Pramāṇayuktanidhi

Pramāṇayuktanidhi (ཚད་མ་རིགས་གཏེར་)

Treasury of Reasoning on Valid Cognition — sakya-pandita’s masterpiece and arguably the most important Tibetan-authored text on Buddhist epistemology. Known simply as “the Treasure” (རིགས་གཏེར་).

Overview

The Treasure is a direct engagement with dharmakirti’s thought. Rather than relying on commentaries, Sakya Paṇḍita quotes the master directly — an apparently unique feature in Tibet. The prime target of the work is Tibetan realism — the attempt by cha-ba and his followers to reintroduce real universals into Buddhist epistemology (dreyfus-recognizing-reality, p. 24).

The root text is written in verse form and was renowned for its brevity, pithiness, and mnemonic quality. It is accompanied by an auto-commentary, though its authenticity has been questioned by some (sonam-introduction-treasury).

The Treasury is arguably the most authoritative and coherent commentarial treatise on the pramāṇa system of Dignāga and Dharmakīrti from a native Tibetan scholar. It is one of the few indigenous Tibetan texts to have received commentary from scholars of almost all denominations of Tibetan Buddhism — at least fourteen extant commentaries are known (sonam-introduction-treasury).

Structure

The text is divided into two main sections (sonam-introduction-treasury):

Part I — Ascertaining the Object (Chapters 1–7): subdivided into three parts: the object of comprehension, the cognitions that comprehend it, and the manner in which cognitions comprehend their objects.

Part II — Ascertaining Valid Cognition (Chapters 8–11): covering definition, perception, and the two types of inference.

Each chapter follows the threefold structure conventional for writers of the period: refutation of others’ systems, presentation of one’s own view, and responding to possible criticisms (དགག་བཞག་སྤོང་གསུམ་) (sonam-introduction-treasury).

Sakya Paṇḍita also marks his own joints. Two kinds of verse punctuate the chapters: the interstitial verse (བར་སྐབས་ཀྱི་ཚིགས་སུ་བཅད་པ་), inserted mid-composition to recapitulate a completed stretch of argument, and the summary verse (བསྡུ་བའི་ཚིགས་སུ་བཅད་པ་), which closes a chapter by restating its thesis. Where a chapter has one, that verse is the shortest available statement of what the chapter was for.

The text chapter by chapter

Each chapter below opens with its summary or concluding verse — the anchoring verse for Chapters 8 and 9, which carry neither marker, is the chapter’s own definition. The dialectic beneath it (opponents named, objections in their own voice, the examples actually used) comes from Khenpo Sodargye’s lectures, summarised chapter by chapter under index; the structural overview is from sonam-introduction-treasury and the interpretive placement from dreyfus-recognizing-reality. Root verses are quoted from the working translations under translations/pramanayuktanidhi/.

Chapter 1: Objects (ཡུལ་བརྟག་པ་)

།ཇི་སྲིད་ཕྱི་རོལ་ཁས་ལེན་པ། །དེ་སྲིད་རྒྱུ་ལ་གཟུང་ཡུལ་ཟེར། །ཤེས་བྱ་ནང་གིར་ཞུགས་པ་ན། །ཡུལ་དང་ཡུལ་ཅན་གྲུབ་པ་མེད།

So long as external objects are accepted, the [object that is the] cause is called the “apprehended object”. But once the knowable is drawn inward, the distinction of object and subject is no longer established. (summary verse)

An object is “that which is known by cognition” (ཡུལ་གྱི་མཚན་ཉིད་བློས་རིག་བྱ།, TR 1.1) — the referent of any cognition whatever, conceptual or not, correct or erroneous. Sapaṇ’s own position is the flat restriction of TR 1.11: specifically characterised phenomena alone are objects of valid cognition (གཞལ་བྱ་རང་མཚན་གཅིག་ཁོ་ན།). Everything else that gets called an object — the generally-characterised phenomenon, the double moon, the floating hairs of eye disease — is cognition itself, mistaken for something standing outside it.

The opponents are cha-ba and the early Tibetan logicians, who held that the conceptual generality (དོན་སྤྱི་) and the non-existent appearance are substantially real apprehended objects with characteristics of their own, existing outside consciousness. Sapaṇ presses two consequences. If they were real objects, the cognitions grasping them could not be erroneous, since a real object grounds a non-erroneous cognition; but both sides agree those cognitions err. And real self-characterised entities standing outside cognition are publicly available: anyone suitably placed should see my conceptual image of a vase as they see the vase itself.

The heart of the chapter is what happens when the opponent retreats to privacy — real, but bound to one’s own mind, as one’s internal organs are. Sapaṇ takes the retreat and pushes it until it destroys language. If my vase-image is tied to my mind and my listener’s to his, with no connection between them, then saying “vase” a thousand times could never raise the concept in him. A shared concept does arise, so the premise is false; the cost of the opponent’s position is the severing of every speaker–listener relation in the world. Pressed to say the two images are identical, he faces a further problem: judging two things similar requires a third party who sees both, as deciding that two friends have similar faces does, and no such observer exists for two private images. The opponent finally concedes that the images are distinct and are treated as one only through error — which is Sapaṇ’s own thesis, conceded (chapter-1-objects).

Nothing outside the two truths is needed to house the leftovers. Ask whether the floating hairs existed and they become an object of thought, hence a generality, hence conventional truth; look at them as cognition and they are a self-characterised cognition, hence ultimate truth. The chapter then runs the Sautrāntika and Yogācāra positions in sequence, and the summary verse ranks them: while external objects are accepted, the cause that imparts its aspect to cognition really is the apprehended object; once the knowable is drawn inward, object and subject fall together. Sapaṇ’s hermeneutical frame of three worldviews (Sautrāntika, Cittamātra, Mādhyamaka) is set out here, and the ultimate tenets of the pramāṇa teachers identified as Cittamātra (sonam-introduction-treasury).

Chapter 2: Cognition (བློ་བརྟག་པ་)

།ཚད་མའི་ཤེས་པ་གཉིས་པོ་ཡང༌། །རང་རིག་ཚད་མ་ཁོ་ནར་འདུས། །ཚད་མ་མིན་པ་གསུམ་པོ་ཡང༌། །མ་རྟོགས་པ་ལས་དོན་གཞན་མིན།

The two valid cognitions are subsumed into self-awareness alone as pramāṇa; and the three non-valid cognitions are, in substance, nothing other than non-realisation. (summary verse)

Cognition is “clarity and knowing” (བློ་ཡི་མཚན་ཉིད་རིག་པ་ཡིན།, TR 2.1) — a mark that holds across every division, conceptual or not, erroneous or not, valid or not. Non-valid cognition is “cognition in which non-deceptiveness is not established” (མི་བསླུ་མ་གྲུབ་ཚད་མ་མིན།, TR 2.18).

Dreyfus argues that this chapter is the central polemical thrust of the whole Treasure: Sa-paṇ collapses cha-ba’s seven-fold typology back into the classical three-fold division by refuting the three innovations — investigative thought (Dreyfus: “correct assumption”; ཡིད་དཔྱོད་), appearance-without-ascertainment (སྣང་ལ་མ་ངེས་པ་), and subsequent-cognition (བཅད་ཤེས་) (dreyfus-recognizing-reality, pp. 389–94).

The argument against investigative thought is a dilemma about reasons. Either the alleged cognition depends on a reason or it does not. If not, it is bare thesis-assertion — a guess — and collapses into doubt. If it does, the reason is genuine or one of exactly three pseudo-reasons, and each of those already has its home: a contradictory reason yields wrong cognition, an inconclusive one doubt, an unestablished one non-realisation. Nothing is left for a fifth category to be. Sapaṇ then works the parallel: if a guess that happens to match fact counts as “genuine” investigative thought, a guess that misses should be erroneous investigative thought — and admitting that destroys the opponent’s own definition, which required accordance with fact, while denying it removes the ground for the genuine case on identical reasoning.

Against subsequent cognition the argument is that the second moment of perception is no less knowledge-generating than the first, since each moment holds a different, momentary object; the “continuum” is a conventional label, and the subject changes as continuously as the object does. Against appearance-without-ascertainment: since perception never determines anything anyway, perceptions that fail to induce certainty are not a special class — and if lack of ascertainment were what made them non-valid, sensory, mental, self-aware and yogic perception would all drop out of valid cognition, leaving inference alone.

Sapaṇ’s own division is threefold — non-realisation, wrong cognition, doubt — separated by mode of apprehension, one in essence, all three finally reducing to non-realisation. Why keep three at all, if classification must proceed by kind? Because the division answers a specific need in Chapter 11: with only “non-realisation” available, the sole reply to a faulty reason would be “unestablished,” and the replies “contradictory” and “inconclusive” would be lost. Against “sound is permanent, because produced,” one cannot say unestablished — produced is established of sound (chapter-2-cognition).

That principle is also what condemns the fivefold scheme, and it is a structural objection rather than a piecemeal one. Enumerations have to be gathered by kind, so that every member returns to a single head, as men, women and paṇḍakas return to “human”; a blade drawn from a sheath belongs to that sheath. Sapaṇ’s three return to non-realisation, because doubt does not penetrate reality and wrong cognition does not penetrate reality. cha-ba’s five cannot return anywhere, since they omit non-realisation altogether — the one head under which the rest would have to gather. The opponents accept the principle of classification by kind and then produce a list that violates it.

The positive account is correspondingly compact, and doubt carries the weight. Doubt is the state in which, while one apprehension is still in force, the grasping of “not that” also arises; it divides into manifest and latent, and manifest doubt further into even (the two possibilities held at exactly equal weight) and weighted (leaning to one side, reasonably or not). Investigative thought is filed under latent doubt on a criterion the Auto-commentary supplies and that decides the whole earlier argument: whatever has neither establishing evidence nor invalidating evidence is latent doubt. The guess about water in the old well has neither. Beneath the three-and-two divisions runs a chain of reductions Sodargye calls pramāṇa’s “secrets,” also given in the Auto-commentary: all non-valid cognition to non-realisation, all valid cognition to perception and inference, inference to perception, perception to self-awareness, and outer objects to the specifically characterised phenomenon.

Mipham Rinpoche’s classification differs on both contested items: he places investigative thought under non-realisation rather than doubt, and treats subsequent cognition as valid — specifically as the result of valid cognition, since it accords with fact and is indispensable on the path of preparation, where the practitioner repeatedly familiarises himself with views already ascertained. Sodargye reports Khenpo Jigme Phuntsok’s ruling that the two masters refute and establish from different focal points and that the content is the same: from the object side, subsequent cognition delivers a generality already known, hence non-valid; from the subject side it accords with the object, and Sapaṇ grants that much himself. Dreyfus notes a sharper irony in the Gelug reception — later monastic textbooks cite Sa-paṇ’s refutation of investigative thought as if it were an endorsement (dreyfus-recognizing-reality, p. 390).

Chapter 3: Universals and Particulars (སྤྱི་དང་བྱེ་བྲག་བརྟག་པ་)

།ཐ་དད་སྤྱི་འདོད་འུག་ཕྲུག་པ། །རྫས་གཅིག་ཡིན་ཞེས་གྲངས་ཅན་པ། །གངས་ཅན་པ་རྣམས་དེ་རྗེས་འབྲང༌། །སེལ་བར་འདོད་པ་མཁས་པའི་ལུགས།

The Owl School (Vaiśeṣikas) hold the universal to be a distinct substance; the Sāṃkhyas hold it to be one substance; the Snow Landers follow them; to hold it as exclusion is the tradition of the wise. (concluding verse)

The chapter fixes the allocation on which the rest of the treatise runs, and Sapaṇ says plainly that he takes it from the Indian treatises rather than from any Tibetan predecessor: universal and particular belong to the object; conceptual and non-conceptual to the subject; exclusion and appearance to the manner of taking the object. Non-conceptual cognition apprehends particulars by way of appearance; conceptual-cognition grasps generalities by way of exclusion. Lose the allocation and appearance is perpetually confused with exclusion.

The refutations run through every way a real universal might be held (problem-of-universals). Against the Vaiśeṣika, who make universal and particular distinct substances: a universal existing apart from its instances would be perceptible, and apart from gold vases and silver vases nothing pervading them can be seen, heard, or found among their atoms. Against the Sāṃkhya, who make them one substance: if a single universal is one with vases of every place and time, then all of them must arise and cease together, since destroying one destroys what it is one with — and if the universal “human” were real, one man’s death would kill everyone alive. gorampa presses the same as a dilemma: when a particular ceases, does the universal cease? If not, universal and particular are not one substance; if so, every other particular ceases with it.

Three further consequences are aimed at the Tibetan realists rather than at the non-Buddhists, because only Buddhists have to accept them. First, the Vaiśeṣika whole-with-parts becomes irrefutable — one real universal pervading all particulars is that doctrine under another word, and dharmakirti refutes the whole-with-parts at length. Second, the senses become conceptual: if the universal sits on every particular and is non-different from it, the eye should see it. Third, contradictories collapse — white and black are both colours, and if “colour” is one with each, they are one with each other, and we should see nothing but blackness.

Sodargye supplies a genealogy for the error that is worth recording and worth checking. ngok-lotsawa studied seventeen years in Kashmir under masters learned in both Buddhist and non-Buddhist pramāṇa; non-Buddhist realism about the universal entered his works, and through him passed to cha-ba and his successors. The criticism is doctrinal only — Ngok’s standing as translator of the Pramāṇālaṃkāra, as founder of the Tibetan teaching-and-debate tradition, and as teacher of vast assemblies is not at issue (chapter-3-universals-and-particulars). The genealogy is presented there as established fact and has not been cross-checked against dreyfus-recognizing-reality. [unverified]

Sapaṇ’s own account: the universal is exclusion of what is not it, an image in the mental continuum with no nature of its own, established on entities only by way of function. It divides fourfold — conceptual generality (དོན་སྤྱི་), name-generality, kind-generality, aggregate-generality — each with its matching particulars; gorampa gives the three-way division of kind (རིགས་སྤྱི་), meaning (དོན་སྤྱི་) and collection (ཚོགས་སྤྱི་) universals (sonam-introduction-treasury). A non-entity can do the work of selection because we mistake it for the thing: were the generality actually the thing’s nature, saying “fire” would send fire from one’s mouth. That mistake is not a defect in the system but the mechanism of it — valid cognition itself operates on a base of error, and the genuine particular has never been spoken and never been thought. gyel-tsap pushes back on where this leaves karma, warning that denying reality to duration undermines karma and its fruits (sonam-introduction-treasury).

Chapter 4: Proof and Exclusion (སྒྲུབ་པ་དང་གཞན་སེལ་བརྟག་པ་)

།སྒྲུབ་དང་གཞན་སེལ་རྣམ་དག་གི །མིག་དང་བློ་གྲོས་གསལ་ལྡན་པ། །རྟོག་གེ་པ་དེ་ཤེས་བྱ་ཡི། །དེ་ཉིད་ལྟ་བ་ཀུན་མཁྱེན་འདྲ།

The reasoner endowed with the eyes of pure establishment-and-exclusion and with luminous intelligence — that one, in seeing the suchness of objects of knowledge, resembles an Omniscient One. (interstitial verse, closing the chapter)

The apoha chapter, and by Sodargye’s reckoning the pivot of the treatise: exclusion is “the heart and eyes of pramāṇa,” and with it in hand the earlier chapters become clear and the later ones easy. Where Chapter 3 treated object and subject, this one treats neither — only the manner in which the subject takes the object. Non-conceptual cognition apprehends by appearance; naming-conceptual cognition grasps by exclusion; each divides into mistaken and unmistaken, giving four.

The argument that perception does not determine its object is built from the object’s side and the subject’s side at once. On the object’s side, a thing has no manifest and non-manifest part: seeing sapphire, its colour, shape, momentariness and producedness are all equally present, and the “properties” are one entity divided only by thought. On the subject’s side, apprehension by appearance is not sequential — everything presents at once, as an image in a mirror. So the question why someone knows a thing is produced but not that it is impermanent has to be answered elsewhere: by deluding causes, and by the fact that certainty is conception’s work, not perception’s. Perception is valid by being unmistaken about its object, and needs no determinative function at all. cha-ba’s school proposed that perception lacks conceptual determination but has non-conceptual determination; Sapaṇ’s reply is that no such doctrine appears anywhere in the seven treatises, and that if it means the eye faithfully sees a form, that is not determination but apprehension.

Certainty then arrives in one of two ways — by perception’s own power, as when seeing red coral one knows at once that it is red, or by another’s power, as when the reasoning “since it is produced, it must be impermanent” supplies what the eye could not. Either way certainty can be drawn out of perception, which is what the objection required.

The chapter’s two structural doctrines are the modes of exclusion. Distinct things grasped as one: sameness of function (all fires burn) or of defining mark (all trees have branches) lets habit treat many as one, and that is what gives a single word its power to refute and establish across a whole class at once — “there is no tree in front” negates sandalwood, cypress and pine together, without enumeration. One thing grasped as distinct: a single entity divides before exclusion into vase, impermanent, produced, blue, non-microphone, without limit, and that division is what makes a three-membered argument possible. Without it one could say only “the pillar is a pillar, because it is a pillar.”

Against exclusion residing in the external object — cha-ba’s position, that the vase’s own entity already excludes non-vase — Sapaṇ argues by parity. If exclusion were in the particular it would become an object of appearance and lose its own defining character. And if exclusion is in the object, appearance is too, both being modes of apprehension; but appearance is the way a non-conceptual cognition arises in dependence on the object through a faculty, so colours could be seen without an eye — and there would be no blind or deaf people at all (chapter-4-proof-and-exclusion). How many exclusions to admit remains a live disagreement: Śāntarakṣita, Devendrabuddhi and kay-drup count three, dharmottara and Sapaṇ two, gorampa three but with cognitive exclusion taken as the entity of cognition itself, and Mipham two or three depending on the work — a divergence that is largely verbal, since all parties deny any exclusion existing independently of cognition.

Dreyfus reads Sa-paṇ’s apoha here as closer to Dignāga than to Dharmakīrti: like dignaga, Sa-paṇ treats language as primarily an internal articulation of the conceptual realm, giving less weight than dharmakirti to the causal-experiential bridge that connects concepts to reality. This is no accident — Dignāga and Sa-paṇ share a polemical situation, refuting realism, Hindu in one case and Tibetan in the other, whereas Dharmakīrti is on the defensive against Hindu charges of arbitrariness and so emphasises the causal-objective grounding (dreyfus-recognizing-reality, pp. 274–75).

The chapter also carries the simile Sapaṇ returns to whenever the gap between seeing and saying has to be closed: sense-cognition is a mute with clear eyes, conception a blind man skilled in speech, and self-awareness the one with all faculties intact who introduces the one to the other. The three are not three minds in one continuum, which would give a single being three continua, but isolates of cognition’s functions.

Chapter 5: Expressed and Expresser (བརྗོད་བྱ་དང་རྗོད་བྱེད་བརྟག་པ་)

།སྒྲ་ཡི་དངོས་ཀྱི་བརྗོད་བྱ་མེད། །ཞེན་པའི་བརྗོད་བྱ་རང་མཚན་ཡིན། །འབྲེལ་པ་འཁྲུལ་པས་བྱས་ན་ཡང༌། །འཇུག་པའི་སྐྱེས་བུ་བསླུ་བ་མེད།

The word has no real expressed; the clung-to expressed is the specifically characterised phenomenon; and though the connection is made through confusion, the person who engages the object is not deceived. (summary verse)

Four candidates could be what language expresses, and Sapaṇ eliminates all four. The particular of the external object fails twice over: names cannot be assigned to particulars in the first place, since they are unmingled and limitless by place, time and form, and a vase perishes several times over while “vase” is being said; and even if assigned, the name could not be used, because the particular it was laid on is gone. Past and future would be inexpressible — yet we speak of Maitreya and of Kāśyapa. The particular of cognition fails for exactly the same reasons. The kind-generality fails because there is no generality in the object to name, and naming one would serve no purpose, since a generality cannot lead us to a thing. And the word-and-meaning generality taken as a real non-associated factor — the position of the early Tibetans and the Sāṃmitīyas — fails because on Mahāyāna analysis no such substance exists outside mind and matter, and a name bestowed on such a factor could no more make a form known than naming a vase could make a yak known.

What is left is the generality taken as unreal. The real expresser is the word-generality, the real expressed the conceptual generality that arises through it; the clung-to expresser and expressed are the particular sound and the particular thing. Both parties, speaker and hearer, take themselves to be dealing in particulars, and both are mistaken; the shared mistake is sufficient for communication to work.

The chapter’s sharpest argument is about the constancy of the error. The opponents fall back on resemblance: name and thing merely resemble each other, and in use we confuse them. Sapaṇ answers that error from resemblance is by nature occasional — twins in identical clothes are confused at first sight, but on closer look one nose is bigger, and in time nobody mistakes them. Name and thing are confused always, in every place and every form. So resemblance is not the cause; the cause is that conception’s own nature is mistaken, and whatever appears to it is therefore incorrect. This is what separates Sapaṇ’s position from the early Tibetans’, who took the generality to be substantial and never saw that conception is mistaken by nature (chapter-5-expressed-and-expresser).

Naming and conceptual identification are conventions, then, and a form of mistake in which all unenlightened beings participate under ignorance. The chapter introduces the division of two levels of discourse — theoretical standpoint (འཆད་པའི་ཚེ་) and practical standpoint (འཇུག་པའི་ཚེ་) (sonam-introduction-treasury) — and that division does real work in every later chapter: in exposition, generality and particular are separated; in application they are fused, and the fusion is what reaches the thing. The mirror analogy carries the point about how a non-entity can function: what is expressed is the reflected image, not the mirror, and by the image one can remove a blemish from one’s own face.

Chapter 6: Connection (འབྲེལ་པ་བརྟག་པ་)

།དངོས་པོའི་དོན་ལ་འབྲེལ་པ་མེད། །འབྲེལ་པ་གཉིས་ཀྱང་སྒྲོ་བཏགས་ཡིན། །སྒྲ་ཡི་དོན་ལ་རང་མཚན་དུ། །འཁྲུལ་པས་འབྲེལ་པའི་རྟོག་པ་སྐྱེ།

In the entity, as object, there is no connection; both connections too are superimpositions. It is through mistaking the word’s referent for a specifically characterised phenomenon that the conception of connection arises. (summary verse)

Because all refutation and proof depend on connection and contradiction, both are examined before the definition of valid cognition itself. The claim of the chapter is startling on first hearing: no relation exists in the nature of things. Two entities each abide in their own nature without a particle’s mingling, so nothing joins them; one entity has nothing to relate to, since nothing relates to itself; and non-entities have no nature to relate with at all. There is no third category in which “not forsaking another” could reside.

Both of the two connections dignaga and dharmakirti admit fail at that level. Identity: one entity cannot bear two phenomena, so the producedness and the impermanence of a pillar have nothing between them to be connected. Causality: if cause and effect are simultaneous there is no causation, and if sequential the two relata are never both present — when the cause exists the effect is as absent as a barren woman’s son, and when the effect arises the cause has ceased. An intermediate link fails too: permanent, it has no function; impermanent, it perishes in the instant it arises and needs a further link, without end.

The non-Buddhist supplements go down to the same one-or-many test and are then absorbed one by one. A stone resting in a bowl is not a third kind of relation but the fruit’s own prior nature as substantial cause with the vessel as cooperative condition. The staff-holder known by the staff is the same case. Conjunction of an impermanent particular with a permanent universal fails because no permanent thing exists to depend on — joining it to an impermanent one is like marrying the barren woman’s son to a yak. Master and servant, husband and wife: not one entity, so not identity-connected, and at any one time bearing no relation at all; across time the servant acts on the king’s distinguishing feature, and that is causal connection.

What survives is a discrimination Sapaṇ presses hard, because everything downstream depends on it. Nominal positing divides into concordant and discordant. The pillar really is momentary, really is produced, and the two really are inseparable in its nature, so positing their identity-connection accords with the object even though the relation is not in it; positing that the pillar is the barren woman’s son does not. The test is the result: the concordant positing obtains its object, the discordant obtains nothing. Mipham’s formulation of the same point is exact — “without the cause the effect does not arise” holds in the external object; the relation of “without which it does not arise” is posited by the mind (chapter-6-logical-connection).

The longest stretch of the chapter is not about relation at all but about momentariness, because the nature-reason needs it. The reason of non-dependence: whatever has arisen is bound to perish and requires no other cause of cessation — an arrow shot into the sky falls of itself, the force that sent it up needing no further agent to bring it down. The vase said to last a hundred days is caught in a dilemma. If its nature changes over those days, destruction is already proceeding with no hammer involved. If it does not, the first day’s vase and the ninety-ninth day’s are indistinguishable, so either the last can abide another ninety-nine and the vase is permanent, or the last day’s vase should perish today. Alongside this stands the invalidating reason, aimed at a different opponent — the Maheśvara school’s claim that although most things are impermanent, rare exceptions are not, as most crows are black yet a few are white. Sodargye records Sapaṇ’s insistence that these two reasons are not a graded path from preliminary to main practice — the reading he rejects, whose holders the lecture summary leaves unnamed — but answers to two distinct wrong views: one that things perish only at the end and through a destroyer, one that a few things do not perish at all.

Causal connection is ascertained by three stages of positive or negative concomitance — the place is seen to have neither fire nor smoke; fire is seen; smoke is seen arising from that fire — and the first stage carries as much weight as the others, since without it the effect might have been present already or have come from elsewhere. This is where Sapaṇ locates the Cārvāka failure on mind and body: they can neither observe an intermediate state in which both are absent nor see mind arise from a body once present; they observe only that mind changes when the body acts.

Chapter 7: Contradiction (འགལ་བ་བརྟག་པ་)

།ཕན་ཚུན་སྤངས་འགལ་གཅིག་པ་བཀག །དེ་ཡིན་དེ་མ་ཡིན་གྱིས་འགོག །ཕན་ཚུན་སྤངས་པའི་དངོས་པོ་གཉིས། །ཡིན་པ་སྣང་རུང་མེད་པས་ཁེགས།

The contradiction of mutual exclusion negates singularity: “it” is refuted by “non-it.” As for the two real entities said to be mutually exclusive, their both being present on a single basis is refuted by the non-apprehension of what should appear. (summary verse)

Contradiction mirrors connection: connection is one phenomenon not forsaking another, contradiction is one phenomenon impairing another — impairing its arising, its development, or the unbroken continuation of its continuum. Like connection, it is posited before the exclusion-cognition and exists nowhere else. Two kinds, divided by whether the contradictory class is a real entity: non-coexistence (ལྷན་ཅིག་མི་གནས་འགལ་), where both terms are real and cannot occupy one place — heat and cold, light and dark, the wisdom of selflessness and superimposition — and mutual exclusion (ཕན་ཚུན་སྤང་འགལ་), where the contradictory class is a non-entity.

The correction the chapter exists to make is that a pillar and a vase are not mutually exclusive. The early Tibetans, followed by kay-drup and later Gelug masters, held that whatever is a distinct entity is a mutual-exclusion contradiction: the vase is not the pillar, so from the pillar’s side it is a non-pillar, and pillar and non-pillar are exclusive. Sapaṇ’s reply turns on the requirement that the contradictory class here be a non-entity. “Non-pillar” means the sheer negation portion, the absence of pillar; the real entities that happen to be non-pillars are no part of the contradictory class. The consequence of the opponents’ rule is fatal — seed and fruit are distinct, so they would be contradictory, and contradictory things cannot be causally connected, so nothing could be produced at all. The first and second moments of a mind are cause and effect and wholly distinct entities; on their view these too would exclude one another. And dharmakirti states plainly that the Buddha’s speaking and his omniscience, distinct isolates, carry no contradiction (chapter-7-contradiction).

Sapaṇ’s replacement is a three-way division among distinct entities. Where one benefits the other, they are connected; where one harms the other, contradictory; where there is neither benefit nor harm, they are merely distinct — as among people, one you get on with, one you clash with, one who is simply a stranger.

Non-coexistence is posited on the continuum, never on the instant, and runs in three moments: contact, the weaker losing its capacity, and the weaker’s continuum failing while the stronger’s continues. That timing generates the chapter’s hardest objection, which Sodargye flags as the kind of question a non-Buddhist might have brought to the gate-keeper of Nālandā. If the fire-particle and the water-particle coexist in the first instant, they can coexist in the second, and “non-coexistence” is a misnomer; if they cannot coexist even for an instant, then producer and produced fall in one instant and cause and effect become simultaneous. Three earlier answers are refuted — the Sāṃkhya’s, that the weaker becomes hidden within the stronger as stars do by day; the Vedists’, that it transforms while its substance persists; the Vaibhāṣika and Sautrāntika claim that the particles never touch. Sapaṇ’s own answer is that momentary coexistence for an instant or two is not a coexistence-contradiction at all, so the first horn never bites.

For the mind no three instants are needed: the moment the wisdom of selflessness arises, the former trace of self-grasping is cut at its nature. That is the model case of the whole category, and it is why the chapter closes with the ironic verse against those who let wisdom and its object of abandonment coexist for two incalculable aeons — a doctrine, Sapaṇ says, that would be a rare marvel indeed. Sodargye resolves the apparent scriptural difficulty by type: each ground’s wisdom eliminates that ground’s objects of abandonment, and the next ground’s obscuration remains only because its antidote is the next ground’s wisdom.

Chapter 8: Definitions (མཚན་ཉིད་བརྟག་པ་)

།མཚན་ཉིད་གོ་བྱེད་དོན་གྱི་ཆོས། །མཚོན་བྱ་གོ་བྱ་བློ་ཡི་ཆོས། །མཚན་གཞི་གོ་བྱ་རྟེན་གྱི་ཆོས།

The definition is the means of making known — a phenomenon on the side of the object; the definiendum is what is made known — a phenomenon on the side of the mind; the illustration is the support for what is made known.

Definition (མཚན་ཉིད་), definiendum (མཚོན་བྱ་) and illustration (མཚན་གཞི་) pervade every object of knowledge, non-entities included: the son of a barren woman raises an image bearing a feature distinct from other sons, and that image is his definition. All three are posited before the exclusion-cognition, and the early Tibetans’ failure to see this is the root of the debates filling the chapter’s first half.

The recurring threat is regress. If the definition of the ox is hump and dewlap, does that have a definition? Say no, and it falls outside the objects of knowledge, contradicting the opening thesis; say yes, and the series never ends, and the definiendum’s definiendum multiplies until the Dongdongs fill the heavens. Sapaṇ’s answer takes the opponent’s own escape route and turns it round. The opponent had said that in establishing sandalwood as a tree by its branch, the branch needs no further branch, because in establishing it is inseparable from the sandalwood. Very well: the animal possessing the dewlap is named “ox,” and “ox” has no second name, yet is not thereby unknowable, because the argument establishes name and thing as inseparable. Three formal arguments exhaust the matter — the root argument, the meaning-establishing argument, and the convention-recalling argument, “because the epistemologists at the outset named it so.” A fourth is never needed; anyone who will not accept the third is simply dull. The definition of the definition is “the phenomenon that eliminates direct contradiction and establishes the meaning-isolate,” and its faults are three and only three: non-pervasion, over-pervasion, impossibility.

sonam-introduction-treasury summarises Sapaṇ here as following cha-ba’s solution, that the name of a definition should have a definition while its meaning-distinguisher need not. Sodargye’s account gives Cha-ba no such concession and instead has the series halt after the second or third application. The two summaries have not been reconciled against the root text. [translation uncertain]

The chapter’s weight, though, falls on the definition of valid-cognition, and four Indian masters are refuted before Sapaṇ states his own:

།མི་སླུ་མ་ཤེས་དོན་གསལ་གཉིས། །རང་མཚན་རྟོགས་པར་དོན་ལ་མཐུན།

“Non-deceptive” and “illuminating an unknown object” agree in meaning, for both realise the specifically characterised phenomenon.

Devendrabuddhi had made the two synonyms; two definitions wholly separate in essence would require two definienda, and two definienda cannot appear in one mind. Prajñākaragupta had assigned one to ultimate and one to conventional valid cognition; then no single general definition of valid cognition could be produced at all. dharmottara had supplied the removal of non-pervasion from outside the words of the definition, and split “realising the true reality” off as the function rather than the definition — but a definition whose faults must be eliminated by other words elsewhere is faulty as posited, and the split is found nowhere in Dignāga or Dharmakīrti. Śaṅkaranandana had made comprehension of true reality the definition; that overshoots, since neither perception nor inference comprehends an object’s reality exactly as it is, and it fails to reach perception, which cannot eliminate superimposition at all.

Non-deception then has three sides — function, agent, and the acted-upon — and the agent side is decisive. It is what disqualifies the two cognitions expelled in Chapter 2. The guess that there is water in the old well, where water in fact is, does not deceive as to function or object; it deceives as to agent, since the agent neither saw by perception nor determined by genuinely effective reasoning. Subsequent cognition knows its object and fails on the same side.

The chapter closes on how ascertainment is reached. Three kinds of cognition ascertain by their own power: object-establishment, whether by force of function or of familiarisation; self-awareness; and inference, since a three-mode-complete reason needs no second reason behind it. Three depend on another — the first cognition of an unfamiliar thing, the inattentive cognition, and cognition with a deluding cause. Neither party in the older debate had it right; both the wholly self-determined and the wholly other-determined accounts forsake Dharmakīrti’s reasoning, whose treatises state both.

Chapter 9: Perception (མངོན་སུམ་བརྟག་པ་)

།མངོན་སུམ་རྟོག་བྲལ་མ་འཁྲུལ་བ། །རྟོག་པ་སྒྲ་དོན་འཛིན་པ་ཡིན།

Direct perception is free from conception and non-mistaken; conception is what apprehends the fusion of word and object.

The longest chapter. It opens by refusing the split cha-ba and the early Tibetans had made between perception and the valid cognition that is perception, the latter defined as “that which, by eliminating superimposition, comprehends a previously unrealised object.” Sapaṇ refuses it on three grounds: no such division appears anywhere in the seven treatises; “previously unrealised” cannot apply to a cognition that engages only what stands immediately before it, since ranging over past and future is conception’s work; and eliminating superimposition belongs to conception alone.

His own definition has two marks and no more — free from all conception, and non-mistaken. Four perceptions follow, divided by object, support and person: sense, mental, self-aware and yogic. The schools differ on the count, the Sautrāntikas accepting all four, the Vaibhāṣikas three, the Cittamātrins ultimately two.

Mental perception gets the most careful treatment, because its structure is odd. It arises without interval from the ceasing sense cognition, and so has no continuum of its own: each moment is produced from a sense cognition rather than from its own predecessor. It is therefore valid — it illuminates the object’s particulars exactly as the sense cognition saw them — yet it cannot generate ascertainment by its own power, and cannot even produce its own essence by its own power. Mipham’s simile carries the point about ordinary usage: the axe does the cutting, yet we say the woodcutter cut the wood, and so we say “my eye saw it” though the mental consciousness does the real illuminating. Two rival accounts are set aside — Jñānaśrī’s alternation of sense and mental cognition, which leaves no reason why one yields ascertainment and the other cannot, and Jinendrabuddhi’s end-of-the-continuum account, which would leave a complete cause without its effect at every earlier moment.

Dreyfus identifies the self-awareness bridge (Dreyfus: “self-cognition”, རང་རིག་) as Sa-paṇ’s most distinctive positive contribution. Where cha-ba’s new epistemology bridged perception and conception by attributing ascertainment to perception itself, Sa-paṇ preserves Dharmakīrti’s bare-sensing model and locates the bridge in self-awareness, present in both perception and conception, which ensures they bear on the same object indirectly. The move goes beyond what Dharmakīrti explicitly says — internal reconstruction of the system rather than literal commentary (dreyfus-recognizing-reality, p. 397).

Yogic perception carries the soteriology, and here the Treasury’s verses are extremely terse: the substance is the Pramāṇasiddhi chapter of the pramanavartika, on the two accumulations as mutual causes, the times of culmination for the three vehicles, and the Buddha as the sole valid being (buddha-as-pramana). The proof-of-rebirth runs in both directions — the beginningless prior limit, since the substantial cause of mind can only be mind and no opponent can produce a reason why it should be otherwise; and the endless future limit, since the causes are complete and the obstruction, the wisdom of selflessness, is absent.

One passage refutes arguments for the right conclusion, and the reasoning generalises past its occasion. Certain masters — Sodargye does not name them — had established past and future lives by self-awareness, using fire-and-smoke and lamp similes. Sapaṇ rejects both, on the ground that each hands the Cārvāka his own position. The fire-and-smoke simile has a thing of one kind producing a thing of another — exactly the Cārvāka claim that luminous cognition is produced from the non-luminous body of four elements. The lamp is worse: once the oil is spent the light stops, which is precisely the view that after death nothing whatever remains. What is needed is an inference complete in the three modes, not a simile that concedes the point (chapter-9-perception).

Sapaṇ also declines the easy defences of omniscience. That the Buddha knew the worms on the body, that he prophesied Nāgārjuna and Asaṅga, that he had clairvoyance — the Pramāṇavārttika dismisses the last outright, since worldly seers have clairvoyance and so do eagles. Omniscience is instead of two kinds. Knowing all that must be known: one is called omniscient for being unerring in what beings actually need, and “all have gathered to listen” is not falsified by the absence of an ox. And knowing all knowables, established by inference — one who cognises the subtlest and most profound principles without error knows the coarser a fortiori, and speech that accords throughout with the facts, with no contradiction front or back, indicates direct seeing, as a man describing his native place in exact detail must have been there. Sodargye is emphatic that “the Buddha is inconceivable” must never be offered to a non-Buddhist as an answer, since every question would collapse into that one word and nothing would be established.

The chapter’s last topic is the result of perception, where the four rival accounts — Mīmāṃsaka, Vaidika, Vaiśeṣika, and certain Śrāvaka schools — share a single error: all take valid cognition and its result as producer and produced, which leaves the result absent whenever the valid cognition is present. Sapaṇ’s own account makes them prover and proven, distinguished by angle rather than by time. The theory of the mind’s natural luminosity (སེམས་ཀྱི་རང་བཞིན་འོད་གསལ་བ་) belongs here too (sonam-introduction-treasury).

Chapter 10: Inference for Oneself (རང་དོན་རྗེས་དཔག་བརྟག་པ་)

།ཕྱོགས་ཆོས་གྲུབ་ཅིང་འབྲེལ་པ་ངེས། །གཏན་ཚིགས་མཚན་ཉིད་སྐྱོན་མེད་ཡིན།

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

inference for oneself is the seeing of a hidden object by a reason possessing the three modes (ཚུལ་གསུམ་), and the definition serves for inference in general, since inference for others reduces to it. Rival counts of the modes run from one to six — the Vedic school’s four, Īśvarasena’s six — and all fail on one principle: if the three are complete the extras are superfluous, and if incomplete no number of extras helps. Sapaṇ’s two conditions contain the three, because ascertaining the connection is done by two valid cognitions, one establishing on the similar class and one refuting on the dissimilar.

Against ngok-lotsawa and Shawa Chösang, who held the similar and dissimilar classes to be directly contradictory and divided substance from isolate to keep a third class from appearing, Sapaṇ argues that the two classes cannot be determined either way. Test on the substance with “there is fire on the mountain, because there is a grey-white thing”: call the grey-white thing the similar class and every grey cloth has fire; call it dissimilar and grey smoke is not fire’s effect; call it both and the third class has appeared, against their own thesis. The deeper charge is that dividing substance from isolate destroys convention altogether. Conduct proof on the bare particular and nothing can be proven, since the pillar is as it is and no one, the Buddha included, can make an impermanent thing permanent. Conduct it before a bare consciousness unlinked to the object and the mind may declare anything at will. Only the fusion permits ascertainment.

The classification of reasons is the chapter’s technical core: nature-reason and effect-reason on the establishing side, non-observation and contradiction-observation on the refuting side. The four non-observations are of nature, cause, pervader and effect, and each carries a restriction Sodargye insists on — the reason works only where seeing is possible, so “there can be no demon before me, because I do not see one” fails, and so does the materialist’s “past and future lives do not exist, because I do not perceive them.” The contradiction-observation reasons are counted at twelve rather than the sixteen Cha-naba proposed, Sapaṇ rejecting the four that take fire’s unobstructed capacity as destroyer, since that capacity has no continuum and destroys nothing. The fallacious reasons are three — unestablished (six kinds on the object, three before the mind), inconclusive (uncommon and common), and contradictory, the last classified by the speaker’s intent rather than by the essence of the thing.

The last third of the chapter, and by far the largest single block in Sodargye’s lectures, treats common renown (གྲགས་པ་) and the contradiction with it. The apoha doctrine has an unexpected consequence here. Since names rest wholly on the intent to speak and nothing in the object fixes them, common renown divides in two: what is already renowned, which cannot be overturned and does real work, and what can be renowned, which pervades every knowable. The Śabda school had argued that the hare-bearer cannot take the name “moon” because it exists; Sapaṇ turns the reason on them twice, as inconclusive — since whatever exists may bear any name, and so may what does not exist — and as contradictory, since existing is precisely the ground on which a thing may be so named. That “moon” already names an incense ingredient, a Tibetan medicinal flower, and the white bodhicitta in tantric usage is the counter-evidence offered. What the argument secures is not licence but the right ground of convention: to deny that a name may be applied is to incur the harm of the world, and the Buddha permitted the giving of names precisely because without them nothing could be told apart.

The chapter also settles scriptural-authority. Scripture is what is pure in the three examinations — unharmed by perception in its manifest part, uncontradicted by inference in its hidden part, free of internal contradiction in its extremely hidden part — and it is subsumed within inference. Sapaṇ’s interstitial verse states the order plainly: what is established by valid cognition is scripture, and to investigate after accepting is the fool’s way while to accept after investigating is the way of the wise. The corollary is stated too, and it is not a small one: words that cannot be established by reasoning stand on a level with any ordinary being’s self-contradiction (chapter-10-inference-for-oneself).

Chapter 11: Inference for Others (གཞན་དོན་རྗེས་སུ་དཔག་པ་བརྟག་པ་)

།།གཞན་དོན་རྗེས་དཔག་རང་ཉིད་ཀྱིས། །མཐོང་བ་གཞན་ལ་སྟོན་པའི་ངག

Inference for others is speech that demonstrates to others what one has oneself seen.

Inference for others is not valid-cognition at all. Speech is sound, sound is insentient, and valid cognition must be a non-deceptive consciousness; the words are called valid only by transferring the name of the effect back onto its cause, since relying on them an unmistaken inference arises in the hearer. The three parties are proponent (སྔ་རྒོལ་), opponent (ཕྱི་རྒོལ་) and witness (དཔང་པོ་), who respectively prove, refute and adjudicate.

The chapter’s method is reduction, applied twice. Sapaṇ admits exactly four points of defeat, two per side: the proponent may fail to state a proof or may state a non-proof; the opponent may fail to state the fault or may state a non-fault. Against the early Tibetans’ sixteen he argues that whatever among them genuinely is a point of defeat is already subsumed in the four, and whatever is not subsumed is no fault at all. He is particularly firm that the witness has none: both parties appointed him, so whatever he says must count, and punishing him would leave nobody willing to serve. Against the twenty-two of Akṣapāda the same move is made item by item — destruction of the thesis, shifting the thesis, the meaningless, the unintelligible, repetition, evasion — nearly every one turning out to be an inconclusive or otherwise fallacious reason, hence already “stating a non-proof.” The principle behind the reduction is that defeats classified by what people actually say would run without limit, since beings’ discursive thoughts are boundless; what can be subsumed should be subsumed, and only what cannot deserves its own class.

The critique of the Nyāya five-membered proof works the other way, finding the scheme superfluous and deficient at once. Superfluous in three places: the thesis fixes neither the subject-property nor the relation between reason and predicate, so stated it produces no comprehension and unstated it is not missed; the application repeats the reason; the conclusion repeats the thesis. Deficient in the one place that matters more — the example is given but no pervasion is fixed upon it. Sapaṇ’s own system fixes it both ways, forward and reverse, and only then is the question decided. Before a dull hearer a two-membered argument stating pervasion with subject-property suffices; before one who already accepts the pervasion, the single line “sound too is produced” is enough, and thesis and example are wasted breath.

Consequence (ཐལ་འགྱུར་) and autonomous reasoning (རང་རྒྱུད་) are then distinguished, with a warning Sodargye repeats: these are not the Svātantrika and Prāsaṅgika of Madhyamaka, the terms coinciding while the meanings do not. The consequence has its own three modes, the first of which takes the opponent’s admission rather than valid cognition as its reason, and its work is to draw a contradiction out of his own words. Reversed, it acquires the definition of an autonomous reason — but Sapaṇ holds that reversal induces the pervasion and not the subject-property, so the meaning of the autonomous reason is induced and the wording is not. The early Tibetans’ fourteen-fold classification of consequences is rejected as superfluous, deficient, and erroneous together: splitting the unestablished into reason-unestablished and pervasion-unestablished adds nothing, since an unestablished reason leaves no pervasion to speak of; the contradictory reason goes unmentioned throughout; and “reason unestablished yet pervasion established” names something that does not occur.

Four replies to a consequence are admissible, against the Snowland masters’ three — “admitted,” “contradictory,” “inconclusive,” “unestablished” — and the proponent has no licence to ramble beyond them. Collapsing the contradictory into the inconclusive would mislead: answer “it would follow that sound is permanent, because it is produced” with “pervasion unestablished” and the hearer concludes that produced things might after all be permanent, a doubt that should never have been raised. Where a charge genuinely cannot be overturned, following the opponent is the way of the wise, as the non-Buddhist Harinanda followed Sakya Paṇḍita back to Tibet (chapter-11-inference-for-others).

Colophon and stated purpose

The treatise closes with Sakya Paṇḍita’s own account of what he was doing, and it is unusually direct. He names six targets: the Sāṃkhya, the Naiyāyikas of Akṣapāda, the Jains, the Vaiśeṣika, the Lokāyata, and — within Buddhism — some Śrāvaka positions together with a portion of the early Tibetans’ views. He names his basis: the Pramāṇasamuccaya, the Seven Treatises, and their auto-commentaries. Sodargye reads “their commentaries” strictly as dignaga’s and dharmakirti’s own, noting that the Auto-commentary cites no other Indian master and cites the early Tibetans only to refute them.

He also states his method, against a practice he evidently found common: he discards the habit of invoking Dignāga and Dharmakīrti as founders in order to present one’s own position as the secret meaning of the seven treatises, and says he composed instead by deciding through reasoning. The verse claiming that he studied under accomplished masters through former lives and in this one comprehends all the treatises on hearing them once or twice is openly self-praise, and Sodargye says so plainly, contrasting it with Mipham’s habit of calling himself an ordinary being.

The Auto-Commentary

Sakya Paṇḍita wrote an auto-commentary (རང་འགྲེལ་) that is much more explicit in its rejection of realism. There are discrepancies between the root text and the auto-commentary version, leading to significant hermeneutical disputes:

• Some scholars questioned its authenticity
• gyel-tsap claimed passages “conflicting with the root-text” had been “introduced by some ignoramus” (dreyfus-recognizing-reality, p. 26)
• Sakya thinkers defend it as authentic and authoritative

Commentarial History

The Treasure was initially received as a “problematic work” by early Tibetan scholars, who found its antirealism difficult to accept (dreyfus-recognizing-reality, p. 24). Dreyfus’s reconstruction (Intro II, pp. 23–28) is that for c. 150 years after Sa-paṇ’s death, the realist interpretations of cha-ba continued to dominate Tibetan epistemological study; even Sa-paṇ’s own direct disciple U-yuk-pa disagreed with the Treasure (p. 33). The revival begins with Yak-don (1348–1414), whose literal commentary defended Sa-paṇ’s antirealism against the realist mainstream and complained “Nowadays in this Land of Snow I am the only [such informed speaker]” (p. 25). Bo-dong, though nominally Sakya, attacked Sa-paṇ as too nihilistic for denying the existence of generally characterised phenomena (p. 25). The conflict crystallised around the authenticity of the Auto-Commentary (རང་འགྲེལ་): because the Rang-‘grel is much more explicitly antirealist than the brief root verses, and because there are real discrepancies between the two, gyel-tsap argued some passages were corruptions “introduced by some ignoramus” (p. 26); Yak-don, Rong-don, gorampa, and sakya-chok-den defended it as authentic. Interest grew from the early 15th century:

1. Yak-don (1348–1414): wrote a literal commentary defending Sakya Paṇḍita’s antirealism
2. Rong-don (1367–1449): further defence
3. gorampa (1429–1489): definitive Sakya commentary; became the standard reading
4. sakya-chok-den (1428–1507): independent, philosophically rich commentary (The Defeater)
5. gyel-tsap (1364–1432): attempted to reconcile the Treasure with realism
6. kay-drup (1385–1438): modelled his Clearing on the Treasure’s outline, claiming to represent its “true intention”

The Treasure is included in the “Eighteen Texts of Great Renown” (གྲགས་ཆེན་བཅོ་བརྒྱད་) studied in the Sakya tradition.

The teaching tradition is still live, and the modern layer is worth distinguishing from the classical one. Sakya study relies chiefly on Rongtön and gorampa, with Jamyang Loter Wangpo’s manual as the standard teaching aid. Khenpo Sodargye’s eighty-four-lesson lecture series, delivered in Chinese from August 2006 and summarised chapter by chapter in this wiki, reads the text through Rongtön and Gorampa while cross-checking almost every contested point against Mipham’s Great Commentary on the Pramāṇavārttika — a Nyingma khenpo teaching a Sakya text, and a reading of it rather than a neutral report. That is where its distinctive value lies and also where care is needed: the refutations of cha-ba in Chapters 1 and 3, and of kay-drup on the reality of universals in Chapter 3, bear most directly on realism-vs-antirealism, and each is argued from Mipham as much as from Sakya Paṇḍita. Khenpo Jigme Phuntsok’s judgement, reported there, is that the vast wording of the Seven Treatises and the Pramāṇasamuccaya is contained in meaning within this treatise, and that content taking days to work through in the pramanavartika a single verse here can settle.

Significance

The conflict over the interpretation of this text is one of the defining features of the Sakya/Gelug philosophical divergence (realism-vs-antirealism). The debate hinges on whether Dharmakīrti (and Sakya Paṇḍita following him) is genuinely antirealist about universals, or whether his system can be read as compatible with moderate realism.

Pramāṇa and Madhyamaka

The Treasure sits at the intersection of the logico-epistemological tradition and Madhyamaka. The Padmakara introduction to Śāntarakṣita’s Madhyamakālaṅkāra describes it as Sakya Paṇḍita’s “celebrated masterpiece” that “strongly reaffirmed Dharmakirti’s antirealism” against cha-ba’s moderate realist interpretation. This antirealist reading was “accepted as normative by Sakyapas and the other non-Gelugpa traditions” (shantarakshita-madhyamakalankara-intro, pp. 29–30).

Mipham, in his commentary on the Madhyamakālaṅkāra, prefers the antirealist interpretation of Sakya Paṇḍita — an alignment that connects the Treasure to the broader Rimé reassessment of shantarakshita’s synthesis (shantarakshita-madhyamakalankara-intro, p. 30).

Sapaṇ himself marks the boundary repeatedly. The analysis in the Treasury is conventional throughout, and the arguments are withdrawn explicitly at the point where Madhyamaka’s ultimate analysis would begin: the partless instant of Chapter 7 holds conventionally and dissolves under the diamond-sliver reasoning; the contradiction of Chapter 7 and the connection of Chapter 6 are denied as objective relations without any claim that cause and effect fail to occur. Where the sequence is stated as a ranking — in Chapter 1’s summary verse — it runs Sautrāntika, then Yogācāra, then a Madhyamaka in which object and mind alike merge into the dharmadhātu.

Sources

• sodargye-pramanayuktanidhi — Khenpo Sodargye’s eighty-four-lesson lecture commentary: main argument, key claims, lesson ranges, and what it can and cannot be relied on for
• index — the same lectures summarised chapter by chapter; the only treatment in the vault covering all eleven chapters at comparable depth, and the source of the dialectic, named opponents and worked examples in the walk-through above
• dreyfus-recognizing-reality — the Treasure is treated as the central work of the Tibetan reception throughout; Introduction II gives the most thorough English account of the text’s reception history; Ch. 3 on Sa-paṇ on existence; Ch. 8 on Sakya antirealism on predication; Ch. 24 on Sa-paṇ’s critique of the new epistemology; Ch. 25 on self-awareness
• shantarakshita-madhyamakalankara-intro — the Treasure as reaffirmation of Dharmakīrti’s antirealism; Mipham’s preference for Sakya Paṇḍita’s reading
• sonam-introduction-treasury — comprehensive introduction with chapter-by-chapter summary, key root text verses quoted in Tibetan, structural overview, bibliography of Tibetan commentaries