Chapter 10: Inference for Oneself

Inference for oneself is knowing the object measured by a reason complete in the three modes, and because inference for others reduces to it, the definition serves for inference in general. The chapter recognises the bases the reason depends on, then classifies the reason itself: Dharmakīrti fixes three, since everything on the refuting side reduces to non-observation, while the establishing side requires two kinds of example and so divides into nature-reason and effect-reason. The Vedic school’s four modes are rejected as either weaker than the three modes or redundant.

Condensed from Khenpo Sodargye’s Chinese lecture transcripts on the Pramāṇayuktanidhi, and rendered into English. Root verses are given as the lectures give them; the prose beneath summarises the exposition. On the provenance and the limits of this rendering, see the series introduction.

Chapters: 1. Objects · 2. Cognition · 3. Universals and Particulars · 4. Proof and Exclusion · 5. Expressed and Expresser · 6. Connection · 7. Contradiction · 8. Definitions · 9. Perception · 10. Inference for Oneself · 11. Inference for Others

---

Inference for Oneself

Two parts: (1) the definition; (2) ascertaining the meaning.

a. The Definition

Inference is twofold; as for inference for oneself, By a reason possessing the three modes one knows the basic meaning.

Inference for oneself is knowing the object measured by a reason complete in the three modes — as when someone who did not know the pillar to be impermanent comes, through the reason “produced,” to determine that it is. This definition also serves as the definition of inference in general, since inference for others reduces to inference for oneself: “knowing a previously unknown meaning by a reason complete in the three modes.” Inference for others is treated in the next chapter.

b. Ascertaining the Meaning

Two parts: (1) the cause of comprehension; (2) explaining the probandum proven by the reason.

1. The Cause of Comprehension

Two parts: (1) recognising the bases that the reason depends upon; (2) classifying the reason that depends upon them.

Recognising the Bases the Reason Depends Upon

Three parts: (1) the subject-property, the basis of the first mode; (2) the similar and dissimilar class, the bases of forward and reverse pervasion; (3) the reason for positing these as bases.

The subject-property, basis of the first mode

The subject-property is the predicate and the subject; Their assemblage, when stated, is the true name; In order to hold to one side of it, Either one alone is given a borrowed name.

In “the pillar is impermanent, because it is produced,” the term “subject-property” (宗法) is used in two ways: for the assemblage of subject and predicate (pillar plus impermanence), and for either member alone. Only the assemblage bears the true name; the isolated subject or predicate bears a borrowed name. This is the point compressed in the Pramāṇavārttika’s phrase “the subject-property is pervaded by its part,” expounded at length in Mipham Rinpoche’s great commentary and confirmed in the Auto-commentary on the Pramāṇayuktanidhi. Here “subject-property,” “object of measurement,” and “probandum” all mean the same thing.

The difference between the true name and the borrowed name

Naming as one wishes, relying on the prior naming, And understanding that meaning — that is the true name; By some particular reason connected with it, Understanding it as another thing — that is accepted as the borrowed name.

A true name is set down by the first namer with no further reason: seeing an animal with hump and dewlap he simply called it “ox,” and thereafter people rely on that name to understand and judge. A borrowed name transfers the name of one thing to another for some reason of similarity, relatedness, or even dislike — calling a slow-witted person “ox,” or a wide-mouthed flat-nosed person “big lion.” A borrowed name cannot stand for the thing it was taken from. (Commentaries also analyse true and borrowed names by time, form, and object; not expounded here.)

Considering the conjoined meaning as the real dharma

The assembled meaning is the name “subject-property”; Before all the wise it is established from the very first.

Why is the assemblage of subject and probandum called “subject-property”? For no further reason at all — this is its true name. Among all the wise, Buddhist and non-Buddhist alike who understand inferential reasoning, the term is universally renowned in this sense, exactly as “ox” is for the humped animal.

The parts as borrowed dharma — three parts: (1) the reason for naming; (2) the necessity of naming; (3) the borrowed dharma’s discordance with the actual.

The reason for naming

The naming depends on a reason of similarity to something else, Or of connection [with it].

Borrowed names are posited for two reasons. (i) Similarity: a black flower resembling a crow is called “the arrow-bearing crow”; a dull man is called a “yak.” (ii) Connection, in four sub-cases — the cause named by the effect’s name (the sun’s rays called “the sun”); the effect named by the cause’s name (liquor called “intoxication” in poetics); the whole’s name taken upon a part (“the house is on fire” when only part burns); a part’s name taken upon the whole (the composite sound called “drum-sound”; the cognition arising from attention, faculty, and object called “sense cognition”).

The necessity of naming — two parts: (1) naming the predicate “probandum”; (2) naming the subject “subject-property.”

The necessity of naming the predicate “probandum”

The predicate takes the name “probandum”: Lest a contradictory reason be wrongly grasped as true, In order to dispel such a wrong notion, Or for the sake of word-combination, it is so named.

Two necessities. First, some hold that a contradictory predicate also possesses a similar-class predicate — reasoning that in “the pillar is permanent, because it is produced,” “produced” is established on the pillar and is itself impermanent, hence of the similar class. If a contradictory inference were admitted as genuine, contradictory reasons would cease to exist and every non-reason would become a reason. Calling the bare predicate “probandum” blocks this. Second, word-combination: Dharmakīrti explains in the Pramāṇavārttika that in Sanskrit positing a predicate as another predicate within a single phrase is unreasonable, whereas “probandum” combines well.

The necessity of naming the subject “subject-property” — two parts: (1) the actual necessity; (2) dispelling the view that it is the same as not naming it.

The actual necessity

If one says “subject,” there is the fault of confusion; If one excludes the [other] extremes, time is wasted; If one says the true name, the metre is lost; So that it may be easily known, one says “subject-property.”

Three reasons for saying “subject-property” rather than “subject.” (i) Confusion: “subject” could refer to the pillar of the first mode, or to the similar-class predicate of the second, or to the dissimilar-class predicate of the third. (ii) Waste: excluding those extremes one by one (“neither the subject of the similar class nor of the dissimilar class, but the matter under dispute”) costs time for no gain; the Auto-commentary answers that when a matter can be put concisely there is no need to circle around it. (iii) Metre: the three modes are stated in one compact phrase — “subject-property; presence; absence in the similar class” — whereas substituting the full true name, “the predicate of the subject one wishes to know,” destroys the word-combination. The Auto-commentary’s image: what can be cut open with a fingernail needs no axe.

Dispelling the view that it is identical with not naming it

[Objection:] “Subject-property” and “subject” are entirely equivalent. [Reply:] In common renown there is a difference.

[Indian masters headed by Śrī Yāpa] They concede the convenience of wording but deny any difference in meaning: “subject-property” covers the subject, the probandum-predicate, and their totality; and “subject” covers exactly the same three. If the scope is identical, renaming achieves nothing.

Reply: the two differ in what is universally known — in custom, not in scope. This is elaborated in the summarising verse.

The name “subject-property,” connected with the predicate, Is the matter under dispute, like “the one with a hand”; The name “subject,” connected with the predicate, Cannot be determined, like “the one with a head.”

Two similes. In Indian poetics “the one with a hand” in principle covers elephants, yaks, lions, humans — anything with a hand — yet by convention it denotes the elephant alone, and no one hearing it wonders whether a yak is meant. “The one with a head,” by contrast, covers every headed creature and settles on none, because usage has fixed no referent. Likewise, on hearing “subject-property” every logician knows the matter under dispute (the pillar) is meant; on hearing “subject” they do not. Since people fuse term-universal and object-universal by custom, the difference is real in use though not in bare scope.

The borrowed dharma does not accord with the actual

The predicate and the subject are not the probandum, Because they do not possess its defining characteristic.

Neither the bare pillar nor bare impermanence is the probandum, because neither carries its defining characteristic. The Pramāṇasamuccaya states five defining characteristics of the probandum and the Nyāyamukha one; summarised here it is “the object of measurement that is identified without having been [already] proven or refuted.” Three conditions follow: not already proven between the parties (else it is a jointly accepted dharma, not a topic); not already refuted by valid cognition (as “fire is not hot” is refuted by direct perception for both parties); and actually brought forward as the object to be measured (otherwise stones, rosaries and the rest are simply not at issue). The bare “pillar” provokes no debate, and the bare “impermanent” provokes none either; only “the pillar is impermanent” does. Hence the combination is the genuine subject-property and probandum, and calling either part alone by that name is licence.

Determining the contextual distinction

The object to be measured, the dharma one wishes to know — When [the reason] is established upon it, that is the subject-property.

The subject-property is the reason’s being established upon the matter under dispute — established either by direct perception (seeing smoke, in “there is fire on the mountain, because there is smoke”) or by inference (knowing through report or further reasoning that the pillar is produced). Where the reason fails on the matter under dispute — “the pillar is impermanent, because it is a yak” — the first mode is simply not established.

The similar and dissimilar class, the bases of forward and reverse pervasion — three parts: (1) refuting others’ position; (2) establishing one’s own position; (3) dispelling disputes.

Refuting others’ position — three parts: (1) stating the opponent’s view; (2) refuting that view; (3) dispelling [the charge] that the refutation is unreasonable.

Stating the opponent’s view

[Objection:] Possessing or not possessing the probandum-predicate Is the similar class and the dissimilar class.

[Early Tibetan epistemologists — Shawa Chösang and the translator Ngok, following the Indian master Śaṅkarānanda] Whatever possesses the probandum-predicate is the similar class; whatever lacks it is the dissimilar class. In “the pillar is impermanent, because it is produced,” whatever is impermanent is the similar class and whatever is permanent the dissimilar class. They hold the two classes must be directly contradictory, since otherwise correct reasoning has no footing.

Holding the two classes to be directly contradictory, And further fearing the appearance of a third class, Some masters, not knowing the object of valid cognition, Explain by dividing substance and isolate.

They call the substance-dharma one in nature or other in nature, And [posit] the isolate as depending on its own isolate.

Fearing that a “third class” (neither impermanent nor permanent, or both) will appear, they divide substance from isolate: when one establishes through the substance, the two classes are determinate on the substance while a third class may appear on the isolate; when one establishes through the isolate, the two classes are determinate on the isolate while a third class may appear on the substance. Their root error is not seeing that all refutation and proof proceed by fusing self-particular and object-universal into one.

Their position has three components: (i) in establishing anything the two classes are determinate and directly contradictory; (ii) substance and isolate are divided so that the third class always falls on the other side; (iii) establishing proceeds in three modes — identical substance (“the pillar is impermanent, because it is produced”), separate substance (“there is fire on the mountain, because there is smoke”), and own-isolate (“the produced that establishes impermanence is a correct reason, because it is complete in the three modes”). Sakya Paṇḍita, while acknowledging the early epistemologists’ standing in the history of the tradition, sets out these views and refutes them one by one.

Refuting that view — three parts: (1) determining two classes is unreasonable; (2) dividing substance and isolate to establish is unreasonable; (3) establishing in dependence on the own-isolate is unreasonable.

Determining two classes is unreasonable

However one divides substance and isolate, The two classes cannot be determined; The substance and isolate that the two involve Are seen to be established before the wise.

Test on the substance with “there is fire on the mountain, because there is a grey-white substance.” Is the grey-white substance the similar class? Then grey-white gas, cloud, mist and cloth would all have fire. The dissimilar class? Then grey-white smoke would not be fire’s effect. Both parts at once? Then a third class has appeared on the substance, demolishing their own thesis.

Test on the isolate with “the produced is the correct reason establishing impermanence, because it is complete in the three modes.” Is “complete in the three modes” the similar class? Then every three-moded reason would establish impermanence, though thousands establish other things. The dissimilar class? Then the reasoning itself becomes a pseudo-reason. Both? Then a third class exists on the isolate, again against their own claim.

Dividing substance and isolate to establish is unreasonable — four parts: (1) if examined, it is unreasonable; (2) it is over-extended; (3) it contradicts [their own] position; (4) it is not the Master’s intent.

If examined, it is unreasonable

If one performs refutation and proof upon the external object, The object of valid cognition cannot be known; If one performs refutation and proof merely before the mind, Existence and non-existence become indeterminate.

Refutation and proof must fuse self-particular and object-universal. Conducted on the bare self-particular, nothing could be refuted or proven at all: the pillar is as it is, and no one — the Buddha included — can make an impermanent thing permanent; as the sūtras say, whether or not a Tathāgata arises, the nature of dharmas abides. Conducted before a bare consciousness with no link to the object, the mind may declare anything at will — permanent, impermanent, empty, non-empty, both, neither. Only the fusion permits real ascertainment, as when others come to know “the pillar is impermanent” or “Śākyamuni Buddha is a valid being.” Since the isolate is posited in the mind and the substance on the self-particular, separating them entirely leaves no ground for refutation or proof at all.

It is over-extended

Not accepting the object-to-be-known as the subject-property, The first of the bases comes to be absent; And if one does not accept the first basis, The defining characteristic of the subject-property is truly hard to posit.

On their account the pillar’s impermanence falls wholly within the similar class, so the matter under dispute is swallowed up and no separate subject-property remains — hence its defining characteristic cannot be posited.

Were the first basis determined as [one of] the two classes, Then for the remaining two bases To determine the presence or absence of forward and reverse pervasion Would be exceedingly hard even to utter.

With the subject-property gone, forward pervasion (presence in the similar class) and reverse pervasion (absence in the dissimilar class) also lose their footing. In a correct reasoning the matter under dispute would become the similar class, and in a contradictory reasoning the dissimilar class — so the subject-property never exists at all.

If the subject-property and the two classes are one entity, Distinguished only by isolate, All reasons fall into either the correct or the contradictory; How then could an uncertain reason be possible?

If subject-property, similar class and dissimilar class are one entity distinguished only by isolate, every reasoning whose matter under dispute is similar-class becomes a correct reason and every one whose matter is dissimilar-class becomes contradictory — leaving no room for the uncertain (inconclusive) reason. But pseudo-reasons are three: contradictory, uncertain, unestablished. Their view reduces them to two.

It contradicts [their own] position

If one accepts that the knowable is determinately two classes, It truly contradicts [there being] three bases; And if one accepts the bases too are [only] two, Then the reasons depending on them become [only] two modes.

Bases and modes correspond: the subject-property depends on the matter under dispute, forward pervasion on the similar class, reverse pervasion on the dissimilar class. Two bases cannot support three modes. Further, without the subject-property the unestablished reason — defined as the reason’s not being established on the matter under dispute — could never arise, and the three pseudo-reasons emphasised throughout Dharmakīrti’s treatises would shrink to two. The opponent dare not accept this.

It is not the Master’s intent

Determining the two — similar and dissimilar class — This the Master does not accept; If, following substance and isolate, One divides each separately, convention is destroyed.

Dharmakīrti states in the Pramāṇaviniścaya, “The name of the [first] mode is the subject-property” — so the subject-property is neither similar nor dissimilar class, and a subject-property besides the two classes must exist. The opponent’s tenet is his own thought grafted onto other masters’ words.

Second, dividing substance and isolate entirely destroys all worldly convention. As shown when treating expresser and expressed: in exposition self-particular and universal may be divided; in entry or application they must be fused. If refutation and proof were conducted on the vase’s self-particular, no one could speak it or fetch it — the moment it was spoken it would have perished and no longer be that self-particular; and if conducted on the bare isolate, an image detached from the thing, nothing said would touch the thing at all. With conventions destroyed, communication and all adoption and rejection collapse.

Establishing in dependence on the own-isolate is unreasonable

Establishing by means of the own-isolate — When examined, this principle has no substance.

Establishing admits only two modes: reason and probandum non-dual and undifferentiated (the nature-reason, “the pillar is impermanent, because it is produced”), or reason and probandum entirely separate entities (the effect-reason, “there is fire on the mountain, because there is smoke”). A third mode — neither identical nor separate, on a bare isolate — does not exist. This parallels the answer given in the eighth chapter to the infinite regress of defining characteristics: the defining characteristic, with its branches and leaves, is established with the tree of the root thesis as non-dual and undifferentiated, not as a self-particular divided from a later isolate. So the opponent’s own example, establishing impermanence by the isolate of “produced,” is in fact established only by fusion with the pillar’s self-particular.

(The stock example “the pillar is impermanent, because it is produced” is a template, not the whole of the subject — much as Newton’s laws are applied constantly without exhausting physics. Master the principle and any example serves.)

Dispelling [the charge] that the refutation is unreasonable

[Objection:] If the two classes are not directly contradictory, All refutation and proof collapse. [Reply:] Refutation and proof are not done by the two classes; They are determined by valid cognition’s contradiction and connection. By the force of the connection between reason and probandum, Existence and non-existence are determined through positive and negative concomitance.

[Early epistemologists] If the two classes are not directly contradictory, then refuting the negandum no longer establishes the probandum and establishing the probandum no longer refutes the negandum, so all establishment and refutation are destroyed.

Reply: the similar and dissimilar class are only the bases depended upon — one looks to see whether the reasoning holds of them. Genuine refutation and proof are determined by valid cognition’s contradiction and connection. Knowing fire and water to be contradictory, one knows that where strong fire is, water is not (as where strong sunlight is, darkness is not); knowing smoke and fire to be joined as “this arising from that,” seen before by direct perception or established by inference, one knows that where smoke is, fire is. In “the pillar is impermanent, because it is produced,” “produced” and “impermanent” stand in positive concomitance, and reversing “impermanent” reverses “produced” — negative concomitance. The existence and perishing of things is settled by these relations, not by the similar-and-dissimilar-class relation the early epistemologists rely on.

Establishing one’s own position — three parts: (1) analysing the similar and dissimilar class; (2) recognising the object of the reason’s refutation and proof; (3) the detailed classification of refutation and proof through exclusion.

Analysing the similar and dissimilar class — three parts: (1) the definition; (2) recognising the similar object; (3) dispelling disputes.

The definition

The subject, by the general predicate to be proven, Being similar or dissimilar — these are the two classes.

The subject-property’s being similar to an object by way of the probandum-predicate is the similar class; being dissimilar by that predicate is the dissimilar class. In “the pillar is impermanent, because it is produced, like the vase,” pillar and vase are similar by way of impermanence; either may be taken as the one that is similar or as the object similar to. Just as I, Devadatta, am similar to Yajñadatta — and one must be able to say in what respect.

Forward pervasion is the reason’s holding of the probandum: whatever is produced is impermanent. Reverse pervasion must be inferred backwards from the thesis, not forwards from the reason: “if it is not the probandum, it is definitely not the reason” — if “impermanent” reverses into “permanent,” “produced” is negated. Saying instead “if not produced, then not impermanent” causes trouble, since Dharmakīrti’s other treatises note that this holds for some reasonings and not others.

Recognising the similar object

The subject-property as the similar object Is the borrowed exclusion, not either of the two.

The similar object is the subject-property — but the borrowed-name subject-property, and by way of exclusion, not by way of the self-particular. It cannot be the true-name subject-property (the combination of the matter to be known and the thesis), for nothing can be similar to a combined dharma; nor can the pillar’s self-particular resemble the vase’s, since both change moment by moment and share no point whatever (as established in the third chapter).

Dispelling disputes

Here there is no fault such as others allege; The intent of the treatises too is just this.

[Early epistemologists] If the similar class is what resembles the matter under dispute, then since nothing resembles itself, the matter under dispute must always fall on the dissimilar side, and every genuine reasoning is ruined.

Reply: granted the matter under dispute is not similar to itself — but it is similar to the example. The pillar is similar to the vase by way of impermanence, so no such fault arises. This is the ultimate view of Dignāga and Dharmakīrti.

Recognising the object of the reason’s refutation and proof

“Substance” means the entity; Apart from exclusion there is no isolate; Therefore, mistaking substance and isolate as one, Refutation and proof are carried out.

“Substance” is whatever has a self-particular — pillar, vase — and not, as the non-Buddhists say (noted in the Auto-commentary), only an assemblage of atoms. The “isolate” is the image appearing before conceptual thought, that is, exclusion; apart from exclusion there is no other isolate. As shown in chapters four and five, the two may be divided in exposition but never in application: without the fusion one could not even say “I am going to fetch water,” since speaker and hearer have different concepts of isolate and substance. Refutation (“before me there is no pillar, for if one existed I could see it”) and proof (“before me there is a pillar, because I see it directly”) alike proceed by mistaking substance and isolate as one. This is why the fourth chapter, on exclusion, is the baseline for the whole treatise.

The detailed classification of refutation and proof through exclusion

Relying on entity and non-entity, By three names-and-meanings, two dharmas are established.

Three names-and-meanings arise: relying on entity (the self-particulars of pillar, vase, ox), on non-entity (the rabbit’s horn, the barren woman’s child, space — no self-particular essence, yet a concept before thought), and on both together (the knowable, which pervades all dharmas). The Pramāṇavārttika likewise says that relying on existence and non-existence three general dharmas appear, the third-class general dharma being what presents itself before conceptual thought. These three exclusions may be divided into three names on the subject side and three meanings on the object side. By them both reasoned-meaning and convention are proven: “the pillar is impermanent, because it is produced” proves reasoned-meaning; “this animal is an ox, because it has a hump and dewlap” proves convention. All such refutation and proof occurs before the exclusion-cognition.

Dispelling disputes — two parts: (1) that examining the two sides is unreasonable; (2) that examining reason and probandum is unreasonable.

Dispelling the dispute that examining the two sides is unreasonable

[Objection:] The grey-white thing in the smoke-reason, And the general knowable of the three-mode reason — When one examines under which class, similar or dissimilar, they fall, Your positing of valid cognition is destroyed.

[Early epistemologists] Take “there is fire on the mountain, because there is smoke” (with its grey-white substance) and “the produced is the correct reason, because it is complete in the three modes” (with its general isolate of the knowable). Each must fall under one class or the other; but examined, neither can be determined — so the later school’s positing of valid cognition collapses.

[Reply:] The two exclusions — smoke and the three modes — Are closely bound up with the grey-white [thing] and the knowable By way of the self-particular; The isolate links the two, so it is reasonable.

Reply: the grey-white thing and the general knowable are not independent dharmas; they are bound to the self-particular (the grey-white is entirely tied to the self-particular of the mountain’s smoke), so no free-floating third class arises on the substance. On the isolate a third class is admitted: the grey-white in consciousness may be fire-born smoke or mere mist, and the knowable in consciousness may include both impermanent dharmas and dharmas like space. So one must ask the opponent which he means — substance or isolate — and answer accordingly. Their error is failing to distinguish exposition (where the two may be divided) from application (where they must be fused).

Dispelling the dispute that examining reason and probandum is unreasonable — two parts: (1) stating the debate and that others’ answers are unreasonable; (2) our own school’s answer.

Stating the debate and that others’ answers are unreasonable

On the two classes — of exclusion, refutation, predicate, and reason — The four debates: others’ answers are mistaken.

Four debates arise, all turning on dividing substance and isolate: on exclusion (whether the reason lies outside the probandum or within it), on refutation (whether the reason refutes the probandum), on the predicate side (“emptiness is a non-implicative negation, because it has no probandum”), and on the reason side (“the produced is the reason establishing impermanence, because it is complete in the three modes”). Other masters answer all four, and Sakya Paṇḍita rejects all four answers.

The first debate: in “emptiness is a non-implicative negation, because after refuting the negandum it draws out no other dharma,” if the reason lies outside the probandum-subject the subject-property fails and forward pervasion is lacking; if it lies outside the negandum, the parallel argument fails likewise. Some masters answer by splitting essence from convention — the reason holds of the essence of emptiness, while the convention of emptiness is an implicative negation (the distinction between implicative and non-implicative negation as treated in the Resolving Doubts: the Lamp Establishing Definitive Meaning). Sakya Paṇḍita refuses this: convention and essence of emptiness are non-dual and undifferentiated, and if they were separate entities the nature-reason’s forward pervasion would not be established.

Our own school’s answer

The convention of a non-implicative negation is itself non-implicative; The general isolate, being a third class, is therefore reasonable.

The convention of a non-implicative negation is itself a non-implicative negation, since the convention of every refuted dharma is posited on the very essence of that negation. And a third class before the exclusion-cognition is entirely admissible: in “there is fire on the mountain, because there is smoke,” the grey thing may be similar-class (a steel furnace’s smoke, though not this mountain’s) or dissimilar-class (gas or mist), so a third class stands before consciousness. The opponent’s worry about a third class in the emptiness argument is therefore superfluous.

The reason for positing these as bases

Whatever is posited as a reason All depends upon the three bases; Though in reality any one of the bases may be incomplete, Before cognition the three bases are accepted.

Every reason, correct or pseudo, rests on three bases: subject-property, similar class, dissimilar class. Correct reasons (effect, nature, negation) are established on them; pseudo-reasons (contradictory, uncertain, unestablished) are diagnosed by which of the three fails.

A basis may be missing in reality yet accepted before cognition. The Pramāṇavārttika’s example: “all dharmas are without self-nature, because they are free of one and many.” Its similar-class example exists in fact (the water-moon, a reflection), and the reason holds of all dharmas; but no dissimilar-class example — a really existent dharma not free of one and many — can be found anywhere in the world. The reasoning is nonetheless correct, because the dissimilar-class example is established before the exclusion-cognition. The Madhyamakālaṃkāra discusses the same principle when treating reverse pervasion.

For refutation and proof carried out through exclusion, The debate-faults about dividing substance and isolate into two classes Do no harm, Because [we] are versed in the object of valid cognition.

Refutation and proof through exclusion is the later school’s method; the division of substance and isolate is the early epistemologists’. Their debate-faults do no damage, because the later school has mastered the object of valid cognition, whereas their divisions generate contradictions precisely because they have not.

Classifying the Reason That Depends Upon Them

Three parts: (1) the correct reason; (2) the pseudo-reason; (3) the fixed number of these reasons. This section corresponds to the science of logical reasons (因类学, rtags rigs).

The correct reason — three parts: (1) the definition; (2) the instance; (3) the classification of the name.

The definition — two parts: (1) identifying the definition; (2) the reason it serves as the definition.

Identifying the definition — three parts: (1) refuting others’ position; (2) establishing one’s own position; (3) dispelling disputes.

Refuting others’ position

From one mode up to six modes — Others’ approximate positings are mistaken.

Rival counts run from one mode to six, and all fail on a single principle: if the three modes are complete the extras are superfluous; if incomplete, no number of extras helps.

[Certain non-Buddhists] One mode only — “the others being unreasonable.”

[Early Tibetan epistemologists] Two modes — the subject-property plus either the forward or the reverse pervasion.

[Vedic school] Four modes — the “antecedent dharma” (the reason existing on the dharma-to-be-known), the “remaining dharma” (existing in the similar class), “general non-observation” (not existing in the dissimilar class), and non-contradiction (unharmed by direct perception and scriptural valid cognition). Their first three fall short of ours: they require only that the reason exist, not that it be established, and only that it not be seen in the dissimilar class, whereas forward and reverse pervasion require a relation of determination. Non-contradiction is redundant, since what is proven by three complete modes can never be harmed by direct perception or scripture.

[Certain Vedic masters] Five modes — the three plus “each being determinate” and “being un-harmed.”

[Master Īśvarasena] Six modes — the three plus “an un-harmed subject,” “the dharma one wishes to state,” and “being determined by the mind.” Īśvarasena was Dharmakīrti’s teacher and had heard the Pramāṇasamuccaya from Dignāga’s tradition; having heard it three times from him, Dharmakīrti judged that his teacher had not grasped Dignāga’s ultimate intent, and Īśvarasena himself gave permission for Dharmakīrti to compose a treatise taking him as the opponent — as Vasubandhu takes Saṃghabhadra in the Abhidharmakośa.

Establishing one’s own position

The subject-property established and the connection determined — That is the flawless definition of a reason.

Two conditions: the reason must be established on the subject (“produced” on sound), and the connection between reason and thesis must be ascertained — either identity of nature or causal connection, expressed as positive and negative concomitance (if produced, certainly impermanent; if not impermanent, not produced). Without them one has only the worldly manner of arguing: “this is unreasonable, because I think it unreasonable, because it might be unreasonable.”

Dispelling disputes

The [alleged faults of being merely] two modes, and of the forward pervasion Drawing out the reverse pervasion — these two faults are absent.

[Early epistemologists] First charge: you require three modes yet define the reason by two — subject-property established, connection ascertained — so your own school is a two-mode school.

Reply: no. The ascertainment of the connection already includes both pervasions, since the connection is established only by two valid cognitions, one establishing on the similar class and one refuting on the dissimilar class; conversely, given the connection, both pervasions follow.

[Early epistemologists] Second charge: a general forward pervasion is deluded, and a special one already subsumes the reverse pervasion — so the two pervasions become one entity, and you are back at two modes.

Reply: the reverse pervasion may indeed be summarised within the forward pervasion by an indirect link, but they are not one entity. Analysed on the isolate, the positive angle establishes the forward pervasion and the reversed angle the reverse pervasion. Only if their essence were absolutely one would the fault follow.

The reason it serves as the definition

What is [merely] determinable is not a reason, because of doubt; What is determined is a reason, because it makes [the probandum] known. Here there is no fault such as others allege.

Is a reason a dharma whose three modes are already determined, or one that can be determined later? Only the already-determined. What is merely determinable still carries doubt now, and doubt cannot make the probandum known; the determined reason has produced certainty and dispelled doubt.

[Objection] Before the three modes are determined it is not a reason; and once determined, the mind produced is a mind produced by a non-reason — so it is not a correct reason either.

Reply: what we accept as the mind grasping the reason is not a mind grasping the bare qualifying dharma “complete in the three modes,” but a mind grasping the qualified instance — the mental state non-dual with the smoke on the mountain, or with “produced.” Merely seeing smoke by direct perception is not yet grasping the reason; grasping occurs when the connection of smoke and fire is present to mind and the inference arises of itself: “there must be fire there, for I have seen this in a stove.” Likewise seeing a worker make a vase is not yet grasping the reason, but through his producing it, “this vase is impermanent” is at once induced by prior familiarity.

The instance — two parts: (1) refuting the untenable part of others’ position; (2) positing the dispute-free own position.

Refuting the untenable part of others’ position

Some accept that [the reason] qualified by the matter under dispute Is the instance of the reason.

[Some masters] The instance of the reason is the reason qualified by the matter under dispute: “the smoke on the mountain”; “the produced of sound.”

If it is a non-affirming negation relative to the disputed matter, then we have no dispute; But [if it is] an exclusionary negation, there is no forward pervasion, And the probandum-predicate would fare the same.

Ask whether the qualification is a non-affirming negation or an exclusionary “only-this” negation. If non-affirming — there is smoke on the mountain, which merely excludes “the mountain has no smoke” — we agree and there is no debate. If exclusionary — solely the smoke on the mountain — then no similar-class example can be found: “there is fire on the mountain, because there is the smoke on the mountain — like what?” With general smoke as the reason, the stove’s smoke serves as example and forward pervasion stands. The same collapse afflicts “sound is impermanent, because it is the produced of sound,” which admits no example, whereas “because it is produced” admits pot and pillar.

The probandum fares identically: “the mountain has the fire of the mountain, because there is the smoke of the mountain” admits no example, since no other fire can serve; and if “impermanent” is an exclusionary negation of sound, no other impermanence can be found.

Positing the dispute-free own position

The instance of the reason is just the exclusion; Here there is no fault such as others [allege].

The instance of the reason is exclusion — the general isolate of smoke, of “produced,” cited without distinguishing time, place, or form, and grasped as one with the posited reason. The objection that “the universal of the reason cannot exist, because it is an isolate” fails: the universal does not exist by way of substance, but by way of isolate or exclusion it is entirely established. Both the definition and the instance of the reason are inseparable from exclusion.

The classification of the name — three parts: (1) the principle of how to distinguish; (2) determining the nature of the divisions; (3) refuting wrong notions about their connection.

Positing other classificatory approaches

This reason, possessing this definition, Divides by category into many kinds.

By the angle of positing the reason: reasons positing a non-implicative and an implicative negation. By the probandum-predicate: reasons establishing a non-implicative and an implicative negation. By connection: reasons of identical nature and of causal connection. By establishing: reasons proving reasoned-meaning and proving convention. By the prover: the reason of common renown (what the world acknowledges, admitted provisionally, treated at length in the last chapter of the Pramāṇavārttika and later in this treatise) and the reason of factually grounded reasoning.

Identifying what is meant here

The qualified Master Dharmakīrti, By way of the formal argument, states a fixed three kinds.

In the Pramāṇavārttika and Pramāṇaviniścaya Dharmakīrti fixes three. All reasons on the refuting side reduce to the non-observation reason (the contradiction-observation reasons included). All reasons on the establishing side require two kinds of example and so divide into nature-reason and effect-reason.

Determining the nature of the divisions — three parts: (1) the non-observation reason; (2) the nature-reason; (3) the effect-reason.

The non-observation reason — three parts: (1) the definition; (2) the classification; (3) the way of determining its connection with the negandum.

The definition

It possesses the three modes in the refutation of the negandum.

The nature- and effect-reasons possess the three modes on the establishing side; the non-observation reason possesses them on the refuting side.

The classification — two parts: (1) the non-appearing non-observation reason, which refutes existence but cannot establish definite non-existence; (2) the contradiction-observation reason, which establishes definite non-existence.

Non-observation of the essence

Non-observation of the essence has four kinds.

The principle: if the dharma existed it would be seen; it is not seen, therefore it does not exist. This applies only where seeing is possible. “There can be no demon before me, because I do not see it” fails, since ordinary beings cannot see demons; so does the materialist’s “past and future lives do not exist, because I do not perceive them” — not to speak of future lives, even tomorrow’s appearances are imperceptible to an ordinary being. To deny a thing one must have grounds that convince both oneself and others, not “I feel it does not exist” or “it probably does not exist.”

1. Non-observation of the nature: “before me a vase does not exist, because I do not see it directly” — my eye-faculty being unimpaired, I would see it. “In this hall there is no other person, because if one existed I could see him” holds for a person but not for an ant.
2. Non-observation of the cause: the effect is denied because the cause is not obtained — at night, “on this lake there is certainly no smoke, because there is no fire.” This works at night only: by day distant fire goes undetected, which is why bandits hiding in gullies cooked by day with barkless (smokeless) firewood and never lit fires after dark.
3. Non-observation of the pervader: “on this bare flagstone there can be no cypress, because there is no tree of any kind.” The pervader is the general category (tree), the pervaded its particulars (sandalwood, aloeswood, pine, cypress); negating the pervader negates all the pervaded.
4. Non-observation of the effect: “under open sky with no obstruction there can be no fire, because there is no smoke.” Obstruction would spoil the inference; from a mountain-top the absence of rising smoke settles the absence of its direct cause.

These four are undisputed. (Whether a given reason establishes or refutes was analysed when teaching the Madhyamakālaṃkāra.)

The contradiction-observation reason — two parts: (1) the incompatible-non-coexistence contradiction-observation; (2) the mutual-exclusion contradiction-observation.

The contradiction-observation [reason] divides into two kinds.

Incompatible-non-coexistence: refuting the untenable part of others’ position

[Opponent:] The four kinds [of the fire whose] heat-touch destroys the cold-touch — Hence sixteen formal arguments are stated. [Reply:] The unobstructed capacity is not fire; Having no continuum, it is not a destroyer.

First distinguish destroyer from destroyed. The destroyed are four cold-dharmas: cold-touch itself; its cause (the unobstructed capacity — cold not yet arisen but able to arise in the next instant); its effect (hairs standing on end); and its pervaded (frost, snow, the cold of winter — general cold-touch being the pervader).

[Early epistemologist master Cha-naba] Four destroyers each remove the four destroyed, giving sixteen: (i) burning fire — the four contradiction-nature-observation reasons (“before me there cannot be cold-touch / the cause of cold / hairs standing on end / frost, because there is fire”); (ii) sandalwood-fire — the four contradiction-pervaded-observation reasons; (iii) rising smoke — the four contradiction-effect-observation reasons; (iv) the unobstructed capacity of fire — the four contradiction-cause-observation reasons.

Sakya Paṇḍita accepts the first twelve and rejects the last four. The unobstructed capacity is only the last instant of the firewood before fire arises; it has no continuum and is not the nature of fire, so it has no power to destroy cold-touch or any of its three. A genuine reason of this type must be a destroyer.

Positing the tenable own position

Though the treatises state many positings, The destroyers are just the first twelve.

Following the Indian master Cha-tari-pa, and using the teaching-notes of Jamyang Loter Wangpo: four destroyed against three destroyers (fire on the side of nature, smoke on the side of effect, sandalwood-fire on the side of the pervaded) gives twelve. Other treatises give ten, nine, or eight.

The four contradiction-nature-observations take fire ascertained-to-measure as reason (not a single spark). Nature-contradiction-nature-observation: “where there is fire ascertained-to-measure there is no cold-touch, because there is fire” — fire and cold-touch being contradictory in substance-nature. Cause-contradiction-nature-observation: “there is no hairs-standing-on-end, because there is fire” — named from the cause, since the effect of cold is not itself directly contradictory with fire, whereas its cause, cold-touch, is. Effect-contradiction-nature-observation: “there is no cause of cold-touch, because there is fire” — named from the effect of the unobstructed capacity, namely cold. Pervader-contradiction-nature-observation: “there can be no frost-touch, because there is fire” — frost being a pervaded of cold, and cold the pervader that contradicts fire. (Mipham Rinpoche’s Science of Logical Reasons names some of these differently.)

The four contradiction-effect-observations take billowing smoke as reason, negating in turn the nature, cause, effect and pervaded of cold. The smoke must be smoke just rising from fire; the Pramāṇavārttika accordingly qualifies smoke as produced from fire and newly risen — lingering smoke from an extinguished fire will not serve, whereas a factory chimney’s rising smoke settles the matter at once. Smoke is used where the fire itself is not seen.

The four contradiction-pervaded-observations take sandalwood-fire as reason. Mipham Rinpoche’s Gateway for the Wise says these may be subsumed under the nature-observations and need not be listed separately.

[Objection] Since sandalwood-fire is only a qualifier of fire, these four are redundant.

Reply: before conceptual thought sandalwood-fire and other fire have different essences, and particular dharmas must be usable as reasons — we do say “there is no darkness here, because there is a butter-lamp,” or “because there is sunlight,” not merely “because there is light.” Forbidding particulars destroys worldly convention.

[Further objection] Then sandalwood-smoke should serve equally.

Reply: not so. If one has not already recognised that the fire is sandalwood, one cannot tell the smoke is sandalwood-smoke; and if one has, singling out the smoke is superfluous. (Some argued in debate that Indian white sandalwood’s scent identifies its smoke; but then one must infer from scent to sandalwood-smoke, from that to fire, and from fire to the absence of cold — needlessly roundabout.)

The mutual-exclusion contradiction-observation

For a true mutually-exclusive contradiction, There is no formal argument of reason and thesis; Yet by its qualified dharma There is a pervader-contradiction-observation reason.

As shown in the seventh chapter, mutual exclusion may be direct (permanent/impermanent, entity/non-entity) or indirect. Where the contradiction is direct, affirming one side has already negated the other, so no formal argument is possible: “the pillar is non-permanent, because it is impermanent” is pointless, since a reason exists to make a thesis known and here nothing remains unknown.

By a qualified dharma, however, a pervader-contradiction-observation reason is available. The direct contradictory of impermanent is permanent, and the pervader of permanent is the non-produced; so “produced” stands in an indirect relation to “permanent,” and “the pillar has no permanent dharma, because it is produced” negates the permanent by contradicting its pervader.

The contradiction [established] by valid cognition’s invalidation Also belongs to the pervader-contradiction-observation.

Two cases fold in here. The invalidating reason: non-Buddhists allow that dharmas are generally impermanent yet claim some particular — Maheśvara — is permanent, as most crows are black yet a few are white; the invalidating reason shows no permanent unchanging dharma is possible anywhere. The non-dependence reason: to one hesitating whether a vase or tree must perish, one argues “it is not the case that it perishes or not uncertainly, because it is of perishing nature and depends on no [other] dharma” — perishing requiring no destroying cause. Both are subsumed under the pervader-contradiction-observation and need no separate heading.

The way of determining the reason’s connection with the negandum — two parts: (1) stating others’ wrong view; (2) positing the flawless own view.

The connection at the level of real entities has been refuted; [The claim] that [the negandum] arises from the assemblage of incompatible-non-coexistence is mistaken.

Chapters six and seven established that neither identical-nature nor causal connection exists on the level of a thing’s essence.

[Certain early epistemologists] Where fire’s four dharmas exist, the four contradictory water-dharmas cannot; and conversely, where fire’s four are absent, water’s four arise.

Reply: the converse does not hold. Fire’s absence does not entail cold-touch water — a state neither cold nor hot may obtain, and where strong fire is absent a merely weak water incapable of answering to fire may lie nearby. Otherwise one could argue “there must be water here, because there is no fire,” or “fire here, because there is no water.”

When the connection’s object reverses, the connected dharma reverses; Therefore the reverse-connection is established before the mind.

No essential connection exists, but a reverse-connection is established before the exclusion-cognition: if the connection-object reverses or is absent, the connected dharma follows. If “impermanent” turns into “permanent,” “produced” becomes non-produced; if fire is absent, the smoke that arises from it is absent. Causal connection has its reverse in non-observation of the cause and of the effect; identical-nature connection has its reverse in non-observation of the nature, non-observation of the pervader, and mutual-exclusion contradiction (“there can be no pine here, because there is no tree”). The contradiction-observation reason, however, needs no reverse-connection at all, since establishing one dharma directly destroys the other — where fire is, cold-touch is not.

The nature-reason — two parts: (1) the actual statement; (2) dispelling faults.

Proving that “it is,” possessing the three modes, With essence non-distinct — that is the nature-reason; Its counterpart cases are subsumed within it.

The nature-reason proves that something is: reason and probandum are non-dual and undifferentiated in essence, as in “the pillar is impermanent, because it is produced,” where the pillar is both produced and impermanent without separateness. (This reasoning was met in the Madhyamakālaṃkāra and the Mūlamadhyamakakārikā without being named as such.)

Two kinds. The purified-of-qualifier nature-reason: “the pillar is impermanent, because it exists” — the Sanskrit for “exists,” bhāva, carries the sense of an entity, so the argument may equally run “because it is an entity”; it is purified of qualifier because no ground beyond natural law can be given for its existing. The qualifier-dependent nature-reason: “the pillar is impermanent, because it is produced” — “produced” implies a maker (carpenter, potter), so it depends on further analysis.

Two counterpart cases are subsumed here though non-Buddhists deny it. (i) The inference from complete causes and conditions to the future effect — “the sprout can be produced in this excellent farmland, because it is excellent farmland.” It looks like an effect-reason, but the field’s capacity to produce the sprout and the field are non-dual, so it is a nature-reason. (ii) “Smoke is impermanent, because it possesses the smoke-cause, the unobstructed capacity” — the momentariness of that capacity and the impermanence of the smoke are established as non-dual.

Dispelling faults

Being one [in nature], it is not tainted by the fault of [being a] separate substance.

[Some masters] The unobstructed capacity is the former instant of smoke and the rising smoke is a later one — separate entities — so this cannot be a nature-reason.

Reply: the rising smoke is produced from that former instant, and what is at issue is the impermanence posited in combination with smoke and its unobstructed capacity — these three are not separate substances. (Taken as the capacity of a different continuum they would indeed be separate, as the Auto-commentary also notes, but that is not what is being established.)

The effect-reason — two parts: (1) the actual statement; (2) dispelling disputes.

Proving that “it has,” possessing the three modes, With causal connection — that is the effect-reason; Its counterpart cases are subsumed within it.

The effect-reason proves that something has, by causal connection rather than identity: “there is fire on the mountain, because there is smoke” — mountain and fire are separate entities, and the smoke genuinely exists apart. “The pillar has impermanence, because it is produced” cannot be an effect-reason, since pillar and impermanence are one entity, not two.

Three kinds. (i) The effect-reason proving by the isolate of the cause — the standard “there is fire on the mountain, because there is smoke” (with smoke qualified as newly risen and unmixed with cloud or mist), posited to refute the non-Buddhist claim of causeless arising. (ii) The effect-reason proving the prior cause — “because smoke curls in the sky, there must previously have been fire,” even if no fire remains; posited to refute the Cārvāka denial that the present consciousness has a prior consciousness. (iii) The effect-reason inferring the effect from the cause-dharma — “at night there is smoke on the mountain, because there is a forest fire”; or inferring form from the sweet taste of sugar in the mouth, since form and taste belong to one group of the four great elements. This third is posited against the Maheśvara school’s claim that things are created by Maheśvara without assembling causes and conditions. Everyday cases abound: the frog’s call before rain, the night-lotus opening in moonlight, dense clouds or ants shifting house presaging rain; and the Daśabhūmika-sūtra’s inference of water from water-fowl.

Dispelling disputes

Because separate entities are posited, there is no fault.

[Objection] When the word “pot” produces the universal of a pot in the mind, or a universal produces the word, you treat this as inferring an effect. But a universal is an unreal entity and language a real one, and a causal relation between the two is unreasonable.

Reply: on the level of actual fact real and unreal entity are neither one nor separate, so no causal relation obtains there. But in dependence on the excluding cognition it does. The wish-to-speak (the thought of uttering “pot”) is the substantial cause; the exertion of tongue, palate and other points of articulation is the cooperative condition; the sound “pot” issues. Before the excluding cognition the universal of a pot and the word “pot” are wholly separate entities, so nothing is unreasonable in positing them as cause and effect there.

Refuting mistaken notions about these connections — two parts: (1) stating the opponent’s view; (2) refuting that view.

Some say connection divides into nine kinds, And the connections able to make known are four.

[Early epistemologist masters such as Tsang Nagpa] Our own school admits only two connections, causal and identical-nature; they admit nine, of which four can serve as proving-reasons.

They first separate connection-object from connection-dharma (fire from smoke; seed from sprout). Each side yields three: the true member, its qualifier-dharma (what has one isolate), and its qualified-dharma (what has two isolates — as “the staff-holder” has the isolates of person and staff, whereas “staff” has one). On fire’s side: fire’s essence; fire’s glowing-red; fire’s burning of things. On smoke’s side: smoke itself; the blue colour of the smoke; the smelling of the smoke’s odour. Pairing each of the three connection-dharmas with each of the three connection-objects gives nine.

Of these, four are held to make known: (i) true smoke with true fire; (ii) true smoke with fire’s qualified-dharma, its burning — but not with fire’s glowing-red, since pervasion on the qualified side does not prove pervasion on the qualifier side; (iii) and (iv) true fire with true smoke and with smoke’s qualifier-dharma, the blue colour — but not with smoke’s qualified-dharma, the smelling of its odour, since where the qualifier exists the qualified must follow.

The positing here is self-contradictory, And does not accord with the intent of the treatises.

Refutation on three counts. First, they themselves accept that fire and smoke stand in causal connection; so denying any relation between smoke and fire’s glowing-red is impossible — otherwise one could not infer glowing-red fire from billowing smoke. Second, denying a relation between fire and the smelling of smoke’s odour is likewise impossible: fire produces smoke, smoke gives off odour, and by that odour one knows its source; deny it and “there is fire on the mountain, because there is smoke” itself fails. Third, inferring fire from smoke’s odour is an ordinary inference of the effect from the cause-dharma, since fire has both the function of producing smoke and that of producing odour. Above all, Dignāga and Dharmakīrti declare that all connections reduce to the causal and the identical-nature; the nine-fold scheme exceeds that scope and contradicts scripture and reasoning alike.

Summary verse:

By the difference of qualifier and qualified, the connection cannot make the probandum-dharma known; Only in dependence on identity of essence or on the other meaning is it known by the force of connection.

A qualifier–qualified connection cannot make a previously unknown probandum known. Only the identical-nature connection (known through the nature-reason) and the causal connection (known through the effect-reason) can. These two are sufficient and necessary: fewer leaves the account short, and all the connections admitted by Buddhists and non-Buddhists reduce to them. Tibetan grammar recognises a dozen or so “connections” (the genitive) — “my hat” and “my house” being the connection of self and what belongs to self, “my essence” the identical-nature connection, “my father” the causal — but by the epistemological view all fall under the two.

With this, the genuine reasons are complete: on the affirming side the nature-reason and effect-reason; on the refuting side the non-observation reason and the contradiction-observation reason. Against any inference whose three modes are incomplete, three answers suffice — “unestablished,” “inconclusive,” “contradictory.”

Fallacious reasons — two parts: (1) the definition; (2) the classification.

The definition

A reason in which the three modes are not established.

Any formal argument in which the three modes are incomplete is a fallacious reason. Three classes: unestablished, inconclusive, contradictory. (These are expounded in the Science of Reasons, here treated on the basis of Jamyang Loter Wangpo’s commentary.)

The unestablished reason — two parts: (1) the definition; (2) the classification.

A reason absent in the subject-property is the unestablished reason.

A reason which simply does not exist, or is not established, on the subject-property. Note that in answering “the reason adduced is not established,” one must be ready to say how — whether unestablished on the external object or before the mind — since the bare phrase conveys little to someone unpractised in three-mode reasoning.

The unestablished reason regarding the external object

The unestablished reason regarding the external object has six kinds.

1. The reason’s essence is not established: “this person has a self, because he possesses the nature of a self” — the permanent autonomous self of the non-Buddhists does not exist even conventionally.
2. The essence of the subject is not established: “ultimate sound is permanent, because it is non-produced” — in the ultimate no dharma has any defining feature, so the subject is no different from a barren woman’s son.
3. The connection of the reason with its locus is not established: “sound is impermanent, because it is seen by the eyes” — sound is heard, not seen. (“A pot is impermanent, because it is seen by the eyes” could be established from one angle.)
4. A part of the reason is not established on the subject: “the sense-cognition that displays a double moon is direct perception, because it is non-conceptual and non-mistaken” — non-conceptuality is established, non-mistakenness is not; since only the combination is the reason, the answer is “unestablished.”
5. A part of the subject is not established on the reason: “sound is impermanent, because it is produced by exertion” — true of the voice, false of the sound of water or of sounds issuing naturally from the four great elements. One must ask whether sound in general or a particular sound is meant.
6. Identity of isolate is not established, in three sub-cases: reason identical with subject (“sound is impermanent, because it is sound” — like “a person is impermanent, because he is a person”; Mipham Rinpoche classes this under the inconclusive reason, since no similar-class example can be found); subject identical with predicate (“sound is sound, because it is non-produced”); predicate identical with reason (“sound is impermanent, because it is impermanent”). A formal argument exists to prove what is not known; where reason and probandum are undifferentiated there is nothing left to prove.

The unestablished reason before the mind

The unestablished reason before the mind has three kinds.

1. The subject being in doubt: “the pot possessed by a ghost-spirit is impermanent, because it is produced,” put to someone without higher perception who cannot tell whose pot it is. The relation of “produced” and “impermanent” is sound; the qualifier spoils it. Likewise “the jewel-light before us exists, because it is seen by direct perception,” put to one who cannot tell jewel-light from lamplight.
2. The reason being in doubt: “on the mountain that has the grey-white thing there is fire, because there is smoke,” put to someone who cannot tell smoke from mist. Subject and thesis are accepted; the reason is doubted.
3. The connection of the reason being in doubt: a peacock’s cry issues from a mountain with three valleys, and one argues “the middle valley has a peacock, because the peacock’s cry issues.” Mipham Rinpoche’s Science of Reasons treats this as inconclusive, since the cry does not settle which valley. It is treated here as unestablished because the opponent has not determined the connection between reason and thesis.

The unestablished reason regarding the disputants

Mipham Rinpoche holds this need not be separated out, and Sakya Paṇḍita’s intention is the same — separating it is merely admissible. Debate proceeds through proponent, opponent, and witness. (The early epistemologists assign six faults to the proponent, six to the opponent, three to the witness and one to all three, sixteen in all; Sakya Paṇḍita rejects so many, and denies the witness any fault at all, since it is unreasonable for the judge to be sentenced. This is treated in the eleventh chapter.)

1. Unestablished with respect to the proponent: “consciousness and pleasure are all not mind, because they have arising and ceasing” — unestablished for the Sāṃkhya, who hold arising to be mere manifestation from the primal substance and ceasing mere concealment within it. (The Sāṃkhya treat pleasure, pain and the rest as insentient; this was clarified when studying the wisdom chapter of the Bodhicaryāvatāra.)
2. Unestablished with respect to the opponent: the Nirgranthas argue against Buddhists that “trees are sentient, because when stripped of bark they die.” If “death” means only the withering of a former continuum, we accept it; but genuine death is the cutting off of the life of a being possessing consciousness, and trees only wither. (Certain beings of the isolated hells take plant form to experience suffering through particular karma, as the sūtras state; apart from these, the genuine life taught in the Abhidharmakośa does not belong to plants.)

Pause-verse:

The subject-property, whether on the external object or before the mind, When it does not abide is the unestablished; Before one disputant or both, when it cannot be established — this is to be analysed and explained.

Truly unestablished on the external object: “the pillar is permanent, because it is produced.” Established on the object but not before the mind: “your future life will be painful or happy, because you have a future life,” put to a materialist; or “the pillar is impermanent, because it perishes moment by moment,” put to a herd-boy selling cow-dung who holds that dung perishes only when burnt — he will answer “not established,” though after three years of study he would see his error. Such cases are not hopeless: by the path of correct reasoning one may lead the other gradually. Unestablished for one party only, or for both: “this pillar is permanent, because it is heard” — neither Buddhist nor non-Buddhist accepts that a pillar is heard.

The inconclusive reason — two parts: (1) the definition; (2) the classification.

The inconclusive is the reason that gives rise to doubt.

The subject-property is established, but the reason gives rise to doubt with respect to the thesis, especially as regards the similar and dissimilar class. It divides into uncommon and common; the Pramāṇavārttika treats both extensively.

The uncommon inconclusive reason

Apart from the subject-property, not established [in either class]; The uncommon inconclusive has four kinds.

The reason is established on the subject-property but on neither class.

1. Absent in both classes, so unseen: “sound is impermanent, because it is heard” — all that is heard has already become the matter under dispute, so no similar-class example exists, and no permanent thing that is heard exists either. (From another angle it is also unestablished, subject and reason being non-dual.)
2. Present in both classes, yet unseen: “this person descended from the sky, because he has eyes” — some who descend from the sky may have eyes, and countless who have eyes did not so descend.
3. Present in the similar class, yet unseen: “sound is impermanent, because it is produced, like the sound of the Veda,” urged before a Vedic scholar who holds the Veda’s sound permanent. The reason does hold of the similar class, and replacing the example with a pot would make the argument sound; but the opponent does not see it.
4. Present in the dissimilar class, yet unseen: “space is permanent, because it is non-produced, like the sound of the Veda,” urged before the same school. Without the example the argument is acceptable to Buddhists, space being an unconditioned dharma; but the opponent cannot see the dissimilar-class case (“if not permanent, then produced, like a pillar”).

A correct reason must pervade the similar class and be absent from the dissimilar class — whether non-existent there or merely unseen. In “sound is impermanent, because it is produced,” “produced” is pervaded by impermanence, though not every impermanent thing is produced; and “produced” is absent from the permanent. Had it belonged to both classes, it would prove nothing.

The common inconclusive reason — two parts: (1) the actual inconclusive reason; (2) the remainder inconclusive reason.

Differing from the subject-property, involving both [classes] in common; Classified as the actual and the remainder inconclusive reason.

Here the reason is established on the subject-property (otherwise one simply answers “unestablished”) but involves both classes, so it over-pervades and proves nothing.

The actual inconclusive reason

Seen in both classes, the other is not cut off.

Four cases.

1. Pervading both classes: “sound is impermanent, because it is an object of cognition.” Impermanent pillars and pots are objects of cognition (similar class pervaded), but so are space and other unconditioned dharmas (dissimilar class pervaded).
2. Pervading the similar class, pervading-or-not the dissimilar: “sound is produced by exertion, because it is impermanent.” Exertion-produced things like pots are impermanent; but among the non-exertion-produced, the sound of water is impermanent while space is permanent.
3. Pervading the dissimilar class, pervading-or-not the similar: “sound is not produced by exertion, because it is impermanent.” Among the non-exertion-produced, water’s sound is impermanent and space permanent; while the exertion-produced (a drum’s sound) is certainly impermanent. Test by both pervasions: a correct reason reverses cleanly (“whatever is produced is impermanent, like a pot”; “whatever is permanent is non-produced, like space”), whereas here no example that is both exertion-produced and permanent can be found.
4. Pervading-or-not in both classes: before the Vaiśeṣikas, who hold the atoms of the four elements permanent and one entity with the tangible, “sound is permanent, because it is intangible.” In the similar class (the permanent): space is intangible, the atom is tangible. In the dissimilar class (the impermanent): pots and pillars are tangible, electricity intangible.

Worldly reasoning is full of such faults — observing only the similar class and never the dissimilar. Much of what passes for argument among writers, in Cārvāka-style works on past and future lives, and even in the courts, would not survive the requirements of epistemology.

The remainder inconclusive reason — two parts: (1) the genuine remainder; (2) the contradictory remainder.

The remainder inconclusive reason does not appear in Dharmakīrti’s or Dignāga’s own treatises but in later commentators such as Devendrabuddhi; Sakya Paṇḍita and Mipham Rinpoche posit it separately, while masters such as Khedrup Je hold it can be folded into the ordinary inconclusive reason.

The genuine remainder is seen in the similar class, But not seen in the dissimilar class.

Mere non-observation in the dissimilar class is not enough; a correct reason requires certainty of its absence there. Where certainty has not arisen, doubt remains and the reason is the genuine remainder. Example: “this person is not omniscient, because he can speak.” On the similar class (the non-omniscient) speakers abound; on the dissimilar class (the omniscient) speakers also exist — one who had met Śākyamuni Buddha would know it — but the arguer has not seen them.

Seen in the dissimilar class, not seen in the similar class — the contradictory remainder.

Here certainty would have made it a contradictory reason outright; lacking certainty, it is posited separately. Example: “this person is omniscient, because he can speak” — the exact reverse of the preceding. In the dissimilar class (the non-omniscient) speaking has been seen; in the similar class (the omniscient) it has not been seen, though the omniscient certainly can speak.

The contradictory reason — two parts: (1) the definition; (2) the classification.

When the reason is established but the pervasion is reversed, it is contradictory.

The reason is established on the subject-property, but wholly contradicts the probandum. The inconclusive reason was analysed by pervading and non-pervading; this one by outright contradiction between reason and probandum.

Division into two by connection

The non-observation [reason] is subsumed in the nature-reason, so the contradictory reason has two kinds.

The non-observation reason folds into the nature-reason — “in the place ahead there can be no pot, because none is seen” establishes only that the convention “pot” is absent there, the substance having long been settled, and a conventional-establishing valid cognition is a nature-reason. Hence two contradictions: of the nature-reason (“sound is permanent, because it is produced”) and of the effect-reason (“sound is permanent, because it is produced by exertion”).

Ascertaining the effect-contradictory reason

An impermanent effect just now produced by exertion — being permanent and unchanging, [the reason] is contradictory. Being impermanent and changing, it is not the same as this; and likewise being unchanging is also unreasonable.

[Certain non-Buddhists] “Sound is permanent, because it is produced by exertion” is a correct effect-reason, not an effect-contradiction: sound is a quality of space, space is permanent, so sound issues from a permanent dharma.

Reply: sound just now uttered by exertion — in singing, in teaching — shows it does not issue from permanent space. A permanent dharma cannot change: a moment ago there was no sound of reviling and now there is; a moment ago no recitation and now there is. Were it permanent it would either sound forever or never sound at all.

[Objection] Sound abides in space all along, but a damp wind obscures it; remove the wind and it issues — as a pot in a pitch-dark room appears the moment a lamp is brought.

Reply: the lamp case works only because everything in it is impermanent — pot, room and darkness alike change. A permanent dharma admits no obscuring, no unobscuring, and no benefit or harm from conditions whatever.

Division into three relative to the formal argument

The non-observation reason is set up as a third, and conversely the contradictory too becomes three.

By connection the non-observation reason folds into the nature-reason; relative to the formal argument it may stand as a third kind, and correspondingly a third contradiction — the non-observation-contradiction — may be added: “on clay that has no pot-bellied shape there exists a pot, because it does not appear.” Examined on the essence of the thing, it reduces to the nature-contradictory reason.

Division by the speaker’s intent — three parts: (1) stating the other school’s view; (2) showing the reasonableness of our own school; (3) dispelling disputes about it.

Some say that, by predicate and by subject and their distinctions, there are four contradictions.

[Early epistemologists] Four contradictions, posited on the essence of the thing:

1. Refuting the essence of the predicate: “sound is permanent, because it is produced” — the permanent cannot be produced.
2. Refuting a distinction of the predicate: “sound is not produced by exertion, because it is produced” — “produced by exertion” (i.e. produced by the exertion of causes and conditions, not merely human force) is a distinction within impermanence, and contradicts “produced.”
3. Refuting the essence of the subject: “the real-entity space is permanent, because it is non-produced” — “non-produced” does not contradict space as such, but a real entity is conditioned, and the conditioned contradicts the non-produced.
4. Refuting a distinction of the subject: “sound is impermanent, because it is not produced by exertion” — impermanent things include both the exertion-produced (a pot) and the non-exertion-produced (water’s sound, natural crystal).

All four fail. Where the essence of the subject is refuted the subject-property is not established, and an argument with an unestablished subject belongs to the unestablished reason, not the contradictory — the Auto-commentary notes that the third case is no different from “ultimate sound is permanent, because it is non-produced.” By the stated definition, the contradictory reason presupposes an established subject-property. And the whole scheme is wrongly posited on the essence of the thing rather than on the speaker’s intent.

Showing the reasonableness of our own school

Explicit and implicit, predicate and subject — the contradictions are brought about by the speaker’s intent.

Four contradictions, all posited from the speaker’s intent:

1. Explicit: “impermanent sound is permanent, because it is produced” — a plain, direct contradiction needing no unpacking.
2. Implicit (indirect): “space is a permanent agent, because it is without function.” “Without function” fits space, being the mark of an unconditioned dharma; but an agent fabricates through function, and having function contradicts being permanent — so the contradiction emerges only on analysis of “agent.”
3. Predicate-stated: “this flower grows in space, because it possesses fragrance.” Fragrance is established of an ordinary flower, but not of one whose source is space; the contradiction lies in the distinguished basis of the predicate, not in the subject.
4. Subject-stated: “the flower in space possesses radiance, because it is without colour.” “Without colour” can be said of a sky-flower; “possesses radiance” cannot — there is no such flower, and radiance belongs to colour, being among the derived visible forms of the form-aggregate in the Abhidharmakośa.

All of this is analysed from the speaker’s side, not from the meaning: on the genuine essence of a thing, two contradictory things cannot coexist, and no real sky-flower stands available for examination.

Dispelling disputes about it

If it be said that genuine reasons are likewise [divisible], the use is trifling, so we set it aside.

[Some masters] If contradiction divides fourfold, genuine reasons should too — explicit, implicit, subject-stated and predicate-stated nature-reasons.

Reply: they may be so divided if one insists — contradiction reversed is connection, and connection is causal or identical-nature, so the two angles exist. But the use is trifling. Dividing contradiction is necessary because many non-Buddhists take contradictions for genuine reasons (holding “the flower in space possesses radiance, because it is without colour” to be a real inference). Dignāga and Dharmakīrti neither made nor forbade the corresponding division of genuine reasons; Mipham Rinpoche’s Great Commentary on the Pramāṇavārttika takes the same tone.

Showing that other classifications accepted are unreasonable

The intent and what is uttered — these two are equally the probandum; the “benefiting-another” contradiction and the rest are not admitted apart from these three.

[Sāṃkhya] “The eyes and so on benefit another, because they are an assemblage, like a bed-cushion.” The subject is the twenty-four principles besides the puruṣa (Gorampa’s epistemological commentary restricts it to the eighteen insentient ones). In their minds they mean the eyes benefit the puruṣa — “benefit oneself” — but they say “benefit another,” fearing that the plain statement would not be accepted by Buddhists; a bed-cushion benefiting the body is offered as illustration.

The argument is contradictory: precisely because the eyes and the rest are assembled things, they cannot benefit a puruṣa, and the bed-cushion illustration fails since a cushion performs no function for any such puruṣa. And this “mouth-contrary-to-mind” contradiction needs no separate class: whatever is intended in the mind and whatever is uttered with the mouth are alike posited as the probandum, so it is already covered by the three kinds above.

The fixed number of these reasons — two parts: (1) the fixed number that excludes the extremes; (2) their functions.

Refuting the other extreme, the fixity: the reasons are posited as a fixed number of four.

One genuine reason and three fallacious. More is unnecessary and fewer impossible. On the subject-property the reason is either not established (the unestablished reason, answered directly with “unestablished”) or established; if established, it either definitely proves the probandum (the genuine reason), definitely contradicts it (the contradictory reason), or leaves it undecided (the inconclusive reason). No fifth case remains.

The reasons of the probandum being four, the functions too should be of four kinds.

1. By an unestablished reason not even doubt arises: “this pot is impermanent, because it is non-produced” — one answers “unestablished” at once.
2. By an inconclusive reason doubt arises: “the pot is impermanent, because it is an object of cognition” — objects of cognition include permanent things like space; or “past and future lives do not exist, because I have not perceived them.”
3. By a contradictory reason one states what is wholly unacceptable: “the pot is permanent, because it is produced.”
4. By a genuine reason acceptance and certainty arise: “the pillar is impermanent, because it is produced” — one assents at once.

2. Explaining the Probandum Proven by the Reason

Three parts: (1) positing the definition; (2) identifying the instance; (3) showing the contradictions that harm the subject-property. The contradictions treated earlier were on the side of the reason; these are on the side of the probandum.

Positing the definition — three parts: (1) stating five defining characteristics relative to negation; (2) establishing one relative to affirmation; (3) dispelling disputes about it.

Stating five defining characteristics relative to negation

The Pramāṇasamuccaya says it possesses five characteristics.

Dignāga’s phrase “single in essence, accepted oneself, not contradictory” yields five: essence, single, oneself, accepted, non-contradictory. Each is stated by negation, indirectly affirming the mark by excluding its contrary.

1. Essence: exclude what is already established. Since the opponent already accepts that sound is heard, nothing is proven by proving it again; the probandum is what was never before verified — as when one knows sound to be produced but has no concept of its moment-by-moment change.
2. Single: exclude the case where probandum and prover coincide, as in “the pillar is impermanent, because it is impermanent.” (In “the pillar is impermanent, because it is produced” the essence of the prover is impermanence, but that notion is not yet established for the proponent, which is why proof is needed at all.)
3. Oneself: exclude taking every principle of the treatise one cites as one’s probandum. Citing “sound is impermanent, because it is produced” from the Pramāṇavārttika does not commit one to every other view in it; one establishes only the thesis one has stated.
4. Accepted: exclude the mouth-contrary-to-mind thesis (the Sāṃkhya “the eyes and so on benefit another, because they are an assemblage, like a bed-cushion”). The genuine probandum is jointly identified by proponent and opponent, and the opponent has the right to question its scope before debate opens — asking, for instance, in what sense “impermanence” or “past and future lives” is meant.
5. Non-contradictory: exclude a probandum already harmed by contradiction. “Sound is non-heard, because it is impermanent” contradicts both common renown and direct perception; where contradiction is present there is nothing left to prove.

Establishing one relative to affirmation

In the Nyāyamukha, the probandum is acceptance.

Stated affirmatively, the probandum requires only acceptance — a view both parties know and one party maintains: that the pillar is impermanent, that there is fire on the mountain.

Dispelling disputes about it

Here there is no fault of non-pervasion, over-pervasion, or sameness-with-the-reason to dispute.

[Some masters] Three faults attach to defining the probandum as acceptance. (i) Non-pervasion: sudden inference — seeing smoke and knowing at once there is fire — has no acceptance. (ii) Over-pervasion: an answer unsuited to the occasion is also an acceptance; asked whether sound is permanent, one replies “there is fire on the mountain,” and that too becomes a probandum. (iii) Sameness with the reason: if the probandum needs acceptance, so does the reason, since “because it is produced” is equally accepted.

Reply. (i) In sudden inference there is a tacit, indirect acceptance — the mind at once affirms fire. (ii) “There is fire on the mountain,” or even “there is a louse on your head,” is a general acceptance; and it naturally becomes a stated thesis as soon as one asks why. (iii) The reason may indeed be accepted, but acceptance is not its defining characteristic; the reason’s defining characteristic is completeness in the three modes. With the three modes complete the point stands whether accepted or not; with them incomplete a hundred acceptances are useless. Only the probandum must be accepted and then proven by the reason.

Identifying the instance — two parts: (1) the actual statement; (2) dispelling disputes.

The accepted instance of the probandum is divided into “true,” “false,” and “proving falsely”; grasping the conventional universal as the particular — this is the instance of the probandum.

[Shawa Chösang and followers] The particular (the true reality of the thing) is the instance. [The translator Ngok and followers] The universal, the bare excluded part, is the instance. Both fail.

The bare particular cannot serve: as shown in the chapter on exclusion, particulars are innumerable — sound’s impermanence has countless particulars across environments, forms and times, and no one can take them all as the instance. The bare universal cannot serve either: proving the universal helps and harms no adoption or rejection, and one may say “sound is impermanent” a hundred times without producing certainty in anyone who holds it permanent.

Our own school takes the combined part of subject and predicate in which conventional universal and particular thing are fused into one. The proponent’s concept clings to the particular sound-impermanence, and the opponent’s refutation is aimed at that same fused dharma.

Dispelling disputes

This establishes the real and the unreal, and there is no occasion for dispute.

[Objection] Take “sound is permanent, because it is non-produced.” If “sound is permanent” is the instance of the probandum, are sound and permanent separate or combined? Separate is unreasonable — neither bare sound nor bare permanence is the probandum here. Combined is unreasonable too — a thing both sound and permanent cannot be found.

Reply: what is established is neither sound stripped of permanence nor permanence stripped of sound. Before the excluding cognition, by their own habit and tradition, non-Buddhists who have never studied impermanence do join the two — sound as a real entity, permanence as an unreal one — and establish “sound is permanent” with universal and particular fused. Since it stands before their mind, it can serve as the object of our refutation, and breaking that wrong view is exactly what is needed.

Showing the contradictions that harm the subject-property — two parts: (1) the general positing; (2) the nature of each.

The general positing — three parts: (1) the definition; (2) subsumption; (3) dispelling disputes.

The definition

Contradiction, once established, conflicts upon the thesis.

The contradictory dharma conflicts with the subject-property the moment the thesis is stated — by contradiction with worldly common renown, with one’s own pledge, with direct perception, or with inference.

Subsumption

The five kinds, summarised, are said to be four; by the object, inference is divided; by this, contradiction is said to have four kinds — direct perception, inference, factually grounded valid cognition; acceptance and common renown, being superimpositions, are therefore assigned within inference.

Dignāga’s Nyāyamukha teaches five contradictions: with inference, with direct perception, with acceptance, with common renown, and with one’s own statement. Dharmakīrti in the Pramāṇavārttika reduces them to four, merging contradiction-with-one’s-own-statement into contradiction-with-acceptance, since both are the inconsistency of what one has held. Sakya Paṇḍita follows him.

By the object: inference is factually grounded reasoning, and within inference fall three — contradiction with factually grounded reasoning (“the pillar is permanent,” refuted the moment one infers from its being produced), contradiction with acceptance (yesterday’s pledge to stay three years overturned by today’s request to leave), and contradiction with common renown (calling the yak a dog or a pig, which everyone objects to). Adding contradiction with direct perception gives four.

These four could be compressed further into direct perception and inference, both being factually grounded valid cognitions; but those two cannot be reduced to one another, even though direct perception is the root of inference. Contradiction with acceptance and with common renown are contradictions with a superimposed positing and are known by inference (that yesterday’s words conflict with today’s is itself an inference), so they belong within inference. Dignāga gives five characteristics, Dharmakīrti four, and Sakya Paṇḍita indirectly points to two — a steadily condensing tendency.

Dispelling disputes

If [you say] that as with inference, so direct perception should likewise divide into two contradictions — [we answer:] the acceptance [case] is solely an effect-reason, and the expressible [case] is not established by direct perception.

[Objection] If inference admits contradiction-with-acceptance and contradiction-with-common-renown, so should direct perception: first denying a double moon and then directly seeing one (contradiction-with-acceptance); and naming a pot “pillar” or a pillar “yak,” which words can express but direct perception will not bear (contradiction-with-common-renown).

Reply: the double-moon case is an effect-reason inference, hence a contradiction within inference, not within direct perception. And what is expressible is established by inference, not by direct perception; the opponent’s appeal to reflexive direct perception fails. So both belong to inference.

The nature of each — two parts: (1) contradiction with direct perception; (2) contradiction with inference.

Contradiction with direct perception

The thesis being opposed to experience — this is contradiction with direct perception.

The thesis conflicts with the speaker’s or hearer’s own experience, and is refuted on the spot: someone plainly wise is called as stupid as a yak; or “sound is not heard, sound is seen by the eyes” — the moment it is said, personal experience answers it.

Contradiction with inference — three parts: (1) with factually grounded reasoning; (2) with conviction; (3) with common renown.

Contradiction with factually grounded reasoning

If the factually grounded reasoning of the three modes conflicts with the thesis, it is contradiction with the reason.

Unlike direct-perception contradiction, which is felt the instant the words fall, this is recognised only through reasoning. “Sound is permanent” may not jar at once, but permanence is the mark of an unconditioned dharma, and sound is conditioned, therefore impermanent. Likewise “there is fire on the mountain, because there is a fine horse.”

Contradiction with conviction — three parts: (1) the definition; (2) the classification; (3) analysing the meaning.

Trustworthy words conflicting with the thesis — the contradiction with conviction.

The thesis conflicts with trustworthy language — one’s own former and later words, or scriptural authority.

It divides into [contradiction with] acceptance and with one’s own statement.

The division follows the Nyāyamukha. (On its authorship: some hold it composed by Śaṅkarasvāmin and translated into Chinese by Xuanzang at Hongfu Monastery, thence into Tibetan; another Tibetan version was made at Sakya directly from Sanskrit. Most Tibetan masters hold it Dignāga’s, though a few Chinese scholars do not.)

Contradiction with acceptance (also called contradiction with scripture) is former and later words conflicting — accepting sound as permanent and minutes later calling it impermanent; or professing ever-growing faith in the guru and, two minutes later, ever-diminishing faith. Contradiction with one’s own statement is a direct or indirect conflict within one’s own words: “my body is very fine, only I take medicine every day”; “my mother is a barren woman.”

Truly explaining the distinction

Others say it is scripture and one’s own statement.

[Some early epistemologists] The two are wholly separate: contradiction-with-acceptance is contradiction with scripture (saying giving yields no wealth, against scripture’s word that it does), while contradiction-with-one’s-own-statement is the speaker’s self-contradiction.

The master accepts a difference of statement. The Pramāṇasamuccaya, by a single simile, sets forth scripture and one’s own statement; the Nyāyamukha teaches another simile.

Dharmakīrti, in the Pramāṇaviniścaya, holds the two not different in substance but only in mode of verbal expression: one is posited by the conflict of former and later acceptance, the other by direct or indirect conflict within one’s own statement, but both are attacking one’s own shield with one’s own spear.

The Pramāṇasamuccaya covers both with a single simile — “the valid cognition possessing the object of measurement does not exist” — since in both cases the valid cognition measured is incomplete, whether one says “my mother is a barren woman” or “sound is permanent, sound is impermanent.” The Nyāyamukha gives contradiction with scripture a simile of its own: “[it is] both a virtuous dharma and one that, after death, does not bestow the fruit of happiness.” The slight separation does not conflict with Dharmakīrti, for the two do differ in isolate.

Dividing by the distinction of scripture

The two valid cognitions not harming the two objects measured, one’s own statement not harming the hidden matter — this is the Buddha’s word, our scripture, said to be subsumed within inference.

Scriptural valid cognition is “the scriptural word pure in the three examinations”: in the manifest part, unharmed by direct perception; in the hidden part, uncontradicted by inference; and in the extremely hidden part — matters of place, form and time that ordinary beings cannot reach — free of any direct or indirect conflict among the Buddha’s own former and later words. The Seven Treatises hold all valid cognition to be direct perception or inference; scriptural valid cognition is subsumed within inference.

Dharmakīrti in the Pramāṇavārttika: “The word of conviction is undeceiving; in general, on these hidden matters, there is also no occasion [for harm], so the wise admit it within inference.” This conviction is not the blind acceptance of a father’s or a founder’s word; the Buddha himself said in the sūtras that monks should accept his words only after examining them as one refines gold.

Therefore scripture established by reasoning is correct valid cognition; if it cannot be established, then, being equal to [contradiction with] one’s own statement, it is accepted as having the nature of obstruction.

Scripture must not contradict the factually grounded reasoning of direct perception and inference; if it did — if the Buddha called the pillar impermanent and it proved permanent — his words would not be trustworthy and his omniscience would fail. Words that cannot be established by reasoning stand on a level with any ordinary being’s self-contradiction.

Harm and obstruction differ. No harm, with obstruction: sound is impermanent and someone calls it permanent — the claim does sound’s impermanence no harm, yet obstructs proof, since he will not accept it. Harm with obstruction: one party establishes sound as permanent and the other as impermanent — the true thesis both harms and obstructs the false. Neither harm nor obstruction is ordinary correct inference. Harm without obstruction does not occur.

Pause-verse:

Not every commonly renowned “teaching” is valid cognition; what is established by valid cognition is scripture. To accept first and examine after is the fool’s way; to examine first and accept after is the wise one’s path.

Every religion holds its own texts to be scriptural valid cognition, but not everything called “teaching” is undeceiving. Genuine scripture is what survives examination by wisdom and inference without harm or contradiction. Hence the order: examine first, accept afterwards. To take up a teaching untested and later withdraw on finding it illogical is the fool’s course; to examine all and then take refuge with diligence is the wise one’s.

Dispelling the equation with other schools’ [scriptures]

[Opponent:] contradicting another’s teaching likewise becomes that [contradiction]. [Reply:] desire and the rest are non-Dharma, freedom from desire is Dharma — all but the nihilists say so with one voice; bathing and desire and the rest, their causes, do not contradict; so such treatises are not scripture.

[Opponent] The Veda says bathing purifies negative karma; to contradict that should equally be contradiction with scripture, since the Veda too is scripture.

Reply: the claim itself is unreasonable, so the texts are not scripture. Apart from the Cārvāka nihilists, all schools accept with one voice that desire, anger and delusion are non-Dharma and freedom from them Dharma. For bathing to dispel the afflictions, either their essences or their causes would have to contradict — but desire is a clinging of the mind and bathing a washing of the body, and the two do not touch. So bathing in the Ganges to remove obscurations accords neither with reason nor with cause and effect.

(This does not impugn Buddhist ablution rites: in empowerment the guru bathes the disciple while reciting the Vajrasattva mantra and the Hundred Syllables, and obscurations are purified through the assembling of ritual, mantra, mudrā and samādhi, water being one condition among many rather than the agent. The non-Buddhist practice invokes no mantra, mudrā or meditation and rests on water alone. The Pramāṇavārttika makes the parallel point that loving-kindness and the rest do not contradict the self and so cannot dispel self-grasping.)

Dispelling the fault that obstruction does not establish the probandum

If it be said: in the obstruction by one’s own statement or by a treatise’s meaning, only doubt is produced; that is a faultless probandum, and so not a fault of the subject-property —

[Objection] Self-contradiction and contradiction-with-scripture merely produce doubt about the probandum; doubt is not itself a fault, so this is no fault of the subject-property.

— this is not that, from within the obstruction, the arising of doubt becomes the fault; it is that those words cannot prove [it], so the probandum is posited as a fault.

Reply: the fault is not the doubt as such but the incapacity of such words to prove anything. Saying first that sound is impermanent and then that it is permanent leaves the hearer hesitant, and “permanent” obstructs the establishing of “impermanent.” Asked whether past and future lives exist and answering now yes, now no, one establishes nothing even conventionally. Obstruction indirectly entails harm to the thesis: where one merely speaks without proving, obstruction may stand without harming, but wherever a thesis is to be established, an obstructive dharma in the end prevents its proof.

The thesis declares one’s own intent; if it is contradictory, the thesis is destroyed. Therefore faulty words, in debate, invite one’s own defeat.

The thesis makes plain what the proponent accepts and rejects; a direct or indirect self-contradiction destroys it. As a little yoghurt turns a whole pot of milk sour, one self-contradicting sentence spoils everything else said. In debate one must remember what one has already accepted and never transgress it, or the opponent will turn it back — like cutting one’s own head with one’s own sword. (Mipham Rinpoche treats the responsibility of speech in the Speech chapter of the Treatise on Ethics for Kings.)

Contradiction with common renown — two parts: (1) the definition; (2) determining its nature.

The definition

The thesis opposing worldly common renown — this is contradiction with common renown.

The world has fixed names — “the thing with a big belly that holds water” is a pot, “the animal with hump and dewlap” an ox — and transgressing them is contradiction with common renown, as calling a train an aeroplane.

Two senses must be distinguished throughout what follows: what is already commonly renowned, and what can be commonly renowned in future. From the first angle the train cannot be called an aeroplane; from the second it can, since the first namer had no reason for the name either. Names are given sometimes with a reason and sometimes casually — Tibetans generally choose auspicious personal names, elsewhere names are given without regard to meaning — and no rule makes an auspicious person’s name “Auspicious.”

Determining its nature — two parts: (1) the brief statement; (2) the extensive statement.

The brief statement

The direct connection of name and meaning accepted by the Śabda masters was already refuted before; that whatever is said at will is verified — therefore common renown too is established.

[Śabda school] Name and meaning have a particular intrinsic connection.

This was refuted in the fifth chapter, on expressed and expresser: were there such a connection, uttering “fire” would set flames in the mouth and uttering “water” would set water flowing. Language has no relation to the particular of the external object; all conventions are posited at will by the namer.

Hence common renown has two kinds: genuine common renown (the humped animal already called an ox, the being who knows speech already called a human) and posited common renown (what may be renowned in future — the same animal called “fine horse”). From the second angle any dharma may bear any name, and no confusion follows, since the positing is established by reasoning.

Dignāga’s Auto-commentary on the Pramāṇasamuccaya supplies the key sentence: “Because it is uncommon there is no inference; it contradicts the common renown already established for the name; therefore that contradiction too is not a subject-property.” Dharmakīrti in the Pramāṇavārttika divides it into two parts, each analysed by refuting the other school and establishing our own.

Refuting the other school’s view

By the reason “because it exists” [the opponent] refutes the statement that the hare-bearer is the moon.

“The hare-bearer” is a poetic name for the moon, from the hare-image seen on its face. [Śabda school] The hare-bearer cannot take the name “moon,” because it exists. (Mipham Rinpoche and others note that the Śabda school does in fact accept that the hare-bearer may be called “the moon”; what they deny is that the name may be transferred to a pillar or pot. But their wording, taken by reasoning, amounts to denying even the first.)

Since they hold name and particular directly related, their own reason turns on them: if what exists cannot be so named, then the hare-bearer, which exists, cannot be named “the moon” either — for the existence of any other dharma is not the hare-bearer’s existence.

By our view the hare-bearer may be called “the moon,” since that designation is complete and denying it incurs the harm of common renown; and by the same reasoning the ox may be called “the moon” too — not by what is already renowned, but by what can be renowned.

This has no example: whatever exists, all names may be spoken of it —

First fault: the reason is inconclusive, having no similar-class example. Whatever exists may bear any name at all; indeed even what does not exist may — the barren woman’s son may be called “the moon” in imagination.

The meaning of the name existing, it may be called “the moon” — therefore it is contradictory.

Second fault: the reason is contradictory, since the very existence of the thing named is the ground on which it may be called “the moon.”

Expressible common renown

Established by the world’s use of a name, what is accepted as such is common renown.

Expressible common renown means any dharma may take a different name — the pot called “the moon,” the moon “the sun.” From the angle of what is already renowned this jars against habit; from the angle of what can be renowned it is entirely admissible. Were name and meaning directly related, one dharma could have only one name, and many names would mean many dharmas — yet people take new names constantly (a Dharma-name from each guru, besides lay name and nicknames), and in some Tibetan regions a person sharing a dead neighbour’s name must change it so the bereaved are not grieved.

Since it [applies] to all knowable dharmas, the [potentially] renowned pervades everything.

What can be renowned pervades every knowable dharma; names may be exchanged again and again without restriction.

[Opponent:] what has a designation is a renowned term; for the not-yet-renowned it contradicts the world. [Reply:] it is fitting for all knowables; using a renowned name is like cooking.

[Opponent] A dharma that already has a name, called by a name never used of it, will strike everyone as transgressing common renown.

Reply: examined by reasoning, other names may be used, since all names are posited by imagination and nothing is established on the essence of the thing. As what may be eaten is called “food” and what may be drunk “drink” before any eating or drinking occurs, so what may be so named may be so named now.

Identifying the common renown already established for the name

What can be commonly renowned is established on the thing; yet in dependence on the name and the intent to speak, if the name is established, the designation is established — so it is called the renown established for the name. Proven by the inference of common renown, a contradiction without error is impossible.

Any name may be established on any thing, the ground being that names are wholly posited by imagination and the intent to speak. This is itself a kind of inference — the inference of common renown: “any dharma may be commonly renowned by ‘the moon,’ because people can posit it by their imagination.” Is there a counter-inference proving that only the hare-bearer may bear the name? There is none; and two conflicting inferences cannot both verify one thing. So the inference of common renown stands unopposed.

Showing that the factually grounded inference is an object separate from expressible common renown

If common renown were established by factually grounded reasoning, its object would be determinate, like an inference’s; [but] named in dependence on the intent to speak, what name is there that cannot be used?

The object of genuine inference is fixed: smoke arises only within fire and not in an ant-hill, and the pillar’s impermanence and “produced” are joined by identity of nature. Were names tied to particulars as the non-Buddhists claim, the object of a name would be equally fixed. But naming rests on imagination — the first namer called this red thing “pillar” for no determinate reason — so the object of the inference of common renown is unfixed and variable. That is the difference between the two inferences.

Refuting the other school’s view

[Opponent:] because it reaches everything, common renown is not applicable to [a fixed] object. [Reply:] this account can find no other example; it is an uncommon reason, like “heard.”

[Śabda school] Common renown applies only to a fixed object, not to all things.

Reply: “the hare-bearer cannot be replaced by ‘moon,’ because it exists” has no similar-class example, exactly like “sound is impermanent, because it is heard.” It is therefore an uncommon inconclusive reason, and a fallacious reason proves nothing.

[Certain Buddhist masters commenting on the Pramāṇasamuccaya] Dignāga’s citation shows only that “moon” applies to the hare-bearer and not to others, so expressible common renown has no example and is an uncommon inconclusive reason.

It was when explaining the universal that the hare-bearer was used as an example, to show that it pervades all knowables; “the daré is not a tree” and the rest are likewise.

Reply: Dignāga was expounding the universal of common renown, using hare-bearer and moon only as an illustration; he nowhere states that “moon” applies to that thing alone. The purpose of the example is precisely to show that such a name may be used of all knowables. In poetics “moon” describes a fair round face (“a face like the full moon”), and people bear it as a personal name — Candrakīrti’s very name means “renowned as the moon.”

Denying this incurs the harm of common renown. “Daré” is the name of a medicinal tree; to say “the daré is not a tree” is like saying “the pine is not a tree” — everyone objects. Just so, to say that some dharma cannot be called “moon” carries the same harm, once one shifts from what is already renowned to what can be renowned. (“Hare-bearer” also renders as “having a hare” in Master Fazun’s translation, perhaps to avoid the suggestion of the moon carrying a hare in pregnancy; either is admissible.)

Explaining the master’s intent

Camphor, mercury, and the rest are commonly renowned by the name “moon”; taking these as examples, it is proven — “the cool-light-bearer” and “moon” share the common reason.

Dharmakīrti’s intent: the hare-bearer may bear the name “moon,” and so may many other things. Among incense materials there is a white ingredient called “moon”; among the substances compounded for inedia practice likewise; among Tibetan medicines a white flower is denoted by “moon”; in tantric practice sun and moon stand for the red and white bodhicitta. In poetics the moon is also “the cool-light-bearer,” from its cool light, and physicians bear names like “Moonlight.” Since the name attaches to so many things, the claim that it attaches to the hare-bearer alone fails.

Note further that “the cool-light-bearer” is a name given with a reason, and “pot” or “pillar” without one; but where the intent to speak is primary, either sort may be applied to anything. A pot may be called “snow-mountain” and a snow-mountain “pot” without harm.

All names used are held in common; if established on one, established on all — this is factually grounded reasoning; and refuting common renown destroys you too.

Any name given with the intent to speak becomes the common property of all things: if “moon” is truly established on the hare-bearer, it is established on the yak, the elephant, anything. This is settled by factually grounded reasoning.

And the opponent’s denial destroys his own position. Asked for a firm ground why “moon” attaches to the hare-bearer alone, he can offer only traditional habit — which we grant as already renowned, while nothing prevents a future namer from applying it elsewhere. If he refutes our view he refutes his own, since his own use of “moon” for the hare-bearer rests on the same footing; and with that, worldly conventions gradually collapse. The Buddha permitted the giving of names precisely because without them the world would be confused into one and no one could tell what to adopt and reject.

Therefore names given according to one’s wish evidently extend to all dharmas; to refute this has the harm of the world, so contradicting common renown is the [author’s] intent.

Names given by imagination reach all dharmas. “Ox” was first given for hump and dewlap, yet from the angle of what can be designated it may be applied to an elephant or a fine horse without harm. To deny this — to insist “moon” belongs to the hare-bearer alone and “ox” to the humped animal alone — incurs the harm of worldly common renown, since naming is posited by conceptual thought and conceptual thought can take anything as its object; this is established both by reflexive awareness and by inference. That is what “contradiction with common renown” means, and it is the ultimate intent of Dignāga and Dharmakīrti.

(The Abhisamayālaṃkāra and other treatises likewise hold all things to be posited by the conceptual designation of the falsely existent and of the truly existent; understanding this loosens the grasping at names as real and helps one recognise that dharmas are like illusion and dream.)

Analysing their intent — two parts: (1) refuting the other school’s view; (2) stating our own school’s view.

Some divide this into two kinds: terminological and conventional common renown.

[Early epistemologist master Tsang Nagpa] Two kinds: terminological common renown, what is clung to merely by the name-feature of a thing (“pot,” “pillar,” “moon”); and conventional common renown, the clinging through name and consciousness to the meaning-isolate of a determinate meaning.

If so, the convention is not this meaning; the definition was likewise refuted before.

The name-feature is a thing’s special term, and can denote only what possesses the corresponding definition — “ox” only the humped animal. But the common renown under discussion extends to any dharma whatever, so their “convention” does not answer to the meaning at issue. And if the name-feature is taken as the name of the definition, that was already refuted in the eighth chapter, where treating definition and name-feature as one entity was shown to entail faults.

[They say:] the directly graspable expressible — terminological common renown too has two.

They subdivide terminological common renown into the directly expressible (the universal — the moon appearing in one’s head) and the graspable expressible (the particular — the genuine hare-bearer possessing the definition), on the model of the four expressers-and-expresseds treated earlier.

Though in exposition there are these two, in application there is no difference.

The division holds in exposition only. As the fifth chapter says, “In exposition one analyses and masters; in entering [application] one mistakes them for one entity.”

Therefore the two accounts — explaining them as the directly-spoken and the graspable — are mistaken.

Failing to distinguish exposition from application in the Pramāṇasamuccaya, they assign the latter to the directly expressible and the former to the graspable — a division that would work equally well reversed, which shows it is idle. The context here is proving a probandum by a reason, that is, application, where particular and universal must be fused. When we give the hare-bearer the name “moon” we do so with the two grasped as one; wholly separated, the term could indicate and denote but never function in genuine use.

Stating our own school’s view

The former is the renown already established for the name; having no [contrary] inference, it cannot be refuted; or, it having no opposing party, being thereby determinate, it can do harm.

The first part of Dignāga’s sentence — “because it is uncommon there is no inference” — refers to the renown already established for the name: it has an uncommon reason, the name being already formed, and no counter-inference exists to refute it. No direct perception or inference in the world harms it; conventionally it is well-grounded and is itself a forceful harming agent, so that denying the designation incurs the harm of correct valid cognition. (Examined by the ultimate truth such common renown is of course not established; that is no objection, since it holds conventionally.) Dharmakīrti explains Dignāga’s view in the last chapter of the Pramāṇavārttika.

The latter explains this common renown’s difference of object from inference.

The second part of the sentence marks the difference of object. Genuine inference has a fixed object — one infers fire by smoke and impermanence by “produced,” never fire by “produced.” The inference of common renown has no fixed object: since “moon” attaches to the hare-bearer it may attach to a person, a yak, any sentient or insentient thing. Its scope is vast where the other’s is fixed.

If one knows this principle, then is established the wise ones’ intent in [teaching] contradiction with common renown.

Once one sees by way of exclusion that all worldly names are posited by beings’ conceptual thought, the inference of common renown is established, and with it the intent of Dignāga and Dharmakīrti in teaching contradiction with common renown. Mipham Rinpoche analyses the topic from both sides: from the side of establishing, since the hare-bearer is called “moon,” all things may bear the name; from the side of refuting, denying the name — whether to the hare-bearer or to anything else — incurs the harm of worldly common renown.

Here ends the tenth chapter of the Pramāṇayuktanidhi*, Examining Inference for Oneself.*

---

← 9. Examining Perception | All chapters | 11. Inference for Others →