1. Trying to Be Fair

MADHYAMAKA, the philosophy of the middle, is one of the principal interpretations of Mahāyāna Buddhist scriptures and has a lineage of several prolific and revered thinkers in India, Tibet, and China, beginning with Nāgārjuna and Āryadeva in about the second century CE and going on to Candrakīrti and Bhāviveka in the sixth century, Kamalaśīla and Śāntarakṣita in the eighth, and a host of illustrious Tibetan exponents, not the least of which was Tsongkhapa (Tsong kha pa) in the fourteenth century. It has fascinated Western writers from the end of the nineteenth century on, including the major figures in Buddhist philology, like Theodore Stcherbatsky, Louis de La Vallée Poussin, and Étienne Lamotte, and even some well-known philosophers, like Karl Jaspers. Practitioners of Buddhism in the West inspired by, or belonging to, one or another Tibetan school often faithfully endorse the hierarchy of Indian thought as found in the genre of Tibetan works known as grub mtha’ or siddhānta, the doxographical literature that maintains that Buddhist philosophy culminates in the Madhyamaka. The point is incontestable: Madhyamaka, whether in the East or in the West, has often been revered for its depth and has stimulated many of the best minds in Buddhism and in Buddhist studies. I’ll attempt a working philosophical introduction and will seek to bring out what I think is promising philosophically. I will not do a survey of the history and literature of the school, as others have done that already and much better than I can here. I will, however, approach that philosophy in a slightly more backhanded way than is usual. To see what is promising we first need to look seriously at a potentially darker side to Madhyamaka, for there are interesting and well-informed contemporary critics of Madhyamaka who have seen deviousness and fraudulence where most everyone else saw depth.

The first article in this direction was that of Richard Robinson (1972) entitled “Did Nāgārjuna Really Refute all Philosophical Views?” For Robinson, the principal complaint was that Nāgārjuna and the Madhyamaka school were attributing to their opponents notions and positions to which these opponents themselves would never agree. The second major article was 20a follow-up to Robinson by Richard P. Hayes (1994), “Nāgārjuna’s Appeal,” in which the author argued that not only did this Mādhyamika regularly misrepresent his opponent’s positions and thus refute a man of straw, but that his key arguments only appear to work because of a systematic equivocation upon the polysemic term “intrinsic nature” (svabhāva). While Robinson saw a strategy of deliberate misrepresentation, Hayes added equivocation to the would-be sins of Nāgārjuna.¹

That there would be this strong negative turn some time or another is to quite a degree understandable. After all, what could be more irritating to a good, serious scholar than a general idolatry of a philosophy that seemed to him to be a series of bad arguments, misrepresentations, and sloppy or deliberate plays on words? The temptation is great to buck the tide of exaggerated claims. Nonetheless, the Robinson-Hayes type of reaction is short on charity, leading to more heat than light. Worse, it is sometimes short on some potentially relevant information on the complicated points it treats. Indeed, as we shall see, the later Indian and Tibetan Mādhyamika scholastic had taken up accusations similar to those Robinson and Hayes are leveling and had some solutions that involved considerable ingenuity and in some cases, I would argue, significant insights. I think it is clear that at least the impatient tone of Robinson’s and Hayes’s articles is unfair: the Mādhyamika philosopher is much, much less of an amateur; or to put it more strongly, he is less of a trickster or fraud than Robinson and Hayes make him out to be.

It would be too involved and technical in the present context to undertake a blow-by-blow analysis of the passages that modern critics of the Madhyamaka cite. Nor fortunately do I think we need to do this, as we can get our points across with a reconstruction of some general strategies in this school’s argumentation. But before we delve into that, it is worthwhile to point out that the argumentation is not just what should make or break this philosophy, or other philosophies, for us. Even if certain of the different sorts of arguments that we find in these Madhyamaka texts might seem unconvincing to us, as they probably often do, it would nonetheless be a mistake to thereby dismiss Madhyamaka thought in general. To take a parallel, I think that many people, other than perhaps certain die-hard analytic philosophers, would think it strangely narrow to dismiss the philosophies of St. Thomas Aquinas or René Descartes purely because of the unconvincingness of the Five Ways or the ontological argument — it would be seen as narrow because somehow these philosophies are more than just those arguments; they involve a certain systematic vision, approach, and method of thinking that is of interest and can be developed further, even if many of the actual 21arguments that Aquinas and Descartes themselves gave might often leave us less than converted. It may be that someone formulates other arguments to arrive at essentially Thomistic or Cartesian conclusions. So I think it may be with Nāgārjuna and the Madhyamaka: even if some of the reasoning that he gave in the second century leaves us puzzled in the twenty-first, the philosophic vision is of interest and could well find support in arguments possibly quite different from those of Nāgārjuna himself. In the last chapter of this book, I suggest some ways in which this update could be pursued. In short, I think the Madhyamaka should be of interest to contemporary scholars, because the system and philosophic vision should be of interest. On the most general level the Madhyamaka is trenchantly asking the question “What is a thing?” This question, as well as the Mādhyamika thinker’s attempted answers, should be of interest to philosophers, be they analytic philosophers concerned with issues of realism, antirealism, and quietism, or so-called continental philosophers, such as the Heideggerians meditating on Die Frage nach dem Ding.

I

For our purposes, that is, to try to reopen the debate on Madhyamaka in a fairer way and bring out some of what seems promising in the Madhyamaka vision, we need to do two things: have a working description of what Madhyamaka is and dispel, to some degree at least, the accusations of sleight of hand and amateurism.

The Madhyamaka is in the first place a philosophy that denies, across the board, that things (whatever they might be) have any intrinsic nature (svabhāva). This lack of intrinsic nature, or what it terms “emptiness of intrinsic nature” (svabhāvena śūnyatā), is considered the ultimate truth (paramārthasatya) or ultimate reality. Now there have been certain interpretations of Madhyamaka that have tended to go in the direction of taking this ultimate truth as a kind of permanent absolute, much more real than the phenomena of our ordinary world, which are supposedly just widespread illusions that the world shares in common due to its general ignorance of this absolute. On the other hand, we also find in Indian and Tibetan literature a carefully developed position that the ultimate, or emptiness of intrinsic nature, is itself nothing more real than the ordinary things that make up our world. To say, therefore, that ordinary things lack intrinsic nature is not to describe a genuine reality lying behind or separate from them but is rather 22to give the final and best account of how ordinary things are; indeed this “no-intrinsic-nature-ness” (niḥsvabhāvatā, naiḥsvābhāvya) is just itself without any intrinsic nature, no more no less.

No-intrinsic-nature-ness is supposedly, however, something that is difficult to realize, something extremely subtle that has a deep effect upon us when we do realize it. Following the usual canonical descriptions, a superficial understanding of “no-intrinsic-nature-ness” even inspires terror, but a genuine understanding is a liberating experience.² Why would this realization be liberating at all, and what would this liberation be like? Of course, there are elaborate scholastic accounts as to how this liberation comes about, who has it, and when. However, I think the fundamental Madhyamaka stance is that people’s thought and language is through and through pervaded with an almost instinctive and confused reification: we grasp at things, and hold that, if they are to exist at all, things must possess consistent intrinsic natures that are what they are completely independently, in themselves. The emotional and ethical life of people is then supposedly directly conditioned by this systematic reification. One of the many surprising Madhyamaka positions is that the ordinary person is also fundamentally in the dark about how he or she reifies ordinary things. Instead of the Madhyamaka advocating a simple acquiescence in the banal, there thus are significant discoveries to be undertaken. Understanding emptiness and being liberated is understanding the ordinary, or the so-called “customary truth” or “customary reality” (saṃvṛtisatya),³ as is, stripped of our all-pervasive reifications of intrinsic natures and related confusions; it is an understanding of saṃvṛtimātra — what is just customary, as is, and no more.⁴

So much for the basic picture of Madhyamaka as I see it. Clearly the key term here is svabhāva, intrinsic nature. Broadly speaking, a recurrent semantic feature of the term, whether in Madhyamaka or in other philosophies, is that svabhāva is something or some property that exists objectively and genuinely occurs in, or qualifies, certain things; it is thus to be contrasted with an appearance that is absent from, or fails to correspond to, the thing themselves. Thus, to take the stock Indian analogy: when a striped rope is seen as a snake, the pseudo-snake that appears is not present in, or corresponding to, the striped rope. In the Madhyamaka texts we find this fundamental sense expressed in terms of designations and their bases: to say that something has an intrinsic nature and is not just a mere appearance or a mere designation (prajñaptimātra) due to language and thought means that it withstands logical analysis and that it is findable or obtainable by reason in the “basis of designation.” The pseudo-snake is obviously not findable anywhere in the striped rope — a fact that any worldling can verify — but if we switch 23to a more sophisticated level, that of ultimate analysis of the mode of being of things, then according to the Madhyamaka, nothing is fully findable in its bases, and in that sense, nothing has intrinsic nature.

Note that the formulation I have adopted in terms of x having a svabhāva implying that x and its properties are findable when one searches logically, or equivalently, that x and its properties have the ability to withstand logical analysis (rigs pas dpyad bzod pa), is not literally what occurs in the texts of Nāgārjuna himself. However, the locution “ability to withstand logical analysis” and variants upon this terminology are prominently used by most major Indian Mādhyamika writers, who say that customary things exist for us only insofar as they are not analyzed (avicāratas), or to take the striking formulation of Śrīgupta and Atiśa, they are “fine [only] when not analyzed” (avicāraramaṇīya).⁵ “Findable/obtainable/perceptible as existent under analysis” figures in such texts as, for example, Madhyamakāvatāra 6.160, where Candrakīrti discusses the so-called sevenfold reasoning (rnam bdun gyi rigs pa) and says that when yogins analyze things, the latter are not found (rnyed pa) to have any of the seven possible relationships to parts.⁶ Under analysis, wholes are neither identical with nor different from their parts, meaning that when we look for, or analyze, what we take to be the intrinsic nature of something like a cart in terms of possible part-whole relations, we come up empty-handed: we don’t find (apprehend/perceive) any coherent, unassailable version of what this cart or its cartness could be. And in that sense, we don’t find any real thing: the customary cart only exists unanalyzed. These ways of interpreting Nāgārjuna are probably present in one way or another in all the important currents of Indian Madhyamaka philosophy and especially so in the later Indian Madhyamaka works, such as those of Candrakīrti, Jñānagarbha, Śrīgupta, Śāntarakṣita, and Kamalaśīla. There is a quasi consensus among commentators on this unfindability under analysis, and I see no reason to deny that, on this very broad characterization at least, they may well have gotten Nāgārjuna pretty much right.

As we shall discuss shortly, Richard Hayes also focused on something like this sense of svabhāva, but instead of speaking of it being a type of analytically findable identity or intrinsic nature, he spoke of it as being identity simpliciter. Thus for him this usage of svabhāva meant just what something is, its identity, as opposed to what it is not, its difference from other things.⁷ This is not far from accurate as a general account but nonetheless lacks a very important feature in Madhyamaka contexts. The term svabhāva can indeed mean identity, what something is — as for example in Abhidharma texts or when Mādhyamikas themselves endorse the generally recognized verity that fire has the svabhāva of heat — but it is always more than that in the polemical 24contexts where it is being refuted by the Madhyamaka. In those contexts, it is an identity that withstands analysis — that is hence real and not just customary; it is consistent and does not dissolve into contradictions when subjected to logical analysis. This is why I prefer to speak of this sense of svabhāva as an “analytically findable intrinsic nature” or “analytically findable identity,” thus bringing out the fact that the Mādhyamika is arguing against real identity, what something really is. In their polemical attributions of svabhāva to “realists,” or advocates of “real entities” (bhāva), the Mādhyamikas always take this svabhāva as involving a reification, a misguided attempt to confer some sort of an ultimate status to things, a bhāvasvabhāva. As I will try to show below, this idea that realists, and indeed we ourselves, are constantly engaged in reification — that is, a type of distorting projection — is the thread that ties the would-be double use of the term svabhāva together. Before we get to that, however, let us very briefly look at Nāgārjuna’s own use of svabhāva.

In chapter 15 of his most important work, the Mūlamadhyamakakārikā, Nāgārjuna develops what seems to be a different use of “intrinsic nature” from that which we have termed “analytically findable identity.” In the first two verses he speaks instead of an intrinsic nature as something that “cannot arise due to causes and conditions” (na saṃbhavaḥ svabhāvasya yuktaḥ pratyayahetubhiḥ) and as that which is “not fabricated and is not dependent on anything else” (akṛtrimaḥ svabhāvo hi nirapekṣaḥ paratra ca).⁸ It is this sense that Robinson would take as the point of departure for his critique, arguing that Nāgārjuna foists upon his innocent opponent an acceptance of an absurd and self-contradictory svabhāva (after all, would anyone actually acknowledge that things are independent of causes and conditions?) and proceeds to an all-too-easy refutation of his opponent by saying that this cannot exist.⁹ Equally it is this second sense of svabhāva that Hayes would seize upon to show that the Madhyamaka arguments’ seeming persuasiveness will evaporate when we diagnose the equivocations between the first and second senses of svabhāva. Let us term these two aspects intrinsic-nature-as-findable-identity and intrinsic-nature-as-independent-existence.¹⁰ As I had mentioned earlier, Hayes spoke of the first sense as identity simpliciter, but it’s worth our while to stress that it is actually a type of analytically findable identity, or analytically findable intrinsic nature. So I’ll deliberately take the liberty of modifying Hayes’ formulation a bit and add the qualification of “findability” to identity.

Now, I think that it is quite clear that Nāgārjuna himself made a conscious attempt to fuse these two would-be separate aspects of svabhāva in some sort 25of mutually implicative relationship. In his Mūlamadhyamakakārikā 18.10, we find:

           pratītya yad yad bhavati na hi tāvat tad eva tat / na cānyad api tat tasmān nocchinnaṃ nāpi śāśvatam //

           Whatever x exists in dependence [upon y], that x is not identical to y, nor is it other than y. Therefore it is neither eliminated nor eternal.

The passage is in effect stating that whatever exists dependently — in other words, whatever lacks independent existence — also lacks findable identity, for being a findable identity means, according to Nāgārjuna, that one should be able to say rationally, in a way that stands up to analysis, that a thing is either identifiable with, or is something different from, the things it depends upon. In short: if x and y are dependent, they do not have independent existence; if they do not have independent existence, they do not have a findable identity. And providing that this “if then” paraphrase of Nāgārjuna’s verse is right, it follows by applications of modus tollens that whatever x and y have findable identity must also have independent existence and will not be dependent on anything. The route from findability to independence is thus short: for Nāgārjuna, a findable identity entails independence. If we add the term “intrinsic nature,” it looks like the following entailment holds for Nāgārjuna: intrinsic-nature-as-findable-entity entails intrinsic-nature-as-independent-existence.

There is another famous passage in the Mūlamadhyamakakārikā that rather clearly supports our contention that Nāgārjuna himself saw a link between findability and independence and thus between the two aspects of the semantic range of the supposedly equivocal term svabhāva. This is the extremely well-known verse 24.18, which makes a series of equivalences or mutual implications, including one between being dependent and being something just simply designated by the mind on the basis of other things, especially its parts:

           yaḥ pratītyasamutpādaḥ śūnyatāṃ tāṃ pracakṣmahe / sā prajñaptir upādāya pratipat saiva madhyamā //

           Dependent arising, that we declare to be emptiness. This [emptiness] is [equivalent to] being a designation in dependence. And it [i.e., emptiness] is precisely the middle way.

Without going into a detailed exegesis of this rich and complex verse, we can fairly readily see that an attempt is once again being made to connect the concepts of dependence and unfindability or, equally, findability and independence. Being a designation in dependence upon something else has to be understood as being just a designation and no more (prajñaptimātra): the thing in question cannot be found if we subject it to analysis. What is dependent, then, is unfindable under analysis.

We’ll come back to the type of argumentation strategies used by Nāgārjuna, but in any case, I think we already should have an inkling that a relatively natural reading of Nāgārjuna is to take him as accepting a very close link between the two aspects of svabhāva. In fact, I strongly suspect that the link is not just a conditional in one direction but is rather a biconditional; in other words, x has independent existence if and only if x is analytically findable. Besides findability under analysis implying independence, it looks to me that Nāgārjuna would also accept the converse, that if something were to be of a genuinely independent intrinsic nature (namely, independent of causes, parts, and all activities to understand it), it would have to be somehow findable under analysis — for example, as something completely distinct from parts or from any kind of causal history, and present in an object independently of any conventions, customs, or cognitive and linguistic processes. It would be a genuine absolute completely other than the relative. The candidates for this sort of absolute would be things like nirvāṇa or the “unconditioned” (asaṃskṛta), and it is not surprising that Nāgārjuna subjects them to a trenchant critique of unfindability under analysis. In short, it looks like, for Nāgārjuna at least, findability under analysis and independence are two equivalent, mutually implicative, notions. If we say that sometimes the use of the term svabhāva seems to highlight one aspect and sometimes the other, that does not mean that term svabhāva is thereby equivocal: we may well have two ways to unpack one and the same concept. If that is right, then the minimum result of our discussion up to this point is that it should begin to look rather doubtful that Nāgārjuna is guilty of the gross equivocation of which he is accused by Hayes. He may perhaps have been wrong, he may have even done something that we cannot easily follow, but he did not just simply play on two different senses.

II

A few words about Robinson’s paper and the charge that the Mādhyamika pursued a sophistical strategy of misrepresenting his opponents as accepting 27an absurd and contradictory notion of svabhāva. Robinson had argued: “The validity of Nāgārjuna’s refutations hinges upon whether his opponents really upheld the existence of a svabhāva or svabhāva as he defines the term.”¹¹ He then proceeded to survey the various opponents’ positions to see whether “Nāgārjuna succeed[s] in refuting all views without making any assumptions that are not conceded by the adherents of the particular view under attack.”¹² I won’t evaluate the actual list of “axioms” that Robinson attributes to the Mādhyamika himself and their nonacceptance by the Mādhyamika’s opponents, as fortunately we need not enter into the details of this rather complicated picture. I maintain that the accuracy of those details is in any case a secondary debate. Instead, what is significant for us here is Robinson’s general line of argument that the Mādhyamika philosopher is just practicing sleight of hand because he attributes to his opponents things they do not accept.

It is in fact quite unfair to accuse Nāgārjuna of deliberate misrepresentation simply because he attributed to his opponent things that the same opponent would reject, even vociferously. Certainly if an opponent did not recognize something attributed to him, that in itself would not necessarily mean that he was misrepresented, for it is a natural and even inevitable part of many genuinely philosophical debates between truth seekers that at some point one group says what the other actually thinks, or what they must think if they are to remain consistent with their own basic principles. Of course, there are good and bad, fair and unfair, ways to do this, but the simple fact of one party adopting such a move in a debate does not in itself mean that it is misrepresenting the other or playing, what Robinson calls, a type of sophistical “shell game.”

The issue can be reformulated: Are Mādhyamikas then being sincere and fair to their opponents, and (often or mostly) doing truth-seeking philosophy, or are they just playing at disingenuous word games, purporting to convey significant truth but actually being quite disinterested in whether anything they say is true at all (what an analytic philosopher like Harry Frankfurt would designate with the technical term “bullshit”)?¹³ I think the texts would tend to back mostly the former view, even if the dichotomy is a little simplistic and the line that we want to draw between truth seeking and sophistry is often shaded. In any case, Mādhyamikas, or at least many Mādhyamikas, were quite aware that misrepresentation was a charge they had to answer, and they tried to answer it (as far I can see) with sincerity. Of course, it might be that they were in the long run unsatisfying in their answers, but it’s very hard to agree that they were just performing a trompe l’oeil or deliberately setting up their opponent with facile self-contradictions.

For example, the eighth-century Indian writer Kamalaśīla, in his Madhyamakāloka, was confronted with the objection that his opponents would not themselves acknowledge the logical reasons that the Mādhyamika was using, and that therefore the reasons would be “unestablished” (asiddhahetu). He replied with rather detailed arguments showing that his adversaries would have to accept the reason, in spite of their vociferous denials, because it was entailed by other propositions that they did explicitly accept. The tactic of argumentation is clear and figures repeatedly in the discussion of the “neither one nor many” argument (ekānekaviyogahetu) in the Madhyamakāloka, from folio 215b to 218a in the Sde dge edition (henceforth “D.”).¹⁴ Indeed Kamalaśīla systematically takes up the objections of numerous sorts of Buddhist and non-Buddhists who argue that they have been misrepresented by the Mādhyamika who alleges that things they accept (God, atoms, space, consciousness, and so on) are neither single entities nor several different entities (and are hence unreal). Kamalaśīla’s reply is always the same: the adversary accepts by implication, or has in fact accepted (shugs kyis na khas blangs pa nyid, khas blangs pa kho na), that the pseudo-entity to which he subscribes is neither one thing nor many different things, because he accepts such and such a property of this entity, and that property in fact implies being neither one nor many. The key Tibetan term shugs kyis na that is used here probably translated a Sanskrit original term along the lines of sāmārthyāt, “indirectly,” “by implication.” One example passage from Madhyamakāloka should suffice to show how such “acceptance by implication” works:

           gang dag gis lus la sogs pa rdul phran bsags pa tsam yin pa’i phyir rdul phra rab rnams so so re re gcig pa nyid du kun brtags pa de dag gis kyang sbyor ba dang bsags pa sogs pa’i chos su khas blangs pa’i phyir shugs kyis na gcig pa nyid dang bral pa nyid kyang khas blangs pa kho na yin te /

           Those who imagine that each atom individually is one thing since the body and other such [gross objects] are just simply collections of atoms do in fact also accept by implication that [the atoms] lack oneness, for they accept that [the atoms] have properties (chos), such as being junctions or collections [of parts].¹⁵

Nor should it be thought that this use of “acceptance by implication,” or perhaps more simply “implicit acceptance,” was an occasional flash in the pan of one Indian thinker; it looks to me to be a more or less basic Indo-Tibetan Madhyamaka argument strategy. In Tibet, the same basic general 29method of attributing positions by implication was often known as presenting what the “opponent’s position ends up being” (khas len pa’i mthar thug pa) or, less literally, “the upshot” of his views. One also finds in certain texts the related notion rigs pas ’phul ba, “forcing [a position upon the opponent] through logical reasoning” or ’phul mtshams kyi rigs pa, literally and suggestively translated as a “logical reasoning that pushes one’s limits.”

Interestingly enough, although the term ’phul mtshams is not, to my knowledge, found in any dictionary, including the three-volume Tibetan-Tibetan-Chinese Dictionary, or Bod rgya tshig mdzod chen mo, a quick search of the Asian Classics Input Project database of Tibetan collected works (gsung ’bum) reveals that ’phul mtshams occurs eighty-two times — if we were to add the variants like rigs pas ’phul, ’phul nus pa’i rigs pa, and so on, we would be into hundreds of occurrences, largely in later commentaries, but also occasionally in Tsongkhapa’s works, like his Dbu ma dgongs pa rab gsal on Candrakīrti’s Madhyamakāvatāra. It seems to have escaped lexicographers, and so it escaped Western researchers on Madhyamaka too.¹⁶ That said, I have the strong impression that it was, and still is, pretty much common knowledge among many Tibetan monks. Indeed, this technique of logically imposing principles upon recalcitrant opponents was, according to the late Geshé Tamdin Rabten, essential in actual Geluk monastic debates on Madhyamaka and was typically used when someone wanted to show that having an intrinsic nature, or equivalently “being truly established” (bden par grub pa), or being truly established as such and such a thing, would imply its being such and such in “complete independence from everything whatsoever” (gang la’ang bltos med).¹⁷

In fact, this implication of complete independence from causes, parts, and also from the cognizing mind is amply attested in the texts and is not just an oral tradition. It figures, for example, in an extract from Se ra Rje bstun Chos kyi rgyal mtshan’s Skabs dang po’i spyi don that I translated a number of years ago.¹⁸ I think the implication in question is a compactly formulated version of the principle that we were stressing earlier in looking at the two verses from Nāgārjuna: namely, that findability under analysis, or findable identity, implies complete independence. What is noteworthy in the present context is that this implication, according to experienced debaters like Geshé Rabten, is known as a case of rigs pas ’phul ba. Of course, it would allegedly hold whether the opponent liked it or not — his protests would just be grounds for more debate, a debate that could, nevertheless, be quite sincere and truth-seeking.

III

Let us now take up anew the problem of the link between findability and independence. While it should be clear that the Madhyamaka, with its strategy of implications, upshots, and imposed principles, is probably a much more sophisticated and interesting philosophy than Robinson and Hayes made it out to be, its use of these implications between findable identity and independence is a particularly important step and is, admittedly, not an easy one to fathom. Perhaps at some point much further down the road we might come to the conclusion that we should give up on it all and go back to detecting misrepresentations, equivocation, and other forms of sloppy thinking. But I think that Robinson and Hayes were far too quick in taking that step. So how does the link work? If we grant, as I think we should, that Mādhyamikas were not cunningly equivocating on svabhāva and misrepresenting their opponents, then how did they think that the seeming gap between findability and independence should be bridged?

There is unfortunately no quick answer. To arrive at a charitable interpretation of Madhyamaka thought that would start to answer that question, we need to be clearer on how the mass of seemingly diverse arguments in Madhyamaka might work together. I can see at least three argument strategies that are relevant in this context, all three present in varying degrees in Madhyamaka texts, although for the purposes of this paper I’ll have to be brief and concentrate on broad outlines rather than on the detailed exegesis of specific textual passages. Here are the three:

       1.   Selective use of pan-Indian philosophical debates

       2.   Etymological and purely semantic arguments

       3.   Nonobvious facts about our mental makeup and way of seeing the world.

We first turn to what I am calling “the selective use of pan-Indian debates.” A good example of this is found in Dharmapāla’s commentary on Āryadeva’s Catuḥśataka, where this sixth-century Vijñānavāda commentator on Madhyamaka examines the possibility of vision and other sensory experiences, and embarks on a detailed analysis on whether the subtle matter of the eye has contact with its object.¹⁹ The problem thus is whether the senses act at a distance. This is a bona fide pan-Indian debate, known as the problem of prāpyakāritvavāda (“action by contact”) and aprāpyakāritvavāda (“action without contact”), and interestingly enough, Dharmapāla accepts the critique of each side against the other.²⁰ This acceptance of the absurdities raised by both sides means that the Madhyamaka can then move to the 31desired conclusion that all possibilities of genuine vision are riddled with faults, and that it is hence impossible that people do really see anything. For Dharmapāla, Āryadeva’s end was best served by showing that both sides of the debate were untenable — the two adversaries’ refutations were both accepted, with the result that the eyes, objects, and perceptions were shown to be without intrinsic nature, unable to resist a thorough examination, and hence unfindable under analysis.

We see a similar strategy at work in the Madhyamaka arguments on causality — for example, whether effects and causes are essentially identical or different. Causality was regularly examined in terms of two alternatives, satkāryavāda and asatkāryavāda, or the “theory of the effect existing or not existing” at the time of the cause; predictably Nāgārjuna’s conclusion is that neither alternative is possible.²¹ We also see this use of pan-Indian themes in Nāgārjuna’s use of the recurring controversy on parts (avayava) and wholes (avayavin) — whether parts are identical to or different from wholes, or whether wholes are somehow more or less real than parts. Here too the conclusion is that none of the alternatives are satisfactory and yet, if something genuinely had intrinsic nature, it would have to fit in with one of the two possible positions. The conclusion is immediate: no intrinsic nature.

What is noteworthy for us is that this peculiar type of acceptance of both sides’ refutations in pan-Indian controversies does make for a bridge between intrinsic-nature-as-independent-existence and intrinsic-nature-as-findable-identity. In Mūlamadhyamakakārikā 18.10, cited above, we saw that when x was dependent on y, the two were neither genuinely identical nor different; at least following Candrakīrti’s interpretation in Prasannapadā, the argument turns on the satkāryavāda-asatkāryavāda debate: the cause would be neither the same nor different from the effect. Mūlamadhyamakakārikā 24.18, it will be recalled, made a linkage between being dependent and being a mere designation (prajñaptimātra), that is to say, between dependence and being unanalyzable, unfindable. Here Candrakīrti explicitly glosses the verse in terms of the problem of parts and wholes — everything has parts and is dependent upon parts but cannot be found as the same or different from the parts.

Whether it is causality or part-whole problems that are at stake, the Madhyamaka very often relies upon the same type of argumentation strategy: use one side of a pan-Indian debate off against the other to show that dependence will imply not being findable, and thus show that findability entails independence. If things were dependent