I start with Sukthankar’s master chart of the recensional history for the epic as a whole:
- Vyāsa’s Bhārata
- Ur-Mahābhārata
- N
- υ
- Śārada
- K
- γ
- ε
- Nepalī Maithili Bengali
- Devanāgarī
- ε
- υ
- S
- σ
- Telugu Grantha
- Malayalam
- σ
The Vyāsa phase of the epic, the so called Jaya Bhārata, began perhaps in an oral tradition, by consensus in the Kuru area, and most likely in the kṣatriya circles, as a lay about war for land and territory, perhaps based on the Ten King Battle of the Ṛgveda (Witzel 2006: 21-24). By the Gṛhya Sūtra period—considerably later than the Śrauta Sūtra period, as Oldenberg has shown, thus perhaps 500-300 BCE11—a Mahābhārata has come into existence, the Gṛhya Sūtra passages linking it with the primary, inner circle of redactors, Sumantu, Jaimini, Vaiśmpāyana, and Paila (omitting Śuka, however).12 Perhaps this marks the first “possession” of the epic by the Brahmans, that of the inner frame, a process seen much more deepened in the outer frame, unfolding as a discourse in the sadas of a Śrauta ritual of the Sattra type, with Vyāsa himself present in the sadas and claiming for itself subsequently the status of the Fifth Veda. It is possible that the śrauta device of the outer frame reflects the real-life setting of Hiltebeitel’s intergenerational Brahman committee, engaged in srauta rituals and redaction of the epic at the same time in one of the new reformist Brahman kingdoms, like the Śuṅga, its Brahman king Puṣyamitra performing two aśvamedhas and committed to the promotion of Śrautism.13 Plausible links, as we will see below, can be surmised, between the first group of Brahmans of this study and this original redaction. There is general consensus that the epic passes into literacy by this stage, by 300-100 BCE when the Brāhmī script had taken shape in North India, providing the Sanskrit sound system a syllabary, devised, as its separation of the vowel and consonant sounds into two classes shows, by the Vedic oral tradition and its svādhyāya institutions.14
With the NR and SR phases, we are on firmer ground, what Sukthankar characterizes as the “incontrovertible fact” (1933: [I] xxxi) about the early history of the epic: they are the two broad and distinct recensions of the epic, each with further different versions later in regional languages. The NR text first breaks into two large families, the ν-and the γ-family texts, the former in the Kuru realm, the area of the origin of the epic, and the latter, an inflated version in the Magadha realm, in the eastern parts of North India, in Bihar, Nepal and Bengal. The shorter ν–family of the Kuru-Pāñcāla area gives us the Śārada text, the basis of the Poona CE.
We have no information in Sukthankar about how exactly the SR rises or is found in the peninsular region in terms of a human agency or other irreducible correlates— script, the physical form of the manuscript. All the same, however, from the evidence of his manuscripts, Sukthankar is able to affirm that its appearance can be dated to a “primitive” ([1927]1933 [I]: Forward vi) moment in the textual history of the epic. Noticing the concord in the Ādiparvan between the Kāṣmīrī version of the NR and the SR texts, Sukthankar notes, “Since I have not been able to discover any traces of “secondary interrelationship” between archetypes K [NR] and S [SR], I consider the agreement between these two archetypes as “primitive,” that is depending upon their primitive connection through the Ur-Mahābhārata” (Sukthankar’s quote marks; my parenthetical gloss)–a concord, further, he sees to be of “supreme importance for the reconstruction of the text” ([1927] 1933:[I] Forward vi-vii).15 We should note that the “primitive” accord between the NR and SR texts, so phrased in his 1927 Forward by Sukthankar, becomes the “impressive fact” (1933 [I]: lxxiv) of the recensional history of the epic in the 1933 Prolegomena, in view of the antipodal locations of the two texts, the NR in Kashmir and the SR in Kerala, and we should further note that the picture of the Brahman migration presented below adequately explains this anomaly.
The SR, having thus risen at an early moment in the history of the epic, differentiates essentially into two versions, the shorter Malayalam text, that came to the Poona editors from the Malabar region of Kerala and the longer, inflated Grantha-Telugu version, the latter forming from an interaction between Sukthankar’s theoretical σ text and the resident Σ-SR text, the σ text being an NR version (not indicated to be so in Sukthankar’s chart but made abundantly clear in his Prolegomena to the Ādiparvan) and coming to Poona from the Tanjavur area of Tamil Nadu.
Thus, Sukthankar noted, “all textual criticism of the Mahābhārata begins with this incontrovertible fact that the text of the Great Epic has come down to us in two divergent forms, a Northern and a Southern recension, texts typical of the Āryāvarta and Dakṣināpatha” (xxxi). Yet this is an issue scarcely addressed in the Mahābhārata scholarship since the publication of the CE (1933-1970). The 18-parvan division of the NR increases to 24 in the SR, the SR being almost a Virgilian response to the Homeric NR, characterized by “precision, schematization, and thoroughly practical outlook” (xxxvi; Sukthankar’s italics) compared to the NR version which is “distinctly vague, unsystematic, and sometimes even inconsequent, more like a story naively told, as we find in actual experience” (xxxvi; Sukthankar’s italics). Sukthankar noted further that “there persists throughout, between the recensions, a distinct and undeniable family resemblance, and there cannot be the slightest doubt that both spring from a common source, albeit a distant and somewhat nebulous source” (xxxvi), remarking in addition that “even in its early phases the Mahābhārata text tradition must have been not uniform and singular, but multiple and polygenous” (Forward: i).
That is, the longer and ornate SR text rises at an early moment in the textual history of the epic—a finding echoed further by Belvalkar (1947: lxiv)16, the other CE editor with an equal breadth of experience with the Mahābhārata manuscripts. However, the inflation does not seem to have been the inherent result of the dynamics of an oral tradition;17 that is, the two recensions do not constitute what are called multiforms in oral theory—a virtual impossibility, considering the epic’s eventual extent. Rather, as Sukthankar noted for the Ādiparvan of the SR version but true for the entire recension, “the excess is due to additions, large and small, distributed almost entirely throughout the [Ādi] parvan” (xxxv). Sukthankar also noted omissions, passages found in the NR texts but not in the SR tradition. Additions and omissions: surely we are by now in a literate world (how can an “omission” occur in a dynamic oral tradition?). In other words, we have a transcript laid out, read and episodes and elaborations added on (or dropped). It would seem thus that the SR text clearly rises as a make-over of the NR text, the Southern redactors creating a sentimental version of a naïve text, by adding passages where they felt necessary and dropping them elsewhere. A famous example shows how this process of addition probably worked in actual practice, providing evidence that complements Hiltebeitel’s (2006) finding,18 solely derived from structural considerations of the” dips” between the main frames that make up the epic, both revealing the subterranean dynamics of the formation of the *Pūrvaśikhā SR text. An insertion of 1612 verses into the SR occurs between adhyāyas 34 and 35 of the CE of the Sabhāparvan, not found in the Śārada version, nor the NR as a whole, and thus relegated in the CE as an appendix (II: [Appendix 1, #121]: 386-422). Edgerton, (1944 [II]: xxx), the editor of the Sabhāparvan for the CE, notes that “it is the longest single insertion…occupy[ing] a full hundred pages of [P.P.S] Sastri’s text…seven adhyāyas…a glorification of Kṛṣṇa put in the mouth of Bhīṣma[.] It is not found in N[orthern] MSS” (My parentheses). In the peroration, Bhīṣma justifies the fitness of Kṛṣṇa to be the Guest King to be honoured at Yudhiṣṭira’s Rājasūya, at the sabha of the Sabhāparvan.
P.L.Vaidya (1969: [I] xlviii) shows that the entire discourse is fabricated from the Harivaṃśa, mainly from adhyāyas 38, 41, 42. One aspect of the Harivaṃśa that Vaidya emphasizes is its dual nature, first as an “organic” part of the great epic, justifying the attribution of “śata-sāhasrī Saṃhitā” (100,000-verse epic) to it, and second, as its khilā, a “supplement”. Yet we find material from the supplement forming sections of the main epic, in the SR, forcing us to conclude that the redactor of the SR must have had the entire epic before him, and that he knew the whole of the epic, the main body of 18 parvans and its supplements, to find or remember a discourse from the supplement suitable to be inserted into an earlier section of the main body of the text, no matter that this introduces in the process awkward repetitions, what Edgerton (xxx) calls “internal duplications” as with Sahadeva’s threats.19
My argument in the rest of the paper is predicated to this incontrovertible fact, that at an early, decidedly “primitive” moment in its textual history, the epic is already found to be present in the peninsular region, the logic of chronology demanding that this be very likely a *Śārada text, 75,000 verses long in its modern CE and nearly the same length at this time, providing the NR text-template for its SR makeover. I will attempt to account for its rise in terms first of its plausible redactors and then the paleography needed for the transcription of the Sanskrit of the epic in an area already widely literate at this time with a script adopted to Tamil phonology, the Tamil Brāhmī syllabary, created by Jains ca. 250 BCE.