Chapter 11: Inference for Others
Inference for others is speech, and therefore not valid cognition proper: speech is sound, sound is insentient, and valid cognition must be a non-deceptive consciousness. It is called valid only by transferring the name of the effect to its cause, since the words produce an unmistaken inference in the hearer’s continuum. The chapter posits the proponent, the opponent and the witness, settles victory and defeat, classifies the language of debate, and ends with the treatise’s colophon.
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
Three parts: (1) the definition of inference for others; (2) positing the persons; (3) classifying the language of debate.
a. The Definition of Inference for Others
Inference for others is the speech that declares to others what one has seen oneself.
Inference for others is speech: the proponent, by reasoning, declares before another the meaning he has himself ascertained. It is therefore not genuine valid cognition — speech is sound, sound is insentient, and valid cognition must be a non-deceptive consciousness. It is called valid only by naming the cause after its effect: relying on this speech, an unmistaken inference arises in the other’s continuum. Saying “there must be fire on the mountain, because there is smoke” produces in the hearer a correct ascertainment of fire; the name of that effect is transferred back onto its cause, the words.
b. Positing the Persons
Three parts: (1) their respective definitions; (2) positing victory and defeat; (3) how the witness establishes the outcome.
1. Their Respective Definitions
The proponent, the opponent, and the witness — in turn they prove, refute, and adjudicate.
The proponent states the proof and establishes his own position. The opponent refutes the other’s position — if someone asserts sound is permanent, the opponent overturns that acceptance and thereby indirectly establishes impermanence. The witness analyses by a correct route and decides victory and defeat when the two cannot decide it between themselves.
In a formal arena all three are present; informal training may dispense with the witness, but a large debate needs one, since the proponent will not always concede the opponent’s overreaching charges. The arrangement matches a court: the judge in the middle renders judgment on the lawyers’ dispute. Roles can blur — an opponent questioned may be forced to establish something, a proponent attacked may rebut — but the division holds: the proponent establishes and should not be long-winded; the opponent refutes, and his own position is beside the point.
2. Positing Victory and Defeat
The two debating parties, through merit and fault, enact subjugation and winning-over. Stated thus, there is victory and defeat; if not, there is neither — so says the Treatise on Debate (Vādanyāya).
The proponent establishes a reasonable position; the opponent points out faults. If the position fails, the opponent subdues him, and the proponent, sincerely convinced, concedes and may be won over — as in the great Indian debates, where non-Buddhists defeated at Nālandā entered Buddhism.
Where the opponent lacks the wisdom, ability, or scriptural and rational basis to rebut, or where the two wrangle without end, the debate reaches no conclusion and there is no victory or defeat. Dharmakīrti’s Vādanyāya (which expounds the twenty-two points of defeat) holds that victory and defeat obtain only when the opponent has actually understood the other’s position and rebuts on that basis; otherwise the two remain on a par, which is unreasonable — a debate should begin from differing positions and end with one victor.
Though one may subdue through lies and deception, that is not admitted as [genuine] victory or defeat.
Subduing by lies or craft is no victory. The Vādanyāya records the claim that “to protect Buddhism it is reasonable for the upright to subdue by deception”; this is rejected. Deception, blows, weapons, arson — none of these uphold the Dharma, and a victory so gained cannot be justified in terms of cause and effect. The typical evasion is citing a scriptural line that does not exist, and then, when everyone looks it up, pleading headache, sleepiness, and reluctance to continue. Buddhism convinces by reason alone; contests of physical strength or intimidation are not debate.
3. How the Witness Establishes the Outcome
Two parts: (1) a general statement of the principle of subduing and winning-over; (2) determining the points of defeat in particular.
General Statement of the Principle of Subduing and Winning-over
What the world calls subjugation and winning-over is to seize and to bestow; but in adopting or discarding a tenet-position, what is admitted is the true person’s subduing [of error] and upholding [of truth].
Worldly subjugation means stripping someone of property, status and reputation; worldly winning-over means bestowing them. In epistemological debate neither giving nor seizing is at issue. On the principle of adopting or discarding a tenet, whoever cannot hold up before genuine reasoning is defeated; whoever expounds coherently with reason and evidence has won.
Victory built on force, prestige, or political power — whether that of ancient Indian non-Buddhists or of later state religions — is worth nothing here. Nor may status decide: “you are the teacher, I the disciple, so I grant you the victory” has no place; teacher and disciple debate on an equal footing, as Sakya Paṇḍita did against non-Buddhists. The witness must understand this, or he cannot do his job.
Determining the Points of Defeat in Particular
Two parts: (1) stating the correct view; (2) refuting mistaken views.
Stating the Correct View
The points of defeat, shared by both parties, are four; this is the intent of those who know reasoning.
A point of defeat (负处) is a place of failure at which the witness must rule the party unreasonable. Following Dharmakīrti and Dignāga, Sakya Paṇḍita admits exactly four: two per side.
The proponent’s two: not stating a proof — asserting “sound is permanent” and, asked why, giving no grounds at all; and stating a non-proof — giving grounds that are no grounds, as “sound is impermanent, because it is a pillar,” or “sound is permanent, because it is a yak.”
The opponent’s two: not stating the fault — the proponent argues “sound is permanent, because it is not produced” and the opponent, though standing, says nothing, which forfeits the very defining character of an opponent; and stating a non-fault — speaking irrelevancies, as replying to “sound is permanent, like space” with “is the yak permanent or impermanent?”
These four exhaust all faults of both parties, and every point of defeat proposed by others reduces to them. This is the key to the whole refutation that follows.
Refuting Mistaken Views
Two parts: (1) refuting the Snowland school’s view; (2) refuting the Akṣapāda (Naiyāyika) school’s view.
Refuting the Snowland School
The proponent has three and three, six faults in all; the opponent likewise has six, three and three; the witness has three, and all three share one — the Snowland masters admit sixteen.
[Snowland school, i.e. the Tibetan earlier epistemologists:] the proponent has six faults — three concerning questioning (being asked and not answering; answering when not asked; stating the opponent’s fault, which is not his business), and three concerning the clearing of faults (not clearing a fault when he should; answering with similar-seeming words that have no grounds; answering what was not asked). The opponent likewise has six — three at the time of questioning (not asking at all; asking what should not be asked, untimely or irrelevant; posing meaningless or self-contradictory questions), and three at the time of stating faults (not stating the fault; stating similar-seeming faults; stating faults that do not fit the case). The witness has three: not distinguishing at all; distinguishing wrongly (shielding a relative); and distinguishing where the debate would resolve itself. The sixteenth is shared by all three: inattentiveness of mind, without which the debate cannot succeed.
An illustration of the witness’s first fault: at a monastery assembly a revered tulku who had never studied epistemology was seated as witness beside a geshe; while the geshe distinguished, the tulku only laughed. Asked afterwards why, he said there were places he could not follow, but it was wonderful to watch.
If they are points of defeat, the former [four] subsume them; if not, they are no grounds for being subdued. The witness has no point of defeat; to insist on it so strenuously is excessive.
[Sakya Paṇḍita replies:] whatever among the sixteen genuinely is a point of defeat is already subsumed in the four; whatever is not subsumed there is no fault at all and gives no grounds for subduing.
Assigning faults to the witness is simply wrong. Both parties have already conceded him: a witness is appointed only after both agree that this person may serve, so whatever he says must count. To punish him is upside-down, and would leave no one willing to serve — the witness himself would end as “the black man,” the loser. Tibetan litigation is the parallel: plaintiff and defendant jointly appoint the judge, and the judge is not punished. A Tibetan saying puts it that the judge wears goatskin while the two he judges wear tiger and leopard skin; ferocious as those two look, they must obey the man in goatskin. To insist strenuously that the witness has a point of defeat is excessive and would make the dispute unresolvable.
Refuting the Akṣapāda (Naiyāyika) School
Destruction of the thesis, shifting the thesis, contradicting the thesis, abandoning the thesis, shifting the reason, and shifting the meaning; the unintelligible and the meaningless, the incoherent, repetition, and the inopportune; deficiency, excess, and the failure to repeat, not understanding, and the inability to rejoin; admitting the [other’s] charge, and evasion, overlooking what is censurable, censuring what is not censurable, departing from the tenet, and the fallacious reason — these are what the master Akṣapāda declared to be the twenty-two points of defeat. But the glorious master Dharmakīrti said that some of them are not points of defeat, and that those which are can be reduced to two [for each party].
The verse lists the twenty-two in brief; each is examined below. Dharmakīrti’s verdict: a portion are not points of defeat at all, and those that are fall within the four (two per party).
Throughout what follows, a point of defeat means the wrong view, clash, contradiction, or error in the other’s speech that we are to block and overcome. The refutations turn on one move repeated: whatever the Akṣapāda school isolates as a separate defeat is either no fault, or else an instance of an inconclusive (or otherwise fallacious) reason — hence already “stating a non-proof” for the proponent, or “stating a non-fault” / “not stating the fault” for the opponent.
1. Destruction of the Thesis
The so-called point of defeat “destruction of the thesis” is just [a case of] stating a non-proof.
[Akṣapāda:] the proponent argues “sound is impermanent, because it is an object measured by the senses”; the opponent, admitting that the universal is both an object measured by the senses and permanent, turns the same reason round — “sound is permanent, because it is an object measured by the senses.” The later position is held to harm the earlier, destroying the thesis.
[Reply:] the counter-position raises a doubt but works no harm — permanence cannot destroy impermanence. If a doubt arises, the reason is inconclusive, since “object measured by the senses” extends to permanent things as well as impermanent; an inconclusive reason is already stating a non-proof. No separate defeat is needed.
2. Shifting the Thesis
The so-called point of defeat “shifting the thesis,” once understood, the refutation reduces it to the inconclusive.
[Akṣapāda:] the proponent, his first thesis (“sound is impermanent, because it is an object measured by the senses”) under attack, sets up a second — “just as the pot does not pervade all phenomena, so too sound does not pervade all phenomena” — and offers it in support. Stacking a second probandum on the first is a defeat.
[Reply:] what he has cited is not a second probandum but a proof: the pot example shows that “object measured by the senses” pervades only the impermanent. Arguing by a proof does not produce two theses. And if that proof is taken as the reason for the first thesis, it is simply an inconclusive reason with no bearing on the probandum — like arguing “there is fire on the mountain” from “the opposite of water is fire.” So it belongs under stating a non-proof.
3. Contradicting the Thesis
The thesis contradicting the reason and the like: if the thesis is stated, it is needless; if not stated, it becomes the three faults of the reason — where the reason has no fault, it is not a point of defeat.
[Akṣapāda:] two kinds. The thesis contradicting the example: “sound is permanent, like the pot” — a permanent thesis requires a permanent example such as space, not the impermanent pot. The thesis contradicting the reason: “all phenomena are distinct entities, because the names of real things are jointly applied” — the separate-natured and the joint are contradictory.
[Reply:] the first is an inconclusive reason, since no homogeneous example can be found — nothing in the world is both permanent and like the pot — hence stating a non-proof. The second dissolves on analysis: if all phenomena are separate-natured, the joint thing is itself included among them and so is also separate-natured; contradiction never gets its opportunity.
Further, is the contradiction spoken aloud or only intended? If spoken, the thesis is superfluous anyway — the five-membered argument’s thesis is unnecessary, since sufficient reason establishes the point unstated, and insufficient reason is not helped by stating it a hundred times. If unspoken, it falls among the three faults of the reason (unestablished, inconclusive, contradictory). Either way no separate defeat is needed.
4. Abandoning the Thesis
Abandoning the thesis belongs within the inconclusive; to posit a second point of defeat is pointless.
[Akṣapāda:] the proponent asserts “sound is impermanent, because it is an object measured by the senses”; rebutted (that reason extends to the universal and space, hence to permanent things), he cannot answer and denies ever having said it. Since he admitted it and now does not, he should be subdued.
[Reply:] his flip-flop shows only that he never gained doubt-free ascertainment of the reason, so it reduces to the inconclusive. Moreover the opponent bears part of the blame: the moment “object measured by the senses” was uttered he should have said “not necessarily” and overturned it; to begin subduing only at the second stage is too late. The everyday parallel: someone freely says a dharma-friend has broken her vows, and when the steward asks, denies having said anything — which is why witnesses matter (in the West even a marriage requires them; Einstein, marrying against his parents’ wishes, gathered witnesses for a simple ceremony so it would be acknowledged). Subdue at the outset; a second defeat category has no meaning.
5. Shifting the Reason
Shifting the reason: if the statement is complete, it is the inconclusive; if not yet finished, it is not a point of defeat.
[Akṣapāda:] the proponent holds “all particular phenomena are of one nature, because they have measure, like a clay bowl.” The opponent objects that pain and pleasure are particulars yet have no measure (measure meaning possession of a body, chiefly of subtle particles). The proponent at once qualifies: he meant only those particulars that have measure, setting pain and pleasure aside. Failing to discriminate at the outset and only discriminating when cornered is a defeat.
[Reply:] ask whether his statement was completed. If completed, the grounds are insufficient and it is an inconclusive reason — hence stating a non-proof. If not yet completed, judgment is premature and it is no defeat, since one thesis may be argued by many reasons (sound is impermanent because produced, because conditioned, because perishing moment by moment), and a differing intention may licence a differing reason.
The illustrative anecdote: a yogin held forth at length to Jigme Phuntsok about how bad ceramic bowls are for the health, then noticed the Dharma King holding an old porcelain bowl painted with the eight auspicious symbols — and immediately added that ancient ceramic bowls are very good for the body.
6. Shifting the Meaning
Shifting the meaning, for the proponent, is the inconclusive; as the opponent’s point of defeat, it is no real fault. What is to be subdued is [covered] by [stating a non-proof]; there is no point of defeat apart from this.
[Akṣapāda:] uttering words unrelated to the real meaning — same in sound, different in meaning — is a defeat. The example turns on a homophone. In Tibetan the word for “permanent” is pronounced as the word for “inference”; in Chinese, cháng as “permanent, constant” is picked up in its other sense of “long.” So “sound is permanent (cháng), because it is not an object of sight” is deliberately misheard. Comparable: “my daughter has gone forth (chūjiā, become a nun)” answered with “your daughter has married out (chūjià)? whom did she marry?”
[Reply:] if the equivocation is the proponent’s own, his proof is incomplete — stating a non-proof, an inconclusive reason. If it is the opponent’s, the fault he states is no fault but a misunderstanding — stating a non-fault. Neither requires a separate category.
7. The Meaningless
The meaningless is [a case of] stating a non-proof; there is in truth nothing else besides this.
[Akṣapāda:] speech with no meaning at all is a defeat. The Auto-commentary’s example is reasoning by the order of the alphabet: “‘ka kha ga a’ is ‘ca cha ja ya,’ because of ‘ta tha da na’”; or “you are Chinese, because you are American, like a Singaporean.”
[Reply:] for the proponent the reason stated is not in the least sufficient and no genuine grounds were given — stating a non-proof, with nothing gained by a separate heading.
8. The Unintelligible
The unintelligible — [whether] meaningless words [is included] — is subsumed within stating a non-proof. If [the proponent] states it meaningfully three times and the opponent still does not understand, that is by no means the proponent’s defeat.
[Akṣapāda:] if something is said three times and still not understood, that is the proponent’s defeat — he says “sound is impermanent because it is produced” three times over and the opponent only answers “Eh? What are you saying?”
[Reply:] distinguish. If the proponent deliberately makes himself unintelligible — using a foreign tongue or unheard-of classical diction, or gabbling so fast no one can follow (one khenpo debated at such speed that nobody could catch a word) — the fault is his, and it is stating a non-proof. But if he states it fully, in good order, each word clear — “sound is impermanent, because it is produced, like the pot” — and repeats it two or three times, then failure to understand shows the opponent is too dull; the defeat is the opponent’s, not the proponent’s.
9. The Incoherent
The incoherent — denying with irrelevant words — belongs within stating a non-proof.
[Akṣapāda:] uttering denials with no connection before and after — first “the pillar is impermanent, because it is produced,” then “the pillar is impermanent, because of space” — to make deliberate difficulties.
[Reply:] this is insufficiency of grounds, so stating a non-proof; no separate heading.
10. The Inopportune
The inopportune is just a reversal of order. If the meaning can be understood, it is not a point of defeat; if it cannot be understood, then it belongs within stating a non-proof.
[Akṣapāda:] reversing the order of the words of an argument is a defeat.
[Reply:] not necessarily. If the meaning still comes through — “the golden yellow pot” and “the yellow golden pot” both convey the intended content — the speech is correct and there is no defeat. If the meaning is lost, it is stating a non-proof.
11. Deficiency
Deficiency is the fault of [lacking] the two members; the absence of the thesis and the like is not a point of defeat.
[Akṣapāda, joined here by the Indian earlier epistemologists, chief among them Vasubandhu:] the five members — thesis, reason, example, application, conclusion — must all be present; any missing one makes the reason spurious and is a defeat.
[Reply:] only the absence of the two required members is a fault, since without them the argument’s meaning cannot be grasped. Chapter Ten gave the definition: “once the subject-property is established and the connection is fixed, that is the unmistaken definition of a reason.” Of the five members, the reason and the example must be present; thesis, application, and conclusion are superfluous (as this chapter states below). Their absence is no fault, so it cannot be a defeat.
12. Excess
Excess speech is subsumed within stating a non-proof; on your own [Akṣapāda] view it is not a point of defeat.
[Akṣapāda:] once the point is established, adding further reasons or examples is superfluous and a defeat — “sound is impermanent, because it is produced, like the pot, like the pillar” (two examples), or “because it is produced, because it is conditioned” (two reasons).
[Reply:] in our tight system a needless addition is indeed a fault, but it falls under stating a non-proof. And on the Akṣapāda school’s own terms it cannot be a fault at all, since they admit five members where none are needed; a school that insists on that much padding has no ground to complain of surplus.
13. Repetition
Repetition: verbal repetition is not a fault; repetition of meaning is subsumed within stating a non-proof. That too, set up as a fault within an argument [is admitted]; but a long narrative is no error.
[Akṣapāda:] two kinds — repetition of words (“sound is impermanent, sound is impermanent”) and repetition of meaning (“sound is impermanent, sound is a perishing thing”). Both are defeats.
[Reply:] verbal repetition needs no separate heading. Where the meaning too repeats, it is already repetition of meaning; where only the words repeat, the problem is small and repetition may even be needed — “if you’re happy, be happy; if you want to eat, eat” carries an implied further meaning, and “pay attention, please! pay attention, please!” serves emphasis. Kadampa teachings allow repetition of words and meaning for matters of importance; Jigme Phuntsok’s Drop of Heart-Advice adds a fourth verse to emphasise practice after three verses on study, contemplation and meditation, while noting that unmotivated repetition is a verbal fault. Within a formal argument, where the logic is strict, repetition does become a fault — but one already covered by the four. And expounding a long or intricate matter at length before the witness, to make the principle clear, is no error.
14. Failure to Repeat
Failure to repeat: if the opponent has not restated the proponent’s words, that is no fault; [but] if what needs to be said is not declared, it is subsumed within the inability to rejoin.
[Akṣapāda:] the opponent should echo back the proponent’s statement — “ah, you say sound is impermanent because it is produced” — before questioning. Sitting silent after three statements is a defeat.
[Reply:] we never agreed to any such teacher-and-pupil recitation, so failure to echo is not a fault. If some principle that genuinely needed stating was left unstated, that is the inability to rejoin — a term applying to both parties, reducing to not stating the fault (opponent) or not stating the proof (proponent). Repetition is used only in special cases, as when the opponent pins down a damaging admission: “this is what you said! you said the sun rises in the west; remember it, do not change!”
15. Not Understanding the Meaning
Not understanding the meaning, for the opponent, is nothing other than the inability to rejoin.
[Akṣapāda:] the proponent states his position — “the animal with a dewlap and hump is the ox” — the witness understands it, the opponent does not; that is a defeat.
[Reply:] like the previous case, this is nothing beyond not stating the fault. Whatever the cause — inability to answer, or unwillingness because he is in a poor mood and will only mutter “mm, is that so” — the opponent’s task was to state the fault and he did not.
16. The Inability to Rejoin
The inability to rejoin belongs to either not stating the proof, or not stating the fault.
[Akṣapāda:] one party, obliged to answer, instead idly sings a doha or a worldly song, paying no heed to the debate.
[Reply:] plainly unreasonable, but not a separate defeat. For the proponent it is not stating the proof (he produces no reason, only the singing); for the opponent, not stating the fault.
17. Evasion
Evasion, when genuine, is not a point of defeat; where there is cunning, it belongs to the inability to rejoin.
[Akṣapāda:] abandoning the debate by pleading business or illness is a defeat.
[Reply:] two cases. If the plea is true — danger to life, a threat to his discipline, or acute illness requiring an ambulance — it is a fact and no defeat; laying a charge on a genuinely sick man is unreasonable. If he is merely cornered by sharp arguments and feigns it — “ah, I have fallen ill, I want to sleep, I must leave at once” — it is a defeat, but one already covered: not stating the fault for the opponent, not stating the proof for the proponent.
18. Admitting the [Other’s] Charge
Admitting the other’s charge, [analysed by] knowing or not knowing, is respectively the inconclusive and not stating the fault.
[Akṣapāda:] the proponent says “you are a thief, because you are a person,” and the opponent answers only “you too” — which concedes that the proponent is a thief and implies that he himself is one. Turning his own fault back on the other is a defeat.
[Reply:] analyse by whether the opponent knows the reason is fallacious. If he knows it, then answering “if being a person makes one a thief, you too are a thief” is perfectly reasonable — and the proponent cannot accept that conclusion, which exposes “person” as a fallacious reason; the correct reply is simply “not necessarily.” If he does not know it is fallacious and treats it as genuine, then he really has admitted himself and others to be thieves, and that is a defeat — subsumed within not stating the fault.
19. Overlooking What Is Censurable
Overlooking what is censurable, for both debaters, is subsumed within the inability to rejoin.
[Akṣapāda:] the proponent uses a fallacious reason, the opponent cannot see it or overturn it and flatly gives up; that is a defeat.
[Reply:] it is the inability to rejoin — not stating the proof for the proponent, not stating the fault for the opponent. No separate heading.
20. Censuring What Is Not Censurable
Censuring what is not censurable is solely the point of defeat of the inability to rejoin.
[Akṣapāda:] fastening a fault on someone who has committed none, forcing it on him by eloquence or sheer insistence, is a defeat.
[Reply:] it amounts to one’s own inability to overturn the other’s position, so it is nothing but the inability to rejoin, subsumed within not stating the fault and not stating the proof.
21. Departing from the Tenet
Their so-called “departing from the tenet” is subsumed within not stating the proof.
[Akṣapāda:] the Sāṃkhya school holds, from the side of real nature, that what exists does not perish and what does not exist does not arise — then also holds that within the universal principle all can appear, so the non-existent arises from it and the existent dissolves back into it. Contradicting one’s own tenet is a defeat.
[Reply:] their claim is self-contradictory before and after, which is stating a non-proof; no separate heading.
22. The Fallacious Reason
The fallacious reason is a point of defeat.
[Akṣapāda:] the three fallacious reasons — unestablished, inconclusive, contradictory — are defeats.
[Reply:] entirely true, and nothing is wrong with it; but all three fall within stating a non-proof and stating a non-fault, so no separate positing is required.
The whole scheme fails on principle as well as in detail: since every person in the world speaks differently, defeats classified this way would run to ten thousand upon ten thousand, and twenty-two would never suffice. What can be subsumed should be subsumed within its kind; only what genuinely cannot be subsumed deserves its own class.
The Twenty-Four Pseudo-Refutations
Pseudo-refutations are all fallacious reasons, analysed from the three aspects of thesis, reason, and example. They are covered here only in outline; the principle is set out in the Pramāṇasamuccaya and clearly in the Nyāyamukha.
The fallacious reasons of contact and non-contact, the eternal, non-statement, and the unborn, the effect-property, the homogeneous, and the heterogeneous, the differentiating, the non-different, and the apprehensible, doubt, understanding-the-meaning, and consequence — all these are stated in the Pramāṇasamuccaya.
Fourteen pseudo-refutations, expounded by Dignāga.
The Nyāyamukha speaks of addition and subtraction, the stated and the unstated, and the reasonable, the respective examples, the probandum, the impermanent, the produced, and the production-of-fault pseudo-refutation. Thus the twenty-four that are admitted Dignāga has already refuted, and Dharmakīrti let stand.
A further ten from the Nyāyamukha, making twenty-four. Dignāga having refuted them from every angle, Dharmakīrti did not take them up again in the Pramāṇavārttika.
Sample refutation, the first of the fourteen. Against the correct argument “the sound of a conch is impermanent, because it is produced by effort, like the pot,” the opponent charges: is the relation between reason and probandum contact or non-contact? If contact, they merge like river water into sea water and become inseparable, so nothing is proven; if non-contact, they are as unrelated as sound and the eye, so no conclusion can ever be reached. The answer: refutation and proof between reason and thesis proceed by exclusion, not by inspecting contact or non-contact one by one — the dilemma has no bearing on correct reasoning.
Concluding verse:
Those who, not knowing these pseudo-refutations, expound the great treatises, mostly fail to grasp the earlier views; hence [I] explain them broadly, set apart.
Without knowing the twenty-four, expounding the Pramāṇasamuccaya, the Nyāyadvāra and the like is difficult, since the opposing views remain obscure. As these views are little used now, Sakya Paṇḍita treats them only roughly.
c. Classifying the Language of Debate
Three parts: (1) classified by person; (2) classified by need; (3) classified by argument-type.
1. Classified by Person
The three persons have six modes of speech.
Proponent: establishing his own position, and clearing away faults. Opponent: questioning, and stating the fault. Witness: rendering judgment, and analysing. Six in all, and each party keeps to his own rules rather than cutting in on the others.
The proponent states his tenet; when the opponent flings overreaching charges, he clears them — briefly. Faced with an argument he accepts, he says “admitted”; otherwise he has only three replies: “contradictory,” “unestablished,” “inconclusive.” In a properly conducted arena the proponent says little beyond these. Trading line for line back and forth means the roles have not been kept straight. The proponent generally need not state faults: if the opponent cannot rebut his established position, he has already won; if the rebuttal succeeds, he has already lost. Anyone who insists on doing the attacking should take the opponent’s seat and rebut properly.
The opponent may question first where the proponent’s words are unclear — “do you mean the object measured by the five senses, or by the mental faculty?” — and then refute. The witness records the earlier stages of a long debate so as not to forget, analyses where analysis is needed, and renders judgment; he may not argue from either party’s side.
Not every debate proceeds by formal argument; convincing by reasoning is common, as two stories show. A monk found a non-Buddhist patron smearing bezoar on his forehead and wearing a conch on his head for good fortune. If bezoar brings fortune, the monk asked, why are cows — in whose chests bezoar is always present — slaughtered and eaten? And if the conch shell is auspicious, why did the conch that lived in it day and night nonetheless die on the shore? The patron had no answer. (A caveat: from the Buddhist side, auspicious substances may well have inconceivable dependent-arisings; the monk’s argument works as skilful means rather than as final metaphysics.)
The second: four men were sent to subdue a yogin who had lived so long in a cave that birds nested in his hair and spiderwebs joined his face to the cave-mouth. Asked why he wore his wolfskin fur-side out, against the natural order, he replied that the fur faced outward on the living wolf, so his way accords with the original order and the usual way does not. Asked instead to debate the essence of Buddhism, he answered that the light of a firefly cannot measure the radiance of the sun, and the discursive mind cannot gauge the Dharma. The remaining two fared no better against his effortless practice. Argument alone, without realisation, looks fragile before such a person.
2. Classified by Need
Two parts: (1) language that establishes one’s own position; (2) language that refutes others’ positions.
Language That Establishes One’s Own Position
Two parts: (1) classifying genuine arguments; (2) classifying pseudo-arguments.
Classifying Genuine Arguments
Two parts: (1) the manner of employing language; (2) the function of the argument.
The Manner of Employing Language
Three parts: (1) refuting others’ positions; (2) establishing one’s own position; (3) dispelling disputes.
Declaring the three modes is the genuine argument.
The genuine argument is probative language possessing all three modes.
This sits within the eight topics of epistemology named at the opening of the treatise: perception and pseudo-perception, inference and pseudo-inference, refutation and pseudo-refutation, proof and pseudo-proof. Dignāga holds that all of epistemology falls within these eight. Chapter Nine treated perception (and indirectly pseudo-perception), Chapter Ten inference for oneself (and pseudo-inference); Chapter Eleven treats proof and refutation, with pseudo-proof and pseudo-refutation incidentally.
Refuting Others’ Positions — two parts: (1) the earlier school’s view; (2) the Snowland school’s view. “Earlier school” here means the Indian earlier epistemologists, Vasubandhu and Bhāvaviveka; the later Indian epistemologists are Dignāga and Dharmakīrti. Dignāga received the lineage through Vasubandhu but, with his teacher’s leave, criticised the earlier school on several points.
Refuting the Earlier School’s View
Of the five members, the thesis, the application, and the conclusion — these three are superfluous; and since the pervasion is incomplete, [the five] are untenable.
[Indian earlier epistemologists:] three members are not enough; a complete argument has five — “sound is impermanent (thesis), because it is produced (reason), like the pot (example); just as the pot is produced, so too sound is produced (application); thereby it is proven that sound is impermanent (conclusion).” (The major premise, minor premise and example of Western logic resemble this, and modern logicians likewise stress the concluding step.)
[Reply:] the scheme is at once superfluous and deficient. Superfluous in three places: the thesis fixes neither the subject-property nor the relation between reason and probandum, so before the other party it is an unestablished nature — stated, it produces no comprehension; unstated, the reason alone proves the point. The application repeats the reason (both turn on “produced”), and the conclusion repeats the thesis (both turn on “impermanent”). Deficient in one crucial place: the example is given (“like the pot”) but no pervasion is fixed upon it. Our system fixes it both ways — if produced, necessarily impermanent, like the pot; if not impermanent, necessarily not produced, like space — and only then is the question decided. The earlier school says what should not be said and leaves unsaid what must be.
Refuting the Snowland School’s View
The two members too, using the third case and the fifth case, lead to the thesis. If the thesis is not stated outright, the statement is incomplete and so one must still ask.
[Snowland masters, the Tibetan earlier epistemologists:] two members suffice, as in “sound too is produced” — “sound” being the subject-property and “too is produced” both reason and conclusion reached. They rely on Tibetan case-endings: the third case (“having-”) and the fifth, the causal or source case (“by-”, with the sense of “from”). (In Tibetan grammar the cases function as connecting particles, marking relation, action, time, analysis, cause and so on.)
[Reply:] however the third and fifth cases are deployed, the language still leads to a thesis, and if the thesis is not stated the statement is unfinished and the other must still ask. “Having-produced” — what follows? “By-produced” — after being produced, what then? Compare “having-eaten” and “by-eaten”: after eating, what does one do? Since the hearer must still ask, judgment is impossible and the claim fails on its own terms. (The argument turns on Tibetan word-order and carries over only roughly into Chinese; whole clauses that are natural in Tibetan cannot be reproduced word for word. Translators who ignore this produce nonsense, as when a Chinese speaker orders “eat tsampa” in Tibetan word-order that no Tibetan can parse. The old lotsāwa-and-paṇḍita partnership avoided this by working from both languages at once.)
Establishing One’s Own Position
Two parts: (1) the manner of proving before the dull; (2) the manner of proving before the wise.
Before the Dull
Before the dull, apply the concise and the full; the order of the two makes no difference; and add a concluding particle at the end.
“Dull” here does not mean incapable — with someone lacking basic knowledge there is no debating at all — but one who grasps the pervasion in a particular instance (“upon the pot, once ‘produced’ is established, it is necessarily impermanent”) without yet extending it to all phenomena. Before such a person prove by a two-membered argument, simple in language and full in content, so that the three modes are recalled.
Two routes. By the homogeneous relation, stating forward pervasion with subject-property: “whatever is produced is all impermanent, like the pot; sound too is produced.” By the heterogeneous relation, stating reverse pervasion with subject-property: “whatever is not impermanent is necessarily not produced, like space; and sound is produced.” Either order — pervasion first or reason first — serves equally.
At the end a concluding particle is best. Tibetan has ample completing particles and the effect is natural; the Chinese equivalent (“yea,” 也) belongs to classical usage and sounds absurd in modern speech, so Jamyang Loter Wangpo’s manual, in Chinese, omits it. The point is that a final particle marks the statement complete.
Before the Wise
Before the wise, rely on the reason alone.
Where the other already accepts that whatever is produced is necessarily impermanent, the single line “sound too is produced” suffices, and he grasps the impermanence of sound without further words; thesis and example are unnecessary. (If he has never stated the pervasion, his acceptance cannot be judged, however clear his mind may be.) The analogy: to someone expert in business who knows gold is valuable, one need say only “this ring is made of gold” — he recalls the relation on the spot, and “this ring is precious, because it is gold, like such-and-such” would be wasted breath.
Dispelling Disputes
[Objection:] The particle “and” in the Pramāṇasamuccaya, analysed and stated [your way], truly contradicts [it]. [Reply:] That is subsumed by the third case; therefore there is no contradiction at all.
[Objection:] Dignāga says in the Pramāṇasamuccaya, “wishing to make another, like oneself, gain ascertainment, one states the probandum by relying on the subject-property and the connection, setting aside the rest” — where the third-case particle “and” links the connection with the probandum, so that subject-property, connection and probandum are inseparable and all three must be present. How then can a single reason suffice before the sharp-faculty hearer?
[Reply:] the third case here means that the connection includes the probandum. Arguing “because it is produced” already contains “impermanent” within it, since a wise hearer knows that whatever is produced is impermanent; the relation is thereby indirectly stated, and the third case is fully honoured. No contradiction with Dignāga arises.
The Function of the Argument
It makes a resultant inference arise in the other.
The argument produces a resultant inference in the opponent’s continuum, and is so called because the name of the effect is posited upon the cause: the words are the cause, the genuine valid cognition arising in the hearer is the effect. Saying “the pillar is impermanent, because it is produced,” we voice what stands in our own continuum; the hearer understands, and an ascertainment accordant with the truth of things is born in him where before there was none.
Classifying Pseudo-Arguments
Two parts: (1) the definition; (2) the classification.
The Definition
Employing a mistaken proof is the definition of a pseudo-argument.
An argument with the three modes incomplete.
The Classification
Divided by meaning, by line of thought, and by wording, the pseudo-argument too is of three kinds.
Faulty in meaning: the three modes are incomplete, as in “sound is impermanent, because it is not produced.” Faulty in line of thought: the three modes are complete and the language sound, but from the standpoint of the hearer’s disposition the reason is unestablished or unpervaded, so he does not accept it (as already discussed under inference for oneself). Faulty in wording: the hearer knows the three modes, but the wording is improper, the language unclear, or the argument badly deployed. The fault in meaning was settled under inference for oneself; the other two belong here, and must be judged against the persons and the arena.
Sound debate therefore requires that the meaning be faultless, the line of thought straight, and the mind free of craft — saying what one means, from an impartial standpoint, whether stating another’s fault or clearing one’s own.
Language That Refutes Others’ Positions
Two parts: (1) genuine refutation; (2) pseudo-refutation.
Genuine Refutation
Stating the fault so as to remove wrong thinking [is refutation]. Divided by object, by essence, and by wording, genuine refutation is therefore of three classes.
Refutation is stating the fault so as to remove the other’s wrong thinking: he did not know he had the fault, and our verbal argument makes him know it or dispels the wrong conception in his continuum.
Divided by object, three: refutation stating a fault in the reason (against “the pillar is impermanent, because it is not produced”); in the definition (against defining a person as “possessing a dewlap and hump,” which is the ox’s definition); in the probandum (against “the pillar is permanent, because it is produced”). Divided by essence, three: fault in meaning (a contradiction in what is asserted, such as sound being permanent and not produced); fault in line of thought (no contradiction in meaning, but the thinking departs from the normal); fault in language (meaning and thinking sound, wording improper). Divided by wording, two: autonomous refutation and consequence refutation. Six by this count, or, matching each essence against each object, nine.
The refutation itself is never at fault; the fault lies in the refuted. The analogy: weapons for killing people, yaks, and goats differ by object, not in their killing-essence, and the classification of refutations follows the same pattern. A refutation with a fault could not refute anything.
On the two wordings: never confuse the autonomous and consequence reasons of epistemology with the Svātantrika and Prāsaṅgika of Madhyamaka — the terms coincide, the meanings do not, and arguing epistemology by “neither one nor many” would be a serious error. The autonomous reason runs “it is such-and-such, because of such-and-such” — “the pillar is impermanent, because it is produced.” The consequence runs “it would follow that the pillar is impermanent, because it is produced,” where “it would follow” carries the flavour of flinging an overreaching charge: taking what the other admits, and forcing him to admit what follows from it.
Pseudo-Refutation
The pseudo-refutation is failing to state the fault of a fault.
The other really has a fault, but one does not state it as a fault — from fear of his anger, from inability, or from fear of threats. It wears the surface of a refutation without performing the function, as a false teacher bears the name without leading anyone into wisdom.
The non-Buddhists and certain Tibetan masters, though they fix a number, are unreasonable. It is just not stating the fault and stating a non-fault; to classify them in detail is beyond conception.
[Non-Buddhists and some Tibetan earlier epistemologists:] pseudo-refutations come to a fixed number.
[Reply:] no fixed number is possible. A speaker may cite a genuine reason and yet praise himself and disparage others, or wander off, or refute by similar-seeming language; twenty-two would not cover it, nor two hundred and twenty, nor twenty-two thousand, since beings’ discursive thoughts are boundless and their speech follows. Subsume instead, as Dharmakīrti does: from the opponent’s side, every pseudo-refutation is either not stating the fault or stating a non-fault. (Refutation is the opponent’s work — the proponent need only establish his tenet and clear charges — so an opponent who does not refute is no opponent.) Subdividing within the two would run on forever: not stating the fault because one is ill, in a poor mood, busy, cheerful after rain, sad after rain, obliged to fetch water, obliged to eat — the list has no end and no use.
3. Classified by Argument-Type
Two parts: (1) autonomous arguments; (2) consequence arguments.
Autonomous Arguments
The autonomous argument was already explained earlier.
Chapter Ten expounded it: the nature-reason, the effect-reason, and the non-apprehension reason are the autonomous reason, whose definition is reasoning complete in the three modes. The distinction from the consequence reason is also treated in the Sword of Wisdom (Don rnam par nges pa’i shes rab ral gri) and in Orgyen Tulku’s Treatise Distinguishing Views and Tenets.
Consequence Arguments
Two parts: (1) positing the consequence; (2) the manner of reply.
Positing the Consequence
Three parts: (1) refuting others’ positions; (2) establishing one’s own position; (3) dispelling disputes.
Refuting Others’ Positions — two parts: (1) stating the other party’s view; (2) refuting that view.
Stating the Other Party’s View
[They say:] consequences are four, [subdivided into] fourteen — thus the Snowland masters admit. Among those that can and cannot be refuted [by reply], genuine and pseudo-consequences are seven each. Six and a half do not induce [an autonomous reason]; half of the seventh does induce one.
[Snowland masters:] the root division is four — reason and pervasion both unestablished; reason alone unestablished; pervasion alone unestablished; reason and pervasion both established.
(i) Both unestablished: before a Buddhist, “it would follow that sound is not an object of measurement, because it is permanent” — the reason (sound permanent) is unestablished, and no pervasion holds between permanence and not being an object of measurement. (ii) Reason alone unestablished: “it would follow that sound is not produced, because it is permanent.” (iii) Pervasion alone unestablished: “it would follow that sound is permanent, because it is an object of measurement.” (iv) Both established, which subdivides into four: both by valid cognition; both by admission; reason by valid cognition and pervasion by admission; reason by admission and pervasion by valid cognition. Three plus four gives seven.
Examples of the fourth set: both by valid cognition — “it would follow that sound is impermanent, because it is produced” (whether or not the other accepts it, valid cognition establishes both). Both by admission — “it would follow that sound is permanent, because it is an object measured by the eye,” which stands only before someone who admits that whatever the eye measures is permanent; no valid cognition could ever establish it. Reason by valid cognition, pervasion by admission — “it would follow that sound is permanent, because it is produced.” Reason by admission, pervasion by valid cognition — “it would follow that sound is not produced, because it is permanent.”
The force of the consequence lies here: where valid cognition establishes a member, no one dares overstep it; where the other has promised something, that promise itself becomes the reason binding him to the conclusion.
Of the fourteen, the seven that can be overturned by reply (by “reason unestablished,” “pervasion unestablished,” “both unestablished,” or “admitted”) are pseudo-consequences; the seven that cannot are genuine. Of the genuine seven, six and a half do not induce an autonomous reason when reversed, because the three modes come out incomplete; half of the seventh (reason by admission, pervasion by valid cognition) does induce one.
Reversal works by exchanging the two members and inverting the content: “it would follow that sound is permanent, because it is not produced” becomes “sound is impermanent, because it is produced.” Why only half of the seventh induces: reversed, the probandum is sound’s impermanence, and masters differ on what that is. Those (chiefly the earlier epistemologists) who take impermanence here as a universal hold universals permanent, so nothing is established; those who take it as sound’s specifically-characterised property find “sound is impermanent, because it is produced” perfectly established. Hence half and half.
Refuting That View
Because the autonomous reason would be divided likewise, because where there is no reason there is no pervasion or contradiction, and because fixing a number for pseudo-refutations is unreasonable, dividing consequences into four is untenable.
[Sakya Paṇḍita:] three reasons against the scheme.
First, parity: if the consequence divides into four on these lines, the autonomous reason should too — but it is divided only as unestablished, inconclusive, contradictory, and never into “reason and pervasion both unestablished” and the rest.
Second, where the reason is absent the pervasion cannot exist, so inspecting the pervasion is idle — like examining whether the faculties still function in a man whose life-faculty is wholly extinguished. Once the reason is unestablished, saying “unestablished” settles the matter; a case of unestablished reason with established pervasion does not occur.
Third, this style of subdivision has no floor: half the reason unestablished, half the pervasion unestablished, then halves of halves, without end. The three — unestablished, inconclusive, contradictory — already include every fallacious reason.
Dividing the unestablished into two is superfluous; [their scheme] is deficient in the contradictory pseudo-consequence; “pervasion established without a reason” is a real error — therefore consequences are not fourteen.
Their scheme is superfluous, deficient, and erroneous at once. Superfluous: splitting the unestablished into reason-unestablished and pervasion-unestablished adds nothing, since an unestablished reason leaves no pervasion to speak of. Deficient: for all the four, seven and fourteen, the contradictory reason goes unmentioned. Erroneous: “reason unestablished yet pervasion established” names a phenomenon that does not exist.
Establishing One’s Own Position
Two parts: (1) the genuine consequence reason; (2) the pseudo-consequence reason.
The Genuine Consequence Reason — two parts: (1) the definition; (2) the classification.
The Definition
Stating what is admitted so as to establish what is not admitted.
Language that takes the other’s own admitted view as the reason and thereby establishes a view he does not admit at all — as when, from his admission that sound is permanent, one forces on him a conclusion he rejects.
The Classification
Three parts: (1) the consequence reason that induces a proof; (2) the consequence reason that does not induce a proof; (3) illustrating these classes.
The Consequence Reason That Induces a Proof
The establishing consequence has the three modes; though it is not a genuine probative statement, reversed, it acquires the definition of a [proof].
The consequence has its own three modes, which the autonomous reason does not share: (1) the opponent admits the reason with reifying grasping — we take his admission, not valid cognition, as the reason; (2) forward and reverse pervasion are established by valid cognition; (3) his thesis has a contradiction with valid cognition. In “it would follow that there is no smoke on the mountain, because there is no fire,” the absence of fire is what he admits, the pervasion (no fire, no smoke) is established by valid cognition, and his thesis is thereby harmed.
Possessing these three modes does not make it a genuine argument — the genuine argument is the autonomous reason complete in the three modes, by which ascertainment arises in one’s own continuum. But reversed, the consequence does acquire that definition, becoming an autonomous reason. Its distinctive work is to take the other’s admission, draw out a contradiction inside his own words, and force him to concede that his admission was mistaken — the same manner in which the Prāsaṅgika flings charges at the proponents of real entities, unanswerable because any answer turns on the answerer.
[Objection:] some say the reversed consequence induces the very words of the establishing [argument]. [Reply:] the consequence statement induces the autonomous reason [only in part]; [taking] the subject-property [as induced] ruins the arguments. The consequence statement induces the meaning of the autonomous reason, but the wording, reversed, is not faithfully [induced].
[Some Tibetan masters:] reversing the consequence induces the establishing argument exactly as it stands, wording and all.
[Sakya Paṇḍita:] reversal induces the pervasion but not the subject-property; taking the subject-property as induced ruins the argument. The Auto-commentary treats this briefly; Nyenchung-pa’s manual is clearer. Before someone who admits there is no fire on the mountain but there is smoke, we charge: “it would follow that there is no smoke on the mountain, because there is no fire” — a genuine consequence, since the absence of fire is his own admission. Reversed, thesis and reason exchange places: “there is fire on the mountain, because there is smoke,” which is now an autonomous reason. Its pervasion is present — where there is smoke there is fire. But the subject-property is not induced, for the same reason the earlier school’s thesis was superfluous: “there is fire” is already drawn out by “there is smoke,” so taking it as a stated thesis is redundant, and a superfluous probandum cannot make a correct argument. Hence the meaning of the autonomous reason is induced; the wording is not.
The Consequence Reason That Does Not Induce a Proof
The genuine consequence, reversed, does not induce the [full] meaning.
Some genuine consequences, reversed, yield incomplete three modes and so induce no proof.
Illustrating These Classes
Know that the effect, the nature, and the definition — these three non-apprehensions do not induce. All other genuine reasons induce a proof. When reversed, four induce their own class, and fourteen induce a different class.
The autonomous reason has twenty kinds: the nature-reason and the effect-reason; the four of nature non-apprehension (nature, effect, cause, and pervader non-apprehension); the twelve apprehension-reasons of the contradiction of non-coexistence (given earlier in the verse “though the treatises posit many, the extinguishers are the former twelve”); and the two of the contradiction of mutual exclusion. Twenty consequence reasons correspond.
Three of the twenty (in fact two and a half) fail to induce an autonomous reason:
- Nature non-apprehension: “it would follow that there is no pot, because the pot is not apprehended” reverses to “the pot is apprehended, because it exists” — worthless, since once the pot is known to be apprehended, confirming it by “because it exists” adds nothing.
- Effect non-apprehension: “because there is no smoke, it would follow that there is no unobstructed-capacity cause of smoke” (the cause of smoke being called unobstructed capacity) reverses to “before me there is smoke, because the unobstructed-capacity cause of smoke is present” — impossible, since the fleshly eye cannot see the moment prior to smoke; knowing the unobstructed capacity while not knowing the smoke does not occur.
- Definition non-apprehension, which is half of pervader non-apprehension: “because it does not possess a dewlap and hump, it would follow that it is not an ox” reverses to “it possesses a dewlap and hump, because it is an ox” — backwards, since one proves the definiendum by the definition, and anyone who already knows it is an ox knows it has dewlap and hump.
The remaining eighteen do induce a proof. (Pervader non-apprehension appears on both sides of this line: its definition-non-apprehension portion fails, while the rest succeeds — “because before us there is no tree, it would follow that there is no aloeswood tree,” the tree being pervader of the aloeswood tree.)
Of these eighteen, four induce their own class and fourteen a different class. Own class, as in the nature-contradiction nature-apprehension consequence “because before us there is the cold sensation, it would follow that there is no fire,” reversed to “before us there is no cold sensation, because there is fire” — the same class throughout. Different class, as in the cause-non-apprehension consequence “because there is no fire, it would follow that there is no smoke,” reversed to “there is fire, because there is smoke” — a cause-non-apprehension consequence becoming an effect-apprehension autonomous reason. The fourteen comprise: effect apprehension; the contradiction-nature-apprehension, contradiction-effect-apprehension, and contradiction-pervaded-apprehension reasons with three classes each (nine); contradiction-pervaded apprehension, pervader non-apprehension, and the nature-reason (thirteen); plus the cause-non-apprehension example above.
The Pseudo-Consequence Reason
Stating what is admitted yet failing to establish what is not admitted.
One states the other’s view but cannot, on that basis, establish anything he does not admit. Such a reason can generally be overturned by reply: I establish a principle, the other attacks it by a consequence, and once I answer on the root principle he has no way to continue.
Dispelling Disputes
[Objection:] if the admitted is a reason, there would come to be a fourth kind of reason; if the admitted cannot serve as a reason, proving by consequence is unreasonable.
[Non-Buddhists:] since the consequence always takes the other’s admission as its reason, you have admitted an “admission-reason,” which would make a fourth kind alongside effect, nature and non-apprehension. And if the admission cannot bear being a reason, then nothing can be established by it, and your consequence arguments collapse. Either way you are stuck.
[Reply:] as a nominal imputation upon the other’s conception, proving by consequence is faultless.
An autonomous reason must indeed be established by valid cognition, and there the admission-reason is inadmissible. But the consequence is a different instrument. People do admit the pervaded while denying the pervader — admitting sandalwood while denying “tree” — and the consequence exploits exactly that, attacking the shield with its owner’s spear: someone who holds “there is no smoke on the mountain, because there is fire” is met with “because there is fire, it would follow that there is smoke,” and must admit the smoke. Relative to the autonomous reason the consequence is a nominal imputation upon the other’s own conception, and its work is to expose his self-contradiction — so no fourth kind of reason is added and no fault follows.
The Manner of Reply
Two parts: (1) refuting the untenable view; (2) stating the tenable view.
Refuting the Untenable View
The Snowland masters say thus: there are three modes of replying to a consequence.
[Snowland masters:] the proponent replies in three modes only — reason unestablished, pervasion unestablished, and admission. The contradictory is not among them.
The contradictory and the inconclusive become one; if so admitted, the fallacious reasons become two. In the end all fallacious reasons would be subsumed within the single “unestablished.” If you say the unestablished must be split in two — likewise [must the others]; or else, [collapsing] the contradictory becomes an error.
[Reply:] setting aside admission, which we also accept, two modes cannot carry the load. The contradictory reason (which involves solely the reverse pervasion, as “not produced” in “the pillar is impermanent, because it is not produced”) and the inconclusive reason (which involves both similar and dissimilar class) would both be answered by “pervasion unestablished,” collapsing into one — so the fallacious reasons reduce to two, the contradictory having been swallowed by the inconclusive. Press further and both reduce to one, since “pervasion unestablished” and “reason unestablished” are alike unestablished.
Should they answer that pervasion-unestablished and reason-unestablished must be kept apart or the debate cannot be made clear, the same demand applies one level down: the contradictory and the inconclusive within “pervasion unestablished” must equally be kept apart. Where there is contradiction, reply “contradictory”; where the reason spans both classes, “inconclusive”; where the subject-property fails, “unestablished.” Calling a reverse-pervasion contradiction merely “pervasion unestablished” misleads: answer “it would follow that sound is permanent, because it is produced” with “pervasion unestablished” — which reads as “inconclusive” — and the hearer concludes that produced things might after all be permanent, a doubt that should never have been raised.
Stating the Tenable View
Therefore the wise reply to a consequence by means of four modes of reply; and when one cannot overturn [the consequence] by reply, to follow [the opponent] is the way of the wise.
Four modes of reply (the four liberating replies): to “the pillar is impermanent, because it is produced” — “admitted”; to “because it is not produced” — “contradictory”; to “because it is an object of measurement” — “inconclusive”; to “because it is heard by the ear” — “unestablished.” The proponent has no licence to ramble beyond these. A brief qualification is allowed where a distinction is at stake: “admitted in conventional valid cognition, unestablished in the ultimate situation,” or “admitted in the relative, not admitted in the ultimate,” which guards against the other conflating the two truths.
Where the proponent genuinely cannot overturn a charge, he should follow the other’s position — as the non-Buddhist Harinanda, defeated by Sakya Paṇḍita, followed him back to Tibet. Feigning illness or urgent business, or turning to harsh words, is the opposite of the way of the wise.
Whoever understands all refutation and proof, the teaching of the treatises of reasoning — those wise ones come to uphold the true meaning of the perfect Buddha’s teaching.
Whoever grasps that unreasonable positions are to be refuted and reasonable ones established has comprehended the stainless teaching of the treatises of reasoning. Two accounts illustrate this. While Mipham Rinpoche was composing his Great Commentary on the Pramāṇavārttika, he beheld Sakya Paṇḍita in a luminous dream, who told him there is nothing in epistemology that is not understood — it is merely included within refutation and proof — and handed him a scripture-volume that turned into a sword in his hand; the whole of learning appeared as a single book before him, and he cut it in two at a stroke. Jigme Phuntsok, finding the Pramāṇayuktanidhi especially deep when he first studied it at Sershul, prayed to the guru and the Three Jewels, and likewise beheld Sakya Paṇḍita in a luminous dream, who blessed him with the Mañjuśrīnāmasaṃgīti; from then on he fully comprehended epistemology and the other sūtra teachings.
The Conclusion of the Treatise
Three parts: (1) the manner of composing the treatise; (2) dedicating this virtue to Buddhahood; (3) stating the author’s name to inspire faith in the treatise.
The Manner of Composing the Treatise
Five parts: (1) stating the author’s surpassing wisdom; (2) stating on what basis it was composed; (3) stating the real necessity for composing it; (4) stating that one should therefore rely on the wise; (5) determining by this reason.
Stating the Author’s Surpassing Wisdom
In former lives, relying again and again [on teachers], he studied deeply under accomplished masters; in this life, hearing but once or twice, Kunga Gyaltsen knows all the treatises.
Sakya Paṇḍita relied through many past lives on masters proficient in sūtra and tantra, studying the epistemological treatises and their commentaries deeply; biographies say he had been a paṇḍita for twenty-five lives, or twenty-seven. Hence in this life he comprehended all the treatises on hearing them once or twice, without effort. The verse is openly self-praise, in the manner of Longchenpa and Khedrubje, and unlike Mipham Rinpoche, who called himself an ordinary being and a fool.
Sakya Paṇḍita is the universally acknowledged emanation of Mañjuśrī in Tibet — Patrul Rinpoche names three great Mañjuśrīs of the Snowland: Longchenpa, Tsongkhapa, and Sakya Paṇḍita. His biography records that a party who went to Wutai Shan to behold Mañjuśrī all dreamed the same night that Mañjuśrī was not there but dwelt in Tibet at Langthang Kar in the form of a paṇḍita, and should be sought there; Sakya Paṇḍita was then at Langthang Kar.
Stating on What Basis It Was Composed
The all-victorious masters Dignāga and Dharmakīrti, the Pramāṇasamuccaya, the Pramāṇavārttika, and their commentaries — having fully understood the principles of epistemology [therein], he composed this Treasury of Reasoning on Valid Cognition.
The basis is the Pramāṇasamuccaya, the Seven Treatises, and their commentaries. “Their commentaries” is generally taken as the auto-commentaries of Dignāga and Dharmakīrti alone: Sakya Paṇḍita did not draw on the Indian manuals of Śākyabuddhi, Devendrabuddhi and the rest, holding that he had comprehended Dharmakīrti’s ultimate intent and that those manuals added nothing. The Auto-commentary on the Pramāṇayuktanidhi cites the scriptural authority of no Indian master besides those two, and cites the Tibetan earlier epistemologists only to refute them.
Jigme Phuntsok held that the vast wording of the Seven Treatises and the Pramāṇasamuccaya is all contained in meaning within this treatise — which is why it is expounded yearly, why all schools of Tibet study it, and why no Tibetan master of any school, Nyingma, Sakya or Gelug, has failed to praise it. Content that takes days to work through in the Pramāṇavārttika a single verse here can settle; without it, the key points of the other treatises are hard to grasp.
Stating the Real Necessity for Composing the Treatise
To crush the Kapila [Sāṃkhya] and the master Akṣapāda, the Naked [Jain] masters and the Owl-son [Vaiśeṣika], the Hedonist (Lokāyata) theorists who hold “beauty of this world,” and the Hearers and the Snowland masters — their base and mistaken speculations: for this he composed this great treatise.
Six targets: (1) the Kapila or Sāṃkhya school with its twenty-five principles (already treated in the Madhyamakālaṃkāra); (2) the Akṣapāda school, the Brahminical or Vedic school; (3) the Owl-son, the Vaiśeṣika, who hold the six categories; (4) the Naked non-Buddhists, the Jains, who hold space to be the finest clothing, so that no other is needed; (5) the Lokāyata “beauty of this world,” valuing only the enjoyments of the present life; (6) some views of the Hearer school within Buddhism, and a portion of the Tibetan earlier epistemologists’ views.
On the last: the earlier epistemologist Shawa Chökyi Sengge had eight disciples named “Sengge,” of whom Tsang Tsöndrü Sengge and Den Mawai Sengge were foremost — the “Sha, Tsang, Den” of Tibetan Buddhism. Sakya Paṇḍita refuted a portion of their views, such as the claim treated in Chapter Four that exclusion and the distinguisher exist in the essence of the external object.
Stating That One Should Therefore Rely on the Wise
Though possessing slight intelligence, not gaining the essence of good teaching; day and night diligent in the treatises, yet knowing them only partially, not ultimately; ever exerting in meditation, yet straying from the path that delights the Buddha — let the people satisfied with [such] in this degenerate age reflect well and rely on the wise.
Three kinds of person must take care and drop their pride. Those with some worldly education who have not gained the heart-essence of the teaching, treating the Dharma as an ornament that never enters the heart. Those diligent day and night who have only half-understood the treatises, having heard them once or twice without reaching their ultimate meaning. And those who exert constantly in meditation but abide in a merely worldly concentration, unheld by renunciation, bodhicitta or non-dual wisdom, so that their practice strays from the path that delights the Buddha and the guru. The danger is contentment — “I practise day and night, I keep reading the treatises, I am learned” — while the actual understanding and practice remain poor. The remedy is to rely on a genuine spiritual guide; study, contemplation and meditation without one cannot succeed.
Determining by This Reason
Discarding the [pretence] that “the reasoning of the seven treatises is just thus” — that hackneyed refrain — I state this principle of the Dharma. Though the wise who know reasoning delight in this, even hearing it many times, it is not the domain of the dull.
Sakya Paṇḍita states his method. He discards the common practice, in India and Tibet alike, of invoking Dignāga and Dharmakīrti as founders in order to present one’s own position as the secret meaning of the seven treatises — a hackneyed move by which people establish their own school under borrowed authority. He composed instead from his innate wisdom together with the wisdom gained by study, deciding by reasoning; so the treatise is not a repetition of others’ views but a further disclosure of the two masters’ intent and a refutation of later error. (The contemporary parallel drawn is plagiarised and reassembled compilations issued under new names — a first manual being hard to produce, a second and third easy, needing only another’s outline plus a little analysis of one’s own.)
The wise have loved the treatise ever since; a truly wise person, not necessarily delighted by a hundred wish-fulfilling jewels, is delighted by this, carrying it wherever he goes or else keeping it in the treasury of his own wisdom, the words memorised and the meaning held in the heart. Those with neither innate wisdom nor present diligence may hear it many times without understanding — one lama at the institute heard it seven or eight times, tutorials included, and understood nothing, though his perseverance was commendable — which is what the last line prophesies.
Dedicating This Virtue to Buddhahood
Having opened the eye of wisdom through good teaching, and rightly shown the true suchness of all that is knowable, by this virtue may [all] gain the wisdom that sees all phenomena, and become a refuge for every sentient being.
The author dedicates: by the Pramāṇayuktanidhi opening the eye of wisdom in fortunate hearers, and by the virtue of setting forth the Buddha’s subtle truth, may all beings to the ends of space illumine the truth of all phenomena, attain perfect enlightenment, and become protectors of all beings.
Stating the Author’s Name to Inspire Faith in the Treatise
This Pramāṇayuktanidhi was composed at Sakya Monastery by the Śākya monk Kunga Gyaltsen Pal Zangpo — born in the Snowland north of India, omniscient in grammar, logic, prosody, poetics, synonymics, and the conventions of language; possessed of the talents of exposition, debate, and composition; who unerringly comprehended the Pramāṇasamuccaya and the seven treatises on epistemology; who gained the radiance of wisdom in scripture, reasoning, and pith instruction — setting aside ornate diction, and expounding in a plain and easily understood manner.
(The great five sciences are logic, grammar, medicine, arts and crafts, and inner Buddhist science; the minor five are prosody, poetics, rhetoric, synonymics, and astrological calculation with drama — counted separately, prosody drops out. All worldly learning falls within the two sets.)
Here ends the Eleventh Chapter, Examining Inference for Others, of the Pramāṇayuktanidhi.