Critical Studies in the Mahabharata (V. S. Sukthankar Memorial Edition, Vol. I) c7.s3: IV. More Text-Critical Notes
Critical Studies in the Mahābhārata (V. S. Sukthankar Memorial Edition, Vol. I) c7.s3
- Language
- English
IV. More Text-Critical Notes
c7.s3.1
The Mahābhārata Textual Criticism is unquestionably making progress: slow but steady progress. There are many things in it still that are obscure, but some things have undoubtedly been cleared up. Thus there seems to be consensus of opinion among scholars now on the following points.
c7.s3.2
The Mahābhārata textual tradition, as we know it, is far from being quite uniform. There are now two main streams of tradition, the Northern Recension and the Southern Recension. Each recension is further sub-divided into a number of provincial versions, which differ *inter se* in many particulars. But the text of the Mahābhārata must have been in a fluid state for a very long time, almost from the beginning. Clearly therefore a wholly satisfactory restoration of the text to its pristine form—even the late so-called *śatasāhasrī saṃhitā* form—would be a task now beyond the powers of criticism. All that we can attempt now is to reconstruct the oldest form of the text that is possible to reach on the basis of the MS. material available. The peculiar conditions of the transmission of the epic necessitate an eclectic but cautious utilization of all MS. classes. The Kaśmīrī version (*textus simpli-cior*) has proved itself so far to be unquestionably the best Northern version; and the Malayālam, which in many respects is superior to the Grantha, is likewise the best Southern version (*textus ornatior*). The agreement between the Kaśmīrī version and the Southern recension (or sometimes even merely the Malayālam version) can be taken as an indication of originality. But contamination between the different versions was inevitable and must be admitted. The Telugu MSS. are generally and the Grantha MSS. frequently contaminated from Northern sources. Even the Malayālam version, which is on the whole free from Northern influence, may show some contamination, in unexpected places. [91] In fact, all versions, with the possible exception of the Kaśmīrī, are contaminated in various degrees. It is, therefore, often a very ticklish question to decide which agreements are original and which secondary. Notwithstanding these difficulties, experience has now shown, the case is not as desperate as it might at first appear. For one thing, there is a considerable bulk of text where the Northern and Southern versions are in full agreement, where there are no variants at all, or—more frequently—only unimportant variants: this part of the text is fortunately certain. There is then the question of the "additional" passages, that is, passages found in only one of the rival recensions. There is only one rational way of dealing with these "additional" passages: they must be carefully segregated from the rest of the text, and examined individually. The onus of proving the originality of these "additional" passages will naturally rest on him who alleges the originality: the documents speak naturally against them, but their evidence is not by any means conclusive.
c7.s3.3
These are some of the principal findings of the Prolegomena. It is assuring to find that these conclusions have been restated and endorsed emphatically by so cautious and judicious a critic as Prof. M. WINTERNITZ in his recent review of the Ādiparvan volume, in the pages of these *Annals* (Vol. 15, pp. 159-175). The outlines of the reconstruction may, therefore, be taken to be correct, and the method of reconstruction sound. There is bound to be difference of opinion as regards details. When there are hundreds of thousands of readings to be considered and weighed, it is natural that all the selections would not satisfy all readers; and there are bound to be small slips in so enormous and difficult a work as this. But the reader has the advantage of having the full critical apparatus before him, prepared with all possible care and presented in a convenient manner. The reader may easily substitute in the text any reading that appeals to him better. Prof. WINTERNITZ has thus shown his preferences, in the article mentioned above, in a certain number of cases, where he differs from me as regards the choice of readings. They are passages, as he tells us, which he came across in reading parts of the Critical Edition with his pupils in the Indologisches Seminar at Prag. He has thus published these criticisms after much thought and discussion with [92] other scholars. I gladly take this opportunity, therefore, to present my view of the case, setting forth the reasons which have guided me in the choice of the readings adopted by me in the Critical Text.
c7.s3.4
1. 3. 60 : *girā vā śaṃsāmi*.
c7.s3.5
I have indeed assumed that agreement between K (strictly speaking, between the original Kaśmīrī version, or at least the Śāradā MS. Ś₁) and S is a sufficient though not a necessary proof of originality. But there is no agreement here between K and S. K₀, it is true, represents the version K in a comparatively pure form (Prolegomena, p. L.); but K₀ is not K; and K₁ is, on the whole, decidedly a better representative of the Kaśmīrī version than K₀ (Prolegomena, p. XLIX). Now here K₀ and K₁ have different readings: K₀ agrees with S, and K₁ with the rest of N! This fact has been ignored by WINTERNITZ. As for the agreement of Ñ₁ and S, I have pointed out in the Prolegomena (p. LIX), that "even the manuscripts of distant Nepal are not wholly free from contamination from some Southern source or sources (direct or indirect)." Thus the agreement between Ñ₁ and S cannot be considered as compelling evidence, by any means. Moreover, it should be borne in mind that this agreement between K₀ Ñ₁ and S concerns merely the *omission of one syllable*; and it is clear that this trifling omission could quite easily take place *utterly independently* in the respective groups. Therefore, even the documental probability in favour of the reading preferred by WINTERNITZ is not at all strong. It is, in fact, considerably weakened by the following (intrinsic) consideration. N has *vā* and a dodeka (hypermetric) line; S omits *vā* and has the eleven-syllable (normal) line. This is a circumstance suspicious in itself. It is by no means certain or necessary that all the pādas of our Triṣṭubhs should be of the eleven-syllable pattern. Some of these hypermetric pādas can indeed be made normal by the omission of one or more of the additional syllables, often merely of an expletive. But there are many lines which defy such athetization; for instance, the second pāda of the very *next* stanza (1. 3. 61; *nāśatyadasrau sunasau vaijayantau*). Such lines ought to [93] warn us against giving hasty preference to these normalizing readings. I have cited elsewhere (Prolegomena, p. XCIII) clear instances of efforts made by redactors to correct hypermetric pādas of Ślokas. Similar efforts are to be found among the variants of our Triṣṭubhs. But as the scheme of the Triṣṭubh is more variable and arbitrary, or at least more complicated and obscure, it is difficult to prove the alterations made by the different scribes or editors. In this particular instance, however, it seems to me, there is no reason to assume that the original pāda was *not* hypermetric, because our pāda is a standard hypermeter—a type of which HOPKINS has cited numerous instances. Cf. No. 6 of the typical, oft-recurring varieties (mentioned by HOPKINS in *The Great Epic*, p. 275):
c7.s3.6
◡ ◡ ◡ ◡ ◡ | ◡ ◡ ◡ ◡ ◡ ◡ .
c7.s3.7
The scansion of our line is :
c7.s3.8
◡ ◡ ◡ ◡ ◡ | ◡ ◡ ◡ ◡ ◡ ,
c7.s3.9
which, it will be seen, is a pāda of the identical pattern.
c7.s3.10
From the examples given by HOPKINS, I will cite here only the following :
c7.s3.11
3. 13. 193 *mamaiva (tau) vāmyau parigrahya rājan*
5. 42. 9 *tatrānu (te) yānti na taranti mṛtyum*
5. 48. 77 *vegena (iva) śailam abhihatya Jambhaḥ*
13. 94. 13 *na (hy) utsahe draṣṭum iha jīvalokam*
Hari. 2, 72. 44 *vi-añjano jano ('tha) vidvān samagraḥ*
c7.s3.12
Are we to omit the bracketed syllables—or some similar syllable—in each case, on the ground that they disturb the metre, if not the sense? We do not possess yet the collations of all these passages, but I am confident that there will always be some MSS. if not versions (or even a whole recension), which omit these extra syllables, for one reason or other. There is this other consideration. The ancients had as good an ear for the rhythm of their Triṣṭubhs as we have, if not better. Why and how was the offending syllable first inserted, and then tolerated by generations of editors? All N MSS. except D₂ contain the hypermeter. Is it not, under these circumstances, more probable that the *original* was a hypermetric line, which was emended by some redactors in conformity with the later ideas of the regular Upajāti metres? [94] This particular hypermeter is moreover antique, for we find (as HOPKINS has pointed out) already an example of it in Muṇḍ. Up. 3. 1. 6 :
c7.s3.13
*yatra (tat) satyasya paramam nidhānam,*
c7.s3.14
with the scansion ◡ ◡ ◡ ◡ ◡ | ◡ ◡ ◡ ◡ ◡ , which differs from our line as regards the quantity of the first two syllables only; ours has an iambic opening, the Muṇḍaka line trochaic. Here, also, we can with impunity omit the bracketed syllable (*tat*), and get a line which is metrically a “better” line, but obviously not the original one. I am, therefore, fully persuaded that in all such cases we have original hypermeters. These old stanzas were not built at all on the pattern of the later, more regular, classical metres, but followed some other finer rhythmic principle, which escapes our methods of rough analysis by syllable-counting.—Now as regards the sense. The exact explanation of the phrase *girā vā śaṃsāmi* is, I admit, difficult. Nīlakaṇṭha offers the facile explanation: *vāśabdaś cārthe* (i.e. *vā* = *ca*), which can hardly be considered satisfactory. Devabodha analyses the phrase into *girāv* + *ā* = *śaṃsāmi* (with *girau*, loc. of *giri*), but that hangs together with his interpretation of the Aśvins as the Sun and the Moon, which is not very convincing. It is possible to interpret the words as ‘*girā vā* *śaṃsāmi* (i.e. *vai* + *āśaṃsāmi*), as is done by some editors. On the other hand, it is also possible to emend *vā* to *vām* (“ye two,” acc. dual), which would give a satisfactory sense. This reading, which is mentioned as a pāṭha by Arjunamiśra, is found in only one of our inferior MSS. (D₂). It seemed to me such an obvious emendation that I did not like to adopt it on such slender MS. evidence, and I have hesitatingly (as shown by the wavy line) set in the text the old Northern reading *vā*, which is unquestionably a *lectio difficilior*. The whole hymn is, however, obscure and full of interpretative and other difficulties. It will have to be studied and dealt with more minutely before we can be sure of its text and meaning. But I am fully persuaded that *vā* (or some such word) did form part of the original line, and it would be a mistake to omit it.
c7.s3.15
1. 3. 145 *Kurukṣetre nivasatām*.
c7.s3.16
Why WINTERNITZ should have any doubts about *nivasatām*, I fail to understand. It is merely an augmentless imperfect, and [95] augmentless imperfects are most common in the Mbh., especially in the case of verbs with prepositions, as in this case. Even WHITNEY (*Sanskrit Grammar*, § 587d) remarks: “Besides the augmentless aorist-forms with *mā* prohibitive, there are also found occasionally in the later language augmentless imperfect-forms (very rarely aorist-forms), which have the same value as if they were augmented, and are for the most part examples of metrical license. They are especially frequent in the epics¹ (whence some scores of them are quotable).” To this, in 1884 (that is now more than fifty years ago), HOLTZMANN added the note:
c7.s3.17
¹ Italics mine.
c7.s3.18
“Beispiele von fehlendem Augmente des Imperfects sind ziemlich häufig, besonders in Compositum und hier wieder in solchen, die mit *auf* a auslautenden Präpositionen, *ava*, *upa*, *apa*, zusammengesetzt sind : aber nicht ganz vereinzelt ist der Mangel nur bei *bhavan* sie wurden 3, 110, 3 = 9970. 11, 27, 14 = 813. 16, 1, 9 = 9 u. a.”
c7.s3.19
He has given the following examples: *prcchat*, *patan*, *cintayan*; *man-yadhvam*, *budhyetām*; *vyapagacchatām*; *adhyavasyan*, *avabudhyata*, *avatiṣṭhata*, *avekṣetām*, *avapadyanta*, *paryavasthāpayat*; *abhyupamantrayat*, *upatiṣṭhetām, upanṛtyanta; abhiṣṭuvan, abhidhāvetām, abhyarcayat; samuttiṣṭhan, utthāpayan, uccārayan; paripālayan, nivartayetām, nirjayat; saṃgṛhṇītām; svīkarot. The Critical Ed. has, besides nivasatām, the following: ṛdhyata 1. 58. 8, avabudhyata 120.11, avatiṣṭhat 189.11, avamaṃsthāḥ (no. v. 1. !) 189. 21, samabhidravat 218. 28. There are many others, less certain. A number of instances will be found in the Tentative Edition of the Virāṭaparvan by Mr. UTGIKAR; the Critical Edition of this parvan, I may add, will again contain quite a few instances of this particular solecism. How many examples are really needed before it is admitted that augmentless imperfect forms occur regularly in the Mbh., and we cease to talk of “correct” and “incorrect” forms? The augmentless forms are quite as “correct” as the other as far as the Mbh. is concerned. WINTERNITZ points out that the “correct” form “is given by the Kaśmīrian transcript K₁, by the Maithilī and Bengali MSS., by Arjunamiśra, and by some Southern MSS.,” and therefore he prefers it. It is a wonder that [96] it is not given by more MSS.; because the tendency is always to change an “incorrect” form into the “correct” one. We have here to look not to the MSS. which have the “correct” form, but to those which have not the “correct” form! Let us look at it in another way. If nivasatām be the original reading, then there are two ways of improving the text: (i) keep vasatām of the original and change the word to nyavasatām and thus make it “correct” (Central sub-recension and Malayālam, independently), or (ii) keep ni- intact, and change the word to the present nivasataḥ (“K” version); both are represented here. If, on the other hand, nyavasatām were the original reading, no one would think of changing it later to nivasatām; and nivasataḥ of K would also be quite unintelligible. The text reading nivasatām thus explains, to a certain extent, how the other readings may have arisen and has therefore been preferred. And I think there cannot be much doubt about its correctness and originality.
c7.s3.20
1. 3. 183 prabrūhi vā kiṃ kriyatām dvijendra.
c7.s3.21
If WINTERNITZ prefers me kiṃ, he must, at the same time, also prefer karaṇīyam adya | yenāsi kāryeṇa etc., found in N (except K B₄ D₂.₅); but he did not realize it. The me kiṃ of the Vulgate cannot be joined to kriyatām dvijendra of K B₄ D₂.₅ S. There is no MS. which reads prabrūhi me kiṃ kriyatām dvijendra | śuśrūṣur asmy adya vacas tvadīyam. The reading of the line proposed by WINTERNITZ will be like an animal with the head of a horse and tail of a donkey! The text reading of the entire line—nay, of the entire stanza—occurs verbatim in K₁. 3. 4. I think, the original Northern reading was probably prabrūhi vā kiṃ, as in text. If vā be useless after prabrūhi, it is more than useless after kiṃ, where WINTERNITZ would have it. It is conceivable that the (original) Northern reading (preserved in K₁. 3. 4. is corrupt, and we ought to give here preference to S. That is another matter. Not being convinced that vā kiṃ of K₁. 3. 4 was “secondary” or corrupt from *kim* vā of S, I have adopted, in the text, *vā kim* underlining it to show that it is less than certain. But *mām kim, mām yat, me kim, me tvam, vākyam* of the Northern MSS. are clearly all secondary, and need not be considered at all.
c7.s3.22
[97] 1. 55. 3 *śrotum pātram ca rājams tvam*.
c7.s3.23
I have taken the Śāradā MS. only as the *norm* for my edition; I have not undertaken to reproduce its text *verbatim*. The *śrotṛpātram ca rājams tvam* of Ś₁ K is opposed by all other MSS. And, moreover, as there is agreement between B D and S—more or less independent versions—on the reading *śrotum pātram ca rājams tvam*, this reading is indicated as the original, and has been adopted, without wavy line; a procedure quite in accord with the principles of text reconstruction laid down in the Prolegomena (p. xci). WINTERNITZ finds *śrotṛpātram ca rājams tvām* of K “better Sanskrit,” but I must frankly confess that I could not make head or tail of it, and I cannot understand it even now. The text reading is the same as the reading of the Vulgate and of the Southern recension, and is besides quite clear, and, as far as I can judge, flawless Sanskrit. It may be translated: “And thou art, O King; a fit person to hear (Vyāsa’s Bhāratī Kathā),” a sense which suits the context admirably. Furthermore, K has here an entirely different reading for the whole stanza. If we adopt the reading preferred by WINTERNITZ, then we must read the stanza as:
c7.s3.24
*śrotṛpātram ca rājams tvām prāpyeyam Bhāratī śubhā |*
*guror vaktraparispandam mudā protsāhatīva mām ||,*
c7.s3.25
which is apreciably different from the text. But the variants of the other versions do not at all suggest that this is the *original* reading. Our text represents the reading of the Southern recension, which has here, in my opinion, the greatest probability.
c7.s3.26
1. 56. 8 : *katham vyatikraman dyūte*.
c7.s3.27
It is natural that W. should demand wavy lines for this verse. I admit the line is most puzzling, and I only lighted upon the solution of the puzzle by accident, after prolonged wrestling with it. An explanatory note was really called for, because the first line is absolutely unintelligible unless one knows the particular meaning of the root *vyatikram* intended here, which is “to wrongly submit or surrender oneself to, wrongly take to (a thing or person, acc.)”. It is a rare meaning, but not unknown. It is given in the dictionaries of MONIER WILLIAMS (ed. 1899, s. v. [98] *vyatikram*), and MACDONELL (s. v. *kram*). It appears not to be given in PW, but PW has (s. v. *kram*) : “verkehrter Weise sich einer Sache (Acc.) hingeben”. One example of the use is (B.) 12. 174. 36 :
c7.s3.28
*atha ye buddhim aprāptā vyatikrāntāś ca mūḍhatām*
*te ’tivelam prahṛṣyanti samtāpam upayānti ca ||*
c7.s3.29
Here *vyatikrāntāś ca mūḍhatām* must mean "those who have surrendered themselves (wrongly) to folly"; no other meaning will fit, as far as I can judge. Our stanza is, therefore, to be traslated thus:
c7.s3.30
"How (indeed) did the two sons of Pṛthā, as well as the two sons of Mādrī, (tamely) surrender themselves to the tiger among men (Yudhiṣṭhira), who was being cheated at dice by those wicked men, and (how did they) follow him?"
c7.s3.31
This strange submission consists in their consenting, in the first instance, to be used as pledges or stakes in that dire game of dice, and acquiescing dumbly to everything that Yudhiṣṭhira did. The Southern recension is unanimous on the point that the doubtful word was some form of *vyatikram*; and the eight readings recorded in our crit. app. show that it must have been some word which was as unintelligible to the scribes and editors as it is to WINTERNITZ. WINTERNITZ is again mistaken in thinking that the S reading is *vyatikramadyūte*. This is the reading not of S but of only two MSS. out of the thirteen Southern MSS. of our crit. app. I myself have followed S in reading *vyatikraman dyūte*, which is the reading of *five* of the Southern MSS. In fact, my text represents the reading of the *majority* of the Southern MSS., and is, therefore in a sense, the true S reading. It is needless to add that *vinirjitam* of "K" and *vyasaninam* of the Central sub-recension, are emendations, made by those who could not make head or tail of the original. For it is clear that no one would change *vinirjitam* or *vyasaninam* to any of the Southern readings, while the reverse would be the most natural thing in the world. It is further worthy of note that while the Northern MSS. have the easier reading, there is no agreement between the Kaśmīrī and the Central sub-recension. They must therefore represent independent alterations of a third original.—Following certain other S MSS. P. P. S. SASTRI has adopted the reading [99] *vyatikramam dyūte* (without recording any v. 1.); but I am inclined to think that this is a wrong analysis of *°kramandyūte* which in S MSS. stands for both *°kramam dyūte* and *°kraman dyūte*.—Now that I have explained how the awkward word is to be interpreted, I have no doubt that WINTERNITZ will agree with me that the text reading is correct.
c7.s3.32
1. 57. 20 : *kriyate *ucchrayo nṛpaiḥ*.
c7.s3.33
There has been extraordinary reluctance among scholars to face the fact that the Mbh. text once contained far more examples of hiatus than what one deems right or reasonable for such an ancient and venerable text. But there is hardly any excuse now for such hesitation. On p. xciii of the Prolegomena, I have given numerous instances of the surreptitious efforts of scribes and redactors to eliminate hiatus, which show that hiatus was an anathema to them. It is well known that the Sandhi is not rigidly observed in the Vedic Saṃhitās, the Brāhmaṇas, and the Upaniṣads, in Pali, and in Prakrit works.
c7.s3.34
Why must the Sandhi then be rigidly observed in the popular epic, a work which has its roots firmly embedded in the soil of the heroic poetry of the Sūtas, and which has throughout preserved vestiges of its humble origin? The only reference (so far as I know), in the poem itself, to the Mbh. being a Kāvya (*kṛtaṃ mamedam bhagavan kāvyaṃ paramapūjitam*) has now disappeared with the Brahmā-Gaṇeśa interpolation in the Ādi, and we are thus left only with a Purāṇa, Itihāsa or Ākhyāna. A specimen of this class of work may surely contain sporadic instances of hiatus, without our being scandalized. It is futile, I may add, to try to apply to this category of *popu-lar* literature the rules of the dialect of the high-brow Śiṣṭas. We may expect every conceivable solecism in the language of the Mbh., and we find, in fact, an amazing variety of them; the hiatus, which was extremely common, is only one of such so-called irregularities. But for this strange modern antipathy towards hiatus, WINTERNITZ would have recognized that the emendation is quite legitimate and certain. WINTERNITZ considers the Northern readings but wholly ignores the Southern, which is fatal to his argument. S does not contain either *ati* (or *abhi*) which is found only in certain N MSS., and is besides superfluous. [100] TG have recast the line: *kurvanty ete dhvajocchrayam*, with *kurvanti* (note the active voice!), that is, a form of the root *kṛ* (like *kriyate* of N) at the head of the pāda. All these versions have the verb, it may be noted, in the *beginning* of the pāda. M contains the same three words as our *kriyate*, *ucchrayaḥ* and *nṛpaiḥ*, but transposes *kriyate* and *ucchrayaḥ*, reading *ucchrayaḥ kriyate nṛpaiḥ*. This is the key-reading, which gives a clue to the original. You can explain everything else, but you cannot explain the transposition in M (which normally ought to agree with TG), unless you start from the hypothetical *kriyate *ucchrayo nṛpaiḥ*. If you argue, on the other hand, that the reading of M is an arbitrary variant, which is in no need of a special explanation by means of an emendation, then there is no need of a critical edition either; because all MSS. contain approximately the same text and the same story, and the few differences could be regarded as arbitrary variations which needed no explanation. With this reading, containing the opprobrious hiatus, you can explain *all* the variations. N inserts a futile and innocuous *abhi* (corrupted in some MSS. to *ati*) to destroy the hiatus. M transposes the *ucchrayo*, which is the chief cause of the trouble; TG recast the original, and substitute a reading which fits badly in the context, with an active *kurvanti* (having no proper subject), after *kriyate* in stanza 19. That is how I regard this little complex of readings, and hence the emendation, of whose correctness I myself have no doubt whatsoever.
c7.s3.35
1. 57. 21 : *hāsyarūpeṇa śaṃkaraḥ*.
c7.s3.36
This reading is undoubtedly, as WINTERNITZ rightly observes, the *lectio difficilior*, but there is not the slightest doubt about its being the original reading, judging by documentary evidence, and therefore, there is absolutely no need of underlining it. It is, in the first instance, the reading of the whole of M, which Winternitz appears not to have realized. I have stated, as my opinion, that the Malayalam version is the best representative of the Southern recension (Prolegomena, p. LXXIII). I am glad to note that Winternitz (Annals, Vol. 15, p. 170) agrees with me, and that even my learned friend Vidyāsāgara Prof. P. P. S. Sastri now endorses it. In the introduction (p. iii) to Vol. IV of his edition [101] of the Southern Recension, Sastri writes : “ Not having been subject to Nāyak influence in any manner whatsoever, the tradition handed down by the Malayalam Manuscripts preserved the Grantha text, in a purer and more unmixed form than even some comparatively early Grantha manuscripts, as the Malayalam MSS. do not at all seem to have come into contact with the Northern Recension till *very recent times*.”¹ Now this reading of the Malayalam version is supported by the reading of the Kaśmīrī version ; for though Ś₁ is missing, yet K₁ (India Office 2137) has the corrupt *hāmyahāmya*° (for *haṃsa*° of the rest of N), and it is well known that in Nāgarī copies of Śāradā MSS., *s* is frequently transcribed wrongly as *m* ; so that we have to read *hāsyahāsya* (dittography). The reading is further borne out by Devabodha’s gloss (the MS. reads *hāsa*° for our *hāsya*°) : *hāsarūpeṇa krīḍāvatārarūpeṇa* (cf. p. 990 of the Crit. Ed. for the gloss), where *haṃsa*, in any case, would not fit, for it is not clear how *haṃsarūpa* would be *krīḍāvatārarūpa*. Further, *śaṃkaraḥ* is the reading of all MSS. except B₅ Dn, which have *īśvaraḥ* (an obvious emendation) and TG which has *vāsavaḥ* (which must likewise be an emendation), originally, perhaps, only a gloss. Here practically the whole of N is supported by M. *śaṃkaraḥ* is here not a name of Indra, but only an epithet, an attribute meaning ‘auspicious, beneficent’. Thus in (B.) 3. 229. 6 *śaṃkara* is used with reference to Skanda : *lokānāṃ śaṃkaro bhava*. In (B.) 3. 201. 29 the epithet *maheśvara* (“great lord”) is applied to Viṣṇu *tvayi .nityaṃ maheśvara*. Further in 1. 58. 43, the epithet *īśa* (“lord”) and *śambhu* (“auspicious”) are applied to god Brahmā : *prabhavaḥ sarvabhūtānāṃ īśaḥ śambhuḥ prajāpatiḥ*. This shows that the words *śaṃkara*, *śambhu*, *īśa*, *maheśvara*, and so on, which are now generally regarded as proper names of Mahādeva, were still not specialized, but were applied to other gods as well, as mere epithets, descriptive of their beneficence or omnipotence.
c7.s3.37
¹ Italics mine.
c7.s3.38
1. 57. 58 : *āvayor dṛśyator ebhiḥ*.
c7.s3.39
Here it is not difficult to see on what authority the adopted reading rests. The footnotes give here the reading of Ñ V₁ B D [102] (except D₅) of the Northern, and of G of the Southern recension. The adopted reading (which is partly supported by the *dṛśyayor* of B₃) is, therefore, the reading of the remaining MSS. K D₅ T M (since Ś₁ is missing, as mentioned on p. 244, at the beginning of the adhy., or as may be seen from the table on p. XXIV of the Prolegomena). The Vulgate lacks here the support of the Granthas version, which has an entirely different reading :
c7.s3.40
*āvayoś ca kathaṃ brahman bhaviṣyati samāgamaḥ* |
c7.s3.41
The Grantha version dispenses altogether with the troublesome *dṛśyataḥ* of the original (or the unsuitable v. l. *dṛṣṭayoḥ* of the Vulgate), thus revealing its secondary character. The line was recast in the archetype “Sigma” (see the pedigree in the Prolegomena, p. xxx). The textual relations here are ideal, being clear as crystal. Winternitz is right in regarding it as pass. part. praes. with active ending. Passive forms with active endings are quite common in the epic. I adduce a few examples to dispel the remaining doubts of sceptics : 1. 11. 15 *iṣyati* (v. l. *iṣyeta*, *arhati* etc.) ; 16. 15 *ākṣipyataḥ* (no. v. l. explanation) ; 34. 13 *mokṣyanti* (v. l. *mokṣyante*) ; 47. 9 *saṃbhriyantu* (v. l. *°bharantu*, *°bhavantu*) ; 48. 12 *dahyatām* (no. v. l. !) 48. 13 *pacyatām* (v. l. *jvalatām*, *dahyatām*) ; 48. 22 *dīryati* (v. l. *śīryati*) ; 53. 6 *vidūyatā* (no. v. l. !) ; 71. 44 *dṛśyet* (v. l. *paśyet*) ; 124. 24 *uhyantaḥ* (v. l. *uhyante*) ; 147. 8 *vyucchidyet* (v. l. *ucchindyāt*, *chidyeta*) ; 165. 24 *hriyasi* (v. l. *gṛhyase*, *hriyase*, *prayāsi*) ; 202. 19 *adṛśyadbhiḥ* (v. l. *adṛśyaiś ca*, *adṛśyau tau*) ; 217. 13 *mathyataḥ* (v. l. *manthane*) ; 218. 49 *praviśīryataḥ* ; 219, 5 *adṛśyan* (v. l. *adṛśyā*, *nyapatan*).
c7.s3.42
1. 92. 2 : *Gaṅgā śrīr iva rūpiṇī*.
c7.s3.43
The configuration of the MSS. as well as the intrinsic merit of the readings are different in the two lines. That is how Ś₁ K₁ appear to be of greater authority in the first line than in the second. The *salilāt* of the text is found in *all* MSS. except Ś₁ K₁ (S only transposing the word : Text *salilāt tasmāt*, S *Gaṅgā salilāt*), and is, therefore, for one thing, obviously far better documented than *śayanāt* of Ś₁ K₁ only. In the second line, therefore, we have, practically, only two readings : *śayanāt* of Ś₁ K₁, against *salilāt* of the rest ; therefore the reading of Ś₁ K₁ has been rightly re-[103]-jected. Such is not the case in the first line. Here we have *three* nearly independent readings (Ś₁ K₁ *Gaṅgā śrīr iva rūpiṇī* : Vulgate G. *strīrūpadhāriṇī* : S *lobhanīyatamākṛtiḥ*, which latter is our *fourth pāda*). Here, while the two Northern readings are somewhat allied to each other, the Southern reading is entirely different, having very little connection with the Northern. None of the readings can be mechanically derived from the other, and intrinsically they are all more or less of the same value. Such being the case, the Northern tradition was, as usual, followed. Winternitz does not say why he thinks *strīrūpadhāriṇī* is better than the other. The reason why I chose *śrīr iva rūpiṇī* (“beautiful like Lakṣmī”) is that it is nearer in sense to *lobhanīyatamākṛtiḥ* (“with a most enticing form”) of S, than *strīrūpadhāriṇī* (“assuming the form of a woman”) of the Vulgate. I have here explained in detail the exact reasons which have weighed with me in making the difficult choice, but as a matter of fact, they are all three epic *iterata* and these *iterata* keep alternating with each other indiscriminately. In such cases, it is impossible to decide, with certainty, which is original and which is secondary, and the matter is also of no great consequence. The uncertainty of the text tradition has been duly indicated by a wavy line.
c7.s3.44
1. 92. 7 : *rājan kanyām varastriyam*.
c7.s3.45
I am glad WINTERNITZ has drawn my attention to a wrong reading which has crept in here, through oversight, and I gladly take this opportunity to publish a correction. The reading I had intended to adopt, or at least I should have adopted, is precisely the one advocated by WINTERNITZ : *rājan kāmyām vara°*, that of the Kaśmīrī version (Ś₁ K₀₋₂.₄). The fact is that just for first two words of the last pāda of this stanza, there is a confusing array of readings in Northern MSS., all meaning, more or less, the same thing : *rājan kāmyām* (Ś₁ K₀₋₂.₄), *rājan divyām* (K₃), *divyām kanyām* (Vulgate), *kāmyām divyām* (B), *divyām kāmyām* (Ñ₁.₃). Under these circumstances, clearly, the Kaśmīrī reading should have been adopted, according to the principles enunciated in the Prolegomena, but through a clerical error *kanyām* (of the Vulgate) [104] has crept in, in place of *kāmyām* (of the Kaśmīrī), which was intended.—I do not agree with WINTERNITZ however when he says that the epithet *kanyā* is “not very appropriate for Gaṅgā”. It would be quite appropriate, as far as I can see. It would imply that she was an unmarried virgin, which is the regular meaning of *kanyā*; compare :
c7.s3.46
1. 57. 63 *uvāca matpriyaṃ kṛtvā kanyaiva tvaṃ bhaviṣyasi*
1. 104. 12 *prādāc casyāḥ kanyātvaṃ punaḥ sa paramadyutiḥ*
(B) 3. 307. 16 *sā mayā saha saṃgamya punaḥ kanyā bhaviṣyasi*
c7.s3.47
It would then answer Pratīpa's objection (1. 92. 6) :
c7.s3.48
*nāhaṃ parastriyam kāmāt gaccheyaṃ varavarṇini.*
c7.s3.49
But *kanyām* is not supported by the MS. evidence, and *kāmyām* is : that is the main point. The constituted text should therefore be corrected accordingly by changing *kanyām* to *kāmyām*.
c7.s3.50
1. 92. 45 : *na ca tāṃ kiṃcanovāca*.
c7.s3.51
Here I do not agree with WINTERNITZ. I am fully persuaded that the text is quite in order. The first mistake WINTERNITZ makes is in thinking that *sa* is the reading of Ś₁ K ; it is actually given in the crit. app. as the reading of Ś₁ K₀.₃.₄ D₅. WINTERNITZ has overlooked the fact that Ś₁ differs from K₁, which is in itself suspicious; and K₂ is also excluded from the group. If *ca* has to be translated by “but”, there is no help for it ; for *ca* has frequently to be rendered in that way. For the Mbh., I may add, *ca* and *tu* are almost synonymous, *tu* having entirely lost the specific adversative force. It is quite clear that there must have been a period in the history of Sanskrit, as it was spoken, in which the two small particles *ca* and *tu* were confused in the mouths of the common people, and were used indiscriminately. That is, in fact, the only explanation of the curious particle *cu* one comes across in certain Aśoka inscriptions, which is evidently a combination of *ca* and *tu*. The Śāradā MS. undoubtedly offers a “better” reading, but, as a comparison of the different versions shows, it is a clear innovation or emendation. That the original had something like *na ca* (of the text) is proved by TG, which has *na tu*, only in a slightly different combination: [105] *uvāca kimcin na tu tām*. M differs from TG, and has a third combination, with a new word *vacanam*, not found in the other versions: *novāca vacanam kimcit*, which has neither *na ca* nor *na tu*, and which does not help to solve our difficulty. The *sa* of some of the Kaśmīri MSS. is not necessary for the context, as the subject *mahipatiḥ* (“the king”) occurs in the same line.
c7.s3.52
1. 92. 50 : *aṣṭeme Vasavo devāḥ*.
c7.s3.53
Practically all the MSS. except those of the D version begin with *aṣṭau*, which makes it probable (if it does not directly prove) that the original must have begun with *aṣṭau*; therefore *imeṣṭau* (one of the readings preferred by Winternitz) must be ruled out, as an emendation, probably of a *lectio difficilior*. The (final) *me* of Ś₁ K₁ and the initial *ime* of some Nāgarī MSS. suggest that the original was the conjectural *aṣṭeme*, which combines the characteristics of these groups, and which is very close to the *aṣṭa me* of Ñ₂. It is also suggested, to a certain extent, by the Bengali *aṣṭau ye*. The conjectural *aṣṭeme*, is in fact,, the hypothetical form to which the various readings *aṣṭau me*, *aṣṭa me*, *aṣṭau ye*, *aṣṭau hi*, and *imeṣṭau* seem to converge. The convergence is not a matter of which a mathematical proof can be supplied. If Winternitz does not see the convergence, he may substitute for the text reading any of the five readings found in the MSS. I did not know which to choose; hence the emendation. The change is not quite unmotivated. The motive appears to have been the seemingly irregular Sandhi (double crasis): *aṣṭau + ime = aṣṭā + ime = aṣṭeme*. But as a matter of fact, there need not have been any irregularity at all in the Sandhi, for *aṣṭeme* may simply be resolved into *aṣṭā+ime*, or *aṣṭa+ime*, since besides the heavier *aṣṭau*, both the forms *aṣṭā* and *aṣṭa* were in use, of which the latter (*aṣṭa*) particularly was used, I think, in all periods of the language (cf. Whitney, *Sanskrit Grammar* § 483). The translation of Winternitz is not quite accurate; at least it is not literal. The stanza has no word for “obtained birth,” which Winternitz interpolates into it, in order to justify the reading *me*, “of me,” which he has further to explain as “in my body”. Translate: “Those gods, the Eight Vasus, had on account of Vasiṣṭha’s curse become men (lit. attained the state of men).”
c7.s3.54
[106] 1. 93. 1 : *mānuṣīm tanum āgatāḥ*.
c7.s3.55
The agreement between S and Ñ V₁ Dn D₁ signifies absolutely nothing for the *original* reading; it is purely the result of conflation. V₁ is, moreover, not a *version*; it is merely a single MS., written in Maithilī script, and as such, capable of showing any reading at random. Again, the agreement of ÑV₁ with Dn and S against B Da is most unnatural and thoroughly suspicious. In other words, the MS. evidence here is extremely confused. Ordinarily I would have adopted the Ś₁ K reading *mānuṣatvam upāgatāḥ*, which has occurred already in the previous adhy. (1. 92. 50), and occurs again in (B) 3. 83. 66; but owing to the *mānuṣīm* of B, which recurs in S, I have given here preference to the reading of the Bengali version. I may point out that it is the reading of the entire Bengali version, and not of a few unimportant MSS., as Winternitz imagines. The readings are these:
c7.s3.56
Ś₁ K *mānuṣatvam upāgatāḥ*
B *mānuṣīm tanum āgatāḥ*
S *mānuṣīm yonim āgatāḥ*,
c7.s3.57
This being the constellation of the readings, *mānuṣīm* and *āgatāḥ* are as good as certain, as suggested by the agreement between B and S. Therefore the real variant is only *tanum* : *yonim*. It is likely that the *original* was *yonim* which was corrupted, in N, to *tanum*, and then the original *pāda* *mānuṣīm tanum āgatāḥ* (preserved in Bengali only) was replaced by the recurrent tag *mānuṣatvam upāgatāḥ*; but it is most unlikely that the latter was the original which was changed in B to *mānuṣīm tanum ā°*, and in S to *mānuṣīm yonim ā°*. In any case, the Kaśmīrī reading must be rejected here as secondary (cf. Prolegomena, p. LXII), as shown by the evidence of B and S. Dn D₁ appear to have got their *yonim* from S (ibid. p. LXVII), and Ñ V₁ possibly from Dn.
c7.s3.58
1. 93. 11 : *devadevarṣisevitam*.
c7.s3.59
Winternitz is mistaken in thinking that the hermitage in question was frequented by Ṛṣis *only*, and not by gods as well. In fact, the talk is not about a hermitage at all, but about the [107] forest (*vanam*) containing the hermitage. Moreover, it was not an ordinary hermitage, but the hermitage of Vasiṣṭha, situated on the slopes of the celestial mount Meru (1. 93. 15 f.) :
c7.s3.60
*Vasiṣṭho nāma iti sa khyāta āpava ity uta* || 15
*tasyāśramapadam puṇyam mṛgapakṣigaṇānvitam* |
*Meroḥ pārśve nagendrasya sarvartukusumāvṛtam* || 16
c7.s3.61
The forest, therefore, in which Vasiṣṭha’s hermitage was situated, was inhabited and frequented by divine ṛṣis (like Vasiṣṭha) *and* by gods (like the Vasus). That is how, in fact, the Vasus found themselves in that hermitage on that fateful day. If the forest were not frequented by gods, the Vasus would not have come there at all. If we adopt the reading *devadevarṣisevitam* (which is plausible) then the compound must be taken to mean "frequented by gods and ṛṣis" (and not by divine ṛṣis, as WINTERNITZ understands it). According to our text, the forest was frequented by gods and divine ṛṣis. As will be seen, there is not much to choose between the readings, but I prefer the text, because only the divine ṛṣis like Vasiṣṭha could reach those regions, not ordinary ṛṣis. Moreover, such adjectives (forming the entire second half of the line) are extremely common in the descriptions of sacred forests, hermitages, tīrthas, with many variants, as may be seen from the following examples from the Āraṇyaka (Bom. Ed.) :
c7.s3.62
3. 82. 41 *devarṣiṭṛsevita*
84. 46 *siddhagandharvasevita*
87. 5 *rājarṣigaṇasevita*
89. 7 *devagandharvasevita*
89. 8 *devarṣigaṇasevita*
90. 21 *brahmarṣigaṇasevita*
156. 10 *siddhacāraṇapūjita* and 145. 41 *devadevarṣipūjita*,
c7.s3.63
which latter has our combination *devadevarṣi*, and in which *deva* cannot be separated from *devarṣi*, for the stanza reads :
c7.s3.64
*tatrāpaśyata dharmātmā devadevarṣipūjitam |*
*Nara-Nārāyaṇasthānam Bhāgīrthyo paśobhitam ||*
c7.s3.65
The adjective *devadevarṣisevitam* (or *°pūjitam*) is used with reference to the most sacred and inaccessible regions like the [108] penance groves on Meru or Gandhamādana, or on the Himālayan peaks. I do not wish to suggest that the simple phrase *devarṣisevitam* would not have sufficed here. I only want to show that there is nothing wrong with the text reading, which is proved by the Kaśmīrī version (Ś₁ K). From the graphical point of view, the difference between the readings (*devā* and *deva-*) is so slight that the documentary evidence actually counts here for very little. It is just owing to this uncertainty that the reading has been underlined in the critical text.
c7.s3.66
1. 214. 9 : *Dharmarāje * atiprityā.*
c7.s3.67
Here again we notice WINTERNITZ's prejudice against hiatus; to which I had drawn attention in the discussion on 1. 57. 20, above. But here my case is stronger still. As far as the Vedic tradition is concerned, *e* (like *o*) remained unchanged before *a* which was generally elided in the written text, but, according to the evidence of metre, must almost invariably in the Ṛgveda and generally in the Yajurveda and Atharvaveda, be pronounced whether written or not. According to the statistics prepared by Vedic scholars, it must be pronounced in the Ṛgveda in 99 per cent of the cases, in the Atharvaveda and the metrical portion of the Yajurveda, in about 80 per cent of its occurrences. This shows that in the older stages of the language, at any rate, any Sandhi between final *e* (or *o*) and initial *a* was rare. The rule becomes more and more rigid as we advance, until with classical authors, ignorance or violation of the rule came to be regarded as a capital blunder. Now the Mbh., whatever its age be, stands unquestionably midway between the Vedic and the classical epoch, and therefore partakes naturally of the linguistic characteristics of both. In the Vedic literature, where the scribes or editors did not dare to add even a single syllable to the received text, the later antipathy towards hiatus shows itself in the efforts to coalesce, in the written text, the adjoining vowels, according to rules of (Sanskrit) grammar, leaving a correct but unmetrical and unreadable line. The subterfuge is however so obvious that no one hesitates to dissolve the Sandhi automatically and restore the hiatus. Much reluctance is felt, on the other hand, by scholars [109] in admitting that the epic text likewise contained originally many instances of hiatus, though of course they are not as frequent as in the Veda. The reluctance is due to two causes. Firstly the Mbh. text looks on the surface so like a classical text that scholars, who have so far studied the text mainly from printed editions, insist on applying rules of classical grammar to the text. The second reason is that the ancient redactors who had not the same compunction about making small alterations in the epic text (as is evident from our critical apparatus) as they had in the case of the Vedic texts, have not resorted to coalescence in order to remove the hiatus, but have boldly added little expletives like *ca*, *tu*, *hi* for saving their precious rules of grammar, a procedure which, as it leaves the line metrically intact, makes the detection of their nefarious interference difficult, if not impossible. Now though the Mbh. looks on the surface like a classical text, there are many peculiarities of Sandhi and grammar—even in the printed editions and the Vulgate—which distinguish it from a classical text. Coalescence unknown to classical usage is seen in *amale 'tmānam* (1. 68. 64), *te 'jñayā* (1. 70. 41), *manyate 'tmānam* (1. 198. 19). We have hiatus in a compound in *Nārāyaṇaurogataḥ* (1. 16. 35); *Pragṛhya* Sandhi in *samupete 'dbhute 'naghe* (fem. du. 1. 14. 5), *jajñate 'stravi śāradau* (1. 57. 88) etc. Double crasis in *pannagābhavan* (1. 21. 6), *Vasumanābravīt* (1. 87, 18), *jāyateti* (1. 11. 13). Hiatus between pādas, caused by the change of *as* to *o* (1. 76. 35) : *jagāma svapuram hṛṣṭo anujñāto mahātmanā*. Frequently we come across Prakritic Absolutives like *gṛhya* (1. 2. 93 ; 9. 19 ; 39. 23, 30 ; 119. 16 ; 123. 12, 16, 50 ; 124. 20 etc.), *toṣya* (1. 1. 109), *cintya* (1. 9. 2), *uṣya* (1. 71. 58), etc. etc.; and the converse (*tvā* for *ya* in *sam-pūjayitvā* (1. 54. 15), *ānayitvā* (1. 66. 12), *ni-ṣṭanitvā* (1. 85. 18), *anu-śiṣṭvānugatvā* ca (1. 133. 24), *prāpayitvā* (1. 189. 25) etc., some of which may however, be explained as absolutives of verbs with separable prepositions. Such a separable preposition we have in 1. 65. 34 :
c7.s3.68
*prati śravaṇapurvāṇi nakṣatrāṇi sasarja ha* |
c7.s3.69
where *prati* is to be construed with *sasarja*, as *pratisasarja* : a line often misunderstood by editors, commentators and translators alike. These violations of (Sanskrit) grammar are not so rare that one has to hunt for them with a microscope. There is an endless list of [110] them. One meets with them at every step. If these and scores of other irregularities do occur as a matter of fact, why could there not be *hiatus* as well? Scholars are not yet familiar with this phenomenon, because they have been dealing, so far, with the clarified text, from which most of these irregularities have been carefully expunged, as is evident from our critical apparatus, by purists who have had the handling of the epic text during a long succession of centuries. I have noted that even BÖHTLINGK, who was otherwise a careful and conscious editor, has in editing Mbh. passages for his *Chresthomathie*, rigidly enforced Sandhi rules, even in prose passages, with a zeal which was worthy of a better cause, where there was not the slightest manuscript authority for doing so. My study of the Mbh. MSS during the last ten years has convinced me now that it is the grammatical and material irregularities of the original that are responsible for quite a large fraction of the mass of variations which we come across in the MSS. The correctness of my reconstructions can be proved only by adducing the entire evidence, which it is not possible to do here, but which may be undertaken later on, when a large part of the text has been dealt with in a like manner. I am fully persuaded that the Critical Edition, as it advances, will supply material for confirming most of my conclusions.—To return to the case under consideration. WINTERNITZ points out that *hi* is found in all N MSS., including S₁, but he ignores that K₁ shows here *tu*, not *hi*? Now K₁ is a *clear* transcript of a Śāradā MS., and agrees with our S₁ to such an extent that I have expressed it as my opinion that it is a copy of a Śāradā original very similar to *our* Śāradā MS. It is, no doubt, a very small and insignificant variation, but how would WINTERNITZ explain it all the same? *hya* cannot be misread as *tva*, even in the Śāradā script. Why should just K₁ show here *tva*? Is it not, perhaps, because a not very distant ancestor of our S₁ and K₁ had still the obnoxious hiatus, as in our text? The Kaśmīrī version does contain many original features and archaisms.
c7.s3.70
1. 215. 2 : *ekām tṛptim prayacchatām*.
c7.s3.71
*prayacchatām* of the text, I may point out, is not entirely wrong; it may be construed with *bhavantau*, understood, a more [111] respectful form of address than the second person. In stanza 5, however, the reading *samprayacchatam* has been accepted, because the subject in the second person (*yuvām*) is expressed. We cannot argue that since in 5 we have *samprayacchatam*, therefore in 2 we must have also *prayacchatam*. Because in the Mbh., which is not and never was polished literature, we do come across such strange yoke-fellows. Changes of subject and changes of construction are frequent. That is the view I had taken when I adopted the text reading.
c7.s3.72
But on second thoughts I decided that *prayacchatam* would nevertheless be a better reading, and the correction now proposed by WINTERNITZ had been already published by me, in the “Addenda et Corrigenda,” at the end of the volume (p. 996).
c7.s3.73
*sasarja yat svatapasā Bhauvano bhuvanaprabhuḥ* |
*prajāpatir anirdeśyaṃ yasya rūpaṃ raver iva* ||
c7.s3.74
The matter is not simple as WINTERNITZ imagines. Of course, with the reading *yaṃ*, any one can see that the relative may be construed with *ratham* in 8. We then get two sentences: “Which (scil. chariot of Arjuna) was fashioned by Bhauvana....by (the power of) his austerities,” and “whose beauty was like that of the sun.” But the reading *yat* is actually found in Ś₁K (except K₀) V₁G (except G₁); i.e. in the Kaśmīrī version (except K₀), agreeing with the Grantha version (except G₁), plus the Maithilī MS. How is it that so many MSS. give what WINTERNITZ considers, an “impossible” reading? The reading is, in fact, not impossible at all; only the construction is a little involved and difficult to understand. By reading *yat* we actually get better sense. Construe: *yasya, raver iva, yat anirdeśyaṃ rūpaṃ, (tat) Bhauvanaḥ....svatapasā sasarja*, “whose indefinable (or incomparable) loveliness (or splendour), like that of the Sun, Bhauvana had fashioned by (the power of) his austerities.” This construction avoids the two disjointed and halting sentences *yaṃ anirdeśyaṃ Bhauvanaḥ....svatapasā sasarja*, and *yasya rūpaṃ raver iva* (or *yaṃ Bhauvanaḥ....svatapasā sasarja*, and *yasya anirdeśyaṃ rūpaṃ raver iva*), which result from the wrong reading *yaṃ* for [112] *yat*, *yat* is almost a perfect example of the *lectio difficilior*, and a regular trap for unwary editors.
c7.s3.75
1. 218. 14 : *jaladhārāmuco 'kulān*.
c7.s3.76
WINTERNITZ is here again mistaken in thinking that *°mucotulān* is the reading of S. It is the reading of only five of the thirteen Southern MSS. (T₁ G₁.₂ M₃.₅); three others (T₃ G₃.₄) read *°muconilān*, three more (M₆₋₈) read *mumocivān*, two (G₅.₆) read as in text (*°mucokulān*). The question is, in fact, what was the original reading of S. That question I have not been able to answer definitely, and I have, therefore, put in, as a stop-gap, the reading of G₅.₆ which seemed to me not improbable, since initial *ā* is curiously enough, sporadically treated in the Mbh. text, like *a*. The reading *°samākulān* is useless; it is obviously a substitute for some reading which was difficult to understand or explain.
c7.s3.77
I have underlined *muco* in the text, but I now think that it is as good as certain. It is documented not only by the whole of S, but also by K₀.₂.₃ Ñ V₁D₁. If *muco* be admitted, then *samākulān* of the Northern group becomes secondary, but with an important residue in the shape of the final *kulān*, agreeing with the final of *°mucokulān* of G₅.₆, rhyming with *°muco-*.
First 77 of 83 passages. Read on
Linked texts (3)
- Mahābhārata — Vyāsa, Sanskrit
- Muṇḍakopaniṣad — Sanskrit
- Harivaṃśa — Vyāsa, Sanskrit