Garfield's translation of and commentary on Nagarjuna's Mulamadhyamakakarika (MMK) should be welcome to all who are interested in Buddhist philosophy, but perhaps especially so to those who wish to introduce Madhyamaka thought to students not well-versed in the Buddhist tradition, and to those wanting a Buddhist philosophical text to recommend to colleagues trained only in the Western philosophical tradition. It is arranged in a way that will prove useful in undergraduate teaching: first the translation of the Tibetan version of MMK is given in toto (1-83), then the same translation is given embedded within Garfield's prose commentary on individual verses (87-359). This makes it possible for students to confront the text directly and grapple with Nagarjuna's arguments on their own before taking in Garfield's reconstruction of them. But it is the audience of Western philosophers that Garfield says (vii, 95) he had in mind in preparing his translation and commentary, so my comments shall principally address the adequacy of the work in this respect. This may be particularly important given Richard Hayes's recent claim ( Journal of Indian Philosophy 22 [1994]: 299-378) that the arguments of MMK are largely fallacious. While Garfield would not have seen Hayes's claims in print before his own work went to press (the two were published roughly simultaneously), it is still an interesting question to what extent Garfield's "deliberately sympathetic" (99) account of Nagarjuna's arguments succeeds in deflecting such criticisms.

Garfield interprets Nagarjuna as steering a middle path between the extremes of absolutism and nihilism. Throughout MMK, Nagarjuna subjects to reductio analysis the various categories that must be used in constructing any theory of the ultimate nature of reality, thereby seeking to show that all things are "empty" (sunya), i.e., devoid of individual essence (ni?svabhava). A key question in Nagarjuna interpretation has always been what to make of this. Supposing that these arguments are sound, what conclusion does Nagarjuna wish us to draw from our supposed inability to theorize reality? According to absolutists, the lesson to be learned is that the ultimate nature of reality utterly transcends rationality and all dichotomizing discourse. According to nihilists, the moral is instead that nothing really exists. Garfield rejects both interpretations. He points out that the absolutist/nihilist dichotomy depends on the assumption that reality can only be characterized through the categories of realist metaphysics that Nagarjuna has sought to discredit. And examination of these categories reveals the further assumption that reality ultimately consists in things that "exist from their own side" (Garfield makes good use of the Tibetan phrase): things bearing strictly independent and fully determinate identity conditions, things whose existence is independent of all contingent human practices such as linguistic conventions. To show that reality cannot be characterized employing these categories without lapsing into incoherence, is to assert neither that reality transcends all attempts at categorization, nor that the very idea of real things is incoherent. Rather it means, Garfield claims, that things can only exist conventionally: through relations of dependence with other things, including such contingent human practices as linguistic conventions. The emptiness of all phenomena is not their utter ineffability, nor their utter non-existence. The emptiness of all phenomena is just their conventional reality, which is itself likewise thoroughly empty.

Garfield does an admirable job of explaining and motivating this interpretive framework. Particularly in his explication of MMK XXIV, he gives an astute and sensitive account of the relations among the Madhyamaka categories of conventional truth, ultimate truth, dependent arising, and emptiness that his framework requires. He is also careful to point out alternative readings of the text that might support those other interpretive frameworks that he rejects. There is, though, one rather curious lapse from his otherwise consistent adherence to his interpretive framework. In commenting on XXI.18 (282), he claims that while our predicament is to be "trapped in conventional reality through the force of the delusion of reification," buddhas have the (to us inconceivable) ability to see things independently of all such concepts and categories. This appears to be positing just the sort of ineffable absolute that Garfield is committed to rejecting. This is particularly surprising given that the verse in question makes no mention of how things appear to buddhas; it merely claims that reification of buddhas is just as misguided as reification of any other category that might be thought to represent the ultimately real. Thus it seems somewhat gratuitous to introduce the notion that there is some distinctive way (presumably some ultimately true way?) that things seem to buddhas.

There is one respect in which Garfield's interpretive framework does lead to readings that are in my view questionable. Anyone who sees Nagarjuna as steering a middle path between absolutism and nihilism must consequently take him to recuperate the conventional at the same time as he rejects the realist notion of an ultimate truth about the ultimate nature of reality. For it is the recuperation of the conventional that allows one to reject absolutism and yet avoid nihilism. Once the notion of an ultimately true characterization of reality has been dismissed as incoherent, if we are to say that things exist in some manner other than as ineffable noumena, we shall have to say that things exist in thoroughgoing interdependence with the conventions that constitute common sense. Things have determinate identity conditions only by virtue of linguistic conventions, which are in turn shaped by a set of contingent human institutions and practices. We might think of this as the "rivers are rivers" moment in Madhyamaka thought.

What strikes me as problematic is Garfield's injection of this "positive" aspect of Madhyamaka thought into contexts where it would appear to be dialectically inappropriate. For instance, Garfield claims to find evidence that in MMK I, Nagarjuna is both rejecting a realist account of causation in terms of necessary connection, and also endorsing a regularity theory of dependent arising that is to be understood as purely conventional. This claim is based in part on the allegedly different treatment of the two terms "cause" and "condition" in this chapter. One place where he sees this difference coming to the fore is in I.4cd, which he translates as:

There are no conditions without power to act.
Nor do any have the power to act.

By "power to act" Garfield understands Nagarjuna to mean a causal connection, in the form of a power or force, that ties together cause and distinct effect. He then explains 4c as claiming that conventionally one may say that conditions (e.g., seed, soil, moisture, and warmth) bring about their effect (the sprout), so that the conditions may be said to have the power to act. But 4d he explains as denying that ultimately there is such a thing as a power to act; the "power to act" of I.4c must be understood as a mere façon de parler that generates no ontological commitment. Thus any account of the relation between conditions and effect must end with the observed regularity or constant concomitance, and cannot aspire to the status of a metaphysical theory of causation, which would necessarily lead to the positing of occult entities such as the power to act.

Likewise, Garfield translates I.10 as:

If things did not exist without essence,
The phrase "When this exists so this will be," would not be acceptable.

As he acknowledges in a note, this verse is usually translated quite differently: "Since things do not exist without essence, the assertion 'When this exists, this will be' is not acceptable." Garfield complains that such a reading "forces an excessively negative interpretation on the chapter as a whole" (119, n.33). Yet it is Garfield who is forcing the translation, in reading the yatas of the Sanskrit original as a conditional. He does this, it seems, in order to have the text at this point explicitly preserve a place in the conventional realm for dependent origination: Nagarjuna is saying to the realist opponent that only by conceding that things do not exist in the ultimate sense (with intrinsically determinate essences), but rather just in the conventional sense (with conventionally determined essences), can one hope to maintain the central Buddhist insight of dependent origination.

My own view is that yatas in I.10 should be read as "since," so that the verse is forcing an unwanted consequence on the opponent: As long as you insist on supposing that there are things with intrinsically determinate essences, you will find yourself unable to maintain this key Buddhist tenet. (Bhavaviveka would insert an "ultimately" before the conclusion.) Similarly, I would follow Candrakirti in taking I.4c as merely bringing out another unwelcome consequence for the opponent: conditions could not be said to be conditions did they not collectively give rise to a causal power or force that originated the effect. This is an unwelcome consequence because in 4ab, Nagarjuna has argued that there is no coherent account of how conditions might give rise to such a causal power. I do not see textual support for Garfield's claim that in MMK I, Nagarjuna is articulating a conventional regularity theory of dependent arising.

I would nonetheless agree with Garfield that the Madhyamika must preserve a place in conventional truth for causal claims. Whether the Madhyamika should develop a full-blown theory of causation (such as a regularity theory) is perhaps best left to Prasangikas and Svatantrikas to sort out. (There is always the worry that theory-building will engender aspirations to the status of ultimate truth.) But the important issue here is whether we should see Nagarjuna as expressly affirming in MMK I that there is a place for causality in the conventional realm. One reason for thinking not is that this would seem to undermine the dialectical force of his arguments. If Nagarjuna's intention is to help his realist opponents overcome their clinging to an ultimate reality populated by things with intrinsic essences, his best strategy would seem to be to pile on absurd consequence after absurd consequence that follows from such a conception of the real. This may not prove fully effective, however, if he indicates at the outset that the conventional realm offers a means of escaping the dilemmas posed by the dialectic. The suggestion is that one must first come to see that rivers are not (ultimately) rivers before one can fully appreciate the fact that rivers are just rivers. Garfield is quite right to see Nagarjuna as intimating the recuperation of the conventional in MMK XXIV. What is not clear is that he would have wanted to explicitly affirm this as early as MMK I.

Our focus so far has been on Garfield's interpretive framework and not on his interpretation of individual arguments. One final point concerning the framework will bring us to the arguments themselves. This point has to do with the question, what the target of Nagarjuna's arguments is. It has been common to object to Streng-style interpretations of Madhyamaka that they take the arguments to be directed exclusively against philosophical theories, and not against the conventional views of common sense. While Wittgenstein might have thought that common sense is perfectly all right as it stands, no Buddhist will share this view. For the Buddhist, the ignorance that is the root cause of suffering must be deeply entrenched within common sense, since suffering is not confined to philosophers but is the common lot of humankind. Those who oppose Streng-influenced interpretations see them as losing sight of this through their assimilation of Nagarjuna to Wittgenstein. Now this is clearly not a significant objection, since one might use selected elements of Wittgenstein's semantics to throw light on Madhyamaka without attributing all of Wittgenstein's views to Nagarjuna. Still it is important to be sensitive to this difference. And Garfield does show concern regarding this issue. In the introduction to his commentary he states that the root delusion of substantialism from which Nagarjuna sees suffering as emanating is not "the product of sophisticated philosophical theory" but is held naively and pretheoretically, so that Nagarjuna's arguments, aimed as they are principally against essentialist metaphysics, will not alone suffice for liberation from suffering, but will nonetheless be of some service (88, n.2). Yet later he says, "The standpoint of emptiness is hence not at odds with the conventional standpoint, only with a particular philosophical understanding of it" (314). Of course he immediately adds that this philosophical perspective is natural and seductive. And in a footnote he then distinguishes between a metaphysical and a non-metaphysical component of common sense. So the picture we end up with is this: Nagarjuna's arguments are directed against philosophical views that naturally grow out of and express systematically what is already inchoately present in universal common sense, though it is present there only as a dispensable part. This picture may well be right. Still its rather scattered presentation leads to some awkwardness.

What might have served Garfield better on this score would have been to explain precisely why the essentialist metaphysics he sees as Nagarjuna's target is so natural and seductive, given our pretheoretical intuitions. For this would better position him to answer the objections of those like Hayes who see Nagarjuna as routinely saddling his opponent with highly questionable assumptions (an objection that the Western philosopher is likely to share). For instance, in his exposition of MMK I.3ab, which reads,

The essence of entities is not present in the conditions, etc.

Garfield explains the argument as claiming that since the essence of the effect (e.g., the heat of fire) is not to be found in any of the conditions out of which it arises (the fuel, friction, etc.) prior to the effect's production, the effect cannot strictly be said to have been produced from those conditions. To the natural question of why the essence cannot be said to have been itself produced from the conditions, Garfield answers that essences are "by definition eternal and fixed. They are independent" (111). But we are not told why. And surely it would seem sensible to suppose that just as the flame is produced by conditions which do not themselves somehow already covertly contain fire, so the heat of the flame is produced in similar fashion. Now in the case of this verse, Garfield is overlooking the possibility that the argument is directed against a satkaryavadin , someone who holds that prior to its production, the effect exists in unmanifest form in its material cause. Read this way, the argument does not require us to agree that essences must be eternal and fixed, for this is a view to which the opponent is already committed. But Nagarjuna does require such an assumption elsewhere, e.g., MMK XV.1-2. And in commenting on that passage, Garfield merely tells us that Nagarjuna holds three views concerning essences: "An essence (or an entity that exists in virtue of possessing an essence) is uncaused, independent of other phenomena, and not fabricated from other things" (221). Once again we are not told why.

This brings us to the heart of the disagreement between Garfield and Hayes. In order to answer the criticism that Nagarjuna is systematically equivocating on svabhava , Garfield needs to explain the source of the svabhava criterion of dharmahood in Abhidharma (something Hayes does on pages 305-7), and then show how this represents a reasonable articulation of common sense realism. This would then allow Garfield to explain why Nagarjuna is justified in attributing to the opponent the view that any account of the ultimate nature of reality must involve things that bear intrinsically determinate essences that consequently cannot undergo alteration. Garfield could thus dispel the appearance that the opponent is a straw person expressly concocted to suit Nagarjuna's dialectical purposes. Some of the elements of such an explanation are to be found here, as when he says "Since the recognition of compounds as unitary phenomena demands conventions of aggregation, to be compounded is, ipso facto, to have a merely conventional existence" (326). But we are not told why anyone would think this to show that compounds are not ultimately real. (Nor is it explained that Nagarjuna shares this view with his Abhidharmika opponent.) Garfield uses the example of a table to introduce the notion that a thing's essence must be thoroughly non-relational in order for that thing to be ultimately real (89-90). The table, he says, is not ultimately but only conventionally real because (1) its existence as an artifact is dependent on certain contingent human practices; (2) its existence is dependent on its parts; and (3) its existence is dependent on the causes (including the material cause) responsible for its origination. But it is far from obvious why any of these should be taken to show that the table is not ultimately real. For instance, we might readily grant that the table would not be identified as such by someone from a culture that lacked such items of furniture; yet a realist would be puzzled as to why this should be taken to show that the table she is currently pounding, rooted as it is in our furniture-manufacturing culture, is not just as real as the rocks and trees outside the window. Now some realists, such as Aristotle, have agreed that artifact kinds are not as fully real as natural kinds, but as Garfield points out in a footnote, the point about intrinsic essences is supposed to hold for all entities, not just artifacts. Point (2) then gets explained (in n.6) in terms of the sorites difficulties that arise in trying to specify the identity conditions for a tree: just how many of its parts must exist together at a given time, how do we draw determinate spatial boundaries around the tree, and the like? This explanation may go some way toward satisfying the Western philosopher. Interestingly enough however, this is one consideration that no Indian Buddhist would have been likely to use in support of the claim that compounds are ultimately unreal. Instead they appear to have relied on a principled argument against all mereological sums. This leaves (3), which we have already seen the realist to be likely to challenge.

The claim that compounded entities are ultimately unreal can, I believe, be given a principled defense. Explicating this defense is crucial to understanding why Abhidharma insisted that only that which bears an intrinsic essence is ultimately real. And unpacking this point is the key to explaining Nagarjuna's strategy for showing that then nothing could be ultimately real. The sorts of sorites difficulties that Garfield relies on at a number of key points do succeed in conveying the flavor of the Madhyamaka critique of realist metaphysics (e.g., 199), but they fail to explain the specific reasons Nagarjuna had for claiming that the doctrine of intrinsic essences is incomprehensible. So in the end I am not sure Garfield has succeeded in giving us all the resources necessary to answer Hayes's criticisms.

Garfield is also not entirely convincing in his account (111-12, 221, 277) of Nagarjuna's use of the notion of parabhava . He sees that Nagarjuna means to argue that if there are no entities with intrinsic essences, there can be no entity whose essence is borrowed from other entities (as the tree's essential properties supervene on the properties of its parts). As Candrakirti puts it, there could be no borrowers if there were no lenders who owned the property they lend out. But Garfield seems to see Nagarjuna as arguing that without intrinsic essences there can be no real differences among entities, so that there can be no "other" from which an entity might be said to borrow its essence. While this is an interesting argument, it is not one that Nagarjuna would have felt compelled to make, given that the opponent already agreed that that which borrows its essence from another entity is not ultimately but only conventionally real.

Also somewhat questionable is Garfield's interpretation of the example of fire in II.3 and the lamp in VI.8-13. He takes the argument of II.2 to be that vision could be ultimately real only if its essence, seeing, were manifested independently of all other existents, such as light and the visible object. He then takes the opponent's introduction of fire in II.3 to be meant to serve as an instance of an entity that performs its characteristic operation on other entities without doing so on itself: If fire can burn other things without burning itself, why cannot vision likewise be said to see other things even though it cannot see itself? But there is some reason to believe that the opponent intends fire to serve as a counter-example to the principle that an entity cannot operate on itself, the principle that Nagarjuna's argument of II.2 relies on. For instance, the commentary on Vigrahavyavartani (VV) verse 33 has the opponent saying that fire illuminates both itself and other things.

The example of the lamp is also illuminated by VV. In MMK VI.8 the opponent introduces the example of the lamp that illuminates both itself and other things, to serve as counter-example to Nagarjuna's argument that origination (the first of three phases of a dharma's existence) cannot originate both itself and a dharma. In VI.9, Nagarjuna responds that where the lamp is there is no darkness, but that illumination consists in the clearing of darkness. Now Garfield takes the point here to be that while the opponent wants origination not itself to be originated, the lamp is itself illuminated and so cannot serve the opponent's purpose. But the opponent does wish origination to itself be originated, namely by itself; the point was to introduce an instance of reflexive operation in order to stop the infinite regress threatened by Nagarjuna's initial argument. Rather, as VV 34 makes clear, the argument is that the lamp, considered qua source of illumination, cannot be said to illuminate itself. For an entity can be said to be illuminated only if it is also capable of existing in the dark, and the lamp ( qua illuminator) never exists in the dark. Garfield is right to see the doctrine of intrinsic essences playing a crucial role in the argument here, but it is not clear that he has the argument quite right.

Also open to question is his interpretation of the argument in I.8 against the (perceptual) object-condition (alambana), which Garfield takes (117-18) to be a version of the Sautrantika time-lag argument for their representationalist theory of perception. While this is an interesting argument, it is not clear that it was current until some time after Nagarjuna. Likewise, Garfield interprets the argument of V.2-5 concerning the relation of essential characteristic ( laka ) and characterized (lakya) as requiring the assumption that both characteristic and characterized exist inherently. But this is a problematic assumption, since the opponent will immediately object that we are here subjecting to analysis the sort of entity that itself exists inherently, and that it therefore makes no sense to demand that the factors our analysis identifies should themselves be supposed to exist inherently. Moreover, it is an assumption that Nagarjuna need not make, since his reductio goes through without it. This is especially important since this then offers an independent argument for the claim that we cannot make sense of the notion of a thing that bears an intrinsic essence.

Among the many interesting and insightful readings of MMK that Garfield offers, I shall mention just two. His discussion of the argument against motion's ultimate reality in MMK I, especially his final summary of the argument (134-35), does a masterful job of bringing out Nagarjuna's reasoning. And in his discussion of the arguments against time in MMK XIX, he makes interesting use of the notion of a meta-time, thereby making far better sense of the argument than others have managed to do.

In fact Garfield's is, in my opinion, the best overall account of Nagarjuna's central work yet to appear in English. His interpretive framework makes clear why Nagarjuna's thought has played a germinal role in Buddhist philosophizing (as well as why it has so often been misunderstood by its critics). His commentary reveals the philosophical sophistication of the Tibetan exegetical tradition (thereby inspiring the reader to wonder what might be found in some of the other commentarial lines to which Garfield alludes). Its chief weakness, in my view, lies in its failure to make as persuasive a case as might be made, for the student of Western philosophy, for the doctrine of emptiness. But perhaps this is not something that can reasonably be expected of a volume that attempts so much else, and so often succeeds so admirably.