Van buitenen

A Contribution to the Critical Edition of the Bhagavadgītā

Author(s): J. A. B. van Buitenen

Source: Journal of the American Oriental Society , Jan. - Mar., 1965, Vol. 85, No. 1

(Jan. - Mar., 1965), pp. 99-109

Published by: American Oriental Society

Stable URL: https://www.jstor.org/stable/597711

A CONTRIBUTION TO THE CRITICAL EDITION OF THE BHAGAVADGĪTĀ

1. A NOTE ON BHAGAVADGĪTĀ 1.10. J. A. B. VAN BUITENEN UNIVERSITY OF CHICAGO

IN THE FIRST CHAPTER OF THE GITA occurs a stanza which has given the Indian commentators more trouble than modern interpreters have found with it. The stanza is spoken by Duryodhana:

aparyāptam tad asmakam balam bhiṣmābhi- rakṣitam/
paryaptam tv idam eteṣām balam bhimabhi- rakṣitam //

Modern translators take (a-)paryaptam in the sense of “(un-) abundant, (in-) sufficient,” and differ 1 cf. PW I, 651 paryāpta “erfüllt, zum Abschluss gebracht, das volle Maass habend, ein hohes Maass er- reichend, geräumig, voll, hinreichend, genügend.”

only in whether they take (a-) paryāptam or bhi(s)mabhirakṣitam as predicate to balam. Se- nart renders it as follows: “Limitée en nombre, c’est en Bhisma que notre armée a sauvegarde; leur armée à eux, sous la sauvegarde de Bhima, est immense.” Edgerton modifies the literal sense of this stanza by parenthetical additions, but his version is essentially the same as Senart’s: (Altho) insufficient (in number), this our host is protected by (the wise) Bhiṣma; on the other hand, (while) sufficient, this their host is protected by (the unskilled) Bhima." However, Duryodhana’s army was not at all in- sufficient in number; and though Western scholars may tend to disregard the monstrous numbers of 11 akṣauhinis on the Kaurava side and 7 on the Pāṇḍava side, Indian commentators were sharply aware of this fact. This accounts for much dis- agreement. Śrīdhara takes aparyāptam to mean aparimitam. Nilakantha, more fancifully, has paryāptam parita aptam parivestitam sur- rounded," and he notes: “The Pandava army is in fact small, as it is limited to 7 akṣauhinis; so it can be surrounded by his own army, which is large, consisting of 11 akṣauhinis, but his own army cannot be surrounded by the others, thus is the sense. So theirs is paryāptam, sc. defeatable by him.” 5 6 66

The oldest commentator, Sankara, does not com- ment on this stanza, a significant fact to which we shall revert. His commentator finds himself forced to construe the sloka as three sentences: on the basis of Anandagiri’s interpretation we should split the sloka thus: aparyāptam tad (sc. eteṣām balam); asmākam balam bhiṣmabhirakṣitam pary- aptam tv idam; eteṣām balam bhimabhirakṣitam. The meaning of paryāptam is then pareṣām pari- bhave samartham. As a second choice he gives for aparyāptam the usual aparimitam adhṛṣyam akso- bhyam. Rāmānuja does not here comment word by word, but has the following summary:

7

2 Emile Senart, La Bhagavad-gîtâ (Paris, 19442), p. 1. 3 Franklin Edgerton, Bhagavad Gita (Cambridge, Mass., 1946), I, p. 5; cf. also his note p. 180.

Sridhara, Srimad Bhagavadgită (ed. K. S. Agāše) Poona (AAS) 1901.

5 Sriman Mahabharatam vol. III, with Bharata-bhāva- dipa by Nilakantha (Poona, 1931), part VI Bhisma- parvan, p. 43.

Srimadbhagavadgita, Anandagiri-krtatikāsamvalita- Sānkara-bhāṣya-samavetā, ed. Kāśīnāth/Apṭe, AAS 34

(Bombay, 1936), p. 11-12.

7

Srimadbhagavadgitā, Vedāntācārya-Šrī-Venkaṭanatha-

“Duryodhana, observing that the army of the Pandavas is protected by Bhima and his own by Bhisma, and having conveyed to his teacher that the other army is sufficient to defeat himself but his own not sufficient to defeat theirs, was down- cast in his heart.” This is precisely the opposite meaning of Anandagiri’s. Commenting on Rāmā- nuja, Vedantadeśika offers no less than four pages of commentary on this interpretation-proof, if proof were needed, how this seemingly transparent stanza exercized the minds of eminent Indian sanskritists.

Vedāntadesika introduces an objector, who states precisely what is the difficulty with Ramanuja’s and all modern translations: “Duryodhana does not consider his army inadequate, as it is protected by such warriors as Bhisma and Drona: for the even- tual death of these warriors is at this point still conditional.” Also, the Pandavas are outnumbered 11 to 7, and, he remarks, the very same Duryo- dhana will tomorrow say: aparyāptam tad asmā- kam balam pārthabhirakṣitam / paryāptam tv idam eteṣām balam parthivasattamāḥ: here, as he correctly explains, we have to construe asmākam aparyāptam “not equal to us,” and not asmakam balam “our army (is insufficient.)” for clearly

C

our army" is not protected by Partha. He goes on to suggest that our text is corrupt: Bhima and Bhişma have been transposed. Restore them to their right places and “then 1. we have proper agreement in the construction their army, pro- tected by Bhima, is not equal to us,’ 2. the preposi- tion tad has its proper sense as remote-demon- strative, 3. there is no conflict between Duryo- dhana’s meaning here and in the other passage.” “9 Although Vedantadeśika brushes the argument probably right on all three counts. To take his away on unconvincing grounds, the objector is point about the pronouns first: if paryāptam and aparyāptam, asmākam and eteṣām, bhima and bhiṣma are contrastive terms, so are tad and idam. kṛta-Tātparyacandrikākhyaṭīkāsamvalita-Srimad - Rāmā- nujācārya-viracitabhāṣyasahita, ed. M. R. sankara and V. G. Apte, AAS 92 (Bombay, 1923), p. 34-35. 8o. c., p. 29 ff. 9o. c., p. 30: tatra bhiṣmabhimasabdayor viparyäsät paṭhabhedaḥ / tada ca bhiṣmābhirakṣitam tad balam asmākam aparyaptam ity anvaye sāmānādhikaranyam tad iti viprakṛṣṭanirdeśasvārasyam duryodhanābhiprāyā- virodhaś ca sidhyati. The same objector also puts for- ward Anandagiri’s interpretation: vākyabhede tv evan yojana, etc. (ib.).

In these two antithetically parallel sentences we either must ignore the contrast between tad and idam, or question the contrast between bhima and bhisma. The original presence in the stanza of tad and idam is assured by the metre. Now, a confusion of the meanings of idam “this” and tad “that” (which all translators in fact assume) 10 goes against the genius of the language; but a confusion of Bhima and Bhisma, neither metrically anchored, is a minimal scribal error. Was it a scribal error? This bring up the objector’s other point. Elsewhere in the epic, cr. ed. 6.47.6, Duryodhana makes a statement which, in the objector’s reading makes perfect sense as to context, and also gives the obvious meaning to tad and idam.11 aparyāptam tad asmakam balam bhiṣmābhi- rakṣitam/ paryāptam tv idam eteṣām pārthivasattamāḥ // With this relevant critical apparatus:

— b/ KD2 balam pārthivarakṣitam; K,D, (marg. sec. m. as in text) G1.3 balam pārthābh°. Da,Ds om. 6cd. c/ K, aparyāptam idam teṣām; D3.4 paryāptam idam ed/ KB Da2Dn D4-7 T1 G, balam bhiṣmābhiraksitam; K, parthivasattama. From the correspondences between 6.47.2-30 and BhG. 1.2-19 it is at once clear that the two text portions resemble each other very closely. Of the 59 half-slokas of 6.47 no less than 19 correspond wholly or in part with 19 of the 36 half-slokas of the Gītā. Although this correspondence in princi- ple raises the question whether the Bhagavadgītā- developing within a Bharata version, and posing and resolving a problem of kinship morality which in effect presupposes the actual killings in the family war-borrowed its preamble from the be- ginning of the Second Day or whether the latter was borrowed from the preamble to the First 10 so ‘yam ghanṭāpathāt pataccarakuṭirapraveśaḥ “this is like a robber entering a hovel by way of the belfry,” that is to say, creating and solving a problem that does not exist, ib. 30. 11 Edgerton’s “this our host” for tad asmakam balam, and “this their host” for idam eteṣām balam, actually translates as though tad and idam were transposed. I doubt strongly whether English this their host could be rendered with Skt. “idam.” 101 Day, 12 it is profitless to address ourselves to this question here: for our purposes it suffices to note that the two texts are parallel, that the two stanzas are parallel, and that the parallelism offers a basis for interpretation. As can be judged from the quoted apparatus, the reading of 47.6d is less than firm. The competing reading is, remarkably, bhiṣmābhirakṣitam, which of course could not be accepted if the same com- pound is accepted in 47.66. There is no better reading for b. On the other hand we will, like the editor, be hesitant to accept as original a super- fluous vocative parthivasattamaḥ, if other MSS. suggest the possibility of a corrupted reading be- hind the elsewhere repeated bhiṣmābhirakṣitam. We also note that some MSS. dropped the second half of the śloka, either by accident, or because it made no sense on account of the repeated bhiṣmā- bhirakṣitam. Still other MSS. preferred in b. another reading. It is significant that the uncertainty of interpre- tation of 1.10, where the reading seems to be firm, is repeated in the parallel 6.47.6, where the reading is very uncertain. The reading accepted by the Editor makes no sense in the context. Duryodhana is giving his warlords a pep talk; he is supposed to be “heartening, gladdening, encouraging” (harsa- yan),13 and a statement that his own forces are inadequate, despite the fact that they are numeri- cally superior and moreover directed by a great warlord, Bhisma, makes no sense whatever. Al- though the reference of Vedantadeśika’s objector has no authoritative standing, his general argu- ment remains valid, and we have excellent evidence to uphold it. In the critical edition of the Bhagavadgītā the remarkable fact emerges that, while the MSS. of the Sarada Kashmir tradition are generally the most authoritative for the earliest text of the epic, including the Bhismaparvan where the Gītā is found, nevertheless they are found to be late and secondary as far as the text of the Gītā is con- cerned.1 14 This means that in Kashmir the text of “this 12 S. K. Belvalkar, Mahabharata, cr. ed., The Bhisma- parvan (Poona, 1947), p. 788 notes on 6.47 chapter… may well belong to the pre-Gītā stage in the evolution of the epic.” 13 consequently the men of Dhṛtaraṣṭra “rejoice ex- citedly”: 6. 47. 21-22. 14 I refer to the discussion of Belvalkar, o. c., Intr. civ ff. on the evidence for the Bhismaparvan as a whole compared with ib. lxxviii ff. for that of the BhG.

102

the epic was sooner fixed than elsewhere, but the text of the Gītā later. The earliest testimony con- cerning the Gītā is that of Sankara, and the Gītā text adopted by the Editor is really Sankara’s text with but 14 highly insignificant variants. It should, however, be noted that Sankara does not comment on the first 57 stanzas of the Gita, which include the entire first chapter. Sankara therefore holds no authority for the present read- ing. His commentator Anandagiri cannot be quoted as evidence: the uncommented portion in Sankara’s Gītā-if these stanzas were originally included at all-was as wide open to adaptation and normalization as any other uncommented Gītā manuscript. The Editor has failed to acknowledge what then must be the most ancient testimony for this part of the text; which is curious, since the authority was known to him.15 It is Bhaskara, the Vedantin, who gives a scholastic word-by-word explanation of the text missing in Sankara. Now Bhaskara’s text has in our stanza another reading, which is the oldest available and which eliminates all the problems found in our line. He reads: aparyāptam tad asmakam balam bhima- bhiraksitam/ paryāptam tv idam eteṣām balam bhīṣmā- bhirakṣitam // And he comments: asmākam yathanirdiṣṭasamha- tānām tat pāṇḍavabalam aparyāptam asamartham abhibhavitum yato bhīmābhirakṣitam bhimasenena pradhanabhutena guptam sa ca bhiṣmāpekṣaya / kimcin nyūnatara ity abhiprayaḥ / idam tv asma- kam balam eteṣām pāṇḍavānām paryāptam abhi- bhāvāya-vināśane samartham ity arthaḥ-yato bhīṣmābhirakṣitam / bhismas tv anekasastrastra- parago nirjitānekamahāsamara ity arthaḥ // There is no evidence that Bhāskara has tampered with his text; we may note in this connection that one Sārada MS. bears his reading out.16 May we prefer his reading? I think we must: 1. it has the authority of the most ancient authority; 2. it removes the inexplicable anomaly of tad and idam; 3. it does justice to the universally accepted supe- riority of Bhisma to Bhima as a warlord; 4. it fits the tongue of Duryodhana addressing his columns, who are superior in numbers, on the first day of 15 o. c., Intr. lxxvii. 16 MS siglum $3. battle; 5. it explains the recurrence of the same line in what is evidently an exhortatory address by Duryodhana in 6. 47. Finally, it helps to explain the variety of read- ings in 6.47.6. We can now more clearly see why parthivasattamāḥ is uncertain: compared to the repeated bhiṣmābhirakṣitam in 6d of many MSS., it represents an early emendation by an easy voca- tive to meet an anomaly which other MSS. solved by dropping 6cd entirely. The reading bhism in 6b necessitated taking aparyāptam in the artificial sense of “aparimitam” (cf. Nilakantha ad loc.), which other scribes could not understand, so that in 6b bhiṣmabhirakṣitam was changed to parthi- varakṣitam (possibly under the influence of parthi- vasattamaḥ in 6d), which in turn may have given rise to pārthābhirakṣitam, though this may have been an independent emendation of bhism. The original source of all later difficulties was of course a proleptic dittography of bhisma for bhima° in 66. This original bhīmā° is still evidenced for the Gita by Bhaskara and one Śārada MS., until in the Gītā “vulgate” transmission too the two com- pounds were transposed. This may have been influenced by the very general bhismā° reading in 6.47.66; or a deliberate change to account for tasya samjanayan harṣam in 12,17 or sheer coinci- dence of repeated dittography. By Ramanuja’s time the reading was definitely established. On this showing we must conclude that Bhā- skara’s reading is the correct one, and that the translation of the stanza should read: “That army, guarded by Bhima, is not equal to us; on the other hand this army, guarded by Bhisma, is equal to them.” 2. THE TEXT OF BHAGAVADGĪTĀ CH. 1. Since Bhaskara is the oldest authority for those portions of the Gīta which Śañkara left uncom- mented upon, his evidence surely demands more 17 apparently understood as tasya, sc. Duryodhanasya, which presumes that D. was less than happy before. Bh. comments: tasya prakrantasya balasya Duryodhanasya vā, which at least offers a reasonable alternative: it is quite intelligible that after Duryodhana had ordered his troops to defend their mainstay Bhisma the latter wishes to inspire confidence by blowing his horn. It is likely that the BhG. has here condensed the corresponding por- tion of 6. 47 which in 21-22 reads: tatas te tavakāḥ sarve hṛṣṭa yuddhaya Bharata / dadhmuḥ sankhān mudā yuktāḥ simhanādāmś ca nādayan // teṣām śrutvā tu hṛṣṭānam kuruvṛddhaḥ pitāmahaḥ / simhanādam vina- dyoccaiḥ sankham dadhmau pratāpavān.

“18

consideration than it has received. Belvalkar, the editor of the Bhagavadgītā makes mention of him, but has made no study of what on even superficial enquiry reveals itself to be the earliest testimony to part of his text: “Bhaskara’s variants are re- ported in our Critical Apparatus on the authority of T. R. Chintamani,” “who has given some details and extracts.” This is an error of method. The relevant portion, principally comprising the first chapter, may be negligible as far as doctrine is con- cerned, a mere preamble, and therefore usually rushed through or forgotten (as Sankara did); but for the purpose of textual criticism it is of great importance. If it can be shown that a number of Bhaskara’s readings are distinctly superior to those of the Vulgate (V), there is at least room for the presumption that variants elsewhere are also sig- nificant. Incidentally, it must be pointed out that for this portion we cannot really speak of a “Vul- gate” before Rāmānuja. We can have no certainty about the readings Sankara had in that portion of the text which he skipped through. In judging the quality of Bh.’s readings we have additional support: for we have a good part of Gītā ch. 1 in the parallel text of MBh. 6.47. In effect, 6.47 is our most ancient, though indirect, authority for Gītā ch. 1, whether it preceded the Gītā in time (as is likely) or not. We find that in a number of instances Bh.’s text is closer to 6.47 than V is. Finally, on the basis of the principle of chrono- logical priority, as adopted by the critical edition. in the case of Sankara’s text, we may assume the greater correctness of Bh.’s text there where there is little to choose between two variant readings. * * * 1.1. sarvakṣatrasamāgame for samaveta yuyutsa- vah. Bh. has the advantage of greater stylistic effect: -kṣetre -kṣetre -kṣatra-. 1.7. nāyakān for nayakā. V demands a tortuous construction, but has the advantage of being a lectio difficilior. We must assume an omitted (ye) nāyakā. 1.8. kṛpaḥ salyo jayadrathaḥ for krpaś ca sami- tim jayaḥ; a jayadrathaḥ in this stanza is probably original, though it is hard to make out where it occurred. Nilakantha reads a jayadrathaḥ in 8d for the meaningless tathaiva ca. 1.8d. saumadattiś ca viryavan for saumadattis 18 o. c., Intr. lxxvii. 103 tathaiva ca. Bh. is no doubt preferable, with the typical strengthening of the last item in a series by means of an adjective; but considering the proba- bility that a jayadrathaḥ occurred in the stanza, not Bh. but Nilakantha may have preserved the more authentic reading. 1.9a. anye’pi for anye ca. 1.9d. Bh. nānāyuddhaviśāradāḥ presents an in- teresting case. V here has sarve yuddhaviśaradāḥ. 6.47.4d in a parallel stanza has a variety of read- ings apparently stemming from an original sar- vaśasträstravedinaḥ, with reasonable support for sarve instead of sarva°. If we accept that 1.9 is a calque of 6.47.4 we surmise an original nānāyu- dhaviśāradaḥ behind nanayuddha, with a variant sarvayudha. Possibly to avoid repitiousness after 9c or by mistake, nanayudha was changed to nānāyuddha, sarva/sarve-yuddha°. 1.10. For Bh.’s undoubtedly correct reading, cf. supra. 1.11a. tu for ca. 1.14a. Bh. appears to read svete for śvetair, a likely error. 1.18ab. Bh.’s pāñcālaś ca maheśvāso draupadeyāś ca pañca ye is completely parallel to 6.47.18 (where v. 1. pañcalyaś) and distinctly superior to V with the stanza-fillers sarvaśaḥ pṛthivipate. 1.19d. vyanunādayat for V vyanunādayan. V here has an obvious lectio facilior for the augmentless imperfect; the latter Bh. reading has however been preserved in 6.47.29. 1.21a/24a. Bh. ubhayoḥ senayor for senayor ubhayor. 1.23. Bh. omits this stanza, which is in fact a repetition of stanza 1.22. 1.27ab. Bh. lists this as a separate half-stanza. Probably this half-stanza was an additional 26ef. The stanza division in V. is anomalous: 27cd ob- viously goes with 28 ab, after which arjuna uvāca, and 28cd with 29ab. Bh. then reads another supernumerary stanza, 29cd-30a-d. We may as- sume the following original division: A. 26-27 ab; B. cd.-28ab; C. 28cd-29ab; D. 29cd-30; this gives the best sense. دو 1.28. Bh. sidamano ‘bravid idam for V viṣidann idam abravit; V looks like an “improvement,’ meant to bring out not merely that Arjuna sat down, but sat down despairing.

104

1.28. Bh. svajanān yuyutsun samavasthitan for svajanam… yuyutsum samu pasthitam. 1.29. Bh. sramsate ganḍīvam for V. gānḍīvam sramsate. Bh. is stylistically superior, esp. in the correct stanza division: there we have the sequence romaharṣaś ca jāyate/ sramsate gänḍīvam. 1.33. Bh. ta eva naḥ sthita yoddhum prānāms tyaktva sudustyajan (“they stand arrayed to fight us”) for V’s facile ta ime ‘vasthita yuddhe. Bel- valkar’s argument for dhanani ca in d fails to convince. O 1.43cd. utsadyate jatidharmahḥ kuladharmaś ca śāśvataḥ for utsadyante dharmaḥ dharmaś ca śāśvatāḥ. O 1.47c. utsṛjya for visṛjya. * .

It is clear from the above comparison that Bh.’s Gītā readings which for this portion of the text are the oldest available are in a number of in- stances preferable to those of “V.” This being the case, Bh.’s readings elsewhere deserve close atten- tion. As will become clear, Bhaskara, writing within a century of Sankara and knowing Śañ- kara’s text, had before him another text which he used. Since his commentary frequently engages in polemics with Sankara, we may presume that the quality of Bhaskara’s readings were not in ques- tion: Bhaskara does not simply have recourse to another text in order to combat Sankara’s interpre- tation; for his contemporaries it could not be an issue which text he used. This should prove that very close to Sankara’s date there existed two vari- ant texts that were equally authoritative. There- fore the question of the “critical edition” of the Bhagavadgită can again be raised legitimately. Edgerton 20 and after him Belvalkar 21 have argued that on the contrary in the majority of significant cases the Vulgate is superior. In the critical edition of the Mahabharata the singular paradox came to light that while for the epic in its entirety the Kashmir recension was on the whole superior to other versions, this is not true for the episode of the Bhagavadgītā. This seemed merely due to the fact that we have a more copious and older apparatus for the Gita than for the entire epic. On the assumption of the existence of one single Vulgate against which the variations of the Kashmir version could be measured, a Vul- gate version moreover which is the oldest known text, Edgerton could come to the belief “that the variant readings of the Kashmirian text are with- out exception late and secondary, and have no bear- ing on the determination of the oldest form of the Gītā.” 22 With the evidence of Bhaskara this judg- ment will now have to be revised. Instead of the assumed relation O→ V→ K, we must now accept the more complicated relation: V(S) 0 Bh ↓ K This relation is required by the fact that Bh. had, besides Sankara’s Vulgate, another text which must be considered a Vorlage of what now survives as the Kashmir version. It is, however, not identical with it, an important fact not recognized by Belvalkar. Belvalkar’s treatment of Bhaskara’s evidence has been inconsistent and inadequate. He writes: “the commentator Jayatirtha, apud 6.7b, records the ingenious manner in which the scholiast Bhaskara (whom, it seems, even Abhinavagupta has men- 3. BHASKARA AND THE SO-CALLED KASHMIR tioned with reverence) changed the traditional text VERSION. Before we can address ourselves to the question of how far Bhaskara’s readings contribute to our knowledge of the “original” text of the Gītā, we have to raise the prior question of his relationship to the Kashmir version. Schrader has the distinc- tion of having brought this version to light.19 But his argument that this version contains a number of preferable readings has not convinced scholars. 19 F. Otto Schrader, The Kashmir Recension of the Bhagavadgita (Stuttgart, 1930). of a stanza from the BG. with a view to get from it pering with the original is actually recorded in the a more suitable sense. Where such “pious” tam- case of one noteworthy Kashmirian writer, one of the other “Kashmirian” variants in the Bhaga- would be justified in viewing with suspicion most vadgita portion of the Bhismaparvan, which seems 20 Review of Schrader’s o. c. in JAOS 52 (1932), p. 68 ff. 21” The so-called Kashmir Recension of the Bhaga- vadgītā,” NIA II, 4 (1939-40), p. 211-251. 22 The Bhagavad Gita, Intr. xiii.

105 to be obviously due to some partisan’s over-zeal at Abhinavagupta as well as Rāmakantha “must have simplification.” This argument is entirely circular as well as aprioristic. It supposes that Jayatirtha was right, which supposes that his text was right and that Bhaskara did not have a text equally right. On this presupposition, which should have been in- vestigated, to base the sweeping generalization that other Kashmirian writers would have tampered likewise and that therefore Jayatirtha’s text was right in the first place closes a circle unworthy of the learned Editor. There is in the first place the question of the identity of the Bhaskara mentioned by Jayatirtha and the Bhaṭṭabhaskara listed by Abhinavagupta. Jayatirtha reacts to Bhaskara’s reading in 6.7, parātmasu samā matiḥ for V. paramātmā samā- hitaḥ, which he calls “fabricated” (kṛtrimaḥ pathaḥ). Jayatirtha knew Bhaskara’s commen- tary, and the Bhaskara whose commentary he knew is the bhedabhedavadin author of our present Bhā- skara MSS. This Bhaskara has a considerable number of readings in common with the Kash- mirian version, which has in 6.7 the very reading to which Jayatirtha objected. Of the Bhaṭṭa- Of the Bhatta- bhaskara mentioned by Abhinavagupta we know only that he had commented on the Gita, but unless we assume that there were two Bhaskaras who commented on the Gita we may safely con- clude that Jayatirtha’s Bhaskara and Abhinava- gupta’s were the same person. Unfortunately we lack the final evidence of Bhaskara’s comment on 6.7, which is lost. The fact that Bhaskara was known to Abhinava- gupta does not make him a Kashmirian. The use of the qualification “Kashmir version” is mislead- ing, if it is taken to mean a version known only in Kashmir and nowhere else. We do not know whether Bhaskara was a Kashmirian. We do know that he had a text different from, sometimes supe- rior to, the Vulgate of Sankara and prior to, and consistently superior to, a version of the Gītā now known from Kashmir sources. We do not even know whether Abhinavagupta himself had first- hand knowledge of Bhaskara’s commentary. His statement that “previous commentators, Bhat- ṭabhaskara, etc., have studied the text in great detail” would on the face of it make it likely that he knew Bhaskara’s commentary. But this is diffi- cult to rhyme with Schrader’s assertion that 23 Bhismaparvan, Intr. lxxxii f. been completely ignorant of what is now the Vul- gate text of the Bhagavadgītā, for Bh. has a num- ber of “Vulgate " readings. If anything, Schra- der’s plausible conclusion that “the Vulgate of the Bhagavadgītā was still unknown in Kashmir by the end of the tenth century” would place Bhaskara well out of Kashmir. For the consequence is that Sankara’s bhāṣya was also not known in Kashmir, and Bhaskara knew that very well indeed.24 Bhaskara knew, and thus authenticates, three works of Sankara, the Brahmasūtrabhāṣya, the Gītābhāṣya and the Upadeśasahasrī. While he is very familiar with the two bhāṣyas, which he attacks time and time again, he is not proven to be familiar with any other work in the Sankara tradi- tion, a fact pointed out by Hacker 25 and very relevant for his date. Elsewhere I have raised the question whether Sankara and Bhaskara were not contemporaries-there is one bit of evidence to suggest that 26 in which case his home would be likely to be South-India. If this evidence be con- sidered insufficient, we can show at least that Bhaskara was known to Vacaspatimiśra, who de- votes several pages to the refutation of his philoso- phy in the Bhamati on 1.1.4 and elsewhere. This would place him, if we accept the traditional dates, at least within a century of Sankara. I believe that the conclusion is unavoidable that in the ninth century there existed a text of the Bhagavadgītā which had equal authority with that used by Sankara; that it existed outside Kashmir; and that it is the prototype of the so-called Kash- mir version. The consequence of this conclusion is that the Kashmir version is late and secondary, not to the Vulgate, but to Bhaskara’s text. A further consequence is that as far as critical criteria are concerned, the Bhagavadgītā portion is 24 I do not believe that we should take references to predecessors mentioned in passing too seriously: Bhās- kara may have been known traditionally as an important commentator and Abhinavagupta’s reference might well be a courtesy to an author he had not read or whose text he had not available. 25 Paul Hacker, Vivarta (Ak. Wiss. u. Lit., Mainz, Abh. G. u. S.-w. Kl. 1953, 5; Wiesbaden 1953). 66 26 See my “Relative dates of Sankara and Bhaskara,” Adyar Bulletin, 1962; B. N. Krishnamurti Sarma in ‘Bhaskara-a Forgotten Commentator on the Gītā,” IHQ IX (1933), p. 663 f., also suggests that s. and Bh. were contemporaries, but I find his belief that Bh. is attacked by S. in his Gītābhāṣya and that Bh. rejoined in some additions to his own bhāṣya hard to follow.

106

Between them the manuscripts cover the follow- not really an exception within the Mahabhārata tradition: that for the Bhagavadgītā too the King portions of the text of the Bhagavadgītā: 1. tradition carries on a text, however deteriorated here and there, that was authentic and of which we have the earliest record in Bhaskara’s bhāṣya. 4. BHASKARA’S GĪTĀ TEXT. The commentary on the Bhagavadgītā by Bhā- skara, entitled in full Bhagavadanuśayānusaraṇa, survives incomplete in two partly overlapping manuscripts, one from a London Library, the other now deposited in the Sarasvati Bhavan Library in Benares.27 The London manuscript is written in Kashmiri script, the Benares one in devanagari. 2. lc. 3a. 56. 6d. 10d. 11. sīdamānam ma klaibyam gaccha kaunteya śreyas cartum te naḥ sthitaḥ sīdamānam tvam mānuṣenopahatāntarātmā viṣādamohābhibhavād visamjñaḥ / kṛpāgṛhītas samavekṣya bandhun abhiprapannan mukham antakasya // vināsino nānuśocitum 18c. 26d. 296. tathaivam 30d. 40a. 4lb. ekaiva nānuśocitum nehätikrama, but V is quoted as variant 43d. gatam (?) 50a. jahātīme 52ab. yada tvam mohakalilam (buddhyā?) vyatitarisyasi 54ab. lacuna in text and commentary 55- 1-47; 2.1-57ab; 3.3-43; 4.1-42; 5.1-29; 6.1-2 and 26-47; 7.1-30; 8.1-28; 9.1-33. Up to and including 3.31 the Gītā stanzas are written out in full; thereafter by pratika with the exception of (the supernumerary) 3.40-42; and of 5.29 and 6.29. In most cases the literalness of the commentary permits us to recover the reading of Bhaskara’s text even though it is only given pratikena. I list hereunder the variants of Bhaskara’s text, following the numbering of V, omitting the vari- ants of ch. 1 (cf. supra). visidantam klsibyam mã sma gamaḥ Pärtha śreyo bhoktum te ‘vasthitaḥ visidantam (omitted; in Bh. the numbering now differs) anāśino nainam socitum tathaiva na tvam socitum nehābhikrama° ekeha °gatim jahätiha yada te mohakalilam buddhir vyatitarisyati (here an interpolation has intruded upon the text: aiśvaryasya sa(ma)grasya dharmasya (ya)sasaḥ śriyaḥ / vairagyasyatha mokṣasya sannām bhaga iti dhvaniḥ //) 57cd 3.3 (lacuna) vākṛteneha vartamy eva; but V is quoted as variant 3. 12a. kāmān 186. 22d. 23a. yadi tv aham na varteya 23c. anuvarteran 276. bhāgaśaḥ 31b. anuvartanti 31d. te vimucyanti 326. nanuvartanti 35d. 38. bhavaty eṣa katham śrnu (from 32 onward stanza is quoted by pratika) paradharmodayad api katham caiva vivardhate / 27 The London MS. was discovered by Dr. V. Raghavan while searching for his monumental Catalogus Catalo- bhogan näkṛteneha varta eva yadi hy aham na varteyam anuvartante sarvaśaḥ anutisthanti mucyante te’pi nānutiṣṭhanti para-dharmo bhayavahaḥ gorum; a preliminary edition of both MSS is under preparation by Dr. Subhadra Jha.

kimätmakaḥ kimācāras tan mamācakṣva prcchataḥ // (this stanza is quoted by pratika but not commenced upon) śrībhagavan uvāca esa sūkṣmaḥ paraḥ satrur dehinam indriyaiḥ saha / sukhatantra ivāsīno mohayan pārtha tisthati // sa eva kaluṣaḥ kṣudras chidraprekṣi dhananjayaḥ / rajaḥpravṛtto mohātmā mānuṣāṇām upadravaḥ // kāmakrodhamayoomayo ghoras stambhaharṣasamudbhavaḥ/ ahamkarābhimānātmā dustaraḥ papakarmabhiḥ // harṣam asya nivartyaisa śokam asya dadāti ca / bhayam casya karoty eşa mohayams tu muhur muhuḥ // (These four stanzas are quoted in full with the concluding comment: iti samkhyamatānusarinaḥ kecanādhiyante / te ca pañca slokā na vyākhyātāḥ) vā paramparakhyātam sanātanaḥ mām evam yo (?) yajñāyārabhataḥ naiḥśreyasakarāv yad eva samkhyāḥ pasyanti yogais tad anugamyate bhuñjan… svasan svapan 39d. 4. 2a. 3b. 146. 23c. 5. 26. 5ab. 8d. 21. 236. 24a. antaḥsukho 24c. sa pārtha paramam yogam 6. 28a. evam yuñjan 28c. brahmasamsleṣam 28d. atyanta (?) 29c. pasyate 37. (omitted?) vimocanāt ca paramparaprāptam purātanaḥ iti mām yo yajñāyācarataḥ niḥśreyasakarāv yat samkhyaiḥ präpyate sthānam tad yogair api gamyate aśnan… svapan svasan vimokṣaṇāt yo’ntaḥsukho sa yogi brahmanirvāṇam yuñjann evam brahmasamsparsam atyantam ikṣate (adds, but fails to comment on supernumerary half-stanza) anekacittovibhrānto mohasyeva vasam gataḥ śāradabhram (hypermetrical) (?) 38a. kaścid (?) 386. 39a. etam me 40d. tāto (but probably tāta) 41b. jāyate dhimatām kule 436. purvadaihikam 43c. tato yatati bhūyo’pi 45b. avaso ‘pi san (?) 7. 16. mām āśritaḥ / madāśritaḥ lla. 186. me mataḥ 206. 23c. 28a. antam gatam balavatam caham āśritya additional half-stanza, uncommented. kaccid chinnabhram etan me tāta kule bhavati dhimatām paurvadehikam yatate ca tato bhūyo avaso ‘pi saḥ madāśrayaḥ balavatam asmi me matam ästhāya antagatam

107108 8. tyaktvā (?) °vṛttinā 56. 8b. 12d. о dhāraṇam 206.

‘yam vyakto ‘vyaktat, or ’nyo ‘vyakto ‘vyaktāt

5. THE TWO TEXTS COMPARED.

As will be clear from the above list of variants, differences between Bh. and V are matters of detail; Edgerton’s remark about the K version, that “differences are relatively very slight and rarely affect the essential meaning of even single stanzas, never of the work as a whole,” applies fully to Bhaskara’s text as well. Its interest lies principally in the help it gives to clarify the crucial relationship between K and V. “28 In a number of cases Bh. seems to me to be superior to V. 2.1c Bh. sīdamānam was more likely to be changed to the more emphatic visidantam than vice-versa; cf. 1.28 3a. Bh. mā with imperative not countenanced by Pāņini; which might possibly account for the Pāṇinean ma gamaḥ of V? 5b. V sreyo bhoktum may be a lectio facilior for Bh. śreyas cartum. 6d. Bh. te naḥ sthitaḥ pramukhe, cf. 1.33; V te ‘vasthitaḥ is more easily explained out of te naḥ sthitaḥ than vice-versa; V continues to presuppose a naḥ or asmākam (Edgerton: “they are arrayed in front of us”). 3.22d Bh. prefers the less regular active vartāmy eva to varta eva which he also knows. Similar cases: 31b. anuvartanti; ib. d vimucyanti; 326 anuvartanti; 43c. yatati these irregularities are normalized in V. 35d. Although one hesitates to change this famous stanza, Bh. sequence “death in one’s own duty is more salutary than prosperity through the duty proper to someone else” makes better sense of śreyaḥ; runs nicely parallel to ab. śreyan sva- dharmo… paradharmat; and does away with the lame ending paradharmo bhayavahaḥ. The close resemblance of paradharmodayad api to para- dharmobhayavahah renders it likely that a corrup- tion of Bh. underlies the variant of V. 39d. Bh. va has to be taken in the obsolescent sense of iva, and thus makes better sense than V ca, which might quite well be a “correction” of a misunderstood va “or.” Bh. translates “knowl- muktvā °gāminā °dhāraṇām ’nyo ‘vyakto vyaktāt edge is obfuscated by this perennial foe, in the form of Desire, which is like an insatiable fire” instead of V (Edgerton) “by this constant foe in the form of Desire and an insatiable fire.” 4.23c. arabh with karman “undertaking, initiat- ing acts” seems slightly preferable as lectio diff. to V acar “performing acts” 5.2b. naiḥśreyasa, less common than V niḥśreyasa. 5.5ab. V appears to be a reinterpretation of Bh.: that which the followers of the reason-method (Edgerton) see is followed by the application of yoga discipline.” This would suggest that “Sam- khya” presents a view, which is thereupon acted upon through yoga; the two are therefore comple- mentary and thus “one” (ekam). V suggests that both s. and y. as such, independently of each other, lead to the same goal (sthānam), and are therefore identical from the point of view of their result. V commits us to taking yogaiḥ in the sense of yogibhiḥ, for which I cannot find other attestations except 12.289.2-3 and 298.8. 5.24a. V yo’ntaḥsukho looks like a facilitation of antaḥsukho. yaḥ, but it is a border case. 24c. It is easier to explain V sa yogi brahmanir- vānam as a reinterpretation of V. paramam yogam, inspired by the labhante brahmanirvanam of 25a, than to treat paramam yogam as a gloss of brah- manirvāņam. The vocative Partha where V has yogi need not make us suspicious; just in this didactic situation where great emphasis is required the alliterative vocative makes fine sense. 6.43c Bh. tato yatati bhuyo ‘pi, normalized in V yatate ca tato bhuyo. 6. CONCLUSION. I have refrained from entering into too great detail in comparing V and Bh.-apart from the first chapter where this detail was required by Bhaskara’s seniority-, because I see little profit in arguing for one reading over another in the numerous cases where there is little to choose. We shall never recover an “original” Bhagavad-

gītā on the basis of the available textual evidence, including Bhaskara’s. The comparison of Bhaskara’s variants with Sankara’s textus receptus is, however, enlightening in so far as it shows how relatively minor the variations still were in a fairly late period, at least five centuries after the final redaction of the text. Clearly, the text of the Cita was stabilized quite early and apparently sooner than that of the large mass of the great epic. This stands to reason. The very fact that Sankara felt impelled to com- ment on the Gītā, a text far from congenial to his central doctrines, should sufficiently show in how high an esteem the Gītā was held as a quasi-philo- sophical, moralistic and religious discourse. Such an esteem supposes separate commentatorial-and therefore editorial-treatment. Already Sañkara refers to conflicting interpretations of evident predecessors (2.11, 21; 3.1; 4.18, 24; 18.6) and such references presuppose a commentatorial tradi- tion antedating Sankara by at least several cen- turies. Thus Sankara himself must have found an already firmly established text, close to a Vul- gate, so well known that he did not feel the need to comment on the beginning portions. If Bhās- kara, who quite obviously knew Sankara’s text, thought nothing of using variant readings, this demonstrates that this was mostly a matter of piety of traditional affiliation with one school of com- mentators rather than another and that the Gītā interpretation was no longer a matter of textual 109 criticism: the variants were compatible and herme- neutically irrelevant. Therefore I strongly doubt if the remaining portions of his bhāṣya, if they ever come to light, would change the picture offered by what is extant. The significance of Bhaskara’s text thus is that it proves the existence of a gen- erally accepted Gītā text well before both him and Sankara. The variations between Bhaskara’s text and the “Kashmirian” manuscripts seem to me to indicate that in Kashmir the Bhagavadgītā was not so stabilized; in other words, that a firmly established separate existence of the text, in the form of one or more authoritative commentaries, was unable to check the natural deterioration of the readings until far later. This would well agree with the relatively late date when distant Kashmir begins to participate in “Indian” philosophy. But in spite of the deterioration, natural under the circumstances, changes took place on the basis of a sound text which we find earlier represented in the transmission out of which Bhaskara wrote. The “Kashmirian” variations, when compared with Bhaskara’s text which is so far older than our oldest mss. and must itself rest on a yet older com- mentatorial transmission, illustrate by their com- parative insignificance the excellence of the preser- vation of the Mahābhārata; and even though we are unable to recover the authentic original of the redacted epic, we are once more assured that the Kashmir transmission brings us closest.