Madhyamaka Wiki

VV 29: Three Readings

18 min read

The claim

Verse 29 of Nāgārjuna’s Vigrahavyāvartanī — “If I had any thesis, that fault would apply to me. But I do not have any thesis, so there is indeed no fault for me” — is the locus classicus of the Madhyamaka no-thesis doctrine. Three of the most important interpreters here read it in substantively different ways: Tsongkhapa denies only theses held with intrinsic existence; Gorampa denies autonomous theses while allowing mere theses (དམ་བཅའ་ཙམ་) and reductio theses; Westerhoff denies theses read with realist semantics. Six further readings are on record. Within the Tibetan tradition, Mabja reads the verse by the two truths; the Ninth Karmapa reads it by perspective (no thesis of one’s own, speech from the perspective of others); and an equipoise reading, which confines the verse to meditative equipoise (མཉམ་བཞག་), is stated by Gendün Chöpel, by Mipham as his translators report him, and, on Lopez’s tentative identification, by Tsongkhapa in his answers to Rendawa. Gorampa argues against that reading when a defender of Dolpopa uses it. Outside the tradition, Oetke universalises the denial, Gokhale denies that the verse asserts anything at all, and Galloway makes it a report of what the Madhyamaka negation always yields. (Nine readings are now catalogued; the page title is kept for link stability.) The readings of Tsongkhapa, Gorampa and Westerhoff do not collapse onto one another, nor do they differ by mere emphasis. Each entails distinct downstream commitments about Madhyamaka method, the Two Truths, and the use of Pramāṇa in realising the ultimate. The contradiction is therefore load-bearing rather than terminological, and it is one of the cleanest verse-level points where the hermeneutical divergence between Geluk and Sakya becomes visible.

Evidence for (the contradiction is load-bearing)

Each reading restricts a different parameter. Tsongkhapa restricts content (theses-with-svabhāva); Gorampa restricts kind (autonomous theses, not mere or reductio theses); Westerhoff restricts semantic register (realist semantics); the Ninth Karmapa restricts perspective (the speaker’s own, not that of others); the equipoise reading restricts occasion (meditative equipoise). These are independent dimensions: a Mādhyamika could in principle hold any subset of the restrictions. The fact that the tradition does not in fact converge on any subset — but instead distributes across them — is evidence that v. 29 is genuinely underdetermined and that the readings are doing real interpretive work.

Each reading entails different downstream commitments.

  • Tsongkhapa’s narrow reading is consistent with — and arguably requires — the integration of Pramāṇa into Madhyamaka. If only theses-with-svabhāva are denied, then conventional theses can be defended by svatantra-anumāna, and the Mādhyamika needs an account of what conventional warrant looks like. Tsongkhapa supplies this with a Madhyamaka-transposed Diṅnāga–Dharmakīrti epistemology. (Cross-reference: atisha-tsongkhapa-pramana-divide.)
  • Gorampa’s kind-of-thesis reading does not by itself settle the place of Pramāṇa; his division of the ultimate does. The approximate ultimate (རྣམ་གྲངས་པའི་དོན་དམ་), which rational analysis reaches by negating the four extremes one at a time, is an ultimate measured against a mind that grasps at truth but only a conventional truth measured against the āryas’ equipoise (gorampa-distinguishing-views-1469 ch. 4). Inference therefore reaches the approximate ultimate and no further. That, rather than v. 29, is where Gorampa stands near Atiśa’s line on valid cognition, and Mipham’s two ultimates have the same shape. The Ninth Karmapa goes further: Prāsaṅgikas establish nothing by valid cognition in either truth (Feast pp. 167, 173).
  • Westerhoff’s two-flavour reading makes the no-thesis claim semantic rather than cognitive: Nāgārjuna does not assert that he has nothing in mind, only that whatever he has in mind is not to be supplied with realist truth-conditions. On this reading the Madhyamaka project can be articulated without reference to a special meditative attainment. It is philosophically congenial to the analytic reconstruction of Madhyamaka but historically discontinuous with both Tsongkhapa and Gorampa.

Twentieth-century Geluk-internal articulation of the equipoise reading (Gendün Chöpel). lopez-madmans-middle-way-2006 gives the cleanest single-paragraph statement of the yogin-in-equipoise reading of v. 29 available here. Klu sgrub dgongs rgyan ¶51: “When one opponent who has assertions debates using scripture and reasoning with an opponent without assertions who abides in a state of meditative equipoise, free from verbalization, whatever answers the latter gives all become mere assertions. Thus, there is no place to fit this view of having no assertions within words, sounds, and, particularly, the reasoning of logicians.” Gendün Chöpel names the equipoise reading outright and states it as a structural feature of what prasaṅga-without-thesis even is. ¶77 supplies the cranes / vow-of-silence analogy that blocks the standard Geluk reply (“‘I have no assertion’ is itself an assertion”): saying “Don’t talk” while flying is not the same kind of talk, because the genre of the utterance matters. ¶80–¶84 spells out the assertions-for-oneself / assertions-for-others distinction with the cakravartin analogy. The Geluk objection — that “in the context of ultimate analysis” is itself a propositional commitment — is met by distinguishing the relevant context-of-assertion, not by claiming that the equipoise-position is verbally self-consistent. The own-system / for-others distinction is one he shares with the Ninth Karmapa (below), whom he does not cite; what he adds is the index to meditative equipoise, which the Karmapa does not use. He is not on a “Gorampa side”: Gorampa argues against the equipoise reading (below), and the one time Gendün Chöpel names Gorampa it is with disapproval, on a different question (Adornment ; Lopez 2006, “The Critics”). Nor is the reading peculiar to him within the Geluk. Lopez finds that Tsongkhapa’s answers to Rendawa refer the absence of a thesis to the same equipoise, so that the two agree on its referent while differing on what follows from it; Lopez identifies the passage only tentatively (commentary on §–85). How independent Gendün Chöpel was is itself disputed among his critics: the editor of the 1997 Geluk refutations calls the Adornment an imitation of Gorampa, Shakya Chokden and Taktsang (Lopez 2006, “The Critics”).

A Tibetan argument against the equipoise reading (Gorampa). In Distinguishing the Views ch. 2 (gorampa-distinguishing-views-1469), after Gorampa cites MMK 7:33 against the Jonang claim that the noncomposite thoroughly established nature truly exists, a defender of Dolpopa answers that MMK 7:33, VV 29 and MMK 22:11 apply only within the context of equipoise. In the aftermath (རྗེས་ཐོབ་), he says, one still distinguishes what truly exists from what does not. Gorampa rejects this outright, on two grounds. If the non-establishment of the noncomposite belonged only to equipoise, so would that of the composite, and the analysis of arising, abiding and ceasing would be pointless. And if the Vigrahavyāvartanī passage referred only to equipoise, the Mādhyamika’s refutation of the realists would belong to equipoise rather than to the aftermath. He then turns lines of Drakpa Gyaltsen on those who hold that reality truly exists yet truth does not appear to the āryas’ equipoise. This is the one place in the sources here where a Tibetan author argues against the equipoise reading rather than for or past it. (Cabezón’s note to the verse, translation n. 151, numbers it v. 24; the verse quoted is the one standardly numbered VV 29.)

Gorampa’s own reading is in ch. 3 (.5.1). Both schools accept the four theses that things arise neither from themselves nor from others and so on as mere theses; the difference is that “Svātantrikas believe in autonomous theses, whereas Prāsaṅgikas believe in reductio theses”. An autonomous thesis is proved by a reason established by valid cognition on a subject established for both parties; a reductio thesis is asserted provisionally, once the proponent sees that there are arguments against the opponent’s position. Gorampa follows his teacher Rongtön’s threefold typology of theses here (Cabezón, translation n. 346). What the Prāsaṅgika lacks, then, is an autonomous thesis, not every thesis, and Gorampa’s reading cannot be placed on the occasion axis at all.

A reading by perspective (Ninth Karmapa). In karmapa-feast-fortunate-1578 the Followers of the Middle Way have no theses or positions of their own. They speak of existence, nonexistence, emptiness and the rest “from the perspectives of others”, in a merely conventional way, as a method of leading others to a reality free of theses. The reasonings that refute arising from the four extremes are not the Mādhyamika’s own position; and since others’ positions do not truly exist either, no reasons are needed to refute them. VV 29 is then quoted (pp. 155–157). In Buddhapālita’s voice the Karmapa adds that he holds no thesis of valid cognition that affirms anything in either truth (p. 173). The reading is not indexed to equipoise; the idea that the genuine Mādhyamika abides in the wisdom of the noble ones is the translator’s gloss (Dewar, translator’s introduction, pp. 26–27).

A reading by the two truths (Mabja). Against those who take the verse to mean that a Mādhyamika holds no view and has no means of knowledge, refuting only by consequence, Mabja answers that no claim can be established from the standpoint of ultimate analysis, while on the relative level Mādhyamikas advance arguments, so that a follower of the Middle Way cannot reasonably be said to have no view (mabja-ornament-of-reason, on the homage, “How the Two Truths Are Known”).

A reading on a different axis altogether (Gokhale). Every reading above restricts what the verse denies — its content, its occasion, its semantic register, its truth-level, its assertion-context. gokhale-ways-of-argumentation-2008 restricts nothing and denies instead that v. 29 makes an assertion: “I am not making any proposition” cannot itself be a proposition without paradox, so the negation must be non-truth-functional (a performative negation in Ruegg’s sense, 1986, p. 235) and the utterance a pseudo-proposition. Neither double negation nor the excluded middle reaches it. This answers the standard Geluk objection on logical grounds, where Gendün Chöpel’s cranes analogy (¶77) answers it on grounds of genre. It also separates Gokhale from Oetke, whom he most resembles in method: both reconstruct analytically without the commentators and both make v. 29 the test case, but Oetke takes the verse to be a proposition and reaches a nihilist reading, while Gokhale takes it to be none and reaches an explicitly anti-nihilist one.

The reading has an earlier and weaker form worth recording, because it shows what the performative account was brought in to do. gokhale-nagarjuna-skepticism already has the shape — Nāgārjuna grants that his statement is as empty as any other and declines a language / meta-language distinction (VV 24), and claims for his statements a logical status differing from his opponents’ while sharing their grammar, since he is proving nothing and has no thesis — but stops there, saying that what that logical status is Nāgārjuna does not make clear, and offering only that a śūnyavādin speaks for the sake of illumination (MMK 22.11). Ruegg’s performative negation is what fills the acknowledged blank in the later paper. Two consequences: the reading is arrived at in two steps rather than one, and the blank is a fair thing to put to Gokhale, since whether the second step is the only way to fill it is precisely what needs to know.

A ninth reading, and an outsider landing on a Tibetan fault line (Galloway). galloway-logical-issues-1989 n. 5 reads v. 29 as the direct consequence of his account of Madhyamaka negation: since the properly Madhyamaka prasajya yields a “zero statement”, a statement with no content, “I have no pratijñā” says that Nāgārjuna makes only such statements. Using Ruegg 1983’s distinction, a pratijñā in sense 1 is a statement with content, which he never makes; in sense 2 it is any verbal formulation, which he makes constantly but only procedurally, to refute others. The reading is not indexed to content, occasion, register, truth-level or assertion-context, and unlike Gokhale’s it does not deny that the verse says anything — it makes the verse a description of the method.

Its evidential value for this page lies in what Galloway does next. He quotes Lopez’s presentation of Tsongkhapa — Nāgārjuna “has no inherently existent position” — and rejects it on the ground that one could then hold any position at all by attaching the qualifier, and Nāgārjuna does not appear to give himself that freedom. The objection is not Gorampa’s own, since Gorampa allows the Prāsaṅgika mere and reductio theses; his recorded objection to the qualifier is that the negation of all views and of all four extremes becomes idle if each is negated only as truly existent (Distinguishing the Views ch. 3). But it is an objection a reader on Gorampa’s side of the dispute could take up, made by a logician with no Tibetan stake, from Ruegg’s pratijñā distinction alone, and evidently without knowing that a six-century Tibetan controversy runs through the same point. An outsider reproducing one side of a framework-internal dispute by an independent route is exactly the pattern wants: it is evidence that the disputants are tracking something in the verse rather than in their institutions. The caveat is that his target is Lopez’s 1987 summary rather than Ocean of Reasoning, which is more careful about the qualifier (see Non-affirming Negation on Ocean Homage .1.1.2), so this is evidence about how the Geluk reading presents from outside, not a refutation of it.

Witnesses against collapse, but not two camps. The traditional sources do not behave as though Tsongkhapa and Gorampa are saying the same thing, and they do not line up behind either. The Ninth Karmapa reads by perspective (above). mipham-introduction-middle-way-2002, as the translators report Mipham, puts the absence of assertion at the level of the ultimate in itself, which the āryas see in equipoise; Prāsaṅgika and Svātantrika differ only in what they assert after meditation (translators’ introduction). apple-jewels-middle-way-2018 reports that Atiśa did not accept that the Madhyamaka posits a thesis (Apple’s introduction), and the eleventh-century General Explanation in his lineage cites v. 29 to deny a thesis even conventionally, the ultimate being indicated “by means of nonimplicative negation” ([739.25–740]). Neither the Kadampa denial nor the Karmapa’s is Gorampa’s position, since Gorampa keeps mere and reductio theses. Tsongkhapa’s reading is therefore one position among several in the broader Indo-Tibetan tradition, not its default.

Indian primary-text grounding. westerhoff-vigrahavyavartani-2010 supplies VV v. 29 as primary text in Sanskrit, so the verse can be read directly rather than only through its Tibetan and modern receptions. Two facts then stand out. First, the autocommentary on v. 29 immediately invokes v. 28’s quotation of MMK 24:10 (the Two Truths structure), so any reading that ignores the framework is exegetically vulnerable. Second, the autocommentary does not restrict the denial in any of the three ways the later traditions impose, so each of those readings is a constructive overlay rather than a transparent rendering. This strengthens the contradiction rather than dissolving it.

Evidence against (the contradiction might be reconcilable)

  • Tsongkhapa and Westerhoff are structurally cousins. Both readings restrict the denied class on principled grounds (content vs semantic register) and both leave room for the Mādhyamika to assert plenty of theses. One could argue that Tsongkhapa is a special case of Westerhoff’s reading (theses-with-svabhāva is one species of the genus “theses with realist semantics”). This dissolves the Tsongkhapa-vs-Westerhoff disagreement but leaves the Gorampa contrast standing.
  • Gorampa and Westerhoff might be triangulated via the Two Truths. If Gorampa’s distinction between theses held autonomously and theses asserted provisionally inside the opponent’s conventions is read as a dialectical reflex of the conventional/ultimate distinction, and Westerhoff’s two-flavour-semantics rejection is read as a semantic reflex of the same distinction, then both are restrictions indexed to the same underlying framework — different surface forms of one structural commitment.
  • Tsongkhapa’s reading has a primary-text foothold elsewhere. jinpa-tsongkhapa-qualms-1999 cites VV 26b in defence of the svabhāva-qualifier reading. The Yuktiṣaṣṭikā and Ratnāvalī contain plausibly Tsongkhapa-friendly passages. Tsongkhapa is not a Geluk fabrication imposed on a hostile text — he has Indian textual support, even if he does not have the whole tradition.
  • Tsongkhapa’s Illuminating the Intent Ch 10 supplies primary grounding for the no-thesis question. Working through MMK I.1 as quoted by Candrakīrti at MA 6.8a, Tsongkhapa defends — against critics who claim Prāsaṅgika has “no thesis” — that Madhyamaka does establish the absence of intrinsic arising. He cites Vigrahavyāvartanī and the Prasannapadā together: the proposition is to be read as “things never anywhere arise from any of the four [extremes],” and the negation is non-implicative, but the negation is established. His gloss on the Prasannapadā line “the inferences have as their effect only the negation of what others propose” is that this concerns propositions held with intrinsic existence, not theses simpliciter. On the Prasannapadā’s “as for us, we do not establish this to be either nonexistent or existent; rather, we refute what others impute to be existent and nonexistent. This is because our wish is to clear away the two extremes and establish the path of the middle way,” Tsongkhapa argues that “establish the path of the middle way” is the establishment of a thesis — the simple non-implicative negation of the two extremes. This is the load-bearing Tsongkhapa-side primary-text basis for the narrow (“theses-with-svabhāva”) reading of v. 29, and it addresses, by direct quotation, the Prasannapadā passages on which the denial of theses is usually rested. The Tsongkhapa-versus-Gorampa disagreement is therefore not a quarrel about which Indian texts to honour but about how to read shared Indian textual data. That is itself support for the claim that framework-internal debate is productive.
  • Tsongkhapa’s Ocean of Reasoning Ch 1 is the earlier (1407–08) and more architectural Tsongkhapa-side anchor for the narrow reading. tsongkhapa-ocean-of-reasoning-1408 Ch 1 develops the prasajya / paryudāsa doctrine in conversation with Bhāviveka’s Tarkajvālā definition and dismantles two opponent positions at once: (a) those who claim prasajya-negation cannot be the conclusion of an argument; (b) those who claim that establishing essencelessness by analytic cognition renders essencelessness itself truly existent. The decisive Tsongkhapa-side move at Ocean, Sarnath ed. p. 42 — “all of these inferences merely have the effect of refuting the conclusions of others … this means that these arguments establish the mere elimination of the existence of essence, but do not establish anything else; it therefore does not mean that they refute the existence of essence, but do not establish its nonexistence” — is the same gloss on the same Prasannapadā line that Illuminating the Intent Ch 10 will rehearse eleven years later, now anchored on the MMK rather than on the MA. This strengthens the Tsongkhapa-side primary-text basis without adding another reading. There is a subtle nuance in Ocean. The Homage .1.1.2 attributes the verbal-qualifier rule to Svātantrikas and disavows it for Prāsaṅgikas, while retaining the substantive mode-of-existence distinction. Tsongkhapa’s narrow reading of VV 29 is therefore “theses-with-essential-mode-of-existence,” not “theses-with-the-word-ultimately.” His reading is more flexible than the Gorampa polemic suggests, but the substantive distinction is not negotiable.
  • Path-dependence. Each reading is calibrated to a downstream purpose (Tsongkhapa to pramāṇa-integration, Gorampa to the reductio method, Westerhoff to analytic reconstruction). One could argue that the three readings are not contradictory but different uses of the same verse — like three different tools made from one piece of material. This is the mipham-introduction-middle-way-2002-style “all are Prāsaṅgikas” line transposed to v. 29.
  • Oetke’s universalist reading sits uneasily. Oetke’s claim that v. 29 follows by universal instantiation from “on the level of highest truth there is nothing of any kind” is a further reading, but it is structurally closer to a deflationary-nihilist gloss than to Tsongkhapa, Gorampa or Westerhoff. If Oetke’s reading is wrong, that does not vindicate any of the others; if Oetke’s reading is right, it relegates v. 29 to a corollary of a separate central tenet, and the other readings become competing glosses on a side issue.

Linked pages

Graph View

Backlinks