The familiar principle that characterizes the basic method of the Buddhist scholastic enterprise--reliance on both scripture and reasoning--has been continuously employed in much of the history of Buddhist thought. Just as their Indian predecessors did in the first millennium, Tibetan writers of the second millennium also invoked this principle frequently and even included it in the titles of texts. (1) Thus, Go rams pa (1429-1489) started his criticism of Tsong kha pa's view in the Lta ba'i shan 'byed by stating that he would offer a brief examination of his rival's system "by using scripture and reasoning." (2) Tsong kha pa (1357-1419), on his part, regarded his Lam rim chen mo as "having been drawn from the path of proper analysis using scripture and reasoning." (3) While these two writers disagreed on doctrinal and philosophical points, the general guideline of using scripture and reasoning as a means of scholastic deliberation was held by both.
Going back to the Madhyamaka tradition in India, in the sixth century Bhaviveka stated in Madhyamakahrdaya that he had described reality that is "endowed with reasoning and scripture," which, "being examined by reasoning, remains unharmed." (4) In the early seventh century, Candraklrti asserted in the Prasannapada, his commentary on the Mulamadhyamakakarika, that even Nagarjuna relied on the same method: "Using reasoning and scripture, the acarya [Nagarjuna] composed this text for the purpose of removing doubts and misunderstanding." (5) In a work that has been generally accepted to be authored by Nagarjuna, (6) Ratndvali attests to the customary practice of demonstrating a point by recourse to both scriptural sources and reasoning: "This has been spoken by the Bhagavat, while the reason is also observed in this case." (7) Indeed, in the Abhidharma texts of comparable antiquity, there was already evidence for the application of this established norm, (8) which both Mahayana and early Buddhist authors observed.
What we have seen in these few instances are merely varied ways of expressing the basic underlying principle. In this specific case Buddhist writers have rarely provided self-conscious and detailed accounts of the different manners in which they applied reasoning and--even less frequently--scripture. Moreover, we can hardly expect that the types of scripture and forms of reasoning employed and the understanding of what counts as reasoning could remain unchanged over such a vast span of time. To understand specific ways in which scripture and reasoning are employed in Buddhist texts and how these uses vary in the course of history, scholars of Buddhism must examine concrete cases of application of scripture and reasoning found in the literature. Between the two basic sources of Buddhist scholastic writing, reasoning has consistently received more scholarly attention, especially in the field of Buddhist philosophy. (9) However, Buddhist writers' deployment of scriptural sources also holds special interest as it contains information about reading cultures of different historical periods. Although citations have often been relegated to footnotes and other peripheral spaces, they are inscriptions in texts that can reveal histories of books in religious communities if they are examined with the kind of vigor to which inscriptions have been subjected in the study of Indian history. We cannot locate many Indian Buddhist authors geographically in the same way that inscriptions can be localized, but it is often possible to place a number of writers in the same text tradition in which later writers are aware of, and influenced by, the range of textual sources that their predecessors referred to. In other words, intertextuality plays a much greater role in the citation of texts than in the inscriptional records and has to be accounted for accordingly.
For the purpose of such a study of citation and intertextuality, I have selected some interconnected portions of the texts composed by several Buddhist writers who belong to the Madhyamaka tradition. The pivotal figure of this exercise is Candrakirti, a writer who continued the earlier Indian tradition of Madhyamaka interpretation and whose work also became particularly influential in late Indian Buddhism and much of second-millennium Tibetan Buddhist thought. The comparative angle is supplied here by a consideration of the citations found in related Tibetan works composed by Tsong kha pa and in the earlier Madhyamaka commentaries, especially those written by Buddhapalita and Bhaviveka. Among these four Buddhist philosophers, each author was acquainted with the writings of every remaining author that predated him, if he was preceded by any. Therefore, they self-consciously regarded themselves as a part of the same philosophical tradition.
In his recent work on the seventeenth chapter of the Prasannapada (see n. 38 below), Ulrich Timme Kragh has demonstrated that Candrakirti incorporated a very substantial amount of material from the earlier Madhyamaka commentaries. While my reading of the eighteenth chapter of the Prasannapada against the earlier Madhyamaka commentaries confirms Kragh's conclusion, the current study will bring Candrakirti's influence on Tibetan Madhyamaka thought into view while also emphasizing a form of intertextuality that is peculiar to the use of scriptural citations. With my narrower focus, I will demonstrate that we can uncover a gradual process of Madhyamaka writers' collection of scriptural citations for the purpose of building a hermeneutic apparatus of its scholastic discipline. Buddhapalita already referred to many scriptural sources, but it is in Bhaviveka's and Candrakirti's treatises and commentaries that the process gained momentum. These two authors also demonstrated a clear attention to the Mahayana sutras. An underlying interest of this study is to discover the changing orientations in Buddhist scholastic practices, especially in regard to the use of the various categories of sutra and sastra literature.
THE DISTINCTION BETWEEN SUTRA AND SASTRA
Madhyamaka treatises and commentaries weave into their philosophical analyses an extensive number of citations. The majority of the texts cited come under the generic categories of sastra and sutra. The use of the term sastra designates here commentaries on sastras as well, since sastras in the narrow sense and their commentaries share a great deal in content and method. Before we examine specific instances, it will be useful to discuss these designations very briefly to understand the significance of these textual categories. For the Madhyamaka writers discussed in this article, the term sastra refers first and foremost to the foundational treatises of their tradition. Thus, in reference to the Mulamadhyamakakarika Candrakirti says that Nagarjuna, "the acarya, has written this Madhyamaka Sastra for the purpose of teaching the distinction between the provisional and definitive sutrantas." (10) Statements of this kind offer a typical articulation of the relation between sastras and sutras, in which sastras are seen as second-order formulations of the contents of sutras, revealing what the uninitiated cannot learn by reading sutras directly. In addition to the Mulamadhyamakakarika, influential Madhyamaka sastras include Aryadeva's Catuhshataka, whose verses had been frequently cited at least since the time of Buddhapalita, and Ratnavali, which received particular attention from Candrakirti.
From Vasubandhu's description of his own Abhidharmakosa as a sastra, (11) Tarkajvala's reference to the author of Madhyamakahrdaya as the author of the sastra, (12) and Candrakirti's citation of the verses of his own Madhyamakavatara, (13) we learn that Buddhist writers recognize as sastras not just canonized works of their tradition but their further expositions as well, including systematic treatises that they themselves have composed. Moreover, Madhyamikas are also engaged in conversation with the sastras of competing Indian text traditions. In the Madhymakavatara, Candraklrti speaks of various sastras of Indian philosophical traditions outside Buddhism, which expound notions of self that he endeavors to disprove. (14) Even within the Buddhist fold, Candraklrti singles out the Yogacara scholars Vasubandhu and Dharmapala and the Buddhist epistemologist Dignaga as teachers who have turned their back on the unique tradition of Madhyamaka thought, although he recognizes them as authors of sastras. (15) Bhaviveka also devoted several chapters of his Madhyamakahrdaya to the criticism of rival sastra disciplines ranging from Samkhya, Vaisesika, Vedanta, Mimamsa to the Buddhist tradition of Yogacara. It is evident from these statements that there is a clear awareness of one's own sastra discipline as one among many and that the disparate sastra traditions are often in conflict with one another.
As is well known, sastras are characterized by the deployment of numerous scholastic devices, while sutras, presented as the teachings of the Buddha or the sayings of his disciples, are more varied in content and style. Within sastras and their commentaries, the citation of passages from sutras fulfills, above all, the purpose of justifying the views put forward in the sastras by appealing to the scriptural status of the passages. The hierarchy of religious authority at work here, which assigns higher scriptural authority to sutras while delegating to sastras the status of authorized interpretations of scriptures, would incline us to think that their respective functions in textual practice bear proportionate relation to the degrees of authority that they enjoy. However, the roles that sutras and sastras play in actual scholastic practice do not follow from their nominal status and even change over time, as the following pages will demonstrate.
The sastra-sutra distinction considered so far will serve as a framework and a starting point for the examination of...