Nagarjuna and the Ratnavali: New Ways to Date An Old Philosopher

JIABS Journal of the International Association of Buddhist Studies Volume 25 Number 1-2 2002 Buddhist Histories Richard SALOMON and Gregory SCHOPEN On an Alleged Reference to Amitabha in a KharoÒ†hi Inscription on a Gandharian Relief .................................................................... Jinhua CHEN Sarira and Scepter. Empress Wu’s Political Use of Buddhist Relics Justin T. MCDANIEL Transformative History. Nihon Ryoiki and Jinakalamalipakara∞am 3 33 151 Joseph WALSER Nagarjuna and the Ratnavali. New Ways to Date an Old Philosopher ................................................................................ 209 Cristina A. SCHERRER-SCHAUB Enacting Words. A Diplomatic Analysis of the Imperial Decrees (bkas bcad) and their Application in the sGra sbyor bam po gnis pa Tradition....................................................................................... 263 Notes on the Contributors................................................................. 341 NAGARJUNA AND THE RATNAVALI: NEW WAYS TO DATE AN OLD PHILOSOPHER* JOSEPH WALSER There have been two prevailing trends among scholars of Buddhism writing about Nagarjuna. The first is to simply assume that Nagarjuna lived sometime around the second century AD. The second trend, dating to the beginning of the twentieth century, is found among scholars who try to provide a firm scholarly grounding for this assumption. These scholars have been discouraged in their efforts to come up with anything definitive concerning Nagarjuna's date or place of activity. Among the more prominent of the early studies we have statements such as the following from Max Walleser's 1923 study: The systematic development of the thought of voidness laid down in the Prajñaparamita Sutras is brought into junction with the name of a man of whom we cannot even positively say that he has really existed, still less that he is the author of the works ascribed to him: this name is Nagarjuna1. Almost eighty years later, the situation has not improved. Surprisingly little in the way of new evidence or new interpretation has been brought to bear on the question of his dates and location in recent scholarship, although there have been a number of works summarizing the available data. The most recent of these summaries, “The Problem of the Historical Nagarjuna Revisited,” by Ian Mabbett provides an excellent survey and analysis of much available scholarship to date. The abstract to his article minces no words in its evaluation of the current state of Nagarjuna scholarship. * My sincere thanks to James Egge, Ikumi Kaminishi, and Gary Leupp who read drafts of this article, and to my wife Radha, who has read and commented on multiple versions. 1 M. Walleser, The Life of Nagarjuna from Tibetan and Chinese Sources, reprint (Delhi: Nag Publishers, 1979),1. The original article appeared in Asia Major, Introductory Volume. Hirth Anniversary Volume. (Leipzig: 1923), 421-55. Journal of the International Association of Buddhist Studies Volume 25 • Number 1-2 • 2002 210 JOSEPH WALSER Nagarjuna, the founder of Madhyamaka, is an enigma. Scholars are unable to agree on a date for him (within the first three centuries AD), or a place (almost anywhere in India), or even the number of Nagarjuna's (from one to four). This article suggests that none of the commonly advanced arguments about his date or habitat can be proved; that later Nagarjunas are more likely to have been (in some sense) the authors of pseudepigrapha than real individuals; that the most attractive (though unproved) reading of the evidence sets Nagarjuna in the general area of Andhra country in about the third century AD2. The rather intractable problem with which scholars have been struggling becomes apparent in Mabbett's account of the sources. Although there is no lack of literary sources discussing Nagarjuna, almost all of the elements contained therein are mythical at best and conflicting at worst. Further, very few details contained in these sources can be corroborated with external evidence. Most of this material comes from accounts that were written with hagiographical interests ahead of historical documentation. Clearly, for those who like certainty, any kind of “proof” of Nagarjuna's dates and place of residence is a long way off. Thus far, the details of Nagarjuna's life have been little more than a passing curiosity to most Madhyamika scholars — a problem which persists but which is assumed to have little bearing on his philosophy (which is the primary object of their interest). This is naive. Any philosophical text needs to be read within its socio-historical context. More to the point, Nagarjuna's philosophy as presented in the Mulamadhyamakakarika (hereafter the “Karika”) is argumentative, and the opponent or opponents are unnamed. The range of interpretations that one may give to any of the arguments in the Karika is limited, at least in part, by the assumptions that one makes about whom Nagarjuna is arguing against. The issue of the identity of Nagarjuna's audience has not become an issue in Nagarjuna scholarship because scholars have tended to read him through the lens of Candrakirti or Bhavaviveka, both of whom assume that Nagarjuna's primary opponent is a Sarvastivadin3. There is, however, 2 I. Mabbett, “The Problem of the Historical Nagarjuna Revisited,” Journal of the American Oriental Society 118 (1998): 332. 3 See, for example, Candrakirti's long explanation of the Buddhist path in his commentary on MMK. 24.4, which seems to come straight out of the 6th chapter of Vasubandhu's Abhidharmakosa. NAGARJUNA AND THE RATNAVALI 211 some reason to question this assumption. Richard Hayes4 has argued that Nagarjuna's arguments against the svabhava theory of the Sarvastivadins do not refute any theory that the Sarvastivadins actually held5. More to the point, Sarvastivada was not just a philosophical school in the abstract, but something that existed as an institutional reality at a specific times and places. In the first two centuries of the first millennium, it was centered in Gandhara and Kashmir, and there is no reason at present to assume that its influence was pan-Indian. Thus, when scholars interpret Madhyamika philosophy as a response to Sarvastivadin tenets they inadvertently import one of three assumptions about Nagarjuna's date and place of residence — assumptions which they are unprepared to defend. If Nagarjuna's opponent in the Karika was a Sarvastivadin, then a) Nagarjuna lived when and where the Sarvastivadins were present; i.e. he lived somewhere in north India and his arguments were directed at a local opponent, or b) Nagarjuna lived at a time and place where there were no Sarvastivadins present and Sarvastivadin philosophy was unheard of, or c) Nagarjuna lived someplace where Sarvastivadins were not present, but which had some kind of cultural contact with northwest India/Pakistan. If the first option is assumed, then it implies an unstated presumption for which no evidence has been given. If the second assumption is true, then we might well question whether Nagarjuna was actually addressing Sarvastivadin claims at all. Finally, if the third assumption is correct, and the influence of Sarvastivadin philosophy (as opposed to Sarvastivada monasteries) extended far beyond the regions occupied by Sarvastivadin monasteries, this thesis too would have to be established with some evidence. None of these assumptions have been argued. Any interpretation of Nagarjuna's philosophy that involves contextualizing his arguments in larger discourses, either Brahma∞nical or Buddhist, involves similar unwitting assumptions about the date and location of the discourse. Philosophical propositions may claim to be universal, but 4 R. Hayes, “Nagarjuna's Appeal” Journal of Indian Philosophy 22 (1994): 311. See, P. Williams, “Some Aspects of Language and Construction in the Madhyamaka,” Journal of Indian Philosophy 8 (1980): 1-45. 5 212 JOSEPH WALSER dialectics are always local. Scholars of Nagarjuna do tend to ignore this factor in Nagarjuna's philosophy, yet all but the most formal treatments of Nagarjuna's logic have to assume something about his audience. Mabbett's conclusions, however, need not be the end of the story. It is the purpose of this paper to argue that if we are willing to accept a fallibilist proof, or an analysis based on partial information, we can come to some kind of solution, albeit a tentative one. Given the pressing need to take some sort of stand on this issue, even a tentative solution is preferable to the present impasse. In the following, I will identify two propositions that have a bearing on the date of Nagarjuna. If both propositions turn out to be true, then we will have placed one event in Nagarjuna's life, his writing of the Ratnavali, within a thirty year period at the end of the second century in the Andhra region around Dhanyakataka (modern Amaravati). My interpretation not only supports Mabbett's “most attractive reading” of third-century Andhra, but will upgrade it to “the most likely reading, given our current state of knowledge.” Let me begin by proposing the two sub-theses that could considerably narrow the range of dates and locations for Nagarjuna. The first sub-thesis is that Nagarjuna, the author of the Mulamadhyamakakarika, was also the author of the Ratnavali. The second sub-thesis is that a Satavahana king was Nagarjuna's patron. Many scholars take both of these propositions for granted, but for our purposes it will be important to review the evidence. By means of these two sub-theses, I will establish a period and a location in which the Ratnavali could have been written, thereby establishing a benchmark event in the life of Nagarjuna. 1. The author of the Mulamadhyamakakarika was also the author of the Ratnavali. Modern scholarship has a problem dating Nagarjuna. The blame for this lies at least partly in the way that modern scholars have set up the problem. The first of Nagarjuna's texts to be discovered, and the one which has attracted the most interest in the West has been the Mulamadhyamakakarika. The name, Nagarjuna is then treated as a synecdoche for “the author of the Karika,” and hence it seems strange for us to ask NAGARJUNA AND THE RATNAVALI 213 whether Nagarjuna wrote the Karika because he is, by definition, its author. It only needs to be mentioned in passing that the West was not the first culture to make these kinds of assumptions, it is merely the first to use this one text as the sole indication of Nagarjunian authenticity. For example, to Yijing, Nagarjuna was first and foremost the author of the Suh®llekha, while for Pure Land Buddhists such as Shinran, Nagarjuna was assumed to be the author of the Twelve-Gate Treatise. Be that as it may, by identifying Nagarjuna as the author of the Karika, modern scholars have painted themselves into something of a corner when it comes to the vexed issue of his date. They need to ground their arguments in his text (because the Nagarjuna they are most interested in is, first and foremost, an author), and yet the only evidence which the Karika offers up is of a logical/doctrinal nature. A.K. Warder6 and David S. Ruegg7 have done an admirable job in constructing a “relative chronology” of Nagarjuna vis-a-vis the development of Buddhist and Naiyayika doctrine, but as the other authors they compare him with have even less secure dates than Nagarjuna, we are left little better off than we started. For better or worse, in India “absolute chronologies” (i.e., a set of dates that can be translated into Gregorian dates), have only been worked out for empires and their political administrators. In order to connect Nagarjuna to a Gregorian year, we must first connect him to an Indian monarch for whom the dates are known. To make this kind of connection we need to find evidence relating to practices or events that leave their mark in the archeological record. Unfortunately, the Karika is a peculiar text in that it focuses so exclusively on classical Buddhist doctrine and logical issues that it has few cultural references that would help us date it8. The only option left to us is to seek evidence in other texts ascribed to Nagarjuna, and this is where our scholarly presuppositions leave us in a 6 A.K. Warder, “Parsva, Vasumitra (II), Caraka and Mat®ce†a” in Papers on the date of Kaniska, A. L. Basham, ed. (Leiden: E.J. Brill, 1968), 331-5. 7 See D.S. Ruegg, Literature of the Madhyamaka School of Philosophy in India. A History of Indian Literature. Vol. 7, fasc. 1 (Wiesbaden: Otto Harrassowitz, 1981). 8 The only such reference that I have been able to find occurs in chapter 17 verse 14 where Nagarjuna refers to a contract of debt (®∞apattra). While this is an interesting reference, it is hardly of any help in dating the work, as this practice is well attested in Kau†ilya's Arthasastra and other Dharmasastras as well as in inscriptions dating from the early centuries of the common era. 214 JOSEPH WALSER bind. While nobody wants to defend the thesis that Nagarjuna only wrote one work, scholars are left in the uncomfortable position of having recourse to only two criteria by which to determine the authenticity of a putative Nagarjunian text: a) the testimony of other (preferably early) authors, and b) similarities of logic, doctrine, style to the Karika. While these criteria have been effectively employed to eliminate texts as authentic Nagarjuna texts, the same criteria are not so conclusive when it comes to establishing a text as authentic. Testimony of other authors, doctrine, logic, and style are, however, the only data we have, and so we must consider the evidence such as it is and make an educated guess as to where the weight of the evidence lies. The criteria of doctrine, logic, and style have proven the most problematic to use. In order to date Nagarjuna, we need specific cultural information, and yet most of the texts that present that kind of information are not concerned with the same doctrine as the Karika, and hence do not display its logic or style. An example of such a text is the Suh®llekha, which though universally ascribed to Nagarjuna, displays little of the interests and penchant for argument of the Karika. There is, however, one text ascribed to Nagarjuna, which contains some sections with logical/doctrinal arguments similar to the Karika and has other sections with significant cultural content. This is a work called The Ratnavali or “The Jeweled Garland.” Because this work contains logical/doctrinal arguments similar to those in the Karika, its ascription to Nagarjuna can be investigated using the criteria stated above. Once Nagarjuna's authorship of this work has been established, its numerous social and cultural references can be used to explore the date and location of its author. 1.1. The Authenticity of the Ratnavali Nagarjuna's authorship of the Ratnavali has been well attested in India, China and Tibet going back at least as far back as the sixth century9. Paramartha first translated the work into Chinese in the sixth century, 9 Christian Lindtner states that the Ratnavali is ascribed to Nagarjuna by Bhavya, Candrakirti, and SantarakÒita, “and many other later authors.” See, C. Lindtner, Nagarjuniana, Studies in the Writings and Philosophy of Nagarjuna, (Delhi: Motilal Banarsidass, reprint 1990), 163. NAGARJUNA AND THE RATNAVALI 215 although he does not name its author. The earliest explicit attribution of this text to Nagarjuna can be found in Bhavaviveka's Tarkajvala, where he quotes verses thirty-five to thirty nine from the fifth chapter of the Ratnavali, introducing them with the words, “the great teacher, Arya Nagarjuna said….”10 Though the dates for Bhavaviveka are even more elusive than are those of Nagarjuna, it seems safe to place him in the sixth century11 — perhaps as a slightly younger contemporary of Paramartha. Candrakirti (seventh century) quotes the Ratnavali a total of sixteen times in his Prasannapada12 and five times in his Madhyamakavatara. Though he never explicitly ascribes it to Nagarjuna in these works, La Vallée Poussin notes that the Ratnavali verse quoted after verse three of the twenty-fifth chapter of the Karika “…est citée Namasaµgiti†ika, ad 96, où la Ratnavali est attribuée à Nagarjuna.”13 Similarly, both Haribhadra, in his 8th century Prajñaparamitopadesasastra, and Prajñakaramati (ca. end of eighth beginning of ninth centuries) in his Bodhicaryavatarapañjika quote from it, but without attribution14. It is clear from the number and the context of these quotations that the Ratnavali was a text held in great esteem by the Madhyamika School. It is not clear what conclusions, if any should be drawn from the fact that so many early scholars felt comfortable quoting it without attribution. Surely Candrakirti knew about Bhavaviveka's attribution of the text to Nagarjuna, and if he doesn't repeat the former's attribution, neither does he deny it. In the eighth-century, Jñanagarbha and Klu'i rgyal mtshan as well as the team of Vidyakaraprabha and [s]Ka ba dPal brtsegs both explicitly attribute the text to Nagarjuna in their colophons, as does Ajitamitra, who wrote the ninth century †ika on the work. In short, the work is attributed to 10 “slob dpon chen po ’phags pa na ga rdsu nas ji skad du.” Peking Tripitaka, v. 96, #5256, 145a. 11 For a discussion of his date, see S. Iida, Reason and Emptiness: A Study in Logic and Mysticism (Tokyo: Hokuseido Press, 1980), 6-12. 12 For page numbers see M. Hahn, Nagarjuna’s Ratnavali (Bonn: Indica und Tibetica Verlag, 1982), 10. 13 L. de LaVallée Poussin, Mulamadhyamakakarikas (Madhyamikasutras) de Nagarjuna avec la Prasannapada commentaire de Candrakirti, Bibliotheca Buddhica IV (Osnabrück: Biblio Verlag, 1970), 524, n. 4 14 Haribhadra cites Ratnavali vs. 98 in his Abhisamayalaµkaraloka (p. 66 in Wogihara's edition) 216 JOSEPH WALSER Nagarjuna as early as the sixth century and this attribution is repeated in the eighth and ninth centuries. While these attributions might seem late, it should be kept in mind that (other than a brief remark by Kumarajiva) Bhavaviveka is the earliest source we have that mentions other texts that Nagarjuna wrote. 1.2. Doctrine and logic in the Ratnavali The doctrinal and logical content of the Ratnavali compares favorably with that of the Karika. The Ratnavali is a very different text than the Karika, and presumably speaks to a different audience. Nevertheless, it does contain a number of points of striking similarity to the Karika. In general, both works are committed to a Mahayana teaching of emptiness. There is also a similarity in the topics dealt with in both works as well as the way these topics are treated. For instance, both works have lengthy refutations of the three times (past present and future)15 as well as arguments about antecedent states of being16. The rather peculiar treatment of Nirva∞a as being neither “bhava” nor “abhava” occurs in both works17 as does the teaching that saµsara is somehow not different than nirva∞a18. The topics discussed, however, do not help us to determine authorship since a rehearsal of topics is precisely what determines a school of thought. To determine authorship, we need to isolate those elements that are likely to be idiosyncratic by determining those elements which were unlikely to have been emulated by his followers. In the following, I identify three areas of Nagarjuna's writing in the Karika that appear to be matters of individual style rather than modes of discourse characteristic of the early Madhyamika school. The three areas are: logical syntax, use of scripture, 15 Ratnavali verses 63-5 and 108-115. Compare with similar arguments in Karika chps. 2,5,7,9,11,19,20, and 27. 16 Ratnavali, verse 47, dealing with prior and simultaneous production (prag- and sahajata) echoes the argument about antecedent states of being in Karika chp. 9 and the discussion of previous and simultaneous causes in Karika chp. 6 (there the terms are purvaand saha-bhavaµ). 17 Cp. Ratnavali vs. 42 with arguments in Karika chp. 25. 18 Cp. Ratnavali vs. 41 and 64 with Karika 25.19-20. NAGARJUNA AND THE RATNAVALI 217 and metrics. I will show that these elements are present in the Ratnavali while absent in the works of Nagarjuna's closest disciple Aryadeva. Though the examples of truly logical arguments are fewer in the Ratnavali than in the Karika, there are a few passages in the Ratnavali whose unusual logical syntax is remarkably similar to prominent verses in the Karika. Compare Ratnavali 365: “Past and future objects and the senses are meaningless, [due to the preceding argument]. So too are present objects since they are not distinct from these two.”19 And the familiar verse from Karika chapter two: “ What has been traversed is not being traversed. What has not yet been traversed is not being traversed. What is being traversed, apart from what has been traversed and what is not yet traversed, is not being traversed.”20 Both passages appeal to the law of excluded middle to eliminate a third term which common sense tells us must exist. Though Aryadeva treats similar topics in his CatuÌsataka and Satasastra, he consistently avoids expressing the same ideas in this form21. There are also a number of verses in both the Karika and the Ratnavali displaying what we must assume to be a rather unusual syntax of the form “if a not b; if not a also not b.” For example: Karika 20.15 “Without partaking of a union, how could cause give rise to an effect? But again, with the partaking of a union, how could cause give rise to an effect?22 19 Hahn, 83: “bdag phan ci dan ci bya zes| ji ltar khyed la gus yod pa| gzan phan ci dan ci bya zes| de bzin khyod ni gus par mdzod||” 20 Poussin, 92: “gataµ na gamyate tavad agataµ naiva gamyate| gatagatavinirmuktaµ gamyamanaµ na gamyate||” 21 Cp. CatuÌsataka v. 374 “About the completed it is said, ‘It exists'; about the uncompleted it is said, ‘It does not exist'. When the process of arising is non-existent, what, indeed, is it said to be?” “jayate ’stiti niÒpanno nastity ak®ta ucyate| jayamano yadabhavas tada ko nama sa sm®taÌ||” [K. Lang, Aryadeva’s CatuÌsataka: On the Bodhisattva’s Cultivation of Merit and Knowledge, (Copenhagen: Akademisk Forlag, 1986) 142-3. Also see Satasastra, chapter 8 in G. Tucci, Pre-Dinnaga Buddhist Texts from Chinese Sources, (Baroda: Oriental Institute, 1929), 65-72. 22 Another example can be found at 6.5. 218 JOSEPH WALSER Compare this to Ratnavali verse 68: “If momentary, then it becomes entirely non-existent; hence how could it be old? Also, if non-momentary, it is constant; hence, how could it become old?”23 Again, this way of phrasing the issue is unusual and I can find no examples of it in the writings of Aryadeva. This suggests that this way of phrasing an issue was peculiar to Nagarjuna and not a way of expressing a thought characteristic of the Early Madhyamika school more broadly. 1.3. Sutra References in the Ratnavali The Karika and the Ratnavali both give a prominent position to the same sutras, and make use of those scriptures in remarkably similar ways. Taking the most obvious examples, some version of the Parileyyaka sutta (Samyutta Nikaya III, 94-99) where the Buddha states that some questions are unanswerable (avyak®ta), is alluded to at a number of places in both works24. Similarly, the teaching that dharmas are beyond existence and non-existence from the Kaccayanagotta sutta plays a prominent role in the Karika25 and also can be seen a number of places in the Ratnavali26. We also find allusions in both works to the Buddha's reluctance to teach as told in the Ariyapariyesana sutta27. That any Buddhist of the early centuries of the Common Era would allude to these sutras is not unusual, but the way that Nagarjuna’s two texts employ these two sutras to justify the teaching of emptiness seems to be a distinguishing feature. There is, however, a reference to a sutra in both the Karika and the Ratnavali which seems to have been unknown even to the early Madhyamika tradition: Karika 18.6 “The Buddhas have provisionally employed the term atman and instructed on the true idea of anatman. They have also taught that any … entity as atman or anatman does not exist.”28 23 Hopkins: “Buddhist Advice for Living and Liberation: Nagarjuna’s Precious Garland Ithica: Snow Lion Press, 1988)”, 103. Also see verse 88. 24 Ratnavali verses 73, 105-6, 115, and Karika 22.14, and chp. 27. 25 It is mentioned by name at Karika 15.7 26 Ratnavali verses 38, 42, 46, and 71. 27 Ratnavali verse 103 and Karika 24.12. 28 Poussin, Prasannapada 355, “atmety api prajñapitam anatmetyapi desitaµ| buddhair natma na canatma kascid ity api desitaµ||” NAGARJUNA AND THE RATNAVALI 219 Ratnavali 103: “Thus neither the self nor non-self are said to be apprehended as real. Therefore the Great Subduer rejected views of self and nonself29. Whenever Nagarjuna says something to the effect of, “the Buddha says…” the Indian commentaries assume that he has a specific sutra in mind. Of the three earliest extant commentaries, the Akutobhaya and the Buddhapalitav®tti are the most conscientious about identifying the source of Nagarjuna's references. The curious fact about their comments on Karika 15.6, however, is that, while they both assume that Nagarjuna is referring to a specific scripture here, they nevertheless seem hard-pressed to identify it. They both quote the “Saleyyaka Sutta” of the Majjhima Nikaya as the source of this Karika verse. The text which they both quote is from a sermon in which the Buddha is explaining to a group of Brahmins which activities of body speech and mind lead to good destinies and which lead to foul. Among the thoughts leading to a foul rebirth are the thoughts: “this world does not exist. The other world does not exist. Beings who are spontaneously produced do not exist, etc.”30 The Akutobhaya and Buddhapalitav®tti take this passage as describing different dispositions of converts (gdul bya = vineya) upon entering the order. The teachings of self and non-self, then are to be seen as antidotes to a specific false view. This is a bit of a commentarial stretch considering the passage's original context. The Saleyyaka Sutta never mentions atman and anatman as beliefs to be abandoned. The question remains why these early commentaries didn't find a better proof-text. Certainly, stanzas 22, 93 or 154 of the Suttanipata’s “A††hakavagga” would have been a better choice. An answer is suggested when we consider the commentaries of Bhavaviveka 29 Hopkins, 109. Cp. Hahn, 40. C.W. Huntington “The ‘Akutobhaya’ and Early Indian Madhyamaka” (Ph.D. dissertation, University of Michigan, 1986), 432. “gdul ba [sic] gan dag la ’jig rten ’di med do ’jig rten pha rol med do|| sems can brdzus te skye ba med do sñam pa’i lta ba de lta bu byun bar gyur pa de dag gi bdag med par lta ba bzlog pa’i phyir bdag go zhes kyang gtags par gyur to||” Buddhapalitav®tti Peking Tripitaka vol. 95, #5242, p. 273b: “de la gdul bya gan dag la ’jig rten ’di med do| ’jig rten pha rol med do| sems can rdzus te skye ba med do sñam pa’i lta ba de lta bu byun bar gyur ba|” Cp. Majjhima Nikaya I. 287. “…natthi ayaµ loko natthi paro loko natthi mata natthi pita natthi satta opapatika…” 30 220 JOSEPH WALSER and Candrakirti. Neither Bhavaviveka nor Candrakirti identify the Suttanipata as the source of this quote. Both consider its source to be a Mahayana text, although they identify two different texts. Bhavaviveka quotes from the Suvikrantavikramin Sutra31, while Candrakirti quotes from the Kasyapaparivarta Sutra32. What is significant here is the textual histories of these two sutras. According J.W. de Jong, the former text is fairly late — the terminus ante quem coinciding only with the dates of Bhavaviveka (sixth century)33. In other words, there is no evidence that the sutra existed prior to Bhavaviveka who mentions it in the sixth century, and hence it is unlikely that Nagarjuna's commentators (much less Nagarjuna himself) could have quoted from it. The story is different with the Kasyapaparivarta. It is, by all accounts, one of the oldest Mahayana texts, or at least it is one of the earliest to have reached China. The oldest translation into Chinese is ascribed to a certain Lou-jia-chan ( ) dur34 ing the second century AD . Hence, it is historically quite possible that this is the sutra to which Nagarjuna is referring. The passage in question, however, does not occur in this earliest translation35. It does occur in the next extant translation (anonymous) finished 31 Prajñapradipa, Peking Tripitaka vol. 95, #5253, p. 233a: “de ltar yan ji skad du rab kyi rtsa la gyis rnam par gnon pa| gjugs ni bdag gam bdag med pa ma yin no| de bzin du tshor ba dan| ’du ses dan| ’du byed rnams dan| rnam par ses pa yan| bdag gam bdag med pa ma yin no|” 32 Poussin, Prasannapada, 358.10: “yathoktam aryaratnaku†e| atmeti kasyapa ayam eko’ntaÌ| nairatmyam ity ayaµ dvitiyo’ntaÌ| yad etad anayorantayor madhyaµ tadarupyam anidarsanam apratiÒ†ham anabhasam- avijñaptikam aniketam iyam ucyate Kasyapa madhyama pratipaddharma∞aµ bhutapraty avekÒeti||” which is virtually identical to a passage in the Kasyapaparivarta. Cp. The Kasyapaparivarta: A Mahayanasutra of the Ratnaku†a Class, Baron A. von Stäel-Holstein, ed. 1926; (Tokyo: Meicho-Fukyu-Kai, 1977), 87, paragraph 57. Translation in Chang, ed. A Treasury of Mahayana Sutras: Selections from the Maharatnaku†a Sutra (Delhi: Motilal Banarsidass Publishers, 1996), 394. “Ego is one extreme, egolessness is the other, and [the two-in-one of] ego-egolessness is the middle, which is formless, shapeless, incognizable, and unknowable. [To realize] it is called the middle way, the true insight into all dharmas.” 33 J.W. de Jong “Notes on Prajñaparamita texts: The Suvikrantavikramiparip®ccha” in Prajñaparamita and related Systems: Studies in honor of Edward Conze, L. Lancaster, ed. (Berkeley: Berkeley Buddhist Studies, 1977) 187. 34 Stäel-Holstein, ed., Kasyapaparivarta, ix. 35 Ibid., 87, paragraph 57. NAGARJUNA AND THE RATNAVALI 221 sometime between 265 and 420 AD36. If Nagarjuna is indeed referring to this passage, then we have to conclude that during the first few centuries of the common era, some manuscripts of the Kasyapaparivarta contained this verse and some did not. Whether or not Nagarjuna was referring to this verse or one from the Suttanipata, the case of the Kasyapaparivarta is illustrative of the status of many texts in early India. Buddhist monks had access to Buddhist scriptures, but not all Buddhist monks had access to all Buddhist scriptures. And, just because a monk had access to a Buddhist scripture, we cannot assume that he had access to the same version that was available to other monks. The fact that Nagarjuna refers to a scripture with which other members of the early Madhyamika school were unfamiliar means that access to his version of that scripture was limited to a few members of the early school — perhaps even limited to Nagarjuna himself since Aryadeva makes no references to this passage. The fact that the Ratnavali refers to a sutra of which other early Madhyamikas seem to have been unaware increases the likelihood that Nagarjuna wrote it. 1.4. Poetic Style of the Ratnavali The final aspect of Nagarjuna's work that I would like to introduce is the issue of his poetic style. The main work on this issue has been done by Tilmann Vetter in a 1992 article analyzing the statistics of the Ratnavali’s metrics and use of conjunctions in comparison with the same statistics from the Karika. His findings are, not surprisingly, inconclusive. The metrics of the Ratnavali do not diverge significantly from those of the Karika37, and while the use of certain particles38 and compounds does differ significantly39, he nevertheless concludes: 36 Ibid., ix “The total number of vipula forms in the Karikas is 160, which is 18% of a total of 884 lines. The 14.4% in the Ratnavali does not diverge significantly from this figure, though the higher number of ra-vipula in the Karikas and the occurrence of other vipula forms should be kept in mind.” T. Vetter, “On the Authenticity of the Ratnavali,” Asiatische Studien 46.1 (1992): 501. 38 “Ca, eva, api, iti, va, punaÌ and tu” Ibid. 501. 39 Vetter finds that the density of particles in the Sanskrit fragments of the Ratnavali is about half of their density in the Karika. Further, in the Karika 79% of the verses do 37 222 JOSEPH WALSER Concluding these remarks on style we might state: The observations are not so strong as to force us to deny the authenticity to the Ratnavali, but if it was composed by Nagarjuna, it is difficult to imagine that it was written in the same period as the Karikas.“40 There is nothing in Vetter's statistics to seriously challenge Nagarjuna's authorship of the Ratnavali, and in fact his analysis provides us with an important suggestion. If the Ratnavali was written later in Nagarjuna's life than the Karika we might be able to explain some of the slight divergences between the two texts. It should be remembered that Sanskrit was probably a secondary language for Nagarjuna, and certainly the highly stylized metrical version used in his works was developed over years of practice. In ordinary speech, the use of compounds would have been less frequent — the conjunctive task being taken over by particles. As the author's poetic style developed over the years, the facility with making compounds would presumably increase. Vetter's statistics, then, do seem to indicate that the Ratnavali is a more mature work poetically if not philosophically. If, then, Nagarjuna did write the Ratnavali he probably wrote it some years after the Mulamadhyamakakarika. This hypothesis gains support when one considers that there are at least two arguments occurring in the Ratnavali that do not occur in the Karika. The first of these concerns the doctrine of momentariness. Ratnavali verse sixty-three begins a discussion of the three times. The argument is similar to those in the Karika until verse sixty-six, when the discussion shifts to the status of the moment (“kÒa∞a”). Verses sixty-six through seventy refute the possibility of momentariness in much the same way as each of the three times is refuted in the Karika. This argument is significant in light of the importance that this notion would play in the future of Buddhist philosophy (especially in the works of Dignaga, Dharmakirti and Ratnakirti) and in light of the fact that the concept is wholly absent from the Karika. The other argument in the Ratnavali that goes beyond the Karika is the argument asserting that the object of desire must be a false not contain compounds, while in the Ratnavali only 51.1% do not contain compounds. (Vetter, 503) 40 Vetter, 504. NAGARJUNA AND THE RATNAVALI 223 construction since the image one attaches to is unitary while the senses that actually perceive it are five-fold41. The latter argument seems to have been picked up by Aryadeva (in his CatuÌsataka vs. 268), although he avoids arguments against momentariness in the Satakasastra. There can be little question, however, of Aryadeva having written the Ratnavali. While Vetter's statistical analysis of the Ratnavali’s style is inconclusive concerning Nagarjuna's authorship, it nevertheless does rule out Aryadeva as the author. It may be noteworthy that the 303 lines of the Sanskrit fragments of Aryadeva's CatuÌsataka as edited by Karen Lang… contain only a percentage of 2.3% vipula (7 on a total of 303 lines), and only ma-vipula. Aryadeva, so it seems, may be safely eliminated as a possible author of the Ratnavali…42. In all, then, the evidence supporting Nagarjuna's authorship of the Ratnavali is strong. It is ascribed to Nagarjuna by a number of sources beginning in the sixth century and shows an affinity for common Madhyamika doctrine. Finally, the Ratnavali contains many of the peculiar stylistic elements found in the Karika which are not found in other authors of the early Madhyamika school — such as Aryadeva, Buddhapalita and the author of the Akutobhaya. 2. Nagarjuna's Danapati was a Satavahana king The second sub-thesis to be established in dating Nagarjuna is that Nagarjuna's danapati and benefactor was a Satavahana king. There are two factors that I would like to offer into evidence in support of this. First, 41 Ratnavali, vs. 351 When [all] five senses, eye and so forth [Simultaneously] apprehend their objects A thought [of pleasure] does not refer [to all of them] Therefore at that time they do not [all] give pleasure. 352 Whenever any of the [five] objects is known [As pleasurable] by one of the [five] senses, Then the remaining [objects] are not so known by the remaining [senses] Since they then are not meaningful [causes of pleasure]. Translation from J. Hopkins Buddhist Advice, 140] Cp. Hahn, 112. 42 Vetter, 501. 224 JOSEPH WALSER the earliest and latest dates for Nagarjuna coincide almost exactly with the range of dates for the Satavahana dynasty. Second, the way that the hagiographical tradition about Nagarjuna appears to have developed points to his association with a Satavahana king as one of its earliest elements. 2.1. Earliest and Latest Dates That Nagarjuna lived during the reign of a Satavahana king must be admitted as a possibility when the factors establishing his earliest and latest dates are considered. Obviously, Nagarjuna is writing at a time when the early Mahayana sutras have already been written. Since the earliest Prajñaparamita sutras are estimated to have been written around 100 BC. we may take this to be an earliest limit date for Nagarjuna. On the other end, the earliest of the datable external sources mentioning Nagarjuna are several translations of the DasabhumikavibhaÒa, attributed in their colophons to Nagarjuna. According to Lamotte: … the Chinese catalogues list among the works translated by DharmarakÒa at Ch'ang-an, between A.D. 265 and 313, a P’u-sa hui-kuo ching . This translation is noted in the Ch’u (T 2145, ch 2, p. 8b 17), and the Li (T 2034, ch.6, p. 63a 23) which remark: “The colophon says that this is an extract from the Dasabhumikasastra of Nagarjuna”. It therefore results that a work by Nagarjuna had already reached China about A.D. 265. 43 Whether or not Nagarjuna actually wrote the Dasabhumikasastra, does not change the fact that two catalogues (both from the sixth century A.D.) record that a work was ascribed to someone named Nagarjuna by 265 A.D. at the latest. This then is the earliest recorded date of an external source mentioning Nagarjuna's name, and as such provides us with a date by which Nagarjuna must have been an established scholar. 2.2. Testimony of Kumarajiva's school A third century date is confirmed in the writings of Kumarajiva and his school. Kumarajiva indicates a third century date for Nagarjuna's death 43 É. Lamotte, The Teaching of Vimalakirti (Vimalakirtinirdesa), Sara Boin, trans. (Oxford: The Pali Text Society, 1994), xcvii. NAGARJUNA AND THE RATNAVALI 225 in a statement at the end of his translation of Nagarjuna's “Biography”44 which claims that, “From that leave taking [i.e., from Nagarjuna's death] until today one hundred years have passed”45. Arguably, the “today” referred to is the time of Kumarajiva's translation of the text. According to Robinson. It would be hard to defend every item in the Biography, but it is easy to show that in substance it represents Kumarajiva's account. Seng-jui mentions the Indian Chronicle(s) (t’ien-chu-chuan), which probably means the biographies narrated by Kumarajiva. Hui-yuan's biographical sketch of Nagarjuna in his Preface to the Great Perfection of Wisdom Treatise agrees with the Biography and many of his allusions are intelligible only with a knowledge of it. Seng-jui mentions the existence of temples to Nagarjuna and AsvaghoÒa, unfortunately without the date that occurs in the Biography. But the literary form and style of the Biography are typically Chinese. It has the standard opening, which states the man's native region and class, and then indicates that the child was precocious and received a good education. The laudatory cliches are purely Chinese and transparently do not stand for Indic originals. Insofar as it is genuine, this Biography must consist of Kumarajiva's oral account as worded by his disciples…. In this case, the point one hundred years after Nagarjuna's death would be sometime during Kumarajiva's residence at Ch'ang-an (A.D. 401-13). Thus Nagarjuna would have flourished in the third century A.D46. The other set of dates for Nagarjuna comes from a disciple of Kumarajiva named San-jwei (So-yei), who places Nagarjuna at the end of the time of the xiang-fa ( = dharma pratirupaka or “Semblance dharma”47). Correlating this information with the dates of Aryadeva recorded by another disciple of Kumarajiva, Ui comes up with a date of “about 113-213 44 TaishoShinshu Daizokyo (hereafter, T.). 2047, lit. “The Chronicle of the Bodhisattva Nagarjuna.” In the rest of this article, I will refer to it simply as the Biography. 45 R. Corless, “The Chinese Life of Nagarjuna,” in Buddhism in Practice, Donald Lopez, ed. (Princeton: Princeton University Press, 1995), 531. 46 R. H. Robinson, Early Madhyamika in India and China, reprint (Delhi: Motilal Banarsidass, 1978), 25. 47 Here I am following Jan Nattier's translation of the term. See J. Nattier, Once Upon a Future Time: Studies in a Buddhist Prophesy of Decline (Berkeley: Asian Humanities Press, 1991), 86-9. 226 JOSEPH WALSER A.D.” for Nagarjuna48. Though this testimony relies on some rather strained calculations, it does suggest that Nagarjuna may have lived in the third century A.D49. The period between the first century B.C. and the third Century A.D. roughly corresponds to the dates for the Satavahana dynasty (the dynasty ends sometime in the first two quarters of the third century). Hence, that Nagarjuna lived during the time of Satavahana dynasty is a strong possibility. 2.3. Nagarjuna's Letters The oldest extant sources testifying to Nagarjuna's connection with the Satavahana dynasty surround two works — the Suh®llekha and the Ratnavali. According to tradition, Nagarjuna wrote these as letters to his patron king. The translations into Chinese and Tibetan are fairly consistent in naming this king. The earliest extant translation of the Suh®llekha, translated by Gu∞avarman sometime after 43150. Presumably, it is 51 Gu∞avarman who gives it the title , which may be rendered as “The Essential Verses (gatha) on Dharma Explained By the Bodhisattva Nagarjuna to King Shan-ta-ka.” This name for the Suh®llekha’s addressee can also be found in the 7th century in Yijing's Nanhai ji-gui nei-fa chuan, where the king is named “ ,” (“shi-yinde-ka”)52. It is possible that both of these translate “Dhanya(ka)†aka” (modern Amaravati) the name of an important Satavahana site in the eastern Deccan. If this is the case, Gu∞avarman and Yijing are telling us 48 H. Ui, The VaiseÒika Philosophy According to the Dasapadartha-Sastra, 2nd ed. (Varanasi: Chowkhamba Sanskrit Series Office, 1962), 43. For Ui's discussion of Nagarjuna's date, see pages 42-46. 49 Using a similar method, one could try to come up with a date for Nagarjuna's birth based on the testimony of works such as the Lankavatara Sutra, the Mahamegha Sutra, or the Mañjusrimulatantra, which claim that Nagarjuna was born 400, 700, and 400 years after the Buddha's Parinirva∞a respectively. Unfortunately, since we know nothing of the authors of these texts, we do not know when they thought the Buddha's parinirva∞a was. Hence, these dates are of little use. 50 Gu∞avarman was born in India in 367 and arrived in China in 431. Source: P. Demiéville, et al., Répertoire du Canon Bouddhique Sino-Japonais, Fascicule Annexe du Hobogirin (Paris, L'Académie des inscriptions et Belles-Lettres, Institut de France, 1978), 252 (q.v. “Gunabatsuma”). 51 T. 1672, 745b. 52 T. 2125, 227c 14-15. NAGARJUNA AND THE RATNAVALI 227 important information concerning the king's capital. Yijing also claims that this king is a Satavahana (Sha- duo-pó-han-na , which he 53 translates as ). Between 560 and 57054, Paramartha translated the Ratnavali into Chinese, though he does not name the author. He does, however, mention its addressee. The title of this translation in Chinese is which can be translated as “Treatise on the Precious Course [Delivered to] King “Righteous.” In this same vein Xuanzang's use of Yin-zhèng (“leading to righteousness”) a century later to translate the Sanskrit name 55 Satavahana (Sha-duo-pó-he ), suggesting that Paramartha may also be using the character “zheng” (“righteousness”) as a (spurious) translation of “Sata” (reading it as being somehow derived from the Sanskrit “sat” = “truth” or “righteousness”) to designate the king to whom the Ratnavali is addressed as King “Sata[vahana]”. A better explanation, though a more complicated one is that Paramartha does use the character to translate the sound “Sata”, but that this is not to indicate the Satavahana dynasty but rather one of the many Sata (Prakrit = “Sada”) kings. There are quite a few Sata/Sada kings mentioned in inscriptions found in inscriptions from Andhra Pradesh. Inscriptions and coins mentioning these kings have been found at Chebrolu, Dhanyakataka, Ramatirtham, Guntupalli, Vaddamanu, Nandayapallem and Velpur56. The identity of these kings is a matter of some debate. Some scholars consider the kings whose last name ends in “Sada” to be rulers in the Satavahana lineage. Others consider them to belong to another dynasty. The debate over this issue seems to revolve around an inscription found at Guntupalli, a village in West Godavari District. The inscription reads as follows: Maharajasa Kalinga(Ma)Hisakadhipatisa MahaMekhavahanasa Siri Sadasa lekhakasa CulagoMasa ma∂apo danaµ 53 Ibid. Mabbett, using Pulleyblank, renders this into its Central Middle Indic equivalent as “sa-ta-ba-xa-nah”. See Mabbett, 336. 54 For a brief biography of Paramartha, see P. Demiéville, (1978), 276. 55 T. 2087, 929a 27. 56 B.S.L. Rao et al. Buddhist Inscriptions of Andhradesa (Secunderbad: Ananda Buddha Vihara Trust, 1998), 54. 228 JOSEPH WALSER “Gift of a Mandapa by Cula Goma, the scribe of Maharaja Siri Sada who belonged to the dynasty of Mahameghavahana and had the title KalingaMahiÒakadhipati”57 This inscription clearly establishes a connection between the Sata kings and Mahameghavaha Kharavela of the Hathigumpha inscription and mentions the extent of his kingdom (namely, the area of Kalinga). D.C. Sircar suggests that the name was Sata, indicating that this king was born to a Satavahana princess, but the form Sada often appears on Satavahana coins and hence is not necessarily a matronym. On the other hand, if we include the Sata kings in the Satavahana dynasty, we have to posit a collinear rule. Whether they were either independent from or under the suzerainty of the Satavahanas, the Sata kings seem to have been confined to coastal Andhra58 throughout their reign which was roughly coterminous with that of the Satavahanas. In short, for the purposes of finding a date and location of Nagarjuna, it will not matter much whether his patron king was a Sata king or a Satavahana as the time period and geographical range coincide with the most important evidence from the Ratnavali (see below). It is likely that Paramartha, being from Ujjain, would have had access to important texts coming from coastal Andhra Pradesh since the two areas were culturally well connected and well traveled since the second century at least. Further, he would have passed through Kalinga on his way to China (he took a sea route). In the Tibetan translations of these works the addressee of these letters is translated as “bDe spyod,” (“good conduct”) in the Ratnavali†ika by Ajitamitra59, as well as in the colophon to the Tibetan translation of the Suh®llekha by Sarvajñadeva60. The meaning of this is word is so close 57 Rao, 109. There are a total of eight Sada/Sata kings mentioned in inscriptions: Sri Sada(sata), Sivamaka Sada (Vaddamanu), Manasada, Mahasada, Asaka Sada, Aira Asaka Sada, Siri Mahasada and Siva Sada. Concerning their territory I.K. Sarma identifies MahiÒaka with the Maisolia region. (Rao, 109-10) 59 Ajitamitra, in the beginning of his commentary on the Ratnavali’s says: “de la ’dir btsun pa ’phags pa klu sgrub ’jig rten mtha’ dag la phan par bzed pas rgyal po bde spyod kyi dban du mdzad nas dam pa’i chos rin po che’i phren ba dgod pa’i nes pa mdzad de dam pa’i spyod pa dan mthun par|” Yukihiro Okada, Die Ratnavali†ika des Ajitamitra, (Bonn: Indica et Tibetica Verlag, 1990), 1. 60 Nagarjuna, Golden Zephyr: Instructions from a Spiritual Friend, L. Kawamura, trans., (Berkeley: Dharma Publishing, 1975), 93. 58 NAGARJUNA AND THE RATNAVALI 229 to Xuanzang's translation for Satavahana (“leading right”) that one cannot overlook the possibility, that it translates Satavahana.61 Most scholars take this to translate the name “Udayana” following Scheifner62, but since there are no Satavahana kings by that name either in the Pura∞ic accounts or in any inscription discovered so far, it is more likely that it is a translation of the name of the dynasty itself. Thus, from the colophons of these translations, we have Nagarjuna's patron identified as one of the Satavahanas whose personal name was something like “Jantaka.” This personal name of Nagarjuna's king is quite common in later Tibetan literature as well. While Mabbett thinks that this may be a version of the surname Satkar∞i, so common among members of the Satavahana dynasty63, this reconstruction cannot account for the fact that both Gu∞avarman and Yijing explicitly represent a nasal sound in their transliterations. Again, it is more likely to be the place name, Dhanya(ka)taka. 2.4. The Elements of Nagarjuna's Hagiography This general agreement among the translators of Nagarjuna's letters about the identity of Nagarjuna's patron king needs to be placed in the larger context of legends about Nagarjuna. Since none of the translators lived during the life of Nagarjuna, we must consider the possibility that their sources of this attribution are the legends about Nagarjuna that were circulating at the time of translation. Therefore, we must assess the hagiographical tradition surrounding Nagarjuna before we can assess the testimony of these translators who likely drew upon it. The earliest extant legends about Nagarjuna were translated by Kumarajiva into Chinese in about 405 C.E64. After that, legends proliferate in Buddhist, Hindu, Siddha, and Jain sources. Although I discuss a number of these sources in what follows, this will not be an exhaustive review of 61 This was suggested by, J.W. de Jong, review of J. Hopkins and Lati Rimpoche, trans., The Precious Garland and the Song of the Four Mindfullnesses (London, 1975), in Indo-Iranian Journal 20 (1978): 137. 62 Scheifner trans., Taranatha’s Geschichte des Buddhismus in Indien (St. Petersburg, 1869), 2 note 2. 63 Mabbett, 341. 64 T. 2047. 230 JOSEPH WALSER all the legends told about Nagarjuna. Much of the bibliographic spadework and analysis of this material has already been done by Mabbett and others65. This article offers instead a new interpretation of the evidence already available. Legends of Nagarjuna were compiled for over a thousand years in Sanskrit, Chinese and Tibetan. When these legends are taken as a group the diversity and range of the stories is somewhat daunting. Even if we only look to these legends for information about Nagarjuna's patron or place of residence, we are left with a number of problems. While most of our sources mention that Nagarjuna's patron was a Satavahana66, there are two dissenting voices in this regard. The first, the Kathasaritsagara (eleventh century) by Somadeva Bha††a, is a reworking of an earlier B®hatkathamañjari of KÒemendra (also eleventh century), and the second is the Rajatarangi∞i by Kalha∞a. The former work seems to be oblivious of any connection between Nagarjuna and a Satavahana king insofar as it has one section of stories devoted to King Satavahana and a separate section for stories related to Nagarjuna, who in turn is the associate of a King Cirayus (“Long-Life”). No place-name is associated with Nagarjuna in this work. The Rajatarangi∞i by Kalha∞a is a court history of Kashmir that is often discussed in modern works on Nagarjuna. Kalha∞a 65 In addition to Mabbett's article, I recommend Phyllis Granoff's, “Jain Biographies of Nagarjuna: Notes on the Composing of a Biography in Medieval India,” in Monks and Magicians: Religious Biographies in Asia, Phyllis Granoff and Koichi Shinohara eds. (Oakville, Ont.: Mosaic Press, 1988), 45-61, and David G. White's Alchemical Body: Siddha Traditions in Medieval India, (Chicago: Univ. of Chicago Press, 1996), 62-77. 66 This is mentioned by Xuanzang. See, T. Watters., On Yuan Chwang’s Travels in India 629 A.D. – 645 A.D (New Delhi: Asian Educational Services, 1988), 201; Ba∞a see Ba∞a, The HarÒa-Carita of Ba∞a, Cowell and Thomas trans. (Delhi: Motilal Banarsidass, 1961), 252; a number of Jain sources including the Prabandhakosa, see Prabandha Kosa, Jina Vijaya, ed., (Santiniketan: Adhis†hata-singhi Jaina Jñanapi†ha, 1991), 84; and the Prabandha Cintama∞i, see Prabandha Cintama∞i of Merutungacarya, Jinavijaya Muni ed. part I, (Santiniketan: Adhis†ata Singhi Jaina Jñanapi†ha, 1933), 119; Abhayadatta's Lives of the 84 Siddhas, see Abhayadatta, Masters of Mahamudra: Songs and Histories of the Eighty-Four Buddhist Siddhas, K. Dowman, trans. (Albany: State University of New York Press, 1984), 115; the Rasendra Mangala, see, White 155; Bu-ston, see Bu-ston, History of Buddhism in India and Tibet, Eugene Obermiller trans. Bibliotheca Indo-Buddhica, no. 26 (Delhi: Sri Satguru Press, 1932, 1986 reprint, 127; and Taranatha, see Taranatha, Taranatha's History of Buddhism in India, D. Chattopadhyaya, trans. (Calcutta: K.P. Bagchi & Co., 1970) 109. NAGARJUNA AND THE RATNAVALI 231 mentions Nagarjuna as living at ∑a∂arhadvana67 during the reign of either HuÒka, JuÒka, or KaniÒka. When we come to the issue of Nagarjuna's place of residence, the legends are much more diverse. Kumarajiva's translation of Nagarjuna's legends mentions a rather vague “South India” (presumably “DakÒinapatha”) a number of times and also mentions that he spent a brief period in the Himalayas68. Some (fifth century) versions of the Lankavatara Sutra69 (and the Mañjusrimulatantra70) claim that a monk whose name sounds like “Naga” will live in Vidarbha71. Xuanzang has Nagarjuna living 300 li to the south-west of the capital of southern Kosala at a mountain called “Black Peak,” or “Black Bee”72. Candrakirti in his commentary on 67 On this site see Stein's note: “∑a∂arhadvana, ‘the wood of the six Saints,' if rightly identified by the glossator as (Harvan grame), is the modern village Harvan, situated about one and a half miles to the N.W. of the gardens of Shalimar near Srinagar. On the hillside to the south of Harvan ancient remains have come to light in the shape of highly ornamented brick pavements, which were dug up in the course of excavations conducted at the site in connection with the new Srinagar waterworks.” See M.A. Stein, Kalha∞a’s Rajatarangi∞i: A Chronicle of the Kings of Kasmir (Srinagar: Verinag Publishers, 1961), 31 note 173. Could “ ∑a∂arhadvana” possibly be used in this story because it is a homonym for “Satavahana?” 68 Corless, 528. 69 This passage does not appear in the earliest version of the Lankavatara translated by Gu∞abhadra in 443 C.E.. It does appear in the versions translated by Bodhiruci (trans. 513 C.E.) and SikÒananda (trans. 700-704). The passage in question, according to Walleser, may possibly have been added in the fifth century C.E. because the section in which it appears contains a verse referring to Maurya, Gupta, and Nanda kings of the Kali Yuga. 70 Translated by J. Hopkins in Buddhist Advice for Living, 13. There are only three substantial differences between this prophecy and that of the Lankavatara Sutra: 1) the number of years that he appears after the Buddha's parinirva∞a increases to 400 years and his life-span increases to 600 years, 2) no place name is indicated, 3) he is the transmitter of the Mahamayuri mantra. The Lankavatara is probably the earlier of the two, and what can be said of it can also be said of the Manjusri Mulatantra as far as its testimony of Nagarjuna is concerned. 71 Hopkins cites 19th century Mongolian scholar Nga-wang-bel-den (b. 1797) who in his discussion of Jam-yang-shay-ba's work, “gives be da (misprinted as pe da) and identifies the place as Vidarbha (be dar bha).” [Hopkins, 10, note a]. Alternately, P.S. Shastri suggests that this “Vedalya” could also be “Dehali” which is a site near Nagarjunako∞∂a, the site of Vijaya Satkar∞i's capital. See I.K. Sharma Studies in Early Buddhist Monuments and Brahmi Inscriptions of Andhradesa, (Nagpur: Dattsons Publishers, 1988), 17. Also see, Mabbett, 335, note 32. 72 Watters, 201. Watters, by using two different Chinese glosses, reasons that “Po-lomo-lo-ki-li” is probably a transliteration of Bhramara-giri (Bee-peak) which is confirmed 232 JOSEPH WALSER Aryadeva's CatuÌsataka says that Aryadeva became Nagarjuna's disciple after travelling in South India, perhaps indicating that Nagarjuna lived there too73. The Jain tradition74 (which is also echoed by Al-beruni75) consistently places Nagarjuna at Mt. Satrunjaya in Gujarat76, while the Buddhist and Siddha traditions consistently place him at Nalanda, Sriparvata77, Kancipuram78, Dhanyakataka79, Godavari80, and Vidarbha. If we add Kalha∞a's assertion that Nagarjuna lived in Kashmir, then we have to admit that Nagarjuna could have lived virtually anywhere in India. Indeed, the range of dates and the conflicting traditions concerning Nagarjuna's residence and royal patronage have led many to dismiss some of these sources or all of them. For instance, in his article, “Sur la formation du Mahayana,” Étienne Lamotte complains: A la tradition qui fait de Nagarjuna un sujet des souverains Satavahana, on peut opposer le témoignage de la chronique cachemirienne qui le rattache aux rois TuruÒka du Nord-Ouest, HuÒka, JuÒka et KaniÒka et lui assigne comme résidence le Bois des six Arhat près de Harwan au Kasmir. Le Kosala méridional n'était point seul à posséder un Sriparvata, c'est- à -dire, en sanskrit, un Mont Sacré: toponyme extrêmement répandu que la Mahabharata et les Pura∞a appliquent à de nombreuses montagnes et qui désigne notament un site du Kasmir. En ce qui concerne Nagarjuna, il est scientifiquement incorrect de retirer de leur contexte pour les grouper artificillement ses prétendues attaches avec le pays Andhra. Les biographies et notices qui lui sont consacrées fourmillent de légendes, plus ahurissantes les unes by the (“Black Bee”) translation. He cites Beal's reasoning that “Black Bee” is a synonym for the Goddess Durga or Parvati, and hence, Po-lo-mo-lo-ki-li is some form of “Parvata” (literally meaning “mountain”). James Burgess, following this lead, identifies Nagarjuna's abode with Sri-Parvata, a well-known mountain on the Krishna River in modern Andrha Pradesh. Watters, 208 73 See Lang, 7. 74 Jain legends of Nagarjuna have been discussed extensively in Granoff, op. cit. 75 Alberuni (writing in 1030), mentions that Nagarjuna lived at a Gujarati site, “Fort Daihak” near Somnath, one hundred years previous to his writing. See Mabbett, 338. 76 This is called “∆hanka” in the Prabandha Cintama∞i p. 119, and “Satruñjaya” in the Prabandhakosa p. 84. 77 Bu-ston, 127. 78 Abhayadatta, Buddha’s Lions: The Lives of the Eighty-Four Siddhas, J. Robinson, trans. (Berkeley: Dharma Publishing, 1979), 75 79 As both Bu-ston and Taranatha assert. 80 This attribution can be found in the Tantra Mahar∞ava. See White, 113. NAGARJUNA AND THE RATNAVALI 233 que les autres et qui concernent au moins quatre Nagarjuna différents...81 [italics mine] If the reports of the later traditions conflict, the question at this point is what to do with the testimony as it has come down to us in these traditions. Contemporary Buddhist scholars lean toward a kind of academic agnosticism when it comes to looking for historical evidence among legendary materials. As in Christianity's “Search for the Historical Jesus,” the “Search for the Historical Buddha,” has told us much more about the early compilers of the Buddhist suttas than about the Buddha himself. Recently, Jonathan Walters has proposed four ways of reading legendary materials, the first two of which are the “Historical Source Mode” (reading the texts for historical information about the subjects which they relate), and the “text of its day mode” (reading texts for historical information about the compilers/readers of the texts). The first of these modes is much maligned in his article in that it assumes that the authors of these legendary texts, “were somehow trying to objectively report historical facts in a would-be nineteenth-century European way.” He concludes, “So long as this assumption remains operative, there is nothing to do except judge the suttas as though they had been compiled by Edward Gibbon; and given that they were not, the impasse reached by scholarship in this mode seems inevitable”82. Although there may be very good reasons to assume an impasse in the “Search for the Historical Buddha,” the same need not be assumed at the start for all figures in the history of Buddhism. This is, of course, not to say that we should read Nagarjuna's legends as if they were BBC press releases. As rich as these legends are, they yield little in he way of historical evidence about Nagarjuna. Nevertheless, I believe that some historical information can be recovered from these texts if we can imaginatively put ourselves in the world of the writers of these texts. In order to interpret these legends, the most productive position is to assume that all pieces of information in the legends were included for a 81 Étienne Lamotte, “Sur la formation du mahayana,” in Asiatica: Festschrift für Friedrich Weller (Leipzig: Otto Harrowitz, 1954), 388. 82 J. Walters, “Suttas as History: Four Approaches to the Sermon on the Noble Quest (Aryapariyesana Sutta),” History of Religions (1999): 259. 234 JOSEPH WALSER reason. The purposive element will be stronger for those elements of the story that occupy a prominent place in the narrative. For those who are uncomfortable with the “intentional fallacy,” I will say merely that we must impute a purposefulness or a strategy to the text in order to interpret it in its historical context. In this way of reading the Nagarjuna legends, we must start with Walter's “text of its day mode.” In general, hagiographers compose their stories with two purposes in mind, spiritual edification and institutional legitimation. Elements of hagiographies put there for spiritual edification tend to echo or illustrate themes found in scripture, such as acts of altruism (Nagarjuna offering up his head upon request in a number of these legends, echoes the kind of radical giving found in the Vessantara Jataka and a number of Mahayana Sutras). Elements of hagiographies put there for legitimation are sometimes more difficult to spot. These fall into two groups. In some stories, the character of Nagarjuna is placed in juxtaposition to a person, place, or theme that is independently famous. For example, Nagarjuna is often said to reside at a place called Sriparvata. Sriparvata was already famous by the time Nagarjuna legends were being written as a powerful and auspicious place. By locating Nagarjuna there, the character of Nagarjuna takes on some of the (in this case, magical) legitimacy already associated with the site. Legitimation also goes the other way. Once Nagarjuna became famous, his association with pilgrimage sites lent an air of legitimacy (and antiquity) to those sites (we may speculate that this is partly responsible for Nagarjuna's association with Nalanda in some of the post-tenth century legends). If I am correct in describing the rationale for the composition of these stories, then we have a powerful tool with which to look for historical evidence. Any detail which is present in a story for the purposes of spiritual edification or for purposes of legitimation may be hypothesized to tell us more about the compilers of the legend than about the subject of the legend itself. These elements should be read in the “text of its day mode” and should not be assumed to tell us anything about Nagarjuna himself. Note that the existence of such a literary device does not prove that there is no factual basis; an element of a story may serve the plot and also happen to be true. Nevertheless, the presence of such devices should make us question the historical accuracy of the information until we have some NAGARJUNA AND THE RATNAVALI 235 reason to think otherwise. By the same token, if an element of the Nagarjuna legend proves to be an early element in the tradition, and if it does not have an obvious role in edification or legitimation, then we have no choice but to assume that it was included into the hagiographies because it was “common knowledge” to the compilers of these texts. This does not mean that the information is objectively true, but rather that the compilers assumed that it was a given, a fact that their readers probably already knew. To contradict this information even in a legend probably would be equivalent to someone writing a legend about George Washington in which he becomes a benevolent ruler of Thailand. Few would buy it because it goes against what we believe is common knowledge. In the following, I will argue that Nagarjuna's association with the Satavahana king was this kind of information — which may be as close as we can come to “proving” his relation with the dynasty. The Nagarjuna legends are diverse, but the diversity seems to stem from a just a few factors. In the following, I will discuss what I see to be four sources by virtue of which the Nagarjuna legends were legitimated. The first two are traditions, originally independent of the Nagarjuna legend, that were drawn into the Nagarjuna legend. The other two sources are thematic elements that can be found in all of Nagarjuna's legends, which take on a life of their own. Almost every element that occurs in Nagarjuna legends can be attributed to at least one of these four sources, while some of the stories have multiple determinations. 2.5. Other Nagarjunas Other scholars who have tried to sort out the details of the Nagarjuna legend have attempted to solve the problem by postulating more than one Nagarjuna, or many authors using “Nagarjuna” as a nom du plume: one Nagarjuna who was a Madhyamika philosopher, one who was a tantric adept, and one who was a medical practitioner. While this hypothesis should not be accepted without question83, it also cannot be completely 83 The multiple Nagarjuna hypothesis has been most seriously criticized by J. Hua, in his article, “Nagarjuna, One or More? A New Interpretation of Buddhist Hagiography,” History of Religions 10 (1970): 139-53. 236 JOSEPH WALSER dismissed. Clearly, there were a number of people harking to the name Nagarjuna in the history of India. But this does not mean that these “other Nagarjunas” were operating under a pseudonym, any more than the modern Telegu actor named Nagarjuna is (Nagarjuna is still a common name in Andhra Pradesh)84. The fact that there were a number of later Nagarjunas, does not, however, help us sort out the details of Nagarjuna's hagiography. We cannot claim that all of the tantric/alchemical elements of Nagarjuna's hagiography belong to a seventh century “tantric” Nagarjuna when these same elements appear in Kumarajiva's forth/fifth century Biography. Furthermore, works ascribed to a Nagarjuna such as the Yogasataka and the Rasendra Mangala do not claim to be written by the same author as the Mulamadhyamakakarika85 and are easy to distinguish. Thus, for the most part, the assumption of other Nagarjunas will not help us much in sorting out the details of his hagiography. There was, however, one other early Nagarjuna (a Jain) who lived in the early fourth century A.D. who was incorporated into the Nagarjuna legend translated by Kumarajiva. The Jain legend could be a source for Nagarjuna's association with SuraÒ†ra/Gujarat in Jain sources and as well as a source for the stories of Nagarjuna's role in compiling the Mahayana sutras. In Kumarajiva's account of Nagarjuna and a monk in the Himalayas we can discern a borrowing from Jain traditions of the Jain Third Council. This occurs shortly after Nagarjuna is ordained and after he has mastered the Tripi†aka. Then [Nagarjuna] sought other texts, but completely failed, so he went to the Himalayas. In those mountains there was a pagoda, and in that pagoda there was an old bhikÒu who gave him the Mahayana texts86. It is conceivable that this brief detail of Nagarjuna's biography was assimilated into the story from the (Svetambara) Jain Ardhamagadhi canonical 84 White mentions a number of these other Nagarjunas. Xuanzang met one of the disciples of Nagarjuna, “who looked thirty despite his 700 years.” Similarly, there are a number of texts of a much later date written by authors named Nagarjuna. The first of these is the Yogasataka datable to the 7th or 9th century. Similarly, the 14th century Rasendra Mangala is ostensibly by a “Sriman Nagarjuna.” White, 75. 85 White, 164. 86 Corless 528. NAGARJUNA AND THE RATNAVALI 237 text, the Nandisutta, where a Jain Nagarjuna (unrelated87) is said to be the disciple of a master named “Himavat.” 35. Homage to Nagarjuna the teacher who was an able srama∞a of Himavant, and who was the memorizer of the earliest (holy texts) and was the memorizer of the interpretation of the Kalika scriptures. 36. Homage to Nagarjuna the canter, who taught the Ogha sruta, who attained the ability to recite in proper order and who was perfectly acquainted with subtlety and subtle things88. In the Jain tradition, as in the Buddhist tradition, there were four “Councils” to determine or confirm the scriptural tradition. The third of these Councils was held at Valabhi, in the first half of the fourth century and presided over by a monk named Nagarjuna. This Nagarjuna, according to the Nandi Sutta passage quoted above, had been the student of a certain “Himavat (“Snowy”), who entrusted Nagarjuna with the memorization of the early Jain texts and the Kalika sruta (texts which are to be read at a specific time). The Nandi Sutta was probably composed sometime in the fifth century89, but the story obviously dates back to the third Jain council itself. From the above, it seems likely that the Buddhist tradition (recorded by Kumarajiva) that Nagarjuna received an important set of scriptures (the Mahayana sutras) from a monk in the “Himalayas” (lit. “Snowy Mountains”), is borrowed from the Jain tradition that a Nagarjuna, who was a student of “Himavant”, memorized two important sets of texts, the Kalika srutas and the Purva (srutas). If the Jain legend of Nagarjuna is indeed the source of the tradition that places the Buddhist 87 If the Nagarjuna of the Mulamadhyamakakarika is the same as the author of the Ratnavali then we can say that he definitely was not a Jain. Ratnavali verses 61-2 discusses the superiority of Buddhism to Saµkhya, VaiseÒika and Jainism insofar as none of these have a teaching that is beyond existence and non-existence. Similarly, in verse 237 Nagarjuna tells the king not to revere other religious specialists (Tirthikas). 88 A. Mahaprajna, ed. Nandi: Prakrit Text, Sanskrit Rendering, Hindi Translation, Comparative Notes and Various Appendixes, (Ladnun, Rajasthan: Jain Visva-Bharati Institute, 1997), 9. V. 35 “kaliyasuya-a∞u-ogassa dhare dhare ya puvva∞aµ| himaµvatakhamasama∞e vaµde ∞agajju∞ayarië||” V. 36 “miü-maddava-saµpa∞∞e a∞upuµvvi vayagatta∞aµ patte| oha-suya-samayare nagajju∞avaye vaµde||” 89 Natubhai Shah, Jainism: The World of the Conquerors, (Portland: Sussex Academic Press, 1998) 17. 238 JOSEPH WALSER Nagarjuna in the Himalayas, then we have grounds to question the claim that Nagarjuna was there. In later hagiographies of Nagarjuna, the connection with the Himalayas is dropped and Nagarjuna is only said to have received these texts from the Naga kingdom. Nevertheless, the element of the story that claims Nagarjuna to be the bearer of an important class of religious texts remains. In terms of the effect of this connection, on the one hand, the character of Nagarjuna receives some authority by a partial merging with the character of the more recently famous Jain Nagarjuna. At the same time, Kumarajiva's story demotes the status of the Himalayan monk/Himavantacarya, thereby taking legitimacy away from the Jain tradition even as it borrows legitimacy from a Jain saint. Nagarjuna learns what he can from this monk, but is dissatisfied and looks for other Mahayana sutras elsewhere. 2.6. The Mahamegha Prophecy and related Sutras One of the best ways to grant legitimacy to a Buddhist saint is to have his birth and career predicted by the Buddha. This was certainly the idea behind the prophecy about the monk “whose name sounds like Naga” in the Lankavatara Sutra. There is another prophecy that may have factored into the construction of the Nagarjuna legends — a prophecy that, in its original context was unrelated to Nagarjuna but was conscripted into the Nagarjuna legend at least by the time of Candrakirti (seventh century). Like the Jain Nagarjuna, this prophecy may also be a source for the legends locating Nagarjuna's residence in Gujarat. On the other hand, we must also consider whether this prophecy could also be the source for the tradition associating Nagarjuna with a Satavahana king. In his Madhyamakavatara, Candrakirti relates the following prophecy about Nagarjuna: Also from the Mahamegha (Great Cloud) Sutra in 12,000 [verses]: “Ananda, this Licchavi youth called ‘Joy-When-Seen-By-All-Beings,' when 400 years after my parinirva∞a have elapsed, will be a fully ordained monk named Naga [who will] spread widely my teaching. Finally, in the world realm called the ‘Pure Illumination,' (Prasannaprabha90) he will become an arhant, 90 Demiéville has, “Suvisuddhaprabhabhumi.” See P. Demiéville, “Sur un passage du Mahameghasutra,” appendix 2 of “Les versions chinois du Milindapañha,” Bulletin de l’Ecole française d’Extrême-Orient 24, (1924): 218. NAGARJUNA AND THE RATNAVALI 239 a Samyaksambuddha, named ‘Jñanakaraprabha.'91 Therefore, by means of this agama [Nagarjuna's prediction] has been necessarily, and unmistakably established92. The section of the Mahamegha Sutra to which Candrakirti is referring has the Buddha talking about the past and future lives of a certain Liccavi youth named “Pleasant-to-See-by-all-Sentient-Beings” (sems can thams cad kyis mthon na dga’ ba,). Versions of the prophecy concerning the lives of this youth also appear in the Mahabheriharakaparivarta Sutra, and the Suvar∞aprabhasottama Sutra. The problem with this prophecy insofar as Nagarjuna is concerned is that, while the earliest translation of the Mahamegha into Chinese93 does mention that a Licchavi youth will be reborn as the monk who will protect the dharma, it does not mention the monk's name. The closest that this translation comes is to say that the Licchavi youth was formerly a mysterious naga king94, named Mahaviryanagaraja95 ( ). The Licchavi is, however, associated with a Satavahana king in a future life96. The Buddha foretells that 1200 years after his death, the Licchavi youth will be reborn to a brahmin in the kingdom ruled by a great South Indian king named Satavahana (So-duo-po-he-na ) whose kingdom is called (SuraÒ†ra — modern Gujarat). He will be born in a village called “shan-fang-shi” on the river “hua-huan” . During this lifetime he will become a monk who, among other things, 91 ye ses ’byun gnas ’od. “yan ’phags pa sprin chen po ston phrag bcu gñis pa las kyan| kun dga’ po li tsa byi gzon nu sems can thams cad kyis mthon na dga’ ba zes bya ba ’di ni na mya nan las ’das nas lo bzi brgya lon pa na klu zes bya ba’i dge slon du gyur nas nai bstan pa rgyas par rab tu bstan te| mthar gyi sa rab tu dan ba’i od ces bya ba’i ’jig rten gyi khams su de bzin gsegs pa dgra bcom pa yan dag par rdzogs pa’i sans rgyas ye ses ’byun gnas ’od ces bya bar ’gyur ro zes gsuns so| de’i phyir ’dis lun phyin ci ma log par nes par grub bo||” L. de la Valée Poussin, Madhyamakavatara par Candrakirti, Bibliotheca Buddhica IX. (Osnabruk: Biblio Verlag, 1970) 76. 93 Translation done by = “DharmakÒema,” an Indian monk who arrived in China in 412. See Demiéville (1978), 243. 94 . T. 387, 1100a7-8. 95 This is Demiéville's reconstruction. See, Demiéville (1924), 225. 96 “The one who at that time was the naga king Mahavirya is now the Licchavi, Priyadarsana, and will become the BhikÒu who protects the dharma.” T. 387, 1100b5-6. See Demiéville, 228. 92 240 JOSEPH WALSER teaches the vaipulyasutra of the Mahayana, supports and lifts up the Dharma, and distributes this (the Mahamegha) sutra throughout the world97. Thus, whoever this person is, he is associated with western India, and a Satavahana king. Given that there are a number of different versions of the Mahamegha Sutra in existence, we cannot rule out the possibility that Candrakirti is actually quoting from the version that he knew, a version that is no longer available. However, given the fact that Nagarjuna's name also does not appear in any other version of this prophecy98, it seems more likely that Candrakirti's statement reflects more of the reading practice of the Buddhist community that he represents than an actual textual variant. Mabbett takes another of the Mahamegha’s prophecies to refer to Nagarjuna. This is the prophecy that occurs at the very end of the sutra and discusses a certain princess who will be the daughter of a “Satavahana” (his reconstruction of ) king on the south bank of the river “K®Ò∞a” ( ) in a town called “Dhanyaka†aka.” ( )99 He concludes, “the Mahamegha Sutra therefore offers us a ‘Naga' and a ‘Nagaraja,' named in proximity to a prophecy about a Satavahana ruler at Dhanyaka†aka”100. Mabbett may be reading this sutra too much through the lens of later Tibetan sources. Bu-ston and the other Tibetan historians do place Nagarjuna at Dhanyakataka, but the version of the Mahamegha Sutra that Mabbett (through Demiéville) cites does not. The “proximate prophecy” to which Mabbett refers, occurs many pages after the prophecies attributed to Nagarjuna by classical sources with nothing to link them. Furthermore, it is clear from the text that the Dhanyakataka story is a prophecy relating a future birth of the devi, who is a character in the story unrelated to the future-BhikÒu/present-Licchavi/past-Nagaraja. Not all traditional authors were convinced that the “Naga” to whom Candrakirti alludes in this prophecy refers so unmistakably to Nagarjuna. 97 Demiéville, 227; T. 387, 1099c-1100a. The Mahabheriharakaparivarta Sutra does give a name to this monk, but that name is “Mindful.” See Hopkins, p. 15. Similarly, the name Nagarjuna is nowhere mentioned in the corresponding prophecy in the Suvar∞aprabhasottama Sutra. See J. Nobel, ed. Suvar∞aprabhasottamasutra: Das Goldglanz-Sutra, (Leiden: E.J. Brill Publishers, 1950), 12-17. 99 The Sanskrit is from Mabbett's reconstruction. See Mabbett, 337. 100 Ibid. 98 NAGARJUNA AND THE RATNAVALI 241 Bu-ston, for one, provides an extended quotation from the Mahamegha Sutra contextualizing Candrakirti's citation, and then adds, “So it is to be read, but it is not clear, whether (this passage) really refers to Nagarjuna.”101 From the passage that Bu-ston quotes, it is clear that his version differs from Candrakirti's, insofar as in Candrakirti's version the monk is named Naga, whereas in Bu-ston's version, the monk bears the name of the Buddha (presumably some form of “Sakya-”). Bu-ston explains that others have made this misattribution based on the fact that Nagarjuna's ordained name is said to have been “Sakyamitra”102. Nevertheless, he remains skeptical. Given that this prophecy probably had nothing to do with Nagarjuna initially, the question of how its subsequent association with Nagarjuna was justified in the minds of its interpreters becomes more significant. Why this prophecy? Was Nagarjuna associated with this prophecy because it has a monk associated with a Satavahana king or is Nagarjuna associated with a Satavahana king because he is associated with this prophecy? In order for Candrakirti to make his interpretation of the text plausible, we have to assume that there was some element of the future Licchavi's life that corresponded to information that was already known about Nagarjuna. Unlike the prophecy in the Lankavatara Sutra that gives specifics of the monks philosophical activities, this prophecy does not tell us anything about the future monk's affiliations except that he is an advocate for the Mahayana and propagates the Vaipulya Sutras. We are not given a name for this monk, so the attribution cannot be on similarity of name. Nagas play a big part in the Mahamegha Sutra (a factor which will be discussed more below), but unlike the Rajatarangini, the particular story in the Mahamegha that is associated with Nagarjuna is not a story about Nagas, except insofar as the monk had been a Naga king two births previously. Neither of these factors alone should have been enough to identify Nagarjuna with this monk. The attribution of Nagarjuna to the prophecy about the Licchavi youth only crosses the threshold of plausibility when these two elements are taken together with the association with the Satavahana king. The future, unnamed monk who in a past life was 101 102 Bu-ston, 129. Ibid. 129-30. 242 JOSEPH WALSER a Naga king, who will teach the Mahayana, and associate with a Satavahana king in his future life, probably did sound like Nagarjuna to Candrakirti. Thus, we should see information about Nagarjuna and the Satavahana king as leading to the association of Nagarjuna with this prophecy, and not that Nagarjuna is associated with this prophecy and therefore becomes associated with the Satavahana king. 2.7. Nagas There are a number of elements that occur in every story related to Nagarjuna, and some elements that have a more isolated occurrence. I will examine two of these elements — nagas and alchemy — to show how they have a bearing on his association with particular kings and place names. Every account of Nagarjuna has some etiological myth related to his name, i.e., some myth relating to nagas or snakes. This is not the place to go into all of the cultural significance of nagas in early India, but suffice it to say that nagas were considered to be creatures of great magical power, who were often conscripted into the service of Buddhism in Buddhist legends. Nagarjuna's connection to nagas usually involves his receiving some gift or boon from a naga king. In the HarÒa-Carita, this is an antidote to all poisons, a gift of the moon. In Kumarajiva's Biography and in the Tibetan historical tradition, the gift is the Prajñaparamita Sutras. In these myths we see an attempt to tie the character of Nagarjuna to some other element desirable to the hagiographer (such as alchemy or Mahayana Buddhism) through the instrument of his name. Other associations made with nagas are more complicated. Phyllis Granoff has identified this theme as one of the threads unifying all Jain biographies of Nagarjuna103. These stories are replete with naga associations. The most obvious of these is the fact that in Jain hagiographies Nagarjuna's father is the naga king, Vasuki. Subtler use of the naga connection is made in Nagarjuna's association with, Stambhana Tirtha. What makes Nagarjuna's association with Stambhana Tirtha possible is the sinuous snakes. Stambhana was in fact revered for being the locus of a magical image of the Tirthamkara Parsvanatha. Now biographies of Parsvanatha 103 Granoff, 47. NAGARJUNA AND THE RATNAVALI 243 are unanimous in pointing out connection between this Tirthamkara and the snake god Dharanendra. Nagarjuna is said to have brought the magical image of Parsvanatha to Stambhana in the advice of his father the snake king, in order to make his elixir, in an act that now must seem almost natural in the associative world of these texts: the son of the snake God brings to the holy site the image of the tirthamkara protected by the snake deity104. As we have seen, nagas are a contributing factor in Candrakirti's association of Nagarjuna with the Mahamegha Sutra. This sutra is primarily a vehicle for transmitting a rain-making mantra. As such, the role of nagas as both listeners of the sutra and as characters in the story is emphasized. In addition to the Satavahana connection, Candrakirti's association of Nagarjuna with the Licchavi youth was probably aided by the youth's past life as the Nagaraja (one cannot help but notice the play-on-words with “Nagarjuna”) Mahavirya. The naga connection played a more critical role in the assimilation of the Nagarjuna legend into the chronicles of Kashmir in the Rajatarangi∞i. In this work, Nagarjuna and his Mahayana followers are credited with leading good brahmins away from the rites of the “Nila[mata]pura∞a,” with the result that the nagas sent the snows to destroy the people. Those who did not adhere to Buddhism and still performed the rites were magically spared, while all of the Buddhists were destroyed. The snows only abated when a certain brahmin, Candradeva practiced austerities to please Nila, “lord of the [Kashmir] Nagas, and protector of the land.” This Nila then reestablishes the rites previously revealed in his pura∞a. The story is then summed up as follows: “As the first Candradeva had stopped the plague of the YakÒas, thus the second brought an end in this land the intolerable plague of the BhikÒus”105. The entire story is a reworking of an older legend contained in the Nilamata Pura∞a106 with Nagarjuna 104 Granoff, 48. A. Stein, 33. 106 See Stein 33, note 184. “K. refers here to the legend told in the Nilamata (vv. 325 sq.) regarding the liberation of the land from the Pisacas. The latter… occupied Kasmir under a sentence of Kasyapa during the six months of winter, while men lived there for the remaining six months only, and emigrated each year before the month of Asvayuja. The deliverance of the country from the Pisacas and the excessive cold was effective after four Yugas through the observance of the rites which Candradeva, and old Brahman, descended from Kasyapa, had learned from the Nila Naga… The story told by K[alhana] 105 244 JOSEPH WALSER imported into the beginning of the story to explain why the Nagas were angry. That there were Buddhists in Kashmir was certainly common knowledge. The detail of Nagarjuna at the head of the Buddhists seems to have been added as a poetic way to connect Mahayana Buddhists (we can assume that it was common knowledge by that time that he was a Mahayanist) with a story about Nagas. However, unlike the Jain stories, Nagarjuna is the villain who is antagonistic to the naga king, Nila. Thus, pending any discovery to the contrary, the associations of Nagarjuna with both Stambana Tirtha and Kashmir should be regarded as serving a legitimating function in their legends and not as fact. 2.8. Alchemy Another element common to all traditions concerning Nagarjuna is that he was an alchemist. At the time that these legends were first composed (ca. fifth century), alchemy was of great interest in the courts and monasteries in India as well as in China. Whether one is trying to sell the Nagarjuna legend to an Indian audience or whether one is trying to export the legend to a Chinese audience, claiming that the saint is an alchemist would have ensured the audience's attention. While the Jain tradition is perhaps the first to actually use the term “rasayana siddha”107 (“alchemist”) to describe Nagarjuna, this idea clearly has roots going back to Kumarajiva's stories of Nagarjuna. In Kumarajiva's Biography, Nagarjuna is credited with making an “elixir” ( ) of invisibility. In the story, he and some friends go to a magician for the formula. The Magician, wanting them to remain dependent on him, doesn't give them the formula, but gives them pills that they are to grind to a paste and put on their eyelids. Nagarjuna smells the resulting paste and guesses its 70 ingredients along with their quantities. The theme of Nagarjuna detecting the formula for an elixir appears again in the Prabandhacintama∞i, where there it is an ointment for flying which he smells under the ruse in i. 178-184 is obviously in particulars a mere rechauffé of the ancient legend. The charitable comparison between the Pisacas and the Bauddhas leaves no doubt as to the source from which K. borrowed it.” 107 For example, Rajasekhara Suri uses this term in his Prabandha Kosa, p. 85. NAGARJUNA AND THE RATNAVALI 245 of washing his master's feet (the ointment works when applied to the feet)108. In Xuanzang, Bu-ston, Taranatha, and the Prabandhacintama∞i, Nagarjuna is credited with turning rocks into gold109. In Xuanzang's account, this is done in order to help a Satavahana king out of financial straits, while in Tibetan accounts, it is done to feed the bhikÒus. Xuanzang reports that, “Nagarjuna had the secret to long life,”110 though the source of this long life is not mentioned. In Bu-ston, Taranatha, the B®hatkathamañjari, the Kathasaritsagara, and Jain sources, he is credited with producing an elixir of longevity. In the Prabhandhacintama∞i, this is in order to prove his perfection of charity. In Bu-ston and Taranatha this elixir is shared with the Satavahana king111 whose life is prolonged thereby. That Nagarjuna is consistently associated with alchemy explains a number of details that we find in biographies of Nagarjuna. Granoff points out that Nagarjuna is associated with Padaliptacarya by virtue of the fact that the Jain master was “the best known of all wizards in the Jain tradition”112. Of course, the naga connection also played a role in the association, insofar as Padaliptacarya was the boon of the snake Goddess Vairothya to his barren parents. Further, according to the Prabandhakosa, Padaliptacarya was really named “Nagendra.”113 Nagarjuna's connection to Padaliptacarya may be one of the rationales behind his association with Gujarat in general and Mt. Dhanka in particular. Padaliptacarya is associated with the mountain and Nagarjuna is associated with the acarya114. 108 Granoff, 49-50. This theme also shows up (predictably) in tantric stories related to Nagarjuna. White mentions two such incidents; one in the Rasendra Mangala, where Nagarjuna promises the Goddess Prajñaparamita that he will turn Sriparvata into gold. On the other hand, in a 14th century Telegu work, the Navanatha by Gaura∞a, the credit for this feat is given to Nagarjuna's student (also named Nagarjuna). See White, 166. 110 Watters, II: 201. 111 In Xuanzang's account the length of the Satavahana king's life is also tied to Nagarjuna's, but no elixir is mentioned. 112 Granoff, 47 113 Ibid. 57. 114 The way to the association of Nagarjuna with Gujarat is opened by his identification with the SuraÒ†®ian monk in the Mahamegha Sutra (Mt. Satruñjaya is in Bhavnagar district, Gujarat). 109 246 JOSEPH WALSER The alchemical connection is also the inspiration for the story in the B®hatkathamañjari and the Kathasaritsagara, where the king is named “Cirayus” (“long-life”). Clearly, the king's name is merely a function of a story about Nagarjuna's alchemical feat of producing an elixir of immortality. Finally, it is worth considering whether Nagarjuna's association with Sriparvata may be an association made by his biographers solely through his association with alchemy, as the name Sriparvata had strong associations with the study of alchemy dating back at least to the fifth/sixth century (when some of the earliest biographies were written). Nagarjuna's association with this site may be nothing more than the association of his alchemy with the most famous alchemical site. In fact, the numerous stories about Nagarjuna's alchemical prowess may even confirm Nagarjuna's South Indian origin. This is because, while there are so many hagiographical details are associated alchemy, the curious fact is that there is no evidence that Nagarjuna was an alchemist. Although there are a number of works surviving in the Tibetan canon which are ascribed to Nagarjuna, according to White, “Of the fifty-nine works attributed to Nagarjuna and translated, in the twelfth through thirteenth centuries A.D. into Tibetan in the Tanjur, none contains any alchemical material”115. This is a curious circumstance for a figure who became the alchemist par excellence not only in his own religious tradition but in the Hindu and Jain traditions as well. No other Buddhist figure has been so widely renowned for alchemy and appropriated into other traditions as an alchemist. Thus, the origin of the alchemical association requires some explanation. In Kumarajiva's Biography, we find three examples of Nagarjuna's magic (only the first of these feats is alchemy proper). The first story is Nagarjuna's mishap with the invisibility potion, the second is his magical battle with a brahmin and the third is his conversion of the south Indian king. At the beginning of each of these stories there is something to tell us that he is in South India. As a matter of fact, of the four times South India is mentioned, three of these introduce a story about his alchemy or wizardry. It should be kept in mind that while there are no 115 White, 70. NAGARJUNA AND THE RATNAVALI 247 Indian sources from the fifth century which explicitly talk about alchemical practices, alchemy was already firmly ensconced in the popular imagination of the Chinese for whom Kumarajiva was writing. In fact, in Ge Hong's Baopuzi116 (ca. 320 A.D.) there is a discussion of an invisibility potion. It is quite possible that the early associations of Nagarjuna with alchemy came from Kumarajiva trying to appeal to Chinese interests. The question remains why this practice would be associated with South India. The answer could be as simple as South India being a vast unknown region to Kumarajiva and hence the appropriate location for exotic heroes. Yet, by the time that Kumarajiva is writing, the trade routes between north and south are well traveled and the exotic South does not seem to be a major theme in the literature and drama of the day. This, coupled with the fact that, there are sources (such as Candrakirti's CatuÌsatakav®tti) which mention South India apart from any mention of alchemy, leaves us with the impression that, for Kumarajiva, Nagarjuna's South Indian origin was probably a fact independent of his association with Alchemy. So where does all of this leave us? Tracing the literary connections in the various legends of Nagarjuna has led us to question the validity of Nagarjuna's associations with Kashmir, the Himalayas, Mt. Dhanka, Stambhana Tirtha, and Sriparvata. Similarly, the stories of Nagarjuna's association with King Cirayus, and with HuÒka, JuÒka, and KaniÒka have also been called into question. The only element of these stories that does not seem to have been put there for specific sectarian/institutional/ideological motivations is Nagarjuna's association with the Satavahana king. As far as his residence is concerned, we are left with three names that occur prominently in Nagarjuna legends — Nalanda, Vidarbha, and possibly Dhanyakataka. Nalanda cannot be taken seriously as a possibility for three reasons. First, it was not a strong monastic center until about 425117, i.e., after Kumarajiva's report that Nagarjuna had been dead over one-hundred years. Second, Nagarjuna's associations with Nalanda are confined to 116 J.R. Ware, Alchemy, Medicine, and Religion in the China of A.D. 320: the Nei P’ien of Ko Hung (Cambridge: M.I.T. Press, 1966) 16.2a. 117 K. Satcidananda Murti, Nagarjuna (New Delhi: National Book Trust, 1971), 50. 248 JOSEPH WALSER Tibetan Buddhist sources that are concerned with placing him in the transmission lineage for the Guhyasamajatantra, a text that was important in the curriculum at Nalanda. Third, Xuanzang and Yijing both spent considerable time at Nalanda and studied Nagarjuna's texts there. It is strange that they would have spent so much time there and yet heard nothing of a man whose works played such an important part in the curriculum. Though absence of evidence cannot be taken as evidence of absence, the silence of the pre-tenth-century sources about a north Indian origin for Nagarjuna should be carefully examined. Kumarajiva was born in Kucha, and at the age of nine, went with his mother to Jibin ( – Kashmir) where he received his early schooling. Presumably, it was in Kashgar that he studied and memorized the texts of Nagarjuna118. If he was between fifty and sixty years old when he translated Nagarjuna's Biography in Changan and testified that Nagarjuna had been dead nearly one-hundred years, we may assume that Nagarjuna had been dead considerably less time than that when he first studied his texts before the age of twenty. Given this, it seems unlikely that he would not have heard any news of Nagarjuna having lived on the same trade route as the places where he (Kumarajiva) studied. By the same reasoning, Xuangzang, Yijing, and Huichao traveled to India during the sixth through eighth centuries and spent considerable time at Nalanda University, and none of them heard stories connecting Nagarjuna with North India or with a North Indian king, while all of them (Kumarajiva included) heard stories connecting Nagarjuna with South India and two of them heard of his association with a Satavahana king119. Thus far, we have shown that all but two of the place-names associated with Nagarjuna are associated with him for reasons of questionable historical value, and that sites and kings in North India are unlikely. The 118 Robinson, 72. In this connection, however, it should be mentioned that Xuanzang visited Dhanyakataka and did not hear any stories about Nagarjuna. I would argue that this case is different from that of Nalanda insofar as Nalanda was still a vibrant university when he visited there (and hence, one should expect some institutional memory of a former master to survive), whereas many of the monasteries around Dhanyakataka were deserted. See Watters, 214. 119 NAGARJUNA AND THE RATNAVALI 249 two remaining sites are in South India. Furthermore, the sites of Vidarbha and Dhanyakataka (provided this latter attribution does not come from the Kalacakra sutra) do not seem to be connected to stories about alchemy or nagas, and should be taken seriously as possible sites for Nagarjuna's residence. Since these two sites had strong associations with the Satavahana dynasty, these sites may also lend their weight to the connection between Nagarjuna and a Satavahana king. The Satavahana connection finds further support in the fact that, while all of the elements in the Nagarjuna hagiography discussed so far have some connection either to nagas or alchemy, the Satavahana dynasty does not have strong connections to either. This is especially noticeable in the Kathasaritsagara, where the legends of Nagarjuna and those of Satavahana are separated. All of the stories about alchemy and nagas go with Nagarjuna while none of these elements are contained in the story of Satavahana. The Satavahana king is mentioned in the Mahamegha legend, but as I argued above, it is unlikely that the Mahamegha is the source of this information. In short, Nagarjuna's connection to a Satavahana king seems to have occurred independent of any of the hagiographical patterns of legitimation we have discussed so far. True, in later hagiographical literature, it is not uncommon for a saint to have interactions with a king, but in most of these legends, the king is unnamed. It does help the legitimacy a saint to be associated with a king, but if this association were made up, we should expect to not see unanimity as to the name of the king. The diversity of the legends about what Nagarjuna did with this king rule out a single, “ur-” source for this information. Hence, we are still pressed to explain why Nagarjuna is associated with this dynasty. While there are a number of legends about KaniÒka as a great patron of Buddhism, the only stories about a Satavahana king being a benefactor of Buddhism occur in conjunction with legends of Nagarjuna. As far as the early Indian literary imagination was concerned, the Satavahana dynasty was probably not the best dynasty to attach your saint to. Until another explanation can be offered, we simply have no choice but to consider that Nagarjuna's hagiographers assumed this information to be common knowledge. Thus, through a long process of elimination, the best reading of the information we have points to Nagarjuna's residence in the Deccan during the reign of a Satavahana king. 250 JOSEPH WALSER 3. The Ratnavali and the Satavahana Dynasty: The Image of the Buddha Thus far, I have established that two facts are likely: that a Satavahana king was Nagarjuna's patron and that Nagarjuna was most likely the author of the Ratnavali. How do these two pieces of information get us closer to determining the date or the residence of Nagarjuna? Simply put, the Ratnavali instructs the king to say a certain ritual formula three times a day in front of an “image of the Buddha,” and to construct images of the Buddha “positioned on lotuses.” If the arguments concerning Nagarjuna's patron and his authorship of the Ratnavali are correct, then the it would have to have been written: a) during the reign of a Satavahana king b) at a time and in a region where Buddhas sitting on lotuses were a motif in use c) at a time and in a region where Buddha images were available as distinct objects of veneration and/or propitiation d) to a king who could have had access to an appropriate Buddha image to recite Nagarjuna's twenty verse prayer Although anthropomorphic images of the Buddha had wide currency around Gandhara and Mathura as early as the first century, during most of the Satavahana dynasty anthropomorphic representations of the Buddha were absent in the Deccan. In fact, very few of the Satavahana kings were alive at a time and a place to meet all of the above criteria for the Ratnavali’s addressee. In the following, I argue that probably the only Satavahana king meeting all three criteria is Yajña Sri, and then only during the years when he ruled from Dhanyakataka. If the Pura∞ic accounts concerning the length of Satavahana reigns are accurate, then the Ratnavali must have been written within a 29-year period somewhere in the area of the lower Krishna River valley. In the Ratnavali there are three verses where Nagarjuna mentions images of the Buddha120. 120 Wooden images of the Buddha are also mentioned in verse 2 of the Suh®llekha: “Just as the wise ones will respect a statue of the Sugata, even though it be made of wood [and] however [unadorned] it may be, so in the same manner, although this composition of mine may be pitiful, may you not criticize it, for it is based on the Sublime Teaching.” See Golden Zephyr, 6. Gu∞avarman's translation does not specifically mention wood, but refers to a “Buddha image which is carved and painted” (“ ”) T. 1672, p. 745b14. However, since Nagarjuna's authorship of this text is more difficult to defend, I will shall limit our inquiry to the relevant verses of the Ratnavali. NAGARJUNA AND THE RATNAVALI 251 Verse 231: You should respectfully and extensively construct Images of Buddha, Monuments, [stupas] and temples And Provide residences, abundant riches, and so forth121. Verse 232: Please construct from all precious substances Images of Buddha with fine proportions, Well designed and sitting on lotuses, Adorned with all precious substances122. Verse 465: Therefore in the presence of an image [of the Buddha123] Or monument [stupa] or something else Say these twenty stanzas Three times every day124: 121 Translation of these verses is from Hopkins, 124-5 and 159. There is no Sanskrit available for any of these verses. Tibetan: 231. sans rgyas sku gzugs mchod rten dan| gtsug lag khan dag gus tshul du| sin tu rgya chen gnas mal sogs| rgya chen phyug pa bsgrub par mdzod|| Variant readings: 231c [Narthang and Peking] “gnas lam” vs. “gnas mal” in Chone, Derge and in Rgyal tshab rje's commentary on the Ratnavali. Hahn, 78. . T. 1656, 498b26-27. 122 232. rin chen kun las bgyis pa yi| sans rgyas sku gzugs dbyibs mdzes sin| legs par bris pa padma la| bzugs pa dag kyan bgyid do stsol|| Variations: v. 232b Narthang and Peking have legs sin whereas Chone and Derge have mdes sin. 232d; Narthang and Peking have two lines: bzugs pa dag la rin po che| kun gyis brgyan pa bgyid du gsol|. The Chone and Derge versions, however, are confirmed by Rgyal tshab rje and Paramartha's translation (below). Hahn, Ibid. Dunne and McClintock write the following note: “The Zhol, Narthang and Peking editions of a slightly different reading. Following those editions, the verse would read as follows: “From all kinds of precious substances, please make well drawn and beautifully proportioned images of buddhas seated upon lotuses and adorned with all kinds of gems.” Dunne, and McClintock, The Precious Garland: An Epistle to a King, (Boston: Wisdom Publications, 1997), 118, note 50. My thanks to John Dunne and Wisdom Press for providing me with a copy of this translation. Paramartha's translation: (var. Ming mss. “ ”) T. 1656, 498b28-28. 123 Both Dunne/McClintock and Hopkins translate “sku gzugs” as “icon,” which is certainly acceptable. For our purposes, however, “icon” could refer to any of a number of non-anthropomorphic representations of the Buddha (such as the empty throne, the Buddhapada, etc.) prevalent in India until the third century CE. It should be noted that a more literal translation for sku gzugs would be “body-image.” Since the word sku is the respectful form for lus = “body,” it is implied that the image the king is to go in front of is an image of the Buddha's body. The phrase is unequivocal in Paramartha's Chinese translation: Therefore, rise up determined and appear before a Buddha or caitya…” 124 465. de phyir sku gzugs mchod rten gyi| spyan sna ‘am yan na gzan yan run| tshigs su bcad pa ñi su ‘di| ñin gcig bzin yan dus gsum brjod|| Hahn, 155. . T. 1656, 504b 12-13. 252 JOSEPH WALSER That these verses refer to actual images of Buddhas (as opposed to Buddhas to be visualized in meditation) is clear from the context. Verses 231 and 232 begin a long list of construction and public works projects for the king to perform. Nagarjuna is clearly not talking about meditation in this section. It is also likely that the image referred to in verse 465 was also a physical image, as this practice of using physical images in a Mahayana ritual context has been found in other sources contemporary with the Ratnavali125. If Nagarjuna lived at some distance from the king, we might refine our criteria further by stating that the motif of a Buddha on a lotus had to have been available at a time and in a place where Nagarjuna could have been aware of it, whereas the king merely had to have access to a free-standing image of the Buddha (preferably one not embedded in a narrative context), in front of which he could perform this ritual. I am of course assuming that Nagarjuna would not have suggested that the king go in front of an image of the Buddha knowing that such a thing did not exist where the king lived. 3.1. The Buddha Image in the Deccan For a Satavahana king to be able to stand in front of an anthropomorphic image of the Buddha (as opposed to an iconic representation) and recite a formula, he would most likely have to have lived in the eastern Deccan sometime after the first century A.D. Though the western Deccan sites of Nasik, and Paithan were centers of Satavahana political activity 125 Paul Harrison writes, “…there can be no doubt that by the second century C.E. some Buddhists were indeed practicing a form of buddhanusm®ti that… included detailed visualization of the physical body of the Buddha, and was accompanied by the use of images. The principle evidence for this is provided by a Mahayana sutra called the Pratyutpannabuddha-saµmukhavasthita-samadhi-sutra… the first translation of which was made by the Indo-Scythian LokakÒema in 179 CE.” Paul Harrison, “Commemoration and Identification in Buddhanusm®ti,” in In the Mirror of Memory: Reflections on Mindfulness and Remembrance in Indian and Tibetan Buddhism, Janet Gyatso, ed. (Albany: State University of New York Press, 1992), 220. It is interesting to note in this regard that while Mahayana sutras such as the Ugradattaparip®ccha and the late BodhisattvapratimokÒa mention a Mahayana ritual similar to the one that Nagarjuna describes in the Ratnavali, the Ratnavali is the only text that instructs the adherent to stand in front of a statue or stupa and not to stand in front of a (human monk?) Mahayana Bodhisattva. NAGARJUNA AND THE RATNAVALI 253 until at least the reign of Yajña Sri Satakar∞i, (170-198 A.D126.) virtually no anthropomorphic images (sculpted or painted) of the Buddha have been found anywhere in the western Deccan during the Satavahana dynasty. Most scholars place the beginning of anthropomorphic representation of the Buddha in the Western Deccan much later, during the reign of Harisena (ca. 450-500 CE) of the Vaka†aka dynasty127. Thus, even if A.M. Shastri is right in claiming that Kumbha Satakar∞i, Kar∞a Satakar∞i, and Saka Satakar∞i were the last three rulers of the Satavahana dynasty who ruled from Vidarbha right up to the beginning of the Vaka†aka dynasty128, it is still unlikely that any of these were Nagarjuna's patron, because none of them would have had access to a Buddha image in that region. For this reason, any king who could have been Nagarjuna's patron would have had to live in the eastern Deccan. 126 Though the controversies surrounding the dates and chronology of the Satavahana dynasty are far from over, throughout this article I will use the dates provided by Shastri. See A.M. Shastri, The Satavahanas and the Western Kshatrapas: A Historical Framework (Nagpur: Dattsons, 1998), 131. Since I am fixing Nagarjuna's dates to the reign of Yajña Sri, one should adjust the dates of the former to correspond to discoveries concerning the date of the latter. 127 See W. Spink, Ajanta to Ellora, (Ann Arbor: Marg Publications, 1967), 7-8. There is one notable exception. It is, however, an exception that perhaps proves the point. Marilyn Leese has documented two anthropomorphic images of the Buddha at cave 3 at Ka∞heri. These images (which she takes pains to prove were carved during the reign of Yajña Sri) are quite small, only about a foot high, and are placed at the top of a pillar so as to be inconspicuous. She attributes their small size to their being modeled after portable images procured through trade with the north. See M. Leese, “The Early Buddhist Icons in Kanheri's Cave 3,” Artibus Asiae, 41 (1979): 93. M.K. Dhavalikar, however, attributes their small stature to another motive: “[The Kanheri Buddha images] have been carved on the top of the pillar. No one can normally see it and it therefore seems highly likely that the sculptor had stealthily carved it without the knowledge of the donor.” M.K. Dhavalikar, Later Hinayana Caves of Western India, (Poona: Deccan College Postgraduate and Research Institute, 1984), 66. Dhavalikar takes this as proof positive that the Buddha image had made it to the Western Deccan by the end of the second century, perhaps in order to support his claim that some of the shrine niches found at Kanheri may have contained wooden images of the Buddha. Be that as it may, the fact remains that no such images have been found. This coupled with the avoidance of any open anthropomorphic representation of the Buddha in stone or in paint, leaves us with the impression that whereas the Buddha image may have been known at this time, its representation was considered somehow distasteful. 128 See A.M. Shastri, “The Closing Phase of the Satavahana Power and Allied Issues,” in Early History of the Deccan, (Delhi: Sundeep Prakashan, 1987): 38-44. 254 JOSEPH WALSER There are only a few places in the eastern Deccan which were home to Satavahana kings. It appears that Pu¬amavi, Sivamakaskandha Gautamiputra and Yajña Sri Satakar∞i may have ruled from Dhanyakataka and Vijaya ruled from Nagarjunakonda (also known as Vijayapura). It is not known from whence the last two Satavahana kings listed in the Pura∞as, Candrasri and Pu¬umavi II, ruled. It is possible that Candrasri continued to rule from Nagarjunakonda as the one inscription mentioning him comes from Kodavolu in Godavari district. Using this same reasoning, however, we would have to place the last Satavahana king far west of Nagarjunakonda, due to the fact that Pu¬umavi II's only surviving inscription was found at Myakadoni in Bellary district, Karnataka129. In the following, I will explore the art history of these regions to determine which of these kings would have had access to an image of the Buddha. At this point a note should be added about the nature of art historical evidence available to us. All of the work that has been done on the relative chronology of art in India during the period that concerns us has been on art carved in stone. The reasons for this are obvious. Images made of materials that decay or break simply have not survived. Clearly, the Buddha could be represented in other media, such as paintings and wooden or clay sculptures. The earliest mention of the figure of the Buddha, refers to a painting130. Similarly, literary evidence for the representation of the Buddha on cloth can be found in the “Rudrayanavadanam” of the Divyavadanam where there is a legend that king Bimbisara allowed his image to fall on a piece of cloth in order that his image might be painted131. Such portable images of the Buddha were popular at the time of Yijing where we have testimony of the use of portable drawings of the Buddha by travelling monks132. Finally, M.K. Dhavalikar notes that there are 129 H.P. Ray, Monastery and Guild: Commerce under the Satavahanas, (Delhi: Oxford University Press, 1986), 40. 130 See R.C. Sharma, Buddhist Art of Mathura, (Delhi: Agam Kala Prakashan, 1984), viii-ix. 131 R.C. Ray 144, Vaidya, Divyavadanam p. 466. 132 “The priests and the laymen in India make Kaityas or images with earth, or impress the Buddha's image on silk or paper, and worship it with offerings wherever they go. Sometimes they build stupas of the Buddha by making a pile and surrounding it with bricks. They sometimes form these stupas in lovely fields, and leave them to fall in ruins. Any one may thus employ himself in making the objects for worship. Again when the NAGARJUNA AND THE RATNAVALI 255 wall sockets for installing wooden images in a number of caves at Kanheri133. Indeed, the issue of what kind of representation is intended in the Ratnavali is ambiguous in the absence of the Sanskrit original. Both the Chinese and Tibetan indicate that the images of the Buddha are to be drawn or painted (Paramartha “ ” Tibetan “bris pa”). Both of these terms, however, could translate the Sanskrit √likh, (lit. “to scratch” but also “to write” or “to draw”). It is possible that the Ratnavali is referring to the practice of scratching a line drawing of the subject on the rock before sculpting134. This translation assumes that the Buddha images were drawings or paintings (as opposed to sculptures or statues). Whether paintings of the Buddha existed during Satavahana times is difficult, at our present state of knowledge, to know. Paintings from Satavahana times have been found in the western Deccan, but none of the Buddha. The Chinese and Tibetan translations do not allow us to reconstruct the Sanskrit with any certainty; the Chinese seems to point to some form of √likh whereas the Tibetan bgyis pa, “to make” suggests some form of √kl®p. While we do not know if cloth paintings were in use at the time of the Ratnavali, we do know quite a bit about the art history of the time. Though there is no need to assume that stone sculpture was the only form of art at this time, we would need to come up with a special explanation of why the anthropomorphic image of the Buddha (on a lotus, no less) should be portrayed in non-stone artworks when it is consciously avoided in stone sculptures. Until such an argument can be made, we must assume that the motifs of non-stone artworks generally mirrored the motifs seen in stone works. Our task, then, is to determine when the motif of the lotus-pedestal first appears in the eastern Deccan. In the eastern Deccan, Art Historical scholarship has really only focused on two sites: Amaravati and Nagarjunako∞nda. I will discuss some of the relative dates of images from these people make images and Kaityas which consist of gold, silver, copper, iron, earth, lacquer, bricks, and stone, or when they heap up the snowy sand (lit. sand-snow), they put in the images or Kaityas two kinds of sariras. The relics of the teacher, and the Gatha of the chain of causation.” J. Takakusu, A record of the Buddhist religion as practiced in India and the Malay archipelago (A. D. 671-695), (Oxford, Clarendon Press, 1896), 150-151. 133 Dhavalikar, 51. 134 Elizabeth Rosen Stone, The Buddhist Art of Nagarjunako∞∂a (Delhi: Motilal Banarsidass, 1994), pl. 187. 256 JOSEPH WALSER sites and assume that the sequence was the same at other sites in the region unless we have a reason to think otherwise. A considerable amount of work has been done on the art sequence at the Amaravati stupa. The most recent work is that of Anamika Roy, who has done a thorough investigation of the epigraphy, art and architecture of that site, in order to determine its chronology. On the basis of her findings, she outlines the development of the site into four distinct phases135. The first phase goes from ca. third century B.C. to first century A.D., and contains no anthropomorphic images of the Buddha. During this time, while quite a number of Buddhist narratives are portrayed in sculpture (both narratives from the life of the Buddha as well as his past lives), an anthropomorphic image of the Buddha is conspicuously and uniformly avoided. In its place, we find the Buddha represented symbolically by the Bodhi tree, the dharma cakra, etc. This avoidance of representing the Buddha anthropomorphically seems to be a Deccan-wide phenomenon and not confined to any particular sect in the Deccan during this period. The second phase spans the first century A.D. and includes the first anthropomorphic representations of the Buddha. Roy lists two examples of this early form of the Buddha in catalogue numbers 187 and 188 of the Madras Government Museum. These are both hybrid representations of the Buddha; images that use both symbolic representations as well as anthropomorphic depictions. Significantly, both depictions of the Buddha from this period have the Buddha sitting on either throne (“paryanka”) or a long seat (“asandi”)136 in abhaya-mudra. The third phase marks the height of Buddhist art at Amaravati and lasts roughly until the second half of the second century. It is during this phase that the majority of the Jataka tales were carved on the rail copings. In this phase, no new anthropomorphic representations of the Buddha appear, and the style again reverts to symbolic manifestations137. 135 These phases are actually a revision of the four phases first proposed by Sivaramamurti, “Amaravati Sculptures in the Madras Government Museum”, Bulletin of the Madras Government Museum, 4 (1956), 26-32. 136 For a discussion of these seats, see Sivaramamurti, 136-7. 137 A. Roy, Amaravati Stupa: a critical comparison of epigraphic, architectural, and sculptural evidence. (Delhi: Agam Kala Prakashan, 1994), 138. NAGARJUNA AND THE RATNAVALI 257 Nevertheless, there is no evidence to suggest that the previously installed Buddha images were taken away during this period. The Satavahana kings who might have ruled over the area during to these two artistic periods (and hence would have had access to an image of the Buddha) were Pu¬umavi I, VasiÒ†hiputra Satakar∞i, Siva Sri, and Sivamakasada. Possibly Gautamiputra Satakar∞i was late enough to be included in this list, although all inscriptions bearing his name locate him in the western Deccan. It is unlikely, however, that any of these kings was the king to whom the Ratnavali was addressed. The reason is that during this artistic phase there is no evidence for the existence of the lotus throne (padmapi†ha) motif in the Deccan area this early. Even at Gandhara and Mathura during the KuÒan dynasty, where the anthropomorphic depiction of the Buddha begins quite early, the vast majority of Buddhas are depicted as sitting on three-tiered rectangular platforms whose flat front face served as a place for an inscription or an additional motif138. Buddhas depicted on lotus thrones in that region tend to be dated to the third century or after139. At Mathura, sometime toward the end of the second century, we find cushions made of kuÒa grass added to the simple pedestal on which the Buddha sits, but no lotus thrones140. Coomaraswamy, places the advent of the lotus throne motif sometime during the second century, but does not offer any more precision as to the time or the place of its advent141. 138 For examples of this motif, cf. N.P. Joshi and R.C. Sharma, Catalogue of Gandhara Sculptures in the State Museum, Lucknow, (Lucknow: The State Museum, 1969). S. Nagar, Gandharan Sculpture: A Catalogue of the Collection in the Museum of Art and Archaeology, University of Missouri-Columbia, (Columbia: The Museum of Art and Archaeology, 1981). 139 See, Nagar. There are, perhaps, some early exceptions from Sikri, which Sir John Marshall dates to the first century CE. See. J. Marshall, The Buddhist Art of Gandhara: the story of the early school, its birth, growth, and decline (Karachi: Dept. of Archaeology and Museums, Govt. of Pakistan, 1973), 56 and plate 50. 140 See R.C. Sharma, plates. 141 Coomaraswamy, Elements of Buddhist Iconography, (New Delhi: M. Manoharlal, 1972 ), 39. Precursors to the padmapi†ha can be found earlier. For example, in the State Museum of Lucknow, there is an image of Hariti whose feet rest on a square base decorated with lotus petals. See N.P. Joshi and R.C. Sharma, Catalogue of Gandhara Sculptures in The State Museum, Lucknow, State Museum catalogue series; no. 3 (Lucknow: Uttar Pradesh State Museum. 1969), fig. 68 (accession num. 47.105). At Bharhut, there are two medallions with reliefs of Sri LakÒmi standing on a lotus rising out of a pur∞a-gha†a and a yakÒi standing on a lotus. See B. Barua, Barhut: Aspects of Life and Art, book III, 258 JOSEPH WALSER Unfortunately, it is precisely this precision that we need if we are to date the Ratnavali from its mention of a lotus base (“padmapi†ha,” “padmasana”, or “kamalasana”). If, however, we can assume that a Satavahana king ruling over either Dhanyakataka or Nagarjunakonda patronized Nagarjuna, then we need only to look for a rough date of the first padmapi†ha in this area to find a lower limit for the composition of the Ratnavali. Roy does not discuss the advent of the lotus pedestal motif in the art of Amaravati, but a review of the documented sculptures from Amaravati containing this motif reveals that each of them belongs to her fourth phase of sculpture and to the second part of the fourth epigraphic phase. There are relatively few sculptures from Amaravati exhibiting this feature. We find it on a pillar (Madras Government Museum [MGM] 247), a frieze decorated with alternating Buddhas and stupas (MGM 256), a drum slab (British Museum [BM] 79) and a railing pillar (BM 11). All of these are dated by Roy to be from the third century or after (Roy's fourth phase), as they all share stylistic features common to whose Buddha images date from the second half of the third century)142. The fourth and final period of Amaravati art, according to Roy, was marked by a change in artistic style. The human forms are noticeably more elongated. Fortunately, there is also a change in epigraphy which corresponds to this stylistic change. The epigraphy becomes more ornate, characteristic letters being a notched “ba” and a “pa” with a descending hook ( and )143. It is the latter development that distinguishes the writing style of Siva Skandha's Amaravati inscription from that of his immediate successor, Yajña Sri 144. Of the four images depicting a Buddha on a lotus from Amaravati, three of them have inscriptions. The inscription on MGM 247 is of little help for (Calcutta: Indian Research Institute Publications, 1934-5) pls. LXVI. 79, LXVII. 80 and LXVIII. 81. Similarly, (and perhaps related) there is a beautiful image of the Buddha's mother, Maya, sitting on a lotus (also rising out of a pur∞a-gha†a) from Sañchi. See Marshall and Foucher, Monuments of Sañci, vol. 2, (Delhi: Swati Publications, 1982.) pl. 41. 142 The same dates are also concluded by Robert Knox for the pieces in the British Museum. Cf. R. Knox, Amaravati: Buddhist Sculpture from the Great Stupa, (London: British Museum Press, 1992), 60, 139-40. 143 See Roy, appendix 4, table 4. 144 Compare tables 3 and 4, Ibid. NAGARJUNA AND THE RATNAVALI 259 dating the image145. The inscriptions on BM 79 and MGM 256, however, do seem to belong to the same period as their sculptures, and Roy assigns both of these inscriptions to the second part of the fourth epigraphical phase (ca. third century A.D.). Though the drum slab (BM 79) containing this motif has an inscription, Roy is somewhat uncertain of her dating of it. Her best guess is that it belongs to the fourth phase of epigraphy at Amaravati: BM no. 79…: Half of the inscription is chipped off. Out of the remaining few letter forms, only one word Bhadanta is intelligible and on the basis of these few letters, we may tentatively date it in the late 2nd or early 3rd century A.D. (Fourth phase)146. The inscriptions must, however, be more recent than the sculpture, because parts of the inscription continue between the heads of the uppermost figures of the frieze. Hence, Robert Knox's comments on the date of the sculpture will be relevant. The extreme, fleshy naturalism of the carving of this relief places it at once in the Amaravati High Period. In the tightly packed, nervously energetic decoration of the slab it falls easily into the 2nd phase of the 3rd century AD147. With all of the examples of the lotus pedestal placed in the fourth epigraphic and sculptural phase of Amaravati, we may reasonably place the writing of the Ratnavali within the same period, because it is only in this phase that we find the motif of Buddha standing and sitting on lotus flowers. To what extent can we translate this into a range of dates? Of key importance to this study is the fact that, on epigraphical and stylistic 145 Roy (p.152) describes this piece as follows: “The carving on this fragment is divided into three panels. The uppermost shows a stupa surmounted by an umbrella and the lower panels show the haloed figure of the Buddha on a lotus pedestal. Between the second and the third panels there are two inscriptions belonging to two different periods. One belongs to the first century BC, while the other belongs to the 7th century A.D. Moreover, the sculpture does not belong to the period of either of the inscriptions. It seems that the first inscription was engraved on the plain octagonal pillar in the 1st century B.C., and that this pillar was then recarved in the 3rd century A.D. Subsequently, in the 7th century A.D. another inscription was engraved on it.” 146 Roy, 198. 147 Knox, 141. 260 JOSEPH WALSER grounds, the dome slab with the Buddha standing on lotuses discussed above belongs to the same epigraphic phase (Roy's IV.2) as the dome slab mentioning the reign of Yajna Sri Satakar∞i. The latter inscription is not by Yajña Sri himself but from an upasaka from Ujjain. 1. Sidham rajño Gotamapu[trasya] Sri-Yajña-[Sa]-taka∞isya saµvatsare … … … vasa-pa 5 divase 8 Ujjayini-upasakena 2. Jayilena … … … mahacetiye … … … karitam … … … 3. … … … Dhanaka†a-cetiya148 Unfortunately, while the inscription tells us that it was donated on the eighth day of the fifth fortnight of the monsoon, the regnal year is missing. Thus, all we know is that this was inscribed sometime during the reign of Yajña Sri Satakar∞i (which, by Pura∞ic accounts, lasted 29 years). As the sculpture on which the inscription is found still uses a non-anthopomorphic representation of the Buddha, we might assume that it was carved near the beginning of the fourth phase of Amaravati art and that it pre-dates our Buddhas on lotuses discussed in the Ratnavali. This allows us to date the Ratnavali no earlier than the reign of Yajña Sri (last quarter of the second century). The reigns of the three Satavahana kings succeeding Yajña Sri were fairly short (Vijaya six years, Candra Sri three or ten years, and Pu¬umavi III seven years). Hence, if Nagarjuna wrote the Ratnavali during the reign of a Satavahana king and during a time when the padmapi†ha motif was available, it would have to be written within a period of fifty-two years. However, not all of these kings ruled from Dhanyakataka/Amaravati. We know from an inscription found at Nagarjunakonda, that Vijaya Satakar∞i had moved the capital to that site which is about one hundred kilometers distant149. While the Buddha image (with or without lotuses) continued to be produced at Amaravati, upstream at Nagarjunakonda, artists and/or patrons showed a reluctance to use anthropomorphic images 148 H. Sarkar, “Some Early Inscriptions in the Amaravati Museum,” Journal of Ancient Indian History 4.1-2 (1971): 8. 149 H. Sarkar, “Nagarjunako∞∂a Prakrit Inscription of Gautamiputra Vijaya Satakar∞i, Year 6,” Epigraphia Indica, 36 (1965-66): 273-275. A number of other inscriptions refer to Nagarjunakonda as “Vijayapuri.” Cf. Vogel, “Prakrit Inscriptions from a Buddhist Site at Nagarjunako∞∂a,” Epigraphica Indica 20, (1920-1930): 22. NAGARJUNA AND THE RATNAVALI 261 of the Buddha at all. In fact, the first images of the Buddha at this site can only be dated to the reign of Ma†hariputra VirapuruÒadatta (236-260 A.D.)150 or later. The first Buddha in a non-narrative context (i.e., carved for the purpose of worship) only comes into existence during the time of Ehuvala Caµtamula (261-285 A.D.)151. Thus, while images of the Buddha on a lotus existed in the Deccan during the reign of these last three Satavahana kings, it is unlikely that such an image was available to any king living at until the time of the second IkÒvaku king (i.e., long after the Satavahana dynasty was over). Though the location of the other two kings is uncertain, it appears from the location of their inscriptions that they also were not at Amaravati152. Until more is known about the reign of these last two monarchs, it would be dangerous to speculate about the availability of Buddha images to them. The only surviving inscription mentioning Pu¬umavi II comes from the eighth year of his reign and is located at Myakadoni in Bellary district in Karnataka. If Pu¬umavi II had in fact relocated to that area, then it is unlikely that he would have had access to an image of the Buddha even at his late date. Nevertheless, we cannot rule either of these last two kings out as possible patrons for Nagarjuna simply because we do not know enough about them. If, however, further investigation finds that they continued Vijaya's rule from Nagarjunakonda, they would be unlikely candidates for Nagarjuna's king. By process of elimination, this leaves us with Yajña Sri Satakar∞i (ca. 175-204 A.D.) as the most likely candidate for Nagarjuna's patron, with Candrasri (ca. 210-213 or 210220 A.D. – the pura∞as do not agree about the length of his reign) and Pu¬umavi II (ca. 213-220 or 220-227 A.D.) as other possible candidates. If Nagarjuna's patron had been a Sata/Sada king (as suggested by Paramartha's translation of the Ratnavali), the date does not change because the few images of the Buddha found elsewhere in coastal Andhra certainly do not predate those of Amaravati. Therefore, the best determination we can make of the composition of the Ratnavali has to be between 175-204 A.D. 150 For the dates of the IkÒvaku kings, see H. Sarkar, “The Nagarjunako∞∂a Phase of the Lower K®Ò∞a Valley Art: A Study Based on Epigraphical Data,” in Indian Epigraphy: Its Bearing on the History of Art, ed. F. Asher and G.S. Gai (New Delhi, 1985), 31. 151 Stone, 17. 152 H. Ray, 40. 262 JOSEPH WALSER or between 210-227 A.D, somewhere in the Lower Krishna Valley, with the earlier dates being more likely than the latter. 4. Conclusion I have argued throughout this paper, the date and location of Nagarjuna is dependent, in part, on our conclusions concerning two factors: Nagarjuna's proximity to the Satavahana dynasty and his authorship of the Ratnavali. Under these conditions, the Ratnavali must have been written during the reign of Yajña Sri, because prior to his reign, there is no evidence for the motif of Buddhas on lotuses in the Satavahana kingdom. This hypothesis will hold until someone either discovers an earlier image of a Buddha on a lotus, until someone establishes that this motif existed in some other medium while it was being avoided in stone, or until someone finds convincing evidence proving that Nagarjuna did not write the Ratnavali.