Ruprecht. Whitney, William Dwight: 1862, The Atharva-Veda Pratisakhya or Saunakiya Caturadhyayika. Text, translation and notes. Varanasi: Chowkhamba Sanskrit Series Office. 1962. Whitney, William Dwight: 1888, Sanskrit Grammar. Second Edition. Delhi: Motilal * Banarsidass. 1962.