1.1 Internal and External historical evidence

[Full title: History and Structure of Svacchanda Tantra; (1) Internal and External historical evidence]

1.1 This dissertation has as its subject the text published under the name of the Svacchandatantra, with the commentary Uddyotah, by Kshemaraja. 1 Svacchanda-bhairava, who forms the ritual subject of the Svacchandatantram, and in his different forms, its narrator and supreme deity, has served as the basis for the title of this text. Manuscript lists and catalogues record the same text under variant names: "Svacchanda-bhairava-tantram, "Svacchanda-bhairava-maha-tantram, " and "Svacchanda-lalitabhairava-tantram. "2 However, the "Svatantratantram" and 1 Pandit Madhusudan Kaul Shastri, ed., The Swacchanda-tantra with Commentary by Kshemaraja, 7 vols., Kashmir Series of Texts and Studies 31, 38, 44, 48, 51, 53, 56 (Bombay: Nirnaya Sagar Press, 1921-1935). (All unmarked text references are to this edition by book (patalam) and page or verse number. The text in this edition contains 3648.5 verses, almost all slokah.) Now reprinted in two volumes with notes, indices and new introduction by Pt. Vrajavallabha Dwivedi, ed., The Svacchandantantram with Commentary 'Udyota' by Ksemara jacarya, Parimal Sanskrit Series 16 (Delhi: Parimal Publications, 1985). 2 For Svacchanda-bhairava-tantram, v., for example, Rama Sankar Tripathi, Descriptive Catalogue of Samskrit Manuscripts in Gakwada Library, Bharat Kala Bhavana Library and Samskrit Maha-Vidyalaya Library, Banaras Hindu University Sanskrit Series 6 (Banaras: Banaras Hindu University, 1971), mss.no.C 4400, p.776; for 1

2 "Svacchanda-paddhatih, "noted in manuscript catalogues represent independent works belonging to a separate tradition not based on the Svacchandatantram. 1 In Nepal and northern India, catalogues record numerous manuscripts of Svacchandatantram written in devanagari, sarada, and newari scripts. Moreover, the devanagari and sarada manuscripts appear to contain both the text and the commentary of Kshemaraja. Presuming a sarada autograph of Kshemaraja, it is probable that sarada archetypes underlie the devanagari manuscripts. The newari manuscripts transmitted without commentary, however, potentially represent a version of the text differing from that utilized by Kshemaraja. The printed edition in the Kashmir Series of Texts and Studies, based, according to the editors' notice, on three fairly uniform sarada manuscripts, has been utilized as the textus receptus for this study. 2 Although the printed text serves as a generally reliable and workable edition, a twofold critical edition remains a desideratum. 3 First, an edition should be prepared of the text and commentary of Kshemaraja, using additional Svacchanda-bhairava-maha-tantram, v. Tripathi, mss.no.3865, p.776; for Svacchanda-ialita-bhairava-tantram., v. The NepalGerman Manuscript Preservation Project, Kaisar Library, Kathmandu, tantra mss.no.68; cf. Svacchanda-lalita-bhairavastotram in Nepalarajakayavirapustakalayasthahastalikhitasamastapustakanam Samksiptasucipatram, Puratpattvaprakasanamala 18 (Kathmandu: Joraganesa Press, 1963), p.167. 1 For information about these, Svacchandatantram, and texts with related names, v. Gopinath Kaviraj, Tantrika Sahitya, Hindi Samiti Granthamala 200 (Lucknow: Bhargava Bhusana Press, 1972), pp. 720-721. 2 On the mss. used for the printed work, v. Shastri, ed., Swacchanda-Tantra 1: i-ii. 3 I am now preparing an historical-critical study of the text of Svacchandatantra.

r' 3 manuscripts, and following the standard principles of textual criticism. Second, using primarily the newari manuscripts, one should prepare an edition of the text transmitted without commentary. This second edition can not follow the principles used for the first, suitable for works of single author, but rather must follow those procedures elaborated for the anonymous, collectively authored works of ancient Indian literature. For remarks in the commentary of Kshemaraja and internal literary criticism indicate that the text of Svacchandatantra had undergone a long and complicated process of transmission before his commentatorial work. 1 In editing such anonymous works, the canons for establishing the earliest, correct reading must be supplemented by the systematic collection of significant variants reflecting successive, independent versions. Z Preliminarily, the distribution of extant manuscripts supports arguments for a north Indian origin of the text of Svacchandatantram, in agreement with what is otherwise known about the geographical popularity of the cult of Bhairavah. Linguistic evidence based on an analysis of manuscripts, however, at present, does not permit an additional specification of the provenance of this text. For this specification, the non-standard, or better, non-Paninian Sanskrit, designated by the later tradition as aisah, i.e., the usage of Isah, the lord, in the corpus of extant 1 V., esp., his remarks in bk.14, p. 120. There Kshemaraja states that he has used old manuscripts to remedy the corrupt readings and interpolations found in the text. 2 For a summary discussion of the differences between the two types of editions, v. Oskar von Hinuber, "Remarks on the Problems of Textual Criticism in Editing Anonymous Sanskrit Literature," in Proceedings of the First Symposium of Nepali and German Sanskritists 1978 (Kathmandu: Institute of Sanskrit Studies, Tribhuvan University, Dang Nepal Research Centre, 1980), pp.28-40.

4 texts would first have to be thoroughly collected and analyzed. 1 Such work might enable the isolating of distinctive features of a "tantric Sanskrit," which, at least by frequency, might be differentiated from those irregularities shared with other forms of non-standard usage, variously designated as Buddhist hybrid, epic, puranic, or vernacular Sanskrit. 2 Even after such a collection of forms, however, any subsequent geographical localizing of the provenance of Svacchandatantram on the basis of features shared, for instance, with a middle Indic dialect, would be unlikely, due to the difficult in localizing the known middle Indic dialects themselves. Currently, the manuscript evidence merely confirms what the substantive history of the content of Saiva texts of this genus show: namely, an assimilation and reworking by more cultivated milieus. And the progressive excision of deviant forms by successive generations of copyists or commentators merely reflects this process. 3 In sum, therefore, linguistic evidence corroborates the other evidence pointing to an origin in a nonPaninian, non-smarta brahmanical dominated tradition. Besides manuscripts and linguistic analysis, information found in other related texts offers another potential source of evidence about the origin and history of Svacchandatantram. In the selfrepresentation of these expressly Saiva traditions, works such as Svacchandatantram constitute the scriptural corpus of revelation 1 In his commentary, Kshemaraja regularly notes these nonstandard forms, and glosses them with the correct usage. 2 On the need for this work, v. Albrecht Wezler, review of The Satsahasra Sarnhita, by J.A. Schotermann, Zeitschrift der Deutschen Morgenlandischen Gesellschaft 135, 1 (1985): 143-146. 3 A tension exists between a conservative respect for scripture, and a need to adapt it either linguistically or ideologically to support altered circumstances. As an example of an ingenious solution that maintains the literal wording but with a transformed sense, v. infra 1.2.3.

150 that fufills the function in the Saiva community that Vedic texts fulfill in the explicitly brahmanical smarta traditions. And, at some time during the growth and transmission of this Saiva tradition, as it developed an independent identity, it applied classificatory schemata to its scriptures. Saiva circles subsequently used and transmitted these schemata in several different ways: they either inserted these schemata into the body of existing texts, used them to elaborate and to incorporate "new" texts, or transmitted them concurrently in oral form. 1 One popular schema organized the scriptures in three branches according to their revealing deity, doctrine, and number of texts: ten dualistic scriptures revealed by Shiva, eighteen dualistic-non-dualistic revealed by Rudrah, and sixty-four nondualistic revealed by Bhairavah. In later presentations of this pattern, the Svacchandatantra, revealed by Bhairavah, naturally, falls among the non-dualistic group of sixty-four scriptures. 2 The historical value of the doctrinal classification, specifically, the labeling of Svacchandatantram as a non-dual scripture, becomes immediately suspect, however, by the very fact that Kshemaraja explicity states that he composed his commentary to refute dualistic commentaries on 1 For information on the classificatory schemata used in tantric texts, v. Teun Goudriaan and Sanjukta Gupta, Hindu Tantric and sakta Literature (Wiesbaden: Otto Harrassowitz, 1981), pp.1-31. 2 V. K.C. Pandey, Abhinavagupta. An Historical and Philosophical Study, 2 d ed. The Chowkhamba Sanskrit Studies 1 (Varanasi: Chowkhamba, 1963): 138 ff; Raniero Gnoli, trans., Luce delle Sacre Scritture (Tantraloka) di Abhinavagupta, Classici delle Religioni, Sezione prima (Torino: Unione Tipografico-Editrice Torinese, 1972), p.70, n.17. The best summary of the traditional sources is given in the commentary by Jayaratha on Tantraloka, in Pandit Madhusudan Kaul Shastri, ed., The Tantraloka of Abhinava Gupta, with commentary by Rajanaka Jayaratha, Kashmir Series of Texts and Studies, 23 (Allahabad: The Indian Press, 1918), 1, vs. 18: 35 ff.

6 Svacchandatantram. 1 And as will be shown, Svacchandatantram exhibits affinities with a wide spectrum of Saiva and non-Saiva texts, and lacks an explicit dualistic, qualified non-dualistic, or non-dualistic credo. Clearly, therefore, for Svacchandatantram, this particular schema represents an ex post facto superimposition reflective only of the later transmission of the text and not the circumstances of its early composition. Similarly, the number sixty-four associated with the nondualistic scriptures revealed by Bhairavah, which is then divided into eight subgroups, appears to be a factitious and retrospective construction. It may have been based in part on traditional numerical associations, such as the eight bhairava, expanded by correlation with the sixty-four yoginyah associated with Bhairavah. 2 Though Saiva tradition expressly associated Svacchandatantram with a subset of eight specifically Svacchandabhairava tantrani in this list of sixty-four, Kshemaraja in his commentary, which is full of citations and references, does not extensively quote these works. 3 This pattern argues that this 1 V. the concluding verses to his commentary, in bk.15, p.146. 2 On the sixty-four yoginyah, v. Marie-Therese de Mallmann, Les enseignements iconographiques de l'Agni-Purana, Annales du Musee Gimet 67 (Paris: Presses Universitaires de France, 1963): chap. 8, esp., p.176, n.9, referring to the sixty-four bhairava. As she notes, only the eight main bhairava appear to be well defined. For a list and description of the sixty-four bhairava, v. Gopinath Rao, Elements of Hindu Iconography 2, 1 (Madras: The Law Printing House, 1916): 180 ff. 3 In his commentary on Tantraloka, Jayaratha, in Shastri, M.K., ed., Tantraloka, 1: 42, quotes the Srikanthi describing this first group of eight as an expansion of Svacchanda-bhairvah: "svacchando bhairavascandah krodha unmattabhairavah / asitango mahocchusma kapalisastathaiva ca // ete svacchandarupastu bahurupena bhasitah / "V. also bk.1, p.10, where Kshemaraja recognizes this tradition, by quoting the Sarvavirah, associating Svacchandatantram, with the first four texts in this group, the Candah, Krodhah, Unmattabhairavah, and others. In his

7 association derived more from later, systematic considerations than intrinsic historical and material connections. If additional texts are discovered in manuscript and studied, however, perhaps more substantive reasons for their affiliation by these related patterns will come to light. Nevertheless, just as it lacks a non-dualistic credo, so Svacchandatantram itself does not refer to this schema of sixty-four Bhairava or eight Svacchanda-bhairava tantrani.1 Another widespread classificatory schema divides the Saiva scriptures into various streams (srotamsi) of scriptural revelation which emanate from the faces of Shiva in his various forms. 2 Thus, in addition to assigning Svacchandatantra to the nondualistic branch of sixty-four Bhairava scriptures, Saiva tradition, as followed by Kshemaraja, further categorizes Svacchandatantram in this related schema as belonging to the daksina, i.e., right or commentary, however, Kshemaraja quotes only a Rurusamgrahah. bk.10, p. 402, (which may by identical to the Saiva Rauravagamah) on the names of rudrah, and more significantly, an Ucchusmabhairavah. bk.7, p.315, bk. 10, p.382, giving the same quote twice, which states that only one pure reality exists. 1 In bk.10, pp. 511 ff, the text does mention the ten dualistic, and the eighteen, dualistic-non-dualistic texts, representing the twenty-eight traditional Saiva scriptures. For a rejection of the historical significance of this division, into dualistic, and non-dualistdualistic, v. Helene Brunner-Lachaux, "On the Classification of Siva Tantras," in Summaries of Papers, International Association of Sanskrit Studies, Fifth World Sanskrit Conference (Varanasi: Rashtriya Sanskrit Sansthan, 1981), pp. 128-129. 2 in the basic form of this model, the Saiva scriptures are emitted from the five faces of Sadasivah; v., for a collection of these classifications, M. Arunachalam, The Saivagamas, Peeps into Tamil Culture.6, Studies in the Agamas 1 (Madras: Gandhi Vidyalayam, 1983), pp. 20 ff, 99 ff. This schema represents the Saiva counterpart of the Vedic model in which the different Vedah and other forms of knowledge are emitted from the four faces of Brahma. For an example, v. Krsnasankara Sastri, et al., eds., srimadbhagavatapuranam 3 (Varanasi: Samsara Press, 1965), bk. 12, vss.34-48: 454 ff.

8 southern, stream of scriptures. 1 Using an adage evidently popular in his own tradition, Kshemaraja ranks the various streams of scripture in the order, Saivasiddhantin, vama, i.e., left or nothern, daksina, and Kaula. 2 Svacchandatantram itself does not allude to this structure, and Kshemaraja explains it more fully only in his commentary on the Netratantram, which describes the presiding deity of each stream: Sadasivah for the Saivasiddhantin, Tumburuh for the left, Bhairavah for the right, and Bhairavah once again for the Kaula.³ The historical value of this schema is weakened, however, because only a single text survives that can be putatively attributed to the left stream. 4 Just as the preceding schemata, so also this model apparently reflects more the conditions of the later transmission of texts in the various Saiva schools than the actual circumstances of their composition. In particular, by dropping the vama, this schema becomes just another expression of the later opposition between the dualistic Saiva-siddhantin and non-dualistic schools of Saivism represented by Kshemaraja. As will be examined in full, much of Svacchandatantram, however, appears to predate 1 V. bk.1, p.8. 2 V. bk.11, p.55: ". saivadvamam tu daksinam daksinatparatah kaulam kaulatparataram nahi. "Saiva, as Kshemaraja explains, refers here to the Saiva-siddhantin texts. 3 V. Pandit Madhusudan Kaul Shastri, ed., The Netratantra with Commentary by Kshemaraja 1, Kashmir Series of Texts and Studies 46 (Bombay: Tatva Vivechaka Press, 1926), bks.11-12, pp. 239-259, and Helene Brunner-Lachaux, "Un Tantra du Nord: Le Netra Tantra," Bulletin de l'ecole francaise d'extreme orient 61 (1974): 148 ff. 4 V. Teun Goudriaan, ed. and trans., The Vinasikhatantra, A Saiva Tantra of the Left Current (Delhi: Motilal Banarsidass, 1985), p.27. "The Vinasikha is therefore of considerable importance as a unique document of the Tumburu tradition of Vamastrotas within early Tantric Saivism.' "

g the split between these dualistic and non-dualistic schools. And, as will be discussed, the Kaula stream, though closely related to the daksina as in this schema, only appears as an independent tradition considerably after the formation of the other scriptural branches. 1 The Svacchandatantram, in characterizing itself, uses two simpler schemata instead of referring to any of the preceding pattern of streams found in many scriptures and commentators. First, in the opening book, the text introduces the commonplace that this particular tantram represents a condensation of a much more extensive and primordial "Ur-tantram. "The revelation of this particular text, in its accessible and abbreviated form, constitutes a particular act of grace. 2 Behind the theological cliche lies a possible historical kernel, for in his commentary, Kshemaraja refers to larger texts specified only as "brhattantrani, "which apparently, in extenso, supplement the data given by Svacchandatantram. 3 Perhaps this citation warrants, with caution, the supposition that larger texts of this particular genre were extant at his time, and were to be taken as the context for explicating parts of Svacchandatantram. But the vague reference gives pause, being inconsistent with Ksemaraja's usual commentatorial practice of citing texts by title. 4 1 V. infra II.15, the summary of bk.15. 2 V. bk.1, pp. 9-10, esp. vss., 6-7. Cf. Shastri, M.K., ed., Tantraloka, 36, vss.1-10, for a similar version of the condensation of revelation based, according to Abhinavagupta, on the Siddhayogesvarimatam. For a Puranic version of this idea, v. Ludo Rocher, The Puranas, A History of Indian Literature 2, 3 (Wiesbaden: Otto Harrassowitz, 1986), pp. 46-47. 3 V., for example, bk.10, p. 223, referring to a Brhattantram, and bk. 10, p. 267, p.440, referring to unspecified Brhattantrani. 4 This may also represent Ksemaraja's shorthand for a general cross-reference to other Saiva texts discussing the same topics: note

10 Second, in its version of the "tantra-avatarah˝book, the account of the origin of the scriptures found in most Saiva texts, Svacchandatantra presents another longer and likely later account of the revelation and transmission of scripture. 1 Despite the relative sophistication of this account, however, it refers neither to groups of scriptures nor to specific individual scriptures. Instead it simply emphasizes the motivation of grace while otherwise summarily repeating the cosmogonic account in which the supreme sound manifests itself at successively grosser levels as described at great length in other books on formula (mantrah) practice, and cosmology. The additional designation of Srikanthah as a primary transmitter of the revelation, a role underscored several times by Kshemaraja in the course of his commentary, appears to be only an inherited feature from earlier Saiva discussions of revelation. 2 Thus this theoretical and generalizing account, though in agreement with the rest of Svacchandatantram, is potentially applicable to any Saiva scripture and offers no significant historical information. Though omitting specific and applicable classifying information in its self-characterization, Svacchanda-tantram nevertheless refers to itself as a tantram. While the terms agamah and tantram both on bk.4, p. 13 his gloss of the text's reference to other scriptures as referring to the brhattantram. In addition, if the text represented an abridgement of a larger, historical work, it might have an appropriate title indicating this derivation. Cf. Rocher, The Puranas, p.67: here as elsewhere in Sanskrit Literature we should expect Brhad-, Laghu-, Vrddha-, etc., to refer to compositions which are secondary as compared to the corresponding titles to which these adjectives have not been prefixed." " 1 Cf. the summary of bk.8, pp. 17-23, in section 11.8. 2 For the literature on Srikanthah the disputed either legendary or historical found of the Pasupatah, v. Jan Gonda, Medieval Religious Literature in Sanskrit, A History of Indian Literature, 2, 1 (Wiesbaden: Otto Harrassowitz, 1977), p.216. Also v. infra the discussion of the Pasupatah in section 1.1.3.

11 designate primary Saiva scriptures, it is often noted that their usage, at least in part, differs regionally: agamah is the normal term in the south, tantram in the north. 1 Athough this differentiation may correctly reflect the later history of the Saiva schools, in which the dualistic Saiva-siddhantin based on the twenty-eight agamah became dominant in the south, and the nondualistic in Kasmir, it in no way reflects the religious conditions of earlier periods. 2 In addition, Svacchanda-tantram, like the other agamah, uses both terms in references interchangeably. 3 Furthermore, even if this distinction were accepted, the use of the term, tantram, would merely argue for an unspecified north Indian origin for Svacchandatantra. In its opening dialogue, the Svacchandatantram offers another self-characterization as the compendium of the great scripture (mahatantram) that has four seats or bases (catuspitham). This refers to another classificatory system, which, as explained by Kshemaraja quoting the Sarvavirah, groups scriptures by one of four topics that they may predominantly treat, or for which they may serve as the basis: mantrah or formula; vidya or (female) formula; mandalam or diagram; and 1 V., for example, Gonda, Medieval Religious Literature in Sanskrit, p.2, n.5. 2 On this misconception, v., for example, Alexis Sanderson, review of Mrgendragama, by Michel Hulin, Bulletin of the School of Oriental and African Studies 46 pt.1 (1983): 161. ³ V. bk.4, p.13, for example, where Svacchanda-tantra refers to other agamah, or bk.4, p.101, to other tantrani. Cf. on this topic Helene Brunner-Lachaux, "Les Membres de Siva," Asiatischen Studien 40, 2 (1986): 89, n.1. 4 Bk.1, p.10.

12 mudra or gesture. 1 The Sarvavirah, has classified Svacchandatantra as a formula text, an understandable assertion given the actual predominance of mantra practice in its content. 2 Kshemaraja proceeds to explain, however, that Svacchanda-tantram, as indicated by the opening verse, actually treats all four topics. This discrepancy suggests that this fourfold classificatory schema represents a later categorization of texts, rather than a compositional principle used in constructing early texts like Svacchandatantram. In this way this fourfold schema resembles the other well known ideal description of an agamah as having four topical sections (padah) treating: kriya or ritual; vidya or doctrine; carya or conduct; and yogah or spiritual exercise. 3 Almost all extant agamah, however, do not conform to this structure. 4 Despite its vast similarity in shared content with these agamah, Svacchanda-tantra does not show any traces of this organization. Thus, this omission indicates that these organizing models stem, once again, from systematizing efforts effected during the later transmisson of the Saiva scriptural corpus. As sources of historical information, the traditional views of scripture, represented in the preceding schemata and selfcharacterizations, should be supplemented by assorted direct and indirect external evidence. First, these Saiva scriptures ostensibly 1 Bk. 1, p. 10: "mudra mandalapitham tu mantrapitham tathaiva ca vidyapitham tathaiveha catuspitha tu samhita. 2 Ibid. ☑ 3 On this typical arrangement see, for example, Jean Filliozat, "Les Agama Civaites," Introduction in Rauravagama 1, ed. N.R. Bhatt, Publications de l'institut francais d'indologie 18 (Pondichery: Institut francais d'indologie, 1961): vii. 4 V. Helene Brunner-Lachaux, "Importance de la litterature agamique," Indologica Taurinensia 3-4 (1975-76): 10%.

13 serve as the authority and basis for the later datable and identifiable Saiva sectarian authors. And by default, their dates often serve as the terminus ad quem for the Saiva scriptures. Accordingly, in the case of Svacchandatantram this terminus is the tenth century and Abhinavagupta, and probably prior to the historical figures in his sampradayah or teaching lineage. 1 Supplementary evidence must then help decide a suitable terminus ab quo. At the outset, it should not be assumed that the composition of anonymous scriptures necessarily terminates at the time when more cultivated and historically self-conscious milieus begin theoretically expounding and elaborating them in works of a qualitatively different nature. A continuing parallel process of fabrication, recomposition, and transmission of scriptures, if not in the same, at least in other circles, should be presumed. And this process argues for prudence in pushing back the terminus ab quo for anonymous texts by a set amount determined by reasoning about the time needed for an evolution, as it were, of ideas to a more complex level, or rather, assmilation by a more sophisticated milieu. Nonetheless, in the case of Svacchandatantram, identical reasoning argues for a considerable prehistory, if not antiquity, in which the text could acquire prestige sufficient to motivate numerous commentaries, and even colonization by competing sects. In the absence of direct evidence such as citations (prior to those of the Saiva authors discussed previously), or inscriptions, dating a text such as Svacchandatantram depends on the dating, or periodization, of the genus of texts to which it is commonly 1 The first historical figure in the non-dual lineage, for whom scriptures such as Svacchandatantram are the supposed scriptural basis is Vasugupta, who is the author transmitter of the source text of this school, the Sivasutrani, dated to the beginning of the ninth century. V. Pandey, Abhinavagupta, p.154.

14 attributed. Inscriptional evidence for other Saiva tantrani falls within the span proposed for the composition of Svacchandatantram: in the eighth century before the historical figures in the tradition of Kshemaraja. 1 A century earlier, the existence of Saiva scriptures, if not the agamah of later tradition can be inferred from the South Indian inscription of the Pallava ruler Narasimhavarman II, in which he refers to the Saivasiddhantah. 2 Certainly, the organized cult of Shiva, and, in particular, Saiva sects, such as the Pasupatah, whose practices and beliefs have strongly influenced works like Svacchandatantram, predate this time by at least another century. 3 And though from other evidence of Saiva religious activity in earlier periods the concomitant existence of oral or even written traditions may be inferred, this evidence cannot warrant retroactive assumption of a corpus of Saiva scriptures at this earlier period. More reasonably, their appearance can not be 1 For the Cambodian inscriptions that attest the existence of Saiva tantrani from the beginning of the ninth century, v. Adhir Chakravarti, "New Light on Saiva Tantrika Texts Known in Ancient Cambodia," Journal of the Asiatic Society (Bengal) 15, 1-4 (1973): 1- 10. 2 E. Hultzsch, ed. and trans., South Indian Inscriptions, 1 in Archaological Survey of India n.s.3 (1890; reprint, Varanasi, Delhi: Indological Book House, 1972): 12, 13-14, vs.5. On the dates of Narasimhavarman, c.690-91- c.728-29, v. T.V. Mahalingam, Kancipuram in Early South Indian History (Bombay: Asia Publishing House, 1969), pp. 109 ff. 3 The Pasupatah can be dated in the most conservative view to the preceding two centuries. For a reexamination of the debate on their date and origin, v. V.C. Srivastava, "The Antiquity of the Pasupata Sect," in K.C. Chattopadhyaya Memorial Volume, Department of Ancient History, Culture & Archaology (Allahabad: Allahabad University, 1975), pp. 109-125, especially p.111: "There cannot be any doubt that the Pasupata sect was a well-established system in India in the early mediaeval period (7 th-12 th cent. A.D.)."

15 placed before the efflorescence of Saivism produced by the patronage accorded by dominant Saiva dynasties in the seventh and eighth century. 1 To complement historical evidence such as inscriptions, anonymous works of Sanskrit literature such as Svacchandatantram can be analyzed for isolated internal clues to their provenance that potentially can be dated or located in an external context. Identifiable and specific historical references naturally merit first consideration. Unfortunately, Svacchanda-tantram, with its predominantly ritual and meditational content, does not refer to specific historical figures or events, nor does it associate its deity with a definite pilgrimage center (pitham).2 Comparable to references to historical events, persons, and places, literal citations from other known and datable works can provide reliable information about the anonymous text in which they are embedded. As might be expected, however, in an anonymous and collective composition that purports to be revealed scripture, paraphrase is the rule and quotation the exception. 3 1 For literature, v. Francois Gros, "Towards Reading the Tevaram," introduction to Tevaram. Hymnes Sivaites du Pays Tamoul, T.V. Gopal Iyer and Francois Gros, eds. (Pondichery: Institut francais d'indologie, 1984), esp., pp.1-liii. 2 Cf., for an example of such a reference used in identifying the provenance of a text, J.A. Schotermann, ed., The Yonitantra (New Delhi: Manohar Publications, 1980), pp.3 ff. 3 Cf. in reference to the Puranic texts, Ludwik Sternbach, The Manava-dharmasastra I-III, and the Bhavisya Purana (Varanasi: All India Kashi Raj Trust, 1974). He notes (p.4) how exceptional it is that the Bhavisyapuranam recopies verbatim the smrti texts.

16 Next in importance as internal indicators come iconographic and architectural details. 1 Many agamah have extensive sections on temple construction, installation of images, and similar subjects, since they functioned in large part as guidebooks for temple priests. While sharing many other topics with these texts, Svacchandatantram, however, treats these subjects only in passing. First, for iconographic evidence, though not matching exactly the meditational form of Svacchanda-bhairava described in Svacchandatantram, a bronze figure of Svacchanda-bhairava in union with Bhairavi has been found at Kangra and dated to the twelfth century. 2 This iconographic evidence corroborates other evidence locating the worship of Bhairavah as the supreme form of Shiva in north India, and specifically the northwest Himalayan region. 3 The composition of Svacchandatantram as the principal scripture associated with the worship of Svacchanda-bhairava might then be inferred to have occurred in this region. This can not be concluded with certainty, however, for evidence attests the 1 Cf. Teun Goudriaan, Kasyapa's Book of Wisdom (Kasyapajnanakandah). A Ritual Handbook of the Vaikhanasas (Utrecht) Disputationes Rheno-Trajectina 10 (The Hague: Mouton & Co., 1965) p.10, n.13: "The best way for finding some chronological clues is to consider the architectural and iconographic date offered by the text." 2 B.Ch. Chhabra, "Svachchhanda-Bhairavi Bronze Image from Kangra, Journal of Indian History 42, 1 (April 1964): 145-148. Cf. the description of Svacchanda's meditational form in bk.2, pp.51 ff. 3 V. Brunner-Lachaux, "Les Membres de Siva," p.114, "...celle du Bhairava a cinque tetes qui est au centre du culte du sivaisme du Nord." V. also Alexis Sanderson, "Saivism in Kashmir," in The Encyclopedia of Religion 13 (New York: Macmillan, 1987): 17: "... in Kashmir ... the principal Saiva cult was...the worship of Svacchandabhairava and his consort Aghoresvari

17 existence for Bhairava cults in other centers of scriptural Saivism, for example, in south India. 1 Second, for architectural evidence, though not describing the construction of temples, Svacchandatantram, as usual in texts of its kind, includes brief instructions for worshipping the deities on the door of the ritual pavilion (yagagrham). It locates the rivers Ganga and Yamuna on opposite sides of the doors, a motif that became popular in temple architecture in the seventh century and was reputedly introduced by Lalitaditya into Kasmir in the eighth century. 3 In addition, Svacchandatantra places Ganga on the left and Yamuna on the right. Kshemaraja defends this placement as proper for texts of the daksina Bhairava stream against previous cornmentators who, in commenting upon this passage, had reversed their stated position to conform to the usual Siddhantin pattern. 4 Elsewhere, it has been argued that the Siddhantin placement represents the standard earlier pattern, and the placement found in Svacchanda-tantra typical and later tantric symbolism. 5 Since temples throughout India, however, show both patterns, this detail unfortunately does not further specify the 1 Cf., for example, V. Raghavan, "Tiruvorriyur Inscription of Chaturanana Pandita," in B. Ch. Chhabra, ed., Epigraphia Indica 27 (Delhi: Archaological Survey of India, 1985): 298. Raghavan cites inscriptional evidence that as late as the time of the Chola king Rajadhiraja II (second half of the twelfth century) the Somasiddhantin doctrine associated with Bhairavah and the KapalikaPasupatah was being preached at this site near Madras. 2 Bk. 2, pp. 13-14. 3 Heinrich von Stietencron, Ganga und Yamuna, Freiburger Beitrage zur Indologie 5 (Wiesbaden: Otto Harrassowitz, 1972), pp. 35 ff. 4 Bk.2, pp. 13-14. 5 V. von Stietencron, Garga und Yamuna, pp.113 ff.

18 date or provenance of the text. 1 Moreover, rather than with external structures, such as a Bhairava shrine or temple, here, as in its other statements about monuments or iconography, Svacchandatantram may be more concerned with internal modes of worship and meditation. Accordingly, later in the text, after prescribing construction of a ritual diagram (mandalam) with colored powders, Svacchandatantram repeats its instructions to perform the same worship at the "doors." of the diagram. 2 Svacchandatantram contains another potential indicator of its origin when it describes the construction of amulets written on birch bark (bhurjapatram).³ Since birch bark was used as a writing material primarily in the northwest Himalayan region, Schrader, for example, argued that the Pancaratrin Ahirbudhnya Samhita, which describes the same kind of amulets, had been composed in Kasmir. 4 But as reported by Buhler, the use of birch bark for amulets and letters had a much wider distribution.5 Therefore, though providing another clue as to the provenance of 1 For an exhaustive list of the distribution of this pattern in 314 temples, v. von Stietencron, Ganga und Yamuna, pp.16 ff. 2 Bk.5, p.23. 3 Bk.9, pp.54 ff. 4 F. Otto Schrader, Introduction to the Pancaratra and the Ahirbudhnya Samhita (1916. Reprint. Madras: The Adyar Library and Research Centre, 1973), p.111. 5 Johann Georg Buhler, Indian Paleography, Appendix to The Indian Antiquary 33 (1904), ed. John Fleet, p.93: "According to a statement made to me by Bhau Daji, birch bark mss. occur also in Orissa, and amulets, written on Bhurja, are still used throughout all the Aryan districts of India."

19 Svacchandatantram, the use of birch bark does not prove a Kasmiri or northwest Himalayan origin. 1 Furthermore, even if this information pointed unmistakably to a Kasmiri origin, it would not provide a certain starting point for the evaluation of Svacchandatantram as a whole. For in an anonymous work of this size, presumably transmitted in different regions over a long period, a detail permits the dating or locating of only a particular topical section of the work; its importance for the remaining sections must be corroborated by additional evidence. Thus, isolated evidence for date and provenance found in specific sections of an anonymous and extensive scripture like Svacchandatantra must be evaluated more conservatively than that found in the unified work of a single, known author. In the same way, judging manuscript readings or interpreting single verses and larger sections in a composite work, requires an awareness of the intention of a lineage of distinct authors, and an understanding of the text's unique history of composition and transmission. First, the text of Svacchandatantra has, as its basic compositional or organizing structure, a division into books (patalam). Just like the choice of tantram to designate the work as a whole, the choice of the term patalam, used throughout the genre to designate sections of a text, does not appear significant. 2 1 V. also Sanjukta Gupta's introduction to the Laksmi Tantra, in Sanjukta Gupta, trans., Laksmi Tantra. A Pancaratra Text (Leiden: E.J. Brill, 1972), p.xxi, against arguments for provenance based on the used of birchbark amulets. 2 On the historical significance of the term patalam in Vedic ritual literature, v. Asko Parpola, trans., The Srautasutras of Latyayana and Drahyayana and their commentaries, Commentationes Humanarum Litterarum, Societas Scientiarum Fennica, 42, 2 (Helsinki, 1968), pp. 43 ff; for references to the etymology and used of the term tantram in medical literature, v.

20 For the most part, these books coincide with the introduction of major new topics by the other main organizational structure of the text, the frame dialogue between Bhairavah, who in this work assumes the role of promulgator, and his consort goddess, who plays the interlocutor. In addition, the goddess lists, in the first book, the major topics to be covered in the work. 1 Many anonymous, encyclopadic works like purani have this short table of contents called a sucih or anukramanika.2 The actual number of topics treated in the text exceeds not only the number of books but also those topics listed separately in the opening anukrarnanika. Thus, in his commentary on this initial anukramanika, Kshemaraja attempts to demonstrate that the listed topics include and, thereby, imply the remaining topics. Similarly, in the fourth book, before discussing a series of previously unenumerated topics, the text supplies another anukramanika.3 While the anukramanika omits many topics, almost every topic is marked by the framing dialogue between Bhairavah and the goddess. And where even this dialogue is lacking, Kshemaraja supplies it, by marking the beginning and end of added topics. 4 He confirms, thereby, the perception that the Reinhold F.G. Muller, "Worterheft zu einigen Ausdrucken indischer Medizin," Mitteilungen des Instituts fur Orientforschung, Deutsche Akademie der Wissenschaften zu Berlin Institut fur Orientforschung 8, 1 (1961): 102. 1 Bk.1, pp. 11-12. 2 On this organizational device in the purani, v. Giorgio Bonazzoli, "Schemes in the Puranas," Purana 24, 1 (January 1982): 149 ff. 3 Bk.4, pp. 142-143. 4 Cf., for example, bk.5, p.10, where Kshemaraja rationalizes the deficient dialogue introducing a topic without an appropriate question as due to the supernumerary grace of the lord: . anu jighrksorbhagavtya asayaparitosayaprasnitamapi padadidiksam nirupayisyan

21 dialogue frame forms the primary and even indispensable structure of the text. Although these internal organizing structures, do not, in themselves, further specify the provenance and date of the text of Svacchanda-tantram, they do clearly identify its genre, and suggest the kind of historical and structural analysis that would further unravel its history. Along with the non-Paninian Sanskrit and simple verses, the lack of a strict internally coherent organization marks the text as the collective and cumulative product of an oral and vernacular rather than literary tradition. In particular, the dialogue frame has permitted associated topics to be easily added to the text. While devices like the anukramanika indicate the work of more careful redactors operating with traditional patterns, inconsistencies throughout the work point to the absence of a single author or editor.