The Gate of Logic

In order to concisely set down what is genuinely true within the topics of proof and refutation, this treatise is composed.

A group of statements built around a thesis and the rest is called "proof." Within it, the thesis follows only the proponent's own wish. What is called a "thesis" is what is to be established — not something a contrary meaning could exclude.

"A group of statements built around a thesis and the rest is called proof": because a group of statements consisting of the thesis, the reason, and the example argues out, for another person, a meaning they have not yet understood, this group is called "proof" in the older Nyāya treatises and elsewhere. It is spoken of, in the singular, as "proof" only to show that they together constitute one single property — from which it follows that whatever proof is missing any part of them has a fault of proof.

The word "within it" either opens the discussion or carries a delimiting sense: within the thesis and the rest. The word "only" carries a delimiting sense. "Following one's own [wish]" shows that the thesis is set up according to the proponent's own wish, without regard to some pre-existing tenet. "Wish" means what is to be established — the thesis is not something already wished to serve as what establishes. Were it otherwise, a fallacious reason or a fallacious example would equally have to be called a thesis. To show how a thesis steers clear of the remaining faults of setting one up, it is said: "not something a contrary meaning could exclude."

Consider a thesis excluded by its own contrary meaning through the very sound of its words — as when one sets up, "All statements are false" [which, being itself a statement, refutes itself]. Or a thesis contrary to a position the proponent has already established beforehand — as when a Vaiśeṣika teacher sets up, "Sound is permanent" [contradicting a settled tenet of his own school]. Or a thesis excluded by a meaning contrary to what is already commonly established through everyday language, where the reason offered is so exclusive that no genuine inference can even be drawn from it — as when one says, "The hare-bearer [the moon] is not the moon, because it exists." Or, with respect to the subject itself: the very property to be established, excluded by a meaning contrary to direct perception or inference already commonly established for it — as when one sets up, "Sound is not something heard," or "A pot is permanent."

As for those who call a thesis contradicting its own reason a fault of "the thesis's own contradiction" — this is not really a fault of the thesis at all. In a case such as setting up, "Sound is permanent, because everything is impermanent," the flaw actually lies in the example — badly formed as an example by difference, since its own connective statement shows that not literally "everything" can be meant. Taken literally, this reason does not even exist, since sound itself is included within "everything"; or, read as meaning only a portion of what is to be established, it cannot be established at all — which is a fault of the reason, not the thesis. The example, too, has a fault here: an example by difference must first show the absence of the thesis's property, and only then state the absence of the reason. It ought to say "everything impermanent," restricted to impermanent things — not "absolutely everything." Stated backwards, as "everything is impermanent," the example itself has a fault.

Thus the thesis and the fallacious thesis have been explained. The reason and the fallacious reason are, for the most part, matters of the property of the subject — this differentiated character will now be shown: the property of the subject, with respect to the homologous class, is present, absent, or both; and with respect to the heterologous class, each of these three again divides into the same three.

Is not the thesis, taken as a whole, what one wishes to establish? Why, then, is "the subject" here taken to mean only the subject-locus? This is no fault: a general term can function for one part of it, as when one says "burn the robe" [meaning only a stain]; or the word "thesis" here simply denotes the property being established. "The property of the subject" refers only to what the proponent and the opponent both already, definitely accept — and the same holds for "present" and "absent" within the homologous class. Why? Because we are relying here solely on the informative cause, where the object is understood only through the power of cognition — not the generative cause, which operates through actual causal efficacy.

If cognition itself is the informative cause, does the reason, as stated, lose its power to establish anything? No: it causes the other party to recall what both sides already, originally accepted — and a reason built only on what both sides definitely accept is, precisely for that reason, well formed. Hence, whenever the two sides do not equally accept a reason, it is not a genuine property of the subject — as when one sets up, "Sound is impermanent, because it is seen by the eye." Or if only the opposing party rejects it — as with "because it is produced," addressed to a follower of the Sound-Manifestation school. Or if there is doubt — as when, on the basis of something like mist, one argues, "The elements have combined and fire is present, because smoke is now seen." Or where the subject itself is unestablished — as when one argues, "The self pervades everywhere, because it produces pleasure and the rest in every place."

Statements of every kind just described fail to be a genuine proof, and the same holds, case by case, for "present" and "absent" within the homologous class. Among the reason, the contradictory reason, and the inconclusive reason yet to be discussed, only a statement definite and commonly accepted by both sides is called "proof" or "refutation" — a statement unestablished for one side, or one that leaves the matter in doubt, is neither, since it too still awaits being established.

In setting up the property of the subject, one ought, by rights, to use some further property as the reason establishing it. What, then, of directly establishing the very existence or non-existence of the subject itself — as when a Sāṃkhya argues, "Prakṛti exists, because particular things are seen to share a general class," or argues, "It does not exist, because it cannot be apprehended"? In such a case, what is actually being set up as the thesis is only that "particular things definitely have a single cause" — not "Prakṛti" itself — so there is no such fault. And arguing for its non-existence likewise provisionally posits "that which cannot be apprehended" as the real subject, so there is no fault there either.

What of using a subject to establish a further subject or its property — smoke establishing fire, or fire establishing touch? Here, one is not setting up "fire" or "touch" as bare things, but establishing their being present together, at that place. Otherwise, using smoke for fire and fire for touch would collapse a portion of the thesis into the reason. Nor does one intend to establish that fire or touch exist as things at all — that much is already known — only that they are present, as properties, at this particular place. Subject and property here are established only relative to one another, as observed together — not confused for a quality and its substance — hence no fault.

It is said again, in verse: A subject is not established by another subject, nor by a bare assertion of a property; this does not establish the subject. Only through a property is another property established — and it is in just this way that something is established with respect to a subject.

Suppose someone argues, "Sound is not permanent, for otherwise karma and the rest would have to be permanent, and the permanent would have to become apprehensible." How is this a property of the subject? This states the opponent's fault by way of both thesis and reason at once — because it uses "would have to" about what is to be established, and because, having first set up the opponent's own thesis, "Permanent, because it lacks resistant materiality," it then merely restates that thesis in reverse, exposing the fault in its reason. And what of setting it up instead as, "Sound is impermanent, because what is produced is not permanent, and what is permanent is not produced"? This is simply a method of stating the example: the example by agreement and by difference, in that order, declare that wherever the thesis's property holds, the reason definitely follows, and wherever it is absent, the reason is definitely absent too.

Here, the connective statement makes the reason "being produced" clear — sound is certainly something produced, not something not-produced — and so this property is certainly the property of the subject. It is said again, in verse: When the reason is stated, the thesis's property follows it; where the thesis's property is absent, the reason is not present. The example, stated with its connective clause, does the work the old five-part form spread across two further members — it is through this connection that the reason is known. By this, the method of proof-by-reversal is also explained: since being-produced is seen among impermanent things and not among permanent ones, one can equally argue, "Sound is not permanent, because it would then have to be not-produced." Direct proof and proof-by-reversal, then, do not need separate explanations of the reason. As for refuting the Sāṃkhya, I have already argued this at length elsewhere, and should not pursue this side dispute further here.

Thus the property of the subject divides into three: present in the homologous class, absent from it, or both — set aside "and" as not implying merely two options. A class that stands close to, and shares equally in, the property to be established is called the homologous class, since anything sharing that property counts as a "class" in this sense. Whatever lacks the property to be established is called the heterologous class — defined purely as its absence, not as "opposed to" or "different from" the homologous class. Defined as "opposed," a reason would be restricted to only one case at a time; defined as merely "different," there would be no valid reason at all. For this reason, "because it is produced" can equally establish both "impermanent" and "without self," without contradiction.

Furthermore, each of these three again has three further subdivisions: within "present throughout the homologous class," the reason may, in the heterologous class, be present, absent, or present in part; and the same threefold division applies within "absent from the homologous class" and "present in part of the homologous class." [Three times three: nine cases in all.] And if, for the thesis "impermanent," there is simply no heterologous class at all — addressing a school that does not accept space as existing — how can one still speak of the reason's absence "there"? If that class does not exist, the reason simply does not operate there, and there is no doubt about it whatsoever — so this raises no real difficulty.

Combined, this yields nine kinds of reason for the property of the subject; briefly, in order: (1) "Sound is permanent, because it is an object of cognition." (2) "Sound is impermanent, because it is produced." (3) "Sound is produced through effort without interruption, because it is impermanent." (4) "Sound is permanent, because it is produced." (5) "Sound is permanent, because it is something heard." (6) "Sound is permanent, because it is produced through effort without interruption." (7) "Sound is not produced through effort without interruption, because it is impermanent." (8) "Sound is impermanent, because it is produced through effort without interruption." (9) "Sound is permanent, because it lacks resistant materiality." These nine are gathered into two mnemonic verses [listing, in order, the nine theses and the nine reasons in compressed, synonym-substituted form — expanded below as a chart, for clarity].

Hence the root verse states: present throughout the homologous class, or present in part of it, and entirely absent from the heterologous class, is a valid reason. Reversing this pattern is contradiction. All that remains is inconclusive.

Here, only two are called a valid reason: present throughout the homologous class and entirely absent from the heterologous class, and present in part of the homologous class and entirely absent from the heterologous class. Only two, likewise, are called contradiction, since they establish the reverse of what was intended: present in the heterologous class, whether throughout or in part, together with entire absence throughout the homologous class. The remaining five, whether counted among reasons or contradictions, are simply inconclusive — a reason that leaves the matter in doubt.

What of "because it is something heard"? It is inconclusive because it is exclusive — belonging to nothing but sound itself. When a reason is this exclusive, every possible way of characterizing the property to be established remains equally doubtful, since the reason exists only as its own sole instance, entirely apart from both classes — nothing discriminates a right conclusion from a wrong one. Were the reason instead present in a portion of both classes at once, it would actually become a genuine, distinguishing reason. Addressing a school that already accepts sound-ness itself as permanent, "being heard" could indeed become valid for them — provided no competing reason, like "being produced," were already available to establish the opposite. Since such a competing reason is available, and both conclusions cannot be true at once, the matter remains genuinely in doubt — and here, one should let the greater force of direct perception and scripture settle the question.

Summarizing the above in verse: whatever is exclusive, or common to both classes, or two reasons mutually and decisively contradictory, or pervasively present throughout — all of these leave the matter in doubt. A reason wrongly establishing a property or the subject itself, whether its manifest nature or a further implication, becomes a contradictory reason, provided nothing else prevents this. Examining the property of the subject: if what the proponent wishes to establish is instead contradicted, this results in wavering doubt or outright reversal; apart from these, there is no fallacious reason.

Thus the reason and the fallacious reason have been distinguished. The example and the fallacious example will now be explained. When the reason is stated, the thesis's property follows it; where the thesis's property is absent, the reason is not present. These two statements are called "the example"; whatever merely resembles them, without truly satisfying them, is a fallacious example.

There are two kinds of example: by agreement and by difference. The example by agreement: "Sound is impermanent, because it is produced through effort without interruption. Whatever is so produced is seen to be impermanent, as with a pot and the like." The example by difference: "Whatever is permanent is seen not to be so produced, as with space and the like." The former both excludes and directly indicates the reason's presence; the latter only checks against overextension — together they establish the inferential relation through both connection and separation. Because of this, even addressing a school that does not accept space as truly existing, one can still show that wherever the thesis's property is absent, the reason is absent as well.

Why does the first example say the reason is followed by the thesis's property, and the second say the reason is absent where the thesis's property is absent, rather than the reverse order? Because stating it in this order is precisely what shows the reason to be present in the homologous class and entirely absent from the heterologous class — reversing the order would misstate the logical direction. It is said again, in verse: were the order reversed, one might use "not-produced" to prove permanence, or "impermanent" to establish "produced" — proving something not actually intended, failing to be pervasive, and disagreeing with the connection and separation actually wished for.

Thus the two examples, proper and reversed, have been explained. Whatever merely resembles them is a fallacious example — any case where a connection or separation is stated for the property, the reason, and the heterologous class, but stated in reversed or otherwise flawed form; or where no explicit connection or separation is made at all, and the two are merely shown present together, or both absent from the heterologous class. In such cases, either the reason or the property may fail to be established or excluded, or both may fail together.

For instance: "Sound is permanent, because it lacks resistant materiality," with a flawed example by agreement — "whatever lacks resistant materiality is seen to be permanent, as with karma, an atom, or a pot" — and a flawed example by difference — "whatever is impermanent is seen to have resistant materiality, as with an atom, karma, or space." Must both examples always be stated in full for a proof to be complete, or may only one be given, as with the reason? Strictly, both ought to be stated, since only this fully shows the reason present throughout the homologous class and absent throughout the heterologous class, correctly guarding against both contradiction and inconclusiveness. But if one portion is already established by other means, stating only that portion still completes the proof — and if, as with the standing example of sound, both sides already agree on both directions, neither need be stated at all, since one can imply the other.

In inference, only this principle is actually at work: if this characteristic is carefully ascertained at the place to be inferred about, then, for every other instance of the same kind, it is recalled as definitely present, and for every place lacking it, recalled as entirely absent — and it is from this that a determinate understanding arises. Hence the root verse: just as one's own understanding has become determinate, so the same determinate understanding is sought to be produced in another — by stating the property of the subject, its concomitance, and what is to be established; everything else is set aside.

The reason is stated to show, for what is to be inferred, that it is a property of the subject; the example is stated to show its inseparable connection to that property; the thesis is stated to indicate what is to be inferred. Beyond these, there are no further members — this dispenses with the older syllogism's further scrutiny and its application and conclusion. One might object: then the example ought not to count as a separate member, since it only clarifies the reason's own meaning. This is true in substance — but the reason's own statement shows only that it is a property of the subject, not its presence in the homologous class or absence from the heterologous class; hence the example must still be stated separately. What merit is there, then, in stating it as its own member? That very completeness is the merit.

One might further object: this should simply follow how everyday speech uses examples, which has nothing to do with a technical account of the reason — so what is the fault in relying on that alone? Answer: an everyday example, merely likening the thing to a similar case, carries no real logical force; it never actually states the establishing relation between reason and property. Why no force? Because, in an everyday example, the property of the subject and the meaning of the thesis need not truly resemble each other in the required way — and even that resemblance would itself need a further example to establish it, leading to infinite regress. Hence the verse: if the reason only resembles the thing to be established, or merely resembles a further implication of it, the example would regress infinitely, and could never properly exclude the heterologous class.

Everyday usage merely shows the thesis and reason both present in some heterologous-seeming case and calls this an example by difference — it never properly establishes that the reason is absent wherever the thesis's property is absent. If merely being a property of the subject were enough to make something a valid reason, even a merely inconclusive property would have to count as valid. This fault persists precisely when the heterologous instances used are not truly of a single, properly excluded kind — as with the very last case in the first and last groups of three in the nine-fold chart above. Hence the three characteristics alone reveal a genuine reason. And although every part of the proof together reveals what is to be established, only one part — the reason proper — is, strictly, called "the reason." Thus the thesis and the rest, genuine and fallacious, have been briefly explained; this whole group of statements is called proof and fallacious proof, set forth, as is fitting, for the sake of awakening others.

For the sake of awakening oneself, there are only direct perception and inference — testimony, analogy, and the rest are all subsumed within these two. It is by these that particular and universal characteristics are respectively known; there is no further object of knowledge, apart from these two, requiring some additional means of knowledge to be devised. Hence the root verse: direct perception excludes conceptual construction; the rest arise from the reason as already explained.

"Direct perception excludes conceptual construction" means: whenever a cognition, with respect to an object such as visible form, is free of every kind of conceptual construction — by class, verbal designation, provisional imputation, false non-differentiation, and other logical approaches — and instead arises, through its own uncommon condition, vividly and distinctly, moment by moment, this is called direct perception. Hence the verse: a thing does not have a single fixed character; a sense faculty does not operate upon everything indiscriminately — what is grasped only by inner realization, apart from words, is the true object-domain of a sense faculty and its form.

The mental faculty, too, has a mode of cognition free of all conceptual construction, operating solely through direct realization. Likewise, the self-witnessing awareness of desire and other afflictive states, and the meditative cognition of practitioners of concentration free of conceptually-derived teaching, are all direct perception. Here, again, there is no valid means of knowledge and its result as two separate things: because this cognition itself arises resembling its object, and functions as though it had an effect, it is provisionally called a "means of knowledge." If the self-witnessing awareness of desire is also direct perception, why is conceptual cognition excluded here? This does not rule out the self-witnessing direct perception present within such states, which remains free of conceptual construction — only the further cognition that grasps some other, external object-portion is not called direct perception.

By this same reasoning, memory, inference, desire-driven anticipation, doubtful cognition, and cognition under delusion, with respect to gross states such as craving, are none of them direct perception, since they all operate through conceptual construction carried over from what was previously apprehended. In just the same way, every cognition within conventional existence — of "a pot," "a number," "a raised gesture," "universal being," "pot-ness," and so on — is fallacious perception: with respect to what is genuinely real, it constructs some further aspect, provisionally combining it with some other meaning, through conceptual construction.

Direct perception has now been explained; inference is next. "The rest arise from the reason as already explained" means: this cognition, other than direct perception, arises from the reason exactly as described, taking that reason's meaning as its object. It is of two kinds: the carefully examining cognition regarding what is to be inferred, arising from either direct perception or a further inference; and the recollection that this reason is inseparably connected with the property to be established, accomplishing, through the power already described, the recognition that the reason is definitely present in the homologous class. Because it infers both what is near and what is remote, both are called "inference" — spoken of in terms of the instrument and the agent of the act, respectively. In just this way, inference for the sake of awakening others, too, is accomplished as proof by nothing other than this.

Hence the verse: a single thing has many properties; any one characteristic does not extend to everything that has it as its locus — only by excluding all the rest does a term determinately pick out the one property meant. In just the same way, what serves as the marker likewise has many properties of its own, yet, without exceeding what it is meant to mark, indicates only that property, nothing else. Why is inference established here as distinct from direct perception above? To show both approaches: here, too, one could speak of inference's own result as itself "inference," and there, too, of direct perception's own cause as itself "direct perception" — neither usage is barred. Proof and fallacious proof having now been explained, refutation and fallacious refutation are next.

The verse states: refutation is a statement of deficiency and the rest; fallacious refutation is what is called the sophistical types. "Refutation is a statement of deficiency and the rest" means: the various faults already described — deficiency in one of the proof's parts, and the rest. Each such statement is called refutation, since each individually shows that the opponent's thesis was not well stated.

As for "fallacious refutation is what is called the sophistical types": this refers to the sophistical, seemingly-similar types of objection — by agreement and the rest — called fallacious refutation, since, for the most part, they are devised to confuse an opponent in the face of an otherwise sound inference. They fail to show the opponent's thesis unsound, because they refute without genuine reasoning, and yet are devised precisely where a genuine refutation belongs — being that same general kind, they are called sophistical types. But if such a device is used against an inference genuinely unsound to begin with, or without recognizing a real fault already present, or precisely to expose that fault, it is not a sophistical type at all, but a legitimate move.

Because it displays the heterologous class wrongly, arguing backwards by misusing the example by agreement, is resemblance-by-agreement; the mirror-image, by misusing the example by difference, is resemblance-by-difference. Wrongly distinguishing a further implication is resemblance-by-distinction. Wrongly forcing things that share one point into total identity is resemblance-by-non-difference. Pointing to some other reason for what is already established is resemblance-by-attainability. Objecting to the meaning by manufacturing an ambiguity in the reason is resemblance-by-uncertainty. Drawing out an unintended meaning for the heterologous class is resemblance-by-implication.

The word "mostly" allows for different kinds of objection, and for cases resembling the fault of an unestablished reason instead. The first four types do not correspond at all with the method of example I myself have set out — they simply follow the everyday method of example, which, though it does not show a reason to be genuinely determinate, still captures its general substance. By using an inconclusive reason like the example by agreement to establish one's own thesis, such an objection's method effectively claims the opponent shares that same property too — amounting, in substance, to a resemblance of a "common" inconclusive reason, or of two mutually contradictory yet equally well-formed reasons. One might ask: if this is meant only to establish one's own thesis, how can "inconclusive," as a charge, itself count as refutation? It is not put forward as the refutation itself — objecting to a reason as inconclusive is simply called "raising inconclusiveness." Since the charge states, within what is expressed, the very fault it points to, there is no real difficulty here.

The same holds elsewhere. If the proof genuinely has the fault of inconclusiveness, or a determinate reason like the example by agreement can genuinely establish something real, then pointing this out does count as genuine refutation. Because of such objections: where the force of direct perception already settles a matter, inference cannot override it. For instance, "Sound is not something heard, just as with a pot" fails, since sound is directly perceived to be something heard — one ought not use a formal argument from "being heard" to try to override impermanence in the reverse case either: were it otherwise, such an argument would have to override even permanence in cases where it plainly shouldn't.

The second version of resemblance-by-non-difference merely resembles an unestablished reason: it groundlessly exaggerates the claim by forcing thesis and reason into one fault. Properly used, it can instead employ a genuinely, commonly-established reason to prove real absence after cessation — if this is truly the intent, it can count as genuine refutation. The third version establishes something contradicting what was to be proved, and so merely resembles refutation, since a property like combustibility is not necessarily decisive; were it genuinely decisive, it could indeed amount to real contradiction. Resemblance-by-attainability leaves matters inconclusive and so merely resembles refutation; but if the offered reason applies equally to something permanent, it can genuinely refute. Its second version, though not strictly pervasive, resembles an unestablished reason for lack of a fitting further class; but if what is to be established is genuinely absent, it too can be genuine refutation — though not in the sense of proving literally everything impermanent.

Resemblance-by-uncertainty takes "produced through effort" as establishing destructibility. Insisting on "generation" specifically exaggerates the claim and merely resembles inconclusiveness; objecting only by distinguishing "generation" within the reason, without raising any real issue with what is to be established, merely resembles an unestablished reason — since the proponent never intended "generation alone" in the first place: whether manifested or generated, both are equally destructible. Resemblance-by-implication raises its objection through an improperly reversed inconclusive claim, and so merely resembles inconclusiveness; but arguing that what is not produced through effort could be either permanent or impermanent, or that only what is so produced is impermanent and nothing else, could indeed amount to genuine refutation.

Objecting that the reason does or does not "reach" its conclusion, and objecting via the three times, are unwelcome statements — resemblance-by-arrival-or-non-arrival, and resemblance-by-absence-of-cause; these resemble a deficient reason.

Both of these, without genuine reasoning, deny that things can have causes at all — one, by treating mere "non-arrival" as equivalent to non-existence even when the three characteristics are fully satisfied; the other, by refusing the name "reason" to anything that precedes what it establishes. Both also carry a self-defeating fault: applied consistently, the same objection would rule out everything equally, including its own reasoning. With respect to what is established purely by verbal statement and reasoning, these resemble a deficient reason; with respect to what is established by real, causal reasons in the world, they resemble an unestablished reason — in both cases, without genuine cause, denying that things have causes. Since neither actually involves a genuine producing-and-produced relation in the naive sense the objection assumes, neither accords with correct reasoning; but raised with genuinely correct reasoning, a similar objection could indeed count as a real refutation.

Objecting that, because the reason was not yet stated beforehand, there ought to be nothing established, is resemblance-by-non-statement. The same, applied to arising and non-arising, is likewise given. Objecting that being-produced applies to only a part of what shares the reason with the example, so what is to be established is unestablished, is resemblance-by-being-produced. Most of these largely follow the pattern already used for the fallacious thesis.

Resemblance-by-non-statement exaggerates the inference's actual scope — this only resembles an unestablished reason, or, in another reading, a deficient reason, since it objects that the argument had not yet been completed as a proof before it was stated. If the objection actually shows the thesis's meaning does not hold, or that no genuine statement was made at all, it can count as real refutation. Resemblance-by-non-arising likewise exaggerates the claim with respect to sound before it arises; but if the absence is genuinely shown at the actual time of proof, it can be real refutation. Resemblance-by-being-produced actually comes in three forms: objecting that a pot's specific "being-produced" is absent in sound merely resembles an unestablished reason; objecting that sound's "being-produced" is absent in a pot merely resembles a contradiction; objecting that this same property is even absent from something permanent, being exclusive, merely resembles inconclusiveness — or it may resemble a fault in the example instead. Why? Because a valid inference is built only on a general property, never a particular one — take the particular meaning instead, and it necessarily differs from case to case, so that no general inference could ever be formed at all.

These sophistical types, for the most part as given by Akṣapāda, are explained here as the nature of fallacious refutation, being the most well-established examples of the kind. What other treatises describe should be sorted and established the same way. These same types, by slight variations in method, could be multiplied into limitless further distinctions, and so are not further listed here. Whatever else is said — of exaggeration, diminishment, false manifestation, non-manifestation, flawed causal reasoning, alternate examples, resemblance of kind, and so on — should all be examined and set aside by this same general approach. Those who devise poor methods of inference this way go on branching out endlessly; since other treatises already discuss this without end, it is not discussed further here.

As for "points of defeat" in debate: whatever the old hetuvidyā masters have said on this either already falls under genuine refutation, or is extremely crude, or is simply unreasonable, resembling mere sophistry — and so it is not recorded here. The categories proposed by other schools should likewise be sorted and established by the same method. In this way, every instance of the discriminated, attached-to way of thinking and the like fails to accord with reason, contradicts the characteristics already described, and is, without exception, a lack of understanding — being extremely far from correct reasoning. I have already, for the most part, subdued this whole class of faulty statements myself, among my own school's [Nyāya] Treatise-Forms and elsewhere; and I have already given a complete account of this same topic in my treatise refuting the old hetuvidyā treatise — so I should stop here.

As a healing medicine of wisdom to open the minds of the intelligent, I have set forth this profound teaching, the gate of right reasoning — so that all who are confused by outsiders' flawed means of knowledge may cross beyond the wrong path and reach the true meaning.