Prolegomena

To the Critical Edition of the Ādiparvan, Book 1 of the Mahābhārata
By Vishnu S. Sukthankar (1933)

Introduction

The need of a critical or (as it was sometimes called) a “correct” edition of the Mahābhārata has been felt (at first, of course, rather vaguely) by Sanskritists for over half a century.¹ It was voiced, however, in a clear and emphatic manner, for the first time, by Professor M. Winternitz, at the XIth International Congress of Orientalists, held at Paris, in 1897, when he read a paper drawing attention to the South Indian manuscripts of the Great Epic and ending with the remark that a critical edition of the Mahābhārata was “wanted as the only sound basis for all Mahābhārata studies, nay, for all studies connected with the epic literature of India”.² The idea received a concrete shape in his proposal for the foundation of a Sanskrit Epic Text Society, which he laid before the very next session of the Oriental Congress (XIIth), held in Rome (1899). Again, three years later, at the following session of the Congress (XIIIth) held in Hamburg (1902), Professor Winternitz reiterated his requisition and endeavoured to impress again upon the assembled savants that a “critical edition of the Mahābhārata was a sine quâ non for all historical and critical research regarding the Great Epic of India”.³

The reception accorded to the various proposals made by Professor Winternitz in connection with his favourite project was not as cordial as might have been expected from an enlightened, international assemblage of Sanskritists. “At first”, writes Professor Winternitz himself, “the idea of a critical edition of the Mahābhārata met with great scepticism. Most scholars were of opinion that it was impossible to restore a critical text of the Great Epic, and that we should have to be satisfied with editing the South Indian text, while the North Indian text was represented well enough by the Calcutta and Bombay editions. Only few scholars were in full agreement with the plan of one critical edition”.

Notwithstanding this general apathy, a committee was appointed by the Indian Section of the International Congress of Orientalists in Rome (1899) to consider the proposal of Professor Winternitz for the foundation of a Sanskrit Epic Text Society, already mentioned. This committee was not in favour of the said proposal. It recommended instead that the work of preparing the critical edition should be undertaken by the International Association of Academies. The London session of this Association, held in 1904, adopted the above suggestion and resolved “to make the critical edition of the Mahābhārata one of the tasks to be undertaken under its auspices and with the help of funds to be raised by the Academies”. In pursuance of this decision, the Academies of Berlin and Vienna sanctioned certain funds earmarked for the Mahābhārata work, with whose help the preliminary work for the critical edition was actually begun.

In furtherance of this project, then, Professor H. Lüders prepared a “Specimen” of a critical edition of the Mahābhārata (Druckprobe einer kritischen Ausgabe des Mahābhārata, Leipzig 1908) with the funds provided for the purpose by the Königliche Gesellschaft der Wissenschaften in Göttingen.⁴ The Specimen, which was meant only for private circulation,⁵ consisted of 18 pages, comprising the constituted text (pp. 1-11) of the first 67 stanzas of the Ādiparvan with their various readings (printed as footnotes) an appendix (pp. 12-17), on a similar plan, containing the text of the Brahmā-Gaṇeśa interpolation (with its variants), and finally a list (p. 18) of the 29 manuscripts, selected exclusively from European libraries, which formed the specimen apparatus criticus.⁶ This little brochure, which must rank in the annals of Mahābhārata studies as the first tentative critical edition of the Mahābhārata, was laid before the Indian Section of the XVth International Congress of Orientalists, held in Copenhagen (1908). The tender seedling, planted with infinite care, did not, however, thrive in the uncongenial European soil. Twenty years later, in 1928, at the XVIIth International Congress of Orientalists, held at Oxford, Professor Winternitz reported that, under the scheme of the International Association of Academies, “except this specimen (Druckprobe) nothing has been printed”.⁷ However, in the interval some preliminary work, such as the classifying and collating of manuscripts had been done by Professor Lüders and some of his pupils (among them my fellow-student and friend Dr. Johannes Nobel, now Professor in the University of Marburg), by Professor Winternitz and his pupil Dr. Otto Stein, and by Dr. Bernhard Geiger (Vienna). The last great World War gave its quietus to this ambitious project, sponsored by the Associated Academies of Europe and America, and finally diverted the attention of European scholars from the Mahābhārata Problem. After the war, the Bhandarkar Oriental Research Institute, then in its early infancy, enthusiastically undertook the work, making a fresh start, fortunately without realizing fully the enormousness of the project or the complicacies of the problem. At a meeting of the General Body of the Institute, held on July 6, 1918, Shrimant Balasaheb Pant Pratinidhi, Chief (now Ruler) of Aundh—the liberal and enthusiastic patron of diverse projects calculated to stimulate research, advance knowledge, and enhance Indian prestige—the president elect on the occasion, easily persuaded by a band of young and hopeful Sanskritists who had returned to India after completing their philological training abroad, with their heads full of new ideas, urged upon the audience the need of preparing a Critical and Illustrated Edition of the Mahābhārata, offering to contribute, personally, a lakh of rupees, by annual grants, towards the expenses of producing the edition.⁸ The donor was warmly thanked for this princely

gift and the offer was gratefully accepted by the spokesmen of the Institute, who in their turn undertook to prepare an edition that would meet with the high requirements of modern critical scholarship. In accordance with this decision of the General Body of the Institute, the late lamented Sir Ramkrishna Gopal Bhandarkar, the doyen of the Sanskritists of Western India and the inspirer of the critical and rigorous scholarship of the present day, inaugurated, in April 1919, this monumental work by formally beginning the collation of the opening mantra of the works of the ancient Bhāgavata sect, which is found also at the beginning of some manuscripts of the Mahābhārata.⁹ नारायणं नमस्कृत्य नरं चैव नरोत्तमम् । देवीं सरस्वतीं चैव ततो जयमुदीरयेत् ॥ Then, on the basis of the promise of the donation of a lakh of rupees by the Ruler of Aundh, the Institute appealed for the very large financial support needed to Indian governments, princes, and men of wealth. Not as many favourable responses were received as might have been expected; but very generous aid was and is being given by some, whose names are recorded elsewhere. The reasons which have induced Sanskritists both here and abroad to undertake this gigantic enterprise are easy to understand. The pre-eminent importance of the epic is universally acknowledged. Next to the Vedas, it is the most valuable product of the entire literature of ancient India, so rich in notable works. Venerable for its very antiquity, it is one of the most inspiring monuments of the world, and an inexhaustible mine for the investigation of the religion, mythology, legend, philosophy, law, custom, and political and social institutions of ancient India. As a result of the researches that have been carried on during the last thirty-five years or so, there is now no doubt whatsoever that the text of the Mahābhārata has undergone numerous changes.¹⁰ The texts of the Northern and Southern manuscripts—to mention only two of the manuscript classes—are widely divergent, and much uncertainty prevails regarding the correctness and originality of the texts preserved by them. The existing editions—which either merely reproduce the version of a particular type of manuscripts, like the Bombay edition,¹¹ or else are eclectic on no recognizable principles,—fail to remove the uncertainty of the text. The present edition of the epic is intended chiefly to remedy this unsatisfactory state of things. What the promoters of this scheme desire to produce and supply is briefly this: a critical edition of the Mahābhārata in the preparation of which all important versions of the Great Epic shall have been taken into consideration, and all important manuscripts collated, estimated and turned to account. Since all divergent readings of any importance will be given in the critical notes, printed at the foot of the page, this

Critical Edition of the Mahābhārata (Poona 1919), published by the Institute, p. v.

edition will, for the first time, render it possible for the reader to have before him the entire significant manuscript evidence for each individual method for scientific investigation of the epic is obvious. The value of this edition will be this. Another feature of the new Since not even the seemingly most irrelevant line or stanza, actually found in a Mahābhārata manuscript collated for the edition, is on any account omitted, this edition of the Mahābhārata will be, in a sense, more complete than any previous edition.¹ It will be a veritable thesaurus of the Mahābhārata tradition.

Under the scheme outlined above, a tentative edition prepared by the late Mr. Narayan Bapuji Utgikar, M.A., and published by the Institute in 1923. Copies of this edition were distributed gratis among leading Sanskritists— Indian, European and American—with a view to eliciting from them a frank expression of their opinion on the method worked out by the then editor-in-chief. The opinions received were very favourable and highly encouraging. The valuable suggestions made by many eminent authorities have been to a great extent followed in the subsequent work.

Collation Of Manuscripts

Collation of the manuscripts is being done, regularly, not merely at the Institute, but also at the Visvabharati of Rabindranath Tagore in Bengal under the supervision of Pandit Vidhushekhara Bhattacharya, and at the Saraswathi Mahal in Tanjore under the supervision of M. R. Ry. Rao Saheb T. Sambamurthi Rao Avl., B.A., B.L. These outside centres were at first intended chiefly for the collation of the Bengali and the Telugu- Grantha manuscripts respectively. But provision has now been made at the Institute itself for the collation of manuscripts written in any of the seven scripts (Śāradā, Nepālī, Maithili, Bengali, Telugu, Grantha and Malayālam), besides Devanāgarī, which are ordinarily required for our Mahābhārata work.

The entire Mahābhārata stands now collated from a minimum of ten manuscripts; many parvans have been completely collated from twenty manuscripts; some from thirty; a few from as many as forty; while the first two adhyāyas of the Ādi, which have special importance for the critical constitution of the text of the entire epic, were collated from no less than sixty manuscripts.

The collation is done by a permanent staff of specially trained Shastris (Northern as well as Southern) and University graduates. For the purposes of collation, each Mahābhārata stanza (according to the Bombay edition of Ganpat Krishnaji, Śaka 1799) is first written out, in bold characters, on the top line of a standard, horizontally and verti- cally ruled foolscap sheet. The variant readings are entered by the collator horizontally along a line allotted to the manuscript collated, akṣara by akṣara, in the appropriate column, vertically below the corresponding portion of the original reading of the “Vulgat”. On the right of each of these collation sheets, there is a column four inches wide reserved for remarks (regarding corrections, marginal additions etc.), and for “additional” stanzas found in the manuscripts collated, either immediately before or after

the stanza in question. Very long "additions" are written out on separate “śodhapātras” and attached to the collation sheets. The collations are regularly checked by a batch of collators different from the one which did the collation in the first instance, before they are handed over to the editor for the constitution of the text.

General Account Of The Manuscripts

It is by no means easy to answer the question how many manuscripts of the Mahābhārata there are in existence; firstly, because, no complete list of these manuscripts has ever been compiled; and, secondly, because the expression “Mahābhārata manuscript”, as ordinarily used, is ambiguous in the extreme; it may apply to a small manuscript of the Bhagavadgītā alone, as well as to a complete manuscript of the Mahābhārata, in several volumes, containing all the eighteen parvans. Moreover, the parvans are mostly handed down separately, or in groups of few parvans at a time, at least in the oldest manuscripts now preserved. Therefore, in taking stock of Mahābhārata manuscripts, it is best to take as unit of measurement a manuscript of a single parvan.

As a very approximate computation, I may state that there are known to be about 235 manuscripts of the Ādi, counting only such as have come within my knowledge from catalogues of private and public libraries accessible to me, as also those manuscripts whose owners have sent them to the Institute for collation or inspection. But this is probably by a long way not the total number of extant manuscripts of this parvan, because there must be quite a large number of manuscripts in private hands, of which we know next to nothing. It has been the experience of most manuscript collectors in India that when one takes the trouble to look for the manuscripts, they turn up in quite astonishing numbers, though they are as for a rule late and of questionable worth. Of these 235 manuscripts of the Ādi, a little less than half (107) are in the Devanāgarī script alone. The other scripts are represented in this collection as follows: Bengalī 32, Grantha 31, Telugu 28, Malayāḷam 26, Nepālī 5, Śāradā 3,¹ Maithilī 1, Kannaḍa 1, and Nandināgarī 1. Of these manuscripts of the Ādi about 70 (i. e. a little more than 29 per cent of the total) were fully or partly examined and collated for this edition. And of these again about 60 were actually utilized in preparing the text. The critical apparatus of the first two adhyāyas gives the collations of 50 manuscripts. Many of these were, however, discarded in the sequel as misch-codices of small trustworthiness and of no special value for critical purposes. At the same time a few other manuscripts (such as the Śāradā and Nepālī codices), which were not available in the beginning, were added to the critical apparatus subsequently. A table given below supplies all the necessary details of the critical apparatus as to where the collations of the different manuscripts begin, where they end, and so on and so forth.

The choice of the critical apparatus is not an easy matter, owing to the astonishing bulk and the amazing variety of the material. The number of exact duplicates among these is decidedly small and almost negligible. An exception to this rule is formed only by manuscripts of commentators’ versions, which show inter se little difference. So that what has been said by Kosegarten with respect to the manuscripts of the Pañcatantra, applies, generally speaking, equally well to the Mahābhārata manuscripts: quot codices, tot textus. Notwithstanding these difficulties, the choice of our critical apparatus has not been entirely arbitrary. Efforts were made to secure manuscripts written in as many different Indian scripts as possible, which is the same as saying, manuscripts belonging to as many different Indian provinces as possible. Old manuscripts, even though fragmentary and partly illegible, were selected in preference to modern-looking manuscripts, though complete, neatly written and well preserved. Within the version, discrepant types were chosen in preference to similar types.² Of the Nīlakaṇṭha version, only three were selected, though it is by far the most numerous group; because, firstly, it is one of the latest versions; and, secondly, it has been edited several times already, though not as well as it should be; and, thirdly, there is little difference between the individual manuscripts of the group. The only important scripts unrepresented in our critical apparatus are: Kannaḍa, Uriyā and Nandīnāgarī.

Besides the manuscripts collated specially for this edition, I have made occasional use of the collations of manuscripts preserved in European libraries made by Theodor Goldstücker, photographic copies of which were presented to the Institute, for use in connection with this project, by the University of Strassburg, through the kind offices of the late Professor Émile Senart, as also of the collations intended for the edition planned by the International Association of Academies and made by the pupils of Geheimrat Professor Dr. Heinrich Lüders, which have been placed at the disposal of the Institute in pursuance of a resolution on the subject passed by the Indian Section of the XVIIth International Congress of Orientalists, held at Oxford, in 1928.³

Sixteen of the manuscripts collated bear dates, ranging from the 16th to the 19th century. The oldest dated manuscript of our critical apparatus is a Nepālī manuscript (N₈) which bears a date corresponding to A.D. 1511. The other dates are: A.D. 1519 (K₈), 1528 (V₁), 1598 (D₂), 1620 (D_a2), 1638 (K₂), 1694 (K₄), 1701 (D_r3), 1739 (Ko), 1740 (B₃), 1759 (B₈), 1786 (B₈), 1802 (D₅), 1808 (D_n2), 1838 (M₈), and 1842 (M₈). The Nīlakaṇṭha manuscripts are not all dated, but they can scarcely be much anterior to the beginning of the eighteenth century, since Nīlakaṇṭha himself

belongs to the last quarter of the seventeenth. Many of the Grantha manuscripts do bear dates, but since they refer to a cyclic era, it is difficult to calculate their equivalents.

Classification Of Manuscripts

The manuscript material is divided naturally into recensions by the scripts in which they are written. Corresponding to the two main types of Indian scripts, Northern and Southern, we get two main recensions of the epic. Each of these recensions is again divided into a number of sub-recensions, which I have called “versions”, corresponding to the different provincial scripts in which these texts are written. This principium divisionis is not as arbitrary as it might at first sight appear. The superficial difference of scripts corresponds, as a matter of fact, to deep underlying textual differences. It is common experience in India that when we have a work handed down in different versions, the script is invariably characteristic of the version.¹ The reason for this concomitance between script and version appears to be that the scribes, being as a rule not conversant with any script but that of their own particular province, could copy only manuscripts written in their special provincial scripts, exception being made only in favour of the Devanāgarī, which was a sort of a “vulgar” script, widely used and understood in India.

While the principle mentioned above is not entirely mechanical or arbitrary, it is also not ideal or perfect. It is often contravened in practice, mainly through the agency of the Devanāgarī, which is the chief medium of contamination between the different recensions and versions. Thus we come across Devanāgarī copies of the commentary or version of Arjuņamiśra, who was an Easterner; similar copies of the commentary or version of Ratnagarbha, who was a Southerner. There are again Devanāgarī copies of the Grantha and the Śāradā² versions. On the other hand, a popular version like that of Nīlakaṇṭha may be copied in any script. I have come across manuscripts of the Nīlakaṇṭha (Devanāgarī) version written in Śāradā,³ Bengali,⁴ Telugu and Grantha scripts. Another cause of disturbance was this. Along the boundaries of provinces speaking different languages or using different scripts, there are invariably bi-lingual and bi-scriptal zones. In these zones there was an ever operating impulse, tending to introduce innovations, obliterating the differentiae and normalizing the text. Never- theless, though nothing is impossible, it would be passing strange if we were to find a copy of the pure Śāradā version written, say, in the Malayālam script, or of the Grantha version in the Nepālī script.

List Of Manuscripts Forming The Critical Apparatus

The manuscripts utilized for this edition of the Ādi are as follows:

I. N(orthern) Recension. (a) North-western Group (ν). Śāradā (or Kāśmīrī) Version (Ś). Ś₁ = Poona, Bombay Govt. Collection (deposited at the BORI), No. 159 of 1875–76. Devanāgarī Group allied to the (Śāradā or) Kāśmīrī Version (K). K₀ = Poona, Bombay Govt. Collection (deposited at the BORI), No. 229 of 1895–1902. Dated V. Saṁ. 1795 (ca. A.D. 1739). K₁ = London, India Office Library, No. 3226 (2137). K₂ = Poona, Bombay Govt. Collection (deposited at the BORI), No. 182 of 1891–95. Dated V. Saṁ. 1694 (ca. A.D. 1638). K₃ = Baroda, Oriental Institute Library, No. 632. Dated V. Saṁ. 1575 (ca. A.D. 1519). K₄ = Poona, Bombay Govt. Collection (deposited at the BORI), No. 565 of 1882–83. Dated Śaka 1616 (ca. A.D. 1694). K₅ = Lahore, Dayanand Anglo-Vedic College, No. 1. K₆ = Poona, Bombay Govt. Collection (deposited at the BORI), No. 209 of 1887–91.

(b) Central Group (Υ). Nepālī Version (Ñ). Ñ₁ = Nepal, in private possession. Ñ₂ = Nepal, in private possession. Ñ₃ = Nepal, in private possession. Dated Nepālī Saṁ. 632 (ca. A.D. 1511). Maithilī Version (V). V₁ = Nepal, Darbar Library, No. 1364. Dated La. Saṁ. 411 (ca. A.D. 1528). Bengalī Version (B). B₁ = Santiniketan, Visvabharati Library, No. 1. Dated Śaka 1662 (ca. A.D. 1740). B₂ = Santiniketan, Visvabharati Library, No. 258. B₃ = Santiniketan, Visvabharati Library, No. 782. Dated Śaka 1681 (ca. A.D. 1759). B₄ = Santiniketan, Visvabharati Library, No. 413. B₅ = Dacca, University Library, No. 485. Dated Śaka 1708 (ca. A.D. 1786). B₆ = Dacca, University Library, No. 735. Devanāgarī Versions other than K (D). Devanāgarī Version of Arjuņamiśra (Da). Da₁ = Poona, Bombay Govt. Collection (deposited at the BORI), No. 80 of A 1879–80. Da₂ = Poona, Bombay Govt. Collection (deposited at the BORI), Viśrāmbāg I, No. 468. Dated V. Saṁ. 1676 (ca. A.D. 1620). Devanāgarī Version of Nīlakaṇṭha (Dn), the “Vulgate”. Dn₁ = MS. belonging to Sardar M. V. Kibe of Indore. Dn₂ = Mysore, Oriental Library, No. 1064. Dated V. Saṁ. 1864 (ca. A.D. 1808).

D_9_ = Poona, Bombay Govt. Collection (deposited at the BORI), No. 234 of 1895–1902. Devanāgarī Version of Ratnagarbha (Dr). D_r1_ = Tanjore, Saraswati Mahal Library, No. 1246. D_r2_ = Tanjore, Saraswati Mahal Library, No. 1199. D_r3_ = Tanjore, Saraswati Mahal Library, No. 1313. Dated Śaka 1623 (ca. A.D. 1701). D_r4_ = Tanjore, Saraswati Mahal Library, No. 1339. Devanāgarī Composite Version. D_1_ = Poona, Bombay Govt. Collection (deposited at the BORI), No. 29 of A. 1879–80. D_2_ = Tanjore, Saraswati Mahal Library, No. 1152. Dated V.Sam. 1654 (ca. A.D. 1598). D_3_ = Tanjore, Saraswati Mahal Library, No. 1360. D_4_ = Tanjore, Saraswati Mahal Library, No. 1126. D_5_ = Lahore, Dayanand Anglo-Vedic College, No.4. Dated V. Sam. 1858 (ca. A.D.1802). D_6_ = Tanjore, Saraswati Mahal Library, No. 1223. D_7_ = Tanjore, Saraswati Mahal Library, No. 1269. D_8_ = Tanjore, Saraswati Mahal Library, No. 1329. D_9_ = Tanjore, Saraswati Mahal Library, No. 1176. D_10_ = Tanjore, Saraswati Mahal Library, No. 1293. D_11_ = Tanjore, Saraswati Mahal Library, No. 1340. D_12_ = Tanjore, Saraswati Mahal Library, No. 1373. D_13_ = Poona, Bombay Govt. Collection (deposited at the BORI), Viśrāmbāg II, No. 191. D_14_ = Poona, Bombay Govt. Collection (deposited at the BORI), Viśrāmbāg II, No. 266.

II. S(outhern) Recension.

Telugu Version (T). T_1_ = Melkote, Yadugiri Yatiraj Math Library MS. (without number). T_2_ = Tanjore, Saraswati Mahal Library, No. 11865. T_3_ = Tanjore, Saraswati Mahal Library, No. 11809. Grantha Version (G). G_1_ = Melkote, Yadugiri Yatiraj Math Library MS. (without number). G_2_ = Melkote, Yadugiri Yatiraj Math Library MS. (without number). G_3_ = Tanjore, Saraswati Mahal Library, No. 11823. G_4_ = Tanjore, Saraswati Mahal Library, No. 11838. G_5_ = Tanjore, Saraswati Mahal Library, No. 11851. G_6_ = Tanjore, Saraswati Mahal Library, No. 11860. G_7_ = Melkote, Yadugiri Yatiraj Math Library MS. (without number). Malayālam Version (M). M_1_ = MS. belonging to Chief of Idappalli, Cochin. M_2_ = Cochin, State Library, No. 5. M_3_ = Cochin, State Library, No. 1. Dated Kollam 1013 (ca. A.D. 1838). M_4_ = MS. belonging to Kallenkara Pisharam of Cochin. M_5_ = Cochin (Jayantamangalam) ; property of the Paliyam family. M_6_ = Malabar (Nareri Mana) ; in private possession. M_7_ = Cochin (Avanapparambu Mana); in private possession. M_8_ = Malabar Poomulli Mana Library, No. 297. Dated Kollam 1017 (ca. A.D. 1842).

2

Detailed Account Of The Manuscripts

S₁

Poona, Bombay Government Collection (deposited at the BORI), No. 159 of 1875–76. Total number of folios 114 (some fragmentary), with about 24 lines to a page; size 12″ × 9¼″. Clear Śāradā characters (of perhaps the 16th or 17th century). Birch-bark (bhūrjapatra).

This unique and valuable MS. was purchased for the Government of Bombay, by Bühler, in Kaśmīr. It is listed on p. xi, and cursorily described at p. 64, of his Detailed Report of a Tour in Search of Sanskrit MSS. made in Kaśmīr, Rājputana, and Central India, a report printed as Extra Number of the Journal of the Bombay Branch of the Royal Asiatic Society for 1877. The lines of writing of the MS. run parallel to the narrow side of the leaf. There are, on an average, 24 lines on a page, and 36 akṣaras (i. e. a little over a śloka) in a line. A page, therefore, contains, on an average, 26 (anuṣṭubh) stanzas. Each folio bears, on its reverse, in the left-hand margin, near the bottom, a cipher representing the serial number of the folio and a signature indicating the title of the work, as well as the name of the parvan. — The MS., which is unfortunately incomplete and fragmentary, must have originally contained at least the first three parvans (Ādi, Sabhā and Araṇya), written, as far as one can judge, by the same hand. The extant portion contains the Sabhā in its entirety, but only fragments of the other two parvans, the beginning of Ādi and the end of Araṇya being lost. The Ādi, which appears to have extended from the beginning of the volume up to fol. 154, is particularly fragmentary; a continuous text begins only from fol. 63 (our adhy. 82). Of the first 62 folios, the extant portion contains only the lower segments (with 10 to 15 lines of writing on each page) of fol. 24–25, 36–37, 39, 47–48, 53–57 and 61–62; the initial 23 folios as also 15 other intermediate folios (viz. 38, 40–46, 49–52, 58–60) are entirely missing; while only 10 of these folios are complete. Folio number 96 is repeated. The Ādi ends at fol. 154a. The colophon repeats the stanzas of the Parvasaṅgraha giving the number of adhyāyas (230) in this parvan, as also its extent in “ślokas”, i. e. granthas (7984). The writing is neat and careful; erasures and corrections are few and far between. Occasionally one comes across variant readings (cf. fol. 115 b), entered (probably by the same hand) in yet smaller letters between the lines; on fol. 116 a, there is a stanza written in the upper margin, which is meant to be added after 1. 162. 15, and which is found, otherwise, only in K₁, in other words is an interpolation peculiar to S₁, K₁. Many of the marginal additions are glosses, which are rather numerous in the first 15 (extant) folios, evidently notes made from some commentary by a student who intended making a careful study of the text. In a few places—perhaps about half a dozen—corrections have been made with yellow pigment. Some of the adhyāyas bear (serial) numbers, written probably by a different hand; the first (legible) figure that we come across is 43, corresponding to adhy. 32 of our edition, involving a difference of 11 in our enumerations of adhyāyas! The last adhyāya number noted in this parvan is 100, corresponding to our adhy. 87: the difference between our enumerations thus rises to 13 in 55 adhyāyas. The Purāṇic raconteur is here called, throughout, Sūta, not Sauti. Moreover, the prose formula of reference generally omits उवाच

(resp. ऊजुः), and gives, as in S MSS., merely the name or designation of the speaker, such as वैशंपायनः. However, from the fact that towards the middle and end of the parvan, the full forms containing उवाच (resp. ऊजुः) do occur sporadically, e. g. 1. 94. 64 (fol. 73 a); 98. 1 (fol. 75 b); 99. 36 (fol. 77 a) etc.: it follows that the usual वैशंपायनः etc. are only abbreviations. The names of the sub-parvans are generally added, in the colophons, agreeing mostly with the corresponding divisions of our edition. The extant fragment begins (fol. 24 a) with the words कश्यपः । विदिता चास्य सङ्कल्पमिर्मे (cf. v. 1. 1. 26. 10). — A facsimile of the folio (154) containing the end of the Ādi and the beginning of the Sabhā is given, facing p. 880.

Ko

Poona, Bombay Government Collection (deposited at the BORI), No. 229 of 1895-1902. Folios 181, with about 15 lines to a page; size 14.7″ × 6.7″. Devanāgarī characters; dated V. Samvat 1795 (ca. A.D. 1739). Old Indian paper.

The MS. contains the first three parvans written in the same hand, the date coming at the end of the Araṇya. The writing is clear and fairly correct; a few corrections of scribe’s errors are noted in the margin, probably by the same hand; otherwise the margins are clean. The colophons give adhyāya numbers sporadically, and names of adhyāyas, sub-parvans or upākhyānas generally. On the last folio (181) of the Ādi is given, in different hand, a list of major parvans with the corresponding number of their adhyāyas and stanzas, in a tabular form.

K₁

London, India Office Library, No. 3226 (2137). Folios 169, with about 33 lines to a page; size 16¼″ × 9″. Devanāgarī characters; dated (possibly) 1783 A.D. Indian Paper.

A moderately trustworthy, though somewhat modern and very incorrect transcript of a Śāradā exemplar. Even the outward form and get-up of this MS. are suggestive of Kaśmīrī origin. The lines of writing, as in Śāradā (bhūrjapatra) MSS. run parallel to the narrow side of the folio. The signatures in the margin are like those found in Kaśmīrī books. The numerous clerical errors, which disfigure every page, betray the writer to be a professional scribe, not thoroughly familiar with the awkward Śāradā script, and still less so with the language of the text, easily misled by the deceptive similarity between certain letters of the Śāradā and Devanāgarī alphabets. He frequently writes म for व (e. g. मंजकुले for वंजकुले); ड़ for त and व for ध (e. g. जुबा for तथा कृष्या); श for म (e. g. प्रकाश्मं for प्रकाशं) or for च (e. g. पांचालीं for पांचालौ); (e. g. क्षुणी for क्षणी); व्य for ध; उ for ड, ड for त, त् for न (e. g. आतुखुरे, सत्पमः and मेडिअल ड़ for subscript त; म for त (e. g. मभिः for मतिः ); ड़ for व्य (e. g. बिष्टा for विष्या). Margins are clean; very occasional corrections, in the body of the text, by yellow pigment. The pages from 42 to 45 are left blank, while 41 b and 46 a contain only a few lines of writing. Besides Ādi, the codex contains also Virāṭa, Bhīṣma and a portion of Anuśāsana (Dānadharma), breaking off at the first half of stanza 39 of adhy. 83 of the Bombay ed. According to statements at the end of the Bhīṣma and the beginning of the Anuśāsana, the MS. was written in V. Samvat 1839 (ca. A. D. 1783), by a Brāhmaṇa named Gopāla, residing in Lakṣmīmātha; but the writing of the volume is not quite uniform. It is, therefore, uncertain, in my opinion, whether the Ādi was written by this same Gopāla, in the said year; contra Eggeling, Catalogue of the Skt. MSS.

in the Library of the India Office, Part VI (1899), p. 1158, who regards the entire volume as written by the same scribe. The colophons, which are short, sporadically give the adhyāya numbers. This is the only MS. of the Ādi belonging to a European Library that was available for collation at the Institute and used for this edition! — The reference लो ह भिद् before stanza 8 of adhy. 1 indicates the intention of the scribe to “illuminate” the MS. by writing the alternate letters (म, णं, बाव), which are missing, in red ink.

K₂

Poona, Bombay Government Collection (deposited at the BORI), No. 182 of 1891–95. Folios 296 (of which 220, 226–30, 232–33, 239–40 appear to be written by a different hand), with about 11 lines to a page; size 10.8″ × 4.8″. Devanāgarī characters (with sporadic pṛṣṭhamātrās); dated V. Saṁ. 1694 (ca. A.D. 1638). Indian paper.

Marginal corrections, as also other corrections in the body of the text, are made by using yellow pigment; the colophons give names of sub–parvans, adhyāya names, and adhyāya numbers sporadically. In the marginal notes one occasionally comes across variants and glosses, and additional passages from MSS. of the central sub–recension (γ). The first folio and a part of the second (the latter stuck on to the original torn) are written in a different hand. On fol. 186 b, three lines are left blank by the scribe. After the four stanzas of “phalasṛuti” mentioned on p. 879, there follow two stanzas of the Parva– saṁgraha, giving the number of adhyāyas (218) and ślokas (8984) and, finally, the date: संवत् १६९४ वर्षे माघशुद्धि १० रवी लिखितमिदं.

K₃

Baroda, Oriental Institute Library, No. 632. Folios 407. Devanāgarī characters; dated V. Saṁvat 1575 (ca. A.D. 1519). Old Indian paper.

This MS. is from Gujarat. At the end of the MS. is given the date: Saṁvat 1575, śrāvaṇa, dark half, 5th day, Abhinandana. MS. written by Nāñjika, son of the Nagar Pandit Kālidāsa of village Kāndalāja, under Saṁkhetakapura (modern Saṅkheda, in Baroda State). For further details, see the colophon given on p. 879.

K₄

Poona, Bombay Government Collection (deposited at the BORI), No. 565 of 1882–83. Folios 237 (not counting the suppl. folios), with about 15–16 lines to a page; size 14.9″ × 6″. Devanāgarī characters; dated Śaka 1616 (ca. A.D. 1694), at the end of one of the subsequent parvans. Old Indian paper.

A carelessly written complete MS., with व for ल, throughout, which is a Southern trait; written by one hand, but preserved in the Collection in two bundles numbered 565 and 566. Supplementary folios at 2, 114, 150, 151, 205 include certain long passages (some from Southern sources), copied by the same hand; notable among them being the Brahma–Gaṇeśa interpolation, whose point of insertion is indicated by a small mark made in the body of the text, and the marginal remark अत्र शेषपत्रभेदं (cf. v. 1. 1. 53). There are some excerpts in margins, intended as glosses. Marginal additions of lines and stanzas are frequent only in the first 35 folios, afterwards few and far between. Corrections are made with yellow pigment. Colophons frequently contain adhyāya names, sub–parvan names, but no adhyāya number. The copyist was Gaṇeśa, son of Trimbaka,

K⁵ Lahore, Dayanand Anglo-Vedic College Library, No. 1. Folios 28 (numbered 1–7 and 9–29), with about 11–13 lines to a page; size 12″ × 6″. Devanāgarī characters, (said to be) about 350 years old. Paper. This MS. is incomplete, ending with 1. 3. 152. It was collated at the Visvabharati, up to 1. 2. 40, and was then reported to be missing.

K⁶ Poona, Bombay Government Collection (deposited at the BORI), No. 209 of 1887–91. Folios 386, with about 8–10 lines to a page; size 12.2″ × 5.8″. Devanāgarī characters. Partly old Indian paper and partly modern European paper. Folios 359 to end are of different paper (modern European, with water-marks) and are written by a different hand. In the margin, corrections of scribe’s errors, additional lines and stanzas (some of them probably omitted while copying), and various readings, which are decidedly more numerous in the beginning. On some folios (after fol. 105) yellow pigment has been used for correction. Here and there, lacunae mark the syllables which the scribe could not decipher, or which were missing in the exemplar. The colophons generally give the adhyāya or sub-parvan names; the adhyāya numbers were added afterwards, perhaps by a different hand, and are often crowded out or squeezed in with difficulty. — Collated up to the end of adhy. 2 only.

N¹ MS. in Nepāli characters from Nepal, in private possession. No specifications of the MS. (such as measurements, number of folios etc.) are available. It begins with a short eulogy (praśasti) of king (śrīmān bhūmahendra) Jaya- simhārāma, at whose bidding the MS. was copied. For a king of that name we have the date (Nepālī) Sam. 516 (ca. A. D. 1395). In the praśasti, he is stated to have built (?) a temple of Paśupati in Nepal. Collations of the MS. were kindly supplied by Rajaguru Pandit Hemaraj, C.I.E., D.P.I., Nepal, who had it collated for the Institute, by local Pandits. — The praśasti reads: ॐ नमो भगवते वासुदेवाय । ॐ नमः प्रजापतयेभ्यः । ॐ नमः कृष्णद्वैपायनाय । नारायणं नमस्कृत्य नरं चैव नरोत्तमम् । देवीं सरस्वतीं चैव ततो जयमुदीरयेत् ॥ वर्णप्रचाराय श्रमेध । [ सरवार्थे ] चिन्तामणिकल्पवृक्षो विराजते श्रीजयसिंहरामः ॥ मत्वा तत्क्षणमङ्कुरं जगदिदं विद्युल्लताचञ्चलं आत्मानं विनिवेश्य धर्मसरणीं दानादिसत्कर्मसु । येनाप्यस्तमत्सरत्वेन मनसा नित्यं हरिं ध्यायता [ ? श्री ]भारते लेखितम् ॥ कायेन वाचा मनसा च यद्यदपापं कृतं तस्य विनाशनाय । सुध्राद्वया श्रीजयसिंहरामो व्यलीलेखद्धरान्तमुत्तमं तत् ॥ प्रेते विष्णुसमागममाय नरकद्वारोपरोधाय च कुट्टङ्कूरयमात्मदर्शनप * । एतद्बुध्ययति स्म पुण्यनिलयं श्रीभारतालस्यं महत् पारः श्रीजयसिंहरामसुकृती सत्वार्थचिन्तामणिः ॥

कायवाक्कायमानैर्विष्णो भर्शि करोति यः । स एव भगवान्विष्णुस्तस्मै नित्यं नमो नमः ॥ दानेनादिलसुद्धं त्रिवशपति [गुरुं] प्रक्षया यो विजित्य सौन्दर्येण्णालयगं रजनिकरमपि हेपयन्नात्मदीप्ता । दोर्दैपेर्णापि भीभं दिवसकरमपि स्पर्षमानः प्रतापेः स श्रीमान्भूभहेन्द्रो जयति पशुपतिस्थापनाम्नेन्द्रम् ॥ Collations begin at adhy. 3. — Collated in Nepal.

N₂ MS. in Nepālī characters from Nepal; in private possession. No further details of the MS. are available. Collations of the MS. were kindly supplied by Rajaguru Pandit Hemaraj (Nepal), who had it collated for the Institute by local Pandits. — Collations begin at adhy. 3.

N₃ From a private library in Nepal. Nepālī characters, written in ink on palm-leaf. Besides the Ādi, the MS. contains also Sauptika-Aiṣṭka and Viśoka-Strī. The last folio of this bundle bears the date (Nepālī) Saṃ. 632 (ca. A.D. 1511). Sent to the Institute for collation, through the kind offices of Rajaguru Pandit Hemaraj (Nepal). The MS. was returned to the owner after a hurried collation, and further details of the MS. are unfortunately not available. — Collations begin at adhy. 14.

V₁ Nepal, Darbar Library, No. 1364. Maithilī characters; dated La. Saṃ. 411 (ca. A.D. 1528). Palm-leaf. No further details of the MS. are available. The MS. has two lengthy lacunae: 1. 68. 74 to 92. 13, and 96. 37 to 127. 21. — Collations of the MS. were kindly supplied by Rajaguru Pandit Hemaraj (Nepal), who had it collated, for the Institute by local Pandits.

B₁ Santiniketan, Visvabharati Library, No. 1. Folios 207; size 28·2″ × 2″. Bengali characters; dated Śaka 1662 (ca. A.D. 1740). Palm-leaf. The name of the scribe, as given in a stanza following the last colophon, is Kṛṣṇarāmadvija. — Collated at the Visvabharati.

B₂ Santiniketan, Visvabharati Library, No. 258. Folios 82, with about 5-6 lines to a page; size 25½″ × 2½″. Bengali characters. Palm-leaf. This fragmentary MS. breaks off at 1. 43. 13, in the middle of the Āstika. — Col- lated at the Visvabharati.

B₃ Santiniketan, Visvabharati Library, No. 782. Folios 199; size 19½″ × 4½″. Bengali characters; dated Śaka 1681 (ca. A.D. 1759). Paper. Name of the copyist, as given at the end of the MS., is Khelārāma Vipra. — Collated at the Visvabharati.

B₄ Santiniketan, Visvabharati Library, No. 413. Folios 164, with about 7–9 lines to a page; size 20″ × 5·2″. Bengali characters. Paper. This fragmentary MS. breaks off at 1. 90. 88, in the middle of Sambhavaparvan. — Collated at the Visvabharati.

B₅ Dacca, University Library, No. 485. Folios 366, with about 7 lines to a page; size 17″ × 3¾″. Bengali characters; dated Śaka 1703 (ca. A.D. 1786). Much faded old Indian yellow paper. The MS., which is well preserved and neatly written, containing a few corrections noted in the margins, was obtained from Malatinagar, Bogra District, Bengal. Collations begin at adhy. 3. — Collated at the Visvabharati.

B₅ Dacca, University Library, No. 735. Folios 346, with about 7 lines to a page; size 19″ × 4½″. Bengali characters. Old Indian yellow paper. Appearance, as well as the script of this MS. (which was obtained from Ula Bisnagar, Nadia District, Bengal), is somewhat more modern than that of B₅; belongs apparently to the beginning of the 19th century. Neatly written and fairly correct; contains occasional brief glosses on margin, apparently by the same hand as that of the copyist. — Collations begin at adhy. 54. Collated at the Visvabharati.

Dᴀ1 Poona, Bombay Government Collection (deposited at the BORI), No. 30 of A 1879–80. Folios 416, with about 7–10 lines to a page; size 15¼″ × 6¾″. Devanāgarī characters. Old Indian glossy paper. Text with commentary of Arjunamiśra; written neatly but extremely corrupt and unintelligible in places, on account of the scribe’s inability to read the exemplar correctly. The MS. has many short and long blanks in the text, which support the latter surmise. It has very few glosses and corrections, but a large number of variants noted in the margin. The text is written in three strips: the upper and lower ones comprise the commentary, while the central band, which has generally a still wider margin, is the (epic) text. The references to speaker (such as वैशंपायन उवाच) and colophons are written in red ink. The colophons give generally adhyāya and sub-parvan names. Ślokas are generally numbered; adhyāyas are almost regularly numbered from adhy. 45 to 109. The MS. is almost consistent in writing जन्मेजय (for जनमेजय) उवाच. Punctuation is most imperfect. In the numbering of the folios, number 2 is repeated.

Dᴀ2 Poona, Bombay Government Collection (deposited at the BORI), Viśrāmbāg I, No. 468. Folios 415, with about 10 lines to a page; size 15·7″ × 6·6″. Devanāgarī characters; dated V. Saṁ. 1676 (ca. A.D. 1620). Indian paper. Text with commentary of Arjunamiśra. The MS. is from Dambal, a Jagir in the Kanarese Districts of the Deccan, and the last folio contains several stanzas in praise of a certain Gopālabhaṭṭa, a learned Pandit of great fame, who got the MS. written:

भावकाङ्क्षितमार्थनासुपनयन्नद्यर्थि योग्यं फलम् साधुप्रत्ययनिर्मलप्रकृतिभागाध्यात्मसत्कर्मणि । स्फूर्ज्जन्सत्त्वनिसर्गतोऽप्ययमेते गोपालविग्रहौ यत्स्वां पश्यति तेन जीवनमिदं लोकान्तरं कल्पितम् ॥ सुखया पीयूषवर्षी निरतिशयदयो दूरदर्शी वर्षीयवा- नन्तर्वाणीप्रवीणीः परमसुगतया वन्दनीयो वदान्यः । कुर्वोणः कीर्त्तिकाख्या हरममृतकरं पारदं शारदं यः श्रीमान्गोपालशर्मा जगति विजयते पुण्यकर्माग्रगण्यः ॥ निर्देशास्सस्य नानाविधविबुधगुणग्राहिणः शिक्षितानां संतोषोत्कर्षबीजात्सरसमुनिकालकल्पकल्याणसंवत् । मित्रे औमाङ्गले प्रतिपदि सहस्रः सुदृढं सौघमोदः पर्वाचे चारुटीकासुघटितमखिलं ले रु[व]चित्रै शुभं स्तात् ॥ प्रणेता शाखाणां निखिलनिगमानां प्रतिनिधि- धुरीणः खे धर्मे गुरुगुरुभ्रदराजसद्दशः । कृती शान्दे शाखे धरणिपतिपूजितमति- र्वंशी गोपालोऽयं जयति किल भट्टः पृथुयशाः ॥

The date of the MS. is given as a chronogram corresponding to V. Samvat 1676 (ca. A.D. 1620). Double daṇḍas in red ink are inserted indiscriminately in the middle of the text. The writing, which is full of mistakes, is uniform but not neat. No corrections are, however, to be seen, the MS. being, perhaps, not much used. Notwithstanding the fact that this MS. agrees, page for page, with D₄, there are many small differences between them; neither can be a direct copy of the other; they must go back to a more remote common source. It appears to be older, and is less corrupt, than D₄. In the numbering of the folios, figure 1 is repeated. The colophons contain the names of adhyāyas and sub-parvans generally; but śloka numbers or adhyāya numbers only sporadically. The MS. has a few blanks in the text and commentary.

D n2 Mysore, Oriental Library, No. 1064. Folios 448, with about 22 lines to a page; size 15¼″ × 6½″. Devanāgarī characters; dated V. Samvat 1864 (ca. A.D. 1808). Paper. Text with commentary of Nīlakanṭha.

D n3 Poona, Bombay Government Collection (deposited at the BORI), No. 234 of 1895–1902. Folios 683, with about 9 lines to a page; size 15·2″ × 7·2″. Devanāgarī characters. Thick Indian paper. Text with commentary of Nīlakanṭha. Bold and clear letters; generally correct; margins are almost clean. Ślokas and adhyāyas are throughout numbered. As in D n1, blanks were filled with invocations and names of various gods. The lemmata do not always fit the (epic) text. Colophons and the references to the speakers (and for some initial folios even daṇḍas) are in red ink, but only up to fol. 470.

D r1 Tanjore, Saraswathi Mahal Library, No. 1246. Folios 448, with about 11 lines to a page; size 15″ × 6½″. Devanāgarī characters. Paper. Text with the commentary of Ratnagarbha. — Collations end at adhy. 2. Collated at Tanjore.

D r2 Tanjore, Saraswathi Mahal Library, No. 1199. Folios 306, with about 10–13 lines to a page; size 16″ × 6½″. Devanāgarī characters. Paper. Text with the commentary of Ratnagarbha. — Collations end at adhy. 2. Collated at Tanjore.

D r3 Tanjore, Saraswathi Mahal Library, No. 1313. Folios 366, with about 11–13 lines to a page; size 16″ × 6½″. Devanāgarī characters; dated Śaka 1623 (ca. A.D. 1701). Paper. Text with the commentary of Ratnagarbha. MS. dated, in the Śaka year 1623 (current) corresponding to Vṛṣa, Sunday the 13th (of the bright half) of the month of Āṣāḍha. — Collations end at adhy. 2. Collated at Tanjore.

D r4 Tanjore, Saraswathi Mahal Library, No. 1339. Folios 108, with about 11–22 lines to a page; size 16″ × 6½″. Devanāgarī characters. Paper. Text with the commentary of Ratnagarbha. This fragment contains only about 90 adhyāyas of this edition. The number of lines on each folio fluctuates with the amount of commentary which each folio contains, and which of course, varies considerably. — Collations end at adhy. 2. Collated at Tanjore.

D₁

Poona, Bombay Government Collection (deposited at the BORI), No. 29 of A 1879–80. Folios 30, with about 16–17 lines to a page; size 12″ × 7·15″. Devanāgarī characters. Fine cream-coloured paper.

For the first 140 folios or so, colophons and part references to speakers (such as वैशंपायन³) are generally in red ink; then occasionally. Colophons sporadically give adhyāya or sub-parvan name and number of adhyāyas (especially towards the end of the parvan); stanzas are not numbered. The MS. is generally correct; margins are clean. — This is a complete MS. of Mbh., copied apparently from different exemplars; some parvans have the commentary of Nīlakaṇṭha, while others contain some old text tradition (e. g. “M” of the Tentative Edition of the Virāṭaparvan). The MS. is of modern date, being written on paper with water-marks. Some of the parvans bear dates at the end, but these seem to be copied from the originals; thus, Śānti (Mokṣadharma) has Śaka 1680, while Dānadharma has Śaka 1675. The last parvan bears the date: १६७७ युगनागम्र्वस्सरै.

D₂

Tanjore, Saraswathi Mahal Library, No. 1152. Folios 340, with about 10 lines to a page; size 13″ × 5½″. Devanāgarī characters; dated V. Saṁ. 1654 (ca. A.D. 1598). Paper. The MS. was written on Friday the 13th of Āṣāḍha śuddha of V. Saṁ. 1654, at Benares by a Brāhmaṇa called Govinda, and belonged to Vāsudevabhaṭṭa. — Collated at Tanjore.

D₃

Tanjore, Saraswathi Mahal Library, No. 1360. Folios 120, with about 10 lines to a page; size 14″ × 6¼″. Devanāgarī characters. Paper. Incomplete, breaking off at the end of adhy. 76 (of our edition), in the middle of the Yayāti episode, which, in this MS. (as in S MSS.), precedes the Śakuntalā episode. — Collated at Tanjore.

D₄

Tanjore, Saraswathi Mahal Library, No. 1126. Folios 235, with about 11 lines to a page; size 16″ × 6¾″. Devanāgarī characters. Paper. Many corrections and additions, the MS. being compared with another of the Southern recension, extracts from which have been written out on the margin, and on supplementary folios. — Collated at Tanjore.

D₅

Lahore, Dayanand Anglo-Vedic College Library, No. 4. Folios 246, with about 12–14 lines to a page; size 12″ × 5″. Devanāgarī characters; dated V. Saṁ. 1858 (ca. A.D. 1802). Paper. — Collated at the Visvabharati.

D6 Tanjore, Saraswathi Mahal Library, No. 1223. Folios 293, with about 12 lines to a page; size 14" × 6½". Devanāgarī characters. Paper. An old MS., but with clear and legible writing; well preserved. — Collations end at adhy. 53. Collated at Tanjore.

D7 Tanjore, Saraswathi Mahal Library, No. 1269. Folios 262, with about 11 lines to a page; size 14" × 5⅞". Devanāgarī characters. Paper. Clear and legible writing; well preserved. — Collations end at adhy. 53. Collated at Tanjore.

D8 Tanjore, Saraswathi Mahal Library, No. 1329. Folios 196, with about 16–18 lines to a page; size 15¼" × 7". Devanāgarī characters. Paper. A comparatively modern MS. — Collations end at adhy. 2. Collated at Tanjore.

D9 Tanjore, Saraswathi Mahal Library, No. 1176. Folios 279, with about 11 lines to a page; size 15¼" × 5¾". Devanāgarī characters. Paper. Fol. 1–2 are badly damaged. — Collations end at adhy. 2. Collated at Tanjore.

D10 Tanjore, Saraswathi Mahal Library, No. 1293. Folios 352, with about 10 lines to a page; size 13½" × 5¼". Devanāgarī characters. Paper. Last leaf torn; well-preserved; clear and legible writing. — Collations end at adhy. 2. Collated at Tanjore.

D11 Tanjore, Saraswathi Mahal Library, No. 1340. Folios 290, with about 11–13 lines to a page; size 14" × 5½". Devanāgarī characters. Paper. Written, perhaps, by four different scribes. — Collations end at adhy. 2. Collated at Tanjore.

D12 Tanjore, Saraswathi Mahal Library, No. 1373. Folios 21, with about 12 lines to a page; size 14¼" × 6". Devanāgarī characters. Paper. Incomplete, containing only the first two adhyāyas. — Collated at Tanjore.

D₁₃ Poona, Bombay Government Collection (deposited at the BORI), Viśrāmbāg II, No. 191. Folios 221, with about 13 lines to a page; size 14·25″ × 6·05″. Devanāgarī characters. Old Indian glossy paper.

Fragmentary, folios 1-7 wanting; begins with भूणदृश्याकृतं पादं (1. 1. 205). Text very similar to Arjunamiśra's; neatly written and generally correct; marginal corrections are few and far between. Adhyāya names or sub-parvan names are given, but the ślokas or adhyāyas are not numbered. The reference to narrators is, at first, given at random as सैमितिश्चाव and सूत उ°, but then the scribe settles down to सूत उ°. The collations are given, as a matter of fact, only from 1. 1. 205 to the end of adhy. 2.

D₁₄ Poona, Bombay Government Collection (deposited at the BORI), Viśrāmbāg II, No. 266. Folios 1-121 (fol. 122-189 of this MS. are found under Viśrāmbāg II, No. 86), with about 15 lines to a page; size 18″ × 6½″. Devanāgarī characters. Old Indian unglazed paper.

MS. No. 267 of the same Collection is of Sabhā with commentary and written by the same hand. — Folio 79 is wanting. Carefully written, has very few corrections, which are made by use of yellow pigment, and a few marginal additions; gives, as a rule, numbers to ślokas and adhyāyas; also mentions generally sub-parvan and adhyāya names, — Collated up to the end of adhy. 2 only.

T₁ Melkote, Yadugiri Yatiraj Math MS. (without number). Folios 195, with about 11 lines to a page; size 16·1″ × 2·3″. Telugu characters. Palm-leaf.

MS. kindly lent by His Holiness the Yatiraj Swami. Contains Ādi and Sabhā, written probably by the same hand; writing clear and correct; adhyāya ends are shown by a small floral (or spiral) design engraved in the right and left margins of the MS.; adhyāyas are regularly numbered, but not the ślokas. It is one of the few Southern MSS. which contain the (Northern) salutatory stanza नारायणं नमस्कृत्य etc

T₂ Tanjore, Saraswathi Mahal Library, No. 11865. Folios 400, with about 6 lines to a page; size 21″ × 1¾″. Telugu characters. Palm-leaf.

Fragmentary; breaking off at the end of our adhy. 181 (corresponding to its adhy. 140); from adhy. 182, it is replaced in our critical apparatus by the next MS. T₃. — Collated at Tanjore.

T₃ Tanjore, Saraswathi Mahal Library, No. 11809. Folios 164, with about 12 lines to a page; size 29½″ × 2¼″. Telugu characters. Palm-leaf.

An old MS., containing the first five parvans; script small, but clear. — Collations begin at adhy. 182; used only to supplement the portion missing in T₂. Collated at Tanjore.

G₁ Melkote, Yadugiri Yatiraj Math MS. (without number). Folios 110, with about 16–21 lines to a page; size 18·7″ × 1·8″. Grantha characters. Palm-leaf. Leaves are very brittle, and worm-eaten in places; large pieces have broken off, leaving many lacunae. The holes for the string have enlarged, perhaps from constant use, destroying some parts of the text, written round them.

G₂ Melkote, Yadugiri Yatiraj Math MS. (without number). Folios 202, with about 15–17 lines to a page; size 14·5″ × 2·1″. Grantha characters. Palm-leaf. The MS. contains the first 4 parvans: Ādi, Sabhā, Aranya and Virāṭa, written probably by the same hand. Slightly worm-eaten; but, on the whole, a well preserved old MS. with clear and legible writing.

G₃ Tanjore, Saraswathi Mahal Library, No. 11823. Folios 316, with about 10 lines to a page; size 16¾″ × 1¾″. Grantha characters. Palm-leaf. — Collated at Tanjore.

G₄ Tanjore, Saraswathi Mahal Library, No. 11838. Folios 477, with about 6 lines to a page; size 19″ × 1¾″. Grantha characters. Palm-leaf. An old and well-preserved MS., with clear and legible writing, but many corrections. — Collated at Tanjore.

G₅ Tanjore, Saraswathi Mahal Library, No. 11851. Folios 320, with about 8 lines to a page; size 19″ × 1¾″. Grantha characters. Palm-leaf. The MS. contains the Sabhā also, probably written by the same hand. A well- preserved old MS., with clear and legible writing. — Collated at Tanjore.

G₆ Tanjore, Saraswathi Mahal Library, No. 11860. Folios 324, with about 8 lines to a page; size 18¼″ × 1¾″. Grantha characters. Palm-leaf. MS. written by Kāśipati, on the 22nd of the month of Kumbha, in the year Krodhi. — Collated at Tanjore.

G₇ Melkote, Yadugiri Yatiraj Math MS. (without number). Folios 217, with about 12–14 lines to a page; size 19·2″ × 2″. Grantha characters. Palm-leaf.

Clear and legible writing; worm-eaten in places. Being a conflated MS, it was discontinued after adhy. 2. It is one of the few Southern MSS. which begin with the (Northern) salutatory stanza, नारायणं नमस्कृत्य etc., added later in the narrow upper margin of the first folio, in very fine writing. Its place of insertion is indicated by a "haṁsapāda", inserted immediately after its first maṅgala stanza (9*). — Collated up to the end of adhy. 2 only.

M₁ MS. from the private library of the Chief of Idappalli, Cochin. Folios 79. Malayālam characters. Palm-leaf. — Secured on loan and got collated kindly by Prof. K. Rama Pisharoti. No further details of the MS. are available. Incomplete MS., ending with adhy. 53, the final adhyāya of the Āstikaparvan. — Collated at Sanskrit College, Tripunittura, Cochin.

M₂ Cochin, State Library, No. 5. Folios 122. Malayālam characters. Palm-leaf. The MS. was returned to the Cochin State Library after collation. No further details of the MS. are available. Incomplete MS., ending with adhy. 53, the final adhyāya of the Āstikaparvan.

M₃ Cochin, State Library, No. 1. Folios 166, with about 12-13 lines to a page; size 19.9″ × 1.6″. Malayālam characters; dated Kollam 1013 (ca. A.D. 1838). Palm-leaf. A modern MS., perhaps less than 100 years old; adhyāya numbers and śloka numbers are given. The adhyāya ends are shown by a floral design, inscribed in the margins.

M₄ MS. from the private library of Kallenkara Pisharam, Cochin. Folios 57. Malayālam characters. Palm-leaf. The MS. was returned to the owner immediately after collation. No further details of the MS. are available. Incomplete, ending with adhy. 53, the final adhy. of the Āstikaparvan.

M₅ MS. from the Paliyām MSS. Library, Cochin. Folios 245. Malayālam characters. Palm-leaf. — Secured for collation by courtesy of Mr. P. Anujan Achan, now Superintendent, Archæological Department, Cochin State.

M₆ MS. from the private library of Nareri Mana, Malabar. Folios 163, with about 10 lines to a page; size 18″ × 1.6″. Malayālam characters. Palm-leaf.

Incomplete MS., adhy. 1-53 wanting (i. e. begins with the Ādivaṃśavataraṇa sub-parvan); writing clear and legible; generally correct; margins are clean. — Collations begin from adhy. 54.

M₁ MS. from the private library of Avanapparambu Mana, Cochin. Folios 170, with about 10 lines to a page; size 20·5″ × 1·8″. Malayāḷam characters. Palm-leaf. Clear and legible writing; leaves are in perfect preservation, not a single leaf being worm-eaten; probably not very old. — Scribe has left many blanks in the writing space, whenever the surface of the leaf was uneven or rugged. — Collated from adhy. 54.

M₂ Malabar, Poomulli Mana Library, No. 297. Folios 183, with about 10 lines to a page. Malayāḷam characters; dated Kollam 1017 (ca. A.D. 1842). Palm-leaf. Collated from adhy. 54.

In view of the great unevenness of the critical apparatus, and of the consequent difficulty likely to be experienced by readers using the critical notes (printed at the foot of the page) in ascertaining what manuscripts have been added, discontinued, or discarded at different points of the text, I append, on the following page, a table which shows at a glance just what manuscripts have been actually collated for different portions of the text. Even the larger lacunae of the manuscripts, which cannot be easily ascertained, have been exhibited in this table. Only such (small) omissions have been, as a rule, ignored as are specifically mentioned in the footnote itself pertaining to the particular stanza, and which are therefore brought to the notice of the reader as soon as he reads the footnote.

TABLE SHOWING THE MSS. COLLATED FOR DIFFERENT PORTIONS OF THE TEXT ²

Testimonia

As testimonia, or aids of a partial or subsidiary character, there are available, besides the numerous commentaries, the following three important epitomes of the eleventh century: (i) the Javanese adaptation Bhāratam (ca. A.D. 1000), (ii) the Telugu adaptation Andhra Bhāratamu by the Telugu poet Nannaya Bhaṭṭa (ca. A.D. 1025), and (iii) the Sanskrit adaptation Bhāratamañjarī by the Kaśmīri poet Kṣemendra (ca. A.D. 1050); as also an important Persian rendering made some centuries later (ca. A. D. 1580) at the instance of that enlightened and sagacious Emperor of India with catholic sympathies, the great Akbar.

The commentaries collated for this edition are dealt with below, under the Devanāgarī versions. Here it will suffice to observe that, even when accompanied by the (epic) text, the commentaries are, for reasons which will be explained later on, evidence only for the actual lemmata and the pāṭhāntaras cited. The absence of commentary on a stanza or a group of stanzas or even on an adhyāya is, in general, no proof that that particular passage was lacking in the text used by the commentator. For, clearly, his text may have contained the passage in question, but he may not have deemed it necessary to comment upon any portion of it. Nevertheless when the commentary ignores a lengthy and difficult passage, then there is a strong presumption that the text of the commentator did not contain the passage. A case in point is the Kaṇīkanti, a passage of 186 lines, which is entirely ignored in Devabodha’s commentary (but hase voked lengthy comments from both Arjunamiśra and Nīlakaṇṭha), and which is missing in the Kaśmīri version.

As regards the old Javanese adaptation, from the reports of Dutch scholars¹ who have studied the original Javanese text, it appears that only eight out of the eighteen parvans of the Mahābhārata have been traced so far; namely, Ādi, Virāṭa, Udyoga, Bhīṣma, Āśramavāsa, Mausala, Mahāprasthāna and Svargārohaṇa. Three of these (Āśramavāsa, Mausala, Mahāprasthāna) were the subject of a doctor dissertation, submitted to the Leyden University by Dr. H. H. Juynboll, as early as 1893. The Javanese original was edited by the doctor in Roman characters and rendered into Dutch. Thirteen years later (1906) the same scholar published the text of the Ādi (with different readings) in Roman transcript.² Of the old Javanese Ādiparvan, only a few episodes have been as yet translated, to wit: the Parvasaṃgraha, the Pauṣya, the Amṛtamanthana, the story of Parikṣit and the Sauparṇa. Unfortunately these translations are not available in India; at least they were not available to me.

The chief value of the Javanese adaptation for us lies in the fact that throughout the old Javanese text are scattered Sanskrit quotations, which appear to have “served as landmarks for writers and hearers or readers”. The text prepared by Dr. Juynboll, which is based upon eight manuscripts, is reputed to be very accurate. But it is admitted that the Sanskrit excerpts in the extant Javanese manuscripts are extremely corrupt, and it is a

question how far the conjectural restorations by the editor correctly represent the original readings. It seems to me likely that in his reconstructions Dr. Juynboll was to a certain extent influenced by the wording of the Vulgate, which is certainly not always original. To give only one instance. On p. 70, the Javanese manuscripts read (in the Śakuntalā episode) : paripațyādayaḥ sunu, hāraṇireṇuguṇditaḥ /, which is corrupt; it conveys no sense. In the text the editor gives: pratipadya padā sūnur, dharaṇireṇu gunthitaḥ /, which is nearly the reading of the Calcutta edition (3040). Though the Javanese manuscripts are palpably corrupt, yet they have preserved the correct paripațya (for pratipadya of the Vulgate), which is the reading of the Śāradā and K manuscripts of our edition. We have here to thank the Vulgate for the pratipadya of Dr. Juynboll's text! Notwithstanding, that the period from which this adaptation dates is comparatively speaking recent, it yet precedes the known date of the manuscripts by several centuries and is hence of considerable importance for critical purposes, as a witness³ independent of and uninfluenced by the main line of our extant Indian witnesses. Most of the Sanskrit quotations of the Javanese text can be traced both in the Northern and the Southern recensions, as may be seen from our Appendix II, at the end of this volume, which contains a concordance of the Javanese extracts with the Critical Edition, the Calcutta Edition, and Sastri's Southern Recension. A few of the quotations are to be traced to the “additional” passages in the Northern manuscripts, but none to the specific Southern “additions”. The conclusion is inevitable that the text of the Sanskrit Ādiparvan used by the Javanese writers must have belonged to the Northern recension, a conclusion already suggested by the sequence of the Śakuntalā and Yayāti episodes, which is the Northern sequence. This does not necessarily mean that the entire Javanese Bhāratam represents the Northern recension. It is quite likely that some of the parvans utilized by the Javanese adapters belonged to the Southern recension. The late Mr. Utgikar⁴ was inclined to think that the Javanese Virāṭaparvan was of the Southern type. The point will have to be re-examined in the light of further evidence. The books were preserved and handed down separately; consequently the genesis of each parvan must be investigated separately. The Telugu adaptation, the Āndhra Bhāratamu,⁵ is a metrical epitome of the Mahābhārata, commenced by Nannaya Bhaṭṭa, a court poet of the Eastern Cālukya king Viṣṇuvardhana, who had his capital at Rajamundry, on the East Coast of India, and who appears to have ruled between 1022 and 1066.⁶ The torso of the Telugu rendering left behind by Nannaya, consisting of a version of the first two parvans and of a part of the third, was completed many years later by two other poets. Nannaya's version is valuable for the light it throws on the condition of the Southern recension—or, strictly speaking, of the Telugu version—in the eleventh century of the Christian era, especially in view of

the fact that Nannaya has included in his poem an accurate rendering of the Parvasam- graha, giving the number of ślokas in each of the parvans of his Mahābhārata.¹ The figure for the stanzas of the Ādi is 9984, which shows that the text used by Nannaya must have been substantially of the same size as that preserved in the extant Southern manuscripts. The poet is reported to have followed the original fairly closely. Notable is consequently his omission of Brahmā's visit to Vyāsa.² Curiously enough, the third old important epitome of the Mahābhārata which we possess, the Bhāratamañjarī by Kṣemendra,³ belongs to the same century as the two epitomes mentioned above, since this Kāśmīrī poet must also be assigned to the middle of the eleventh century.⁴ Bühler and Kirste have given in their Indian Studies, No. 2 (pp. 30 ff.), the results of a careful comparison of Kṣemendra's abstract with the Bombay text of the Mahābhārata. They show that Kṣemendra's text contains both additions and omissions as compared with the latter.⁵ Of the omissions they note: adhy. 4, 24, 45-48, 66, 94, 139, and parts of adhy. 141 and 197 of the Vulgate. Of these, adhy. 4 is, as pointed out by Bühler and Kirste, a short introductory chapter, a variant of adhy. 1; adhy. 45-48 are a repetition (with variations) of adhy. 13-15; adhy. 66 is a variant of the preceding adhyāya; adhy. 94 is a variant of adhy. 95 (prose), which is selected by Kṣemendra for his purpose;⁶ finally, stanzas 44 to end of adhy. 197 are a repetition of a part of adhy. 169. The reason for the omission of these adhyāyas is thus clear: they are mere repetitions. The remaining adhyāyas, which are missing and whose omission Bühler-Kirste could not account for, namely, adhy. 24, 139, and 141 (stanzas 1-19) are also missing in many of our Mahābhārata manuscripts and have accordingly been omitted in the constituted text as well. To these must be added the important omission of adhy. 140 of the Vulgate, the Kanīkanīti, which is likewise omitted by Kṣemendra, an omission which appears to have been overlooked by Bühler and Kirste. The collaborating authors felt justified in concluding that the omissions and additions "are just such liberties as any Kāvya poet would take in making a similar abridgement." They were also of opinion that the original cannot have differed very essentially from our current texts, that is, the Vulgate. This is correct up to a certain point. A comparison with the different versions shows that Kṣemendra's version agrees, as was to be expected, most closely with the Śāradā. On comparing the divisions of the Mañjarī with those given in Bombay or Calcutta editions of the Mahābhārata, Bühler and Kirste were struck by the fact that the Mañjarī divisions agreed better with the course of the narrative; and they give examples to show that the arrangement of the Mañjarī is more logical. That is quite natural, because the old Northern manuscripts, which this edition

follows, fully support the arrangement of the Mañjari, whereas the divisions adopted in the Vulgate are secondary and quite corrupt.

The Persian translation¹³ of the Mahābhārata, made in the reign of Emperor Akbar, being still unedited, could not be consulted. A very full account of this rendering has, however, been given by the late Dr. Sir Jivanji Jamshedji Modi in a paper read before the First Oriental Conference at Poona in November 1919 and published in the Annals of this Institute.¹⁴ Of all the Sanskrit works Akbar got translated, the Mahābhārata, it appears, had his most earnest attention.

Several eminent poets and scholars had a hand in translating the Great Epic of India into Persian. The Ā'īn-e-Akbarī gives the following names: Naqīb Khān, Maulānā ‘Abdu’l-Qādir Badāyūnī, and Shaikh Sultān of Thanesar, to which the Muntakhab-u’t-Tawārīkh adds the names of Mullā Sherī, and Shaikh Faizī (the brother of Abu’l-Fazl).

“Badaoni translated”, we are informed by Sir Jivanji,¹⁵ on the authority of contemporary chronicles, “two out of the eighteen sections. Mullā Sherī and Naqīb Khān did a part of the work and the rest was completed by Sultān Hājī of Thanesar. Shaikh Faizī converted their ‘rough translation into elegant prose and verse, but he did not complete more than two sections.’ Sultān Hājī, then revised these two sections and verse. Not only did he do so, but he also revised his work which formed a large share of the work.” Quoting Badāyūnī, Sir Jivanji continues: “The Hājī aforesaid revised these two sections, and as for the omissions which had taken place in his first edition, those defects he put right, and comparing it word for word was brought to such a point of perfection that not a fly-mark of the original was omitted”! The preface to this translation was from the pen of that gifted courtier of Akbar who has left us such an admirable account of the Emperor’s reign, Abu’l-Fazl. This Persian version appears to have been a free rendering of the original, made by Muslim poets and scholars at the Court of Akbar, to whom the sense of the original had been explained by Hindu pandits, under the orders of the Emperor.

There are numerous other vernacular abstracts of the Mahābhārata besides the Telugu abstract mentioned above, but most of them are of a late date. Moreover, they are all far too free to be of much use to us in reconstructing the text of the Mahābhārata.

Besides these abstracts and adaptations, there are parallel versions of certain passages or even of whole episodes to be met with in other works. Thus we have a parallel version of the Śākuntalā episode (adhy. 62 ff.), in the Padmapurāṇa;¹⁶ of the Yayāti episode (adhy. 71 ff.), in the Matsyapurāṇa;¹⁷ of the story of Ruru (adhy. 8 ff.), in the Devībhāgavata; of a portion of Samudramanthanā (adhy. 16 f.), again in the Matsyapurāṇa; of a portion of a cosmogonic passage (1. 60. 54 ff.), in the Rāmāyaṇa.

There is more distant connection between our Sauparṇa (adhy. 14 ff.) and the pseudo-vedic Suparṇādhyāya.¹⁸ Some of the stanzas of the Ādi are cited, with or without mention of the source, in the Tantravārttika of Kumārila Bhaṭṭa (e. g. our 1. 1. 209), as also in the Bhāṣyas of Ācārya Śaṁkara (e. g. our 1. 1. 37). A few of the sententious stanzas (e. g. our 1. 74. 1 ff.) recur, with variation, in Buddhist literature,¹⁹ while stray stanzas are to be found again in the Khilas of the Ṛgveda (e. g. our 1. 53. 22 f.)²⁰, the Manusmṛti²¹ (e. g. our 1. 3. 94) and the Bṛhaddevatā²² (e. g. our 1. 59. 12). One of our stanzas (1. 119. 6) has been cited in the Dhvanyāloka of Ānandavardhana, as by Maharṣi Vyāsa. There are probably many stanzas which remain to be identified. It is perhaps well to add in this place that a certain amount of caution is necessary in making any critical use of citations of stray Mahābhārata stanzas we meet with again in other works. We must, in the first place, bear in mind that most of the other works have yet to be properly edited. Even in critically edited texts we must take into account the various readings of the passage in question in the manuscripts collated. Then in the case of citations we must allow for failures of memory; since in ancient times the stanzas were almost invariably quoted from memory, and the quotation was never compared with the original. Moreover we must never forget that probably from time immemorial there have existed local versions of the Mahābhārata. The citations made even by very old writers were from these local versions. A citation by a writer of the eighth century or even the sixth century proves nothing for the Ur-Mahābhārata, that ideal but impossible desideratum; though the citation is far older than our manuscripts, it is evidence only for the text of the local Mahābhārata in the eighth, respectively the sixth century, notwithstanding that the differences between the various recensions and versions of the Mahābhārata must diminish as we go back further and further.

Pedigree Of Adiparvan Versions

Vyāsa’s Bhārata Ur-Mahābhārata

Explanation Of The Sigla Used In The Above Pedigree

N is the ultimate source from which all versions of the Northern recension are, directly or indirectly, derived.

ν is the lost archetype of the North-Western group, appreciably shorter than any of the other known versions (textus simplicior).

K is a specific Devanāgarī version allied to the Śāradā (or Kaśmīrī) version (sharply distinguished from other Devanāgarī versions), of which one MS. (K₁) is the direct copy of a Śāradā original. The version is largely contaminated from MSS. of the (central) sub-recension (Y), and in part, also from some unknown Southern sources. Exact provenance of the version is unknown.

Y is the intermediate (inflated) source from which all versions of the central sub-recension are derived (comprising the Eastern and Western groups), occupying a position intermediate between the North-Western and the Southern groups. It contains a considerable number of secondary additions (including repetitions), as also a very large number of verbal alterations and corruptions.

ε is the lost archetype of the Eastern group (comprising the Nepāli, Maithili and Bengali versions), which is free from the additions and alterations made later in certain Devanāgarī MSS.

S is the ultimate source from which all versions of the Southern recension are, directly or indirectly, derived and which is appreciably longer than N, and far more elaborate (textus ornatior).

σ is the lost archetype of TG, containing a large number of corruptions and secondary additions, from which M is free.

The Two Recensions

The textual criticism of the Mahābhārata proceeds from the incontrovertible fact that the text of the Great Epic has been handed down in two divergent forms, a Northern and a Southern recension, texts typical of the Āryāvarta and the Dakṣiṇāpatha. With the realization of this patent contrast began the Mahābhārata textual criticism nearly fifty years ago, when Pratap Chandra Roy brought out his popular edition of the Mahābhārata (1883–96), under the auspices of the Dātavya Bhārata Kāryālaya. A brief account of the controversy to which the publication of this edition of the Mahābhārata gave rise is to be found in Roy’s writings¹. We are told there that the appearance of his edition was hailed by The Hindu of Madras, that great bulwark of Dravidian Hinduism, in its issue dated November 22, 1885, with the publication of a bellicose letter, headed “Another edition of the Mahābhārata”, purporting to give an account of the proceedings of a public meeting held at Mayaveram, and containing an outspoken and trenchant criticism of Roy’s edition by one Mr. Sreenivasa Sastrial. This worthy gentleman thought Roy’s edition to be “sadly defective in the text and that this defect is detrimental to the religious interests of the many portions supporting the Advaita and Vāśishta-advaita (sic) doctrines, but unfavourable to the Sakti worshippers of the North, have been omitted”. “It was sad, therefore,” bemoaned this aggrieved protagonist of the Southern Recension, “that the generous gentleman of the North, Protapa Chandra Roy, that undertook to edit the text, should decline the responsibility of editing the text as correctly as possible and to compare various manuscripts of the text from Southern India.” Mr. Sreenivasa Sastrial, it is reported, “instanced one or two portions of the Mahābhārata, omitted in the Calcutta edition, which can be proved by indisputable testimony to have existed in the earliest copies of the work.” One wonders, where and how this esteemable gentleman could have got hold of “the earliest copies” of the work; or rather, just how early were the copies he was referring to. “Again, many verses”, complained this Vaiṣṇava propagandist, “quoted by the great philosophers of the South in support of their respective doctrines, are not to be found in Mr. Protapa Chandra Roy’s edition”¹

The reply of Protapa Chandra Roy is not altogether without interest. He ruefully admitted—what we must even now admit—that “there can be no edition of the Mahabharata, how carefully edited soever, that would please scholars of every part of India. . . .Like other ancient works that have come down to us from century to century by the method of manual transcription, large interpolations have been inserted in this great work.² To settle, at this fag-end of the nineteenth century, what portions are genuine and what otherwise, is, except in a very few instances, simply impossible”. With highly commendable

objectivity, Roy then proceeds to enunciate a critical principle, which, simple—nay, obvious—as it is, many a reputable scholar of India will find difficult to appreciate even at the present day. “ I know of no method”, wrote Roy, nearly fifty years ago, “except that of taking that only as undoubtedly genuine which occurs in all the manuscripts of the East, the North, the West, and the South!” “As far as my edition is concerned”, he continued, “it is substantially based on that of the Royal Asiatic Society of Bengal, published about forty-five years ago under the superintendence of a few learned Pandits of Bengal aided, as I believe, by an English orientalist of repute. . . Manuscripts had been procured from all parts of India (the South unexcepted) and these were carefully collated. Although edited with such care, I have not, however, slavishly followed the Society’s edition. I have compared it carefully with the Maharâjah of Burdwan’s text in the Bengalee character which was edited with still greater care. About 18 manuscripts procured from different parts of India (the South not excepted) were carefully collated by the Burdwan Pandits before they admitted a single śloka as genuine. I have very frequently referred to this Burdwan edition also for checking the Society’s text. . . Besides the published texts, I have now and then referred to certain manuscripts. These, however, are all of Bengal. I am willing to consult any approved manuscript of Southern India. . . I conclude by repeating that I have no complaint against Mr. Sreenivasa. On the other hand, I freely admit that an edition like the one projected by him will be a valuable accession to the libraries of all scholars in India and in countries out of India. Only the same remarks that he has applied to my edition will, I am confident, apply to his, when a Pundit of Northern or Western India takes it up for notice or review, unless, of course, the learned Sastrial includes, without critical examination, every passage bearing on both the Advaita and the Cākta worship. I may assure Mr. Sastrial, however, that in that case, in his attempt to please every body he will, like the painter in the fable, please none, particularly among readers of judgment and critical discrimination. The fact is, that the divergences of manuscripts are so great that it is perfectly impossible to produce an edition that could at once satisfy both Aryâvarta and Dākshinâtya. ” That edition, alas, so bravely and enthusiastically planned by Mr. Sreenivasa Sastrial, to which reference is made in the above extract, appears never to have seen the interior of any printing establishment ! I have quoted Protáp Chandra Roy in extenso, not merely because of the interesting sidelight his remarks throw on the question of the different editions of the Mahâbhârata, projected or planned, in or just before his time, but also because of some remarkably sound principles of textual criticism, briefly, but clearly, propounded therein by him. Protáp Chandra Roy had grasped the Mahâbhârata Problem in all its essentials. But the time was not yet ripe for the actual preparation of a critical edition of the Mahâbhârata.

The differences between the two recensions of the Mahâbhârata must not be underrated. Between them there lies, to start with, the irksome barrier of scripts. It is no exaggeration to say that in India to the Northerners, the Southern versions written in Southern scripts, ordinarily speaking, were and are sealed books; on the other hand, the Southerners, with the possible exception of a few learned Pandits—who, in fact, after a half-hearted admission of epic poetry into the realm of literature, cheerfully leave the

study of the bulk of the Mahābhārata text to their less gifted brethren—could not and cannot decipher the Northern scripts, perhaps with the exception of the Devanāgarī. When one laboriously surmounts this initial obstacle, and starts to compare the two recensions, one finds, to one’s surprise, that the difference between them begins, as a matter of fact, with the very division of the Mahābhārata into its various parvans! Against the commonly accepted, conventional division of the epic into eighteen books (parvans), there is the Southern division into twenty-four.³ More surprising still is the fact that the Ādiparvan itself, the very first book of the epic (with which alone we are, in fact, here concerned), is sub divided in Southern manuscripts into three (Ādi, Āstika and Sambhava), or at least into two (Ādi and Sambhava) separate major parvans.⁴ Let me emphasize that it is the main large divisions (parvans) of the epic I am here referring to, and not the hundred (sub-)parvans (also called upaparvans or antaḥparvans). The sub-parvans, in point of fact, could not come into question here at all. Only the Northern manuscripts, as a rule, mention in their colophons the names of the sub-parvans; the Southern manuscripts ignore (as far as I can say at present, uniformly) this detail, very rarely mentioning, in their colophons, the name of the corresponding sub-parvan.⁵ We have, therefore, no means of knowing precisely the number and the limits of the sub- parvans in the Southern scheme, except, of course, the meagre and ambiguous data of the Parvasaṁgraha (Ādi 2) itself.⁶ It is true that the Southern (printed) editions (not excepting Professor P. P. S. Sastri's critical edition of the Southern recension, as far as it has gone) follow the division of the epic uniformly into the conventional eighteen books.⁷ But in so far as they do that, the editors, it seems to me, must be overriding knowingly (but without giving the fact inex pedient prominence) the clear and unmistakable testimony of Southern manu- scripts. They prefer to sacrifice the Southern manuscript tradition and make their editions harmonize with the data of the Parvasaṁgraha: always a grave blunder; because, clearly, the data of the Parvasaṁgraha can be manipulated far more easily than those of the manuscripts of the text. The Parvasaṁgraha, if compiled, originally, on the basis of some Northern version,⁸ would certainly not fit the Southern recension exactly, even when the Parvasaṁgraha was first compiled.

The difference between the recensions does not end there by any means, unhappily. The manuscripts of the two recensions show numerous other, big and small, discrepancies: discrepancies in the spelling of most ordinary words (e. g. N नीका: S नीळा or नीळा), especially of proper names (e. g. N नैमिष: S नैमिष); in the readings of words, phrases, lines, stanzas, groups of stanzas (passim); in the sequence of all these elements (passim); in the relative position of single adhyāyas or of a small group of adhyāyas (passim); in the relative sequence of whole episodes (e. g. the Śakuntalā and Yayāti episodes, Ādi 62 ff., and 70 ff.). What is more disconcerting still is that the recensions show also complicated displacements of portions of adhyāyas; cf., for example, the long notes on 1. 106. 11 (p. 474 f.), and 1. 144. 20 (p. 624). Besides these variations in spellings, readings and sequences, there are additions (or omissions, just as one may happen to regard them) of single lines (often “inorganic”, i. e. such as can be added or omitted with no effect upon the grammar or continuity), of short passages (passim) and long passages comprising more than a hundred lines (cf. App. I, No. 55, a passage of 125 lines, setting forth the story of the Kāśi princess Ambā). These additions (respectively omissions) and verbal variants sometimes go to such a length that, at times, there emerges in the end an entirely different story. Compare, for instance, the two versions of the highly popular episode “Rape of Subhadrā” (Subhadrāharaņa) in adhy. 211–212 of our edition and passage No. 114 of App. I (comprising over 460 lines!)⁹. We find that the Southern version of this story is enriched with many entirely novel and startling features, such as Arjuna’s masquerading as a peripatetic monk (yati), or his fierce battle with the Yadava forces led by Vipṛthu, which he, of course, routs, alone and unaided, or rather merely with the help of his newly acquired, valiant and resourceful wife, who acts as his charioteer! A notable feature of the Southern recension is that it is considerably longer than the Northern. The constituted text of the Parvasaṁgraha (1. 2. 96) gives 7984 “ślokas” (that is, probably, what is technically called granthas) as the extent of the Ādi: सप्त श्लोकसहस्राणि तथा नव शतानि च । श्लोकाश्च चतुरशीतितर्द्वौ ग्रन्थो महात्मना ॥ The extent of the Vulgate is computed to be about 8460 “stanzas”. The length of the Southern text of the Ādi edited by Professor P. P. S. Sastri is given by himself as 9984 “stanzas”, slightly in excess of his own Parvasaṁgraha figure (M. 1. 2. 102), which differs as regards this figure (as in many other figures in adhy. 2) from our edition. This latter figure (9984) is perhaps a trifle in excess of the presumable extent of the (normal) Southern recension, since P. P. S. Sastri’s text contains some clear instances of interpolation (from Telugu, Tamil and even Northern sources),¹⁰ which need not necessarily be put down to the already swollen account of the Southern recension. The difference between the Vulgate and Sastri’s text is about 1524 “stanzas”. But even the common Southern text, which will be appreciably shorter than Sastri’s, may confidently be

eighteen-parvan division, which does not harmonize with the data of the colophons of the Southern MSS., which have the twenty-four-parvan division.

reckoned to contain approximately 1300 “ślokas” (i. e. granthas) more than the longest Northern version of the Ādi! This excess in the Southern recension is not due to the addition of any single lengthy passage or just a few of such passages even, though there are undoubtedly among them some fairly long passages. The excess is due to additions, large and small, distributed almost evenly throughout the parvan. Not only is the Southern text thus appreciably longer than the other, the story itself of the Southern recension, as compared with that of the Northern, is, owing to many of these additions, much richer in details, leaving little or nothing to the imagination of the reader or the hearer. Thus, for example, in the Northern recension, the father of Satyavatī or Matsyagandhā (Vyāsa’s own mother) is a nameless king of fisher-folk, making a living, on the banks of the Yamunā, by fishing. This is rather unsatisfactory. That the name of Matsyagandhā’s father—he is really only her foster-father, according to the fable—should not have been preserved, seems a shocking piece of negligence on the part of the historian, that is, the story-teller, since history as it is narrated (as has been well said) is a kind of roman à thèse. The Southern recension here comes to our help. It has carefully procured the name of the foster-father of Kālī Matsyagandhā alias Satyavatī: it was Uccaiḥśravas (a high-sounding Aryan name), if we are to believe the Southern recension. He was named after the great snow-white Stallion of the Gods, which came out of the ocean when it was being churned for Ambrosia by the Gods and the Titans. Then again, the Purohita sent by the Yādavas to the forest retreat of Pāṇḍu in the Himalayas was a Kāśyapa. He was required, of course, to perform all the little Aryan rites for the Pāṇḍavas. Moreover, it is best that kings always have their Rājaguru by their side, to advise and help them on all occasions. The Northern recension does not even tell us that the Yādavas had sent any Purohita at all to Pāṇḍu’s hermitage; so there, no question of his name arises. But a really illuminating instance of the richness of information furnished by the Southern recension is supplied by an “additional” adhyāya¹¹ in this recension, which gives us some new and interesting chronological details about the Pāṇḍavas themselves. These details disperse that haze of uncertainty and vagueness which overspreads the ‘ordinary account.’ The Southern recension informs us that when the Pāṇḍavas first arrived at the forest retreat, after the death of their father, Yudhiṣṭhira was exactly sixteen years old, Bhīma fifteen, Arjuna fourteen, the twins thirteen. We are further told exactly how long the Pāṇḍu brothers stayed at the Kaurava Court, in the Lac House (Jatugṛha), in Ekacakrā, at the Court of the Pāñcāla King, then again in Indrapraṣṭha, and so on. Yudhiṣṭhira died at the ripe old age of 108, which is a mystic number. Arjuna was younger than Kṛṣṇa by three months, which was also exactly the difference between the ages of Kṛṣṇa and Balarāma. And so on and so forth. Almost all these useful details are lacking in the Northern recension, and I doubt whether they can even be reconstructed from the meagre data of this recension on these points.

The Southern recension impresses us thus by its precision, schematization, and thoroughly practical outlook. Compared with it, the Northern recension is distinctly vague, unsystematic, sometimes even inconsequent, more like a story rather naively narrated, as we find in actual experience.

The Southern recension of the Ādi at least is thus not merely longer, but also fuller, more exuberant, more ornate than the Northern. It may therefore be fitly styled, in relation to the Northern, the textus ornatior.

Notwithstanding these and other discrepancies, there there persists throughout, between the recensions, a distinct and undeniable family resemblance, and there can be not the slightest doubt that they both spring from a common source, albeit a distant and somewhat nebulous source. Follow the course of these divergent streams as far back as one will, the elusive source seems to recede still further and lose itself in the mists of antiquity.

It was pointed out above that a noteworthy feature of the Southern recension was that it was appreciably longer than the Northern. The character of the principal additions may be seen from the following list of some of the more important and lengthy passages peculiar to the Southern recension, whose texts are given in Appendix I.

(1) No. 9 (S except M₁): God Śiva (Rudra) drinks up the poison (hālāhala) which exudes from the mouth of Vāsuki, while the Devas and Asuras are churning the ocean for Ambrosia (samudramanthanana); comprising 19 lines. (2) No. 45–48 and 51: Additions to the Śakuntalā episode (together 231 lines). (3) No. 52: Mādhavī is introduced on the scene during the discourse between Yayāti and his grandsons, in the Yayāti episode (43 lines). (4) No. 55: Anticipation of the story of the Kāśī princess Ambā (125 lines). (5) No. 59: Sūrya persuades Kuntī to have sexual intercourse (21 lines). (6) No. 67: Details of the early life of the Pāṇḍavas in the Himalayan retreat (46 lines). (7) No. 68–69: Pāṇḍu’s death and many funeral orations (together 123 lines). (8) No. 78 (S, and by conflation K₄ Da₁ Dn D₂.₄.₅): Details of a battle between the Kurus and the Pāñcālas, and capture of Drupada (119 lines). (9) No. 79: Anticipation of the account of the birth of Draupadī and Dhṛṣṭadyumna; and account of the birth of Drupada (together 194 lines). (10) No. 87–89: Additions to the Hiḍimba episode (69 lines). (11) No. 91–93: Additions to the Bakavadha episode, including a detailed account of the fight between the two well-matched giants, Baka and Bhīma (106 lines). (12) No. 95: Drupada bemoans the loss of the Pāṇḍavas, and is consoled by his Purohita; decides, at the advice of the Purohita, to celebrate the Svayaṁvara of Kṛṣṇā, in the hope that the Pāṇḍavas might turn up (74 lines). (13) No. 100: Story of Nālāyanī narrated by Vyāsa to the Pāñcāla king, to justify the polyandrous marriage of the Pāṇḍavas (118 lines). (14) No. 101: Story of Bhaumāsvī related on the same occasion (22 lines). (15) No. 103: Mimic warfare between the Kauravas and Pāṇḍavas aided by Pāñcālas (219 lines).

(16) No. 108: Dhṛtarāṣṭra crowns Yudhiṣṭhira king before despatching the Pāṇḍavas to Indraprastha (58 lines). (17) No. 111: Description of Nārada, who comes to visit Yudhiṣṭhira (55 lines). (18) No. 113–115: Expansion of the Subhadrāharaṇa (562 lines!). (19) No. 116: Arjuna's welcome on his return from exile (28 lines). These passages alone comprise 2250 lines or 1125 stanzas approximately!

The discrepancies between the two recensions, as already observed, are so numerous and so multifarious, that any attempt to enumerate and classify them must remain incomplete and unsatisfactory. Nevertheless it may be useful to begin a cursory survey of the divergences, noting at the same time the typical characteristics of the Southern “additions”, characteristics which recur with fair frequency in the Ādi, and which are likely to reappear in other parvans. These notes may prove useful for distinguishing between the different “hands” which have been at work in shaping this imposing monument of Indian antiquity, when the entire text has been treated in the manner proposed here, and we have sufficient data for undertaking a minute and systematic study of the variations and evaluating them.

The deviations of the Southern recension from the Northern (taking for purposes of exposition the latter to represent the norm) are of the following kind.

1. Variants of isolated words or phrases, (a) unimportant and (b) important.

(a) Unimportant, such as one comes across in line after line. They are far too numerous to be listed even approximately completely, but from among them we may single out these for specific mention: (i) fluctuations in the spelling of proper names, e. g. S नैमिष (N नैमिष), रोमकर्षण (लोमकर्षण), अलीक (अलक), इल्लिण (इल्लिण), हस्तिनपुर (हास्तिनपुर), उच्चथ्य (उत्तथ्य), महाभिषक् (महाभिष), तण्डु (तंडु), etc., etc. (ii) variations mainly due to mere transpositions of words, e. g. S महर्षेः सर्वलोकेषु पूजितस्य महात्मनः (N ०⁴ पूजितस्येह सर्वलोके म⁰) 1. 1. 23; प्रतीपस्तु ततो राजा (ततः प्रतीपो राजा स) 92. 1; etc., etc. (iii) unremitting variation of: monosyllabic particles and verse-fillers, which are among the most unstable elements of the received text, such as न च, वै, तु, ह, हि, [अ] अथ; common adverbs and conjunctions, such as ततः, तदा, तथा, तत्र, यतः, यदा, यथा, यत्र, अत्र, अतः, इतः, अथ, चापि, चैव; and prepositions अधि-अभि-अति, प्रति-परि, अनु-उप, etc., etc. (iv) substitution of metrically equivalent synonyms, or words and phrases of similar significance; e. g. प्रयु-विभु-(अ)भिभू; ऋषि-मुनि; द्विज-विप्र; राज्य-राष्ट्र; पन्नग-भुजग; नरेश्वर-नरोत्तम-नराधिप-नरर्षभ; ब्रह्मवित्तम-वेदवित्तम-विप्रसत्तम; पन्नगेश्वर-पन्नगोत्तम (and similar compounds with ईश्वर and उत्तम); ब्रह्मविशारद-शास्त्रविशारद-धर्मविशारद-वाक्यविशारद; चरखिन्-तरखिन्-तपस्विन्; समरदुर्मदौ-समरदुर्जयाँ; कर्यपियामि-क्षेमप्रियाम्यां-कीर्तिपियाम्यां; अमिततेजसा-अमितमहाद्यशा; वाराङ्गना-शुचिस्मिता; चारुदर्शना-चारुलोचना-चारुहासिनी-वरवर्णिनी-(अ)सर्ववर्णिनी; धर्मवृत्तां वर-शास्त्र (शास्त्र)विदां वर; अर्शवर्धना-संशय; etc., etc. (v) substitution of equivalent epic iterata; e. g. उवाच तदनन्तरं, पुनरेवमाभाषत; परस्परोपजिघांशा, परस्परद्वेषिणः; निःश्वसन्तं यथा नागं, श्वसन्तमिव पन्नगं; etc., etc. For other examples, see

Hopkins's collection of “Parallel phrases in the two Epics” in the Great Epic, pp. 403 ff. (Appendix A).

(b) Important variants, which make a considerable difference in the sense, and of which the critique must take account. Of such variants, relatively speaking, there are only a few; e. g. discrepant divisions of the epic into parvans and adhyāyas; variants of the titles of the sub-parvans (e. g. S प्रायोपवेशनः N मृगसङ्गमय), of the numbers of adhyāyas and ślokas in the Parvasamgraha (the figures for ślokas differ, at times, by thousands). — An example of a different character from another part of the Ādi is the variant वर्षः मास in the stanzas which refer to the duration of Arjuna’s exile. According to the Northern recension it is thirteen years; according to the Southern, only thirteen months! Cf. 1. 204. 28. ध नो द्वादश वर्षाणि (S वै मासान्, मासानि etc.) ब्रह्म(S ब्रत)चारी वने वसेत्; 205. 30 वने द्वादश वर्षाणि (S मासानां, मासानि, मासान् हि).¹² — Then we have in 1. 3. 21 the variant बहु-चन्दे. How was the infinitive really made? — And so on.

2. Larger variations between continuous passages, as a whole, the total extent remaining approximately the same.

We find them (a) mostly in the long lists of names: e. g. of the hundred sons of Dhṛtarāṣṭra (adhy. 108), of ancient kings (1. 1. 166 ff.), of serpents (1. 52. 5 ff.), of kings present at Draupadī’s svayamvara (adhy. 177); but (b) also when there are transpositions of whole or parts of adhyāyas (e. g. the prose genealogy, adhy. 90); or again (c) when there is free paraphrase of a passage (passim).

3. Expansion of the text in S without materially altering the nature of the contents or the course of the narrative.

(a) By multiplication of the items of a list. For instance: In adhy. 20, S (with K₄ marg. Dn D₄ marg.) adds seven lines of praise to an existing hymn (स्तोत्र), addressed to Garuḍa. In these lines, Garuḍa is identified, in turn, with all the principal gods, and with everything that is pre-eminent in the world: 299* त्वं विभुसूक्ष्मः सूर्यः परमेष्ठी प्रजापतिः । त्वमिन्द्रस्त्वं हयमुखस्त्वं हरस्वं जगत्पतिः । त्वं मुखं पञ्जो विप्रस्त्वमग्निः पवनस्तथा । त्वं हि धाता विधाता च त्वं विष्णुः सुरसत्तमः । त्वं महान्निधिः शाश्वतस्त्वं च महेश्वरः । त्वं प्रभास्मन्तमित्तस्त्वं नक्षाणमुत्तमम् । त्वं गतिः सततं त्वत्तः कथं नः प्राणुयाद्वयम् ।

In adhy. 64, S (with K₄ Dn D₁.) gives an additional short list of sciences in which the Ṛṣis in Kanva’s penance grove were proficient: 586* शब्दच्छन्दोनिरुक्तेभ्यः कालज्ञानविशारदेः । द्रव्यकर्मगुणश्रेष्ठा कार्यकारणवेदिभिः । जल्पवादवितण्ड्यैश्च न्यायग्रन्थसमन्वितैः । नानाशास्त्रेषु मुख्येभ्यः शुभावं स्वनमीरितम् ।

In adhy. 74, an additional passage (of 7 lines) in S (with D₈) harps on the well-worn theme of the evils that attend on anger: 745* तस्मात्क्रोधेन: श्रेष्ठ: कामक्रोधौ न पूजितौ। क्रुद्धस्य निष्फलायेष दानयज्ञतपांसि च। तस्मात्क्रोधेन यज्ञोऽयं दानं महत्तल्लम्। न पूतो न तपस्वी च न यज्वा न च धर्मकृत्। क्रोधस्य यो वरं गच्छेत्क्षय लोकद्वयं न च। पुत्रभृत्यसुहृन्मित्रभार्यौ धर्मेक्ष सत्यतः। तस्येतान्यपयस्यान्ति क्रोधघोरलिष्य निश्चितम्।

In adhy. 165, a Southern passage expands in hyperbolic language the list of edibles and other commodities (such as wines, clothes and blankets) furnished by Vasiṣṭha's Kāmadhenu, by the addition of 6 more lines: 1753* बाप्यस्तोनस्यैव राशया: पर्वतोपमाः। निष्ठाचानि च सूपश्च दधिकुल्यास्तथैव च। कूपोश्च घृतसंपूर्णोन्मक्ष्याणां राशायस्तथा। भोजनानि महाहोणि तत्र तत्र सहस्रशः। इक्षूनभु च लाजांश्च मैरेयांश्च वरासवान्। वस्त्राणि च महाहोणि कम्बलानि सहस्रशः।

In adhy. 213, the Southern recension furnishes us with a supplementary list of items in Subhadrā’s dowery, which, taken along with what has gone before, exhausts almost all the things worth possessing in this world: 2082* काम्बोजारुडबाहीकलिन्पुञ्जातांश्च भारत। सुवर्णकृतसंहान्मण्डनाद्विनादितान्। श्वेतचामरसंहान्मक्षेश्यौरलंकृतान्। जात्यश्वाणां सहस्राणि पञ्चाशत्तरदृष्टी तदा। 2088* भूषणानां तु मुख्यानां शतभारं ददौ धनम्। युक्ताहाराणि शुभ्राणि शतसंख्यानि केशवः। प्रवालानां सहस्रं च तथाप्यन्यापि भारत। सुवर्णपादपीठानां महाहॊत्तरणान्यीत्था। पर्यङ्गणां सहस्रं च ददौ कन्याधने तदा।

(b) By anticipation or repetition of stories, motives or discourses. For example: (i) the miraculous birth of Kṛṣṇā and Dhṛṣṭadyumna is narrated twice in S: in adhy. 155 and in App. I, No. 79 (after adhy. 128); (ii) the theme of the amusing experience of a maiden, who, on praying to Mahādeva for one husband five times, was granted, as a boon, five husbands at one time—a story which seems to have been very popular in the South—is used, with variation, in S, no less than three times in the course of the Ādi; cf. adhy. 157, 189 and passage No. 100 (of App. I);¹³ (iii) account of the tragi-comic experiences of the Kāśi princess Ambā, who was passed on in turn by Bhīṣma to Sālva and Sālva to Bhīṣma, repeatedly, like a shuttle-cock, a story which is really the subject-matter of the Ambopākhyāna in Udyoga 173 ff. (Bom.

(d), apparently a favourite piece, is anticipated in passage No. 55 of App. I (cf. adhy. 96) and forms a bulky addition of 125 lines!

(iv) the future of the royal family, which is the subject-matter of the additional dialogue between Bhīṣma and his step-mother, Satyavati, in S, in passage No. 57 (of App. I) is only a continuation and repetition (with v. l.) of the discourse between the same parties in adhy. 99.

(v) Sūrya's warning to Karṇa about the designs of Indra to supplicate Karṇa in the disguise of a Brahman, in passage No. 60 of App. I, which is an anticipation of the story told in Araṇya 300 (Bom. ed.).

(c) Additions in S, due to the explicit mention of the observance of the correct and complete Brahmanic ritual and ceremonial on the proper occasions. Thus, in adhy. 68, at the birth of Bharata: 625* यथाविधि यथान्यायं क्रियाः सर्वास्स्वकारयत् ।

In adhy. 92, at the birth of Śamtanu: 921* तस्य जातस्य कृत्स्नानि प्रतीपोऽकार्य्यवत्प्रभुः । जातकर्मदि विप्रेण वेदौक्तैः कर्ममिच्छता । नामकर्म्म च विप्रास्तु चकुः परमसंस्कृतम् । शतनोर्वर्त्त्नीपाल वेदौक्तैः कर्ममिच्छदा ।

In adhy. 100, at the birth of Dhṛtarāṣṭra and Pāṇḍu: 1084* तयोर्जन्मक्रियाः सर्वा यथावदुपवैशः । कारयामास वै भीष्मो ब्राह्मणैर्वेदपारगैः ।

In adhy. 115, we have, likewise, with reference to the Pāṇḍavas themselves (App. I, No. 67, lines 13–14, 20–27): न भवेत्क्रियाहीनाः पाण्डोः पुत्रा महाबलाः । पाण्डोः प्रियहितान्वेषी प्रेष्य त्वं पुरोहितम् । तमागत् द्विजश्रेष्ठं काश्यपं वै पुरोहितम् । पूजयामास विधिवाण्डुः परपुरञ्जयः । पृथा माद्री च संहृष्टे वन्दुवेंदं प्रशंसताम् । ततः पाण्डुः क्रियाः सर्वाः पाण्डवानांमकारयत् । गर्भाधानादिकृत्यानि चौलोपनयनानि च । काश्यपः कृतवान्वर्षमुपाकर्म्म च भारत । चौलोपनयादृर्ध्वं वृषभाक्षा यशस्विनः । वैदिकान्यध्यने सर्वे समपद्यन्त पारगाः ।

In adhy. 124, at the royal tournament: 1412* रङ्गमध्ये स्थितं द्रोणमिवाथ नरर्षभाः । चकुः पूजां यथान्यायं द्रोणस्य च कूपस्य च । आशीर्भिश्च प्रयुक्ताभिः सर्वे संहृमानसाः । अभिवाय पुनश्शास्त्रान्बलिपुष्पैः समाचितान् । रक्तचन्दनसंसिप्तेः स्वयमर्चन्त कौरवाः ।

(d) Expansion in S of existing scenes by the addition of speeches or detailed descriptions and by other digressions. Examples:

(i) in App. I, No. 9, in the account of the churning of the ocean (samudra- manthana), we are incidentally told of the drinking of poison by Śiva, which had exuded from the mouth of Vāsuki during the churning; (ii) in 998*, we have nine additional lines depicting the humiliating treatment meted out to Bhīşma at the court of the king of Kāśi, during the Svayamvara of his daughters; (iii) passage No. 59 (of App. I) depicts the persuasion of the shy and reluctant Kuntī by Sūrya for intercourse, by alternate threats and promises, like a real Don Juan; (iv) in passages No. 68–69, the Southern recension has tried to develop a very pathetic scene indeed, depicting the death of the father of the heroes, Pāņdu: an incident the perfunctory notice we find preserved in the Northern recension. At the sight of the corpse of her husband, Kuntī falls to the ground in a swoon, like a felled tree. Then the five brothers come up in a single file, and in the order of their ages, and recite their little mournful dirges: Yudhişţhira gets 8 lines, Bhīma 7, Arjuna 4, the twins (in chorus) only 3 lines together.¹⁴ Then follow long-winded farewell orations by Kuntī, Mādrī and the rest of the company, which are followed by a touching scene describing Mādrī mounting the funeral pyre; (v) passage No. 78 gives, in 119 lines, the details of a fight, which, in the Northern recension, at least originally, is disposed of in two lines! The latter I consider adequate treatment, taking everything into consideration; (vi) passage No. 93 is a Southern addition of 37 lines giving fuller details of the titanic struggle between Bhīma and the cannibal Baka; (vii) 1737* adds a hymn (in Trişţubh metre and pseudo-vedic style) by Vasişţha, addressed to Sūrya, when Vasişţha presents himself before that luminary on behalf of Saṁvaraņa; (viii) 1828*ff. describe in turn the discomfiture of each of the suitors for the hand of Draupadī; (ix) passages No. 100–101 add to the existing stock two new anecdotes—alternative explanations—narrated by Vyāsa to prove to Drupada and his son, that the polyandrous marriage proposed by Yudhişţhira, though apparently immoral and illegal, is a most righteous and necessary union, being pre-ordained by the gods themselves for the accomplishment of their cosmic plans: these are the well-known legends of Nālāyanī and Bhaumāsṿi; (x) passage No. 106 gives an almost complete inventory of the presents Drupada gave to the Pāņdu brothers when they left with Draupadī, for the Kaurava Court, at the invitation of Dhṛtarāşţra. The Northern recension ignores this huge mass of presents completely; (xi) passage No. 110 is a farewell scene containing short orations by Kṛşņa, Yudhişţhira and Kuntī, when Kṛşņa sets out for Dvārakā;

(xii) passage No. 112 contains the farewell of Arjuna to Citrāńgadā, telling her that she must not give way to sorrow in his absence, as they would meet again soon at the Rājāsūya, which is going to be performed by Yudhiṣṭhira: a prophetic utterance! And so on and so forth.

(e) Additions of little ethical, moral and sententious maxims, to which S, permeated as it is by a conscious didactic purpose, is particularly partial. We meet with the same old proverbs over and over again in S: 595* अन्यथा सन्तमात्मानमन्यथा सद्भु भाषते । स पापेनाष्टृतो मूर्खस्तेन आत्मापहारकः । 605* पिता रक्षति कौमारे भर्ता रक्षति यौवने । पुत्रस्तु स्थविरे भावे न स्त्री स्वातन्त्र्यमर्हति । मन्थुग्रहणा विश्वा न विश्वाः शस्त्रपाणयः । अग्निर्दहति तेजोभिः सूर्य्यो दहति रश्मिभिः । राजा दहति दण्डेण ब्राह्मणो मन्थुना दहत् । क्रोधितो मन्थुना हन्ति वज्रपाणिरिवासुरान् । 780 परभार्य्या स्वसा ज्येष्ठा सगोत्रा पतिता खुषा । अपरा भिक्षुकावस्था अगम्याः कीर्तिता बुधैः । 782* कश्चिदार्त्तवियो हन्याच्छखमन्यं निकृन्तति । यच्छछ्याद्भिर्द्दहति मनसा हन्ति वै द्विजः । 804* पुत्रार्थे भर्तृपोषणार्थे स्त्रियः सुधाः स्वयंसुवा । अपतिष्ठापि या कन्या अनपत्या च या भवेत् । तस्या जन्म वृथा लोके गतिस्तस्या न विद्यते । 833* गुरोर्व्वचनं पुण्यं स्वर्गमायुष्करं नृणाम् । गुरुप्रसादात्रैलोक्यमनवशासच्छत्तकतुरः । 956* न च कुर्य्याद्भरौ दैन्यं शाठ्यं क्रोधं तथैव च । जैक्यं च मत्सरं वैरं सर्व्वचैतत्त्र कारयेत् । मातरं पितरं चैव विद्वान्सं च तपोधनम् । क्षमावन्तं च देवेन्द्र नावमच्येत् बुद्धिमान् । शक्तस्तु क्षमते नित्यमशक्तः क्रोशते नरः । दुर्ज्जनः सज्जनं द्वेष्टि दुर्बलो बलवत्तरम् । रूपवन्ततरूपीं च धनवन्तं च निर्धनः । अकर्मी कर्मिणं द्वेष्टि धार्मिकं चाप्यधार्मिकः । 1019* वाचा दत्ता मनोदत्ता कृतमङ्गलवाचना । निर्दिष्टा तु परस्त्रैव सा त्याज्या परचिन्तनी । 1101* बालो हि द्वादशाद्धर्षाज्जनो यत्करिष्यति । न भविष्यत्यधमोऽत्र न प्रशस्यति वै दिशः । 1189* नास्तिस्तूत्यति काष्ठानां नापगानां महोदधिः । नान्तर्कः सर्व्वभूतानां न पुंसां वामलोचनाः । एव्वं तृष्णा तु नारीणां पुरुषं पुरुषं प्रति । अगम्यागमने स्त्रीणां नास्ति नित्यं शुचिशुचिसिते ।

1423* अङ्गोज्ञिमेग्रहात: क्षत्रश्ममनो लोहसुरयिम् । तेषां सर्वेगे तेज: स्यात् योनिषु शास्यति । (f) Additional stanzas in S with, perhaps, a certain amount of sexual appeal, bearing the taint of later decadence. Examples: App. I, No. 89 (lines 3–4) with reference to Hiḍimbā: शुभं हि जघनं तस्या: सुवर्णमृगमेखलम् । न तत्पृष्ट मुदा मृद्गमीमेसनो मृदुमुदु: । No. 48 (lines 78–79), describing Śakuntalā: कराग्ग्रमितमध्यगां तां सुकेशीं संहतस्तनीम् । जघनं सुविशालं वै ऊरु करिकरोपमौ । No. 55 (lines 46–47), describing Ambā: पीनोन्नतकुचद्वन्दा विशालजघनेक्षणा । श्रोणीभरलसग्गमा राकाचन्द्रनिभानना । Then also 929* ज्ञातमात्राभवोक्खां गद्गतीररुहे वने । प्रकीर्णकेशी पाणिभ्यां संस्पृशन्तीं शिरोरुहान् । 1189* पुत्रं वा किल पौत्रं वा कासाञ्चिञ्छतरं तथा । रहसीह नरं दृष्ट्वा योनिरेक्तिथ्यते तत: । More sentiments of this type, occurring in the Grantha version alone, are: 1937* राकाराशङ्कवदना: पद्मिनीजातिसंभवा: । पद्मगन्धा: पद्ममुखा: पद्मपत्रनिभेक्षणा: । मुक्तामुक्तकुकर्णाश्च ससविन्दुल्लाटिका: । पीवररत्नमयगता: शङ्ककुण्डलनासिका: । कुन्दोद्योधकुंकेयोयो मुखाङ्कलमुगयमा: । मृञ्झरीमोललतिका द्यावर्तनिभनामिका: । विपुलश्रोणिफलका रम्यात्माभ्युम्युगम्का: । The lengths to which the Muses lead these Southern poetasters may be judged from the following interpolation in certain Grantha manuscripts of the Ādi. 1334* with 1335* reads: व्यूढश्छाम्वरां दृष्ट्वा रम्युगुहाह्वलां नृप । पीनोत्कुचद्गां दृष्ट्वा तामृपिथक्क्मे तत: । Cf. also lines 73–74 of passage No. 100 (of App. I): मारो मां बाधतेत्यर्थमनुगृहातु मां भवान् । तर्पय मममाक्षिणि गुह्यं मां बाधतेऽनिराम् । 4. Southern additions which alter the purport of the fable as narrated in the Northern recension. Made apparently with the object of correcting the laxity of sexual relations implied in the old narrative. In adhy. 67, the royal Purohita, a handy person, quickly but surely and secretly, performed the marriage of Duḥṣanta and Śakuntalā, in order to legitimize Bharata, the eponymous ancestor of the Bhāratas, who has given his name to the country of his birth and to the Great Epic of India, altogether an important personage in ancient Indian history :

610* पुरोहितं समाहुय वचनं चेदमब्रवीत् । राजपुत्र्या यदुक्तं वै न दृष्ट्वा कर्तुमर्हसे । क्रियाहीनो हि न भवेन्न स्म पुत्रो महाद्युतिः । तथा कुरुष्य शाल्लोकं विवाहं मा चिरं कुरु । एवमुक्तो नृपतिना द्विजः परमयन्त्रितः । शोभनं राजराजैति विधिना कृतवान्विजः । शासनाद्विप्रमुख्यस्य कृतकौतुकमङ्गलः ।

In adhy. 77, the marriage of Yayāti and Śarmiṣṭhā is celebrated semi-secretly, in a secluded corner of the Aśoka grove, in the palace grounds, with the usual bakshîsh to the Brahmins, in the presence of counsellors, chaplains, priests and so on, but unknown to Devayānī! All this was done to legitimize Puru, the eponymous ancestor of the Pauravas:

807* ऋत्विक्पुरोहिताचार्यैर्मन्त्रिभिस्त्रैव संवृतः । कृत्वा विवाहं विधिवद्वस्वा ब्राह्मणदक्षिणाम् । पुण्ये नक्षत्रसंयोगे मुहूर्ते द्विजपूजिते ।

In passage No. 114 of App. I, Subhadrā and Arjuna were likewise secretly and hastily married in the presence of gods, ṛṣis, and elders, while Balarāma was away from the scene, to legitimize Abhimanyu (the father of the famous Parīkṣit and grandfather of Janamejaya, to whom the epic was narrated). Cf. lines 281–286 of the passage:

महेन्द्रशासनात्सत्वं सहिताश्च महर्षिभिः । विवाहं कारयामासुः शक्रपुत्रस्य शाश्वतः । अरुन्धती शची देवी रुक्मिणी देवकी तथा । दिव्यस्त्रीभिस्म सहिताः क्रिय्यां भद्रां प्रयोजयन् । महर्षिः काश्यपो होता सूयसा नारदादयः । पुण्याशिषः प्रयोक्तारः सर्वे शास्त्रं तदाज्ञयाने ।

Most Grantha manuscripts (G₁-₅ of our critical apparatus: क, ख and ग of Sastri's) have a passage to show that Parāśara and Matsyagandhā were secretly but regularly married. Cf. passage No. 36 of App. I, which is a somewhat lengthy passage describing with circumstantial detail the nuptial ceremony at which the ancestors of both the bride and the bridegroom are invoked, all the details of the regular Hindu marital rite are scrupulously gone through, and the marriage is solemnized in the presence of Vasiṣṭha, Yājñavalkya and other great Ṛṣis living in the Naimiṣa forest!

5. Additions in S, due to the filling out of lacunae (real or imaginary). Examples of such additions are: 482* which gives a summary of the last five parvans of our Mahābhārata text, and passage No. 79 (of App. I) giving an account of Drupada's birth. It appears, from the latter account, that Drupada was born in the same miraculous way as two of his contemporaries Droṇa and Kṛpa, due to the perturbation of his father at the sight of a beautiful Apsaras. Ascetics involuntarily emitting semen at the sight of heavenly nymphs, broad-hipped, fat-breasted, fair-clad, pleasure-fraught, and the miraculous germination of the semen into human beings, is the regular Purāṇic apparatus for the generation of the great men of the past, about whose birth nothing exciting was specially known to the chronicler.

6. Multiplication of fights and battle-scenes. I have drawn attention above to the expansion of the description of a battle (in which the Kurus and the Pāṇḍus capture Drupada), and of a fight (between Bhīma and Baka).¹⁶ Analogous to it is a battle scene described in an additional adhyāya in the Southern recension (App. I, No. 103). This stages a little miniature replica of the Great War to come. Here also Kurus plan the destruction of the Pāṇḍavas, who are residing in Kām̃pilya as the guests of Drupada. A regular council of war is held, in which Śakuni and Karṇa advocate hostilities, while the nameless son of Somadatta counsels peace and conciliation. The bellicose party has the upper hand in the council chamber. The Kuru army marches against the Pāṇḍavas with their allies, the Pāñcālas. The Kaurava forces are, of course, easily repulsed. No great damage is done. The status quo is immediately restored: things go on just the same as before, as though no battle had ever taken place. There is also no other reference to this battle in the whole of the Mahābhārata. The present parvan does not offer much scope for the full development of this tendency. We shall probably meet with it again in the battle-books (6–9).

7. Omissions in S, as compared with N. These are quite numerous and scattered almost evenly over the whole parvan, but short and contextually unimportant, as a rule. An exception is the somewhat lengthy Śvetaki episode (App. I, No. 118). Since, on the one hand, all reference to the episode is missing in the whole of the Malayālam version, as also in some manuscripts of each of the remaining two versions (T G) of the Southern recension, while, on the other hand, those T G manuscripts that do contain some mention of it insert a variant version at an entirely different place, therefore the episode may legitimately be considered a Northern interpolation which has insinuated its way, by conflation, into some Southern manuscripts. It is a story in true Purāṇic style. King Śvetaki sacrificed with such phenomenal zeal and keenness that his priests, in the end, refused to sacrifice any more! Śvetaki practised penance on the Himālayas with the object of making Rudra his sacrificial priest. Rudra, however, excused himself, asking Śvetaki to apply to Durvāsas, who was his part-incarnation (am̃śa). Durvāsas completed the sacrifice, and Śvetaki poured libations of clarified butter into the fire for twelve years continuously. As a result, Agni had a severe attack of indigestion! He refused after that every offering, and became enfeebled. At Brahmā’s direction, he set the Khāṇḍava forest on fire, and tried his best to burn the forest down; but the denizens of the forest put the fire out, over and over again. He reported his discomfiture to Brahmā, who then asked him to betake himself to Arjuna and Kṛṣṇa, the part incarnations of Nara and Nārāyaṇa, with whose help alone Agni would be in a position to burn the Khāṇḍava forest.

It should be made clear that the variants and passages cited here are merely by way of illustration, and comprise only a small fraction of the total number of deviations.

The presence of an astonishingly large number of additions, some of which are undoubtedly late and spurious, should not be allowed to impair our appreciation of some real merits of the Southern recension. It would be, in fact, a grievous error to ignore on that account the Southern recension or underestimate its value. This recension is an

indispensable aid for controlling the deviations of the Northern recension, both in point of readings and sequence. In comparison with Y, it has unquestionably preserved a very large number of original readings, proved by actual agreements between S and v, as well by their intrinsic merits. The superiority of the Southern recension in comparison to the Vulgate may be said to be quite evident. It may, however, quite easily happen that in a particular instance, the whole of the Northern recension is corrupt, and the true reading is preserved only in the Southern recension.¹⁷ An instance of this is 1. 214. 5. The Vulgate reads (B. 1. 222. 5):

अध्येतारं परं वेदान्प्रयोक्तारं महाध्वरे । रक्षितारं शुभाल्लोकॉनुलेभिरे तं जनाधिपम् ॥

Nilakantha's gloss is: परं अध्येतारं परस्य ब्रह्मगोऽधिगन्तारम् । वेदान् वेदानाम् । The stanza has been translated by Manmath Nath Dutta as follows: "Having obtained him as their king, they obtained a monarch who was devoted to the study of the Vedas, who was a performer of great sacrifices, and who was the protector of all good works". Protap Chandra Roy's translation reads similarly: "And the subjects having obtained Yudhisthira as their king, obtained in him one that was devoted to the study of the Vedas, one that was a performer of great sacrifices, and one that was the protector of all good people".

But the translations of both these scholars are generally free and arbitrary. As it stands, the stanza can be translated only as follows: "They (i. e. the people) obtained for a king, one who studied Brahma (para), employed the Vedas in a great sacrifice, and protected the blessed worlds".

This pedestrian stanza will satisfy most people as it has satisfied a long succession of critics, commentators and translators in the past. About it one can only say that there are worse stanzas in the Mahābhārata. Only a reader endowed with a fine sensibility and critical acumen will feel that there is something amiss here. We are face to face with the danger of acquiescing in a sense which might satisfy us, but which would not have satisfied the ancient writer. The Northern variants do not offer much help; even the Śāradā and K manuscripts have substantially the same readings. It would, consequently, not be easy to reconstruct from this sad wreck of a Dīpaka, the epigrammatic original, which is preserved intact only in the Southern recension, which the constituted text here follows (1. 214. 5):

अध्येतारं परं वेदाः प्रयोक्तारं महाध्वराः । रक्षितारं शुभं वर्णो लेभिरे तं जनाधिपम् ॥

No glosses, translations, exegetical notes, and such other accessories are necessary for the elucidation of this stanza; for it is self-luminous. The correctness of the Southern reading is confirmed by the very next stanza (1. 214. 6), which is also an epigrammatic period of the same type:

अधिष्ठानवती लक्ष्मीः परायणवती मतिः । बन्धुमानखिलो धर्मस्तेनासीत्पृथिवीक्षिता ॥

It should thus seem that the infidelities of the Southern recension are confined mainly to a tendency to inflation and elaboration. In parts unaffected by this tendency,

it is likely to prove, on the whole, purer, more conservative and more archaic than even the best Northern version. The Southern variants, therefore, deserve the closest attention and most sympathetic study.

After this brief survey of the interrelationship between the two recensions, we shall proceed to the consideration of the various provincial versions, into which each of the recensions breaks up.

The Archetype υ.

The Śāradā Version.

The archetype υ comprises the two versions: Śāradā and “K”. The Śāradā version is represented in our critical apparatus by the fragmentary codex Ś₁, belonging to the Bombay Government Collection (No. 159 of 1875-76), which seems to be the only extant genuine representative of the old version of Kaśmīr. The manuscript, which is undated, may be three to four centuries old. For reasons which will appear in the sequel, I have made the Śāradā version the norm to follow.

The text of the Ādi (as of other parvans of the Mahābhārata) according to the different printed editions, as is well-known, varies considerably, not merely as regards the readings, but also as regards the extent. The length of the Ādi, according to the Calcutta edition, as already observed, is estimated to be about 8460 “stanzas”, of the Bombay edition, 8620,¹ of the Madras edition (i. e. Sastri’s Southern Recension) 9984 (according to Sastri’s data), of the Kumbhakonam edition 10889.² Now, in a statement following the colophon (or forming part of the long colophon) of our Śāradā manuscript, the length of its text is given as 7984 in a stanza cited from the Parvasaṅgraha; cf. the accompanying facsimile of fol. 155 a of the Śāradā codex. To judge by the amount of textual matter which an average folio of the fragmentary Śāradā codex holds, this estimate of its extent appears to be approximately correct. Assuming then that to be the length of the Śāradā version, it becomes the shortest known version of the Ādi, and may, therefore, appropriately be called the textus simplicior.

While it is the shortest extant version, it is a demonstrable fact that it contains relatively little matter that is not found, at the same time, in all other versions of both recensions. It is clear, therefore, that it must contain, relatively, less spurious matter than any other known version. That is precisely the main reason why it is taken as the norm for this edition.

Since our codex (Ś₁) is fragmentary, it must be considered a piece of singularly good fortune that there has been preserved at least one nearly complete Devanāgarī manuscript of the Ādi, namely, India Office No. 2137, that may, as will presently be shown, be used, without hesitation, to supplement the missing portions, since it undoubtedly is a moderately trustworthy, though comparatively late and slightly contaminated and incorrect transcript of a Śāradā exemplar.

Further particulars of the Śāradā version will be found under the account of the "K" version.

The accompanying facsimile of a page of the Śāradā codex (fol. 155 a) contains the end of the Ādi and the beginning of the Sabhā. The Parvasaṃgraha stanza, mentioned above, giving the extent of the Ādi, will be found in lines 2–3 of the facsimile.

The "K" Version.

This version, as already explained, is a specific Devanāgarī version, closely akin to the Śāradā version and clearly differentiated from the (so-called) Devanāgarī version.

The affinity of the manuscripts comprising this version is illustrated by the following concordant readings, selected at random. The references are to adhyāyas and ślokas.

1.2 Ko.₂₈ महद्भिः : rest (mostly) ब्रह्माद्भिः. 1.8 K V₁ लोमहर्षणिखाच : others सूत उ°, सौतिर्°, सूतः. 1.49 K V₁ B₁m संक्षेपतोऽब्रवीत् : others संक्षेपम°, संक्षिप्य चात्र°. 1.51 K विचक्षणा : rest (mostly) मनीषिणः. 1.192 Ko₈ श्रुतपाप्मान : rest शुभकर्मणः. 2.76 K यक्षाणामथ रक्षसां : rest (mostly) यक्षाणां च महौजसां. 4.10 K पुरस्कृता : rest पुरःसराः. 8.2 K D₂ झुवकत्वं यतोऽभव : others सुनकं समजीजनत्, धुनकस्तु सुतोऽभवत्, etc. 8.16 K अष्टोदशे : others सूराशमे, etc. 10.2 K D₂₈ (by transp.) सदैव भुजंगं हन्यां : rest हन्यां सदैव भुजंगं. 13.1 K D₂₈ भवीहि : rest वदस्व. 13.25 K D₂₈ बान्धवै : others बन्धुभिः, etc. 17.9 K D₂₈ ततो नानाप्रहरणे : others नानाप्रहरणैर्गमेभिः, etc. 19.4 K D₂₈ ब्रह्मर्षेर् बहुसाहस्रैः सर्वैर्नानाविधेरपि : others सर्वैश्च बहुसाहस्रैर्नानास्त्वैः समाहृतं, etc. 24.1 K D₂ लिखितं : rest om. नु. 24.14 K D₂₈ महाद्युतिः : rest महाबलः. 55.8 K क्षुद्र : others क्रूरः, क्षुद्रः. 55.35 K जातवेदयै : rest हव्यवाहनै. 56.14 K कीर्तयेत् : rest आवयेत्. 57.2 K नेदीयपिर्वनन्दनः : rest पौरवनन्दनः. 57.8 K D₂ रम्यश्च : others पुण्यश्च, etc. 57.43 K D₂ गृहान् : rest गृहं. 58.3 K D₂ कीर्तयिष्यामि : rest कथयिष्यामि. 58.40 K D₂ सपर्वै : rest संनिधौ. 59.29 K D₂ वीर्यवान् : others दानवः, etc. 60.6 K N₅ D₂ श्रमे : पुत्रास्तु बहवः (by transp.) : rest अनेजेस्तु बहवः पुत्राः. 60.52 K संयूत् : rest पुत्रात्ः. 62.6 K D₂ राज्यं प्रशस्तिः : rest राजनि शासति. 64.29 K D₂ ददशें : rest जगाम. 67.30 K D₂ निधाय कायमं त्वच : others विनिधाय ततो भारं, etc.

68. 69 K D₅ ध्रश्वा (or ᵒात्) : rest यात्ता. 71. 41 K D₅ वस्त : rest मित्र. 74. 7 K alone transp. कुमारा: and कुमार्यैः. 76. 33 K D₅ घुष्विणिति : rest सुमद्यमर्धा. 150. 18 K वात्सल : rest विनम्र; etc., etc., etc.

Further examples of the concordant readings of the K version will be found below. It was remarked above that K₁ (= India Office 2137) was a manuscript of Kaśmīri origin, exhibiting specially near affinities with Ś₁, so much so that K₁ may be regarded as a copy of some Śāradā original. The Kaśmīri character of K₁ was already fully recognized by Professor Lüders, who had utilized it in the preparation of his specimen³ of a critical edition of the epic mentioned above, although he had no genuine representative of the Kaśmīri or Śāradā version to compare it with.

The affinity between Ś₁ and K₁ is documented by a mass of readings, of which the following (selected at random) will serve as illustrations. The references are to adhyāyas and ślokas.⁴

27. 15 Ś₁ K₁ पुरंदर: : rest शतकक्रतुः (synonym!). 28. 24 Ś₁ K₁ ततः : rest नदीः. 29. 4 Ś₁ K₁ धरा अरेण : others भरान्तरेश, etc. 30. 7 Ś₁ K₁ प्रदिष्टेव: : others प्रतिगृल्हतां, प्रगृह्य, संगृह्य, आयुह्य, etc. (original hypermetric!). 31. 6 Ś₁ K₁ पंजरक: (corrupt) : others पिंजरक:, पंजरक:, etc. 37. 25 Ś₁ K₁ रक्षितव्य: पिता द्विति (corrupt) : others रक्षितव्या: प्रजाशख्या, etc. 42. 7 Ś₁ K₁ नेतरौ रोच्यमान्हं : others न भरेयं च तामई, etc. 44. 2 Ś₁ K₁ भवात् : others सदा, तथा, ततो, etc. 45. 5 Ś₁ K₁ अवदन् : rest अनुवन् (synonym). 45. 19 Ś₁ K₁ तयोर्दिते : others नराधिपें, etc. 131. 3 Ś₁ प्रसुशुभे (corrupt) : rest पशुष्पते:. 131. 13 Ś₁ चरन् : rest शनैः. 154. 24 Ś₁ K₁ जाहव्यो : rest भागीरथ्या (original has double crasis!). 206. 3 Ś₁ K₁ एककक्षा: (corrupt?) : rest कचकक्ष: (G₁ श्रदद्का:). 218. 48 Ś₁ K₁ मन्दराच्छिखरं महन् ( = 47°) : rest गिरेः शृंगं सहस्रशा (or ᵒशा:).

The above are examples of concordant readings of Ś₁ and K₁. As instances of adhyāya division and numbering may be pointed out that adhy. 42–44 and 46 of the constituted text (comprising adhy. 46–48 and 50 of the Vulgate) are numbered in Ś₁ K₁. 54–56 and 59 respectively, and are so numbered in no other manuscript hitherto collated; further, after only the third stanza of our adhy. 40, both manuscripts (Ś₁ K₁) interpolate the figure 51, Ś₁ marginally inserting, at that place, an additional colophon: इज्यादिपर्वैकपत्राशमोध्यायः. Likewise, after 1.165.34, Ś₁ K₁ insert, an additional colophon, not found in any other manuscript.

Among “additional” passages peculiar to Ś₁ K₁ may be mentioned 1735*. K₁ is, however, by no means, a direct copy of Ś₁. There are numerous discrepancies between them. Notably, there is a big lacuna in K₁ in adhy. 47–48, where Ś₁ is intact.

## PROLEGOMENA

Again at 1. 107. 26; 154. 10, 11; 175. 4 and other places; Ś₁, which generally omits the verbs उवाच (resp. उचुः) in the short prose formulæ of reference to the speaker, does show these verbs, while they are lacking in K₁. 1. 208. 14 is an exception where both Ś₁ and K₁ have उवाच. Ś₁ K₁ exhibit also numerous minor differences in their readings; e. g. 1. 36. 22; 38. 21, 36; 41. 29; 46. 11; 98. 9; 125. 3, 16; 128. 2; 133. 24; 195. 11; 200. 3. After 1, 144. 17, Ś₁ has a colophon which is missing in K₁.

These agreements and differences show that while Ś₁ and K₁ are closely akin, their text is not identical. Neither Ś₁ nor K₁ is a direct copy of the other. They are independent witnesses, a circumstance which adds greater weight to their arguments.

I shall now cite some readings (also selected at random) which Ś₁ shares with the K version, Ś₁ and K standing together against all other manuscripts (barring, of course, conflated specimens). The references are to adhyāyas and ślokas.

28. 18 Ś₁ K क्षसनेन : rest क्रथनेन. 29. 1 Ś₁ K ततो जाम्बूनदो भूत्वा : rest जाम्बूनदगयो भूत्वा. 32. 3 Ś₁ K दशाणैः : rest गोकर्णै. 32. 12 Ś₁ K वा : rest मे. 33. 20 Ś₁ K ॰दृश्यते (K_s ॰थ्रिष्यते) : rest ॰रक्षमा. 36. 21 Ś₁ K D_s गवि जातो : rest तिग्मतेजाः. 38. 2 Ś₁ K D_s कृतोऽन्घया : rest कृतः शपन. 38. 14 Ś₁ K गुणापृति : rest समादितं. 42. 7 Ś₁ K हि : rest च. 94. 31 Ś₁ K D_s गृहान्निभो : others गृहान्तिकं, गृहं विभो, etc. 94. 93 Ś₁ K सहितध : rest समेताक्ष. 118. 1 Ś₁ K यथा विधि : rest विशेषतः. 124. 23 Ś₁ K D_s मनुष्या : rest मनुजाः. 128. 12 Ś₁ K D_s जाम्ब्यामाद्दृसत्तरे : rest भागीरथादृ° (double crasis!). 128. 15 Ś₁ K D_s कालिच्या : others भान्वबी, मान्वी, etc. 142. 23 Ś₁ K भुज्याभ्यां साधु योग्यक्य : others भुजयाः सारमर्षप, etc. 155. 13 Ś₁ K पुनः परिचरन्तृदा : rest स तं पर्यचरत्पुनः. 162. 6 Ś₁ K सोडमालस्सं : rest om. सः. 163. 7 Ś₁ K D_s ॰कोभवत् : others ॰के वभो, etc. 168. 3 Ś₁ K D_s पालिर्व्व : rest भारत. 169. 18 Ś₁ K D_s ते ततः : कोपात् : others ते महेष्वासाः, etc. 170. 9 Ś₁ K D_s सर्वलोकषु ससग : rest सर्वलोकपराध्वं. 177. 5 Ś₁ K D_s प्रकीर्तिता : rest समागताः. 181. 37 Ś₁ K D_s ॰ते च लंघिपते : others ॰लेऽभिगरछति, etc. 181. 40 Ś₁ K D_s प्राविशत्रात्त शीभसुः : others प्राधाणेः प्राविशत्तत्र ; etc., etc.

These concordances are sufficient for postulating the archetype v, comprising the versions Śārada and K, a hypothesis which will be confirmed by further agreements which are mentioned below.

The K version, though comprising manuscripts akin to each other and clearly distinguishable from those of the Devanāgarī version, is by no means—as is natural—quite homogeneous. Only K₀.₁ represent the version K in a comparatively pure form, while the

rêmaining manuscripts of the group (i. e. K₄-s) are really nothing more than misch-codices, being conflated either with Y or with S. On the other hand, just owing to this conflation, some of the other composite Devanâgarî manuscripts (particularly D₄-s) have so many features in common with K, that they may as well be separated from D and classed under K.

The contamination of K₄-s with Y is illustrated by the following passages: No. 14 of App. I (found in K₄ marg., and Ñ V₁ B D); No. 41 (in K₄-s and Ñ₂-s V₁ B D except D₅); No. 42–43 (in K₄-s and Ñ V₁ B D except D₅). K₄ includes passage No. 61 (of App. I) and 1131*, like Ñ B D. The contamination of K₄-s with Y is illustrated by 116*, 119*, 122*, 124*, 125*, 128*, 132*, 137*, 139*, 142*, 143*, 144*, 145*, 151*, 157*, 160*, 162*, 166*, 167*, 168*, 172*, 173*, 189*, 190*, 191*, 221*, 228*, 245*, 281*, 305*, 354*, 372*, 405*, 416*, 417*, 438*, 487*, 490*, 523*, 536*, 564*, 692*, 694*, 824*, 1000*, 1035*, etc., etc.

The contamination of K₄-s with S is exemplified by the following among other facts. K₄-s contain 22*, K₄-s 25*, K₄ 49*, K₄ (suppl. fol.) passage No. 55 (125 lines) and No. 100 (118 lines), of App. I: all of these are Southern passages.

K₄-s, moreover, contain the Brahmâ episode³ in adhy. I (a slippery passage, which migrates from place to place), while K₄-s have found place even for the venerable elephant-headed Gaṇeśa, who is unquestionably a late Northern intruder. In K₄ these interpolations are written out on separate folios (called here शोधपत्रा), and inserted at appropriate places, which shows the interpolations on the high road to recognition as genuine parts of the Mahâbhârata.

Important omissions which distinguish v (really only Ś₁ Ko-s) from all other manuscripts are these: (i) the adhyâya giving a naïve account of the birth of Duhṣalâ (Bom. adhy. 116), which uncommonly looks like being an afterthought (App. I, No. 63); (ii) a passage of about 25 lines describing how Droṇa’s son Âśvatthâman is given flour mixed with water, which he drinks in the belief that it is milk (App. I, No. 75); (iii) an adhy. (Bom. adhy. 139), in which there is an incidental allusion to the installation of Yudhiṣṭhira as Yuvarâja, and which is repetitious and incoherent (App. I, No. 80); (iv) the so-called polity of Kâṇi(ṇ)ka, Kâṇi(ṇ)kaṇîti (Bom. adhy. 140), which is a replica (naturally with many additions, omissions and variant readings) of the advice given by Bhâradvâja (apparently a gotra name of this very individual) to Śatruṁjaya, and duly communicated by Bhīṣma to Yudhiṣṭhira in the Śânti (App. I, No. 81); (v) the crossing of the Ganges by the Pâṇḍavas (Bom. adhy. 149), a superfluous adhyâya, which only serves to confound the already confused geography of the narrative (App. I, No. 85).

These five passages are found in all manuscripts collated except Ś₁ Ko-s, but it is worthy of note that even apart from their omission in v, the documentary evidence with regard to at least two of them, is confused and unsatisfactory. No. v (crossing of the

Ganges) is inserted in different groups of manuscripts at different points of the text. In No. iv (Kanikaniti), on the other hand, most of the Southern manuscripts repeat, after the interpolation, the immediately preceding portion of the original, apparently in order to restore the context broken by the intrusion of extraneous matter.⁶ Of important additions in v, I can recall only one, that of an adhyāya of 42 lines, at the very end of the Adi (added probably as an Appendix), which is a variant, abbreviated version of the Svetaki interpolation. It is found only in Ś₁ Ko._1-4 and therefore cannot even be said to be characteristic of the whole of v (App. I, No. 121). Instances of small additions are Nos. 349*, 449*, 451*, 516*, 565*, etc., found in K with or without some Devanagari manuscripts; while 969*, 1855*, 2077*, etc. are found in Ś₁ K, with or without some Devanagari manuscripts: all these passages are missing in B S.

That Ś₁ and K are not identical but independent (though allied) sources, may be concluded, for example, from 449*, 452*, 491*, 492*, 516*, 565*, 750*, 866* etc., which are found inserted in some or all manuscripts of the K version, but which are conspicuous by their absence in Ś₁ (sometimes with K₁). It was remarked above that v is the shortest of the extant versions of the Adi. Let us examine, without bias, this feature of v. Those passages that are lacking in v, in comparison with the other versions, cannot all be omissions in v, whether accidental or intentional.

They cannot be intentional omissions, notwithstanding that these missing passages are mostly of inferior character, intrinsically worthless, repetitious, superfluous, or finally such as scholars have already (even before the discovery of this version) marked as likely interpolations.⁷ For, this Saradā (Kasmiri) version of the Ādi is not an abstract or an adaptation. It claims to be the unabridged text itself, in all its fullness, and I see no sufficient reason to doubt the a priori presumption that it is not an abridged version. The explanation that primarily with the very object of excising what seems to us to be superfluous or repetitious matter, an abridgement might have been intentionally made in the past by some Kasmiri redactor or a syndicate of redactors, would be a grotesque distortion of Indian literary and religious tradition. No one in the past found the epic text too long. Far from it. It was perhaps not long enough. Taking away something from the received text of the Mahabhārata and passing it off as the original work is a thing categorically different from adding something to it. To add small details here and there, embellishing and amplifying the original, would be merely a gentle and lowly service ad majorem gloriam dei. Even long pieces may sometimes be added, if they are actually found in other Mahābhārata manuscripts; and occasionally, even if they are not found in the current manuscripts, provided there is at least oral tradition to support their claims.

No doubt the received text contained difficulties and obscurities and repetitions. But they would be merely due to corruptions of the text; the difficulties could be solved and the purpose of the repetitions explained by a really learned Pandit, who knows and understands everything.

That the omissions cannot be the result of a preconceived plan to shorten or to improve the text, follows further from two other facts: firstly, enough digressions and superfluities still remain in v, which would have all been swept away in pursuance of the alleged plan; and, secondly, v has its own interpolations, albeit they are few in number and short in extent, such as 349* (in K V1 Da D2.4), 451* (K Dn D1), 565* (K except K2), 1499* (Ś1 K D8), 1735* (Ś1 K1 only), 1855* (Ś1 K N1), 2077* (Ś1 K except K2 and Ñ1.3 V1 D2.5), etc.

While these so-called “omissions” cannot be all intentional, they can also not be all accidental. The text is continuous and complete in itself. It has no apparent lacunae, as it surely would have had, if the omissions had been due to fortuitous loss or destruction of some intermediate folios of a parent manuscript.

It may further be pointed out that many of the apparent “omissions” of v, in relation to T or the Vulgate (i. e. Nīlakantha’s text) are confirmed by the rival recension, the Southern recension; e. g. the Gaṇeśa episode (App. I, No. 1), or the anticipation of the list of the hundred sons of Dhṛtarāṣṭra (No. 41), or again the story of the birth of Abhimanyu (No. 42), or finally the anticipation of the story of the birth of Karṇa (No. 43) in the Sambhavaparvan. These passages are omitted in S no less than in K.

In these instances, moreover, the intrinsic probability is wholly on the side of those manuscripts that lack these accretions. It is unnecessary to dilate on the Gaṇeśa episode, which, on the face of it, is a later addition, and which has been dealt with so often by different critics.⁸ As for the two passages, Nos. 42–43 of App. I, it is sufficient to observe that the adhyāya in which they occur is meant to be a mere list of the dramatis personae, in which each actor in the great drama is identified as the incarnation of some god, goddess, or titan, taking this or that part in one momentous phase of an all-embracing cosmic movement. The adhyāya being originally a mere (metrical) list (as it is in the Southern recension),⁹ such stories as the account of the birth of Abhimanyu and Karṇa are wholly out of place here, and could not possibly have belonged to the original scheme of the adhyāya. The contrary supposition only stultifies the original writer, making him out to be an irresponsible lunatic, scarcely a desirable conclusion from the orthodox view-point.

Likewise many of the apparent omissions in v in relation to the Southern recension are confirmed by other Northern versions; e. g. the anticipation of the birth of Kṛṣṇa and Dhṛṣṭadyumna (App. I, No. 79), or the Nālāyanī episode (No. 100), or the account of a battle between the Kauravas and the Pāṇḍavas (No. 103), which are peculiar to S. In other words, these “omissions” are documented by the whole of N.

BY PROLEGOMENA

One notable feature of v to which I must now draw attention is its frequent agree- ment with S against Y, especially in the matter of isolated and even unimportant readings, scattered throughout this parvan. I shall cite a few (out of the hundreds of possible) instances to exemplify this interesting and important characteristic of v. The readings of conflated manuscripts, which serve only to confuse the issue, have been ignored; the references are, as usual, to the adhyāya and śloka.

1. 138 K V₁ S दुस्सहं (Text) : BD सुदुःसहं. 1. 144 K S पाण्डवं युध्यमानं : BD मण्डलशरत्तनं. 1. 208 K S चत्वार एकतो वेदाः : V₁ BD एकतश्चतुरो वेदाः. 20. 2 K S विष्णवो विष्णुपदनां (hypermetric!) : others विष्णुरुपां विनता, etc. 21. 10 K S 'घ्न° : N V₁ B D (mostly) 'गण°. 26. 9 K S नदीजलविषोषिणं : N V₁ B D (mostly) समुद्रजलशोषिणं (or °षि वि)णं.¹⁰ 34. 13 S₁ K S उत्पत्स्यति महातपाः : N V₁ B D (mostly) भविष्यति तपोधनः. 37. 5 S₁ K S पवित्रा : N V₁ B D (mostly) क्षीप्रगं. 39. 16 S₁ K S गुह्योरागरतस्तमः : N₁., V₁ B D स्थापयेते प्रगुह्य वै. 41. 17 S₁ K S पूर्वं : N₁., V₁ B D तीव्रं. 64. 10 K S त्वन्ति रावं (or रावेष्टु) विदहयाः पदपदैः सहिता मृदुः N V₁ B D (mostly) त्वन्ति राघमरामुद्रपदया महुलियपसः 68. 14 K S तणणदिवववेला : N V₁ B D (mostly) बालार्कसमतेजसा. 76. 22 K S पुरुषेन : N B D (mostly) द्वेष्यः पुंसाम्. 77. 4 K S सूयैः : N B D (mostly) सुग्री. 94. 12 S₁ K S सूर्यैर्काशो : N₁ V₁ B D (mostly) सूर्यैपोटमृत्. 100. 6 S₁ K S रातो : N₁., V₁ B D (mostly) सार्धं. 119. 8 S₁ K S मा दृश्यति (irregular) : N₁., B D मा द्राक्षीत्सर्ष (regular). 138. 17 S प्रासादाशरमनां निलैः : N₂.s V₁ B D तथेव वासजजननी. 141. 4 S₁ K S दुंडेडे रासशानो : N V₁ B D (mostly) दुंडे रसक्षा वै. 142. 18 S₁ K S प्रवृत्ताः : N V₁ B D (mostly) रक्षा. 143. 38 S₁ K S सेनायाच महालयनः : N V₁ B D प्रतिष्ठा महादय्यः. 159. 20 S₁ K S पृथिवी कूपः : N₂. V₁ B D भूमिवालयनः. 176. 5 K S कुरनन्दनः : N V₁ B D (mostly) पाण्डुनन्दनः. 182. 9 S₁ K S माद्रीसूतः सहदेवो जयचयः : N₂. V₁ B D पश्चादयं सहदेवरसखी. 187. 20 K S तत्समग्रवीड्राजा : N V₁ B D तमग्रवीत्तो राजा. 189. 23 S₁ K S भगवन् : N V₁ B D भुवनम्. 193. 1 S₁ K S चिन्तयामि : N B D चिकित्सामि. 196. 4 S₁ K N₁ S बहु : N₂.s V₁ B D बहु. 199. 12 S₁ K N₁ S घोषस्थितम् : N₂.s B D वाचयागतानम्. 199. 19 S₁ K N₁ S भर्तीरः : N₂.s V₁ B D नगरै.

Such extensive agreements in petty verbal details must necessarily be, in the main, an original inheritance, and could never be, in their totality, the result of contamination or conflation, as one may vaguely imagine they are; because to acheive them would necessitate more expenditure of energy than an ancient Indian redactor or reciter or commentator of the epic would bargain for. And even if one or the other of them had the requisite amount

PROLEGOMENA lv

of energy to use in this way, it would appear to him to be a ludicrous waste of it. We in the present century are apt to get nervous and irritable over misprints and variant lections. But an ancient Indian scribe, redactor or even commentator, not to speak of the common reciter (pāṭhaka)—if I read aright Indian literary history—was not perturbed in the least by a little difference in wording or in sequence, especially if the variant did not give an appreciably better or appreciably worse sense. The enormous and complicated critical apparatus assembled here, moreover, can leave us in no doubt as to the attitude of the custodians of the epic tradition towards paltry verbal details: it was that of total indifference.

Addition or omission of passages is, I may add, a variation of an entirely different order. If a reciter or commentator came across, in another manuscript, an additional passage, there was every chance of his copying it down somewhere, either in the margin of his own copy, or on a supplementary folio; for there would be, in his mind, always present the possibility that the passage in question was some part of the original that his own manuscript had unaccountably lost. How else, forsooth, could the passage get into the other manuscript?

In my opinion, therefore, this fact of the concord between v and S in small details, coupled with the almost entire lack of agreement as regards the additions peculiar to v or S, is the strongest argument imaginable for the independence of these two versions, and consequently for the primitive character of their concordant readings. It is needless to point out that this is a factor of supreme importance for the reconstruction of the original.

The text of v is throughout of such a character as to inspire confidence. Its conservatism is proved by its preserving archaisms and the `lectio difficilior` (e. g. अभिमोच्ये 1. 2. 144; श्लोकधर्म¹¹ 1. 2. 177, 189; काम्या adv. “frankly” 1. 10. 6; कल्पस 1. 98. 13; समुद्धे 1. 98. 18), often in a corrupt form, while other manuscripts have discarded them in favour of modern forms or easy paraphrases. It is well known that, for purposes of textual reconstruction, the mechanical corruptions of a stupid but faithful copyist are to be preferred to the intelligent copyings of a less faithful one.

Again, v is often the only version that has preserved the correct reading: e. g. 1. 2. 102: यत्र धृतार्णवे मग्नाद्रौपदीं नौरिवार्णवात्। तारयामास तांस्तत्त्मीणश्लाघ्या दुर्योधनो नृपः। पुनरेव ततो धूते समाहूयन्त पाण्डवात्॥ where the Vulgate version reads (1. 2. 138 f.): यत्र धूतार्णवे मग्ना द्रौपदी नौरिवार्णवात्। गच्छन्भ्रातिष्वृंदान्स्तान् दारवा दुर्योधनो नृपः॥ पुनरेव ततो धूते समाहूयत पाण्डवान्। while Sastri's reading is (1. 2. 108 f.): यत्र धूताखवे मग्ना द्रौपदी नौरिवार्णवे। गच्छन्तभ्रातिषृंदान्स्तान् दारवा दुर्योधनो नृपः॥ पुनरेव ततो धूते समाहूयत पाण्डवान्।

It is Draupadī who, like a canoe, rescues the Pāṇḍavas, who were submerged in the ocean of the dice-play. The correctness of the text reading, which is based on that of K, is proved by a stanza in the Sabhā (B. 2. 72. 3), which is the source of our stanza: अम्बुवेऽस्मनि मङ्गलानामनुष्ठिते निमज्जताम् । पाञ्चाली पाण्डुपुत्राणां नौरेषा पारगाभवत् ॥ Compare also the following three versions of 1. 166. 23 (=B. 1. 176. 27; M. 1. 174. 29):

K, N, V, B D = Vulgate ततो राजा परिक्रम्य यथाकामं यथासुखम् । निवृत्तोऽन्तःपुरं पार्थ प्रविवेश महामनाः ॥

v = Text अन्तर्गतं तु तद्राज्ञः तदा ब्राह्मणभाषितम् । सोऽन्तःपुरं प्रविद्रयाथ संविशेश नराधिपः ।

Southern Recension अन्तःपुरे गतो राजा श्रुत्वा ब्राह्मणभाषितम् । सोऽन्तःपुरं प्रविद्रयाथ न सस्सार नराधिपः ।

Obviously, the stumbling block was अन्तर्गते of the constituted text, which is a lect. diff.; here it means “forgotten”, a meaning cited in our dictionaries generally as an uncommon meaning given only by Indian lexicographers! Unless one here assumes v to be original, it is impossible to explain this divergence of v, γ and S, both of which give a possible though weak sense. An unbiased comparative survey of the different versions leads one to the conclusion that the Śāradā (Kaśmiri) version is certainly the best Northern version, and probably, taken as a whole, the best extant version¹ of the Ādi, a conclusion not based on abstract considerations, but one that may be verified inductively and pragmatically. As is natural, this version is, not by any means, entirely free from corruptions and interpolations. These must be carefully corrected and controlled with the help of the other versions, particularly of those of the rival recension.

Sub-Recension γ.

This sub-recension comprises the four versions: Nepāli, Maithili, Bengāli and Devanāgarī, and is represented by a very large number of manuscripts; it is, in fact, the most numerous group. Instances of readings which distinguish γ from v S, have been adduced above (p. LIV), to show the agreement between v and S against γ. The versions comprising this sub-recension have, moreover, quite a considerable number of “additional” passages in common, which clearly differentiate it from other versions. Noteworthy is the substitution of a lengthy passage of 56 lines (App. I, No. 61) for 1. 105. 4–7, giving a detailed account of the marriage of Pāṇḍu with Kunti and Mādrī. This detailed account is obviously secondary. On no other supposition can one, it seems to me, account for the circumstance that Śi Ko-s and S should agree in having a short version of the episode for being embedded in a portion of descriptive text with minimal variation. For, while it is inconceivable that two (more or less) independent groups of manuscripts such as Śi Ko-s and S could arrive at the same short account independently of each other, it is at the same time, extremely improbable that either group (Śi Ko-s or S) should have copied the short summary from the other, discarding altogether its own original detailed account.

The secondary interrelationship of the various versions comprising sub-recension 𝛾 is documented sufficiently clearly by their having in common quite a large number of lengthy passages which are missing in ν S, and which, on independent (intrinsic) grounds, have been or may be declared spurious. The following passages, given in App. I, are instances of such interpolations:

(1) No. 12 (Ṅ₁.₂ V₁ B D), a duplicate and superfluous description of the ocean, a similarly worded description having occurred only in the preceding adhyāya; (2) No. 14 (K₄ Ñ V₁ B D except D₅, D₂ on suppl. fol.), a short Purāṇic story relating how Sūrya resolves to burn the world down, whereupon Garuḍa, at Brahmā’s behest, brings his brother Aruṇa over to the east that he might act as Sūrya’s charioteer, shielding the world from the heat of the enraged Sun—a digression suggested by the casual mention of Aruṇa in adhy. 14; (3) No. 41 (K₈.₄ Ñ₂.₈ V₁ B D except D₅), a list of the hundred sons of Dhṛtarāṣṭra—an anticipation of adhy. 108, whose occurrence here (like that of the two following interpolations in the same adhyāya), as has been explained above, is obviously contrary to the original plan of the adhyāya; (4) No. 42 (K₈.₄ Ñ V₁ B D except D₅), an account of the scene which was enacted in heaven before the birth of Abhimanyu, a story which is really meant to explain the mystery of his premature death; (5) No. 43 (K₄ Ñ V₁ B D except D₅), the open secret of the mysterious birth of Karṇa, which is an anticipation of adhy. 104; and, finally, (6) No. 81, lines 193–230 (K₄ Ñ V₁ B D T₁), meant to be a summary of the Jatugṛha episode, which is is, however, a garbled and incoherent version of the original story.

The view that ν and 𝒱 may stand in genetic relation to each other does not receive much support from the facts of the case. Neither ν nor 𝒱 can be derived from the other. Each possesses original features that the other lacks, as is evidenced by their alternate agreement with S, even in the matter of petty verbal details. All these coincidences need not, of course, be original. Some could be indeed secondary changes, made independently in the same direction; others again may possibly be explained as the result of contamination. There will remain still an obstinate residue of agreements between ν and S, or between 𝒱 and S, that must be set down as the expression of the ultimate connection of the respective concordant versions through the lost original source.

Contamination between ν and 𝒱, owing to the contiguity of the areas in which the respective versions were current, was inevitable, and must, in any case, be assumed to have existed; on the other hand, contamination between 𝛾 and S cannot be altogether denied.

Particularly interesting is a small group of passages of doubtful character, to which reference has already been made. These are certain passages that are common to 𝛾 and S, and are missing in 𝕊₁ Ko-₈ only; in other words, they are found in all manuscripts collated except 𝕊₁ Ko-₈, for example, the Kaṇikanīti. There is usually other evidence against the passages. Thus the secondary character of the Kaṇikanīti is quite unexpectedly confirmed; firstly, by the illogical repetition in certain Southern manuscripts (T₂ G₂.₄.₅) of two preceding adhyāyas (129–130); and, secondly, from the fact there is no reference to the Kaṇikanīti in Kṣemendra’s Bhāratamañjarī, in the Javanese version, as also in Devabodhas’s

8

commentary. It may, of course, happen that in particular cases there is no collateral (confirmatory) evidence of this character available; e. g. in the puerile account of the birth of Duhṣalā (App. I, No. 63). Here the evidence of documentary and intrinsic probability is almost equally balanced; and documentary probability points in one direction, while intrinsic probability points in the other.

I have in such cases hesitatingly followed γ (=S, K), taking into account, on the one hand, the superiority of γ in general trustworthiness, and, on the other hand, the special characteristics of Y and S, which are versions rather of the inclusive than of the exclusive type, prone to amplification and elaboration. Fortunately for us such cases are comparatively rare.

The rejection, on the evidence of γ alone, of the whole of the incoherent adhy. 139 of the Bombay edition (our App. I, No. 80), an adhyāya which contains only some needless repetition, besides minor absurdities, would not have called forth any comment from me, but for the fact that with its omission disappears the only reference, I think, in the whole epic to this alleged installation of Yudhiṣṭhira as heir apparent to the throne of Hāstinapura. The Kaśmīrī version, which omits the entire adhyāya containing the reference, unexpectedly justifies the indignant outburst of Holtzmann (Das Mahābhārata, Bd. 2, p. 33): “Geradezu Fälschung ist es, wenn 1, 139, 1=5517 behauptet wird, der blinde Dhṛtarāṣṭra hate mit Uebergehung seiner eigenen Söhne den Yudhiṣṭhira zum Kronprinzen (yuvrāja) ausrufen lassen.” One of the main objects in interpolating this adhyāya seems to have been to exonerate Arjuna from the blame or sin of fighting with his own guru (Ācārya Droṇa) in the Great War, by making the Ācārya himself exact from his pupil in the presence of all his kinsfolk—for no reason that is adduced or can be seen — the solemn but senseless promise that he (Arjuna), when challenged, would not refuse to fight with Droṇa. Cf. B. 1. 139. 13:

आचार्यदक्षिणां देहि ज्ञातिग्रामस्य पश्यतः । ददासीति प्रतिज्ञाते फाल्गुनेनाभवीदृढः । युङेऽहं प्रतियोत्स्यो युष्मांस्त्वध्यानघ । तथेति च प्रतिज्ञाय द्रोणाय कुरुपुंगवः । उपसंगृह्य चरणौ स प्रायादुत्तरां दिशम् ।

There is no reference to this alleged promise in the sequel. And originally a different solution of the dilemma was obviously imagined. To Arjuna's question (Gitā 2. 4):

कथं भीष्ममहं संख्ये द्रोणं च मधुसूदन । इषुभिः प्रतियोत्स्यामि पूजार्हावरिसूदन ॥ the reply of Bhagavān Śrī Kṛṣṇa is (Gitā 2. 19, 32, 38): य एनं वेत्ति हन्तारं यश्चैनं मन्यते हतम् । उभौ तौ न विजानीतो नायं हन्ति न हन्यते ॥ अथ चैत्त्वमिमं धर्म्यं संग्रामं न करिष्यसि । ततः स्वधर्मं कीर्तिं च हित्वा पापमवाप्स्यसि ॥ सुखदुःखे समे कृत्वा लाभालाभौ जयाजयौ । ततो युद्धाय युज्यस्व नैवं पापमवाप्स्यसि ॥

Archetype ε.

This archetype is represented in our critical apparatus, by the three closely allied versions Nepāli, Maithili and Bengali; probably together with Uriyā (belonging to Orissa), of which version, however, no manuscripts were available for collation.

The Nepāli Version.

The Nepāli version is represented in our critical apparatus by the three manuscripts N₁, N₂ and N₃.³ The version is closely allied to the Bengali, with which the agreement of one or the other of the three manuscripts is almost constant. That even the manuscripts wholly free from contamination from some Southern source or sources (direct or indirect) follows, for instance, from 224*, 1096*, 1246*, 1470*, 1569*, 1748*, 1768*, 1778*, 1788*, 1828*, 1910*, 1957*, 2133*, etc., etc., as also passage No. 112 of App. I—interpolations common to S and some of the Nepāli manuscripts. One of these manuscripts (N₂) happens to be the oldest of the dated manuscripts (A. D. 1511) belonging to our critical apparatus.

The Maithili Version.

Of the Maithili version, which is the version of North Bihar, only one manuscript (V₁) was collated for this edition. V₁ and K agree sporadically against all other manu- scripts (cf. for instance, 1. 1. 8, 49, 162), but such agreements are few and far between, and it would not be safe to draw from them any far-reaching conclusion regarding the relationship of V₁ and K. As in 306*, 321*, 328*, 346*, 378*, 418*, 450*, 541*, V₁ agrees, on the other hand, with the typical Bengali—Devanāgarī group against all other manuscripts. V₁ contains 1548*, a Southern passage, found otherwise only in Dn D₁₄₅.

The Bengali Version.

The Bengali version of sub-recension γ was studied more carefully than either the Nepāli or Maithili. The study of this version was facilitated by the extreme courtesy and kindness of Pandit Vidhushekhara Bhattacharya, who has, now for many years, kindly and unselfishly supervised the work of our collation centre at the Visvabharati, a centre organized by Professor M. Winternitz, when he was residing at Bolpur as a Guest Professor in Rabindranath Tagore's University. With the co-operation of a select batch of advanced students, Pandit Vidhushekhara has been good enough to supply the Institute regularly with carefully prepared collations of a large number of valuable old Bengali manuscripts in the rich collection of the Visvabharati, as also of other manuscripts placed at his disposal by different Bengali Institutes and scholars, among the latter, my kind friend Professor Sushil Kumar De, of the University of Dacca. Of the large number of manuscripts thus collated, ultimately five were selected for inclusion in the critical apparatus of the edition. Notwithstanding considerable variation in these manuscripts as regards petty verbal details, the material appears sufficient to settle the text of this important version.

The Bengali version is closely allied to the Vulgate, but is unquestionably superior to the latter in so far that it is happily free from a large number of late accretions which encumber the Vulgate. Of such "omissions", exhibiting the superiority of the Bengali version, the following will serve as illustrations: (1) The entire Brahmā-Gaṇeśa complex in adhy. 1, of which the Bengali version contains not the remotest trace. The spurious character of this passage has been discussed and demonstrated so often that it is unnecessary to dilate upon it here.⁴ (2) The short dialogue of 8 lines (71*) between Paraśurāma and the shades of his ancestors, in the beginning of adhy. 2, which is wholly unnecessary here, and is, as a matter of fact, only an excerpt from a detailed description of the principal Indian tirthas, which occurs in the Araṇya (B. 3. 83. 29 ff.). (3) A short passage of only six lines (cf. App. I, No. 13), which represents a somewhat feeble attempt (as unnecessary as it is unsuccessful) to fill out an apparent lacuna in the original.⁵ (4) A long interpolation (App. I, No. 78) of 119 lines in adhy. 138 (Bom. ed.), which gives an inflated account of the defeat and the ultimate capture of Drupada by the Pāṇḍavas. It is one of the miniature Bhārata-yuddhas—mere by-play for the benefit of the gallery—which expand and embellish the Southern recension and the Vulgate. The older version disposes of the battle in two lines, which, taking everything into con- sideration, is after all perhaps not a very inadequate treatment, as already remarked. (5) More than usual interest attaches to another omission in the Bengali version, which concerns a well-known and popular scene describing the discomfiture of Karṇa at Draupadī's svayaṃvara, which is commonly believed to be one of the main reasons why he always entertained feelings of such deep and implacable hatred towards Kṛṣṇā (Draupadi), and lost thereafter no opportunity to hurt and humiliate her. This passage deserves a detailed consideration. Ramesh Chandra Dutt, who had to make a very careful selection of the incidents of the epic in compressing the story, has made this scene the centre of his poetic account of the marriage of Draupadī, and given a vivid rendering of the passage in his Epic of the Bharatas: > “Uprose Karna, peerless archer, proudest of the archers he, > And he went and strung the weapon, fixed the arrows gallantly, > Stood like Surya in his splendour and like Agni in his flame,— > Pandu's sons in terror whispered, Karna sure must hit the aim! > But in proud and queenly accents Drupad's queenly daughter said: > ‘Monarch’s daughter, born a Kshatra, Suta’s son I will not wed.’ > Karna heard with crimsoned forehead, left the emprise almost done, > Left the bow already circled, silent gazed upon the Sun!” The situation is, undoubtedly, full of dramatic possibilities. Just at the moment when the prize was going to be snatched away from the heroes of the epic by an upstart,

The brave little Draupadi comes to the rescue and snubs openly, in the presence of the assembled princes, the semi-divine bastard, the understudy of the Villain of the piece, the unwanted suitor, who thereupon withdraws discomfitted; and everybody breathes a sigh of relief. A tense scene!

Unfortunately, this melo-dramatic interlude, to judge by the documentary evidence, appears to be the handiwork of a very late Vyāsāid, as it is found only in K₄ N₂ Dn D₂.₄.₅, that is, one manuscript of the K group, one Nepāli manuscript, and three composite Devanāgarī manuscripts, besides the Nīlakaṇṭha version! All of these are late and inferior or conflated manuscripts. It is missing, on the other hand, not only in the Śāradā version and the Southern recension (as in the case of many of the interpolations of the Vulgate), but for once, also in the entire Bengali version!

It might seem a piece of sheer vandalism or perverseness to omit this seemingly beautiful little passage, which has won its way into people's hearts, from an edition of the Great Epic of India, relying merely upon documentary evidence. A little reflection will, however, convince any one that the loss to the epic is not as serious as one might, at first, suppose, since it is a palpably faked and thoroughly unreal situation. If one thinks about it at all, one fails to understand how Draupadi, who was, after all, then only an unexperienced maiden in her teens, had recognized the King of Aṅgas (whom she had probably never seen before) and known him for the son of a coachman, unfit to wed a princess. He had been invited by her father. At least he was given a seat of honour among the princes. He is specifically named by Dhṛṣṭadyumna among the suitors (1. 177. 4). Moreover, it does not appear as if the bride elect had much choice or voice in the matter, at the time of these elaborate and formal state functions notwithstanding that they were called svayaṁvaras. She had to wed any competitor who excelled in the particular proficiency test which had been arranged by her father or guardian. She was viryaśulkā: she was given by her guardian to the highest bidder, the price paid being heroism, or rather proficiency in marksmanship. This is quite evident from the words of Yudhiṣṭhira, addressed later to the Purohita of Drupada (1. 185. 23 f.):

प्रदिष्टशल्क़्का द्रुपदेन राज्ञा सात्त्वेन तथादृषा । न तत्र वर्णेषु कृता विवक्षा न जीवशिल्पे न कुले न गोत्रे ॥ कृतं सज्जैन हि कार्मुकेण विद्वेन लक्ष्येण च संनिधौ । सेयं तथानेन महात्मनेह कृष्णा जिता पार्थिवसंघमध्यै ॥

We accordingly find, as a matter of fact, that without murmur or hesitation, she follows an unknown and apparently undistinguished Brahman boy—Arjuna in disguise—who happens to have hit the mark. She does not know him from Adam, but she makes no inquiries about his status or lineage. Even if this were regarded as a case of romantic love at first sight for the handsome and heroic bowman (which it certainly is not), she never opens her lips when Yudhiṣṭhira proposes that she should be the common wife of the five brothers, which must have shattered her romance to smithereens, but quietly submits to (what is made to appear) as a most unusual and unnatural, if not a shocking, proposal, and from which even her old father and brother recoil with perplexity and amazement. It seems to me, therefore, that the documentary evidence is amply supported here by intrinsic probability.

Examples of other less important "omissions" in the Bengali version which distinguish it from the Devanāgarī are: 54*, 60*, 71*, 152*, 171*, 274*, 277*, 689*, 1171*, 1205*, 1222*, 1270*, 1614* (proverbs, one of them being a citation from Manu), 1714* (a short list of sacred rivers), 1788*, 1827*, 1841*, all of which occur in the Vulgate, but are missing in the Bengali version. Occasionally Bengali manuscripts agree in their recension, standing in opposition to Ś₁ K (with or without D); e. g.: 1. 22 B S शुचिः K (mostly) D (mostly) विर्ष. 1. 42 B S आत्मवान् : Ko. 2-4 D (mostly) एव च. 7. 3 B D (mostly) S परान् : K कृळान्. 39. 10 B D (mostly) S तत् : Ś₁ K (with a few D) पुनः 64. 29 Ń B D S जगम् : K ददरी, etc., etc., etc. Other examples have been cited under the description of the K version. In these cases, I have, as a rule, given preference to the S, on the postulated principle of the originality of the agreement between independent versions, adopting in the constituted text, the concordant reading; but owing to the circumstance, that sporadic contamination between B and S, as a whole, cannot be altogether denied and that there are, as a matter of fact, some Bengali manuscripts that influence of the Southern tradition, even in the matter of minor readings, it is impossible to be perfectly certain about the originality of a reading common to B and S. I am, however, of opinion that the probability is always on the side of the concordant reading, though the evidence of this agreement may be rebutted by other considerations, such as intrinsic probability or the evidence of pertinent testimonia.

The Devanāgarī Version.

The Devanāgarī script plays in the Mahābhārata textual tradition the important rôle of being the commonest medium of the contamination of different Mahābhārata versions. A Devanāgarī manuscript of the Mahābhārata may, in fact, contain practically any version or combination of versions. Of the four "Devanāgarī" scholiasts whose commentaries were collated for the Ādi, Arjuniśra is certainly an Easterner, and bases his commentary on the Bengali text; Ratnagarbha appears to be a Southerner, and his text is evidently a blend between the Northern and the Southern texts; while Nīlakaṇṭha is quite definitely a Westerner, though he seems to have written his commentary in Benares. The provenance of the fourth and the last commentator mentioned above cannot be determined with certainty; but it might be surmised that Devabodba was a "Northerner"; in any case, his text (to judge by the lemmata in his commentary) shows remarkable affinities with the North-western or Kaśmiri version (v). Most of the Devanāgarī manuscripts, as already remarked, are eclectic on no recognizable principle: now they approach the Southern tradition (S), now the purer Northern (v). If any one were to maintain that just this composite text was the original, a patchwork of disjointed ancient passages, which had later split up into the Northern and Southern recensions (as might easily be implicitly assumed by the protagonist, say, of Nīlakaṇṭha's version), it would be a thesis difficult to substantiate. It

seems more natural to regard, as already observed, the Devanāgari as a sort of "vulgar" script (like the Latin in Europe), the script understood by the savants all over India, into which many of the local versions were, from time to time, transcribed, a circumstance which facilitated contamination and conflation.

It has been mentioned above that the Devanāgari version contains many more interpolations than even the Bengali. It would be no exaggeration to say that the Devanāgari manuscripts, which are by far the most numerous of Mahābhārata manuscripts, are, at the same time, the least important of them, with the possible exception of those of the adjoining version, Telugu.

The Devanāgari Version of Arjunamis'ra.

This is in a sense a misnomer, because this Devanāgari version, as already remarked, is nothing but a Devanāgari transcript of the Bengali version. Instances of the concord of B and D*a* will be found under: 1. 4. 6; 7. 13; 8. 22; 10. 2; 11. 7; 26. 38; 33. 25 f; 111. 4; 141. 21; 143. 6; etc., etc.

The name of the commentary is variously given as (Mahā)Bhāratārtha(pra)dipikā, and Bhāratasaṁgrahadīpikā.¹ The commentary on the different parvans has been handed down singly or in groups of a few parvans at a time. Complete manuscripts of the manuscripts, which are written in Bengali or Devanāgari characters, have various dates in the seventeenth or later centuries; the earliest hitherto reported date is V. Saṁvat 1676 (ca. A. D. 1620). Arjunamiśra, who styles himself Bharātācārya in the colophons of his commentary, was the son of Īśāna, who was a "Reciter" (pāṭhaka) or "Prince of Reciters" (pāṭhakarāja) of the Mahābhārata, and who appears to have borne, like his son, the title Bharātācārya. Arjunamiśra is cited by name by Nīlakaṇṭha once in his commentary on the Mahābhārata (ad B. 3. 291. 70) and was, therefore, certainly anterior to Nīlakaṇṭha, who belongs to the last quarter of the seventeenth century. Arjuna, in turn, mentions among his predecessors: Devabodha, Vimalabodha, Śāṇḍilya, Sarvajña-Nārāyaṇa (also known as Nārāyaṇa-Sarvajña or merely Nārāyaṇa). He appears to have based his scholium closely on that of Devabodha, from whose commentary Arjuna often cites, verbatim long extracts, without specifically naming the source. Arjuna wrote also a commentary on the Puruṣasūkta, to which he himself refers in the Dīpikā on B. 14. 25. 26. Telang² surmises that he is posterior to the Vedantist Śaṁkarācārya; and Holtzmann³ assigns him to the thirteenth or fourteenth century, both without mentioning any cogent reasons for their assumptions. Arjuna has treated the Harivaṁśa as an integral part of the epic, elaborately defending this position; his commentary, therefore, embraces the Harivaṁśa also.⁴

Following the example of my predecessors, I have utilized Devanāgarī manuscripts of his commentary and treated his version as a sub-division of the Devanāgarī version. The two Devanāgarī manuscripts utilized by me are, however, extremely corrupt. Moreover, the text they contain is evidently contaminated from the Vulgate, as proved by the glaring discrepancies that exist between the readings of the text and the lemmata in the commentary (e. g. 1. 1. 17, 22). This corruption of the Arjumanīśra manuscripts, I could not explain at first, but now it is clear that it is due to their being faulty tran- scripts of Bengali originals. Two such Bengali manuscripts⁵ (unaccompanied by the epic text) were sent to me subsequently by my kind friend Professor Sushil Kumar De of the University of Dacca from the collection of the Dacca University. These manuscripts are far superior, as is but natural, to the Devanāgarī manuscripts. It would seem, therefore, expedient to secure and use, whenever possible, good old Bengali manuscripts of Arjuna- miśra’s commentary, treating his version as an offshoot of the Bengali version (with the symbol Ba); or, still better, such Bengali manuscripts of his commentary as are unaccompanied by the epic text. The reason of the last precaution will be presently explained.

A word of caution is here necessary in regard to what are cited in the critical notes as the readings of Arjumanīśra. The readings found in the (epic) text accompanying the commentary have, as a rule, been taken to represent the readings of Arjumanīśra. The commentary was consulted by me only occasionally, in case of doubt or difficulty, or when a pāṭhāntara was noticed during a hurried perusal of the commentary. It is, therefore, more than likely that, since the (epic) text of our Arjumanīśra manuscripts is conflated with particular with the Nilakantha type, some errors in our readings have crept in.⁶ Such errors can, however, be rectified only by carefully working through the whole commentary word for word, and comparing the lemmata with the (epic) text of the manuscripts. Even then one can, of course, be sure only of the words and passages actually cited by the scholiast.

In passing, it may be mentioned that the practice of combining text and commentary in one manuscript is probably not very old. It is almost certain that the autograph copy of the commentator was not made up on the tripartite system of combining the epic text and commentary in such a way that text occupies a central strip of the folio, while the commentary is written in two narrow strips, one at the top and the other at the bottom of the folio, which is the prototype of the Bombay potḥi-form editions. The scholiast must have written his commentary, certainly at first, on separate leaves, especially in the case of voluminous texts like those of the two epics. Accordingly the commentaries of Devabodha and Vimalabodha have been handed down always unaccompanied by the epic text. Those of Arjumanīśra and Nilakantha, on the other hand, are generally accompanied by the epic text, but the two Dacca manuscripts (lent to me by Professor De), as was mentioned above, contained only the commentary. The two elements—text and commentary—appear to have been combined into the tripartite form by professional scribes. If this combination was done under the supervision of the commentator or at

least in his lifetime, there is some chance of the scribe's reproducing, in an approximately correct form, the text of the commentator. But if the combination is made independently of him and especially if made some time after the death of the commentator, there is every chance that the scribe would combine the commentary he was copying with some text known better to himself than to the scholiast. In the latter case, therefore, it must remain doubtful how far the epic text of such a manuscript resembles the text actually commented upon by the scholiast. It is consequently best to use always texts of the epic text, though it is an extremely laborious process to collate such a manuscript with any given Mahābhārata text; but we eliminate in this way automatically all chances of avoidable errors of commission and omission.

The Devanāgarī Version of Nilakaṇṭha: the Vulgate.

Nilakaṇṭha, considered until lately, at least in India, as the most trustworthy guide for the exposition of the Mahābhārata, was a Brahmin scholar of Mahārāṣṭra, with the surname Caturdhara (modern Chaudhari), son of Govinda Sūri and Phullāmbikā, residing at Kūrparagrāma (modern Kopargaon) on the Godavari.⁷ Nilakaṇṭha wrote his commentary on the Mahābhārata (and another work called the Gaṇeśagītā), in Benares, in the last quarter of the seventeenth century. He appears to be the author also of a work called Mantrarahasyaprakāśikā.

At the beginning of his commentary on the Great Epic, Nilakaṇṭha tells us that before writing his scholium, the Bhāratabhāvadīpa, he had compared many copies of the Mahābhārata, collected from different parts of India, with a view to determining the "best" readings and even consulted the scholia of old authorities:

बहुन्‌समाहृत्य विभिम्नदेश्यान्‌कोशाम्न्निखिलान्‌च पाठान्‌मम् | प्राचां गुरुणाम्मधुल वाचमास्मृत्य भारतभावदीपः ||

We accordingly find that he occasionally mentions (in about 125 places) variant readings and additional passages found in different provincial versions (most of which can be identified among the readings of the manuscripts comprising our critical apparatus), and cites (as a rule, without naming the source) the explanations given by other scholiasts— information, scanty though it is, yet of immense interest and value for the history of the received text. Variants cited by Nilakaṇṭha will be found in the footnotes under: 1. 1, 1, 3, 4, 6, 8, 13, 19, 22, 41, 80, 100, 118, 129, 185, 188; 2. 6, 64, 243; 3. 19, 149, 189; 4. 1, 5, 9, 11; 1. 13, 2, 29; 14. 8, 16, 16, 10, 33 (found only in Od!); 18. 11, 19; 6. 24, 9, 27, 35, 28, 24; 30. 5 (not found elsewhere!); 11, 32, 18; 33. 20; 38. 30; 39. 11, 46, 29, 29 (not found elsewhere!); 49. 4, 17; 50. 9–12. 17; 51. 4 (not found elsewhere!); 5; 58. 34; 54. 8, 3; 55. 57, 21, 22, 78; 58. 35 f., 50; 59. 54, 62, 10; 68. 38; 69. 26, 70. 3, 19, 46, 71, 81, 81, 82. 8; 87. 12; 88. 22; 89. 51, 92, 43, 102, 23, 109, 10, 12, 15 (not found elsewhere!); 110. 33, 114. 2; 117. 9; 118. 9; 120. 10 (Nilp समन्वायात् as in text; om. through oversight; cf. B. 1. 130. 10); 124. 32; 125. 2; 131. 8 (''Gauḍapāṭha''); 141. 7; 143. 12; 148. 10; 150. 15; 153. 3, 154. 2, 13; 155. 28, 34, 49; 158. 14 (mentions Devabodha!); 46; 161. 4 (not found elsewhere!); 168. 25; 169. 20; 170. 21; 171. 7; 178.

9; 186. 1; 188. colophon (mentions S interpolation, the Nālāyani episode); 190. 5; 191. 18; 192. 10, 27; 197. 14; 199. 19, 30; 206. 2; 207. 23; 214. 9, 11; 218. 31; 33 (not found elsewhere!); 219. 3; 221. 5 (not found elsewhere!); 223. 17. The readings of Nīlakaṇṭha’s own text are, as a rule, inferior; our text readings will be found mostly among Nīlakaṇṭha’s pāṭhāntaras.

Nīlakaṇṭha refers to Devabodha, Vimalabodha, Arjunamiśra, Ratnagarbha, and Sarvajña-Nārāyaṇa, in the course of his comments on the different parvans. To Devabodha, who is one of the oldest (if not the oldest) commentators of the Mahābhārata hitherto known, he refers while commenting on 1. 158. 14 (=B. 1. 170. 15.):

न नंदंसाः शृण्वो वा न च देवाज्ञाअजः। कुबेरस्य यथोषीणं किं मां समुप्तसर्पय॥ इति प्राचीनः पाठो देवबोधादिभिर्व्याख्यातत्वात्।

Not a single word of this stanza, as cited here, is commented on, however, by Devabodha! The only word in Devabodha’s scholium which might possibly have been taken from some reading of the stanza before Devabodha is दाशूनाः (= पक्षिणः) and that does not occur in the reading of the stanza cited by Nīlakaṇṭha. The mention of Devabodha by Nīlakaṇṭha here, is, therefore, surely _honoris causa_. Such mistakes by commentators are far too frequent to cause surprise or need comment.⁸ It is, however, noteworthy that the reason Nīlakaṇṭha assigns for considering this as an _ancient_ variant is that it had been commented on by Devabodha and others. This shows that Nīlakaṇṭha held Devabodha in high esteem, and reckons him among the _ancient_ authorities. What Nīlakaṇṭha regards as “ancient” (_prācīna_) is of course a matter for speculation. Nevertheless I do not think that he would have called Devabodha a “prācīna” commentator, unless the interval between them was at least four or five centuries. Nīlakaṇṭha refers to Devabodha again in B. 7. 82. 2: मधुपर्का: मधुपर्कसमये पठन्त इति देवबोध: Arjunamiśra he cites in his comment on B. 3. 291. 70: जाह्रुण्यान् त्रिगुणदक्षिणान् इत्यर्जुनमिश्र:

Since Arjunamiśra also cites Devabodha, we can arrange the three commentators in an incontrovertible sequence: Devabodha—Arjunamiśra—Nīlakaṇṭha.⁹

The text used or prepared by Nīlakaṇṭha is a smooth and _eclectic_ but inferior text, of an inclusive rather than exclusive type, with an inconsiderable amount of Southern element.

As instances of simplification in the Vulgate, I may cite: 1. 2. 144 Text चाक्षिमौः (Vulga. वा विभोः cf. 1. 13. 20; 41. 21); 2. 189 श्लोकानां (श्लोकानां); 10. 6 कामया (कामां मा; cf. 1. 187. 6); 37. 10 दिवं सत्यज्जेव विधितः (शवं स्कन्धे प्रतिहितैः); 39. 16 दित्स (दीदैहि); 45. 16 बाल एवानिवाणतोऽसि (‘निष्पिक्तस्त्वं); 62. 12 सस्छ° (सर्वे°); 96. 16 शृगुणाणां (सर्वेषां); 122. 5 प्रवृशृति (हरष्युत्); 122. 42 तदत्तं (तदेत्तन्); 139. 18 विल्ज्जमानेवता (लज्जमानैव ललना); 150. 8 वस्ती: (रजनी:); 221. 1 शुक्के (बद्धौ); etc., etc.

Instances of the correction of solecisms in the Vulgate are: 1. 2. 93 Text यद्य (Vulg. गृहीत्वा); 9. 2 चिल्प (स्थूला); 119. 8 मा द्रक्ष्यति (मा द्राक्षीत्सर्वे); 181. 25 अहनन् (अभवदीत्); 134. 1 प्रवीढि (वदख); etc., etc. I add a selection of Southern passages which were interpolated into the Northern recension by Nīlakaṇṭha or by one of his immediate predecessors in the field: 263*, 299*, 473*, 513*, 598*, 700*, 701*, 722*, 857*, 863*, 963*, 977*, 1037*, 1054*, 1062*, 1066*, 1069*, 1100*, 1101*, 1169*, 1211*, 1548*, 1768*, 1828*, etc., etc., as also passage No. 56 of App. I. Nīlakaṇṭha’s text has acquired in modern times an importance out of all proportion to its critical value,¹⁰ to the utter neglect of far superior texts, such as the Kāśmīrī or Bengalī. Nīlakaṇṭha’s guiding principle, on his own admission, was to make the Mahābhārata a thesaurus of all excellences (culled no matter from what source). At the beginning of his commentary on the Sanatsujātīya, Nīlakaṇṭha naïvely remarks (Bom. ed. Udyoga 42): उद्योगपर्वणि सनत्सुजातीये भाष्यकारादिभिव्याख्यानत्वेनप्रतितनपुस्तकेषु च स्थितत्वात् श्लोका गुणौषसंहारन्यायेनैकीकृत्य व्याख्यायते ¹¹ That Southern manuscripts were utilized by him is incontrovertibly proved, for instance, from the fact that he cites at the end of his comment on Ādi 196 (Bom. ed.), the Nālāyanī and Bhaumāśvī episodes (in two adhyāyas), which are typical Southern interpolations, not found in any Northern manuscript: अत्र यत्रैवैषा द्वयोरितिहासा त्रिपथगां नदीमिवत्येनो नारायणमुख्यानमन्यैरुख्यानात्वेनकः कचिन्पुस्तके पठ्यते ¹² Characteristically the scholiast speaks only in general terms (कचिन्पुस्तके) without furnishing any further information about the manuscripts in question. But, fortunately, he is not always so reticent. Thus he mentions specifically the Bengalī version, while commenting on B. 1. 145. 20 (अत्रोत्तराधे गौडपाठ एव दृश्यते) and elsewhere; cf. his notes on B. 3. 119. 3, and on 6. 43. 1 (गीता सुगीता कर्त्तव्या इत्यादयः साक्षीः पञ्च श्लोका गौडैर्न पठ्यन्ते). It must be said to his credit that there is at least one place where he honestly confesses his inability to understand the confused textual tradition, and that is in his comment on B. 1. 22. 1: नागाश्च संविढं कृत्वेति द्वादशश्लोकमध्यायै केचिच पठन्ति । काविद्राख्यान् श्लोकान्पूर्वचैव च पठन्ति । अन्ये तु षड्भा- नवब्धपीलंग्य कोषवृद्धि न प्रतीमः ¹³ The (printed) editions of Nīlakaṇṭha’s version leave much to be desired. They have arbitrarily changed many of the readings and added a certain number of lines which are not found in the Nīlakaṇṭha manuscripts hitherto examined. Instances of lines or stanzas with which modern Pandits have enriched most of our (printed) Northern editions and which are lacking even in the Nīlakaṇṭha manuscripts, are besides a (Southern) passage of 21 lines given in App. I (No. 112) and another of 9 lines (998*), the following short interpolations:

27* शतसहस्रैश्छ्रोकाणां पुण्यकर्मणाम् । उपाख्यानैः सह श्रेयं श्रव्यं भारतमुत्तमम् ॥ B. 1. 1. 101f. 146* संशसकानां वीराणां कोट्यो नव महात्मनाम् । किरीटिनाभिनिष्कृम्य गमिता यमसादनम् ॥ B. 1. 2. 261 148* धृतराष्ट्रस्य पुत्राश्च तथा पापाणयोजिनः । नारायणाश्च गोपालाः समरे चित्रयोधिनः ॥ B. 1. 2. 262 224* शौनकस्तु महासत्वः सर्वभागंवनन्दनः । जातसत्पत्ति चेव व स्चितः स्थिरयशास्त्रतः ॥ B. 1. 8. 3 314* गुरुर्हि सर्वभूतानां ब्राह्मणः परिकीर्तितः । B. 1. 28. 4 752* मरणं शोभनं तस्य इति विद्रुजना विदुः । B. 1. 79. 13 1048* उक्त्वा जन्मकुलं मर्ह नास्ति दाशसुतेति च । B. 1. 105. 9 1099* खल्पमेघ यथा वृतं दानं बहुगुणं भवेत् । अधर्मे एवं विप्रषे बहुदुःखफलप्रदः ॥ B. 1. 108. 12 1805* आहारिश्चर्ययं नूनं प्रीति वो वर्धयिष्यति । B. 1. 184. 19 1957* इत्युक्त्वा प्रययौ राजनिंदुरः सं निवेशानम् । B. 1. 200. 26 2043* आर्य पशुपतेः खानं दर्यानादेव मुक्तम् । यत्र पापपि मनजः प्राप्नोत्यभयदं पदम् ॥ B. 1. 217. 35

It would, however, hardly repay, now, the trouble to re-edit, from manuscripts, the version of Nilakantha, as there are far better versions that could be edited instead, for instance, the Kasmīrī.

The manuscripts of the Nilakantha version (which show among themselves slight discrepancies) contain a number of lines which are not found in any of the other versions (except occasionally in a few manuscripts of the composite Devanāgarī version); e. g. 102*, 147*, 276*, 412*, 493*, 574*, 699*, 765*, 838*, 1270*, 1457*, etc. They belong perhaps to the oral tradition which, at one time, had probably as great value and authority as the written text.

Nilakantha has misunderstood the text, and given doubtful, far-fetched or fanciful interpretations at: B. 1. 1. 52 (मदुः=मञ्जः!), 275 (नकुलकः), 2. 33 (शौनक=श्श्रुतम्!), 17. 12 (कलशः), 23. 15 (Vedantic interpretation); 27. 8 (मनःसंर्पणै), 37. 15 (the difference between हेतु and कारण); 43. 22 (गर्दैः); 47. 11 (श्वेतकाकैः); 50. 3 (भक्षणं=आचार्ये!), 61. 11 (श्वेतकटः); 63. 90 (संहिता); 131. 52 (अवसीदेत); 164. 9 (context); 166. 10 (एकतमं); 232. 1-7, 19 (esoteric meaning); etc., etc.

Nilakantha’s stanza (B. 1. 145. 20): प्राज्ञः प्राज्ञमलापक्षः प्रलापपक्षमिदं वचः । प्राज्ञो प्राज्ञः प्रलापज्ञः प्रलापज्ञं वचोऽब्रवीत् ॥

which appears to be sheer nonsense is so in fact. No other version, as far as I know, contains this mystifying repetition. The explanation of the stanza given by Nilakantha is childish, to say the least.

The stanza containing the unintelligible word कुलिश (v. 1. कलिञ्ज), which Nilakantha has great difficulty in explaining:

ततो दुर्योधनः शूरः कुलिशस्य मते स्थितः । पाण्डवान्विषयोपायैः राज्यहेतोरपीड़यत् ।

looks uncommonly like one of the kūṭaślokas, said to be interspersed by Vyāsa at different places in his poem, in order to puzzle and confuse his divine amanuensis, but is, un-

fortunately, nothing of the kind. The passage is only one of the common instances of “conflate” readings. The stanza cited above is the Southern variant (473*) of the Northern stanza, which, in our edition, reads (1. 55. 8): ततो दुर्योधनः क्रूरः कर्णश्च सहसौबलः । तेषां निग्रहणार्थानिचविधिष्यास्ते समारम् ॥ The पुलिङ्गस्य in the former stanza is only a mislection of the original कणिकस्य (often mis-written कणिक्स्य, कणिकस्य), which is the Southern equivalent of कणिकस्य, the reference being, no doubt, to the minister or statesman (mantrin) Kaṇika (named after the famous authority Kaṇika or Kaṇiṅka cited in the Arthaśāstra of Kautilya), who appears only once in the epic, and that expressly for the purpose of expounding his political philosophy to the Kauravas. As another instance of conflation which has had a rather disastrous effect on his text, I may cite Nīlakaṇṭha’s version of the story of Dīrghatamas. The addition has been made in such a manner that one sentence of the original has remained hanging in the air and cannot be construed at all! The story begins at B. 1. 104. 9. All goes well till stanza 28: अहोऽयं मिश्र्मर्यादो नाश्रमे वस्तुमर्हति । तस्मादेनेन वयं सर्वे पापात्मानं त्यजामहे ॥ Then we read 29: इत्यन्योन्यं समाभाष्य ते दीर्घतमसं मुनिम् । पुत्रलामा च सा पत्नी न तुष्टापि पतिं तथा ॥ “Having spoken thus among themselves, they [ scil. the inmates of the hermitage ] to the anchorite Dīrghatamas. Then that wife also, having (already) obtained sons (?) (from him) did not (seek to) please the husband.” Bhīṣma, who is narrating the story, then goes on quite unconcernedly to speak about the wife (of Dīrghatamas) Pradveṣī or Pradviṣaṇtī; about the maryādā made by the exasperated Dīrghatamas, and so on. But what the inmates of the hermitage (āśramavāsinaḥ) did to Dīrghatamas, we never learn from the Vulgate. All modern translators try to eke out a sense by interpolating into the text some words to complete the sense. A reference to the constituted text and the critical notes will, however, show that the text of the Vulgate is conflated; it is a most clumsy blend of interpolations from two entirely different sources (Y and S), which, as is but natural, alters the situation considerably and confuses the narrative hopelessly. By athetizing either passage we get a tolerable text; by athetizing both we get the original, which is the constituted text.

The Devanāgarī Version of Ratnagarbha.

The critical notes contain only specimen collations of this version, which is a blend between the Northern and Southern recensions. Like the Telugu manuscripts, which will be described presently, it is eclectic, following now the Northern tradition, now the Southern. It seems to be an attempt to combine the two recensions by superposition, like the Kumbhakonam edition. Its composite character may be seen from 24*, 25*, 27*, 114*, 138*, 149*, 170*, etc., etc. It contains the additional passages of the Southern recension, as well as the Gaṇeśa episode, which latter is found only in late Northern

(Devanāgarī) manuscripts: exactly like the Kumbhakonam edition. The collation of this version was discontinued after the second adhyāya. The version may be safely ignored as useless for critical purposes.

The Devanāgarī Version of Devabodha.

A commentary older and more important than the Arthadīpikā of Arjunamiśra, and one more neglected still, is the Jñānadīpikā of Devabodha, cited here as Cd. Devabodha is certainly earlier than Vimalabodha, Arjunamiśra and Nīlakaṇṭha, all of whom cite him with great respect, and probably earlier than Sarvajña-Nārāyaṇa and Vādirāja. He is, therefore, most likely, the earliest commentator of the Mahābhārata hitherto known, and, in my opinion, also the best. The commentary is in any case most valuable, and its evidence, both positive and negative, of supreme importance for the constitution of the text.

The Jñānadīpikā is a concise ṭīkā; that is, a running commentary, explaining, as a rule, only the difficult words and passages in the text. Occasionally it offers explanations of constructional obscurities and grammatical difficulties, and gives the gist of passages; in the latter case, usually, under citation of entire verses (i. e. half ślokas) from the text. The extent of the commentary on the Ādi is given in one manuscript as 1400 granthas. The homage which Arjuna pays to Devabodha in the Introduction to his scholium is not a mere matter of form. Arjuna has in fact based his commentary largely on that of his predecessor. He has copied very large portions of Devabodha's commentary, sometimes verbatim, sometimes in extract. Moreover even when the two commentaries differ, the influence of Devabodha is plainly discernible. In fact, the Arthadīpikā may be considered as a revised and enlarged edition of the Jñānadīpikā. The similarity of the names is suggestive and worthy of note.

Unlike the commentaries of Arjunamiśra, Nīlakaṇṭha and Ratnagarbha, that of Devabodha is unaccompanied by the epic text. The question what was Devabodha's text cannot, therefore, be answered with any high degree of certainty. The entire Southern recension and even the Vulgate may, however, be definitely ruled out. There remain the Bengali, Śāradā and “K” versions. With the latter two, the pratīkas of Devabodha seem to show greater affinity than with the Bengali version. For instance, Devabodha has no comment on any of the six adhyāyas (including the Kaṇikanīti) of the central sub-recension (γ), which are missing in Śāradā and K. Worthy of special note is the absence of all reference to the Kaṇikanīti in Devabodha's commentary, since the passage has evoked lengthy comments from both Arjunamiśra and Nīlakaṇṭha. Still greater probative value has an addition which is peculiar to the Kaśmīrī version. This version adds at the very end of the Ādi a supplementary and superfluous adhyāya,—an addition which is only a variant of the well-known Purāṇic tale of Śvetaki's sacrifice, occurring earlier in the course of the same parvan. Curiously enough, the king who is called Śvetaki in the first version is here called Śvetaketu! That the version of Devabodha contained this additional adhyāya is revealed by the concluding remark of Devabodha's commentary on the Ādi: श्वेतकिर श्वेतकेतुरिति नाम. This remark will not apply to any version which has not the additional adhyāya peculiar to the Kaśmīrī version. These considerations tend to show that the version of Devabodha was of the Śāradā-K type. And the inference is confirmed by many minor agreements, which need not be cited here.

The Composite Devanāgarī Version.

The fourteen manuscripts (D₁-₁₄) comprising this version are misch-codices of small trustworthiness and of no special value for critical purposes. Consequently, half of them (D₈-₁₄) were discontinued already after adhyāya 2. The characteristics of these manuscripts may be briefly noticed here.

D₁ is akin to D₉ and looks uncommonly like a Nīlakaṇṭha manuscript minus the commentary. Yet it differs conspicuously from the ordinary Nīlakaṇṭha manuscripts by the unaccountable omission of the entire Brahmā-Gaṇeśa complex (that is, both the visit of Brahmā and the employment of Gaṇeśa as a scribe, which arises out of this visit) as well as the description of the battle in which the Pāṇḍavas capture Drupada and hand him over as gurudakṣiṇā to their preceptor, Ācārya Droṇa (App I, No. 78). The omission of these episodes points rather in the direction of Bengal, since Kāśmīr is excluded by the mass of other interpolations which D₁ contains, as also by the almost complete lack therein of readings peculiar to S₁ K. The manuscript may be a blend of Bengali and some composite Devanāgarī manuscript or manuscripts. — D₂ (like D₅) is akin to K₈-₆ and might have been with advantage classed with them; see, for instance, the critical apparatus pertaining to the list of the contents of the Araṇyaparvan in adhy. 2. — D₃ is palpably under Southern influence, to prove which it is sufficient to point out that it transposes the Śakuntalā and Yayāti episodes, a transposition which is quite peculiar to the Southern tradition. — D₄ contains notably large additions from Southern manuscripts, additions which are either entered on the margin or, when the marginal space would not suffice, written on supplementary folios. The Southern influence is illustrated by the following passages: 587*, 594*, 596*, 598*, 599*, 602*, 603*, 604*, 605*, 609*, 610*, 611*, 612*, 613*, 617*, 621*, 623*, 624*, 628*, 629*, 630*, 633*, 634*, 635*, 637*, 670*, 671*, 713*, 715*, 1255*, 1256*, 1257*, and scores of others. Cf. also the following passages given in App. I: 35, 46-48, 52, 53, 55, 56, 59, 64, 67-69, etc., etc. — D₅ (like D₂) often stands in opposition to other manuscripts of this composite class, agreeing with K₈-₆, with which it might have been with advantage classed. Like K₄-₆, it contains Southern additions as well, e. g. 1565*, 1579*, 1580*, etc., and passage No. 89 of App. I. — The manuscripts D₆-₇ were discontinued after adhy. 53. Frequently, they are found to be in opposition to the Vulgate and agreeing with the manuscripts of the e group. They also show 230*, which is a Southern passage.

D₈-₁₄, as already remarked, were collated only as specimens for the first two adhyāyas and discontinued thereafter. — Of these, D₈-₁₂, ₁₄ are palpably under Southern influence, as is evidenced by their containing one or the other of the following typical Southern insertions: 18*, 21*, 22*, 24*, 32*, 42*, 45*, 48*, 49*, 56*, 80*, 81*, 89*, 114*, 117*, 138*, 149*, 170*. — D₁₃, which is a fragmentary manuscript, beginning almost at the end of adhy. 1, is used in this edition practically only for adhy. 2, as it is discontinued at the end of that adhyāya. The text shows strong affinities with the version of Arjuna-miśra. — The text of D₁₄ is a complex. It contains some old readings such as are preserved only in the Kāśmīri manuscripts, but also an extraordinarily large number of individual readings, not found elsewhere (cf. 1. 1, 50, 63; 2. 101, etc.). At the same time, it is contaminated from some Southern source, perhaps the Malayālam version!

The Devanāgarī manuscripts of the Mahābhārata in the Tanjore Library seem to have been all copied during the régime of the Maratha Chiefs of Tanjore, and are a blend of the Northern and Southern recensions, and, as such, of little value for text-critical purposes.

The Telugu Version.

The Telugu version, situated as it is on the boundary line which divides the Northern from the Southern recension, was particularly open to contamination from the Northern tradition. We accordingly find that the majority of Telugu manuscripts are eclectic on no recognizable principles, presenting somewhat the aspect of a mosaic of the texts of the Northern and Southern recensions, not unlike the Kumbhakonam edition. T₁ is one of the extremely few Southern manuscripts which contain the (Northern) salutational stanza नारायण नमोस्तु etc. For the Northern element in the make-up of T₁, cf. 29*, 30*, 96*, 97*, 98*, 106*, etc., etc. As compared with T₁, T₂ shows a purer Southern tradition and has distinct leanings towards the Grantha version. — T₂ only replaces the fragmentary manuscript T₁, which breaks off at the end of adhy. 181.

Important variants of one other Telugu manuscript (Tanjore 11809) are now given by Professor P. P. S. Sastri in his edition of the Southern recension. It does not differ appreciably from our Telugu manuscripts.

The Grantha Version.

The Grantha version is the version of the Tamil country, and is written in the so-called Grantha script. It is one of the two important Southern versions, the other being the Malayāḷam. The Grantha version—to judge by the manuscripts utilized for the Critical Edition, and for Professor P. P. S. Sastri’s Southern Recension—is more heavily interpolated than the Malayāḷam, and is also more influenced, on the whole, by the Northern recension.

For the beginning of the Ādi, we get, temporarily, the sub-groups G_1-2 and G_3-6, but soon the configuration changes to G_{1, 2, 4, 5} versus G_{3, 6}. The latter group (G_{3, 6}) represents the purer Southern tradition, agreeing with M against the other Southern manuscripts, whereas the four MSS. G_{1, 2, 4, 5} are not merely heavily interpolated but stand palpably under Northern influence. All Grantha manuscripts are probably contaminated (directly or indirectly) from Northern sources in different degrees. G₃ shows, on the whole, little Northern influence, but 419*, 494*, 698*, 1310*, 1312*, 1885*, 1975*, and passage No. 73 of App. I, show that even G₃ is probably not entirely free from contamination, since all these (Northern) passages are missing in M.

Sastri’s edition of the Southern recension gives the (most important) variants of five Grantha manuscripts of which three, क, ख and ग (the latter being Sastri’s “principal text”) are identical with our G₄, G₅ and G₆ respectively. Extracts from a Grantha manuscript belonging to the Royal Asiatic Society of Great Britain and Ireland (Whish Collection, No. 65) have been given by Professor Winternitz¹ and compared with the text of the Bombay edition (Śaka 1799). The passages which differ from the Bombay edition

have been underlined in his extracts, and the corresponding passages of the latter are given opposite each line: a convenient arrangement which shows, at a glance, the relation of the two texts to each other for the passages excerpted.¹ The clearest proof² of the contamination of G_{1. 2. 4. 5} from Northern source is furnished by 294*, a Northern passage, added in this sub-group irrelevantly before 1. 20. 1. The two lines comprising this passage must have been interpolated in a remote ancestor of G_{1. 2. 4. 5} by a clumsy scribe, who had missed the right place by four stanzas, and have remained there ever since, fortunately. Another rather transparent interpolation in G_{1. 2. 4. 5} from a late Northern source is a passage referred to already, No. 14 of App. I, which describes the circumstances under which Aruna becomes the charioteer of the Sun, an irrelevant digression. Cf. also 1373*, 1375*, 1377*, and passage No. 76 of App. I. The sub-group contains an amazingly large number of interpolations, which have not been found, so far, elsewhere, and of which a few may be mentioned as illustrations: 320*, 322*, 326*, 330*, 337*, 345*, 351* (third line!), 357*, 363*, 364*, 368*, 371*, 373*, 382*, 386*, 387*, 388*, 406*, 519*, 584*, 636*, 705*, 741*, 755*, etc., etc. But the Grantha version itself is inclined to admit freely new lines. Instances of rather lengthy interpolations of G are furnished by passages No. 35–39, 73 and 93 of App. I. Most of the interpolations are however short, consisting, as a rule, of less than 10 lines, e. g. 500*, 501*, 502*, 504*, 507*, 509*, 510*, 511*, 520*, 552*, 569*, 570*, 693*, 814*, 841*, 897*, 1259*, 1268*, 1312*, 1313*, 1316*, 1319*, 1320*, 1372*, 1435*, 1441*, 1447*, 1448*, 1452*, 1453*, 1476*, 1489*, 1531*, 1541*, 1542*, 1543*, 1544*, 1545*, 1547*, 1550*, 1551*, 1596*, 1597*, 1604*, 1631*, 1658*, 1666*, 1707*, 1868*, 2009*, 2040*, etc., etc. G₇, which is one of the few Southern manuscripts containing the (Northern) mantra नारायणं नमस्कृत्य etc. is, like T₁, a typical blend of the Northern and Southern tradition, and was, on that account, discontinued after adhy. 2. Its composite character may be seen from: 29*, 30*, 96*, 97*, 98*, 106*, 145*, etc., etc.

The Malayālam Version.

This is the version of Malabar, the Southernmost extremity of India. It is, in my opinion, the best Southern version. It is not only largely free from the interpolations of σ (=T G), but appears to be also less influenced by N than σ, wherein lies its importance for us. Instances of additional passages found in G (with or without T), but missing in M, are: 443*, 500*, 501*, 502*, 504*, 507*, 503*, 510*, 511*, 520*, 552*, 569*, 570*, 691*, 693*, 814*, 839*, 841*, 897*, 1259*, 1268*, 1310*, 1312*, 1313*, 1316*, 1319*, 1320*, 1447*, 1448*, 1452*, 1453*, 1476*, 1489*, 1531*, 1541*, 1542*, 1543*, 1544*, 1545*, 1547*, 1550*, 1551*, 1563*, 1566*, 1596*, 1597*, 1604*, 1658*, 1666*, 1751*, 1868*, 1872*, 1893*, 1896*, 1935*, 2006*, 2007*, 2009*, 2021*, 2024*, 2032*, 2040*, 2052*, 2053*, 2062*, 2071*, 2106*, etc.; and the following passages of App. I: 35–39, and 73.

M₁ often stands in antagonism to M₂₋₄, sometimes agreeing with manuscripts of the Northern recension; and is, therefore, an untrustworthy guide. M₁₋₄ are incomplete manuscripts, ending with adhy. 53; in other words, with the Āstikaparvan. M₅₋₈ replace these manuscripts in the Sambhavaparvan, which is the name under which the remaining portion of the Ādi is known in the Southern recension. This practice of writing the two portions of the Ādi in separate volumes is worthy of note, as an archaic survival. It is, in my opinion, the reflex of some half-forgotten factor connected with the compilation of the Ādi, and seems to me to be text-critically highly important. It should seem that the South has never completely assimilated the (Northern) division of the epic into the conventional eighteen parvans.

Instances of additional passages which distinguish M from all other versions are: 407*, 453*, 800*, 801*, 842*, 970*, 1051*, 1052*, 1278*, 1437*, 1438*, 1613*, 1678*, 1709*, 1871*, etc.

M₅₋₈ constitute really one manuscript, as is proved, for instance, by their repeating the following indubitable clerical errors: (i) in 1. 85. 25, M₅₋₈ repeat inconsequentially the words पूजयन्तीह लोके नाप्ताधयः; (ii) in 1. 154. 13, they omit 13ᵃ and 13ᵇ, transposing 13ᶜ and 13ᵈ, which they read as one line; (iii) in 1. 193. 1, they all read the meaningless क्षत्तारे विवृते दृ (Text स्वत्तारे विवृतं प्रति); (iv) they read 1. 213. 4ᵃ⁻⁵ᵇ erroneously after stanza 31 of adhy. 212; (v) in 1. 213. 6, M₅₋₈ omit the words च यशस्किनः of the text, for which M₇ shows a lacuna. Instances of readings peculiar to M₅₋₈ are (reference to adhyāya and śloka): 58. 6 M₅₋₈ समाजय्यः : rest समापेयुः 106. 2 M₅₋₈ समतोषयत् : rest समतर्पयत् 157. 9 M₅₋₈ शंकरस्तुष्टः : rest भगवोत्सृष्टः

Conflation in M₅₋₈ is suggested by 1. 209. 19, where M₅₋₈ have both the Northern reading and the Southern reading.

It may be added that the cases cited are merely by way of illustrations. A careful study of the critical apparatus would easily furnish scores of other instances.

This version has several striking agreements with S₁, a fact all the more impressive, because M, a Southern version, hails from the province at the opposite end of India from version; for instance, Malayālam supports S₁ (against T G) in omitting the spurious parts of adhy. 128–129 of the Bombay edition.

Winternitz has published, in Devanāgarī transcript, portions of a fragmentary Malayālam manuscript belonging to the Royal Asiatic Society of Great Britain and Ireland (Whish Collection, No. 158), which contains twelve chapters of the Sambhava-parvan.³ The extracts contain the beginning of the Pūruvāmśānukīrtana (our adhy. 90), the passage referring to Śakuntalā and the birth of Bharata (our 1. 90. 27–34), and the end of the adhyāya (our 1. 90. 93–96). The manuscript correctly shows the Southern transposition of the Śakuntalā and Yayāti episodes. It is interesting to observe that this manuscript also further shows the anticipation of 1. 89. 1–16, before the Yayāti episode, which is found in our Malayālam manuscripts (cf. note on p. 282) and in the conflated MSS. G₄.₅ (cf. note on p. 992), and which is text-critically highly important.

Readings or features which are peculiar to M or such as distinguish M from G (with or without T) will be found under: 1. 1, 35, 45, 122, 128, 168, 176, 179, 184, 189; 2. 160; 4. 4; 7. 10; 24. 1; 36. 3; 39. 2, 16; 53. 31; 54. 6; 57. 81; 61. 98; 67. 28; 68. 16, 51; 69. 9; 73. 33; 77. 9; 78. 23; 80. 2; 84. 14; 86. 1; 92. 45; 93. 14; 94. 9, 27, 32; 95. 8; 96. 2, 57; 98. 5, 12; 113. 22; 117. 5, 23; 119. 30; 123. 39; 129. 9–11 (om. in M); 132. 1; 136. 1; 138. 10; 139. 11 (om. in M); 142. 19; 150. 10, 26; etc., etc.

With regard to the versions described above, it must be frankly admitted that they do not, by any means, form water-tight compartments. The isoletional boundaries, as is natural, do not coincide, but are independent of each other; in other words, the textual peculiarities, which are, in final analysis, the real basis of our classification, never have, as a matter of fact, an identical area of distribution. The manuscripts cannot always be squeezed into the same moulds consistently. Thus, for instance, in the beginning of the Ādi, the Grantha version, as already remarked, shows two sub-groups G_1–3 and G_4–5; but soon the configuration changes and, from about adhy. 25 onwards, we get the grouping G_{1. 2. 4. 5.}: G_{3. 6.}. Not only that. Individual manuscripts, groups, or even versions often overstep the boundaries of their particular recension. Thus, for example, on the one hand, G_{1. 2. 4. 5.} frequently agree with N V₁ B D; M agrees with Ś₁; Ś₁ and Dn agree with S: against other manuscripts of their respective recensions.

These discrepancies, as is shown in the sequel, are due chiefly to two different causes: firstly, initial fluidity of the text; and, secondly, subsequent contamination or conflation. As regards fluidity: to conceive of the Epic of the Bhāratas—or for that matter, of any true epic—as a rigid or fixed composition like the dramas or poems of Goethe or Milton, or even of Kālidāsa or Bhavabhūti, would be manifestly grotesque. Such a view can originate only in a fundamental misconception of the origin, growth and function of epic poetry.

In the case of the Mahābhārata, we find, however, the fact of the fluidity of the original reflected in the tradition as preserved even to this day. Only a very late interpola- tion in some inferior Devanāgarī manuscripts speaks of the text as having been written down by Gaṇeśa to the dictation of Vyāsa, a fantastic story that we may ignore with an easy conscience. On the other hand, we are plainly told that the epic was first published, at an elaborate sacrificial session, in the form of a free recitation by Vaiśampāyana, a direct pupil of the author, before king Janamejaya and the assembled guests. It was again recited by Sūta (or Sauti), who had heard it only at the first recitation, and somehow committed the whole poem to memory. After just one single hearing, he obviously could not reproduce such a voluminous text verbatim et literatim. In the beginning, therefore, it is clear that the poem, which was committed to memory, was recited freely, as faithfully as the particular reciter could contrive. This mode of transmission is not calculated to preserve rigid textual purity in any high degree, without stringent precautions, such as were adopted in the case of Vedic texts, but which never existed, as far as one knows, in the case of the epics. This fact also we find unexpectedly preserved by tradition (1. 57. 74 f.). Vyāsa, we are told, taught his Bhārata to his five pupils: Sumantu, Jaimini, Paila, Śuka, and Vaiśampāyana. And the five rhapsodists—the direct pupils of the author—it is reported, published five separate versions of the epic:

संहितास्तैः पृथग्वेन भारतस्य प्रकाशिताः । As is well known, there is preserved a work which actually passes for the Ásvame- dhaparva of the Bhārata of Jaimini (whether it is actually so or not) and which is totally different from our Ásvamedhaparvan. Here, I think, we have a clear glimpse of the early history of the text. Two facts emerge rather clearly out of the chaos: firstly, the text was originally committed to memory and recited freely; secondly, different rhapsodists recited differently. This has indeed been assumed by many writers on the subject.¹ All that is quite natural and intelligible. As a matter of fact, from generation to generation, from place to place, from bard to bard, the wording, even the contents, would vary a little, until the text is committed to writing, which is the beginning of a different phase in its history. The view that the epic has reached its present form by a gradual process of addition and alteration receives strong support from the fact that this process is not stopped even by scriptal fixation.² The study of the manuscripts themselves, which belong to a very late phase in the evolution of the text, shows that texts must have been constantly amplified and altered by conflation. Such derangements, it may be observed, do not totally destroy, as might be imagined, the value of our division of the manuscript material into recensions and versions, but merely complicate its use and interpretation.

CRITICAL PRINCIPLES FOLLOWED IN THE CONSTITUTION OF THE TEXT

As already remarked, the Mahābhārata versions when they first come within our ken appear already dispersed in several distinct groups. The original, from which all these versions are derived, is itself preserved in no authentic copy contemporaneous with, or even reasonably close to, its period of composition. We can only reconstruct the original, approximately, by comparative methods. We recognize today, as already explained, two recensions, descended from the original, each recension embracing a plurality of versions, each version being divided into a multiplicity of sub-groups. The ultimate problem is to unify, as far as possible, this manuscript tradition: to evolve, by comparative methods, a form of the text that will explain this phenomenal wealth of divergent and conflicting texts, and justify it. Before I elucidate the critical principles followed in preparing the constituted text of the Ádi, I must review briefly other principles of textual criticism and textual reconstruction, and discuss the applicability of these principles to the Mahābhārata Problem.

The Classical Model

The method that naturally presents itself first to our mind is the time-honoured method of Classical Philology.³ The older school of classical philologists distinguished four stages in the work of preparing a critical edition of a classical text: (1) Heuristics,

i. e. assembling and arranging the entire material consisting of manuscripts and testimonia in the form of a genealogical tree; (2) Recensio, i. e. restoration of the text of the archetype; (3) Emendatio, i. e. restoration of the text of the author; and, finally, (4) Higher Criticism, i. e. separation of the sources utilized by the author. Excellent as this method is for the purpose for which it is devised, it should not be forgotten that it depends ultimately upon their being a more or less complete concatenation of copies and exemplars reaching finally back to a single authentic (written) archetype; and, consequently, can be applied to the Mahābhārata with great limitations.¹ Indeed our ideal is the same as that of the classical philologist: restoration of the text, as far as possible, to its original form. But the original of a Sanskrit poem and that of a classical poem: how entirely different they are! Particularly, in the case of the Mahābhārata, where, one may well ask, is the original of a whole literature? In the Mahābhārata we have a text with about a dozen, more or less independent, versions, whose extreme types differ, in extent, by about 13,000 stanzas or 26,000 lines; a work which, for centuries, must have been growing not only upwards and downwards, but also laterally, like the Nyagrodha tree, growing on all sides; a codex which has been written in nearly a dozen different scripts assiduously but negligently copied, chiefly as a source of religious merit, through long vistas of centuries by a legion of devout and perhaps mostly uneducated and inefficient copyists, hailing from different corners of a vast sub-continent, and speaking different tongues; a traditional book of inspiration, which in various shapes and sizes has been the cherished heritage of one people continuously for some millennia and which to the present day is interwoven with the thoughts and beliefs and moral ideas of a nation numbering over 300 million souls! The classical philologist has clearly no experience in dealing with a text of this description, an opus of such gigantic dimensions and complex character, with such a long and intricate history behind it.

The Difficulties Of Mahābhārata Textual Criticism

The capital difficulty of the Mahābhārata problem is just this that there are hardly any clear objective criteria which may enable us to discriminate with precision and certainty between the data of the rival recensions, to evaluate correctly and confidently the amazingly large mass of variants. Only an inconsiderable fraction of these variants represents clear “mistakes”, which can be corrected with confidence. As a rule, the variant readings, if they are not mere synonyms, convey a slightly different meaning, but almost always a possible meaning. From the grammatical point of view also, they are both equally valid. One of the variants may be a trifle more suitable than the other; for instance, in the discrimination between the Simple and the Periphrastic Future, or the Parasmaipada and the Ātmanepada. But can we legitimately premise that the original must necessarily have been quite flawless from the point of view of the Pāṇinian grammar? Is it not at least likely that the supposed solecism may be a genuine lapsus calami of the author, or (should that supposition be considered inadmissible or unacceptable) that the usage fluctuated? Then again, as we have seen, there are numerous passages, short and long, that are found in one recension and are lacking in the other, what I call “additional” passages. No

convincing proof can in general be given to establish either the originality or the spuriousness of any given passage of this type. What may fairly be regarded as inter- polations are in general so ingeniously fashioned and so cunningly fitted in that, except under very favourable circumstances, the intrinsic (contextual) evidence is inconclusive. For these and other reasons it is not always easy to correlate the divergent recensions, to discriminate between the variants, and to constitute a wholly unobjectionable single text.

This difficulty has its origin in the circumstance that in the Mahābhārata manuscript tradition, perhaps as much as in any literary tradition, the textual critic is faced with a bewildering profusion of versions as also with an amazing mixture of versions. Contrary tendencies have been at work in the evolution of the text. While, on the one hand, some elements have been working, from the earliest times, for the development of different types; on the other hand, there were not wanting elements that operated against the evolution of sharply differentiated types. To understand the phenomenon of this luxuriant growth and indiscriminate fusion of versions, one must appreciate certain details of historical moment, certain special factors in the transmission of the Mahābhārata, traits which distinguish our work from every other known text except the Rāmāyaṇa and perhaps other similar ancient epopees.

Let us examine closely the character of the differences between the two recensions to start with.² The differences are of three kinds. Broadly speaking, each recension differs from the other, firstly, in point of readings of the common stanzas; secondly, in point of additions (or omissions) of short and long passages; and, thirdly, in point of sequence of the text-units. How do these differences at all arise? Our first thought would be to attempt to explain the additions or omissions as the result of conscious editorial revision, or of clerical error, or partly of one and partly of the other. But the frequent differences in sequence, especially when no material gain is perceptible in either arrangement, rather support the explanation suggested above that both recensions are, in final analysis, independent copies of an orally transmitted text. The suggestion is confirmed by the consideration of the variation of the first type, namely, minor differences in the readings of the stanzas common to the two recensions, which confront us step by step throughout the parvan, nay, throughout the epic, as the partial collations of the other parvans now available at the Institute clearly show. It will be found for one thing perfectly useless to try to derive mechanically one set of readings uniformly from the other. Hundreds and thousands of the minor readings are nothing more than mere synonyms or paraphrases, grammatically and semantically equivalent, but graphically totally unrelated. They, therefore, cannot be all corruptions, in the ordinary sense of the word, of a written archetype. The vast majority of these variants cannot again be due to the zeal of a purist trying to correct the solecisms of the received text, or to the whim of a minor poet endeavouring to polish its diction or style. Had that been the case, we should find that the enthusiasm of the reformer had evaporated long before he had reached the middle or at least the end of the first parvan. The

herculean task of cleansing the Augean stables would be child's play compared to a systematic purification of the Mahābhārata text, according to later standards. Under these circumstances, however great might be the divergence between the two recensions in the beginning, it is bound to vanish or at least diminish towards the middle or the end of the poem. We find, on the other hand, as already remarked, that the stream of variation flows with unabated volume from the beginning to the end of the epic. This fact can in no way be reconciled with the hypothesis of a single uniform revision (or a series of them either) of a fixed and rigid text.

All the difficulties in the explanation of this phenomenal variation vanish, however, as soon as we assume that the epic was handed down from bard to bard originally by word of mouth, as is clearly implied by tradition. That would explain, without any strain or violence, the existence of the mass of variants, of differences in sequence, and of additions or omissions. If the text has been preserved, for any considerable period of time, only in memory and handed down by word of mouth, those are just the changes that could not possibly be avoided. It is evident that no great care would be lavished on the text by these custodians of the tradition to guard it against corruption and elaboration, or against arbitrary emendation and normalization : to reproduce the received text, which was not guarded by canonical authority or religious sanction, with any degree of precision would be neither attempted by the bards nor required of them. Whenever and wherever the text was then written down—and it was probably written down independently in different epochs and under different circumstances—these transmissions by word of mouth must have contaminated the written text and introduced innumerable variations in it. The assumption of some such complicated derangement, beyond the normal vicissitudes of transmission, is necessary to account for the abnormal discrepancies and strange vagaries of the Mahābhārata manuscript tradition. In other words, we are compelled to assume that even in its early phases the Mahābhārata textual tradition must have been not uniform and simple, but multiple and polygenous.

Moreover, a study of the critical apparatus shows that there has intervened a long period in the history of the Mahābhārata in which there was a free comparison of manuscripts aud extensive mutual borrowings. A natural and inevitable source of confusion of the tradition has always been the marginalia, comprising glosses, variae lectiones and additions. The copyist of a manuscript with such accretions copied and sometimes the marginal. It may be incidentally remarked that an examination of the marginalia shows that the variant readings are taken mostly from manuscripts belonging to the same version, or at least the same recension. But there is no reason, theoretical at any rate, why readings of the rival recension could not creep into a manuscript of the text by the medium, say, of a popular commentary such as Nīlakaṇṭha's. And, as a matter of fact, we do find, occasionally, readings of the opposite recension noted in the margins of manuscripts. Under these circumstances it was inevitable that the true reading, especially if it was a lectio difficilior or an archaism or a solecism, would be partly suppressed, being preserved to us in one or two manuscripts only.

Furthermore, that texts may be improved by a comparison of manuscripts is not by any means a modern discovery. The process has been known and practised for ages:

The difference is merely in our ideas of what is meant by "improvement" of the text. I have cited above the instance of Nīlakaṇṭha, who himself says that he had collected and compared Mahābhārata manuscripts from different parts of India in order to ascertain the "best" readings. The other commentators also, Devabodha, Arjunamiśra and Ratna-garbha, cite pāṭhāntaras and speak of apapāṭhas ("bad readings"). These they could have got only from a comparison of different manuscripts.

The texts favoured by the ancients appear to have been of the inclusive, rather than of the exclusive, type. This is proved in the case of Nīlakaṇṭha by a remark of his cited above, where, he naively admits that he had put together the stanzas which had been commented on by the ancient Bhāṣyakāras, and others he had found in modern manuscripts, with the idea of making a "thesaurus of excellences." The remark does not apply to any means exclusively to the Sanatsujāta episode, to which it is appended, at any rate as far as Nīlakaṇṭha is concerned. In the Ādi, we have abundant evidence that he has borrowed, according to his fancy, passages, short and long, from the Southern recension. The critical notes will show that his text includes a large number of Southern passages which are not found in any other Northern version, such as, for example, the catalogue of forest trees, which serves in a modest way for a description of the sylvan scenery amidst which Uparicara Vasu finds himself:

513* अरोकैम्पकश्रेणितिलकैरेतिरतिमुकैः । पुंनागैः कर्णिकारैश्च बकुलैर्दिव्यपादपैः । पनसनारिकेलैश्च चन्दनैश्चार्जुनेतथा । एतैरन्यैर्महावृक्षैः पुण्यैः स्वादुफलैर्युतम् ।

At one place, as was shown above, Nīlakaṇṭha has disfigured his text in his frantic attempt to squeeze into it a lengthy (Southern) passage containing some details which did not fit into his own text. This he has done, be it noted, at the risk of making his text wholly unintelligible, without a word of apology or explanation. Professor Winternitz, while criticizing Dahlmann's Das Mahābhārata, has pointed out this incongruity³: “The story . . . which relates how Dīrghatamas is insulted by his wife Pradveṣī, and how he consequently establishes the fixed rule (maryādā) that henceforth a woman shall always have to adhere to one husband, whether he be alive or dead, and that a woman who goes to another man shall go to hell, thus forbidding any kind of remarriage of widows . . . is strangely out of place⁴ in a chapter treating of Niyoga.” As was pointed out above, in consequence of the intrusion of this foreign matter, the first half of the stanza of the original text is separated from the second half by 27 lines. That in itself is, however, not a very serious matter in Mahābhārata textual tradition, where such transpositions are a common occurrence. But in the present instance, this transfer has had the unexpected and undesirable result that the subject of the sentence, which was left behind in the first half of the stanza, remains to the end without its predicate, which latter, being shunted off to such a remote distance, was furnished with a new and entirely different subject! The effect of this arrangement on the original story may be easily imagined.

Conflation is in general not so easy to detect and prove as in the case of Nīlakaṇṭha. We can date Nīlakaṇṭha with fair accuracy. Again Nīlakaṇṭha, who is one of the latest of our commentators, has himself vouchsafed some information as to how he has prepared his text. We have no such reliable data in the case of the majority of the manuscripts or versions of our critical apparatus.

Take, for instance, the case of the sub-group G₁.₂.₄.₆ of the Grantha version. In opposition to other manuscripts belonging to the same recension and even the same version, G₁.₂.₄.₆ contain, as shown above, an astonishingly large number of passages which are found otherwise only in some inferior manuscripts of the Northern recension. Now is this a case of contamination of the four MSS. G₁.₂.₄.₆ from a Northern source; or are the common passages a remnant of the lost archetype, which were somehow lost in the remaining manuscripts of the Southern recension?⁵ There is apparent agreement here between independent versions. But is this agreement original? The clumsy interpolator of a remote ancestor of G₁.₂.₄.₆ happens to have supplied us with the means of answering these questions. He has left behind, quite unintentionally, an impress of his “finger-prints,” so to say, by which we can easily and confidently trace him and examine his handiwork. The said manuscripts contain a Northern stanza (belonging to manuscripts of class γ) — a mere string of attributes of Garuḍa—wedged in at a place where it can be construed neither with what precedes nor with what follows. This proves incontrovertibly that these four manuscripts G₁.₂.₄.₆ have been compared with some Northern manuscripts, and makes it highly probable that the other doubtful stanzas, which they have in common with the Northern recension, have crept into their text in the same surreptitious way. At least this is the most plausible explanation of the anomaly. But even such confirmatory evidence is not always available.

The reader need not be sceptical about the possibilities of such indiscriminate conflation and addition. The critical apparatus, if closely scrutinized and properly understood, will reveal numerous instances of a similar character. Even a close study of the Kumbhakonam edition, prepared in our own times by two excellent Southern Pandits, will throw some light on the mentality of the old redactors of the Mahābhārata : parallel and even contradictory versions are placed quite unconcernedly side by side, regardless of the effect on the reader, regardless of the fact that sentences are left hanging in the air, that passages do not construe. Here one notices above all the anxiety that nothing that was by any chance found in the Mahābhārata manuscript should be lost. Everything was carefully preserved, assembled in a picturesque disarray.

Another important fact that must be kept in view in dealing with these interpolations is this. The older the borrowing and the more interesting the passage borrowed, the wider will be the area over which it will spread in its new habitat. It then becomes difficult to prove the borrowing.

Thus there is a certain group of passages which are found in all versions except in Ś₁ and K (that is, in the group ν), for example, the Kaṇikanīti.² In the particular case of the Kaṇikanīti, there appears to be sufficient extrinsic and intrinsic evidence to make it

highly probable that the passage is spurious, and the corresponding agreement between some of the (more or less) independent versions is unoriginal. There are indeed yet more difficult cases, where the evidence pro et contra of docu- mentary and intrinsic probability is equally balanced, as far as we can at present judge. In such cases we are forced to look for small things which look suspicious and lead us to probabilities, not facts. The problem is clearly not solved by formulating a priori a hypothesis as to the interrelationship of the different versions and fix the text in terms of some preconceived formula; for instance, by assuming as absolutely independent a certain number of these divergent versions, and laying down an arithmetical rule that whatever is common to two or more of such and such versions must be original. In this method, we can easily deceive ourselves and others; for the results arrived at will appear sounder than in reality they are. Even though the formal operations may be a piece of flawless logic, nevertheless the results, being based on premises possibly unsound though apparently clear and definite, may be wholly fictitious. The study of the manuscripts themselves must first teach us what their interrelationship is. And they unmistakably indicate that their interrelation- ship is of most complex character. The critical apparatus is a veritable labyrinth of complicated and intermingled versions, each with a long and intricate history of its own behind it. We have unfortunately no single thread to guide us out of the maze, but rather a collection of strands intertwined and entangled and leading along divergent paths. With the epic text as preserved in the extant Mahābhārata manuscripts, we stand, I am fully persuaded, at the wrong end of a long chain of successive syntheses of divergent texts, carried out—providentially—in a haphazard fashion, through centuries of diaskeuastic activities; and that with the possible exception of the Śāradā (Kaśmīrī) version, which appears to have been protected by its largely unintelligible script and by the difficulties of access to the province, all versions are indiscriminately conflated. Now it goes without saying that the genetic method (operating with an archetype and a stemma codicum) cannot strictly be applied to fluid texts and conflated manuscripts; for, in their case, it is extremely difficult, if not utterly impossible, to disentangle completely, by means of purely objective criteria, their intricate mutual relationships. The documentary evidence is no doubt supremely important, but the results, arrived at from a consideration of the documentary probability, must be further tested in the light of intrinsic probability. No part of the text can be considered really exempt from the latter scrutiny, when we are dealing with a carelessly guarded text such as we have in the present instance. A careful study of the critical notes will show—if, indeed, the foregoing remarks have not made it abundantly clear—that all the problems which present themselves for solution in editing any text from manuscripts are present in the case of the Mahābhārata on a colossal scale and in an intensified form. We must, therefore, clearly recognize that a wholly certain and satisfactory restoration of the text to its pristine form—even the so-called śataśāhasrī saṁhitā form—may be a task now beyond the powers of criticism.

Critical Editions Of The Different Versions

No doubt, in view of some of these difficulties, one scholar has suggested that to expedite and facilitate the work, we should, as a first step, before any attempt is made

to constitute the final text of the Mahābhārata, critically edit all the different versions.¹ That, it must be said, is a rather tall order, as any one will admit, who has any practical experience of editing the Mahābhārata in any shape or form, critical or otherwise. But perhaps funds and workers—not to speak of patience—can be found to edit a dozen or more lakhs of stanzas comprising the dozen or more versions of the Great Epic. There remains, however, yet another and a more fundamental difficulty, which appears to have wholly escaped the attention of the learned critic. The difficulty is that it is practically impossible to edit even a single version of the Mahābhārata—or for that matter of any other text—wholly satisfactorily, without considering the entire evidence, that is, without, at the same time, consulting the readings of all other versions. Suppose we examine six manuscripts of a version (Grantha) in order to prepare a critical text of that version. It may happen that four of them (G1.2.4.5), which are conflated manuscripts, have a “secondary” reading, while only two (G3.6) have the correct reading. In these circumstances, the true character of the variants could never be inferred from the readings of this version (G) itself; it would be shown only by other versions (T or M or N). In fact, there is no way of finding out whether any of the manuscripts of a particular version are conflated (if they happen to be conflated) without consulting the other versions. And, if for the editing of each of the individual versions, we have to scrutinize and weigh the entire evidence, we might as well get busy with the work of preparing the final text, assuming of course that a final (critical) text has to be prepared.

That consideration apart, even if we assume, for the sake of argument, that all the dozen or more versions lie before us in a critically edited shape, our main task is not made any easier on that account. One has to go through the same mental processes in picking out or reconstructing the correct readings, whether, as at present, the variae lectiones are concentrated on a single page of the critical edition or have to be searched in a dozen or more different provincial editions, arranged round about the critic in a semi-circle. Preparing all these different editions would not by itself give us the correct readings. Some of them, moreover, would but slightly differ form each other, for instance, the editions of the Bengali and the Devanāgarī versions; and it would mean useless duplication of labour. All that is really needed to facilitate our work is a critical edition of the Southern recension. An attempt to supply that need is now being made by Professor P. P. S. Sastri in his edition of the Mahābhārata, referred to already.

The Vulgate As Base

Another high authority, while full of apparent admiration for the way in which the work is being done at present at the Institute, has with much pathos and eloquence deprecated this hastily prepared, eclectic text. All that we need to do at present, according to this scholar, is to reprint the Vulgate, giving merely the variae lectiones of the manuscripts collated and leaving each individual reader to constitute his own text, unhampered and uninfluenced by the obtrusive personality of some editor who stands like a monitor between the reader and his author. The learned critic is evidently of opinion that any average reader, who picks up an edition of the Great Epic for casual study is better qualified to reconstruct the text than the editor who has made a special study of the

LXXXIV. PROLEGOMENA

problem! That is a paradox natural to the subtle mentality of the learned critic. But we need not take it too seriously. Whatever the Average Reader might or might not be able to do, I beg to submit that the Critical Reader, like the learned scholar whose opinion I am quoting, would not be any the worse off, if he is put in possession of this “Recension of Poona”.¹ For, who and what is to prevent him from constituting his own text from this critical edition? Whoever makes the text—even if Bṛhaspati himself were to come down and constitute the text—the Critical Reader would undoubtedly reject it as it would surely not fit in with his ideas of what is right and what is wrong. The Critical Reader has the same freedom of action whether he has before him the critical text or the Vulgate. The Vulgate, as far as I can judge, is no better suited for serving as the base than the present text.

It may, however, be that the hesitation of the learned authority is really due to a categorical objection to interfering in so definite a manner with the received text. Should that be the case, it is certainly difficult to appreciate the veneration of this scholar for the form of a text which was made up, probably, also in great haste but with inadequate and insufficient materials, only in the last quarter of the seventeenth century, that is, only about 250 years ago. It is surely illogical to assume that a text which has been built up largely on unscientific conjecture is now beyond the reach of conjecture.

A simpler and more probable explanation still of the hesitating attitude of the learned critic might perhaps be that his theoretical misgivings are based on a rather hasty study of both the Vulgate and the critical text. For, the text of the Vulgate is so corrupt and so obviously contaminated that it would be a criminal neglect of his duty for any intelligent editor now to reprint the Vulgate, when he has at hand the material to control its vagaries and to correct its absurdities.

One Selected Manuscript As Base

No doubt to remedy the inherent defects in the last method as also to avoid the dreaded saṃkara of pramāṇas, it has been suggested by other scholars that the best course would be to select one manuscript, the best manuscript extant (of any version presumably) and print it, with minimal change, correcting only the obvious and indispensable clerical errors and adding the variants of the collated manuscripts.² This expedient, though unquestionably simple and “safe”, and in most cases indubitably effective, fails totally in the present instance, for two reasons: firstly and chiefly, owing to the negligible age of our manuscripts, which are barely five hundred years old; and, secondly, owing to the systematic conflation which has been carried on through ages of revisional and amplificatory activity. By following any manuscript—even the oldest and the best—we shall be authen- ticating just that arbitrary mixture of versions which it is the express aim of this method to avoid!

This suggestion, however, has special interest, because the principle underlying it has now been, partly and timidly, put into practice by Professor P. P. S. Sastri, in preparing his edition of the Southern recension, whereas the three foregoing methods are mere castles in the air of theoretical critics.

A Critique Of Professor Sastri'S Method

Professor Sastri's edition is an excellent demonstration of the inadequacy of the underlying principle, which has been repeatedly advocated, showing up its defects as nothing else could. What Professor Sastri set out to do is (to quote his own words): "to print the text as it is in the original palm-leaf, liberty being taken only to correct scriptorial blunders,¹ to weigh the different readings in the additional manuscripts and choose the more important ones [scil. readings] for being added to the text by way of footnotes".² How difficult it is to carry this out verbatim in practice and at the same time to present a half-way readable text may be realized when we see how Sastri has had to doctor his text. A few examples may be added to elucidate the point. To begin with, Sastri does not follow the parvan division, nor the adhyāya division, of his basic manuscript, adding and omitting colophons arbitrarily, in order to reach some imaginary norm. Secondly, he adds an adhyāya of 40 lines after his adhy. 164, which is not found in his manuscript! Thirdly, he omits one whole adhyāya of 40 lines after his adhy. 180, where all Southern manuscripts, without exception (including his own exemplar) have it, and is moreover unaccountably silent about the omission! Fourthly, in one place (his adhy. 122) he has omitted fourteen lines of the text of his manuscript and added instead thirteen lines which are not found in any Southern manuscript!³ Fifthly and lastly, in yet another place (his adhy. 214) he has added an interpolation (upākhyāna) of 114 lines of which not a single line (as actually printed in Sastri's edition) is to be found in any of the six manuscripts utilized by him! These are some of the things that an extremely orthodox Southern Pandit actually does when he sets out with the avowed object of printing up a Southern manuscript as it is, correcting only "scriptorial blunders." I will not here speak of a certain number of spurious lines which appear to have crept insidiously into his text from the Vulgate and whose existence even he probably does not suspect.⁴ The changes mentioned first are of a different order: they have been made by Sastri consciously and intentionally.

Let me not be misunderstood. I do not blame Sastri in the least for taking such liberties with his manuscript, which is a tolerably good manuscript (though probably not very old), but has its faults like any other manuscript. I myself have had to proceed similarly, only more thoroughly, more systematically. Our methods are similar in practice, though not in theory; that is, in his theory. Sastri's text is eclectic (an epithet often used by critics with a tinge of reproach, the ground of which it is not easy to perceive): as eclectic as any other Mahābhārata text, printed or in manuscript, that I have seen. I have adduced the above instances chiefly to show what correcting merely "scriptorial blunders" in Mahābhārata textual criticism really ends in.

Thus it will be seen that the method of printing a Mahābhārata manuscript as it is, viewed as a rigid principle, is a deplorable failure. The lateness of our manuscript material

and the peculiar conditions of transmission of the epic are responsible for the defection. They force upon us an eclectic but cautious utilization of all manuscript classes. Since all categories of manuscripts have their strong points and weak points, each variant must be judged on its own merits.

What Is Then Possible?

The Mahābhārata problem is a problem sui generis. It is useless to think of reconstructing a fluid text in a literally original shape, on the basis of an archetype and a stemma codicum. What is then possible? Our objective can only be to reconstruct the oldest form of the text which it is possible to reach, on the basis of the manuscript material available¹. With that end in view, we must examine as many manuscripts—and above all as many classes of manuscripts—as possible, and group them into families. We must try to ascertain and evaluate the tradition of each family, eschewing late and worthless material. We may then consider the relation of these traditions in regard to the variae lectiones, and the genuine and spurious parts of the text. Beyond that, we have to content ourselves with selecting the readings apparently the earliest and choosing that form of the text which commends itself by its documentary probability and intrinsic merit, recording again most carefully the variants, and the additions and omissions. A little critical remaniement of the text need cause no alarm. For, as I have already observed, it is hardly logical to assume that a text which is largely based on conjecture is now beyond the reach of that principle. Of course there will always remain many doubts, but that consideration should not prevent us from correcting those parts which can be corrected with confidence; moreover, that limitation applies to our comparatively well preserved classical texts, despite the guarantee of the careful editings they have undergone. However, owing partly to the fluid character of the original and partly to the fragmentary and inadequate information we possess as regards the origin, growth and transmission of the text, it is incumbent on us to make Conservatism our watchword. We must abstain from effecting any change which is not in some measure supported by manuscript authority.²

The Method Of Reconstruction Explained

The method I have followed in reconstructing the text cannot, unfortunately, be presented in the shape of short general rules. I shall endeavour, however, to explain it as briefly as possible.

The main principle underlying all speculation as to authenticity is the postulated originality of agreement between what may be proved to be (more or less) independent

versions: The principle I have tried to follow religiously—and I hope I have never deviated from it—is to accept as original a reading or feature which is documented uniformly by all manuscripts alike (N=S).

For instance, we frequently come across three-lined stanzas, one of whose lines is an “inorganic line”, that is, a line which can be added or omitted without detriment to sense or grammar. These seemingly superfluous lines, if proved by both recensions, have not been deleted; they have been kept scrupulously intact. A more important instance is of the initial adhyāyas of this parvan. The connection between adhy. 1–3 and what follows, as also the connection between the three adhyāyas inter se, is of most loose character. There is further the suspicious circumstance that adhy. 4 begins precisely in the same way as adhy. 1; both adhyāyas have in fact the identical opening (prose) sentence: लोमहर्षण पुत्रश्रवाः सूतः पौराणिकौ नैमिषारण्ये शौनिकस्य कुलपतेर्होदशवार्षिके सत्रे । In other words, adhy. 4 begins as though nothing had gone before! The prose sentence seems to fit better the context of adhy. 4 than the context of adhy. 1; but that is not material to my argument. It would have been possible to athetize the first three adhyāyas in order to remove this anomaly, relegating them to the Appendix. But as all the four adhyāyas are handed down in exactly the same form (with the usual amount of variants) in all manuscripts of both recensions, they were left perfectly intact. Here we have an old conflation of two different beginnings. They were not harmonious in juxtaposition, but each was too good to lose, in the opinion of the ancient redactors. They therefore put both in, making but a poor compromise.

Another passage that may be thought to need some radical treatment is the account of the cremation of Pāṇḍu and Mādrī. We are first told that the king died in the forest, and Mādrī mounted the funeral pyre and was burned with him (1. 116. 31). After this we read that their “bodies” (śarīre) are brought to the capital of the Kurus (1. 117. 30), and an elaborate royal funeral funeral takes place. In the account given in the following adhyāya (118), from the description of the anointing and dressing of the king’s body, and from the remark that the king looked as if he were alive (1. 118. 20): आच्छन्नः स तु वालोर्मिजिर्वामिव नर्त्तमः । it is clear that that no former burning is imagined. After Pāṇḍu had been burned with his favourite queen Mādrī on the funeral pyre, there could not have been (as Hopkins³ has justly pointed out) much corpse left or not enough to dress and smear with sandal paste! But the manuscripts do not render us any help here. The passage is handed down in identical form in all manuscripts of both recensions.

The above examples will show that the diaskeuasts did not always employ any great art—I may add, fortunately—in conflating two discrepant accounts of an incident, which is by no means an easy task. To resolve such anomalies, however, is beyond the scope of this edition, since the entire manuscript evidence unanimously supports the conflation, which is too old and deep-rooted to be treated by the ordinary principles of textual criticism. If we went about, at this stage of our work, athetizing such passages as were self-contradictory or as contradicted the data of some other part of the epic, there would not be much left of the Mahābhārata to edit in the end.

I give in a footnote⁴ the text of a hundred selected stanzas for which no variants, or only unimportant variants, have been recorded in the critical notes; of these about

Adhy. 1 आर्षं पुरुषमीशानं पुरुषं पुरुहूतम् । ऋतमनेकाक्षरं ब्रह्म व्यक्ताव्यक्तं सनातनम् ॥ २० Adhy. 26 ददर्श तं पिता चापि दिव्यैर्दिव्यैर्विरंगमम् । तेजोवीर्यबलोपेतं मनोमारुतरंहसम् ॥ ६ तमागतमभिप्रक्ष्य भगवानकवचसदा । विदित्वा चास्य संकल्पमिदं वचनमब्रवीत् ॥ १० प्रज्ञादिर्थमर्थर्ममो गरूडस्य तपोधनाः । चिकित्सितं महत्स्रं तदनुष्ठानमुत्तमैः ॥ १३ पक्षिराजस्ततोमा श्रामकरत स शैलराट् । युमोक्ष पुष्पवर्षं च समालिंगितपादयः ॥ २२ Adhy. 27 कश्यपस्य दिजातेश्च वे पछिराव शुतः । भभ्रष्यः सर्वभूतानामवभ्रमबलकम् ॥ २ विषपोडर्गं पुराणस्य येनायं त्वं परिपृच्छसि । श्रृणु मे वदतः सर्वमंतर्सेक्ष्यतो दिज ॥ ४ यजतः पुत्रकामस्य कश्यपस्य प्रजापतेः । साहाय्यमृत्विजो देवा गंधर्वमुख्यैः किल ॥ ५ दत्तस्तेनं काले तु देवी दाक्षायणी शुभा । विनता नाम कल्याणी पुत्रकामा यशस्विनी ॥ २४ तपश्चास्या व्रतपरा जाता पुत्रवने शुचिः । उपचक्राम भतारं पादुवाचम कश्यपः ॥ २५ Adhy. 28 तमुवाचाप्यथो देवी वरदोसीति चेत्स्वगू । स वत्रे तब बिडस्यमन्यं त्वरमम ॥ १३ Adhy. 31 सुपर्णमागां शापस्य मात्रा चैव श्रुतैन च । विनतायास्तथा लोकं पातितमंदन ॥ २१ Adhy. 32 ततमवीत्सलहृति तप्यमानं पितामहः । किमिदं कुरषे शेष प्रज्ञानां त्वरवि कुरु ॥ ६ जानामि शेष सर्वेषां प्राणां ते विविक्षितम् । मातुश्चाप्यपराधादे भ्रातृणां ते महद्वयम् ॥ १३ Adhy. 35 ततः प्रवृति तां कन्यां वासुकिः पर्ययरक्षत । जरत्कार्वं खसारे वै परं धर्ममवाप च ॥ २ Adhy. 38 संदिष्य कुशलप्रश्रं कार्यमृत्विजस्तम च । शिष्यं गौरमुखं नाम शीलवंतं समाहितम् ॥ २४ Adhy. 39 दंष्ट्रया स नागेंद्रः काश्यपेन महात्मना ।

अदशद्रक्षमध्येल न्यग्रोधं पक्षिगोत्तमः ॥ ४ भसीभूतं ततो दृक्षं पक्षिगेन्द्रस्य तेजसा । भस्म सर्वे समाहृत्य काश्यपो वाक्यमब्रवीत् ॥ ७ विधर्वाच पक्षिगेन्द्र पश्य मेद्रिन्मन्नवतो । श्मृदं संजीविन्यानेने परयतत्वं सुर्जंगम ॥ ८ Adhy. 40 ततो नृपे तक्षकस्तेजा हते प्रयुज्य सर्वेः परलोकसक्रियाः । शुचिर्द्धिजो राजपुरोहितस्तदा तथैव ते तस्य नृपस्य मक्रिविः ॥ ५ Adhy. 41 वायुममो निराहाराः शुष्यन्नहरहर्रदुनिः । स ददर्श पितृनार्ते लभ्यमानोपभुखान् ॥ ३ Adhy. 42 पुत्र दिद्यासि संत्राप्त इमं देहं यदुच्छ्रया । किमध्ये च त्वया ब्रह्मन् कृतो दार्सग्रहः ॥ ३ एवं द्वा दृष्टा तु भवतः शकुन्तानिल लंभतः । मया निवत्सिता बुद्धिर्मद्राचर्यालिपतामर्हः ॥ ५ Adhy. 45 चातुर्वर्ये सधर्मस्र स कृत्वा पर्ययरक्षत । धर्मतो धर्मविद्राजा धर्मं विष्वात्मनिव ॥ ७ Adhy. 46 ततस्तत्सिद्ध दिवसे सप्तमे समुपस्थिते । राज्ञः समीपं प्रजग्निः काश्यपो गरुडश्चत ॥ २४ Adhy. 48 तक्षकस्तु स नागेंद्रः पुरंदर्विनिवशनम् । गतः झैलेन राजानं दीक्षितं जनमेजम् ॥ २४ अजर्क्षं निपतत्स्त्री नागेषु भृष्टु दुःखितः । अल्पशेषपरीवारो वासुकिः पर्यतंच्यत ॥ २९ Adhy. 49 तत आहुय पुत्रं स्वं जरत्कारुर्मुजंगमा । वासुकेनागराज्यस वचनादिदमब्रवीत् ॥ १ ततः स वासुकेर्वीच्यमपनीय मनोज्वरम् । आलाभं चालयद्वेष्टुं जगाम त्वरतो भुश्रम् ॥ २५ जनमेजयस्य तं यक्षं सर्वः ससुद्धि युगेः । मोक्षाय भुजंगनाशासको द्विजोत्तमः ॥ २६ स सात्वपचयवत्तीको यज्ञादजनमुद्गमम् । कृतं सद्सैर्वेबुधिः पूर्वदिक्षमप्रमेः ॥ २७ Adhy. 53 स यक्षः पाण्डवेयस्य राज्ञः पारितक्ष्यस्य ह । प्रीतिमांस्त्राभवदाजा भारती जनमेजय ॥ १० ऋत्विभ्यः सदसस्येभ्यो वै तत्रान्नान्समागतः ।

वैशम्पायन प्रदीर्घं शतशोऽथ सहस्रशः ॥ १२ पुनरागमनं कार्यमिति चेन्न वचोऽब्रवीत् । भविष्यसि सदसो मे वाजमेधे महाक्रतौ ॥ २५ Adhy. 54

नक्षत्रपौर्णसे सरलं धर्ममेवाप्नुयामः । पूजयामास राजेन्द्रः शालवृक्षेण कर्मणा ॥ १२ Adhy. 55

शृणु राजन्यथा मेदः कुरुपाण्डवयोस्त्वभूत् । राज्यार्थे शालवृक्षो वनवासमेव च ॥ ४ Adhy. 56

कथितं वै समासेन त्वया सर्वं द्विजोत्तम । महाभारतमाख्यानं कुरुणां चरितं महत् ॥ १ कथं धर्ममृतो श्रेष्ठाः कृतो धर्मस्य भवितुम् । अनघैः परमां केशं सोढ्वाप्यस्मि युधिष्ठिरः ॥ १ एतदाचक्ष्व मे सर्वं यथार्हवृत्तं तपोघन । यद्यच्च कृतवन्तस्ते तत्र तत्र महाराजाः ॥ १२ मद्वर्यैः सर्वलोकैस्तु पूजितस्य महात्मनः । प्रवक्ष्यामि मतं कुलं व्यासस्यामिततेजसः ॥ १२ Adhy. 58

त्रिसप्तकुलः पृथिवीं कृत्वा निःक्षत्रियां पुरा । जामदग्न्यमुपसृतपे महेंद्र पर्वतेतमे ॥ ४ कर्मणि च नरव्याघ्र धर्माविष्टानि मानवाः । धर्मज्ञेषुगुपसृप-तक्षकवर्मपरायणाः ॥ २१ इत्युक्त्व्या स मही देवा राजा राजनिमृत्य च । आदिदेश तदा सर्वान्बुद्धानुभ्यक्कुलवर्यम् ॥ ४५ अथ शकादयः सर्वे इब्वा शुद्धरौवेनाः । तस्यमव्यं च पञ्च तव च तस्य ते जगुस्सदा ॥ ४८ Adhy. 59

भक्ष्योपमानसाः पुत्रा विदितः प्रणयपर्णयः । मरिचिरत्रिसितो पुलस्त्यः पुलहः क्रतुः ॥ १० प्रहादः पूर्वज्ञेपोहा संहादसदन्ननरम् । अनुहादस्तुदीर्णशोभलाश्च शिविबाष्पोली ॥ १८ अमृतं भानुगाणा गो गणगंधर्वसत्तमम् । अपत्यं कपिलायास्तु पुराणे परिकीर्तितम् ॥ ५० Adhy. 60

कृतोः ऋतुसमानाः पुत्राः पतंगसङ्घधारिणः । विद्वताक्षेषु लोक्रेषु सत्तत्वपरायणाः ॥ ८ द्वादशादित्यादिः पुत्रः शक्रतुल्या नराधिप । तेषामयमतौ विष्णुपर्ववे लोकाः प्रतिष्ठिता ॥ ३५ Adhy. 61

तेषामयमतयो यस्तु चतुर्थः परिकीर्तितः । श्रेणिमानिति विख्यातः क्षिति राजर्षिसत्तमः ॥ ४९ इक्षुरुतेवैक्षैहौर्ते देवैपिर्वाधिद्धं भारत । अंशांशो समुत्पन्नं भारतद्वाजमयोजिनम् ॥ ६३ 12

शकुर्णानां यत्त्वासीद्राजा लोके महारथः । द्वापरं विद्धि तं राजानमुर्धमनर्दनम् ॥ ७२ अरिहायास्तु यः पुत्रो हंस इत्यभिबिश्रुतः । स गणगंधर्पतेजैवं कुरुवंशविषवैनः ॥ ७७ Adhy. 64

नरनारायणानं गङ्गेयोषमितम् । मतवहिंसयुत्तं प्रिवेश महदनम् ॥ २४ Adhy. 71

तानपुनर्जीवियामास काव्यो विद्यावभयात् । ततस्ते पुनरथाथ योपयांचक्रिरे सुराणू ॥ ७ Adhy. 73

कृतवधिंचे प्राप्ते हृदरामा दिवाकौसः । कथाचशील तां विद्या कृतार्थी भरतर्षम ॥ १ इद्धा दुहितरं काव्यो देवयानी ततो बले । बाहुभ्यां संपिरेपख्य दुखितो वाक्यमब्रवीद् ॥ २८ Adhy. 76

इयं च मे सखी दासी यत्राङ्गार्थे तव गामिनी । दुहिता दानवेन्द्रस्य शर्मिष्ठा वृषपर्वणः ॥ ९ Adhy. 81

तस्य वित्तीयोर्यशसः सलकीर्तयेहलमनाः । चरितं श्रीदुर्मिछामि दिवि चेह च सर्वशः ॥ ८ Adhy. 91

अथ गङ्गा सरिच्छ्रेष्ठा समुपायतामृगम् । तस्या वासः समुत्कृतं मारुतेन शशिश्भमू ॥ ४ Adhy. 93

स बारुणिरूपसेवे तस्मिनभरतसत्तम । बने पुण्यकृतज्ञं श्रेष्ठः सादुमूफलकोके ॥ ७ अस्याः शीरं पिबेन्मेध्यः सादु यो वै सुमध्यमे । दशा वर्षसहस्राणि स जीवैस्तीरयौनः ॥ १९ दत्तानुकुला तु सा देवी जुतोम सुमध्यमा । तत्पुत्रावानुवावाश्री भर्तारं दीप्ततेजसम् ॥ २० Adhy. 94

स कदाचिदिगं यातो यमुनाभितो नदीम् । महीपतिरनिर्दयमाजिघृतमुत्तमाम् ॥ ४९ ततः कदाचिच्छेल्लोतं शतगुं ध्यानमास्थितम् । पुत्रो देवतोऽभ्येल पितरं वाक्यमब्रवीद् ॥ ५४ Adhy. 96

सुहृदां यतमानमानासः सह चिकिर्वसेकः । जगामास्तमिवादित्यः कौरव्यो यमसादनम् ॥ ५८ Adhy. 102

वाहनानि प्रहनानि मुदिता श्रृषप्रदक्षिणः । गन्धवति च माल्यनि रसवन्ति फलानि च ॥ ३ Adhy. 117

तथा विवृद्धूसदृशानां महान्वैकरोतिवडवः । न कश्चिद्रकरोदीथमनवमर्दुमुदुवः ॥ १२ Adhy. 119

Adhy. 125 अवस्थां च सहिते प्राप्यौ धृतराष्ट्रजितम् । दुर्योधनमभिप्रवृत्तिर्धर्मराजमन्वहम् ॥ १२

Adhy. 127 तमालोक्य धृतराष्ट्रस्त्वा पितुर्गौरवमब्रवीत् । कर्णाभिर्धिक्शिराः शिरसा समवन्दत ॥ २ ततः पादावबन्धाच पदान्तेन संभ्रमम् । पुत्रेति परिपूर्णार्घ्यमब्रवीदग्रसारथिः ॥ ३ ततो दुर्योधनः कोपादुत्पपात महाबलः । प्रापुयषचनात्मनादीदृत् इव दिपः ॥ ९

Adhy. 132 पाण्डवा धृतराष्ट्रेण प्रेषिता वारणावतम् । उल्लसे विहरिष्यन्ति धृतराष्ट्रस्य शासनात् ॥ ६ स सर्वे रासभयुक्तेन स्वन्दनेनानुगामिना । वारणावतमैभ्रत्येय यथा यासि तथा कुरु ॥ ७

Adhy. 138 तप्तायुक्ता वयं दाशदिने वृषप्रपाश्रिताः । कां दिशं प्रतिपत्स्यामः प्राप्ताः केशमनुत्तमम् ॥ २९

Adhy. 139 गच्छ जानीहि के स्वेते शेरेते वनमाभ्रिताः । माधुषी बलवान्नान्नो प्राणं तर्पयतीव मे ॥ ८

Adhy. 152 स तदभ्युपपादाय गतो बकरनं प्रति । तेन नूनं भवेदेतत्लोकहितं कृतम् ॥ १७

Adhy. 158 शिरोरुहेषु जग्राह माल्यवच्च धनंजयः । आप्रुयन्ति चक्रार्थीय सोऽरुणादत्तमहाव्रतम् ॥ २० अक्लाभिन्ना विचित्रोऽथ दधो मे रथ उत्तमः । सोऽढै चित्ररथौ भूत्वा भ्रात्रा दुर्योधनभवम् ॥ २७

Adhy. 159 गर्क्च न बलमस्माकं भूय एवाभिवर्धते । यतस्ततो मां कौन्तेय सदार्द्र मनुरविशत् ॥ ११

Adhy. 162 सहस्रांशुं ततो विप्रः कृताञ्जलिरुपस्थितः । वसिष्ठोऽधमिति प्रीत्या स चात्मानं न्यवेदयत् ॥ १७

Adhy. 170 गर्भोनामिं यदा सूर्यं श्रृणुणां प्रत पुत्रकाः । तदात्ममूणां गर्भा भवा वर्षात्तं धृतः ॥ ३

Adhy. 192 अथ दुर्योधनो राजा विमना भ्रातृभिः सह । अश्वाश्वाक्षा मातुलेन कर्णेन च कृणेन च ॥ १

Adhy. 194 विक्रयं च प्रशंसेन्न क्षत्रियस्य विदो पते । रुश्को हि धर्मः शूद्राणां विक्रमः पार्थिवधर्म ॥ १८

ते बलेन वयं राजनमेहता चतुरङ्गिणा । भ्रमश्य दुपदं सीमानयानयेह पाण्डवान् ॥ २९

Adhy. 198 दित्या जीवन्ति ते पार्थ दिश्या जीवति सा पृथा । दित्या द्रुपदकन्या च लभ्यवन्तो महाबलः ॥ ५

Adhy. 199 यदा तु मन्यते वीरः कुन्तीपुत्रो युधिष्ठिरः । भीमसेनार्जुनौ चैव यमौ च पुरुषर्षभौ ॥ ३ रामकृष्णौ च धर्मज्ञौ तदा गच्छन्तु पाण्डवाः । एतौ हि पुरप्रयाण्यावां प्रियेहेति ॥ ४

Adhy. 200 पाञ्चाली भव्रतामेका धर्मज्ञौ यदाशिनी । यथा वात्र नेदः स्यात्तथा नीतितिंवीताम् ॥ १७

Adhy. 202 चन्द्रद्विती महराहारा नक्षत्रप्राणि दिवौकसः । जग्मुर्विषादं तदर्थं दृष्ट्वा युम्होपदव्योः ॥ २६

Adhy. 203 ततो ददृशुरासीनं सह देवैः पितामहम् । सिद्धैर्महर्षिभिश्चैव समन्ततापरिवारितम् ॥ ३ एवं चतुर्मुखः स्थाणुर्महादेवोऽभवत्पुरा । तथा सहस्रनेत्रश्च बभूव बलवृदनः ॥ २६

Adhy. 205 तेषां मधुजलिसंधानां पञ्चनाममितिजासाम् । बभूव कृष्ण स्वर्वैषां पाशर्वायां वशवर्तिनी ॥ २

Adhy. 206 तत्राभिषेकं कृत्वा स तपस्थिला पितामहाम् । उत्तिष्ठितुर्जलादाञ्जलिंकार्यचिकीर्षया ॥ १२

Adhy. 210 प्रतिगृह्यार्जुनः सर्वसद्युमुन च पाण्डवः । सर्वैव बाहुदेवेन दृष्टवान्नोदवम् ॥ १०

Adhy. 211 पौराश्च पादचारो यानैश्चाष्टार्थस्था । सदारः साधुनात्राश्च शतशोऽथ सहस्रशः ॥ ६ एते परिरुक्षाः कीर्तिगिर्नर्वैव पृथक्पृथक् । तमुत्सवे रैवतके शोभयाञ्चकिरे तदा ॥ १२

Adhy. 212 ते समासाद्य सहिताः सुधर्माणमितिः सभाम् । समापालस्य तस्वेनावत्प्योः पार्थिवकमभूत् ॥ १०

Adhy. 220 स गत्वा तपसः पारं देवदुष्करुम भारत । जगाम पितृलोकाय न ह्यमे तत्र तत्फलम् ॥ ७

Adhy. 225 यदा प्रसप्तो भगवान्माहादेवो भविष्यति । दुष्ष्ये तदा प्रदास्यामि पाण्डवास्त्राणि सर्वशः ॥ १०

thirty have no variants at all, while the remaining (seventy) show only insignificant variants, such as transposition, substitution of synonyms, and so on. The number of the latter class of stanzas could naturally be easily augmented, by increasing the latitude of permissible variation. Being handed down uniformly in all manuscripts alike, they may be regarded as authentic (as least as far as manuscript evidence goes), forming, so to say, pieces of firm bedrock in the shifting quicksands of Mahābhārata poetry. As such they will be valuable for the study of epic style, diction, vocabulary and so on. To return to the question of text reconstruction. The rule arising out of the agree- ment between independent recensions or versions is easy to comprehend and simple to apply; only its sphere of operation is rather restricted. Difficulties arise when there is fluctuation; and that is the normal state. When there was fluctuation, the choice fell, as a corollary of the previous rule, upon a reading which is documented by the largest number of (what prima facie appear to be) more or less independent versions, and which is supported by intrinsic probability. Diagrammatically we might represent the types as follows: (i) N₁=S=Text. (ii) N=S₁=Text. (iii) N₁=S₁=Text. N₂ S₂ N₂ S₂ N₃ S₃ N₃ S₃ etc. etc. etc. etc. The presumption of originality in these cases is frequently confirmed by a lack of definite agreement between the discrepant versions. The commonest application of this rule is when Ś₁ K or B (with or without D) agree with S against their own agnates. Numerous examples of this type of agreement have been adduced above (pp. LIV, LXII).

Occasionally we get “double” agreement, that is, agreement between two or more groups of each recension (N₁=S₁ and N₂=S₂); for example, when (1) Ś₁ K=M, and simultaneously B=TG, or (2) Ś₁ K=TG, and simultaneously B=M. Here one of the agreements must, generally speaking, be accidental, since both can hardly be original; and either may be adopted, if they have equal intrinsic merit. Owing to the much greater correctness and reliability of Ś₁ K, I have, as a rule, adopted the readings of this group, other things being equal. When the two recensions have alternate readings neither of which can have come from the other and which have equal intrinsic merit (N : S), I have, for the sake of consistency and with a view to avoiding unnecessary and indiscriminate fusion of versions, adopted, as a stop-gap, the reading of N. This rule is of very common application, since one constantly comes across readings which are but paraphrases of each other and between which it is impossible to discriminate. Examples of such alternative readings are:

N S 1. 23 महर्षेः पूजितस्तेषु सर्वलोके महात्मनः । महर्षेः सर्वलोकेषु पूजितस्य महात्मनः । . 1. 23 1. 51 अन्धे धारयितुं परे केचिद्ग्रन्थस्य धारणे . 1. 51 54. 3 इतिहासान्ध्र समन्वयः इतिहासार्थ समवेगाः . 54. 3 57. 30 पुण्यवनशास्त्रे शाश्वताः सर्वं श्रीयाधः पुण्यवृक् . 57. 30 60. 9 दशरथजायतेद्धृणाद्द्विजशृणगवात्प्रपिः । अङ्कुष्ठाङ्गुष्ठिहृच्छर उत्सुकौ भगवात्प्रपिः । . 60. 9 60. 10 वामादवायतेद्धाक्षुकौ तपस महात्मनः । महर्षैस्तस मार्गो तु वामाङ्गड़ाजायत्त । . 60. 10

N S 65. 20 श्रुग्रं तापयामास शक्रं सुरगणेष्करम्। कम्पयामास देवेन्द्रं पुरा शक्रं महातपाः। 65. 20 65. 35 एतानि यस्य कर्माणि तथाहं मृशयुद्दिजे । एतान्त्यन्यानि कर्माणि सुरां देव बिभेम्यहम् । 65. 35 65. 35 तथाशपय मी विभो तथा पश्च सुरेश्वर 65. 35 66. 2 तपसा दगधकलिश्वम् तपसोप्रेण कर्शितम् 66. 2 66. 3 अभिवाद्य ततः स तां ततोऽभिवाय्य सा तस्मै 66. 3 66. 9 कृतकार्या ततस्तूर्णमच्छच्छच्छसंकृदम् । शक्रसंसदमागतच्छकृत्वा कार्योशपत्तेः। 66. 9 73. 4 वने चैत्ररथोयमे आरामे नन्दनोपमे 73. 4 106. 9 करेणजोषिथ मध्यस्थः श्रीमान्यौरेद्रो गजः। गजराजः करेणुभ्यां यथा मध्यगतस्तथा । 106. 9 107. 20 इत्थुक्त्वा भगवान्न्यास्त्था प्रतिविधाय च । एवं संविय्य कौरव्य कृष्णद्वैपायनस्त्था । 107. 20 200. 9 नारदस्सथ देवर्षिराजगाम यदृच्छया। आयथौ धर्मराजं तु द्दद्धकामोट्थ नारदः । 200. 9

When the above tests break down or when they give only a negative result, the expedient adopted by me was to find a reading which best explains how the other readings may have arisen. The true reading in this case has often proved to be a lectio difficilior, or an archaism or a solecism, the desire to eliminate them being the cause of the variation. Here follow some examples of variation due to the lectio difficilior:

1. 57. 7 ऊषः “udder” (v. 1. क्रुद्धः, ऊर्षे, ऊध्रेः, ऊषेः, ऊडः, ऊउः, क्रुद्रुः, गूढः) 2. 57. 29 मच्छिष्ठः proper name (v. 1. मावेष्ठः, सरुष्ठः, मावेछः, etc) 3. 96. 16 शुप्राणो from शुभ्र “shining” (v. 1. सर्वैर्षो, हेमानो) 4. 98. 13 कम्यस (doublet of कनीयस्) “younger” (v. 1. कमहः, मा गमः, कनीयस्) 5. 98. 18 समुद्र (v. 1. समुद्रै, समुद्र, समुद्रै, etc.) 6. 102. 18 बाधिभो (v. 1. ब विमो, बोधिताः, चान्तिताः) 7. 103. 13 नास्स्रीयो (v. 1. नान्थीयं, नाम्यं श्रये, नाम्सुयां, नान्थश्रियां, etc.).

EMENDATION

Emendation has played a very inconspicuous rôle in the preparation of the constituted text. Interpretation has in general been given preference over emendation. Even in the case of corrupt passages, the reading of some manuscript or other gives sense, though it may not be the original sense, not even a wholly satisfactory sense. Precipitate emendation is, however, to be deprecated; for experience has shown that but a small portion of scholars' corrections are really amendments. Moreover, in this special case, we know, as yet, too little about the epic idiom and the epic world altogether; as also about the vicissitudes of the epic text. Besides, who can say that the original was linguistically uniform, and conformed to any particular norm? What would be the style of a work which in the main is obviously a compilation?

The text, as it has been fixed by me, contains about 35 emendations. The corrections are generally very slight, being concerned mostly with single isolated words, never with whole passages. Wherever even a single letter has been added, omitted or altered, without the authority of any of the manuscripts, I have inserted an asterisk (*) in the text.

Only in very few instances do the emendations effected in this edition make any difference to the sense; e. g. 1. 41. 5 गर्तैऽस्र*तोषाणमिच्छतः, where the word (आ)र्तम् has been

added to the pāda, a word found only in D₂; the other readings are: गर्तैस्त्वाणै, गर्तैस्त्वाणै, गर्तैस्त्वाणै, गर्तेषु त्वाणै, गर्ते स्वराणै (hypermetric!), गर्तैडूसिङ्गाणै, गर्तेस्त्वाणै, seven combinations, each having a different syllable between तै and त्वा ! In a few cases the emendation affects merely some grammatical form of the stanza in question; e. g. 1. 86. 5 अस्मिञ्च्जीबी *नगृहस्य निव, where the readings for नगृहस्य are निगृहस्य, अपगृहस्य, अपगृहस्य (corruption of last?), गुणवांश्च, गुणवांश्चैव, न गृहस्येव (hypermetric!).

But the large majority of our emendations concern merely metre and sandhi. My study of the manuscript material led me to the conclusion that there was an ever growing antipathy, firstly, to hypermetric pādas, in fact to any form of metrical irregularity; and, secondly, to forms of sandhi not sanctioned or countenanced by Pāṇini’s great grammar. In particular there is noticeable a strong aversion to hiatus, even where it was permitted by rules of grammar. Hiatus between pādas also came to be disapproved and was removed by such expedients as that of adding a meaningless हि, नु or च at the beginning of the posterior pāda.

Manuscripts betray the surreptitious efforts of the scribes and redactators to eliminate hiatus (sometimes even when it is grammatically permissible) in the following instances among others: 1. 2. 91 (between pādas) वनवासश् । उद्दुष्या पथि संगमः; 2. 130 पाण्डवा आयुधान्युत् (8 readings); 2. 150 वच राज्ञा उद्दध्यः; 2. 212 तत आभ्रामयत्वायां; 9. 11 (between pādas) *भरणा । उद्विग्नः; 15. 2 इृह्यश्व अपूज्ययनः; 21. 3 काल आद्युय वचनं; 33. 18 भविष्यन्त्यस्स विप्र एवं; 36. 7 स ऊर्ध्व (v. l. स ऊर्ध्वै, स दृध्वे°); 41. 8 गर्तैस् असिञ्षोधामुखाः; (v. l. साधुखुः); 45. 13 (between pādas) कुडूङ् । उत्तरायमिजायत (v. l. सोत्तर°, सोत्तर°, उत्तरा°); 50. 17 राजा वदत्या ऋत्विजी (v. l. सविक्वो) हयवाहः; 60. 4 मरीचिरचिरा अत्रिः (v. l. *राधात्रि, *राधात्रिः, *राधात्रिः); 65. 24 (between pādas) दुदुध्वे । उमे (S °धेर्थ्योमे°); 72. 22 N देव इन्द्र (S देवाहेन्द्र°); 76. 18 ऋषिश्चपुथ्रिषुदुध्वे (v. l. *सृष्टिधृवि°, *सृष्टिधृ°, *सृष्टिधृ°, *सृष्टिधृवि°); 83. 3 श्लोक अन्तनन्तव्व (v. l. भ्रान्त°, भ्रान्त°, भ्रान्त°); 84. 13 महतो अजय वै; 85. 8 तथा अशीतिं परिवत्सराणि (v. l. तथाषीतिं च परि°); 94. 38 मेने आत्मानं (आत्मान°, सोत्मा°, स्वत्मान°); 96. 42 (between pādas) कालेन् । अकात्तम् (v. l. स्वत्तका°, स्वत्काम°, सोलका°, स्वत्तका°, सोत्तका°); 98. 3 अनवेल्ली मई (v. l. सहै, सहै, *सहै, *साहिस भै); 99. 15 च अभितुदुधुवे, 99. 39 (between pādas) *त्रतोपेता । उपेयात् (v. l. दुयेयात्); 100. 2 निशीथे आगमिष्यति; 101. 3 स आश्रम (v. l. वाश्रम°, त्वाश्रम°); 103. 5 (between pādas) कन्या । अनुरूपा (v. l. साधु°, त्वसु°, साधु°, बहु°, etc.); 107. 32 (between pādas) जनपदस्यायें । आत्मानें (S₁, K₁ स्वात्माणै°); 109. 7 (between pādas) *तेजा । ऋषिवि (M₀-S ऋषि°); 109. 21 (between pādas) च । अधर्मीडं च; 110. 28 यदि आवां; 112. 31 वा त्रशुद्धलाता; 114. 38 देवर्षीणां (v. l. *यत्मान°, *यत्मानो, *यिद्यात्मां, *यर्षीणां); 116. 25 तच्छेष्टा अनुमन्यन्तां; 148. 1 विदित्वा अपकर्षयं; 152. 19 सर्वे आजसुः; 157. 13 त्तथा उक्ताः (6 readings); 183. 3 (between pādas) तांश्च । उपोविषानां (N ins. अपि); 218. 11 (between pādas) वासुदेवश्च । अप्रतिष्ठो. — It is evident that sandhi was originally more flexible. It is only in later phases of literature that writers make a shibboleth of it.

Similar efforts to correct hypermetric lines may be seen from: 20. 2 विनतौ विषणवदनां (v. l. विषण्णवदनां विनता, विसन्नवदनां, विषण्णवदनां कदन्) 78. 23 त्वरितं सकाशं काव्यस्य (v. l. शीघ्रं and द्रुतं for त्वरितं; also पितुः सकाशं त्वरितं, त्वरया काव्यसकाशं) 92. 4 करवाणि किं ते कल्याणै (v. l. किं ते करोमि कल्याणि, करवाणि किमयाहं, किं ते कल्याणि करवै) 94. 74 बलवत्प्रसप्तमताम् (प्राप्तमताम बलवत्, तत्र चापम्°, तत्र चार्म°)

Owing to the increasing sensitiveness to solecism, we find likewise different efforts made, independently of each other, to purge the text of what came to be regarded as stylistic blunders or corruptions in the ancient text. Examples of attempts made to

remove solecisms are: 1. 190 वे च वर्तन्ते (v. 1. वर्तन्ते ये च, ये वर्तन्ते च); 2. 93 हरणं शुश्रूषा संग्रामे (v. 1. गृहीत्वा हरणं प्राप्ते, दत्त्वा चारणं तस्मै); 7. 26 पुलोज्मस (v. 1. पुलोमसु, °ब्रह्म, °स्रोथ); 9. 2 चिन्त्व (v. 1. स्थूला, मध्या); 21. 6 प्रक्रारणम् (‘फणिनोर्दम°’, चार्योडम्°); 43. 14 इयात् (v. 1. अगात्, अयात्); 46. 37 शेषेवैचनं (v. 1. शक्षिणो वाक्ये); 48. 24 सा ब्राह्मणान् (v. 1. ब्राह्मणान्); 96. 44 यथा दुहितरवैच (v. 1. ययावहुदुर्देषैव); 123. 16 त्रयोपकरणं गुप्र (v. 1. करणमाही); 124. 24 उद्वान्तौ pass. pres. part. (v. 1. उद्वान्तै); 141. 7 नयिष्यामि गृह्य वासरी (v. 1. वासखडाद्य); 154. 24 भागीरथ्याम्प्रसुत्तरे (v. 1. जाह्वयामहमुपु); 165. 24 बलाद्रुयसि मे नन्द (v. 1. हियसे त्वं बलाद्गुयरे etc.); 169. 20 दाचार तेजस (v. 1. दत्त्रे महौजस etc.); 184. 18 धरन्ति (v. 1. वसन्ति, कद्रन्ति, चरन्ति, प्रियन्ति, जीवन्ति, वदन्ति, हरन्ति!); etc., etc. I add examples of hypermetric pādas (generally with the scheme ‿–––‿–), which are the result of emendation: 1. 30. 7 प्रतिगृद्यतामिदानीं मे*; 1. 155. 35 *अवलिंसे मे मुखं ब्रह्मन्; And, finally, examples of hiatus as the result of emendation:

51. 8 अर्थो* इन्दः स्वयमेवावगाम 116. 25 तज्ञयेषा *अनुमन्यानां 57. 20 क्रियते *उच्छ्र्यो नृपैः 119. 11 तथैत्येकै *अम्बिकया 98. 8 अन्तर्वैली *अहं मात्रा 147. 2 रोर्वयो *अनाथवत् 99. 15 त्वया च *अभिहितुवे 148. 1 विदिस्वा *अपकर्षवै 100. 2 निशीथै *आगमिष्यति 157. 13 पञ्चकृत्वस्त्वया *उक्तः 103. 5 श्रूयते यादवी कन्या *अनुरूपा कुलस्य नः। 207. 17 कुले *अस्मिन्नभुत्त ह 110. 20 नाहं भाष्विदं मार्गं *अवीर्ययुपचोशिते । 214. 9 धर्मराजे *अतिषीत्वा 110. 28 यदि *आवां महाप्राज्ञ 224. 5 संतप्यमाना (sing.) *अभितो.

It is important to remember that emendation has been resorted to merely for the purpose of unifying divergent and conflicting manuscript evidence, never in opposition to clear and unanimous testimony of manuscripts. The emendations are thus not amendments of the text in the ordinary sense of the word, made in order to eke out a better sense when the manuscripts yield no sense or an unsatisfactory sense; they are rather an effort to find, so to say, a hypothetical focus towards which the discrepant readings converge.

THE “ADDITIONAL” PASSAGES

The uniformity of the interrelationship of the different manuscripts, versions or recensions, as has been already explained, is disturbed chiefly by comparison and conflation of manuscripts. A constant and fruitful source of confusion, as was pointed out above, has always been the marginalia. A more dangerous and troublesome source was the practice of incorporating into one’s text—without stating the source and without much explanatory comment—passages found in other versions. It may be surmised that celebrated places of pilgrimage like Ujjayinī,⁵ Rāmeśvaram, Kāśī, and others, with recitations of the epics held periodically in their famous shrines, have played an important rôle in the dissemination of the knowledge of local versions among the pious visiting pilgrims, whose number undoubtedly included the bards and the professional reciters of the epics.

Much light is thrown on the origin of these misch-codices by the MS. K₄ a manuscript belonging to the Bombay Government Collection deposited at the Institute. In this manuscript we find long extracts from other cognate versions (such as T) as also from the Southern recension, written out on separate folios and inserted at appropriate places in the body of the manuscript, with the words अत्र शोधपत्रमर्कं written on the margin of the original folio, near the place where the passage is to be interpolated. Should this manifest happen to be copied again and should the copyist insert the passage at the place indicated by the previous scribe, the interpolation would become an integral part of the new text which is externally absolutely indistinguishable from the rest of the text. This leads us to the question of "additional" passages in general. Our attitude with regard to them is quite clear, in my opinion. The first and foremost source of our knowledge as to what the Mahābhārata comprises, is and must remain the manuscript evidence itself. For example, the question—which seems to trouble a great many people, judging by the inquiries on the point received at the Institute—whether the Uttaragītā, Gajendramokṣa and Anusmrti are parts of the Mahābhārata, must be answered by the manuscripts themselves. If none of our manuscripts contain these passages, it is prima facie evidence that they are not parts of the Mahābhārata. There is nothing to suggest that our Mahābhārata manuscripts have suffered any serious loss at any time. There never was any lack of manuscripts, many of which were preserved carefully in temples, and which must have been copied repeatedly, for the enhancement of merit. There is no evidence of any break in the tradition at any time or any place, within the confines of India at least. The probable inference is that our manuscripts contain all that was there originally to hand down, and more. What late writers and commentators have said about passages not found in our manuscripts is always a matter of secondary importance; it cannot ipso facto nullify or override the primary evidence of manuscripts. Such extrinsic testimony has only local or personal value; it can always be rebutted by the evidence of the Mahābhārata manuscripts. Likewise, whether an episode, adhyāya, passage, stanza or line may be regarded as belonging to the Mahābhārata or not must primarily depend upon whether the manuscripts contain it. Extrinsic evidence, in so far as it is valid, will principally hold good only for the period or locality to which it belongs. Intrinsic evidence may be considered; but, being of a subjective character, it must be used with caution. Our primary evidence being the manuscripts themselves, we are bound to view with suspicion, as a matter of principle, any part of the text which is found only in one recension, or only in a portion of our critical apparatus. Therefore, the evidence for such passages as are contained only in one manuscript, or a small group of manuscripts or versions, or even in a whole recension must be pronounced to be defective. Consequently, all lines belonging to one recension only, and a fortiori such as pertain to a combination of manuscripts amounting to less than a recension, for which there is nothing corresponding in the other recension and which are not absolutely necessary for the context—all lines, in short, with a defective title—have been placed in the footnotes or the Appendix, pending further inquiry regarding their credentials. Such passages are not all necessarily spurious. There might be a hundred good reasons why the questionable passages are missing in a particular recension or version. It might conceivably be, for instance, that the shorter recension represents (as a certain

scholar has said) "a mutilated and hastily put together composition of the Middle Indian Redactors, who could not lay their hands on all manuscripts of the Mahābhārata".⁶ The shorter version might again be, theoretically, a consciously abridged or expurgated version. Or, more simply, the omission might be due to mere oversight of some scribe who had quite unintentionally omitted the defaulting passage and this mistake of the first scribe had been perpetuated by the other copyists. And so on and so forth. But all these are mere possibilities. All these reasons in general and particular must be adduced and proved, or at least made probable, in any given case. Moreover, the manuscripts clearly show that there has been in progress, through centuries, constant comparison of manuscripts. In view of this circumstance, the explanation that the omission of a passage in a whole version might be due to a scribe's omission loses much of its force. Omission is as much a fact in Mahābhārata textual tradition as addition. And it is fair to demand of a person who alleges the authenticity of such one-recension passages why the rival recension does not contain it.⁷

The general condemnation of a recension or version that it is mutilated, merely on the ground that it lacks certain passages that are found in a rival recension or version, is entirely meaningless; for the argument might easily be reversed, so that the controversy will resolve merely into mutual vituperation. What I mean is this. From the fact that one of the recensions, say N, does not contain a certain passage or a certain set of passages found in another, say S, it is illogical to argue that N is a mutilated version; because such an argument can with equal cogency be applied to S, in regard to certain other passages that are missing in S but found in N. The point is so important and at the same time so difficult to grasp that I shall endeavour to make my meaning clearer with the help of a concrete illustration. My contention is this. From the fact that the Southern recension contains, say, the Nārāyaṇi episode (App. I, No. 100), which is missing in the Northern recension, it would be illogical to argue that the Northern recension is defective or mutilated; because one can, with equal cogency, seek to establish the mutilation or defection of the Southern recension by pointing, say, to the Gaṇeśa passage, which is found only in certain Northern manuscripts and is entirely missing in the Southern manuscripts. The argument could have been employed with greater semblance of reason and plausibility, had there been only a mere plus or minus on either side, but is entirely without cogency in the present instance where there are both additions and omissions on both sides.

Originality and authenticity are, unfortunately, not the prerogative of any single recension or version or manuscript.¹ They must be established, laboriously, chapter by chapter, line by line, word by word, syllable by syllable.² The optimistic view that any extant manuscript, however old and trustworthy, of some favoured version or recension, could give us, with a few additions and alterations, the text of Vyāsa's Bhārata or Mahābhārata is the index of a naive mentality and does not need any elaborate refutation.

The argument in favour of any particular recension or version or text is frequently sought to be strengthened by a reference to the authority of the Parvasamgraha (Ādi 2), a weak reed on which every tyro leans rather heavily in the beginning, and it would be well to examine the argument here.

The Parvasamgraha Argument

Until lately high hopes had been entertained that the Parvasamgrahaparvan (Ādi 2) would supply the clue to the solution of the perplexing question of the reconstruction of the original Mahābhārata. But the paradoxical situation created by the circumstance that two different editors of the Virāṭaparvan, both of whom rely mainly on the data of the Parvasamgraha for establishing the originality and authenticity of their respective texts, have produced critical editions of that parvan which differ by no less than 1467 stanzas,³ has created grave misgivings in the minds of unbiased critics as to whether the Parvasamgraha can render us any help at all in reconstructing the text of the Mahābhārata, and these misgivings appear justified by the facts of the case.

The exaggerated importance which the late Mr. Utgikar was inclined to attach to the numerical data of the Parvasamgraha, was, I believe, mainly, if not wholly, due to his mistaken belief that there was complete agreement between the two rival recensions in all material particulars as regards the text of this adhyāya. This erroneous and wholly unfounded notion seems to have been induced by the ambiguous and thoroughly misleading character of the text of the Kumbhakonam edition, which claims to be an edition “mainly based on South Indian texts”, but presents a version of this adhyāya which has been unblushingly copied from the Bombay and Calcutta editions, ignoring wholly the Southern divergences, which are quite considerable.

Not only are there discrepancies between the two recensions as regards the numbers of the adhyāyas and ślokas in the various parvans, there is no complete agreement even between the different versions of the same recension. Take for instance, the case of the Ādiparvan itself. Our constituted text (following the Śāradā codex) gives (1. 2. 96) the number of ślokas in the Ādi as 7884. But this is not the only reading of that number. For the digit representing the thousands alone, the choice lies between seven, eight, nine and ten! There can, therefore, be no doubt that the text of this adhyāya also has been tampered with and designedly altered, from time to time in various

ways, in order to make it harmonize with the inflated versions of a later epoch. It will thus have to be admitted that the Parvasamgraha argument is of secondary importance and must not be pressed too far. Be that as it may, it is extremely problematic whether we could make any use whatsoever of the Parvasamgraha enumeration of ślokas in the case of the Ādi at least, because it will be difficult to compute the exact extent of this parvan and that for two reasons. Firstly, because this parvan, as is well known, contains two lengthy prose adhyāyas (3 and 90). Taking the figure of the Parvasamgraha to represent the exact extent of the whole of the Ādi, it is not clear how the prose portions were computed by the compilers of the Parvasamgraha. Most of the modern computers add the number of stanzas to the number of their respective prose sections, and arrive at the length of the Ādi in ślokas! But this is bad arithmetic. P. P. S. Sastri offers a solution which is more ingenious than convincing. He holds the compiler of the Parvasamgraha down to the letter of his statement. The Parvasamgraha tells us, says Sastri, merely the number of ślokas which the different parvans contain. Nothing is said about the prose sections. He therefore ignores the prose adhyāyas in computing the extent of the Ādi, and is satisfied that his text exactly agrees with the data of the Parvasamgraha! The other difficulty in the way of using the Parvasamgraha figure in the case of the Ādi is that this parvan contains a large number of Triṣṭubh stanzas, which again introduce an element of uncertainty in the computation. Was each Triṣṭubh stanza counted as one śloka; or did the Bhārata-cintakas (mentioned in 1, 2, 172) compute the exact equivalent of the long-metre stanzas in ślokas? It is difficult to say. The difference in the reckoning will be, however, between 40 and 50 per cent of the total! As a very rough estimate, the Ādi may contain something like 500 long-metre stanzas. This factor alone would introduce a difference of about 225 stanzas! These are some of the obvious difficulties in the way of making any practical use of the figure recorded in the Parvasamgraha for text-critical purposes. The computation may have some value in the case of a parvan in which there is no prose at all, which is almost wholly in anuṣṭubh metre, and for which finally the Parvasamgraha figure is certain, the manuscript evidence being unanimous. It is quite within the range of probability that the apparent extent of the critical text of a parvan may fall appreciably below or rise appreciably above the figure recorded in the Parvasamgraha, as is actually the case with other editions. Moreover, unless it can be made probable that the compilation of this “Table of Contents” is nearly contem- poraneous with the present redaction of the Great Epic, these discrepancies will be without much cogency in matters relating to the constitution of the text. The value of a manu- script, version or printed text of the Mahābhārata must not be thought to depend exclusively or even mainly upon its agreement with or discrepancy from the numerical data of the Parvasamgraha. It must in final analysis be regarded as depending upon the place it occupies in a logical and convincing scheme formulated to explain the evolution of the different extant versions and types of Mahābhārata manuscripts. It should further be carefully borne in mind that even if there be exact agreement as to extent between the Parvasamgraha and any constituted text, this fact alone is no guarantee of the absolute correctness of the entire text, line for line, because the same

number of stanzas could be made up in innumerable different ways by accepting and rejecting stanzas of doubtful authenticity and uncertain documentation, of which there is always a plentiful supply in every parvan. The difficulty will finally not be solved even if we happen to light upon a unique manuscript which agrees with the Parvasaṃgraha exactly as to the number of stanzas in any particular parvan and we should adopt its text verbatim; because there is every probability that while it satisfies the one criterion of extent given by the Parvasaṃgraha, it may not satisfy, in every respect, other and more exacting critical tests, when compared line by line and word by word with other extant manuscripts.

In the above discussion I have implicitly assumed, as is done by most writers on the subject, that the word śloka in the Parvasaṃgraha chapter has the usual meaning “stanza”. This interpretation was called into question by the late Mahamahopadhyaya Haraprasad Śastri, who offered a new interpretation, which I cannot but think is far more plausible, although I do not agree with all the conclusions he deduces therefrom.

The really valuable discovery of the Mahamahopadhyaya, in my opinion, is that the word śloka cannot mean here stanza or verse or anything of the kind, but must denote (as in the parlance of scribes and vendors of manuscripts) a unit of measurement of written matter, comprising 32 syllables or akṣaras⁴. The difficulty of computing prose passages and the long-metre stanzas mentioned above finds a satisfactory solution at once in this interpretation of the word “śloka”. And that is moreover the only interpretation of the word which, as far as I can see, can successfully solve that difficulty, in view of the circumstance that the text is heterogeneous, consisting of ślokas, prose, and long-metre stanzas. But in this supposition we shall have to count, not only the actual text (consisting of prose and verse), but the whole of the written matter. And that enumeration, whether it be 7884, 8884, 9884 or 9984, will include not only the text properly so called but also the colophons and the hundreds of the prose formulaic references (like वैशंपायन उवाच), besides perhaps the captions of adhyāyas, sub-parvans and parvans, and even the numerical figures denoting the numbers of ślokas, and so on.

The number of adhyāyas in our edition (225) does not tally with the number given in the Parvasaṃgraha (219), any more than in any of the previous editions: the Calcutta edition of the Ādi has 234 adhyāyas, the Bombay editions vary between 234 and 236, while the Kumbhakonam edition reaches the astonishing figure 260, though the Parvasaṃgraha figure in the case of each of these latter editions is the same, 227.

It may be pointed out that the adhyāya division in our extant manuscripts is extremely arbitrary. The average length of our adhyāyas should be about 35 stanzas; but adhy. 12 and 22 of our edition contain only 5 stanzas each, while adhy. 57 (to mention only one instance) has over 100 stanzas. As regards the contents of the adhyāyas also there is much inconsistency. Thus we frequently find that one adhyāya ends with the remark that a certain person spoke as follows, and his speech, which may be quite short, forms the beginning of the following adhyāya. Then again the manuscripts are far from being unanimous in the matter of marking the colophons; they show in fact wild

σ PROLEGOMENA

fluctuations. Even the reading of the Parvasaṁgraha figure is not entirely free from doubt (e. g. our Śāradā codex gives the number of the adhyāyas as 230!), though the reading 218 seems highly probable.

Under these circumstances, nothing would be easier than to manipulate the colophons, by arbitrarily combining the conflicting data of the different recensions or versions or even manuscripts and arriving at any required figure. This has actually been done by Professor P. P. S. Sastri in his edition of the Southern Recension, which thereby achieves the dubious distinction of being the only edition of the Ādiparvan in which the adhyāya number agrees exactly with the Parvasaṁgraha figure but the colophons are mostly at the wrong places. This procedure is the less excusable in his case as he is at great pains to create the impression that he is just reproducing the text of one selected manuscript, correcting only "scriptorial" blunders. Now his basic manuscript (अ = our G₅) divides the Ādi into two separate major parvans, Ādiparvan and Sambhavaparvan, with 40 and 200 numbered adhyāyas respectively, which makes a total of 240 adhyāyas, and which is nearer the Kumbhakonam figure (260) than the Parvasaṁgraha figure (218). While correcting "scriptorial blunders", Professor Sastri has, so to say, spirited away 22 colophons before our very eyes.

A more careful study of the manuscript evidence may tend to reduce the discrepancy between the constituted text and the data of the Parvasaṁgraha as regards the number of the adhyāyas, or at any rate may enable us to account for the difference, though at present it seems impossible to harmonize the manuscript evidence (consisting of the actual colophons) with the Parvasaṁgraha.

INTERPOLATION

There has been an extraordinary reluctance among scholars to face the fact that the Mahābhārata manuscripts may contain and do contain quantities of spurious matter. But there is now no excuse for such recalcitrance. The critical apparatus of this edition contains a unique record of hundreds of lines which are evidently and unquestionably spurious. Here is a list of passages from our Appendix, each found in one manuscript only: App. I, No. 2 (in K₆ marg.: containing 4 lines); No. 4 (K₈: 14 lines); No. 5 (B₄: 23 lines); No. 7 (G₁: 4 lines); No. 16 (K₄: 9 lines); No. 25 (D₅: 4 lines); No. 26 (B₄: 6 lines); No. 31 (K₄: 27 lines); No. 34 (K₄: 6 lines); No. (D₈: 21 lines); No. 66 (D₄: 47 lines); No. 70 (G₁: 8 lines); No. 74 (B₁: 9 lines); No. 94 (D₄: 31 lines); No. 98 (D₄: 50 lines); etc., etc. These are passages from the Appendix alone, to which many of them have been relegated on account of either their length or their irrelevancy; but the foot-notes contain hundreds, nay thousands, of lines of precisely the same character. Then there are also lines which are found in only two or three manuscripts, of which I have counted some 300 instances. A number of new additions have been now given by Professor Sastri, who has examined other Telugu and Grantha manuscripts for his edition of the Ādi in the Southern recension. And I am fully persuaded that if we examine yet other manuscripts, we shall still find fresh passages which had never been seen or heard of before. No sane person would maintain that these are all original passages lost in all manuscripts except the few late and inferior manuscripts in which they happen to occur.

It is not always easy, as has already been remarked, to prove that these “additional” passages are interpolations. The epic metre is easy to imitate; the epic grammar is flexible; the epic style is nondescript. The additional lines are generally fashioned with skill, and fitted in with cunning. The following interpolated stanzas, by a poet aspiring after higher things, in fancy metre and classical style are rather exceptional: 1859* भीम उवाच । रे भुजुओ यदि भुवोल्लसितं न किंचि- त्तार्किक स्फृहृहाजनि सुतां प्रति पार्वतस्य । जज्ञे स्पृहृहाय कथमागतमागतं वा प्राणार्थिके धनुषि तत्कथमग्रहोऽभूत् ॥ कस्य द्रोणो धनुषि न गुरुः स्वस्ति देवव्रताय मन्दाभ्यासः कुरुपतिरयं श्रीसमुर्थंविर्वलालेः । रे कर्णायाः श्रुणुत मधुरां ब्राह्मणस्यानु वाणीं राधा यन्मं रचयेतु पुनर्विद्धमप्यस्य विद्धम् ॥

An interesting instance of a passage which is betrayed by its contents is an extravaganza in some Grantha manuscripts. This bizarre interpolation¹ describes among other things, with circumstantial detail, the marriage of Parāśara and Satyavatī ( alias Matsyagandhā). At this ceremony, the shades of the ancestors of both the bride and the bridegroom are invoked, all the details of a regular Hindu marital rite are minutely observed, and the marriage is solemnized in the presence of Vasiṣṭha, Yajñavalkya and other great Ṛṣis living in the Naimiṣa forest, with the distribution of bakṣheesh to Brahmins. It is an interesting speculation whether credulity can go so far as to regard even such passages as an authentic part of the original Mahābhārata or Bhārata of Vyāsa, just because the passage is found in some Mahābhārata manuscripts.

The foot-notes contain a rare selection of passages that are either palpably absurd, sometimes contradicting the immediate context, or else have little connection with the context in which they lie embedded: quotations, glosses, fanciful additions of details, the jetsam and flotsam of Mahābhārata poesie.

These bewildering fluctuations in the text are quite unique, being peculiar to the Mahābhārata. They are not found in the manuscripts of the Vedic literature or in those of grammatical, philosophical, or rhetorical texts or of the works of the classical poets and dramatists. This only proves that the Mahābhārata was peculiarly liable to inflation and elaboration.

When I say that the Mahābhārata manuscripts contain quantities of spurious additions, I intend no disparagement or condemnation of the text or of the manuscripts. The process is normal, inevitable and in a wider sense wholly right. If the epic is to continue to be a vital force in the life of any progressive people, it must be a slow-changing book! The fact of expurgation and elaboration is only an outward indication of its being a book of inspiration and guidance in life, and not merely a book lying unused and forgotten on a dusty book-shelf. Those are probably just the touches that have saved the Mahābhārata from the fate of being consigned to the limbo of oblivion, which has has befallen its sister epics like the Gilgamesh.

To give only one illustration. The awkwardness of the sexual relations of some of those epic characters of bygone ages must have been indeed a puzzle and a source of constant tribulation to the reciter of the epics (Paurāṇika), who was called upon to narrate, explain and justify those old-world stories to his devout and impressionable audiences, in the course of his recitations, which were, in the post-epic period, nothing more than edifying popular sermons. It is then no wonder that the shrewd ones among these pastors of the people, these professional keepers of their morals, should have occasionally taken the bull by the horn, so to say, and boldly added or substituted, bona fide, details which harmonized better with their own conceptions of right and wrong or with those of their pious flock.

A PROBLEM IN “TEXTUAL DYNAMICS”

After what has been said above, it is needless to add that the constituted text is based on all versions of both recensions and prepared on eclectic principles. I have given in the text whatever in each case appeared to be supported by the balance of probabilities, but all important deviations in the manuscripts are noted in the critical apparatus, so that every reader has, at his disposal, the entire material for controlling and correcting the constituted text, where necessary. All important elements of the text—lines, phrases, significant words and even word-parts—that are less than certain, are indicated by a wavy line printed below them. Slight differences in the spellings of words, of proper names (e. g. नैमिष : नैमिष) and some minor details (such as the expletives or the prose formulae सूत उवाच, सौतिरुवाच, सूतः etc.) are ignored for this purpose. This device is, by nature, hard to apply strictly, and there are bound to be many inconsistencies in its application. I have retained it all the same with the express object of obviating all false sense of security. This wavy line, running through the entire length of the text is, to my mind, the symbol and constant remembrancer of this essential fact in Mahābhārata textual criticism that the Mahābhārata is not and never was a fixed rigid text, but is fluctuating epic tradition, a thème avec variations, not unlike a popular Indian melody. Our objective should consequently not be to arrive at an archetype (which practically never existed), but to represent, view and explain the epic tradition in all its variety, in all its fullness, in all its ramifications. Ours is a problem in textual dynamics, rather than in textual statics.

To put it in other words, the Mahābhārata is the whole of the epic tradition: the entire Critical Apparatus. Its separation into the constituted text and the critical notes is only a static representation of a constantly changing epic text—a representation made for the purpose of visualizing, studying and analyzing the panorama of the more grand and less grand thought-movements that have crystallized in the shape of the texts handed down to us in our Mahābhārata manuscripts.

What Is The Constituted Text?

To prevent misconception in the mind of the casual reader, it is best to state at first what the constituted text is not.¹ The editor is firmly convinced that the text

presented in this edition is not anything like the autograph copy of the work of its mythical author, Maharṣi Vyāsa. It is not, in any sense, a reconstruction of the Ur-Mahābhārata or of the Ur-Bhārata, that ideal but impossible desideratum. It is also not an exact replica of the poem recited by Vaiśampāyana before Janamejaya. It is further wholly uncertain how close it approaches the text of the poem said to be recited by the Sūta (or Sauti) before Śaunaka and the other dwellers of the Naimiṣa forest.

It is but a modest attempt to present a version of the epic as old as the extant manuscript material will permit us to reach with some semblance of confidence. It is, in all probability, not the best text of the Great Epic, possible or existing, nor necessarily even a good one. It only claims to be the most ancient one according to the direct line of transmission, purer than the others in so far as it is free from the obvious errors of copying and spurious additions. It may be regarded, if the editor has done his work properly, the ancestor of all extant manuscripts, or, to be precise, of the manuscripts examined and collated for this edition. The constituted text cannot be accurately dated, nor labelled as pertaining to any particular place or personality. Since our manuscripts are comparatively modern, our text cannot claim to be very old. It goes without saying that (precisely like every other edition) it is a mosaic of old and new matter. That is to say, in an average adhyāya of this edition (as of any other edition) we may read a stanza of the second century B.C. followed by one written in the second century A.D. Sometimes the gap will occur in the middle of a line, precisely in every other edition. This unevenness and these inequalities are inevitable, conditioned as they are by the very nature of the text and the tradition.

The Vulgate text of the Mahābhārata is fairly readable and will appear in places, at first sight, to be even “better” than the critical text, because the former has been purged by the continuous emendations of scholars for centuries. A whole army of anonymous scholars and poets must have worked at the text to make it smooth and easy of comprehension, and to increase its popularity and usefulness by adding to it interesting anecdotes, incorporating into it current and popular versions and explanations, bringing it in a line with the ethical, moral, religious and political ideas of essentially different ages.

The reader will find that the constituted text is by no means smooth. It contains fresh instances of loose and archaic linguistic forms and constructions, anacoluthons and lack of syntactical concord. There remain many contradictions and superfluities. There is evident lack of finish in the hidden parts. These blemishes—if they be blemishes in epic poetry, which is dynamic poetry, with no necessary pretensions to niceties of style, in the narrower sense of the term—must have been inherent in the old poem. Where they are met with in the critical text, they are not speculative fiction; they are documented by the manuscripts themselves or at least are inferable from them with a high degree of probability.

For the shortcomings mentioned above, the constituted text has merits also. It cleanses the text of puerile modern accretions and obvious errors of repetition, which lengthen and weaken the text. It solves a certain number of textual riddles (bogus kūṭas), which were the outcome of long standing corruptions and unskilful conflation. It rescues from undeserved oblivion many an authentic archaism, which had been gradually ousted in the course of transmission of the text. Sooner than print up the text of one manuscript, however reliable it may be, declining to shoulder the responsibilities attaching to the work of an editor, I have ventured on the perilous path of text reconstruction, in the hope and belief that it will present a more faithful picture of the original than any extant manuscript could do. That to prepare such a text is a phenomenally difficult task, no one can realize better than the editor himself. It is as certain as inevitable that in preparing a text like this the editor will frequently make blunders, even gross blunders. It is to be feared that there is no royal road in this incomparably difficult field. The only path left open to us by which we may return to the original Mahābhārata or Bhārata is the rough, narrow, scientific foot-path of repeated trial and error. More than one attempt will probably have to be made before the ideal is attained. It will, therefore, be prudent not to claim too much for the first critical edition, nor to expect too much from it.

Other Editions

Of the old editions it must be said that they are creditable performances, but they lack the critical apparatus. We do not know on what manuscripts they are based, according to what principles the editors have prepared the text, information essential on account of the wild fluctuations of the manuscripts. That is why they have been almost wholly ignored in the present edition. The editio princeps (Calcutta 1836) remains the best edition of the Vulgate, after the lapse of nearly a century. The later text editions, as is unfortunately too often the case with our editions, add to the editio princeps only a fresh crop of spurious lines and misprints. The well-known pothi-form Bombay editions (published by Ganpat Krishnaji in Śaka 1799, and Gopal Narayan in 1913, and others), which include Nīlakaṇṭha's scholium, are supposed to represent Nīlakaṇṭha's text; but they contain many readings and lines which are not to be found in the Nīlakaṇṭha manuscripts, and are therefore not wholly reliable. The Kumbhakonam edition, which is said to be “mainly based on the South Indian texts”, is a fine representative of the composite Telugu version; it has been of immense help to me in the study of what may be called “conflate” readings. In former years its chief value lay in that it gave the reader glimpses, however imperfect and confused, of the important Southern recension. It is now rendered obsolete and superfluous by P. P. S. Sastri’s new edition of the Mahābhārata, which will presently be described, and which is unquestionably a better representative of the Southern tradition.

The Grantha edition (Sarfojirajapuram 1896) and the old Telugu edition (Madras 1855) were not examined: they are not likely to contain anything of high importance that is not found in the other editions or manuscripts collated for this edition.

The editions accompanied by vernacular translations, which form a very numerous class, are mostly bad reprints of one or the other of the earlier (printed) editions and may be completely ignored here; they are perfectly useless for critical purposes.

The new edition¹ of the Southern recension of the Mahābhārata by Professor P. P. Subrahmanya Sastri of Madras, now in the course of publication, which has been referred to several times already, is a laudable attempt to supply a long-felt want. He deserves the cordial thanks of all lovers of Sanskrit literature in general and of the Great Epic in particular, for his courageously undertaking such a stupendous and exacting task and pursuing it steadfastly, single-handed, during the scanty leisure permitted by his official duties as Professor of Sanskrit in the Presidency College of Madras, and Editor of the Descriptive Catalogue of Sanskrit Manuscripts in the extensive library of the Saraswathi Mahal at Tanjore. The edition is in no sense rendered superfluous by the Critical Edition, although most of the information it contains is or will be included, in some shape or other, in the present edition. The gulf between the Northern and the Southern recensions is so vast, that it is extremely difficult, if not practically impossible, to reconstruct the Southern text, completely and correctly, from the critical notes of this edition.

The principles on which the text of this edition of the Mahābhārata is prepared have been set forth and briefly discussed above. The editor, it was pointed out, fondly cherishes the unfounded belief that he is printing a Grantha manuscript as it is, but consciously and unconsciously he has introduced so many important innovations, that the text, as a whole, must be pronounced to be eclectic as eclectic as any text—at least as far as the Ādiparvan is concerned—published so far. For far less important deviations from the manuscripts have I condemned, above, the editions of the Vulgate. Judged as an eclectic edition, it must be pronounced to be inferior. The principle Sastri has laid down is a simple one to follow; in fact nothing could be simpler: he is to print the text of a selected manuscript as it is, only correcting clerical errors. And it is to be greatly regretted that he does not follow rigorously this principle. He constantly flouts it, in pursuit of some imaginary norm. Clear as his principle is, his actual procedure is somewhat paradoxical. He has left innumerable minor “inferior” readings in possession of the text (when he could have with perfect confidence, if not certainty, put into his text the correct readings), because he ostensibly wants to present the text as it is in one selected manuscript; on the other hand, he has light-heartedly, on utterly insufficient grounds, effected very substantial additions (in one instance extending to 140 lines), omissions and other unwarranted alterations (such as transpositions of adhyāyas), in the utterly mistaken (though unquestionably bona fide) belief that he is correcting only the “scriptorial blunders” of his exemplar, when they are in reality (as is shown by the evidence of cognate versions) nothing of the kind.

The subtitle “Southern Recension” is perhaps a trifle ambitious, at least as far as the Ādi is concerned; because, firstly, he has utilized only six Southern manuscripts (1 Telugu and 5 Grantha), even less than the number (18) of the Southern manuscripts collated for our edition; and, secondly, he has completely ignored one whole Southern version, the important Malayālam version, in my opinion, the most important of Southern versions.

Further, it may be questioned whether the edition deserves to be called a critical edition at all, since, as was pointed out above, the editor is avowedly aiming only at reproducing the text of one manuscript, categorically renouncing the obligation of the textual critic to restore the text, as far as possible, to its original form.

The inclusion in Sastri's text of a certain number of stray lines and even a few lengthy passages which are peculiar to the Northern recension and absolutely foreign to the Southern,² throws much light on the unconscious process of the growth of the epic and the irresistible influence which the Vulgate exerts on a text that is coming into being, in other words, on that subtle process of textual osmosis (if I may term it so) by which the epic texts have become conflated. Sastri’s explanations in his Introduction as well as his procedure elucidate much of the psychology of the ancient scribes and redactois, who have in the past shaped our Mahābhārata texts for us. Unconsciously he seems to have worked on the identical principles on which the ancient scribes have worked. His edition is a true lineal descendant of the Mahābhārata manuscripts of South India.

In preparing Appendix I of this edition (in which there is a strong preponderance of the Southern element), I had to go rather carefully over Sastri’s text of the Ādi, when I came across far too many inaccuracies in the passages for which I checked his text and critical notes with the collations of the manuscripts common to our critical apparatus. The critical notes of the edition leave much to be desired. He has mostly shown correctly the additional passages in the manuscripts examined by him; but he fails, as a rule, to note the transpositions, omissions, and above all repetitions, which often are, critically, highly significant, probably again in the erroneous belief that they are negligible “scriptorial blunders”. Some of them are undoubtedly so, but not all. Likewise he has not always shown correctly the additions and omissions of the colophons, and yet he is evidently most anxious to reach the number 218, given by the Parvasamgraha. All deviations, however trivial they may seem to him, he should have scrupulously noted, as a matter of principle, because he must realize that with his utterly negligible critical apparatus—comprising only five or six manuscripts out of a total of more than three hundred manuscripts of the Ādi—it is wholly impossible for him to understand and explain the full significance of all the textual features and anomalies of the manuscripts examined by him. I will not take him to task for the numerous wrong readings which have inadvertently crept into his text, because I know, from personal experience, that it would be a physical impossibility to combine any high degree of accuracy with the pace at which he is compelled to bring out the volumes. But it is inevitable that the discovery of such inaccuracies should give rise to a sense of insecurity and suspicion in the mind of the reader in respect of those matters that he has to take from the editor on trust.

The minor deficiencies pointed out here do not, however, detract materially from the many merits of the work, from the incalculable advantage we derive from having a Southern version of an entire parvan in Devanāgarī transcript, printed in handy volumes, because the Southern manuscripts are really most inconvenient for the purposes of rapid consultation. I should be indeed very ungrateful if I did not frankly admit that Professor Sastri's edition has been of immense help to me, personally, for the study of the Southern recension, and I have no doubt that it will also help other workers in the field in future.

There remains for me the pleasant duty of recording all the encouragement and assistance I and my colleagues on the Mahābhārata Editorial Board have received from different quarters in the course of our labours in this connection.

To Shrimant Balasaheb Pant Pratinidhi, B. A., Ruler of Aundh, whose liberality made it in the first instance possible for the Institute to undertake this ambitious project—the greatest philological enterprise undertaken in India within living memory—I have to tender on behalf of myself and other people like myself interested in the study and regeneration of our great National Epic, our most sincere and cordial thanks. For the numerous marks of personal kindness with which the Chief Saheb has favoured me, in this connection, on all occasions, I have to offer him the expression of my profound gratitude. His unflagging zeal and irrepressible optimism have helped me to carry on the work in the face of heavy odds. The Chief Saheb has been pleased to enliven the dry and scientific character of the work by contributing to this edition excellent paintings of scenes selected from the Great Epic, paintings especially prepared under his expert guidance and supervision, for the purposes of this edition.

I have next to record the gratitude of the promoters of this scheme to various distinguished donors: the Imperial Government of India; the Provincial Governments of Bombay, Madras and Burma; the Governments of H. E. H. the Nizam of Hyderabad, H. H. the Maharaja of Mysore, H. H. the Gaekwad of Baroda; the Chief of Phaltan and other enlightened and patriotic Rulers and Chiefs of Indian States; the University of Bombay; and diverse other generous donors: who have all rendered valuable financial assistance to the scheme and contributed their share to that measure of success which has already been achieved. In this connection I must not forget to mention the kind offices of my old friend the Honourable Mr. Mukundarao R. Jayakar, M. A., Bar-at-law, Member of the Legislative Assembly, whose selfless interest in the success of this project has moved him to exert his influence for enlisting the sympathy and securing the help of some of the distinguished donors mentioned above.

I must next record my grateful thanks for help of various kinds I have received from my colleagues on the Mahābhārata Editorial Board, namely: Prof. S. K. Belvalkar, M. A., Ph. D., I. E. S.; Prof. A. B. Gajendragadkar, M. A., B. E. S.; Mr. P. V. Kane, M. A., LL. M.; Principal R. D. Karmarkar, M. A.; Prof. V. G. Paranjpe, M. A., LL. B., D. Litt.; Prof. V. K. Rajavade, M. A.; the late Mr. N. B. Utgikar, M. A.;

OVIII PROLEGOMENA

Prof. P. L. Vaidya, M. A., D. Litt.; Mr. V. P. Vaidya, J. P., B. A., Bar-at-law; Prof. M. Winternitz, Ph. D.; and the late Rev. Father R. Zimmermann, S. J., Ph. D. No Board of which I have been a member has worked, ever since its inception, more smoothly and harmoniously.

But I desire to make a special mention of my indebtedness to Mr. V. P. Vaidya, Bar-at-law, of Bombay, and the late Rev. Father R. Zimmermann, S. J., whose advice and ready help accompanied my labours from the time I first undertook the responsibilities of the work. The interst of my late lamented fellow-student and friend Father Zimmer- mann in this project did not flag even as he lay, in 1931, in a Nursing Home at Feldkirch, waiting prepared to meet his Maker! Nothing encouraged me more in the early stages of this arduous and fascinating work than the active and unwavering interest with which these two friends followed it.

Nepal and Kashmir in the North and Tanjore and Travancore in the South are known to contain vast treasures of unpublished and valuable Sanskrit manuscripts; and the course of Indological studies of the last two or three decades may be said to have been dominated by discoveries of outstanding importance made during that period in the three last mentioned centres. On the other hand, in regard to the large and well-stocked public and private libraries which are known to have been in existence in the country, Nepal decidedly appears not to have contributed its quota to the stock of fresh material which is now required for unravelling further the tangled skein of the history of Indian literature. Satis Chandra Vidyabhushana and Haraprasad Sastri among Indians, and Sylvain Lévi and Giuseppe Tucci among Europeans have undoubtedly done valuable pioneering work, but in view of the immense possibilities, what has been achieved thus far must be said to be tantalizingly little.

Under these circumstances, we cannot be sufficiently grateful to Rajaguru Hemaraj Pandit, C. I. E., Director of Public Instruction, Nepal, through whose good offices the doors of the rich store-house of the Nepāli material were thrown open to us—material which is all but inaccessible to Indologists—and we have been placed in a position to publish, for the first time in the history of Mahābhārata studies, collations of valuable Nepāli manuscripts. This supremely unselfish and profoundly learned patron of Sanskrit studies has really done more than merely supplying to the Institute, free of cost, collations of Nepāli manuscripts available to him in local libraries. Realizing that there were valuable manuscripts to be had outside Kathmandu, the headquarters of the Rajaguru, he caused a search to be made, at his own expense, throughout that distant outpost of Hindu culture and civilization, for old Mahābhārata manuscripts, and the find of the valuable MS. Ns, the oldest of the dated manuscripts of our critical apparatus, is the unexpected and welcome fruit of the Rajaguru's exertions in the cause of Mahābhārata research. Only those who know the difficulties in the way of obtaining any manuscript from Nepal will be in a position to appreciate fully the debt which the editor and the other members of the Mahābhārata Editorial Board, and beyond that the whole world of Indologists, owe to the Rajaguru. Sanskritists have much to hope for from the dispassionate efforts of this truly patriotic and cultured Rajaguru, who loses no opportunity of placing his immense learning and unbounded resources freely at the disposal of all serious workers in the field of Sanskrit research.

In connection with other help that has been received from extra-mural collaborators, I must put on record our special obligation to Pandit Vidhushekhara Sastri Bhattacharya of the Visvabharati, and to M. R. Ry. Rao Saheb T. Sambamurthi Rao Avl., B. A., B. L., of the Saraswathi Mahal, Tanjore. These gentlemen have been good enough to supply the Institute, for many years past, with carefully prepared collations of manuscripts which are in their charge or which were kindly procured by them, on loan, for the purpose, unselfishly supervising the work of their collation centre, at great sacrifice of their time and labour. To Professor K. Rama Pisharoti, then Principal of the Sanskrit College at Trippunittura in Cochin State, I am indebted for the collations of Malayalam manuscripts for the first two adhyāyas of this parvan.

My special thanks are due to the Managing Committees and Trustees of the following libraries and institutions for supplying me with the manuscripts required by me and allowing me to retain them as long as necessary: the Adyar Library, the Baroda Oriental Institute, Benares Sanskrit College, Mysore Oriental Library, Shri Yadugiri Yatiraj Math (Melkote, Mysore) and the India Office (London). The latter deserves special mention as the only European library I know, which sends out freely its Indian manuscripts, on loan, back to India, for the use of Indian scholars. A few manuscripts were sent to me by my kind friends Professor Sushil Kumar De, Head of the Department of Sanskrit and Bengali, Dacca University, and Professor Bhagavadatta of the Dayananda Anglo-Vedic College, Lahore, to whom I wish to thank for this kind help. I am obliged also to Sardar Kibe of Indore for the loan of a Nīlakaṇṭha manuscript. The Chief of Idappalli, Mr. Anujan Achan, Mr. Kallenkara Pisharam, all of Cochin, as also the Proprietors of the following estates in Cochin, Poomulli Mana, Avanapparambu Mana, Nareri Mana, have put me under heavy obligation by sending me freely Malayalam manuscripts in their possession, for collation, at a time when it was rather difficult for me to secure any Malayalam manuscripts at all.

I desire further to express my gratefulness to various scholars who have followed the publication of the fascicules of this volume with keen interest, periodically publishing reviews of them in the Journals of different learned Societies, reviews expressive of their interest and appreciation: to wit, Professors Banerji Sastri, Barnett, Belloni-Filippi, Charpentier, S. K. De, Edgerton, R. Fick, Jayaswal, Konow, Krishnaswami Aiyangar, Lesny, Kalidas Nag, Weller, Winternitz and others. These kind reviewers have adopted uniformly a most courteous and sympathetic tone in their reviews. Their sympathy and courtesy have always reminded me of those classic lines of Bhartṛhari:

परगुणपरमाणून्पर्वतीकृत्य नित्यं निजहृदि विकसन्त: सन्तः कियन्त:।

I must next record my thanks for the ungrudging assistance I have uniformly received from the members of the permanent staff of the Mahābhārata Department of the Institute. Mr. S. N. Tadpatrikar, M. A., Supervisor of Collations, was always by my side, helping me with useful suggestions, when I constituted the text of the Adi. Mr. Tadpatrikar has been associated with the work, in various capacities, since 1919. He had assisted my predecessor, the late Mr. Utgikar, in preparing the Tentative Edition of the Virāṭaparvan and seeing it through the press. The compiling of the critical notes (printed at the foot of the page) was entrusted by me to Messrs. B. G. Bhide and D. V.

Naravane. For the conscientious manner in which these two gentlemen have discharged their duty, I feel greatly obliged, since it is a most tedious and trying piece of work to collect the variant readings from the different collation sheets, and to arrange, in a prescribed form, according to stringent rules of sequence and enunciation, that ponderous mass of variants which is and will remain the unique feature and abiding achievement of this edition. The Śāradā codex was collated by the Head Shastri of the Mahābhārata Department, Shankar Shastri Bhilavadikar. The comparative paucity of printing mistakes in this volume is largely due to the vigilance and conscientiousness of the Collator and Reader, K. V. Krishnamurti Sharma, Sastri, of Erode (South India). These and other members of my staff have uniformly worked with exemplary zeal and untiring patience, to make a success of this edition, and I gladly take the opportunity of putting on record their loyal help and willing co-operation.

It is but right that I should also mention here that the Manager and the expert compositors of the renowned Niranaya Sagar Press have rendered ungrudgingly every assistance in carrying out the typographical arrangements which appeared to me best suited for the purposes of the work, meeting requirements that would have tried the patience and exhausted the resources of any other press in India.

Last but not least, I must express my profound gratitude to my revered Guru Geheimer Regierungsrat Professor Dr. Heinrich Lüders of the University of Berlin. What little merit there may be in the present work is due wholly to that excellent though somewhat rigorous and exacting training in philological methods which I had the benefit of receiving at his hands in the Indogermanisches Seminar, as a student in the University of Berlin. It is my firm conviction that there is no living scholar who has a deeper insight into the history of the Indian epic and the complicacies of its tradition than Geheimrat Lüders. It was, therefore, an unlucky day in the annals of Mahābhārata studies when, for lack of sympathetic co-operation and adequate financial support, he must have been compelled to abandon his epic studies, and our Great Epic lost the benefit of redaction at the hands of one of the greatest living philologists. His early Mahābhārata studies, Über die Grantharecension, Die Sage von Ṛṣyaśṛṅga and the Druckprobe have been to me like beacon lights in the perilous navigation of the Mahābhārata Ocean. May this work be to him a small recompense for the great trouble he has taken to initiate me in the mysteries of textual criticism!

August, 1933. V. S. SUKTHANKAR

Editorial Note (Fascicule 2)

In this fascicule the following additional MSS. have been used:

Newāri Version (Ñ). Ñ₁, Ñ₂, Ñ₃=Three Newāri MSS., in private possession, secured on loan, for purposes of collation, by courtesy of Rajaguru Pandit Hemraj, C. I. E., of Nepal.

Bangāli Version (B). B₅=Dacca University Library MS. No. 485, dated S'aka 1708 (ca. 1630 A. D.).

Malayālam Version (M). M₃=Pāliyam MSS. Library, Cochin, secured by courtesy of P. Anujan Achan Esq, Cochin.

Ñ₁ begins with a short praś'asti addressed to the king (ś'rīmān bhūmahendra) Jayasimharāma at whose bidding the MS. was copied; for him we have the date (Newāri) Sam. 516 (ca. A. D. 1395). — Besides the Ādiparvan, Ñ₃ contains Sauptika-Aiṣika, and Vis'oka-Strī parvans. The last folio of this bundle bears the date (Newāri) Sam. 632 (ca. A. D. 1511). — Ñ₂ agrees, as a rule, with the MSS. of the Maithili-Bangāli (V₁ B) group. Ñ₁,₃ strangely enough, show frequently features which they share with K and S. It is difficult to say, at this stage, whether these two MSS. (Ñ₁,₃) should be taken as true representatives of the Newāri version or whether they have in some way been influenced by the Kāś'miri version: in either case they are valuable MSS.

From adhyāya 26 (of the critical text, corresponding to adhyāya 30 of the vulgate), begin the collations of an old fragmentary Śāradā MS. of the Ādiparvan, written on bhūrjapatra. This unique MS., belonging to Bombay Government MS. collection (No. 150 of 1875-76), was purchased in Kaś'mīr by Buhler (vide Extra Number of JBBRAS. 1877, p. 64) and is now (along with other MSS. of the Government collection) preserved at the Institute.

August 1928.

V. S. SUKTHANKAR.

Editorial Note (Fascicule 3)

This fascicule presents, as far as I know Ādiparvan, which extended from the be- for the first time in the history of Mahā- ginning of the volume up to folio 155, is bhārata studies, collations of a Śāradā particularly fragmentary. A continuous manuscript of the Mahābhārata. The text begins only from folio 63 (adhyāya 87 unique and valuable manuscript (S¹) added of the Bombay edition). Of the first 62 here to the apparatus criticus of the folios, the extant portion contains only Ādiparvan (cf. v. 1. 26. 10) belongs to lower segments (with 10 to 15 lines of the Bombay Government Manuscripts writing on each page) of folios 24–25, 36– collection, bearing the identification no. 37, 39, 47–48, 53–57 and 61–62; while 159 of 1875–76. It was purchased for the initial 23 folios as also 15 other inter- the Government by Bühler in Kaśmīr and mediate folios (viz. folios 38, 40–46, 49–52, is cursorily mentioned by him at p. 64 and 58–60) are entirely missing.

listed at p. xi of his Detailed Report of a Such is the lamentable condition of the Tour in search of Sanskrit MSS. made in only genuine extant representative of the Kas’mīr, Rajputana and Central India, old Kaśmīrī version of the Ādiparvan, a which was printed as Extra Number of the version which must be pronounced to be Journal of the Bombay Branch of the Royal far superior not only to the vulgate but Asiatic Society for 1877. also to the Southern text and far older than either of them. Under these circum- stances it must be considered a piece of singularly good fortune that there has been preserved at least one complete Devanāgarī manuscript of the Ādiparvan that may, without hesitation, be treated as a mode- rately trustworthy, though comparatively modern (? end of the 18th century) and incorrect transcript of a Śāradā exemplar. This manuscript is the India Office codex No. 2137, our K₁.

This birch-bark (bhūrjapatra) codex Even the outward form and get-up of measures 12 in. by 9½ in. and comprises the India Office codex are suggestive of 339 folios. The characters are Old Śāradā, Kaśmīrī origin: the lines of writing, as in of perhaps the sixteenth or seventeenth Śāradā Bhūrja manuscripts, run parallel to century. The lines of writing, as is usual the narrow side of the folio; the signatures in Śāradā manuscripts, run parallel to the in the margin are like those found in narrow side of the leaf. There are on an Kaśmīrī books; the dimensions of the average 24 lines on a page, and 36 akṣaras glossy paper are 16 in. by 9 in. The (i. e. a little over a śloka) in a line. A numerous clerical errors which disfigure page, therefore, contains on an average every page betray the writer to be a pro- 26 (Anuṣṭubh) stanzas. Each folio bears fessional scribe with a modicum of the on its reverse side, in the left-hand margin, knowledge of Sanskrit, frequently and a cipher representing the serial number of easily misled by the deceptive similarity the folio and a signature indicating the title of the work and of the parvan.

The manuscript, which is unfortunately incomplete and fragmentary, must have originally contained at least the first three parvans (Ādi, Sabhā and Araṇya), written, as far as one can judge, by the same hand. The extant portion contains the Sabhā in its entirety, but only fragments of the other two parvans, the beginning of Ādi and the end of Araṇya being lost. The

between certain letters of the S'āradā and Devanāgarī alphabets: he commits such blunders as writing ड्ढ for त्थ and म्तुम्: for सत्तमः. The India Office codex must be a tran-script not merely of a S'āradā codex, but, as a careful comparison with S¹ would show, of an exemplar very closely allied to our S'āradā codex. The affinity between these two codices not merely relates to the general division into and numbering of adhyāyas, and additions and omissions of stanzas, but extends to innumerable minor details (where frequently the two manu-scripts stand together against the rest), and, perhaps, even to a few clerical errors (compare, for instance, 1. 29. 4 S¹K¹ अरा श्र्रेणाº for अरात्रश्रेणाº; 1. 31. 6 S¹K¹ वंजवकः for पिंजरकः).

As an illustration of adhyāya division, it may be pointed out that adhyāyas 42–44 and 46 of the critical text (corresponding to adhyāyas 46–48 and 50 of the vulgate) are numbered in S¹ and K¹ as 54–56 and 59 respectively, and are so numbered in no other manuscript hitherto collated. Further, after the third stanza of adhyāya 40, both manuscripts (S¹K¹) interpolate (the figure) 51, S¹ marginally inserting at that place an additional colophon: इत्यादिपर्वैकपञ्चाशत्तमोऽध्यायः.

Of textual additions, the most notable is that of a short adhyāya of about 24 stanzas, appended at the very end of the Ādiparvan, which, with the exception of S¹K¹, is found only in Ko and K⁴.

The following list of concordant readings of S¹ and K¹, selected at random, will further illustrate how close the affinity between these two codices really is: 1. 27. 15 S¹K¹ पुरंदः: the rest शत्तक्रतुः. 1. 28. 24 S¹K¹ ततः: the rest नदीः. 1. 29. 4 S¹K¹ अरा श्र्रेणाº: text अरात्रश्रेणाº. 1. 30. 7 S¹K¹ प्रदिष्टतम् (K¹ ºतद्): MSS.

प्रतिगृहताम् (text), प्रगृह्याº, संगृह्याº आगृह्याº etc. 1. 31. 6 S¹K¹ वंजवकः: MSS. पिंजरकः (text), पिंजरकः etc. 1. 37. 25 S¹K¹ रक्षितव्यः (K¹ ºव्यं) पिता लिपिः: text रक्षितव्याः प्रजास्तथा. 1. 42. 7 S¹K¹ नेतरो रोचयन्त्यां: text न भरेंर्व्य च ग्रामहमत्. 1. 44. 2 S¹K¹ भयात्: MSS. तदा (text), तथा, ततो etc. 1. 45. 19 S¹K¹ तं (K¹ य) घोषिदितं: the rest नरधिषं. These represent a very low percentage of the instances that might have been collected. I shall now cite some readings, also selected at random, which S¹ shares with the entire K group, S¹K standing to-gether against all other manuscripts: 1. 28. 18 S¹K (K² marg.) श्वसनेन: the rest कथनेन. 1. 29. 1 S¹K ततो जांबूनदो भूत्वा: the rest जंबूवनभूतो भूत्वा. 1. 32. 3 S¹K दशार्गे: the rest गोकुले. 1. 32. 12 S¹K बा: the rest मे. 1. 33. 20 S¹K रुष्यते (K³ रिष्यते): the rest वाले. 1. 36. 21 S¹K दांर्गने: the rest तिमदवेजा. 1. 38. 14 S¹K कुरादग्रतः: the rest समादितं. These two concordances (taken in com-bination with those already cited in the Foreword to fascicule I, p. IV) are, in my opinion, sufficient, if not to establish, at least to suggest, three things relating to the K manuscripts: firstly, that these manuscripts belong together, forming a group distinct from the ordinary Deva-nāgarī manuscripts; secondly, that they are affiliated to the Kaśmīrī version, as represented by our S'āradā codex; thirdly and lastly, that of them K¹ is a transcript of a S'āradā manuscript very closely allied to our S'āradā codex. The value of these conclusions will be

readily admitted by any one who realizes the importance of the Kāśmīri version for the reconstruction of the text of the Ādiparvan (if, indeed, not of the whole Mahābhārata), and who bears in mind the extreme scarcity and inaccessibility of old Śāradā manuscripts.

As a corollary to these conclusions it follows that the India Office codex (K₁) could, with due caution, be tentatively utilized to supplement the Śāradā codex, where there is a lacuna in the latter, especially when the reading of K₁ is supported by other manuscripts of the K group. That such a procedure would not be wholly unwarranted even when the support from other MSS. of the K group is lacking, may be seen from the following illustration.

The constituted text of the Parvasamgraha gives (1. 2. 96) the number of ślokas in the Ādiparvan as 7984, as against 8884 of the vulgate text. In the Śāradā codex the corresponding pages are lost. The variae lectiones for this passage are more numerous than one could have wished. For the digit representing the thousands alone, the choice lies between seven, eight, nine and ten! The majority of manuscripts seems to favour the figure eight. The figure for the actual extent of the printed texts of the Ādiparvan has been variously computed: C. V. Vaidya (The Mahābhārata, Appendix) 8466 Calcutta edition 8479 Lele’s Mbh. edition with Marāṭhī translation 8621 Kumbhakonam edition 10889. In the presence of these figures, doubt may legitimately be entertained regarding the correctness of the reading adopted in the critical text, representing so low a figure as 7984, especially as the text reading appears to be based on that of an extremely small group of manuscripts, K₁. M₁. S. The presence of K₁ in this group raises, however, the presumption that it may represent the reading of the Śāradā version. This presumption is unexpectedly verified by our Śāradā codex; for, although the Parvasaṁgraha is missing in it, the stanza is repeated at the very end of the Ādiparvan in S¹ (and K¹), agreeing almost verbatim with the corresponding stanza of my Parvasaṁgraha. Here is a tran-script of the stanza as found in the colophon of the Śāradā codex: सप्त श्लोकसहस्राणि तथा नव शतानि च ¹ श्लोकाश्च चतुरासीति ग्रन्थो ह्यो (sic) महात्मना ॥

The figure given by me, 7984, may, therefore, without any hesitation, be taken to represent the extent of the Kāśmīri version at least of the Ādiparvan, which, it may be added, is therefore the shortest version of the parvan hitherto known.

In passing I may point out that even the variations mentioned above show, if indeed the critical apparatus has not done so in sufficiency, that it would be a grave mistake to regard the Parvasaṁgraha as the one immutable factor in the chequered history of the Mahābhārata text. There can, I think, be no doubt that the text of this adhyāya also has been tampered with and designedly altered, from time to time, in various ways, in order to make it harmonize with the inflated versions of a later epoch.

March 1929. V. S. SUKTHANKAR.

Editorial Note (Fascicule 4)

The section of the Ādiparvan published in this fascicule (4) is interesting, from the view-point of the textual critic, mainly for two reasons: firstly, because of the far-reaching divergence—met with for the first time in the Ādiparvan—between the two recensions as regards the sequence of adhyāyas or adhyāya-groups; and secondly, because of the stupendous additions—there cannot be much doubt that they are additions—found made in the Southern recension to the well-known Śakuntalā episode.

How different the arrangement of the subject matter of adhyāyas 57–90 is in the Northern and Southern recensions may be seen at a glance from the Table of Contents (viṣayānukramaṇī) of the Kumbhakonam edition of the Mahābhārata, which includes a convenient concordance of the adhyāyas in the Bombay and the Kumbhakonam editions. The differences between the divergent recensions may be summarized thus. In the first place, the Śakuntalā and the Yayāti episodes change places with each other: the Northern order is Śakuntalā-Yayāti, the Southern Yayāti-Śakuntalā. Bharata, the son of Śakuntalā and eponymous ancestor of the Pāṇḍavas, lived long after the time of Yayāti, a very early king, according to the chronicles, only the tenth from the Prajāpati. The Southern arrangement, we observe at once, presents an orderly sequence. In combination with the setting of these two episodes in their correct perspective, we may consider the Southern dissection of the genealogical adhyāya (89) into two sections, widely separated from each other, of which the first section (dealing with the genealogy from Pūru to Bharata), formed into a separate short adhyāya of about 19 stanzas, is placed, in the Southern recension, between its Yāyāta and Śakuntala, while the second section (from Bharata to S’āmtanu) is incorporated bodily in the final adhyāya of the Śakuntala. We thus get in the Southern recension an altogether better sequence of the subject matter in adhyāyas 57–89: first, the story of Yayāti; then the genealogy from Yayāti’s son Pūru to Bharata; and finally, the story of Bharata (or the Śakuntala), including the genealogy from Bharata to S’āmtanu. As against this we have in the Northern recension (which the constituted text follows): first, the Śakuntala (or the story of Bharata); then the story of Yayāti; and finally, the genealogy (in one stretch) from Yayāti’s son Pūru to S’āmtanu. Logically, therefore, the Southern arrangement of the whole of this section is much superior to that of the rival recension; only it looks, in comparison with the other, a trifle artificial, as though it were an afterthought, conceived and carried out by a diaskeuast.

It may be incidentally mentioned that in the constituted text (as in the Northern recension) there is a palpable hiatus between adhyāyas 69 and 70. The thread of the narrative dropped at the end of adhyāya 69 seems to be resumed at adhyāya 89 (or, strictly speaking, at stanza 17 of that adhyāya), after skipping the entire Yayāti episode. The situation is this. Adhyāya 69, which is the final adhyāya of the Śakuntala, ends with the

remark of Vais'ampāyana that he will now enumerate the names of the more impor- tant kings among the descendants of Bharata. Instead of a list of the succes- sors of Bharata, there follows in our text (as in the Northern recension) the story of Yayāti, of which the initial adhyāya contains, besides the argument of the fable, the genealogy of the Solar dynasty from the Prajāpati to Yayāti, but not a word about the descendants of Bharata. Notwithstanding that the Southern re- cension transposes the S'ākuntala and the Yayāta, this hiatus is not removed: owing to the circumstance that this recension further dissects and dis- places adhyāya 89, with the result that the portion of this adhyāya which does contain a list of the descendants of Bha- rata stands now just before the above- mentioned remark of Vais'ampāyana, in- stead of standing, as it should, after it. Thus Vais'ampāyana’s connecting remark is again left hanging in the air. The trans- position in the Southern recension, then, was not made with a view to remedying this defect. The context can be restored, as far as I can judge, only by deleting, in the Northern recension, the Yayāti episode (which, it may be noted, finds no mention in the Parvasaṃgraha, not in any of the numerous versions or manuscripts examined and which, moreover, is but a replica of the story as narrated in the Matsyapurāṇa) together with the first 16 stanzas of adhyā- ya 89 (containing the genealogy from Pūru to Bharata), which latter are in a way the connecting link between the Yāyāta and the S'ākuntala. There remains now only one final trans- position to consider, that of adhyāya 90 of our text. This adhyāya, which is in prose and—be it noted—has a separate phalaśruti, contains again a complete genealogy of the Solar dynasty from the Prajāpati to the Pāṇḍavas, or rather to the sons of Janamejaya, the grandson of Arjuna. This genealogy, which differs in part from the metrical genealogy (adhyāya 89) and which occurs in our text after the S'ākuntala and the Yayāta, is placed in the Southern recension between (adhyāya 89) and adhyāya, in other words, occurs in the Southern recension before, in the Northern recension after, what may be termed the Section of Purāṇic Genealogy (adhyāyas 57–89). The phalaśruti at the end of the adhyāya strongly suggests that the adhyā- ya was borrowed from an older source and incor- porated en bloc in our text at the time of its last redaction or at some subsequent stage of its development. As regards the additions to the S'akun- talā episode, their extent may be realized by comparing the lengths of this episode in the Bombay and the Kumbhakonam editions. The former contains only about 325 stanzas, while the latter has over 590. The constituted text is of about the same length as the Bombay text, only a trifle shorter. The Southern text is, therefore, nearly twice as long as the constituted text. Whichever version is the original one, the difference between them is as- tounding. When there is discrepancy, as for instance in this case, between the two recensions, it is difficult, as a rule, to give a strict proof of the originality of either version. It is, perhaps, as easy to conceive that one recension has interpolated the addi- tional lines as that the other recension has accidentally (or even intentionally) omitted the lacking lines. The probability lies, in my opinion, always in favour of the shorter version; but, it must be admit- ted, it is in general no more than a proba-

bility. Instances do occur, however, where the intrinsic evidence is so strong as to be decisive, determining in favour of the shorter version. Of this character are two clear instances in the present fascicule where a Southern editor, in the interests of morality and piety, has, out of misdirected zeal, carefully recast two passages of the original text, which dis- close the lax sexual life and the erratic marital relations of some epic characters and which must have sorely outraged his sense of moral rectitude. One of the instan- ces occurs in the Yayāti episode. Accor- ding to the Northern recension S'armiṣṭhā was in reality no more than a concubine of Yayāti; and their sons Druhyu, Anu and Pūru were born out of wedlock. The entire course of the narrative implies clearly such a state of things. Yet we find in the Southern recension an additional passage (807*) stating that the marriage of Yayāti and S'armiṣṭhā was celebrated with pomp and ceremony, in the presence of counsellors, priests, ācāryas and domestic chaplains, with distribution of profuse largesse (dākṣiṇā) to Brahmins! The other instance (610*) occurs in the S'akuntalā episode. Here the interpolator startles us by making Duḥṣanta, against the unanimous evidence of tradition, summon his domestic chaplain in the her- mitage of Kaṇva to solemnize his marriage with S'akuntalā “in order that his son of great lustre may not be born without ceremonies”!

These little retouchings in the Southern recension are, however, wholly eclipsed by a wild extravaganza in the Grantha version (cf. Kumbh. ed. vol. 1, pp. 110 ff. = our App. I, Nos. 36–39). This fantastic inter- polation of about 90 stanzas describes, among other things, with circumstantial detail, the marriage ceremony of Parāśara and Satyavatī. At this ceremony the ancestors of both the bride and the bride- groom are invoked, all the details of a regular Hindu marital rite (of mediaeval times) are minutely observed and the marriage is solemnized in the Naimiṣa forest, again with the distribution of profuse largesse (dākṣiṇā) to Brahmins! It will, I think, be readily conceded that in the three instances just cited it is not a mere question of an elusive factor of ambiguous character that may be inter- preted on the one hand as an interpolation in one recension or on the other hand as an omission in the other, according to the view-point or predilection of the critic. To refute the charge of interpolation in the Southern recension, one must establish that that version of the story alone is right and the rest of the entire Indian tradition is wrong, which is obviously an untenable proposition. There is a further implication involved in the assumption of the authenticity of the Southern version. Not only would the Northern version in that case be defective, it would be corrupt in the extreme and calumnious to boot, nay even blasphemous! Could such a charge against the Northern recension be conceivably substantiated? Certainly not. Then the only alternative is to con- clude that in these instances at any rate the epic text has in Southern Indīa been surreptitiously altered by some over pious Vyasād of the South. This is, in other words, a palpable instance of a literary fraud, albeit that it is a pia fraus. However laudable the motives of the interpolator may be and however venal such transgression may appear from the purely human stand-point, this propensity to alter an inherited text, perverting its

sense, is obviously fatal to any claim of superiority that might be set up on behalf of the Southern recension in questions concerning textual purity and integrity. It puts this recension at once on its defence whenever it differs from the Northern. Returning for a moment to the question of the puerile additions to the S'akuntalā episode in the Southern recension, they appear now in a somewhat different light. To the reckless editor who does not hesitate to introduce changes into a text so as to alter its purport, it would be the most natural thing in the world to add small details here and there, embellishing and amplifying the original : that would be merely a gentle and lowly service for the greater glory of God. If a few more unequivocal instances of this character could be found, we should be justified in concluding that even after its final fixation in the North our epic was subjected in the South to a systematic diaskeuasis, during which the text was altered, amplified and even expurgated on a large scale. It is fair to add that in all probability the Northern recension likewise contains some flagrant additions and alterations. The vulgate text contains, for instance, a lengthy, superfluous adhyāya towards the end of the Ādiparvan which is missing in the Southern recension and which must, there- fore, be discarded as a Northern interpola- tion. But that only means that we must build up the critical text on both recensions, using each to control and correct the other. Only that portion of the text which is documented by both recensions may be considered as wholly certain and authentic; the rest is doubtful, in varying degrees.

March 1930. V. S. SUKTHANKAR.

Editorial Note (Fascicule 5)

The section of the Ādiparvan dealt with in this fascicule (5) corresponds roughly to adhy. 96–161 of the Vulgate text, and includes among other things the early life history of the Pāṇḍavas : the circumstances of their mysterious birth in a forest retreat, their life at the Kaurava court in Hāstina- pura, their subsequent exciting experiences and perilous adventures, ending with their encounters with two Rākṣasas, Hiḍimba and Baka.

The constituted text, which closely follows the Kāśmīrī version, is here also con- siderably shorter than the Vulgate, not to speak of the Southern recension, and con- tains some notable omissions. Important among these are the following passages, cited here according to the Bombay text: adhyāya 116 {128. 34 to end {129. 1-35 {138. 6-62 {139 {140 {141. 1-19 149

Adhyāya 116 of the Vulgate text relates the story of the birth of Duhśalā. The epic narrated in great detail the circum- stances of the birth of the hundred sons of Dhṛtarāṣṭra, in still greater detail the circumstances of the birth of each of the five sons of Pāṇḍu; but Duhśalā, the only daughter of Dhṛtarāṣṭra, was treated with scant courtesy. This short adhyāya of 18 stanzas, which, following the Kāśmīrī ver- sion, I have omitted, made good the defi- ciency. It naïvely describes how Duhśalā was born from a superfluous bit of flesh left over while Vyāsa was engaged in dividing into one hundred parts the hard lump of flesh brought forth by the pregnant Gān- dhārī. This poor doggerel is clearly an afterthought, concocted by some pedantic epigoni tormented with the question how Gāndhārī could have had a hundred sons and a daughter when the great sage Vyāsa had said only that she should have a hundred sons:

गान्धारराजदुहिता शतपुत्रैति चानघ । उक्ता महर्षिणा तेन व्यासेनमिततेजसा ॥ The loss to the Vulgate, I cannot but think, is a gain to the epic.

The next passage, consisting of about 31 stanzas of adhy. 128, and 35 stanzas of adhy. 129, gives a somewhat confused ac- count of the various unsuccessful attempts made by Duryodhana to kill Bhīma, con- taining the suspicious story of Bhīma's adventures in the Serpent World. No one who compares the constituted text (based on the Kāśmīrī version) with the Vulgate and the Southern recension can doubt that the Kāśmīrī version alone presents the correct text, while the others are secondary and conflated.

The long interpolation of 56 stanzas in adhy. 138 gives an inflated account of the defeat and capture of Drupada by the Pāṇḍavas after the Kauravas had failed in their attempt. The description was evi- dently spun out expressly with a view to glorifying the popular heroes Arjuna and Bhīma at the expense of the much maligned Duryodhana and the other Kuru princes. The older version disposes of the battle in two lines (which, taking everything into account, I consider a very adequate treat-

ment), and divides the glory impartially between all the pupils of Droṇa alike!

The omission of the whole of adhy. 139, which contains only some needless repetition besides minor absurdities and contradictions, would not have called forth any comment from me but for the fact that with its omission disappears the only reference in the epic to the alleged installation of Yudhiṣṭhira as heir apparent to the throne of Hāstinapura by Dhṛtarāṣṭra. The Kāśmīri version, which omits the entire adhy. containing this reference, fully justifies the indignant outburst of Holzmann ( _Das Mahābhārata_ , Bd. 2, p. 33): “Geradezu Fälschung ist es, wenn 1, 139, 1 = 5517 behauptet wird, der blinde Dhṛtarāṣṭra habe mit Uebergehung seiner eigenen Söhne den Yudhiṣṭhira zum Kronprinzen ( _yuvarāja_ ) ausrufen lassen!” He is probably also right when he adds: “Ich bin überzeugt, dass von dieser ganzen Kindergeschichte keine Zeile alt und ächt ist, und dass wir hier keine Ueberarbeitung, sondern Neudichtung vor uns haben”. But so much of the “Kindergeschichte” as is now left in the critical edition has insidiously filtered its way into all our MSS. and there is no way of dislodging it by any known canon of textual criticism.

Many readers will no doubt miss the notorious Kaṇīkanīti (adhy. 140 of the Bombay text), but its spurious character is borne out by its omission not only in Kṣemendra’s Bhāratamañjarī and in the Telugu and Javanese adaptations of the Mahābhārata but also in the scholium of Devabodha, who has not commented on a single word of the 93 stanzas comprising this adhy., although both Nīlakaṇṭha and Arjunamiśra have written lengthy notes on it in their respective scholia. This delectable piece of political philosophy or political wisdom is, moreover, only a replica (naturally with many additions, omissions, and variant readings) of the advice given by Bhāradvāja (which appears to have been a gotra-name of this very Kaṇīka or Kaṇīkā) to Śatruñjaya and duly communicated to Yudhiṣṭhira by Bhīṣma in the Śānti; it will reappear, therefore, in a slightly different garb in its proper place in due time.

The last important omission is that of another short adhy. (149 of the Bombay text) of 15 stanzas, which relates how the Pāṇḍavas were seeking a ford on the Ganges when suddenly a secret agent of Vidura appears on the scene and conducts them to a little boat “as swift as mind or wind”, which safely carries them across the river. The position of this adhy. in the Vulgate text is evidently wrong; for in the very next adhy. (150 of the Bombay text) there is a reference to the Pāṇḍavas again crossing the Ganges. They would surely not have crossed the river twice in such a short time. The correct place of this (interpolated) adhy. is after stanza 19 of adhy. 150, where the Southern recension places it. Some Northern version had evidently copied the adhyāya in the first place from a Southern exemplar and inserted it at a wrong place, as often has happened in the case of these interpolations—fortunately so, because these displacements are frequently the only surviving indications of these unauthentic accretions.

Here I may draw attention to a difficult text-critical problem which arises in connection with these passages. They have been rejected by me mainly on the evidence of the Kāśmīri version, because they are lacking practically only in this version. The question naturally arises: are they omissions (in the version in which they are missing) or are they additions (in the versions in which they are found)? The

intrinsic evidence is, in my opinion, strongly against their originality. The least that can be said about them is that they are utterly superfluous; the poem as a whole is decidedly better without them, for their only effect is to lengthen and weaken the text. Moreover, no good reason can be adduced why the passages should have been omitted in the Kāśmiri text. There is nothing objectionable in them; they contain instructive and moderately entertaining matter of an innocuous character, matter quite in harmony with the general tenor of the epic. The presumption of unauthen- ticity, on the other hand, is confirmed by the fact that they are found in different versions at different points of the text, frequently also with partial repetition of the preceding matter. Professor Jacobi has gone into the question of such repetitions in the Rāmāyaṇa text and has clearly demonstrated that after a lengthy inter- polation some portion of the original text preceding the interpolation was repeated verbatim for the purpose of rehabilitating the context disturbed by the intrusion of extraneous matter, "damit die Hörer wieder irgendwie in den Zusammenhang hinein- kämen" (Das Rāmāyaṇa, p. 34). Our repetitions are to be judged similarly; they are the fingerposts at interpolations. One lengthy and important passage among those cited, as has been observed, is actually missing in several ancient testimonia (cf. p. 574 below).

Here therefore we are confronted by a very difficult case where the evidence pro et contra of documentary and intrinsic pro- bability is equally or almost equally balanced. Now it would not do to form some a priori hypothesis as to the interrela- tionship of the versions and fix the text in terms of some preconceived notion about it. The study of the documents themselves must teach us what their interrelationship is. And they unmistakably indicate that this interrelationship is of a very complex character. In fact I am now fully persu- aded that with the epic text as preserved in the extant Mahābhārata MSS. we stand at the wrong end of a long chain of succes- sive syntheses of divergent texts carried out in a haphazard fashion through cen- turies of diskeuastic activities; and that with the possible exception of the Kāśmiri version all other versions are indiscriminately conflated. Now it is evident that the genetic method cannot in strictness be applied to conflated MSS.; for in these cases it is extremaly difficult to dis- entangle completely by means of purely objective criteria their intricate mutual in- terrelationships. The documentary evidence is supremely important, but the results arrived at from a consideration of the documentary probability must be further tested in the light of intrinsic probability. No part of the text can be considered really exempt from the latter scrutiny when we are dealing with a carelessly guarded fluid text such as we have for the Mahā- bhārata; that the text was fluid and carelessly guarded is now incontestably demonstrated by the hundreds of variants which fill every page of this edition. There was every inducement and opportunity for interpolation and conflation. The dis- covery of even such sporadic contamination between "independent" versions does not destroy the value of our division of the manuscript material into recensions and ver- sions, but only complicates its interpretation.

V. S. SUKTHANKAR.

March 1931.