Chapter 3: Inference for Others

Inference for others is the statement of a reason complete in the three modes, and it is called an inference only because the cause takes the name of its effect: the words are not themselves a cognition, but they produce one. The chapter first determines the probandum, which must be one’s own nature still to be established, admitted as one wishes within one’s own system, and un-negated by perception, inference, scripture or common renown. It then turns to the means of proof: the property of the position, and the reason proper.

Condensed from Fazun’s Chinese translation-and-exposition of the Pramāṇasamuccaya and its auto-commentary, and rendered into English. Root verses are given as Fazun gives them; the prose beneath summarises his exposition. On the provenance and the limits of this rendering, see the series introduction.

Chapters: 1. Perception · 2. Inference for Oneself · 3. Inference for Others · 4. Example and Pseudo-example · 5. Apoha · 6. Sham Rebuttals

---

Chapter on Inference-for-Others

Two main divisions: (1) the direct statement, and (2) the analysis.

1. The direct statement

Inference for the sake of others clearly reveals to [another] the object that one has oneself seen.

Just as one oneself, by a reason complete in the three characteristics, gives rise to the cognition of a thing together with its reason, so, to make another give rise to that cognition, one states a reason complete in the three characteristics. This statement is called “inference-for-others” because the cause is given the name of its effect — the words are not themselves a cognition, but they produce one. If even one of the three characteristics goes unstated, the statement is incomplete.

2. The analysis

Two parts: (i) determining the probandum, and (ii) determining the means of proof. The first divides into stating one’s own position and refuting rival views.

(一) Determining the probandum

(1) Stating one’s own position

Regarding the statement of the probandum (sādhya) therein: [the probandum] is accepted as the object [referent] of the [cognition that is the] activity of the reason — it is the probandum, not the means of proof.

[The probandum] possesses [the proponent’s] own nature [the thesis-property still to be established, not one already established for both parties]; it is as one oneself wishes [admitted within one’s own system], and is what is [explicitly] wished [to be proven]. And it is un-negated — by perceptible fact, by inference, by trusted scripture, and by common renown — with respect to its own subject (the dharma-bearer).

[Question] What is the probandum of inference-for-others? Reply: the object that is the meaning of the reason. Among the proof-members the thesis is not admitted as a means of proof, since it gives rise to doubt rather than settling it; but because it is the reason’s meaning and displays that object, it must still be established.

Three qualifications. “Only its own nature” — the nature still to be established, not one already established for both parties. This alone frees the thesis from the not-fully-established reason and the fallacious example, neither of which is stated as a nature that ought to be established. “As one wishes” — relative to what one’s own system admits, without regard to others’ systems; and even within one’s own system one posits selectively, not everything one holds. “Un-negated” — the dharma-bearer, as qualified by the property one wishes to establish, must not be contradicted by perception, by inference, by one’s own scripture, or by common consent. Otherwise the thesis is fallacious: “sound is inaudible” (contradicted by perception), “a pot is eternal” (by inference), “a valid cognition cannot establish the object of cognition” (self-contradiction). Where the term is so uncommon that no inference is available, common usage still negates: “the hare-bearer is not the moon” fails, because “hare-bearer” is a settled name for the moon.

These examples state only the negation of the own-nature of the property; the same corner shows that the differences of property and dharma-bearer are negated too. Thus the whole does not exist apart from its parts, since apart from “low and high” there is no separate designation; and the part does not exist apart from the whole, since otherwise it should not be perceptible. [The example of a substance that would be the non-substance of qualities is marked translation-uncertain.]

The classic case of both own-natures being negated is “all verbal meanings are false.” If the speaker’s own utterance falls under “all,” then both the speech-nature and the falsity-nature are removed at once. If instead he exempts his own statement, then “all other speech” is the difference of the dharma-bearer and “the falseness of that speech” the difference of the property — and since his own statement would then be false, it conversely shows other speech to be not-false, contradicting both differences. (Fazun flags this passage as still requiring study.)

(2) Refuting rival views

Three parts: against the Naiyāyikas, against the Vādavidhi, and again against the Naiyāyikas.

A. Against the Naiyāyikas

[Naiyāyika:] As for this [statement of ours], “[the thesis] is what displays the probandum” — by [that phrase’s] sense, [the thesis] is made merely [a presenter of the probandum], not [itself] a means of proof.

[Reply:] If so, then an unestablished reason and example would also become the probandum, since they too present [something] as a probandum.

If the [bare predicate-]property is what is to be proven, there is no similar-property [example]; if the [bare] property-possessor (dharma-bearer) is the probandum, the reason adduced for it is purposeless.

[If the qualifier “only” is] placed before [— “only what displays the probandum is the thesis” —] the determinate specifier is purposeless; if [placed] the other way [— “what displays the probandum is only the thesis” —] it is not what is wished.

[Naiyāyika] The thesis is what displays the probandum, i.e. displays what is not yet established. Reply: then the not-fully-established reason and example must also count as probanda, since they too present something not yet established. “Sound is eternal, because it is tangible, like a cognition” — the reason states something to be established, so on the Naiyāyika account it should be a thesis; likewise “because grasped by the eye, impermanent.”

Press further: is the property the probandum, the dharma-bearer, or both? If the bare property (impermanence as such), there is no similar-property example available for it. If the bare dharma-bearer, the reason is purposeless, since “sound” is already established for both parties and needs no proof. If both, both faults follow together.

Press again on the placement of the specifier “only.” Put in front — “only what displays the probandum is the thesis” — the word is idle, since “displays” already excludes the means of proof that draw out the reason. Put after — “what displays the probandum is only the thesis” — then what is stated for the sake of reason and example becomes thesis too, which is not what the Naiyāyika wishes.

B. Against the Vādavidhi (by parallel argument)

Likewise, to [the Vādavidhi’s] stating [the probandum as the thesis]: [the Vādavidhi claims] “[the fault] cannot arise, because of [the qualifier] ‘thesis’ (pakṣa).” [Reply:] this [thesis, glossed as] “the object under analytical consideration” merely reverses “what is not wished.”

The faults just stated against the Naiyāyikas carry over to the Vādavidhi, since on its account the reason-to-be-established and the fallacious example would likewise become theses. [Vādavidhi] The fault does not arise, because we do not say bare “probandum” but “thesis-probandum,” and the word “thesis” adds the sense “the wished, under consideration.” (Fazun notes the commentary reads: “by adding [the word] ‘thesis,’ it does not arise.“)

Reply: “the wished under consideration” does no more than reverse “the unwished,” and mere reversal of the unwished cannot mark off the probandum from anything else. “Eye-grasped” can equally be put under consideration as to whether it is a reason or not — so it too would have to be called a thesis. With this the thesis and the fallacious thesis are settled.

C. Again against the Naiyāyikas

[The claim] that “because the thesis contradicts the reason, it is a fault of the thesis” is not reasonable. Those poorly trained [in framing proofs] state that [reason] by way of a [botched] dissimilar-property example.

This [permanence of sound] is not [validly] gauged as the probandum by means of its reason, nor is it [a reason genuinely] possessing contradiction.

[Naiyāyika] “Sound is eternal, because everything is impermanent” shows a thesis contradicting its own reason, and this is a thesis-fault. Reply: both halves of the claim fail. Those not well-trained state this formula by way of the dissimilar-example device: the sense intended is that, since nothing is eternal, the dissimilar example displays the absence of the probandum and hence the absence of the reason-property.

Nor does the reason gauge the probandum here. “Sound” is itself included within “everything,” so “not-all-ness” cannot serve as the reason — it is unestablished, being a part of the thesis. And if sound counts as a part and so as “not-all,” so does every other item; “not-all” then holds universally, which is not what the opponent wants. If it is objected that impermanence-ness negates eternity-ness and so this is a thesis-fault, the reply is that it does not negate sound’s existence: “sound exists” is granted, so the formula is a dissimilar-example inference. Taking “apart from sound, the rest is everything” fails the same way — a part-of-the-thesis reason, like “impermanent, because impermanent,” or “all is non-existent.”

[Objection] But “all aggregates necessarily serve another, because they are aggregates,” and “the conditioned is impermanent, because it is conditioned,” plainly take a part of the thesis-meaning as reason. Reply: those do not take all aggregate-dharmas or all conditioned-dharmas as the thesis — that would leave no example — but the eye and the like, a part of them. [Objection] Then “sound” should become uncommon. Reply: it has no capacity to. “Audible-ness” is reasonably uncommon, since upon sound there is no further sound; the present case is not like that. So the defect is a fault of the reason (and of the example), not a fault of the thesis.

The Vādavidhi subsumes the case under the contradictory fallacious reason, but it answers to neither of its two contradictions — contradiction of the thesis-meaning and contradiction of accepted doctrine. Both of those will be shown not to be genuine contradictions either.

(二) Determining the means of proof

Two parts: clarifying the pakṣadharma, and properly clarifying the reason.

(1) Clarifying the pakṣadharma

A. Direct exposition of the pakṣadharma

Whatever appears as a reason for the probandum-property — since most [such fallacious appearances turn on the pakṣadharma] — therefore, before [explaining] the meaning of “reason” and the rest, the pakṣadharma (the property of the position) is to be set out at length.

With respect to the similar class [the reason is] present or absent [or both]; thus the pakṣadharma has three kinds. And with respect to the dissimilar class too, each [of these] is [further divided] by the distinction of present, absent, or both — [hence nine].

Because it is the [conjoined] establishing [of both] that is the probandum, the property and the property-possessor taken singly — though not [the probandum] directly — are, since each is one part of it, spoken of as “probandum” by nominal designation.

Because [the pakṣadharma] is so termed from [being] a property established for both [parties]: [a property] established invertedly for both or for either one, or a doubtful property [whose establishment is in question], is not accepted [as a pakṣadharma].

[Reasons] not accepted [as a pakṣadharma] are, all of them, unfit even as a refutation. [But] when fully established for both [parties], [a reason] serves both to refute and to prove; and the rest are accepted [as such only] when re-established by another [reasoning]. [The Tibetan reads “unfit even as a refutation”; the Chinese 不許皆能破 reads the opposite — against all of these there is refutation.]

A property-possessor (dharma-bearer) is not established by a [bare] property-possessor; nor does that [property-possessor] establish a property. Rather, a property establishes a property, and because [it] establishes [thus], the property-possessor too is likewise [established — as qualified by that property].

Since reasons and fallacious reasons mostly turn on the pakṣadharma, it is set out first. It divides nine ways: present, absent or both in the similar class, each crossed with present, absent or both in the dissimilar class.

[Objection] The dharma-bearer is already established for both parties, so it is not what one wishes to prove — how then can there be a “property of the position”? Reply: no fault. The probandum proper is the conjoined whole; property and dharma-bearer taken singly are its parts, and are called “probandum” only nominally, as one says “the fire burns the garment.” And the designation “pakṣadharma” is posited precisely because the property is established for both proponent and opponent.

Four ways the pakṣadharma fails: both-parties-unestablished (兩俱不成) — “sound is impermanent, because it is grasped by the eye”; one-party-unestablished (隨一不成) — “because produced,” urged against the Śabda-manifestation theorist (聲顯論), for whom sound is manifested and not produced; doubt-unestablished (猶豫不成) — proving “there is fire” from a place where one doubts whether what one sees is smoke; locus-unestablished (āśrayāsiddha, which Xin-hui renders “locus-unestablished”) — “because pleasure and the like exist everywhere, the self exists pervadingly,” where the subject itself is not established.

None of these is admitted as a pakṣadharma. Where a property is fully established for both as a non-pakṣadharma — “eye-grasped-ness” — it functions as a refutation; where it can be established as a pakṣadharma — “because it differs by difference of condition” — it functions as a proof. The remainder are settled only on re-establishment: established as pakṣadharma, they prove; established as non-pakṣadharma, they refute. Fazun notes the agreement with the Nyāyamukha (理門論): “Only commonly-accepted, decisive words are called proof or refutation; not mutually-unestablished, doubtful words, which still await establishment.”

Khenchen Kunga Wangchuk’s annotation commentary (ཚད་མ་ཀུན་བཏུས་ཀྱི་མཆན་འགྲེལ་ཚིག་དོན་གསལ་བྱེད་) resolves the divergence recorded in the verse above in favour of the Tibetan. It reads: དེ་ལྟ་བུའི་ཕྱོགས་ཆོས་མ་གྲུབ་པས་སྒྲུབ་བྱེད་དུ་མི་འདོད་པ་ཐམས་ཅད་སུན་འབྱིན་དུའང་མི་རུང་ངོ་། — “all those not accepted as proofs, their property of the position being unestablished, are unfit even as a refutation.” It continues that a reason thoroughly established as the property of the position for both parties is fit both to refute and to prove, and that those failing through a defect of mind — the doubtful and the rest — are admissible if some further reasoning establishes them. The four failures of the pakṣadharma are given there in the same order and with the same examples. The Chinese 不許皆能破 is the divergent witness.

On what proves what: [Objection] “proving fire by smoke” is a dharma-bearer proving a dharma-bearer. Reply: no. Nobody doubts that fire exists; what is proved is “fire here.” If what the smoke belongs to were the probandum, a part of the thesis-meaning would be doing duty as reason; and if seeing smoke in the locus already meant seeing fire, there would be nothing left to infer. So one must say “in this place” — the place is the dharma-bearer, and property and dharma-bearer are not here set up on the model of quality and quality-possessor.

[Objection] “Proving heat by fire” is a dharma-bearer proving a property. Reply: the same “in this” applies. Even granting “in all smoke one sees fire,” if fire’s heat-differentia is the probandum then examining the place is idle, and one is again taking part of the thesis as reason: fire’s difference would be the probandum while fire as universal is the reason. To hold that, one would have to admit the fire qualified by the place as probandum — so why favour the fire-difference and shun the place-difference?

[Sāṃkhya] “Pradhāna has unity, because individual things are seen to have positive concomitance” (Xuanzang: “because they have a common kind”). This posits, as thesis, “individual things definitely have one cause,” with “a clod and the like definitely have one cause” as example — that is, it posits another property as the thesis. The verse’s “the property-possessor too is likewise” shows that what is proved is the dharma-bearer as qualified by that other property. Reason-property and probandum-property stand in the relation “without the one, the other does not arise.”

B. Resolving a latent objection

[Objection:] “By a statement of thesis and reason, a consequence is flung at what [the opponent] does not accept [— so the proof works even without an established pakṣadharma].” Know that this is [merely] “drawing out [a consequence]” (prasaṅga), for one applies [it] through that [opponent’s own] gateway.

The reason is stated as following the probandum [positive concomitance]; and “where [the probandum] is absent, [the reason] is absent” is [also] to be stated. Therein, the example — bearing the fifth-case [ablative] ending — is to be applied together with the reason.

When [an inference] is stated as a “dismiss-and-return” (avīta) about some [locus], it turns out reason-less; for [a valid reason] is established [only] from the two examples [similar and dissimilar]. Therefore [a valid reason] is none other than the “possessing[-the-modes]” (vīta) [reason].

Because a consequence (prasaṅga) is not [based on a genuine] pakṣadharma, [it operates] from what [the opponent] has previously accepted. Likewise, the stating of a fault by [means of] the reason and thesis is to be known as a refutation [not a probative proof].

Formulae such as “sound is non-eternal, because otherwise action and the rest should be eternal, and because the eternal should be perceptible” are not proofs but consequences drawn through the opponent’s own gateway. Since he first posits “eternal, since formless,” the reply merely posits a counter-thesis to expose the fault in his reason.

The pakṣadharma is itself the reason-nature. In “sound is impermanent, because the produced is not eternal, and because the eternal is not produced,” there is no explicit statement of the pakṣadharma but there is the example-characteristic, and the pakṣadharma is displayed by the conjunction — “sound is produced, not non-produced.” If the counter-refutation (avīta) were stated as something separate, it would turn out reason-less, since it does not arise from refutation and, lacking the reason, would come out other than intended.

Against the Sāṃkhya use of avīta: if difference could be got from “having non-existence as antecedent,” positive concomitance would become unobtainable and would be a wrong superimposition. Since one plainly observes positive concomitance and does not observe mere difference, what is observed is “having unity as antecedent” — one universal cause. Deny that and there is no positive concomitance at all. Hence avīta is not a separate kind of reason: per the Tibetan, apart from the vīta (correspondence) reason there is no other.

Further faults follow for the Sāṃkhya and the Vaiśeṣika if such consequences were probative: pradhāna would be non-existent; enjoyer and enjoyed would collapse into quality and quality-possessor; substance would be non-existent, qualities would become substance, and substance non-substance. But these are not counter-refutations, “because [the reason] would not be a pakṣadharma, and because the probandum is prior” — they are refutations by exposure of fault, drawn through the gateway of reason and thesis.

Illustrating with the Sāṃkhya’s own case: the reason “one effect,” adduced to establish pradhāna via the concomitance of the guṇas, is indeterminate if admitted (the seen effects such as joy would then become the enjoyer), and if not admitted the root reason fails, since what is not enjoyed is not a quality. (Fazun flags this passage as still requiring study.) The parallel Vaiśeṣika consequence: if correspondence with substance-ness makes something substance, the qualities admitted to be non-substance become substance, since the inherence of correspondence is one and pervades all — one would have to admit substance is just correspondence, or correspondence just substance, and non-substance dharmas would have substance. That is a thesis-fault, substance being refuted by non-substance. The opponent cannot escape by calling the reason indeterminate, since no further substance-ness corresponding to substance-ness is ever seen.

(2) Properly clarifying the reason

Two parts: establishing one’s own position, and refuting rival views.

A. Establishing one’s own position

Having made [the point] “[they are] non-different — just that [property],” one shows [it] to another. “Similar class” refers to [a class sharing] the [probandum-]property; here too one does not claim [the property] to be different [though sound and pot differ as things, one does not claim the impermanence on each to be different].

The two [forms of] dissimilar class — “other than [the probandum-property]” and “contradictory to it” — in both, since the reason is absent [from them], there is no contradiction [in establishing the probandum]; [the probandum] is arrived at through [the reason’s] exclusion [from the dissimilar class]. [The Tibetan reads these two as the valid forms of the dissimilar class; the Chinese reads them as two refuted views.]

[A reason that is] “not absent in the similar class” — though it is a single characteristic — reasonably operates over many objects [impermanent, selfless, suffering]; its determinately running through the universal is not [a fault of] confusion.

“Object of valid cognition,” “produced,” “impermanent”; “produced,” “audible,” “effort-born”; “impermanent,” “effort-born,” “non-tangible” — these are the nine [reasons], with [theses such as] “permanent” and so forth. [This is the wheel of reasons, hetucakra.]

Those two that are present in the similar class and absent where [the probandum] is absent [the dissimilar class] are the [valid] reasons. The reverse of these are the contradictory [reasons]; all the others are inconclusive.

[A valid reason] is wished to be stated as a single number, whereas regarding the contradictory [pair] there is doubt [the “contradiction-determination,” viruddhāvyabhicārin]; likewise, from two doubt[-producing] reasons no single determinate [conclusion] is seen.

Those possessing two characteristics [doubt and determination] are admitted as “two-by-two,” not as singletons. [Yet] when [reasons] are many but not contradictory, even very [many] apply to a single [conclusion].

Those properties that are “uncommon,” or “common,” or “non-errantly contradictory” [contradiction-determined] are, in every case, reasons that engender doubt toward that [thesis].

For whatever possesses the property of the position [pakṣadharma]: [if it] harms [the thesis] by reversing what one wishes to know [a contradictory reason], or [if it] becomes doubt [an inconclusive reason] — apart from these, there is no [other] fallacious reason.

The property and the property-possessor, whether [in] their own-nature or [in] their difference — because [the reason] establishes [the opposite] invertedly, harming that [probandum], it is a contradictory [reason].

[There are] the common and the particular [difference-]property, and the probandum [taken] in two [ways] — [hence two groups]. Of these, the first two are doubt[-reasons]; the remainder, in dependence on the probandum, divide into three [valid, contradictory, doubtful].

Because [the first two], by [variously] operating or not operating [as pervaders in both classes], fail to establish negative pervasion and positive concomitance — [Objection:] “should not [a reason] that definitely operates toward one [class] and is not [definitely] reversed become a [valid] reason?”

[Reply:] Becoming both [permanent and impermanent] is impossible; nor can an existent be neither. By [its] exclusion from wherever [the probandum] is absent, [the reason] establishes the meaning to be proven.

That one [reason operates over one class]; the others are established over two. Were [a reason] to establish [positively] just as it operates, then from [the example of] white form, quality, and support, following the reverse sequence one would [have to] cognise three, two, one [all at once].

Because [the reason] is shown in [terms of] “one[-ness],” [a reason operating contrary to] the reversal of that probandum is the contradictory; but [a reason] that [merely positively] operates wherever the contradictory [class is found] is confused about its [the probandum’s] particulars.

Similar class. Whatever shares the universal of the probandum-property is the similar class. Sound is impermanent, and pots too are impermanent, so pots are the similar class of sound’s impermanence; and the reason “produced-ness” runs through both. “Non-different — just that” marks extreme similarity, not numerical identity: the reason-property on the pot is just the reason-property on sound. “Present in the similar class” means the reason is definitely there — not that it is there throughout, since a reason may be narrower than its thesis.

[Objection] If “non-different, just that” is what makes something a similar class, then this itself should count as “presence in the similar class,” and the difference between them goes unstated. Reply: there is a difference — the impermanence on the pot is not the present probandum — but one does not claim a difference here, so “present in the similar class” is entirely in order.

Dissimilar class. Two rival accounts: “the remainder belonging to the dharma-bearer” (as in “non-merit”), and “merely the negation of presence, i.e. absence.” If the dissimilar class were the mere remainder of the similar class, no reason would work: “produced-ness,” present in the similar class, is also present in the rest, e.g. in suffering. What fixes the dissimilar class is exclusion — where the property is absent, one understands the probandum to be absent. The illustration and its limit: from “there is fire-warmth here” one knows there is no cold-touch; but from “there is no warm-touch” one cannot conclude cold, since a neutral place leaves it undetermined.

So the dissimilar class is only the absence of the similar class, and the valid reason is the one “not absent in the similar class” — that is, present in it. “Produced-ness” establishes impermanence and equally establishes selflessness, being present in both. [Objection] It is a universal, so it should be doubtful. Reply: not every universal is a doubtful reason. “Pot-ness” is produced-ness, and produced-ness is seen on non-pots too; but “produced-ness” is never seen anywhere impermanence is absent. Bearing the reason-characteristic in every place, it is determinate. “Following the universal is not confusion.”

The nine pakṣadharmas (hetucakra). Three states in the similar class, each crossed with three in the dissimilar class:

• similar-present, dissimilar-present

• similar-absent, dissimilar-present

• similar-both, dissimilar-present

• similar-present, dissimilar-absent

• similar-absent, dissimilar-absent

• similar-both, dissimilar-absent

• similar-present, dissimilar-both

• similar-absent, dissimilar-both

• similar-both, dissimilar-both

The middle items of the top and bottom rows are the valid reasons; the middle of the left and right columns are the contradictory; the remaining five are inconclusive. Set out as nine cells (columns for the three similar-class states, rows for the three dissimilar-class states), the standard formulae are: (1) “sound is eternal, because an object of cognition”; (2) “impermanent, because produced”; (3) “effort-born (勤勇無間所發), because impermanent”; (4) “eternal, because produced”; (5) “eternal, because audible”; (6) “eternal, because effort-born”; (7) “not effort-born, because impermanent”; (8) “impermanent, because effort-born”; (9) “eternal, because non-tangible.” The summary verse lists them: “object-of-cognition, produced, impermanent; produced-ness, audible, effort-born; impermanent, effort-born, non-tangible — in dependence on [theses such as] eternal etc., the nine.”

If a thesis such as “impermanence” has no dissimilar class at all — against a system that posits no ether and the like — the objection “how can the reason be said to be absent there?” is answered: if no such locus exists, the reason does not operate there, no doubt arises, and there is no fault.

Valid, contradictory, inconclusive. Two of the nine are valid: present throughout the similar class and absent throughout the dissimilar (#2), and present-or-absent in the similar class with absent throughout the dissimilar (#8) — the middles of the first and last triads. Two are contradictory, establishing the inverse: present or both in the dissimilar class with absent throughout the similar — the first and last of the second triad. There is also a further contradictory that establishes something harming the probandum, as in “all distinct things are of aggregate-nature and can serve another, because they are aggregates, hence necessarily serve another”; since such a reason can establish many meanings it still falls within the two contradictories and does not escape the nine. The remaining five are inconclusive, having the sense of doubtful reasons.

“A single number.” Reasons are stated as one number, of one kind — as with “present in the similar class.” Doubt over a contradictory pair is still counted as one: upon sound, “produced-ness” and “audible-ness” together leave one in doubt whether sound is eternal or impermanent. (Xuanzang renders the line “do not say that two mutually-contradictory characteristics, gathered in one place, are still reasons,” taking it as a refutation of the Vaiśeṣika “one substance-number”; the Tibetan rather explains why two numbers are not stated.) Likewise, “not eye-grasped” and “being a manifest thing” together settle that sound is neither substance nor action.

[Objection] If they are stated as one, doubt-reasons and determinate reasons cease to be two different things. Reply: “those possessing two characteristics are admitted as two-by-two, not as singletons” — possessing both doubt and determination is not a single characteristic. The doubt-characteristic here is contradiction-determination (相違決定, viruddhāvyabhicārin).

Many reasons, many theses. [Objection] If each single reason is non-contradictory and so a valid reason — “impermanent because produced,” “impermanent because effort-born” — how do many reasons yield one determination? Reply: “many, if not contradictory, especially operating toward one.” Many reasons doing one and the same job are spoken of as one, like many people saying “on this matter we agree.” (Fazun notes: this is the converse of the earlier point that one reason — produced-ness — can establish many theses.) [Objection] Then two reasons doing several jobs, “impermanent because produced; selfless because suffering,” would be doubtful like contradiction-determination. Reply: no — each of those reasons can do many jobs, and operating over more than one thing is entirely in order. Where distinct reasons establish several probanda, they work by mutual dependence, running together in concomitance as one reason or combining to make up the reason-characteristic; both are proofs, so no fault.

The uncommon reason. [Objection] Since “both” occurs, there ought to be four kinds of inconclusive; what of “audible”? Reply: it is uncommon (不共, asādhāraṇa). An uncommon reason makes every difference of the probandum-property doubtful, since only that subject has that nature and the reason is one-sidedly excluded from everything else. Common reasons, having no excluder, are likewise doubtful, being non-contradictory with respect to both classes. Where the “both” portion is present, the reason is determinate, since it excludes the rest — that is the difference. Against an opponent who admits an eternal “sound-ness,” “audible-ness” could pass as a reason only if nothing displayed “produced-ness” as a reason for impermanence; since a mutually-contradictory meaning is available, it remains a doubtful reason. In such cases the force of perception and scripture is decisive, and determination is sought there.

The contradictory reason. [Question] What is the contradictory reason that harms what is wished? Reply: a reason of the opposite. It is a bad objection to argue that, because “audible-ness” establishes eternity, “produced-ness” should be unable to establish impermanence: that treats a reason co-operating with an eternity-establishing reason as one with it. (Fazun flags this passage as requiring study.)

Fazun’s note distinguishes the two carefully: contradiction-determination grasps two different reasons establishing two theses and can establish neither, so it is a doubtful reason; the property-difference contradictory reason (法差別相違因, dharmaviśeṣaviparītasādhana) uses one reason to establish two opposite theses and definitely establishes the opposite of what was wished.

A reason is contradictory when, by inverted establishment, it harms the own-nature or the difference of property or dharma-bearer. Examples: “because effort-born, it is not sound”; “non-audible-ness.” The showcase of both being contradicted at once is “all speech is false, because it is expressive, like expressive speech in a dream” — the speaker’s own utterance and the nature he states are both false, and he disparages the general meaning; that is contradiction of own-nature. If he exempts himself, it is contradiction of the difference. (Xuanzang and Xin-hui render the last two lines “this becomes a contradictory reason, if there is nothing it harms”; the sense is that the reason harms the wished thesis and is therefore named contradictory.)

A latent objection is then raised and dismissed: “not internal-aggregate-ness, therefore not serving another,” taking aggregate-ness as corresponding to the inner nature of the eye and the rest. Its reason-characteristic fails, the similar class of the two not being established; it establishes the contradictory class “used by all others,” and so falls under the previous contradictory.

The threefold division of reason-cognisers. Cognisers of the probandum-property are of only three kinds: (1) common, (2) particular (uncommon), (3) those of both parts. The nine divisions derive from these three. “The first two are doubt”: the common and the particular each give rise one-sidedly to doubt — “sound is eternal / impermanent, because an object of cognition,” or “because audible.” “The rest, in dependence, divide into three”: those of both parts split, in dependence on the probandum, into valid, wrong and doubtful reasons.

Why? Because the first two fail both positive concomitance and reversal, and so can neither exclude one class nor establish one. Neither has left aside a part: the common has not left aside presence in the dissimilar class, the uncommon has not left aside absence in the similar class. (Xin-hui’s “each lacks one characteristic” is easier.) [Objection] Then should not a reason that has positive concomitance and reversal toward one class become valid? Reply: if such existed, yes — but “becoming both is impossible; nor can an existent be neither.” Nothing is both eternal and impermanent with respect to one meaning, and nothing whatever is neither.

Exclusion, not affirmation. [Objection] How can a “both-parts” reason, operating over contradictory meanings, cognise anything? Reply: it works through the gateway of exclusion, not affirmation. Whatever it lacks is excluded, and through that gateway it operates over all or one of the probandum-class and is a reason of true cognition — “sound is impermanent, because produced,” “because effort-born.” Analogy: from the fine fragrance of the blue lotus one establishes fragrance-ness, quality-ness, supporting-ness, non-quality-ness and the rest, since without the fragrance these are not seen; the later terms are wider than the earlier, so absence of the later entails absence of the earlier, and the earlier establishes the later. Were the reason to work by positive establishment instead, then from the white-colour quality-locus one would have to reverse the sequence and cognise three, two, one at once.

Where one-ness is not established, one has contradiction-determination and a doubtful reason: for one who admits an eternal universal (an eternal sound-ness), sound characterised both by audible-ness and by produced-ness leaves eternity or impermanence undecided. But produced-ness and effort-born-ness are one and the same reason — as, by an upright form and a bird perching on it taken together, one knows there is a tree.

The reverse-of-the-probandum. A “both-parts” reason ought to operate with two characteristics; once one is excluded it operates on the rest for the probandum-meaning, so establishing the opposite makes it contradictory. Illustration: “because there is fire here, there is cold-touch.” Fire merely negates cold-touch and does not positively establish warmth; yet it establishes non-cold-touch, and by establishing the contradictory class it counts as contradictory — while not being contradictory in the strict sense.

[Objection] If produced-ness is non-different across effort-born and non-effort-born, then, there being no reversal anywhere, the cognition of impermanence should not have to await the negation of “eternal.” Reply: “if taken as following positive operation, one is confused about the difference.” Produced-ness runs concomitantly over both effort-born and non-effort-born, so with respect to sound both are doubtful: “because produced, effort-born” and “because produced, not effort-born” are equally positable; likewise effort-born-ness yields both “a quality” and “not a quality.” [Objection] Then “impermanent and selfless, because produced” should be doubtful too, both occurring. Reply: that would follow only if produced-ness were found on selflessness apart from impermanence. The difference lies in mutual exclusion — effort-born and non-effort-born are mutually exclusive, impermanence and selflessness are not. So produced-ness is seen at the loci of impermanence and selflessness and nowhere they are absent, and is in general determinate; only where the alternatives are mutually exclusive is it doubtful.

There is no separate “impermanent” over and above refuting “eternal”: the probandum-cognition operates by reversing the whole dissimilar class. Yet the reason need not be present throughout the whole similar class, nor can it be apprehended as present everywhere — even valid reasons cannot abandon their own connected class and have a reason-cognition arise elsewhere.

Reasons taken as one. For the sound of verbal expression, produced-ness plainly shows a cognition of its own origination, and, as with bodily expressive action, one knows it to be effort-born. [Objection] Then should the first two (common and uncommon) also be apprehenders of a single function? Reply: no — they need another property shown before determination arises, whereas “both-parts” reasons exclude the rest by themselves and, where doubt remains, become determinate by adding the other property. The stump example: seeing the upright thing and a bird perched on top, one knows it is a tree-stump; a single mark could not decide it. So against the eternal-sound theorist, “own-origination” and “produced-ness” together prove effort-born-ness, and are taken as one.

Even so, doubt can arise from what a reason operates over: “this sound, because produced — is it effort-born, like pots, or not effort-born, like lightning?” One does not add “or like ether,” because produced-ness has already excluded the eternity of ether; with respect to impermanence the reason is therefore not doubtful. Reasons that have already excluded certain matters are not re-established upon them by those matters. The same holds for all affirmations.

Finally, negation: saying “this is not eternal” is not simply saying “it is impermanent” — otherwise one would have to state the pot’s own name and the rest. In “sound is eternal, because non-tangible,” the “not” is a negation resting on doubt, whereas the earlier case rests on contradiction. Where permanent abiding is not admitted, one refutes “everlasting abiding” relative to what the opponent posits: the thing exists now but does not abide at the prior and later limits, so calling it impermanent is faultless. And a reason like “non-resistant” in “eternal, because non-resistant” is of the third kind — doubtful in dependence on ether and action, not in dependence on atoms and pots — and so, relative to this probandum, it excludes the other property “resistant”: “sound, because non-resistant — has it function or not?” It reverses neither “eternal” nor “impermanent.” In dependence on the probandum’s “both parts,” such a reason can be called valid, inverted or doubtful. With this the reason and the fallacious reason are stated.

B. Refuting rival views

Two parts: (i) refuting their account of the reason, and (ii) refuting their account of the fallacious reason. The first divides into four: against the Vādavidhi, the Naiyāyikas, the Vaiśeṣikas and the Sāṃkhyas.

(i) Against their account of the reason

A. Against the Vādavidhi

[The Vādavidhi holds:] “the reason is the property [of the subject shown to be] absent wherever the similar [class] is absent.” Formulating [it] thus — “[the reason] is precisely that which is absent where the similar [class] is absent” — the uncommon [reason] would [wrongly] become a [valid reason for] “permanent.” [If instead] one determines [it] from absence [in the dissimilar class] alone, then only an equally-pervading [reason] would become a reason.

Nor is “present only in the similar [class]” admissible, for in a single [reason] one sees many aspects [it is the “without-which-not” of many probanda]. [Objection (Vādavidhi):] “[the reason] shows [the probandum] by its own nature.” [Reply:] merely stating the reason has not the power [to display the probandum’s invariable concomitance].

[Objection (Vādavidhi):] “[the reason states its characteristic] according to the real connection [of reason and property].” [Reply:] [yet you] have not stated “by which [aspect], with respect to which [probandum].” Like [something] that has transcended the means of cognition, [such a reason] is not seen as a [real] establisher of the meaning.

“Eye-grasped” and the like, [said] of sound, would [absurdly] follow as a reason for [its] impermanence. By saying “[it is] not shown [to be fallacious],” [the Vādavidhi] has [still] not stated “in what way” and “with respect to what.”

[Objection (Vādavidhi):] “[merely] co-occurring [with the property in the continuum], [the reason] does not display the [invariable concomitance] for the probandum.” [Reply:] not so — [it] comes [to be known] from the two examples; once impermanence and the like are [already] seen [from “produced”], would [you] even admit some other probandum?

The Vādavidhi defines the reason by absence in the dissimilar class. Dignāga tests three placements of the word “only,” and each fails.

1. “Merely absent in the dissimilar class” — then “eternal, because audible” becomes valid, since audible-ness is not seen on pots and the like, which lack eternity.
2. “Only absence in the dissimilar class” — this escapes the audible-ness fault, but now only equally-pervading reasons qualify: produced-ness could make impermanence understood, while effort-born-ness could not, since impermanence is seen even without it.
3. “Present only in the similar class” — this escapes both, but leaves nothing except the probandum’s similar class “without which the reason does not arise”; yet a single reason is the “without-which-not” of many probanda, as produced-ness is for impermanence, selflessness and the rest.

Next, the account of “similar-class-co-operating.” Does the reason state its characteristic as displayed, or as the real meaning? If as displayed, “because effort-born” cannot display the probandum’s “without-which-not” — stating the reason has no such power. If as the real meaning, the Vādavidhi still has not said by what characteristic and with respect to what; if the reason displays by itself, none of this is seen. A reason not held by its own characteristic “transcends the cognizer” and cannot make the object known. Worse, if merely holding the substance-nature suffices to denote the characteristic, then “eye-grasped-ness” — eye-grasped sound being itself impermanent — would co-operate with impermanence and count as a reason.

[Vādavidhi] What is not displayed simply fails to be a reason. Reply: by what, and how, is it “not stated”? And for what thing does it fail to be a reason? Neither is explained. [Vādavidhi] Through the gateway of “absent in the dissimilar, without-which-not,” the reason has not displayed the probandum. Reply: not so — it is known from the two examples, similar-property and dissimilar-property, which gather the external matter; as it is said, what displays the determinate relation is the example. And if produced-ness already displays sound’s impermanence, what further probandum would one want an example to establish? The claim about “what is not stated” is unreasonable.

B. Against the Naiyāyikas

If the similar property [example] is the means of proof, then [the verbal] members [thesis, reason] would not be [reasons, since they operate separately], nor [is it established] by the fifth[-case]. [If you say] “the statement [establishes it],” it would [have to] be made specific. Since [the statement] is unfit to be the means of proof, both [the direct and the remote establisher] would turn out to be something else.

That [hearing-cognition] is not a similar-property reason [for it arises from the statement, not the example]; nor is [it] free of the two faults [direct and remote]. Even for the “timely [thesis]” there is no other [option]. Because [the members] harm [one another] by their respective characteristics, even a conceptual positing [of one of them as the reason] is impossible.

Therefore, [resorting] even to the sixth-case [genitive — “the reason of that property”], the specification too would be purposeless. [Naiyāyika:] “Not every statement [is admitted as a reason].” [Reply:] Then make the [particular statement] you do admit [as a reason] specific.

Without a purpose, it is not admissible for [the meaning] to go beyond the literal sense of the word. For a mere enumeration [of members] taken in general, there can be no [special] exception [singling out the one valid reason]. And [reasons merely] empty of the appearance [of fallacy] would, being alike to the other [valid] property, [wrongly] turn out to be [valid] reasons.

[Naiyāyika] The example stated from the similar property is itself the reason that proves the probandum. The refutation proceeds point by point.

If the similar-property example is the means of proof, the verbal members (thesis, reason) cannot be reasons, since they operate separately. Pleading that simultaneous statement makes them operate together, “like an army,” fails: thesis and reason do not state the same meaning. Nor can the fifth-case (ablative “from”) be got: since the members are stated separately, the property is not a reason “from that,” and no qualifier–qualified relation is seen.

If instead the statement made from the similar-property example is the reason, it must be specified — for cognition, desire and the rest all arise from that statement too. Refusing the specification leaves the means of proof “unreasonable” (the commentary: “non-existent”), since cognition and desire are not displayed in the statement of what establishes the probandum, and so are not establishers.

Then the dilemma of proximate and remote. If the establisher is proximate, only the listener’s notion is the reason; if remote, the cognition of hearing the similar property also becomes a reason. Pleading that the hearing-cognition is not a reason concedes the point: it arises from the statement, not from the similar-property example, so it is no similar-property reason. The same dilemma recurs for the similar-property example-statement itself: proximate, and the similar-property cognition becomes a reason; remote, and the listener’s cognition stated at that time becomes one too. Appeal to a “timely thesis” does not help — the timely thesis, displayed from the similar-property example, would also become a reason; and if there is some “other” not so displayed, the later members stated in dependence on the example become reasons too. The last escape, that the later members do not become reasons because the stated characteristics mutually harm one another, destroys even a nominal positing: if any one of the obtained members has such a harm, the conceptually-included members go the same way and doubt results. (Xin-hui’s “those included by conception” still requires study.)

Resort to the sixth-case (genitive) — “the rest is a characteristic of that property” — cuts off the doubt but is not faultless: with the genitive in place, “the establisher of the probandum” becomes a redundant word. [Naiyāyika] It is said for specification, since not every similar-property statement is a reason. Reply: then which do you admit? No differentiating meaning has been given. [Naiyāyika] Since there is no difference in the sound itself, the specification is made to convey the meaning, as thesis and reason may turn out contradictory. Reply: without a purpose it is inadmissible for the meaning to exceed the literal sense of the word — so specify the similar property, not the word’s meaning. [Naiyāyika] “The establisher of the probandum” is itself the specification. Reply: that is not a specification but “only a different gateway,” since all members establish the probandum.

[Naiyāyika] Establishing the dissimilar property in reverse is what establishes the probandum; what cannot establish it are the fallacious reasons, and similar properties are not fallacious reasons, hence are valid. Reply: “in the universal there is no such particular.” All confused similar-and-dissimilar-properties are fallacious reasons and cannot become valid: there is no moment in the particular sphere of becoming the universal, and one cannot first list something as particular and afterwards regard it as universal. And if being merely “empty of fallacy” made a reason valid, then in “sound is eternal, because audible, like sound-ness” the example and similar property would have to be free of all fallacious reasons — as would the pot dissimilar-property in that formula, and the dissimilar property in “sound is eternal, because effort-born, like lightning.” Others who add the earlier specification fare no better: they too must use the genitive to cut off doubt and must hold that what is merely empty of fallacy is valid. The reason-characteristic they state is unreasonable.

C. Against the Vaiśeṣikas

[If “the property possessing the probandum” means the property possessing the probandum-universal,] the universal itself would absurdly become the [reason-]property. [If it means the bare] property or property-possessor — these are not the probandum [being already established]. [If it means] the property possessing the particular probandum, [that] would be obtained [even] for another’s inconclusive [reason]. Moreover, the proof establishing “space and the like exist” could not [even] be set up [for the subject is not established, so its property-possession cannot be known].

[Vaiśeṣika] “The property in conjunction with that is the reason.” Since “that” plainly refers to the stated probandum, the claim is that only the probandum is what the reason is conjoined with. Does “that” mean the aggregate, the property, or the dharma-bearer?

• The aggregate: then the universal-nature becomes the reason-property. In “sound is impermanent,” the universal impermanence-property would have to be stated as the reason, there being no other property of the universal — as in “it is a horse because it has horns,” where horned-ness, the universal’s property, becomes the reason.
• The property: then its produced-ness and the rest do not exist, there being no further produced-ness upon impermanence.
• The dharma-bearer: already fully established, so nothing to prove.

Nor is impermanence-ness alone the probandum: “the sound of impermanence-ness,” “sound’s impermanence-ness,” “sound is impermanent” are all results of discrimination, so only the aggregate-meaning is the probandum. Property and dharma-bearer singly “are not the thesis,” having the faults just given.

If the Vaiśeṣika retreats to “only sound’s quality belongs to the probandum,” then, since neither universal nor dharma-bearer has another conjoined property, he must admit only the impermanence-property as probandum — and then sound’s common and uncommon reasons alike become “properties conjoined with the probandum,” so “the uncommon too gets established as a property conjoined with the thesis,” and “sound is eternal, because audible” becomes a reason. Specifying “only the property conjoined with the probandum” is purposeless, since there was no doubt about its becoming a dharma-bearer. And if only impermanence is the probandum, then properties other than sound’s are conjoined with it too: the property of pots, and “eye-grasped-ness,” would have to be stated as reasons.

“Furthermore, the inconclusive” — errant reasons also qualify. “Eternal, because non-resistant” would count as conjoined with that, since in no class is there a definite connection. The Vaiśeṣika would then have to say “conjoined with that, similar-property, and non-errant” — where “similar-property” sets aside the other property and audible-ness, and “non-errant” sets aside errant properties. But the earlier faults still stand. And in proofs such as “space and the like exist,” no thesis-conjoined or probandum-conjoined property is fully established, since the dharma-bearer is not; the proof-formula has no connection at all, and the reason fails outright.

D. Against the Sāṃkhyas — the intermediate verses

[Since pradhāna is a partless unity,] the probandum-property too would become the means of proof. [The reason] is not a [valid] verbose statement, nor is [a concise one] stated. The primal substance (pradhāna) does not possess positive concomitance [on the example, tree-bark and the like]. The statement, too, cannot measure [pradhāna]. [The reason “single cause”] contradicts the thesis-meaning. Since there is no dispute [that effects have causes], mere existence [of pradhāna] is not [the probandum]. [The individual modifications] differ, like tree-bark and so forth.

[Objection (Sāṃkhya):] “positive concomitance is established throughout?” [Reply:] [Even granting] one possessing a single-kind concomitance — otherwise [too] there is but one effect; and the material cause likewise turns out so. [The effects, joy and the like,] are not external, since both [alternatives — taking them as different or non-different from sound] have faults. Even if admitted, [the reason] is contradictory. [It] would also become inconclusive. The example, lacking the probandum — that [example] too does not exist. [Counter-refutation] is not established by [mere] distinction [from another’s position].

The means of obstruction [refutation] are not [only] two [example-contradiction and admission-contradiction]; the two [avīta] formulas, too, are unreasonable. [To take] “the last” as the reversal [the opposite] — to eliminate the one by it is unreasonable. Nor, like the vīta, is [the avīta] possessed of five [members]. By refuting the opponents one by one — if at the last one does not [directly] assert one’s own position, then, since the opponents’ answers are endless, [everything] would absurdly collapse into “dismiss-and-returns.” These are interstitial verses.

[Sāṃkhya] Inference-for-others divides into “correspondence” (相應, vīta) and “counter-refutation” (反破, avīta), and the object-domains of the vīta statement are fivefold by the distinction of thesis and the rest. (Fazun notes his setting-out is fuller than the root text.) Both kinds are to be tested against the object of valid cognition; and not all five members are reason-members.

Take their standard proof: “Pradhāna exists, because individual things are seen to have a common kind” (Tibetan: “because they have positive concomitance”). If pradhāna’s existence is the probandum, the proof fails at once — pradhāna “is not known as an object of valid cognition,” and, not being a universal-object, admits of no inference, as chapter 2 established. Further faults follow in series:

• Probandum-property = establisher. The probandum-property is of the same kind as the establisher, with no difference; otherwise “because it can burn” should equally establish sound’s impermanence.
• Over-broad statement. An extremely broad statement is no establishing reason, the establishing reason being brief. And they have not said that the brief statement is the reason — holding all members to be establishers, they leave the difference unstated.
• The example. Tree-bark and the like do not possess a single-common-kind pradhāna; and even if they did, the undisplayed has no positive concomitance, so the example is fallacious. The statement of the example is spoiled too, since it cannot display pradhāna and positive concomitance upon tree-bark; and there is no unifying measure by which “as the example, so the probandum is established.”

The Nyāyamukha requires a connection, and would posit as thesis “individual things definitely have one cause.” That posits its own faults. It contradicts the thesis-meaning: if all individual things have one cause, tree-bark and other distinct things, having many causes, contradict the thesis; deny that, and the example vanishes. Nor can “having existence as antecedent” be the probandum: “because there is no dispute” that effects have causes, mere existence is not at issue.

The reason also carries the standard faults:

• Unestablished. There is no single kind and positive concomitance among individual things. [Sāṃkhya] But they have one and the same effect. Reply: one also sees things with a one-kind concomitance whose remainder likewise have but one effect — and things with only one effect even without one-kind concomitance. If sattva and the other guṇas are said to be seen producing one effect by concomitance, this is indeterminate: sounds and the like are not different from the effects, and the world also shows the eye-cognitions and so forth. Since these are effects of a direct cause, one should infer the kind by that, not by the domain of sound. And joy and the other effects, seen after the cognitions, would prove the three guṇas upon the puruṣa as readily as upon external objects. Taking joy as an effect of external objects such as sound fails either way: if the effects arising with the ear are grasped as non-different, it wrecks the doctrine that each sense attaches to its own object; if as different, it wrecks the inference to a single common kind. Sound is in any case only nominally posited by aggregation, as action is nominally posited upon the aspects. Specifying makes matters worse (it harms the puruṣa’s manifold own-nature); not specifying leaves no distinct counter-refutation, and universal and particular alike collapse into counter-refutations.
• Contradictory. Even granting concomitance in a single common kind, one does not see many antecedents possessing a single common kind; trees achieve no single one, and they are not the antecedents of tree-bark, since trees have their own direct cause.
• Inconclusive. Even granting trees as the antecedent of tree-bark, tree-bark, potsherds, pottery, ornaments and the rest would all have to possess the pleasure-kind and many-property antecedents.
• Example lacking the probandum. On the thesis “individual things have a prior single cause,” one should not admit that all have pradhāna as antecedent but should say tree-bark has trees as antecedent. That example does not exist, tree-bark being itself within the individual things. The remaining two members (application 合 and conclusion 結) fail likewise.

Counter-refutation (avīta). Refuting each rival position one by one and then upholding one’s own is called avīta. (Per the Tibetan: better named “refute-and-return” or “dismiss-and-return” — dismissing others’ positions, then returning to establish one’s own.) The Sāṃkhya gives two means of refuting: example-contradiction and admission-contradiction.

First, mere refutation of another’s thesis does not establish one’s own. Saying “sound is eternal, because non-resistant” is inconclusive — it wrecks the opponent’s thesis without establishing “sound is impermanent,” since that too is inconclusive. The river analogy: when a river is in flood, one rules out “snow-mountain melting” and the other causes and concludes “rainfall” — but not by refutation alone; it is known by the flood’s own difference. Otherwise, ruling out “rainfall” and the rest would equally establish “thunderbolt” as the cause.

Second, the means of refutation are not two only: displaying the fault of an incomplete member also obstructs. Third, the two avīta formulas fail. Their positive-concomitance avīta runs: “if it arises from non-existence, then, there being no place of arising, things should turn out different” — meaning that things whose place-of-arising has not gone first should be obtainable as non-different. (Fazun corrects the text twice: the case never admitted is “no-place-of-arising going first with things each different,” and the objection concerns not stating a one-kind concomitance.) Not stating difference does not make a one-kind concomitance follow immediately.

Nor does the argument negate a one-sided determinate unity: they have not stated it, and its reverse is equally obtainable; where the place-of-arising has not gone first one sees no one-sidedly determinate unity, so one gets a “universal-particular” (common-uncommon) reason, with the many concomitances and counter-refutations coming out no different. But reversing the probandum is reasonable: non-different dharmas have unity as antecedent, being one-sidedly determinate, and their opposite would turn out different — hence “the last one is the opposite.” Things connected as one kind are plainly seen to have unity as antecedent, not non-existence, and not Īśvara. Fazun notes that Xin-hui reads the last point the other way (“there is no difference” from the correspondence-reason).

Finally, the Sāṃkhya claim that avīta too has five members has no connection with the “last” characteristic by which it was defined. And if what survives the one-by-one refutation of every opponent is admitted as “the final one,” then, opponents’ answers being endless, everything collapses into counter-refutation. So the avīta inference-formula has no coherent characteristic — and this refutes not only the Sāṃkhya reason but the vīta/avīta division itself.

(ii) Against their account of the fallacious reason

Three parts: against the Vādavidhi, the Naiyāyikas and the Vaiśeṣikas.

A. Against the Vādavidhi

[Examining the fallacious reason, from the Vādavidhi:] The divisions of “unestablished” are not stated; the [divisions of] the errant [inconclusive] are also not stated. There is also the “uncommon” [inconclusive]; and there is the non-errant contradictory [contradiction-determination] too. [Of the Vādavidhi’s two examples of “contradiction”:] one [the Vaiśeṣika’s] is [really] a “common-similarity” [inconclusive, seen in both classes], the other [the Sāṃkhya’s] is “unestablished and uncommon.” [It] states neither the ground of [genuine] reversal nor [the reason] that harms what is wished.

[Vādavidhi] The fallacious reason is threefold: unestablished, inconclusive, contradictory — given by examples only, with no characteristics stated. “Impermanent, because eye-grasped” is unestablished; “eternal, because non-resistant” is inconclusive; and for contradiction, two borrowed cases: the Vaiśeṣika “impermanent, because arising from the sense-faculty” (Xin-hui: “because grasped by the sense-faculty”), and the Sāṃkhya “there is effect in the cause, because what exists then arises.”

Reply: the divisions of “unestablished” — one-party-unestablished and the rest — go unstated, as do the divisions of the inconclusive; the uncommon and contradiction-determination are missing, though on Dignāga’s own account there are six kinds of inconclusive. Both alleged contradictions are misclassified. The Vaiśeṣika case is “common-similarity” (i.e. common-inconclusive): “arising from the sense-faculty” is seen in both the eternal and the non-eternal, so it is inconclusive, not contradictory. The Sāṃkhya case is unestablished and uncommon: “what has already arisen, arising again, is not really existent” establishes that the effect-dharma exists within the already-arisen, not that it arises from the already-arisen; and even granting it, it is only-there, hence uncommon, and it establishes no opposite meaning, so it is not contradictory. Having stated neither the ground of genuine reversal nor the reason that harms what is wished, their account is deficient — and what cannot be subsumed under their categories cannot be pronounced non-fallacious either.

B. Against the Naiyāyikas

[Against the Naiyāyikas:] their “having errancy” and “cut-off” [categories] are [just] the inconclusive; there is no [further] distinction.

[Naiyāyika:] “a [reason] contradicting [their] accepted tenet is the contradictory.” To that — “the other contradiction” is likewise not [a separate thesis-fault, but a fault of the establisher].

[Naiyāyika:] “[a reason] listed in order to settle whatever timely [matter] is at hand is ‘time-alike’.” To that — [reply:] it becomes inconclusive [for it is seen in both permanent and impermanent].

[Naiyāyika:] “because [it] proves [something] no different from the probandum, it is ‘probandum-alike’.” To that — [reply:] [it is] not unestablished [for the reason “non-tangible” is established on sound].

[Naiyāyika:] “stating [the reason] belatedly is ‘over-due’ (past-time).” To that — [reply:] [it is] either an incomplete statement [lacking the reason-member], or [a later statement that] does establish the probandum [— so in neither case is it a fallacious reason].

The Naiyāyikas list five fallacious reasons: having-errancy, contradictory, time-alike (時相同), probandum-alike (所立相同), and over-due (過時). Each is taken in turn.

Having errancy is simply the inconclusive, with no further distinction drawn. Errant from what? Neither “excluding the other” nor “timely” supplies an answer. If the universal-and-particular of sound is said to be the quality of both, and the objection is that the universal is not sound because it blocks a determination, the same blocking applies to the probandum-kind. The reason operates on the probandum’s absence, its remainder, or its contrary, and one understands it by the meaning; it is not errant in every class.

Contradicting one’s own accepted tenet. The Vaiśeṣika’s “not a producer, because non-resistant, like the atoms” is a fault of the establisher, and it incurs “the other contradiction” as well, not merely being a fallacious reason: “because non-resistant” is unestablished (a reason-fault), while “not a producer” contradicts the prior admission (a thesis-fault). The objection that a formula cannot suddenly bear both unestablished-ness and contradiction, and that this disparages the author, does not hold: even so, if the fault is displayed by another property’s contradiction it should be called unestablished, and the thesis-fault is “no other” — setting aside the unestablished meaning of one’s own thesis, no separate thesis-contradicting fault remains. The Sāṃkhya’s “pradhāna is not a cause, because it has no thought” is likewise: since nothing other than pradhāna is left to eliminate, the reason is uncommon, hence inconclusive; and even taking “apart from individual things is the other,” it is only the establisher’s contradiction, since it establishes the opposite, not a contradiction of one’s own thesis.

Time-alike. “The self is eternal, because not different from the body” (Fazun: per Xin-hui, read “because different from the body”). Here the self is not examined by that mark at all, and the reason is simply inconclusive: things eternal like ether and things impermanent like pots both show it, so the question “eternal or impermanent?” remains open.

Probandum-alike. “Sound is eternal, because non-tangible, like a cognition.” Since “non-tangible” is fully established on sound, this is not an unestablished reason; and what makes it “no different from the probandum”? Here it is the example that is alike to the probandum. One might construe “no different from the probandum” as meaning “no different from the probandum-example,” but not all such construals can be had, and even granting it the reason still fails; it is a general statement rather than a case of an “example no different from the probandum.” (Fazun notes the several lines above seem corrupt, and that the text should read: it is not merely an “unestablished” fallacious reason.) Genuine unestablishment is otherwise: a reason-property unestablished on the dharma-bearer, one-party-unestablished between proponent and opponent, or doubt-unestablished.

Over-due. “Sound is impermanent, like a pot” — asked for the reason, one adds “because produced.” This is either an incomplete statement or a later statement made for determination, and neither is a fallacious reason. If incomplete, the reason-member is simply missing, and what is non-existent cannot be fallacious. If stated later for determination, the reason-characteristic is sound and does establish the probandum. Being present in the proper reason-locus is not the mark of a fallacious reason.

C. Against the Vaiśeṣikas

According to the Vaiśeṣikas [who posit three fallacious reasons — unestablished, undisplayed, and doubtful-non-existent]: [their] “non-existent” [reason] turns out [to be] either unestablished or contradictory; and [their account] states the reversal [the “uncommon”] incompletely.

[Vaiśeṣika] The fallacious reason is threefold: the unestablished, the undisplayed, and the doubtful-non-existent.

Their “unestablished” is not really their own category: it means either absence on the example or a not-fully-established meaning, and is said to be for the sake of the reason-meaning (“or” being a word of examination), so the reason is already fully established on the example. If instead it means the probandum’s being unestablished, then calling it a matter of the reason-meaning is a contradiction. As for “it should display ‘fully-established’ as antecedent” — at the time of connection it is already well established. And “non-existent” means absence, not doubt: “horned” is already established, so one states the example upon it — “if it has horns, then it is a horse or a cow.”

[Vaiśeṣika] The “non-existent” reason is just non-existence, and a two-meaning reason becomes doubt. Reply: a reason that is merely non-existent cannot make anything known; what it in fact does is establish the opposite. By “horned,” the cognition “not-a-horse” arises rather than the intended one, so the reason is of contradictory nature. A reason with doubt should be called doubtful. And their account “states the reversal incompletely”: “a cow, because horned” is not merely doubtful but uncommon — as with “the six categories are audible-ness,” which is absent in all the contrasted classes and doubtful over many domains. Saying it is a universal because it is seen in both distinct properties, “like the upright thing,” is unreasonable: it appears so only because there is nothing it reverses. Being one substance it raises no doubt in itself, yet with respect to earth and the rest it is doubtful, since all of them reverse from it. So one cannot say every other doubtful reason is uncommon. (Fazun flags these meanings as still requiring study.)

Many further contradictory and inconclusive reasons should be listed. “Is the body produced by the combination of earth and the rest, or by things’ respective kinds?” yields doubt, since five causes are obtainable and production by a different kind is refuted — so their “doubt-sūtra” should be called incomplete. And in general, perception decides: from the white colour of the utpala one definitely sees quality-ness, supporting-ness, non-quality-ness and the rest. That is not doubt, so non-decisive reasons should not be classed wholesale as doubtful reasons within perception; they are merely deficient. Finally, if the “non-existent” reason is absence and the “doubtful” reason is doubt, whose absence and whose doubt? Nothing is stated. Their account says too little — as when someone posits “sound is impermanent” without stating “it is eye-grasped,” leaving one in doubt about the object, or as one establishes fire by smoke. The Vaiśeṣikas’ fallacious reason is hard to establish. With this, the reason and the fallacious reason have been stated.

---

← 2. Inference for Oneself | All chapters | 4. Example and Pseudo-example →