Classic and Contemporary Buddhist Works

Nāgārjuna’s Middle Way - Introduction

Mūlamadhyamakakārikā

The Mūlamadhyamakakārikā (MMK) by Nāgārjuna (ca. 150 C.E.) is the foundational text of the Madhyamaka school of Indian Buddhist philosophy. It consists of verses constituting twenty-seven chapters. In it, Nāgārjuna seeks to establish the chief tenet of Madhyamaka, that all things are empty (śūnya) or devoid of intrinsic nature (svabhāva). The claim that all things are empty first appears in the Buddhist tradition in the early Mahāyāna sūtras known collectively as Prajñāparamitā, beginning roughly in the first century b.c.e. Earlier Buddhist thought was built around the more specific claim that the person is empty: that there is no separately existing, enduring self, and that the person is a conceptual construction. Realization of the emptiness of the person was thought to be crucial to liberation from saṃsāra. The earliest Mahāyāna texts go considerably beyond this claim, asserting that not just the person (and other aggregate entities like the chariot) but everything is devoid of intrinsic nature. While they assert that all things are empty, however, they do not defend the assertion. Nāgārjuna’s task in MMK is to supply its philosophical defense.

As is usual in texts of this nature, the arguments are presented in highly compressed form and so are extremely difficult to comprehend without a commentary. This is due to the nature and purpose of such texts. A kārikā is a work in verse form that contains a concise formulation of some (often philosophical) doctrine; the kārikās are the individual verses making up the work. Texts of this sort were originally used because it is easier to memorize information when it is put in verse form. The regular cadence that results when a verse is constructed out of its four feet (referred to as a, b, c, and d), each consisting of eight syllables, serves as an important mnemonic aid. On the other hand, it would be difficult to clearly formulate and fully defend a sophisticated philosophical thesis within the form’s constraints. But texts of this genre were not composed with that end in mind. The original expectation seems to have been that the student would commit the verses to memory, recite them to the teacher to demonstrate mastery, and then receive an account from the teacher that fully explained the content of each verse. In time these explanations of individual teachers came to be written down in the form of prose commentaries. It is text plus commentary that together are meant to do the work of formulating and defending the philosophical thesis in question. Memorizing the verses would have given students the outline they need in order to remember the full details of the system spelled out in the commentary.

We know of four Indian commentaries on MMK: the Akutobhayā (author unknown), the Madhyamakavṛtti by Buddhapālita, the Prajñāpradīpa by Bhāviveka, and the Prasannapadā by Candrakīrti. They do not all agree on the interpretation of every verse, and some provide more detailed explanations of particular points than others. But they generally agree on such things as what the argument of a particular verse is and which specific views are the subject of refutation in a chapter. And without this information one would be free to read any number of different interpretations into the verses. Of course we cannot be certain that any of the classical Indian commentaries reflect Nāgārjuna’s original intentions. But it would be presumptuous on our part to suppose that we knew better than they what Nāgārjuna really meant.

Our translation of the verses has been guided by the commentaries. This applies to more than just the question of which English term to choose where the Sanskrit is ambiguous. In many cases a translated verse will contain some material in square brackets. These are words the Sanskrit equivalents of which are not in the original verse itself but without which the verse simply does not make sense. When we supply such bracketed material, it is because the commentaries make clear just what has been omitted. That there will be such omissions in the verses proper is understandable given the constraints imposed by the verse form discussed above. We should add that we have tried quite hard to keep the number of square brackets to a minimum; we have, in other words, been fairly liberal in our interpretation of what is “in the original verse itself.” Where the context seems to make abundantly clear that a certain term has been omitted just for the sake of brevity, we supply its English equivalent without the use of square brackets. But those who wish to check our translation’s fidelity to the Sanskrit original might wish to consult an earlier version that was published in The Journal of Indian and Tibetan Studies, where square brackets are used in a more rigorously scholarly fashion.

Rather than translating any one of the commentaries, we have provided our own running commentary to our translation of the verses of MMK based on the four classical Indian commentaries. We have tried to keep our interpretive remarks to a minimum. Seldom do our elucidations go beyond anything stated by at least one of these authors. It is our hope that the arguments will speak for themselves once the larger context has been properly spelled out. We do each have our own preferred ways of understanding Nāgārjuna’s overall stance and how his arguments are meant to function. But we have tried to avoid using this translation as a vehicle to promote our own views on these matters.

Each chapter of MMK contains an analysis of a particular doctrine or concept, usually one held by some rival Buddhist school. The text as we have it does not tell us whether Nāgārjuna supplied titles for each chapter, and if so what they were. We have generally used the chapter titles supplied by Candrakīrti. But in a few cases where we thought it would be more informative, we employed the title supplied by another commentator.

At this point some general introductory remarks concerning Nāgārjuna’s goals and strategies might not be amiss. In MMK Nāgārjuna is addressing an audience of fellow Buddhists. (In the other work generally accepted as by Nāgārjuna, the Vigrahavyāvartanī, his interlocutors also include members of the non-Buddhist Nyāya school.) Of particular importance is the fact that his audience holds views that are based on the fundamental presuppositions behind the Abhidharma enterprise. Abhidharma is that part of the Buddhist philosophical tradition that aims at filling out the metaphysical details behind the Buddha’s core teachings of nonself, impermanence, and suffering. A number of different Abhidharma schools arose out of significant controversies concerning these details. They held in common, however, a core set of presuppositions, which may be roughly sketched as follows:

1. There are two ways in which a statement may be true, conventionally and ultimately.

a. To say of a statement that it is conventionally true is to say that action based on its acceptance reliably leads to successful practice. Our commonsense convictions concerning ourselves and the world are for the most part conventionally true, since they reflect conventions that have been found to be useful in everyday practice.

b. To say of a statement that it is ultimately true is to say that it corresponds to the nature of reality and neither asserts nor presupposes the existence of any mere conceptual fiction. A conceptual fiction is something that is thought to exist only because of facts about us concept-users and the concepts that we happen to employ. For instance, a chariot is a conceptual fiction. When a set of parts is assembled in the right way, we only believe there is a chariot in addition to the parts because of facts about our interests and our cognitive limitations: We have an interest in assemblages that facilitate transportation, and we would have trouble listing all the parts and all their connections. The ultimate truth is absolutely objective; it reflects the way the world is independently of what happens to be useful for us. No statement about a chariot could be ultimately true (or ultimately false).

2. Only dharmas are ultimately real.

a. To say of something that it is ultimately real is to say that it is the sort of thing about which ultimately true (or false) statements may be made. An ultimately real entity is unlike a mere conceptual fiction in that it may be said to exist independently of facts about us.

b. The ultimately real dharmas are simple or impartite. They are not products of the mind’s tendency to aggregate for purposes of conceptual economy. They are what remain when all products of such activity have been analytically resolved into their basic constituents. They may include such things as indivisible material particles, spatio-temporally discrete occurrences of color and shape, pain sensations, particular occurrences of basic desires such as hunger and thirst, and individual moments of consciousness. (Different Abhidharma schools give somewhat different accounts of what dharmas there are.)

c. All the facts about our commonsense world of people, towns, forests, chariots, and the like can be explained entirely in terms of facts about the dharmas and their relations with one another. The conventional truth can be explained entirely in terms of the ultimate truth.

3. Dharmas originate in dependence on causes and conditions.

While not all Abhidharma schools hold that all dharmas are subject to dependent origination (pratītyasamutpāda), all agree that most dharmas are. And since anything subject to origination is also subject to cessation, most (or all) dharmas are also impermanent.

4. Dharmas have intrinsic nature (svabhāva).

a. An intrinsic nature is a property that is intrinsic to its bearer—that is, the fact that the property characterizes that entity is independent of facts about anything else.

b. Only dharmas have intrinsic nature. The size and shape of a chariot are not intrinsic natures of the chariot, since the chariot’s having its size and shape depends on the size, shape, and arrangement of its parts. The size and shape of the chariot are instead extrinsic natures (parabhāva) since they are not the “its own” of the chariot but are rather borrowed.

c. Dharmas have only intrinsic natures. A characteristic that a thing can have only by virtue of its relation to another thing (such as the characteristic of being taller than Mont Blanc) is not intrinsic to the thing that has it. To suppose that the thing nonetheless has that characteristic is to allow mental construction to play a role in our conception of that which is real. For it requires us to suppose that a thing can have a complex nature: an intrinsic nature—what it itself is like apart from everything else—plus those properties it gets by virtue of its relations to other things. To the extent that this nature is complex, it is conceptually constructed by the mind’s aggregative tendencies.

d. A given dharma has only one intrinsic nature. Since dharmas are what remain at the end of analysis, and analysis dissolves the aggregating that is contributed by mental construction, a given dharma can have only one intrinsic nature.

5. Suffering is overcome by coming to realize the ultimate truth about ourselves and the world.

a. Suffering results from the false belief that there is an enduring “I,” the subject of experience and agent of actions, for which events in a life can have meaning.

b. This false belief results from failure to see that the person is a mere conceptual fiction, something lacking intrinsic nature. What is ultimately real is just a causal series of dharmas. Suffering is overcome by coming to see reality in a genuinely objective way, a way that does not project any conceptual fictions onto the world.

Nāgārjuna does not deny that this is what dharmas would be like. Instead he rejects the further implication that there actually are dharmas. His position is that if there were ultimately real things, they would be dharmas, things with intrinsic nature; but there cannot be such things. Not only are the person and other partite things devoid of intrinsic nature and so mere conceptual fictions, the same holds for dharmas as well. This is what it means to say that all things are empty.

Given the nature of this claim, there can be no single argument that could establish it. Such a “master argument” would have to be based on claims about the ultimate natures of things, and given what would be required to establish that such claims are ultimately true, this would involve commitment to intrinsic natures of some sort or other. Nāgārjuna’s strategy is instead to examine a variety of claims made by those who take there to be ultimately real entities and seek to show of each such claim that it cannot be true. Indeed the commentators introduce each chapter as addressing the objection of an opponent to the conclusion of the preceding chapter. The expectation is that once opponents have seen sufficiently many of their central theses refuted, they will acknowledge that further attempts at finding the ultimate truth are likely to prove fruitless.

This expectation is based in part on the fact that Nāgārjuna employs a number of common patterns of reasoning in his refutations. Once one has seen how a particular reasoning strategy may be used to refute several quite distinct hypotheses, it becomes easier to see how it might apply as well to one’s own preferred view concerning some metaphysical issue. Some patterns that occur particularly often in MMK are the following. It is important to note that in each case the hypothesis that is being refuted is meant by the opponent to be ultimately true.

Infinite Regress: This is meant to show that hypothesis H cannot be true, since the same reasoning that leads to H would, when applied to H itself, lead to a further hypothesis H΄, a similar process would lead to hypothesis H", and so on. But H was introduced in order to explain some phenomenon P. And a good explanation must end somewhere. So H cannot be the correct explanation of P. For examples of this style of reasoning see 2.6, 5.3, 7.1, 7.3, 7.6, 7.19, 10.13, 12.7, 21.13.

Neither Identical Nor Distinct: This is meant to refute a hypothesis to the effect that x and y are related in some way R. If they were, then x and y would have to be either two distinct things or else really just one and the same thing (under two different descriptions). But if x and y were distinct, then x exists apart from y. And if x exists apart from y, x is not characterized by R. So it cannot be ultimately true that x bears R to y. If, on the other hand, x and y were identical, then x would bear relation R to itself, which is absurd. Where R is the relation “being the cause of,” for instance, it would be absurd to suppose that some event could be the cause of itself. For examples of this style of reasoning see 2.18, 6.3, 10.1–2, 18.1, 21.10, 22.2–4, 27.15–16.

The Three Times: This is meant to refute a hypothesis to the effect that x has some property P. For the hypothesis to be true, x must have P at one of the three times: past, future, or present. But, it is argued, for various reasons it cannot be true that x has P at any of the three times. Quite often the third possibility—that of the present moment—is eliminated on the grounds that there is no such thing as a present moment distinct from past and future. The present is, in other words, a mere point without duration; what we think of as an extended present is conceptually constructed out of past and future. But in some cases the third possibility is ruled out on the grounds that the ultimately real dharmas must be impartite simples. For examples of this style of reasoning see 1.5–6, 2.1, 2.12, 2.25, 3.3,

7.14, 10.13, 16.7–8, 20.5–8, 21.18–21, 23.17–18.

Irreflexivity: This is usually deployed when the opponent seeks to head off an infinite regress by claiming that an entity x bears relation R to itself. The principle of irreflexivity says that an entity cannot operate on itself. Commonly cited supportive instances include the knife that cannot cut itself and the finger that cannot point at itself. Nāgārjuna utilizes and supports this principle at 3.2, 7.1, 7.8, 7.28.

Nonreciprocity: This is meant to refute a hypothesis to the effect that x and y are in a relation of mutual reciprocal dependence—that x is dependent on y in a certain way and y is dependent in the same way on x. Instances of this may be found at 7.6, 10.10, 11.5, 20.7.

We have used the La Vallée Poussin edition (LVP) of MMK as the basis of our translation of the verses, though where Ye’s more recent edition (Y) differs substantially from the former, we have generally followed the latter. All references to Candrakīrti’s commentary are given with the pagination of the Prasannapadā in the former edition (LVP). Citations from the other three commentaries are from the Pandeya edition (P). Since the Sanskrit of these commentaries is Pandeya’s reconstruction, in all doubtful cases we checked the Tibetan version. References to MMK are always by chapter and verse; thus “See 1.7” refers the reader to verse 7 of chapter 1. Abbreviations for the titles of other texts we regularly refer to are given at the beginning of the bibliography. Those with an interest in the text-critical study of MMK might wish to consult the following:

MacDonald, Anne. 2007. “Revisiting the Mūlamadhyamakakārikā: Text-Critical Proposals and Problems.” Studies in Indian Philosophy and Buddhism (Tokyo University) 14: 25–55.
Saitō, Akira. 1985. “Textcritical Remarks on the Mūlamadhyamakakārikā as Cited in the Prasannapadā.” Journal of Indian and Buddhist Studies 33(2): 24–28.
———. 1986. “A Note on the Prajñā-nāma-mūlamadhyamakakārikā of Nāgārjuna.” Journal of Indian and Buddhist Studies 35(1): 484–87.
———. 1995. “Problems in Translating the Mūlamadhyamakakārikā as Cited in Its Commentaries.” In Buddhist Translations: Problems and Perspectives, edited by Doboom Tulku, pp. 87–96. Delhi: Manohar.