Chapter 4 Buddhapālita and Bhāviveka: The Division between Svātantrika and Prāsaṅgika¶

The previous chapter ended with a question. When Madhyamaka refutes an opponent, should it use only consequences, establishing nothing itself, or also offer an independent argument? In the fifth and sixth centuries, Buddhapālita and Bhāviveka took different paths over this apparently technical issue. Candrakīrti's seventh-century intervention then produced the pattern later called "Svātantrika" and "Prāsaṅgika." This chapter traces the dispute and its central issues; the next develops Candrakīrti's position in detail.

One customary warning is necessary. The terms Svātantrika and Prāsaṅgika are later Tibetan doxographical labels; Buddhapālita, Bhāviveka, and Candrakīrti did not call themselves by these names. The labels identify a real Indian controversy, but the neat division into "two schools" carries the systematizing imprint of later traditions, especially Tibet; see Part XI.¹

1. The Issue: Should Madhyamaka Offer Arguments of Its Own?¶

Return to the methodological problem of the previous chapter. A consequence argument accepts the opponent's premise that "something has intrinsic nature," derives absurdity, and thereby shows that the premise fails, without proposing a Madhyamaka thesis. Is "showing only the opponent's contradiction" sufficient? Or should Madhyamaka go further and use the era's standard logic (Dignāga's three-member argument of proposition, reason, and example) to establish through an autonomous inference that "dharmas lack intrinsic nature"?

Buddhapālita said that it need not. Bhāviveka said that it must. The disagreement begins here.

2. Buddhapālita and Bhāviveka¶

Biographical information about both thinkers is extremely limited. Almost nothing reliable is known of Buddhapālita, and even his dates can be inferred only approximately from his influence on Bhāviveka. His later prominence derives chiefly from Candrakīrti's defense, which raised his consequence method to orthodoxy. Without Candrakīrti, Buddhapālita would probably have remained an obscure commentator rather than a representative of a distinct position. A few stories about Bhāviveka survive in Tibetan sources but are similarly difficult to verify, and reliable historical details remain scarce.

Buddhapālita, around the fifth or sixth century, wrote a commentary on the Mūlamadhyamakakārikā using pure consequence. Take "dharmas do not arise from themselves." He argues that if something arose from itself, it would already be itself, making further production unnecessary and generating endless repetition. This absurdity follows from the thesis of "self-arising"; it is not an alternative claim that Buddhapālita establishes. He merely exposes the contradiction within the opponent's presupposition and asserts nothing himself.

One fact about the materials must be set out here. Buddhapālita's commentary on the Mūlamadhyamakakārikā has no ancient Chinese translation, and his words entered Chinese by one road only, transcription by a critic. The Prajñāpradīpa, in order to refute Buddhapālita, had first to quote him, so that Bhāviveka's attack is what preserved Buddhapālita's own words. Every utterance ascribed to Buddhapālita below comes from a transcription of that kind, and that the transcription is an adversary's work carries an advantage of its own, since these are words both parties acknowledge as evidence. In the transcription, Buddhapālita's refutation of the corner of arising from another runs to a single sentence: production by another likewise is not so, because it would follow that in every place everything arises. The sense is that if arising from another can be established, and whatever differs from the fruit counts as another, then anything may be produced out of anything. No proposition, no reason, no example; the whole passage does one thing only, driving the opponent's thesis out to its own absurd consequence. One further transcription matters as much: how is there making in conventional truth? There is making because self and other, the whole assembly of conditions, are established each in dependence on the rest. Making stands at the conventional truth and is established by mutual dependence. The consequence camp never intended to cancel the everyday level, and what the two camps hold in common is more than what they contest.²

Bhāviveka, around the sixth century, found the procedure unsatisfactory and raised several objections. First, merely revealing absurdity is incomplete; his own words are that you state neither reason nor example, but merely lay down a position and charge the other side with a fault. Under Dignāga's logic, a proper refutation must state a proposition, reason, and example and clearly provide both the reason and its illustration. Second, Buddhapālita failed to state "why" and did not block possible replies. Third, a bare consequence appears to imply a contrary commitment: saying "does not arise from itself" seems to permit "arises from another." Bhāviveka names this danger the inversion in which what is to be established and what establishes it are turned about: let a consequence be read as laying something down, and its proposition and its reason may be inverted just as they stand and turned back for the opponent's own use.² Better, then, to state one's own position openly.³

Bhāviveka therefore held that Madhyamaka too should present independent arguments, or autonomous inferences. This carries a price. To construct an argument both sides accept, the proposition and concepts must be commonly acknowledged and already established. Bhāviveka consequently permits conventional phenomena to be "established by their own characteristics" (svalakṣaṇa) sufficiently to serve as a shared basis of argument. This concession is the place Candrakīrti would later concentrate his attack: at the conventional level it leaves a small space for "self-establishment." Part II, Part 19, Chapters 1 and 2 relocate the point at issue: the price of the concession falls not on the striking surface but on the supply of a commonly admitted basis; see the close of section 4 of this chapter and section 3 of 5.5.

Bhāviveka's principal works include the Prajñāpradīpa, his commentary on the Mūlamadhyamakakārikā and the main text criticizing Buddhapālita; the Tarkajvālā, which directly debates non-Buddhist opponents; and the Karatalaratna, translated into Chinese by Xuanzang. The Karatalaratna supplies a standard autonomous argument: the proposition is that "ultimately, internal and external dharmas do not possess intrinsic nature"; the reason is "because they depend on conditions"; the analogous example is "like an illusion." It is a standard three-member inference with its own proposition, reason, and example and exemplifies his demand that Madhyamaka argue positively.

The first words of that inference call for a separate account. Bhāviveka's autocommentary says that the thesis is accordingly set up with the qualifier "in true nature," and again that it is with respect to ultimate truth that the conditioned is declared empty, and not with respect to the conventional. The qualifier nails the striking surface down upon the reading in terms of real nature, and the conditioned things that the cowherd knows and admits along with everyone else, eyes and the rest, are left untouched throughout. Part II, Part 19, Chapter 1 rules that the procedure does the same work as this book's range clause: both place assertions at the everyday level and at the level of names and conventions outside the scope, and they differ only in manner, the range clause drawing its boundary by type while the qualifier marks each sentence with a modifier as it goes. Chapter 6 of the same part rules further that it is the step which comes nearest to the theorem-shaped reading among all the schools' readings of the two truths. The danger lies in reification alone: should "true nature" be promoted from a mark upon each sentence into a place, as though there were a domain called the ultimate, the reading has fallen back into the two-domains reading that was ruled out. Bhāviveka's own usage stops on the side of the mark.⁴

The meaning of svalakṣaṇa, "own characteristic," in Bhāviveka must be clarified for Candrakīrti's objection to make sense. In Dignāga's fifth-century epistemology, an own characteristic is a thing's irreducibly particular feature and the immediate object of direct perception—the present, individual reality not reached only through conceptual inference. Bhāviveka requires both parties to accept the subject of an argument as "already established." Such establishment implies a degree of cognitive independence: an own characteristic. Candrakīrti argues that once something is said to possess an own characteristic, even qualified as conventional, it has an inherent feature that is not established merely through convention. This conflicts with the fundamental emptiness of intrinsic nature. Bhāviveka regards "conventional own characteristics" as an expedient; Candrakīrti sees a small door opened to intrinsic-nature theory.

3. Candrakīrti's Counterattack: Pure Consequence and No Establishment¶

In the seventh century Candrakīrti defended Buddhapālita and attacked Bhāviveka; the next chapter treats the details. His decisive claim is that Madhyamaka cannot possess an "independent argument." Such an argument requires both parties to regard its subject and terms as "established by a valid instrument of knowledge." Madhyamaka, however, holds that everything is completely empty and accepts nothing as "established by an own characteristic," not even conventionally. By allowing conventional own characteristics in order to argue, Bhāviveka secretly admits that "things establish themselves" and therefore abandons emptiness. Madhyamaka can use only consequences: take premises the opponent already accepts and derive a conclusion the opponent cannot tolerate, while establishing nothing itself.⁵

Candrakīrti's central chain of reasoning, principally in the Prasannapadā, his commentary on the Mūlamadhyamakakārikā, can be reconstructed as follows. An autonomous argument requires both parties to accept that something is "established by a valid instrument of knowledge" (pramāṇasiddha). For an object to be so established, it must possess a degree of epistemic self-establishment—an own characteristic. Madhyamaka holds that nothing possesses an own characteristic, even conventionally. "Convention" consists only of linguistic and social designation, not features inherent in an object. If Madhyamaka offers an autonomous argument, it must accept conventional own characteristics; this contradicts emptiness. Therefore it can use only consequences.

The reply includes an implicit methodological consideration. Pure consequence does not require Madhyamaka to accept the opponent's epistemological framework; it need only derive contradiction within that framework. An autonomous argument requires Madhyamaka itself to accept an instrument of knowledge, exposing it to the question "through what is the instrument established?" and the infinite regress of 5.3, section 4. The Prāsaṅgika strategy is to fight entirely within the opponent's framework without introducing epistemological commitments of its own.

"Refuting without establishing" is thus preserved to the extreme: even a commonly accepted basis on which an argument might stand is relinquished. This pushes the no-thesis line of Nāgārjuna and Āryadeva in 5.1 and 5.2 to its conclusion.

4. What Are They Actually Disputing?¶

Stripped of its technical shell, the controversy asks a single question: does emptiness leave any foothold at the conventional level?

On the surface, Bhāviveka's concern is reasonable. If nothing is established and argument has no common foothold, how can communication occur or anything be demonstrated? Candrakīrti's concern is equally serious. Once argument admits even a trace of own characteristic, emptiness springs a leak. One preserves the operability of debate; the other the thoroughness of emptiness. Each preserves something and loses something.

Beneath their disagreement lies a problem Madhyamaka has never fully explained. Ultimate truth analyzes and eliminates intrinsic nature; what exactly is conventional truth? Is it merely a label for the fact that "phenomena continue to function," or does it possess epistemological status and content of its own? More difficult still: does the act of ultimate analysis itself belong to conventional or ultimate truth? If conventional, every concept and term used in the argument, including the conclusion "emptiness," is established only conventionally; this approaches Bhāviveka's demand for shared foundations. If ultimate, by what is the analysis performed, and where is its own foundation? The tension was not invented by Candrakīrti or Bhāviveka; it is embedded in the framework of the two truths. Candrakīrti's Prāsaṅgika system does appear more internally coherent within Madhyamaka because it avoids the gap of Bhāviveka's "conventional own characteristics." But internal coherence means only that the deductions contain no obvious fissure. It does not dissolve the tension itself. When one says "ultimate analysis reveals that everything lacks intrinsic nature," the statement itself must be made at some level, and how the identity of that foothold is to be determined is a question neither Buddhapālita nor Bhāviveka faced directly.

Part II, Part 18 has closed the question. Chapter 6 of that part takes up Piṅgala's comment on the two truths: the ultimate meaning is in every case had by way of speech, and speech is conventional; therefore, if one does not rely on the conventional, the ultimate meaning cannot be spoken. Speech about the ultimate begins from the conventional, and the conceptual grammar and logic employed at the meta-level are themselves instruments of the conventional level. This is the design of the two-truths framework and not a hole in it. The identity of the foothold is settled accordingly: the act of ultimate analysis is performed at the conventional level, while the constraint it discloses belongs to the ultimate level. The commonly admitted basis Bhāviveka wanted and the establishing of nothing that Candrakīrti guarded each find a place under this answer; see Part II, Part 18, Chapters 5 and 6, and Part 19, Chapter 7.⁶

5. Further Divisions within Svātantrika¶

Svātantrika itself was not uniform. Scholarship commonly distinguishes two branches:

The key difference concerns treatment of the conventional. The Sautrāntika–Madhyamaka line represented by Bhāviveka holds that external objects exist conventionally and independently of cognition. Yogācāra–Madhyamaka, represented by Śāntarakṣita and others, accepts the consciousness-only claim that conventionally only consciousness and its representations exist and that external objects are not independent of consciousness, then proceeds to say that consciousness itself lacks intrinsic nature.

Śāntarakṣita, around the eighth century, was the most influential figure in the latter branch. His Madhyamakālaṃkāra is the programmatic work of the synthesis. After Śāntarakṣita's death, his disciple Kamalaśīla went to Tibet at the invitation of Trisong Detsen and represented the Indian side in the famous Lhasa debate, around 792–794, opposing Chinese Chan monks on gradual versus sudden awakening. Tibetan histories record an Indian victory, after which Kamalaśīla's gradual path became an early Tibetan orthodoxy. This was a decisive event in Tibetan Buddhism's formation; see Part XI.¹

This history means that Tibetan Buddhism's earliest philosophical foundation was Yogācāra–Madhyamaka Svātantrika, not Prāsaṅgika. Tsongkhapa's later ranking of Candrakīrti's Prāsaṅgika as highest explicitly reordered a landscape previously dominated by Svātantrika.

6. The Two Systems and the Resulting Landscape¶

The Svātantrika–Prāsaṅgika division appears methodologically as a dispute over independent arguments, but beneath it lies disagreement over whether emptiness can accommodate a conventional foothold. Svātantrika divides further into Sautrāntika and Yogācāra lines; the latter dominated Tibetan Buddhism's first century. The Geluk tradition's elevation of Prāsaṅgika to the highest view is Tsongkhapa's reordering of Madhyamaka history; see Part XI.

This book again records a recurring observation. The entire controversy unfolds within the enterprise of "dismantling attachment through debate." It disputes how that enterprise should operate and whether it can proceed without leaving any position. From the perspective repeatedly stated in Part III and 5.3, this shows that Madhyamaka had fully become a scholastic debating tradition: even "how to keep arguing without establishing any claim" required a separate controversy. Compared with the silent, nonsystematic original path discussed in 3.5 and 3.6, this is another intellectual environment. The next chapter develops Candrakīrti's Prāsaṅgika Madhyamaka in detail.

References¶