Classic and Contemporary Buddhist Works

Nāgārjuna’s Middle Way - Selections


Dedicatory Verse

anirodham anutpādam anucchedam aśāśvatam |
anekārtham anānārtham anāgamam anirgamam ||
yaḥ pratītyasamutpādaṃ prapañcopaśamaṃ śivam |
deśayāmāsa saṃbuddhas taṃ vande vadatāṃ varam ||

I salute the Fully Enlightened One, the best of orators, who taught the doctrine of dependent origination, according to which there is neither cessation nor origination, neither annihilation nor the eternal, neither singularity nor plurality, neither the coming nor the going [of any dharma, for the purpose of nirvāṇa characterized by] the auspicious cessation of hypostatization.

This verse serves not only as a dedication of the work to the Buddha but also as an announcement of purpose. One often finds at the beginning of an Indian treatise a statement indicating why one should read it: how one will benefit from its contents. Nāgārjuna does not explicitly claim here that this work will help one achieve liberation from saṃsāra (it is Candrakīrti who says this is the purpose of the text), but what he does say suggests that is the intention behind his work.

The verse begins with the famous eight negations: “neither cessation nor origination” and so on. (Our English translation reverses the word order of the Sanskrit original in order to make the meaning more easily intelligible.) These negations are said to describe the content of the Buddha’s central teaching of dependent origination (pratītyasamutpāda). The verse thus claims that when we say everything is subject to dependent origination, what this actually means is that nothing really ceases or arises, nothing is ever annihilated nor is there anything eternal, that things are neither really one nor are they many distinct things, and that nothing really ever comes here from elsewhere or goes away from here.

Some of this would come as no surprise to Nāgārjuna’s fellow Buddhists. For instance, the claim that nothing ever really moves (discussed in chapter 2) was widely accepted by Buddhist philosophers as one consequence of the impermanence of existents; the idea that dependently originated entities form a causal series was thought to explain why it appears to us that there is motion. Likewise “Neither annihilation nor the eternal” echoes the Buddha’s claim that dependent origination represents the correct middle path between the extremes of eternalism and annihilationism. This is discussed in chapters 15, 17, 18, 21, 22, and 27, though in ways that go considerably beyond what had been the orthodox understanding. But the claim that there is neither cessation nor origination (discussed in chapters 1, 7, 20, 21, and

25) would have come as a shock to many, since dependent origination was thought to involve (and explain) the origination and cessation of ultimately real entities. And while “neither one nor many” will have a familiar ring to many Buddhists (the Buddha did say that the person in one life and the reborn person in another are “neither identical nor distinct,” e.g., at S II.62, S II.76, S II.113), the standard Abhidharma account of dependent origination relies on the notion that there are many ultimately real dharmas that are mutually distinct. So when (as in chapters 6, 14, and 27) Nāgārjuna claims that what are thought of as two distinct things can ultimately be neither one nor many, this will surprise quite a few.

The purpose is not to shock, though. Instead, the commentators tell us, the point of understanding dependent origination through these eight negations is to bring about nirvāṇa by bringing an end to hypostatizing (prapañca). By hypostatization is meant the process of reification or “thing-ifying”: taking what is actually just a useful form of speech to refer to some real entity. Because the doctrine of dependent origination plays so central a role in the Buddha’s teachings, Abhidharma scholars developed a complex web of concepts designed to explicate it. The suggestion is that the eight negations are meant to remind us that conceptual proliferation can distract us from the real goal—liberation—and perhaps even serve as a barrier to the achievement of the cessation of suffering. (See 18.6, as well as chapters 24 and 27.) But these negations (as well as other allied negations) are not to be accepted because some wise person has told us so. MMK consists of philosophical arguments meant to refute such things as cessation and origination. This work would then be designed to help foster liberation by enlisting the tool of philosophical rationality in the task of putting in their proper place the sorts of conceptual distinctions developed by other Buddhist philosophers. The “proper place” of these concepts is in the toolkit carried by every skillful Buddhist teacher, to be used when appropriate given the circumstances of a particular suffering being. (See 18.5–12.)

1. An Analysis of Conditions

This is the first of several chapters investigating the concept of causation. It is important to note at the outset that in classical Indian philosophy causation is usually understood as a relation between entities (“the seed, together with warm moist soil, is the cause of the sprout”) and not, as in modern science, between events (“the collision caused the motion of the ball”). It begins with a statement of the thesis: that existing things do not arise in any of the four logically possible ways that causation might be thought to involve. The Ābhidharmika opponent (i.e., a member of one of the Abhidharma schools) then introduces a conditions-based analysis of causation, which is a version of the second of the four possible views concerning causation. The remainder of the chapter consists of arguments against the details of this theory that entities arise in dependence on distinct conditions. In outline the chapter proceeds as follows:

1.1 Assertion: No entity arises in any of the four possible ways: (a)  from itself, (b) from a distinct cause, (c) from both itself and something distinct, or (d) without cause.
1.2 General refutation of arising on possibilities a–d
1.3 Opponent: Entities arise (b) in dependence on distinct conditions of four kinds.
1.4 Refutation of relation between conditions and causal activity
1.5–6 Definition of “condition” and argument for the impossibility of anything meeting the definition
1.7–10 Refutations of each of the four conditions
1.11–14 Refutation of thesis that effect arises from conditions


na svato nāpi parato na dvābhyāṃ nāpy ahetutaḥ |
utpannā jātu vidyante bhāvāḥ kva cana ke cana || 1||

1. Not from itself, not from another, not from both, nor without cause:
Never in any way is there any existing thing that has arisen.

This is the overall conclusion for which Nāgārjuna will argue in this chapter: that existents do not come into existence as the result of causes and conditions. There are four possible ways in which this might be thought to happen, and he rejects all of them. According to the first, when an effect seems to arise, it does so because it was already in some sense present in its cause; its appearance is really just the manifestation of something that already existed. The second view claims instead that cause and effect are distinct entities. The third has it that cause and effect may be said to be both identical and distinct. The fourth claims that things originate without any cause; since there are thus no causes, an originating thing could not be said to originate either from itself or from something distinct—it does not originate from anything.

We follow Ye 2011 and accordingly diverge from translations that follow the La Vallée Poussin edition, in reversing the order of the second and third verses of this chapter. (This ordering is clearly attested to by the Akutobhayā and the commentaries of Buddhapālita and Bhāviveka.) On this reading, general arguments against all four views are given in the next verse. But in his comments on this verse Bhāviveka anticipates by giving arguments against the four views. He says, for instance, that the fourth view would mean that anything could be produced from anything at any time, something we know is false.

na hi svabhāvo bhāvānāṃ pratyayādiṣu vidyate |
avidyamāne svabhāve parabhāvo na vidyate || 2||

2. The intrinsic nature of existents does not exist in the conditions, etc.
The intrinsic nature not occurring, neither is extrinsic nature found.

According to the Akutobhayā, 2ab gives the argument against the first possibility mentioned in verse 1, that an existent arises from itself (the view known as satkāryavāda). The argument is that if that out of which the existent arose were really that existent itself, then it should have the intrinsic nature (svabhāva) of the existent. But this is simply not the case. Indeed as all the other commentators point out, if this were the case, then arising would be pointless. For instance we want to know the cause of fire because we want to produce something with its intrinsic nature, heat. If that nature were already present in its cause, then it would be pointless to produce fire. For then in order to feel heat we would only need to touch unignited fuel.

Again according to the Akutobhayā, 2cd gives the argument against the second possibility mentioned in verse 1, that an existent arises from something distinct from itself (asatkāryavāda). This would mean that the existent must borrow its nature from its cause, thus making its nature something that is extrinsic (parabhāva). The argument is that in the absence of the intrinsic nature of the existent in question, its extrinsic nature is likewise not to be found. This is because in order for something to exist, its intrinsic nature must occur: There is, for instance, no fire without the occurrence of heat. And something cannot be in the position of borrowing a nature from something else unless it exists. So an existent cannot arise from something distinct. (For more on satkāryavāda and asatkāryavāda see chapters 10 and 20.)

The third possibility is to be rejected on the grounds that it inherits all the faults of the first and second. And according to the Akutobhayā, the fourth is false because it is one of the extreme views rejected by the Buddha. (Other commentators give more philosophically respectable reasons to reject this view.)

catvāraḥ pratyayā hetur ārambaṇam anantaram |
tathaivādhipateyaṃ ca pratyayo nāsti pañcamaḥ || 3||

3. [The opponent:] There are four conditions: the primary cause, the objective support, the proximate condition,
and of course the dominant condition; there is no fifth condition.

The commentators represent this as the view of a Buddhist opponent, someone who holds the second of the four possible views about the relation between cause and effect mentioned in verse 1. Candrakīrti has this opponent begin by rehearsing the reasons for rejecting the first, third, and fourth views. On the first, origination would be pointless, since the desired effect would already exist. We seek knowledge of causes because we find ourselves wanting to produce something that does not currently exist. The third view is to be rejected because it is the conjunction of the first and second, and we already know that the first is false. The fourth view, that of causelessness, is one of the absurd extremes said to be false by the Buddha (M I.408, A I.173). But, the opponent claims, the second view was taught by the Buddha and so should not be rejected.

The classification of four kinds of condition is the Abhidharma elaboration of the Buddha’s teaching of origination. (See AKB 2.64a.)

(1) The primary cause is that from which the effect is thought to have been produced—for example, the seed in the case of a sprout. (2) Only a cognition has an objective support, namely its intentional object, that of which it is conscious. A visual cognition has a color-and-shape dharma as its objective support, an auditory cognition has a sound, etc. (3) The proximate condition is that entity or event that immediately precedes the effect and that cedes its place to the effect. (4) The dominant condition is that without which the effect would not arise. After criticizing the basic notion of causation, Nāgārjuna will take up each of these four types in turn: primary cause in verse 7, objective support in verse 8, proximate condition in verse 9, and dominant condition in verse 10.

Candrakīrti sets the stage for verse 4 by having the opponent answer the question raised by 2cd as follows: “Then, given such a refutation of the view that origination is by means of conditions, the view will be entertained that origination is by means of an action (kriyā). The conditions such as vision and color-and-shape do not directly cause consciousness [as effect]. But conditions are so called because they result in a consciousness-producing action. And this action produces consciousness. Thus consciousness is produced by a condition-possessing, consciousness-producing action, not by conditions, as porridge [is produced] by the action of cooking” (LVP p. 79).

kriyā na pratyayavatī nāpratyayavatī kriyā |
pratyayā nākriyāvantaḥ kriyāvantaś ca santy uta || 4||

4. An action does not possess conditions; nor is it devoid of conditions.
Conditions are not devoid of an action; neither are they provided with an action.

This “action” is supposed to be the causal activity that makes the cause and conditions produce the right kind of effect. It is supposed to explain why only when a seed is planted in warm moist soil does a sprout appear (and why a sprout doesn’t arise from a stone). But if this action is the product of the co-occurrence of the conditions, and thus may be said to possess the conditions, then presumably it occurs when these conditions are assembled. But is this before or after the effect has arisen? If before, then it does not perform the producing activity that makes an event an action. If after, then since the effect has already been produced, the producing activity is no longer to be found. And, adds Candrakīrti, there is no third time when the effect is undergoing production, since that would require that the effect be simultaneously existent and nonexistent, which is a contradictory state.

If, on the other hand, one were to say that the action occurs independently of the conditions, then we would be unable to explain why the productive action takes place at one time and not at others. The action, being free of dependence on conditions, would be forever occurring, and all such undertakings as trying to make a fire would be pointless.

Given that one cannot specify a time when this action occurs, it follows that it does not ultimately exist. And from this it follows that it cannot be ultimately true that conditions either possess an action or do not possess an action.

utpadyate pratītyemān itīme pratyayāḥ kila |
yāvan notpadyata ime tāvan nāpratyayāḥ katham || 5||

5. They are said to be conditions when something arises dependent on them.
When something has not arisen, why then are they not nonconditions?

naivāsato naiva sataḥ pratyayo ’rthasya yujyate |
asataḥ pratyayaḥ kasya sataś ca pratyayena kim || 6||

6. Something cannot be called a condition whether the object [that is the supposed effect] is not yet existent or already existent.
If nonexistent, what is it the condition of? And if existent, what is the point of the condition?

These two verses explain in greater detail the argument of verse 4. The supposed conditions for the arising of a visual cognition—functioning eyes, presence of an object, light, and so on—cannot be said to be conditions at the time when the visual cognition does not yet exist, since they have not yet performed the productive activity required to make them be what are properly called “conditions.” But when the visual cognition does exist, no productive activity is to be found. We might think there must be a third time between these two, a time when the visual cognition is undergoing production. But while we could say this about a chariot, it could not hold of something ultimately real such as a cognition. A chariot might be thought of as something that gradually comes into existence when its parts are being assembled. But precisely because we would then have to say that during that process the chariot both exists and does not exist, we must admit that the chariot is not ultimately real. That we can say this about a chariot shows that it is a mere useful fiction.

This pattern of argumentation, which we might call the “argument of the three times,” will figure prominently in chapter 2. The point of the argument as applied to the present case of origination is that for those who hold that cause and effect are distinct (proponents of the view known as asatkāryavāda), the producing relation can only be a conceptual construction. According to asatkāryavāda, cause and conditions occur before the effect arises. To claim that the effect originates in dependence on the cause and conditions, we must take there to be a real relation between the two items. But that relation is not to be found in either of the two available times. As for the third time, it holds only with respect to conceptually constructed entities such as the chariot. It follows that the relation of production or causation must be conceptually constructed. It is something that we impute upon observing a regular succession of events, but it is not to be found in reality.

na san nāsan na sadasan dharmo nirvartate yadā |
kathaṃ nirvartako hetur evaṃ sati hi yujyate || 7||

7. Since a dharma does not operate whether existent, nonexistent, or both existent and nonexistent,
how in that case can something be called an operative cause?

Candrakīrti explains that by “operative cause” (nirvartakahetu) is meant primary cause, the first of the four kinds of conditions identified in verse 2. A dharma is an ultimately real entity, something with intrinsic nature. The argument is that in order for an entity to perform the operation of producing an effect, it must undergo change, going from the state of not yet having produced the effect to the state of having produced the effect. But an ultimately real entity, a dharma, cannot undergo change when it exists, since its existence just consists in the manifestation of its intrinsic nature. Nor can it undergo change when it does not exist, since at that time there is no “it” to serve as the subject of change. As for the third option, that the dharma is both existent and nonexistent, the commentators explain that this thesis inherits the defects of the first and second theses and that moreover the properties of being existent and being nonexistent are mutually incompatible.

anārambaṇa evāyaṃ san dharma upadiśyate |
athānārambaṇe dharme kuta ārambaṇaṃ punaḥ || 8||

8. A dharma, being existent, is said to indeed be without objective support.
Then why again posit an objective support in the case of a dharma without an objective support?

The object of a mental state such as a visual cognition is said to be the objective support (ālambana-pratyaya) of that cognition. To call this a kind of condition is to say that the cognition cannot arise without its object. The argument against there being such a condition is once again like that of verses 6–7. At the time when a cognition exists, its supposed objective support cannot be said to produce it. Only something that does not yet exist can be produced.

Note that this argument differs from the time-lag argument that Sautrāntikas use to support a representationalist theory of perception. Both arguments rely on the fact that the objective support exists before the cognition. But the Sautrāntika argument uses this fact to argue that the cognition cannot be directly aware of what is called its objective support. The argument here, by contrast, uses this fact to prove that what is called the objective support cannot be said to be a causal condition of the cognition.

anutpanneṣu dharmeṣu nirodho nopapadyate |
nānantaram ato yuktaṃ niruddhe pratyayaś ca kaḥ || 9||

9. Destruction does not hold when dharmas have not yet originated.
Thus nothing can be called a proximate condition; if it is destroyed, how can it be a condition?

The argument here is also similar to that of verses 4–7, only this time directed against the idea of a proximate condition (samanantarapratyaya), the third of the four types of condition. The proximate condition can perform its function neither before nor after the arising of the effect. A proximate condition must undergo destruction in order to bring about its effect: It would not be the immediately preceding condition unless it went out of existence before the effect arose. But before the effect has arisen, it has not yet undergone destruction. And once it has undergone destruction, since it no longer exists, it cannot be said to be productive of an effect.

bhāvānāṃ niḥsvabhāvānāṃ na sattā vidyate yataḥ |
satīdam asmin bhavatīty etan naivopapadyate || 10||

10. Since things devoid of intrinsic nature are not existent, “This existing, that comes to be” can never hold.
“This existing, that comes to be” is one standard formulation of dependent origination, the Buddha’s doctrine of causation. The “this” in the formula is identified by the Ābhidharmika as the dominant condition (adhipati-pratyaya), the fourth type of condition mentioned in verse

2. The claim here is that there can be no such dominant condition for things that are ultimately real. The argument is that anything that did originate in accordance with the formula would lack intrinsic nature. We saw it claimed in verses 4–7 that there is no third time when an ultimately real effect is undergoing production. This is because for something to be ultimately real, it must bear its own intrinsic nature and not borrow that nature from other things, in the way in which a chariot borrows its nature (e.g., its size, shape, and weight) from the natures of its parts. And this in turn means that something that is ultimately real must be simple in nature. Something simple in nature either does exist or does not exist; there is no third intermediate state when it is coming into existence. Only things that are not ultimately real, such as a chariot, could be said to undergo production. Hence the formula “This existing, that comes to be” cannot apply to things that are ultimately existent.

na ca vyastasamasteṣu pratyayeṣv asti tat phalam |
pratyayebhyaḥ kathaṃ tac ca bhaven na pratyayeṣu yat || 11||

11. That product does not exist in the conditions whether they are taken separately or together.
What does not exist in the conditions, how can that come from the conditions?

athāsad api tat tebhyaḥ pratyayebhyaḥ pravartate |
apratyayebhyo ’pi kasmān nābhipravartate phalam || 12||

12. If that which does not exist [in them] is produced from those conditions,
how is it that the product does not also come forth from nonconditions?

The argument so far has focused on the conditions. Now it turns to the effect but makes similar points. Here the view in question is that the effect is distinct from its cause and conditions. In verse 11 the difficulty is raised that there is then no explanation as to why this particular effect arises from these conditions. Candrakīrti gives the example of the cloth that is said to arise from the threads, loom, shuttle, pick, and so on. The cloth is not in these conditions taken separately, for the cloth is not found in the separate threads, the loom, etc., and if it were in each of them, then it would be many cloths, not one. Nor is the cloth in the conditions taken collectively or in the assembled state. For when the threads are assembled, the cloth as a whole is not found in each of the many threads that are its individual parts. Consequently the cloth and its conditions must be said to be utterly distinct. In verse 12 it is pointed out that it would then be equally sensible to expect the effect to arise from anything at all—that is, from what would ordinarily be identified as nonconditions with respect to that effect. (Cf. verse 3cd.) For as Bhāviveka points out, threads are just as distinct from curd as they would then be from cloth, so we should expect to be able to get curd from threads.

phalaṃ ca pratyayamayaṃ pratyayāś cāsvayaṃmayāḥ |
phalam asvamayebhyo yat tat pratyayamayaṃ katham || 13||

13. The product consists of the conditions, but the conditions do not consist of themselves.
How can that which is the product of things that do not consist of themselves consist of conditions?

Here the view in question is that the product or effect, while distinct from the cause and conditions, arises from them in that it consists in them or is composed of them. (The Nyāya school held this view.) It differs from the view in question in verses 11–12 in that it restricts the term “condition” to just those things that the effect can be said to be made of. The example used by the commentators is that of the threads and a piece of cloth. Now we can say that the cloth is made up of the threads. But it is not true that a thread is made up of itself. The thread is in turn made up of its parts, such as its two tips and the intermediate parts. But if something is composed of something else, the intrinsic nature of that thing should be found in what it is composed of. For instance the color of the cloth should be found in the threads. And the property of being composed of threads, while found in the cloth, is not to be found in the threads. A thread does not consist of itself; it consists of its tips and the other parts. So the view in question cannot be correct.

tasmān na pratyayamayaṃ nāpratyayamayaṃ phalam |
saṃvidyate phalābhāvāt pratyayāpratyayāḥ kutaḥ || 14||

14. Therefore neither a product consisting of conditions nor one consisting of nonconditions
exists; if the product does not exist, how can there be a condition or noncondition?

As verse 13 showed, the effect cannot be said to be made up of its conditions, since the effect could derive its nature only from things that do not in turn derive their nature from yet other things. The alternative would be to say that the effect is made up of nonconditions. If the cloth is not made up of threads, then perhaps it is made up of straw, which is the condition with respect to a mat but a noncondition with respect to cloth. But this is obviously absurd. So there is no plausible account of the origination of a real effect. And in the absence of a real effect, nothing can be said to be either a condition or a noncondition.

25. An Analysis of Nirvāṇa

Nāgārjuna’s examination of nirvāṇa comes in response to the objection that his doctrine of emptiness would rule out the existence of the state that is supposedly the aim of

the Buddha’s teachings. He responds first by arguing that the same consequence follows from the thesis that there are non-empty things and then by attempting to show that no statement concerning nirvāṇa could be ultimately true. In doing the latter he follows the precedent of the Buddha’s teachings on the so-called indeterminate questions, and the chapter concludes by showing how the doctrine of emptiness can be viewed as an elaboration of the Buddha’s treatment of those disputed points. In outline it runs as follows:

25.1 Objection: If everything were empty there could be no such thing as nirvāṇa.
25.2 Reply: Nonexistence of nirvāṇa also follows from existence of non-empty things.
25.3 Assertion: Nothing can be asserted concerning nirvāṇa. 25.4–6 Refutation of possibility that nirvāṇa is an existent 25.7–8 Refutation of possibility that nirvāṇa is an absence
25.9–10 Tentative solution: Nirvāṇa is neither an existent nor an absence.
25.11–14 Refutation of possibility that nirvāṇa is both an existent and an absence
25.15–16 Refutation of possibility that nirvāṇa is neither an existent nor an absence
25.17  Rejection of four possible views concerning the existence of the Buddha in nirvāṇa
25.18  Application of the same analysis to saṃsāra, which likewise is not existent, an absence, etc.
25.19  –20 There is not the slightest gap between nirvāṇa and saṃsāra. 25.21–23 This analysis likewise disposes of the other indeterminate questions.
25.24 Soteriological consequence: the halting of hypostatization. No dharma was taught by the Buddha.


yadi śūnyam idaṃ sarvam udayo nāsti na vyayaḥ |
prahāṇād vā nirodhād vā kasya nirvāṇam iṣyate || 1||

1. [Objection:] If all this is empty, there is neither origination nor cessation.
Due to abandonment or cessation of what is nirvāṇa then acknowledged?

The opponent raises another objection to the claim that everything is empty. If this were true, then there could ultimately be neither the arising nor the disappearance of phenomena. This much Nāgārjuna has already asserted in 1.1. But in that case, it seems there could be no such thing as nirvāṇa. For nirvāṇa is said to be of two types, with and without remainder. The former involves abandonment of the defilements, so that cessation of rebirth is assured but still involves psychophysical elements resulting from past karma, so one is still embodied. The latter comes about when one’s karma is exhausted, so that the causal series of psychophysical elements is destroyed. Both involve cessation. The former involves the cessation of false views of an existing “I,” while the latter involves cessation of the psychophysical elements. If neither arising nor cessation ultimately occurs, then it seems one cannot attain either form of nirvāṇa, since both require the arising and cessation of really existing things. Consequently the claim that all is empty is incompatible with the teachings of the Buddha.

yady aśūnyam idaṃ sarvam udayo nāsti na vyayaḥ |
prahāṇād vā nirodhād vā kasya nirvāṇam iṣyate || 2|

2. [Reply:] If all this is non-empty, there is neither origination nor cessation.
Due to abandonment or cessation of what is nirvāṇa then acknowledged?

To this Nāgārjuna replies that if we instead believe there are things that are non-empty, then we shall be unable to explain how nirvāṇa is possible. For then arising and cessation are impossible. Bhāviveka and Candrakīrti both explain that this is because something that has intrinsic nature (and hence is non-empty) cannot undergo origination or destruction. This reply might appear to be a tu quoque. But Candrakīrti states that those who hold the doctrine of emptiness do not have this difficulty. And Bhāviveka says all sides agree to the conventional truth of the claim that nirvāṇa is attained. Since he thinks the only truths Mādhyamikas may assert (apart from the doctrine of emptiness) are conventional truths, this means he also believes they can escape the objection of the opponent. The reason for this will emerge in the remainder of the chapter.

aprahīṇam asaṃprāptam anucchinnam aśāśvatam |
aniruddham anutpannam etan nirvāṇam ucyate || 3||

3. Not abandoned, not acquired, not annihilated, not eternal, not ceased, not arisen, thus is nirvāṇa said to be.

In his comments, Candrakīrti quotes a verse attributed to the Buddha to the effect that when all phenomena have ceased, then the notions of “exists” and “does not exist” are impediments to the cessation of suffering. Related ideas are to be found in the Nikāyas. In the Aggi-Vacchagotta Sutta (M I.483), the Buddha says that since enlightened ones have cut off all roots of rebirth, one cannot say of the postmortem enlightened ones that they will be reborn, that they will not be reborn, and so on. (There being no such person, the question simply does not arise.) And in the Kaccāyanagotta Sutta (S II.17, III.134–35) the Buddha says that “exists” and “does not exist” are equally inappropriate extreme views. (Nāgārjuna referred to this sūtra in 15.7.) Putting together the thoughts expressed in these two passages, one can perhaps say the following about “final” nirvāṇa (cessation without remainder). Since the causes of further rebirth have ceased, the liberated one will not be reborn; the causal series of psychophysical elements that constitutes one’s life-series will come to an end at death. So one cannot say that the liberated one exists after death. This is often taken to mean that “final” nirvāṇa amounts to utter annihilation, that the liberated one does not exist after death. And of course this makes nirvāṇa sound distinctly unappealing to many. But on the view being presented in these sūtra passages, that response would be mistaken. Since there is no owner of the elements making up the causal series, it would be inappropriate to describe the ceasing of the causal series as “I will not exist.” Hence neither “exists” nor “does not exist” can be said.

This much virtually all Buddhist schools would probably agree on. But Nāgārjuna has something deeper in mind. What that might be will emerge in the remainder of the chapter. Nāgārjuna conducts his examination by considering whether nirvāṇa might be an existent (i.e., a positive being, bhāva), an absence (a negative being, abhāva), both, or neither. In this he is following the standard logical format of the catuṣkoṭi or tetralemma.

bhāvas tāvan na nirvāṇaṃ jarāmaraṇalakṣaṇam |
prasajyetāsti bhāvo hi na jarāmaraṇaṃ vinā || 4||

4. Nirvāṇa is not, on the one hand, an existent; if it were, its having the characteristics of old age and death
would follow, for there is no existent devoid of old age and death.

It is an orthodoxy for Buddhists that all existents are characterized by suffering, impermanence, and nonself. These are said to be the three universal characteristics of existing things. Being subject to old age and death is the standard specification of what it means for something to be impermanent. This specification is also meant to bring out a connection between impermanence and suffering, since it is universally acknowledged that old age and death are unwelcome phenomena. Because nirvāṇa is supposed to be the cessation of suffering, it follows that it could not be characterized by old age and death.

bhāvaś ca yadi nirvāṇaṃ nirvāṇaṃ saṃskṛtaṃ bhavet |
nāsaṃskṛto hi vidyate bhāvaḥ kva cana kaś cana || 5||

5. And if nirvāṇa were an existent, nirvāṇa would be conditioned,
for never is there found any existent that is not conditioned.

The argument here is that all existents are subject to origination, duration, and cessation. So if nirvāṇa were an existent, it would likewise be subject to origination, duration, and cessation. This is obviously incompatible with the claim that nirvāṇa represents the permanent cessation of suffering. There were Abhidharma schools that included in their list of dharmas or ultimate reals certain unconditioned dharmas. The Vaibhāṣikas, for instance, held that space and the two types of cessation were ultimately real unconditioned entities. It can, however, be claimed that these are not to be thought of as existents but rather as absences, so their inclusion does not conflict with the claim that all existents are conditioned. Space, for instance, is defined as what lacks resistance. But see verse 5.2 above, where the example of space is brought under a general rule that is said to hold for all existents (bhāva).

bhāvaś ca yadi nirvāṇam anupādāya tat katham |
nirvāṇaṃ nānupādāya kaścid bhāvo hi vidyate || 6||

6. And if nirvāṇa were an existent, how could one say that nirvāṇa is nondependent?
For never is there found any existent that is nondependent.

The motivation behind calling nirvāṇa nondependent is presumably that this is the only way of insuring that it represents a permanent cessation of suffering. If it were said to depend on conditions, then its continuation would be contingent on those conditions continuing to obtain. The difficulty with calling nirvāṇa nondependent, though, is that this conflicts with the Buddhist orthodoxy that every existing thing originates in dependence on causes and conditions.

yadi bhāvo na nirvāṇam abhāvaḥ kiṃ bhaviṣyati |
nirvāṇaṃ yatra bhāvo na nābhāvas tatra vidyate || 7||

7. If nirvāṇa is not a [positive] existent, how will nirvāṇa be an absence?
Where there is no existent, there is no absence.

According to Bhāviveka, the argument here is directed at the Sautrāntikas, who held that nirvāṇa is a mere absence. (The term we translate here as “absence,” abhāva, we elsewhere render “nonexistent”; we make this change because to do otherwise would wrongly suggest the idea that there is no such thing as the state of nirvāṇa.) Candrakīrti identifies the target as the view that nirvāṇa is the absence of the defilements and birth. The argument against this is, according to Candrakīrti, that then nirvāṇa would be just as impermanent as defilements and birth are. To this it might be objected that nirvāṇa would still have the sort of permanence that is desired; while it would have a beginning in time, it would not have an end. But Candrakīrti claims the view leads to the absurd consequence that nirvāṇa could be attained effortlessly: Since each occurrence of a defilement or of birth is impermanent (like everything else), it ceases regardless of effort. Thus the absence of each defilement and birth will occur regardless of whether or not one strives to attain nirvāṇa.

yady abhāvaś ca nirvāṇam anupādāya tat katham |
nirvāṇaṃ na hy abhāvo ’sti yo ’nupādāya vidyate || 8||

8. And if nirvāṇa is an absence, how can nirvāṇa be nondependent?
There is no absence that exists without dependence.

If we suppose there to be such a thing as an absence, then we must say that its occurrence is dependent on other things, namely those things of which it is the absence. The Nyāya school puts this in terms of its rule: no absence without an existing counterpositive. By this rule there cannot be such a thing as the absence of the horns of a hare, since the horns of a hare do not exist. (There can, though, be the absence of horns from the head of a hare.) But this makes the occurrence of an absence contingent on its counterpositive existing at some place or time. So if the opponent calls nirvāṇa an absence, this once again contradicts the claim that nirvāṇa is nondependent.

So far we have been told that nirvāṇa is not an existent and that it is also not an absence. One seemingly logical response might be to combine these two claims and say that nirvāṇa is neither existent nor an absence. This is just what is proposed, and defended on the basis of the authority of the Buddha, in the next two verses. But we will see that this does not represent Nāgārjuna’s own view, since it is one that he will reject later, in verses 15–16.

ya ājavaṃjavībhāva upādāya pratītya vā |
so ’pratītyānupādāya nirvāṇam upadiśyate || 9||

9. That which when dependent or conditioned comes into and goes out of existence,
that, when not conditioned or dependent, is called nirvāṇa.

prahāṇaṃ cābravīc chāstā bhavasya vibhavasya ca |
tasmān na bhāvo nābhāvo nirvāṇam iti yujyate || 10||

10. And the teacher taught the abandonment of coming into and going out of existence.
Thus it is correct to call nirvāṇa neither existent nor an absence.

The reference of 10ab appears to be to Sn verse 514. Candrakīrti explains that by “coming into and going out of existence” is meant the state of coming and going through a succession of births and deaths. Such a state arises on the basis of the conditions of ignorance and so on as light arises in dependence on the lamp, and it is conceptualized in dependence on the psychophysical elements, as the long is conceived in dependence on the short. Nirvāṇa is said not to be conditioned by ignorance, etc., or not to be conceptualized in dependence on the psychophysical elements. In that case it, being the mere nonoccurrence of conditioning through ignorance, or the mere nonoccurrence of conceptual dependence on the psychophysical elements, cannot be said to be either an existent or an absence. The reasoning here seems to be that of the Personalism (Pudgalavāda) school. This school held that the person, while ultimately real, is neither identical with nor distinct from the psychophysical elements on the basis of which it is named and conceptualized. Given that nirvāṇa is the state of the person when no longer conditioned by or dependent on the psychophysical elements, it stands to reason that nirvāṇa should be thought of as a state that likewise defies classification in terms of the dichotomous concepts of existent and absence.

At this point the text appears to be endorsing the view that nirvāṇa is neither an existent nor an absence. In the next four verses it takes up and rejects the view that nirvāṇa is both an existent and an absence. This might look like support for the view that it is neither. But in verses 15–16 the “neither” option is rejected. This makes it clear that the endorsement of “neither” in the present verse represents the position of an opponent, not Nāgārjuna.

bhaved abhāvo bhāvaś ca nirvāṇam ubhayaṃ yadi |
bhaved abhāvo bhāvaś ca mokṣas tac ca na yujyate || 11||

11. If nirvāṇa were both an existent and an absence,
then liberation would be an absence and an existent, and that is not correct.

The Akutobhayā points out that there is mutual incompatibility between the existence of something and its absence occurring at the same time. Candrakīrti adds that liberation would then be both the arising of composite things and their ending. The same thing cannot arise and end at the same time. So one cannot say that nirvāṇa is both an existent and an absence.

bhaved abhāvo bhāvaś ca nirvāṇam ubhayaṃ yadi |
nānupādāya nirvāṇam upādāyobhayaṃ hi tat || 12||

 12. If nirvāṇa were both an existent and an absence,
then nirvāṇa would not be nondependent, for it would depend on both.

If nirvāṇa is to be ultimately real, then it must be nondependent—that is, something that is not named and conceptualized in dependence on other things. But a nirvāṇa that was both an existent and an absence would be named and conceptualized in dependence on existent composite things and on their absence. And that is clearly impossible.

bhaved abhāvo bhāvaś ca nirvāṇam ubhayaṃ katham |
asaṃskṛtaṃ hi nirvāṇaṃ bhāvābhāvau ca saṃskṛtau || 13||

13. How can nirvāṇa be both an existent and an absence?
For nirvāṇa is noncomposite, and existents and absences are both composite.

For the meaning of “composite” (saṃskṛta) see chapter 13.

bhaved abhāvo bhāvaś ca nirvāṇa ubhayaṃ katham |
tayor abhāvo hy ekatra prakāśatamasor iva || 14||

14. How could nirvāṇa be both an existent and an absence?
For they do not occur in the same place, just as with light and darkness.

Since darkness is the absence of light, to say that nirvāṇa is both a positive existent and an absence is like saying that there can occur both light and darkness in the same place at the same time. The commentators have already said in commenting on verse 11 and verse 12 that existence and absence are mutually incompatible. Nāgārjuna explicitly makes that point here with the example of light and darkness.

 naivābhāvo naiva bhāvo nirvāṇam iti yāñjanā |

abhāve caiva bhāve ca sā siddhe sati sidhyati || 15||

15. The assertion “Nirvāṇa is neither existent nor an absence”

is established only if there were established both absence and existent.

Nāgārjuna here returns to the view that was apparently endorsed in verse 10, that nirvāṇa is neither an existent nor an absence. The claim now is that it also must be rejected. The argument is that this “neither” thesis could be ultimately true only if sense could be made both of the thesis that nirvāṇa is an existent and the thesis that nirvāṇa is an absence. Since those two theses have already been rejected, it follows that “neither” must be as well. The reasoning is that since the “neither” thesis is purported by its proponent to be ultimately true, it must be understood as a negatively phrased positive characterization of nirvāṇa, one that describes it by saying what it is not. But if there is no such thing as the way it is not, then the thesis cannot hold.

If we think of this situation in terms of classical logic, we might suspect that Nāgārjuna is committing a logical error here. He has just rejected the thesis that nirvāṇa is neither an existent nor an absence. The negation of “neither p nor not p” is “either p or not p.” And for the latter to be true, at least one of the two statements p and not p must be true. But in verses 4–8 we were told that both “nirvāṇa is an existent” and “nirvāṇa is an absence” are to be rejected. Has Nāgārjuna become confused by the logic involved in negating the negation of a disjunction?

According to Candrakīrti’s explanation of the argument, Nāgārjuna did not commit a logical error here. The reason is that there are two ways in which a statement can fail to be ultimately true. One way is for it to be ultimately false. If p fails to be ultimately true by being ultimately false, then not p is ultimately true. But the other way is for p to be about something that simply does not really exist. If p is actually not about anything at all, then it can be neither ultimately true nor ultimately false, because it really has no meaning at all (at least not from the perspective of ultimate truth). In other words, in order to say that not p is ultimately true, we have to be able to imagine how it would be possible for p to be ultimately true. The statement p must really be about something in order to be true or to be false. And what was presumably shown in verses 4–8 is that “nirvāṇa is an existent” and “nirvāṇa is an absence” cannot be ultimately true; it was not shown there that these statements are ultimately false. If “nirvāṇa is an existent” and “nirvāṇa is an absence” cannot be ultimately true, then the negation of their disjunction, “nirvāṇa is neither existent nor an absence,” likewise cannot be ultimately true.

naivābhāvo naiva bhāvo nirvāṇaṃ yadi vidyate |
naivābhāvo naiva bhāva iti kena tad ajyate || 16||

16. If nirvāṇa were found to be neither an existent nor an absence,
then by what is it revealed that it is neither existent nor an absence?

To claim that ultimately nirvāṇa is neither an existent nor an absence is to claim that it has this character. The question here is how this could possibly be known. If the psychophysical elements on the basis of which the person is conceptualized have been abandoned, then it cannot be an object of consciousness. Were it thought that it can be cognized by means of the cognition of emptiness, then insofar as the latter involves the absence of all hypostatization, it likewise cannot be grasped as corresponding to the concept “neither an existent nor an absence,” since this is itself an instance of conceptual proliferation. Thus there could be no reason to hold this thesis.

We have now seen reason to reject all four possible views concerning the ontological status of nirvāṇa. The next two verses show that there is a Buddhist precedent for this way of rejecting all four of the lemmas under consideration in verses 4–16.

paraṃ nirodhād bhagavān bhavatīty eva nājyate |
na bhavaty ubhayaṃ ceti nobhayaṃ ceti nājyate || 17||

17. It is not to be asserted that the Buddha exists beyond cessation, nor “does not exist” nor “both exists and does not exist,” nor
“neither exists nor does not exist”—none of these is to be asserted.

tiṣṭhamāno ’pi bhagavān bhavatīty eva nājyate |
na bhavaty ubhayaṃ ceti nobhayaṃ ceti nājyate || 18||

18. Indeed it is not to be asserted that “The Buddha exists while remaining [in this world],”
nor “does not exist” nor “both exists and does not exist,” nor “neither exists nor does not exist”—none of these is to be asserted.

As Bhāviveka makes explicit, the reference here is to the indeterminate questions (avyākṛta) discussed at S III.112, M I.483–88, and S IV.374–

402. These are questions to which it was commonly assumed an enlightened person would know the answer. They include such questions as whether the liberated person continues to exist postmortem, whether the world is eternal, whether the life-force is identical with the body, and so on. Their consideration is usually put in the form of a tetralemma: Is it that p, not p, both p and not p, or neither p nor not p? The questions are called “indeterminate” because for each such possibility, the Buddha rejects that thesis without embracing any other. This has led some modern scholars to suppose that the Buddha does not always obey the laws of classical logic. To reject p, for instance, would seem to commit one to not p, yet the Buddha rejects this as well. But the example of the fire that has gone out (M I.487–88) shows that the Buddha takes each of the four possibilities to involve a false presupposition, for example, that there ultimately is such a thing as the Buddha who might be said to exist, not exist, etc., after cessation. Since this presupposition is false, one can reject the claim that the Buddha exists postmortem as well as the claim that the Buddha does not exist postmortem without violating any law of classical logic. A similar treatment would allow Nāgārjuna to avoid the charge that he contradicts himself when he says (10cd) that nirvāṇa is not to be called either an existent or an absence and also (15–16) that nirvāṇa is not to be said to be neither an existent nor an absence.

na saṃsārasya nirvāṇāt kiṃ cid asti viśeṣaṇam |
na nirvāṇasya saṃsārāt kiṃ cid asti viśeṣaṇam || 19||

19. There is no distinction whatsoever between saṃsāra and nirvāṇa.
There is no distinction whatsoever between nirvāṇa and saṃsāra.

nirvāṇasya ca yā koṭiḥ koṭiḥ saṃsaraṇasya ca |
na tayor antaraṃ kiṃ cit susūkṣmam api vidyate || 20||

20. What is the limit of nirvāṇa, that is the limit of saṃsāra.
There is not even the finest gap to be found between the two.

The same reasoning that leads to the rejection of the four lemmas with respect to nirvāṇa applies as well to saṃsāra. Since all things are, according to Nāgārjuna, empty of intrinsic nature, it follows that ultimately there is no such state as saṃsāra. For in order for saṃsāra to be something about which ultimately true claims could be made, there would have to be ultimately real mental forces that could produce it. And if all things are empty, then there are no mental forces that are ultimately real. Consequently one cannot say that ultimately saṃsāra exists, does not exist, and so forth. Note, however, that this says nothing about the conventional status of nirvāṇa and saṃsāra. A Mādhyamika can still hold it to be conventionally true that nirvāṇa and saṃsāra are very different states, that the former should be sought while the latter should be stopped, and so on.

paraṃ nirodhād antādyāḥ śāśvatādyāś ca dṛṣṭayaḥ |
nirvāṇam aparāntaṃ ca pūrvāntaṃ ca samāśritāḥ || 21||

21. The views concerning what is beyond cessation, the end of the world, and the eternality of the world
are dependent [respectively] on nirvāṇa, the future life, and the past life.

Among the indeterminate questions the Buddha refused to answer are questions concerning whether there is a state of being following the cessation of such composite things as persons, whether the world is limited in space, and whether the world has limits in time. These questions all presuppose one or another answer to the question whether nirvāṇa has a beginning and an end. The argument of chapter 11 was to the effect that there can be no prior and posterior parts of saṃsāra. And in that chapter it was claimed that the same analysis applies to all supposed existents. (See 11.8.) Here its application to the case of nirvāṇa is being utilized.

śūnyeṣu sarvadharmeṣu kim anantaṃ kim antavat |
kim anantam antavac ca nānantaṃ nāntavac ca kim || 22||

22. All dharmas being empty, what is without end, what has an end?
What is both with and without end, and what is neither without end nor having an end?

kiṃ tad eva kim anyat kiṃ śāśvataṃ kim aśāśvatam |
aśāśvataṃ śāśvataṃ ca kiṃ vā nobhayam apy atha || 23||

23. What is identical with this, what is distinct? What is eternal, what noneternal?
What is both eternal and noneternal, and what is then neither?

To say of all dharmas that they are devoid of intrinsic nature is to say that there are no ultimately real entities. And since a statement can be ultimately true only by virtue of correctly describing an ultimately real entity, it follows that no possible view concerning nirvāṇa and the person who attains it can be ultimately true. Notice the inclusion here of a question that was not mentioned earlier—the question of identity and distinctness. One might, for instance, wonder whether the enlightened person is identical with the person who sought enlightenment or is instead some distinct person. Given the present understanding of nirvāṇa, such a question cannot arise.

sarvopalambhopaśamaḥ prapañcopaśamaḥ śivaḥ |
na kva cit kasyacit kaścid dharmo buddhena deśitaḥ || 24||

24. This halting of cognizing everything, the halting of hypostatizing, is blissful.
No Dharma whatsoever was ever taught by the Buddha to anyone.

Since it follows from the universal emptiness of all dharmas that there is ultimately nothing to be cognized, and suffering is said to result from hypostatization (see 11.6), it follows that the realization of emptiness is “blessed” or the cessation of suffering. Of course it also follows from this that the Dharma, the teachings of the Buddha, contains no single statement that is ultimately true. But this, says Candrakīrti, presents no difficulty for the Mādhyamika. For to the extent that the Buddha’s teachings are useful in helping us overcome suffering, they are conventionally true.

Some modern scholars take the text to end here; they claim that the remaining two chapters are later additions and not the work of Nāgārjuna. In support of this claim they point out that the earliest of the existing commentaries, the Akutobhayā, might seem to have ended at this point. What are presented, in currently available editions of this commentary, as its last two chapters (i.e., commentary on chapters 26–27) are for the most part just the verses themselves, with no elucidatory comments. It might also be said in particular that chapter 26 presents no distinctively Madhyamaka views. Still, both Bhāviveka and Candrakīrti took the last two chapters as authentically Nāgārjuna’s work. We take no stand on this controversy.