Yoga Sutra 1.07 right knowledge is either direct perception, inference, or authority

Sūtra I.7

Right knowledge is either direct perception, inference, or authority

What then are the five kinds of mental process, tainted or pure? They are right knowledge, illusion, logical construction, sleep, memory. All the mental processes are included in these.

Right knowledge is either direct perception, inference, or authority. The process called right knowledge (pramāṇa, also proof) is divided into just these three, the first division of right knowledge being direct perception. Now the definition of direct perception is given. It is put first because the other two presuppose it.

By a sense channel: the sense referred to is one of the five senses of perception like hearing, not one of the senses of action (e.g. speech) which produce actions but not thoughts. The word could mean either but here he wishes to explain mental processes, so he is referring only to senses like hearing, which produce them. When mind is transformed by a thought, in verbal or some other form, the sense (by which the form is received) is the channel, the gateway. So the mind, through that gateway of sense, is transformed into the form of an external thing, which is both a universal and a particular, and is coloured by it. As a result of that colouring of the mind, the mental process receives an impression (prati-mudrā) as if from a seal (mudrā); the impression is of a thing which is both a universal and a particular, but it is determined mainly by the particular aspect, and this is called direct perception.

(Opponent) A sense like hearing, and the mind, can reflect both the universality (i.e. class) and the particularity of an object; why is this supposed to be concerned mainly with determining a particular?

(Answer) Because rejection and acceptance are concerned with particulars. It is not that the universal is not recognized, but it is taken as secondary. When it is said ‘blue colour’, the main thing is to specify blue, but it is the colour (universal) which is blue, this being a secondary meaning. So ‘the word pain’, ‘the word happiness’, and ‘the word dull’, and the mental process (of perception) are concerned mainly with a particular.

Moreover, perception of universals includes cases of doubt and illusion. Perceiving a universal, one may doubt whether this is indeed the particular he is seeking. Illusion, again, arises from memory of a different instance of the universal he is now perceiving. But doubt and illusion cannot maintain themselves when one has come to know the particular rightly, and so it is said that direct perception is concerned mainly with determining a particular.

Even where one wishes only to determine a universal, whether for instance this is a cow or a horse, inasmuch as one alternative is discarded and the other is settled on, the main thing is determination of which particular universal it is. Everything is both a universal and a particular in relation to other things.

(Opponent) But a universal is only something unreal (a-vastu), projected (adhyāropita) like mirage water on desert ground, and not an object of the senses.

(Answer) If so, there would be no regularity in the universals projected. Nothing can be projected unless perceived before. Someone who has never seen water does not project mirages in the desert.

(Opponent) He might project (adhyāropayet) something he did not remember.

(Answer) If so, one might project (adhyasyet) a quality of sound on to what had colour (i.e. anything could be projected on to anything).

(Opponent) Colour and sound are fields of different senses, so no such projection (adhyāsa) can take place.

(Answer) No; you have said they are unreal (a-vastu), so they can not be distinct particulars. Once you admit distinct particulars, they must be real. Again, unless we accept direct perception with its distinctions of place, time and other circumstances, practical life will collapse. There would be no memories ‘I saw this on that occasion and that at that time, and so I can make out what it is’. There can be no memory of subjects and their predicates unless they have been directly perceived. No one in fact lives by fantasy (kalpanāpodha) remote from the field of practical life. There it is proper to accept perception only as it is generally recognized.

(Opponent) The yogic vision from trans-conceptual (nir-vikalpa) samādhi does not fit into the definition through a sense-channel. Furthermore, the knowledge of happiness, passion and so on has not previously passed through a sense-channel; so some other means of right knowledge must be supposed for them, unless they are to be called direct perception (and the definition amended).

(Answer) The difficulty is avoided by observing that direct perception is an idea (pratyaya) in Puruṣa, there being no mental process of direct perception which is not an idea. It is only as arising from an idea in consciousness (caitanya) that it has the power of perception at all. So it is now going to be said by the commentator, The result is an awareness by Puruṣa of the mental process not distinguished from it. This being so, in the case of the knowledge of happiness or passion etc., whether tainted or pure, it comes down to an idea in Puruṣa not distinguished from these things, so it is certain that they are cases of direct perception.

When it is said that perception is through a sense-channel, this is (not a definition but) merely a corroboration (anuvāda) of what happens in ordinary perception in the world. Otherwise, the knowledge possessed by the yogin, and by the Lord, which does not depend on sense-organs, would not come under direct perception. But in fact, when the limitation of the taints has been removed, there is simultaneous perception of all things, however subtle or remote, by the mind-sattva, which goes to all objects and all fields. It will be said (sūtra III.54) Knowledge-born-of-discrimination, having all, and all times, without sequence, for its object, is called Transcendent (tāraka). So the definition of direct perception is not to be restricted to what comes through sense-channels.

Coloured by an external thing is to rule out illusion.

Doubt is ruled out by the fact that right knowledge is concerned mainly with ascertaining a particular.

(Opponent A) Some say that right knowledge (pramāṇa, also proof) is its own result.

(Opponent B) But others maintain that the result is not right knowledge, but something else, namely a development of right knowledge. As for instance the result of the right knowledge which perceives a jar is an idea (buddhi) of putting it aside, or picking it up, or disregarding it altogether. Since these ideas of picking it up, or disregarding it altogether and so on, are concerned with good points of the jar, or its defects, or absence of either, they are quite distinct from the bare perception of the jar which is the right knowledge of it. Therefore the result of the right knowledge relating to an object is something different, namely a development of the right knowledge, and it relates to a different object.

(Answer to Opponent B) It is not reasonable to say that the result of chopping up a khadira tree is the picking up or throwing away of the khadira wood. For the picking up or throwing away are themselves results of connection or separation, and one cannot speak of the result of a result – that would be dividing the result into two, and there would be a gap.

If results were to have results, it would mean that actions would not be needed, inasmuch as action is something done with a view to a result. If a previous result is going to bring about a result even without action, people will not engage in action needed to bring to fulfilment anything needing much effort. Nor would they search for means, because they would not be needing action.

(Answer to Opponent A) Nor can the result be the action (for instance perception or inference) itself. For (if it were) action would never come to an end. Action leads to the result, and if the result were no more than the action itself, results would not be desired, because action is inherently troublesome.

(Answer to both opponents) Therefore the result is not simply the right knowledge itself, nor is it some other kind of right knowledge different from it. So he says, The result is an awareness by Puruṣa of the mental process, not distinguished from it.

Not distinguished from it, not distinguished from that mental process called right knowledge; it is similar to it, being of its form.

Awareness by Puruṣa, this is Puruṣa’s awareness, for it is Puruṣa’s taking on the form of the process.

A result is not merely the material as it is, independent of any action. On the contrary, the result is the taking on by that material of a definite other state; it is produced by an action and appears with the cessation of the action, arising as the action attains its end. For only with fulfilment in the result does action come to its end. The awareness by Puruṣa of the mental process, not distinguished from it, is the main result. But side-effects – the (changes in a) substance resulting from an action apart from the main action, or some (further) action resulting from the (changed) substance – may allowably also be figuratively called results, since they bring out clearly the main result, by mutual reinforcement.

(Opponents) What you call the result is all purely figurative like that, for you accept as the result something not distinguished from the mental process. Puruṣa is pure, and so its assumption of that form is unreal. When a piece of red lac is put against a crystal, the crystal does not really become red.

(Answer) True. But still the main result is that alone, for it is seen to be so, and we do not see any other main result. Result means all that is produced by actions and agent, and it all ends in this. So it is defined as the result.

(Opponent) If the principal result, namely taking the form of the mental process, is only something unreal (mithyā-bhūta), there would be no desire for it.

(Answer) True. And experience is a purely illusory conformity to the mental process (mithyā-bhūta-vṛtty-anugata). So it will be said (sūtra III.35): Experience is an idea which does not distinguish between sattva and Puruṣa, though they are absolutely separate. And the commentator will say (close to II.6), When there is awareness of the true nature of these two, how would there be experience? For this reason, right vision (saṃyag-darśana) is followed by complete cessation of the mental process.

But those who believe results to be real will not come to investigate them through right vision; there will be no release for them, because the requisite detachment will be absent. And release is not a consummation of action, because that would mean it would be impermanent.

(Opponent) It is not established that there is any cognizer (boddhṛ) of buddhi, or (if there is) that it is Puruṣa.

(Answer) We shall describe later how Puruṣa is the witness of buddhi. Now those who suppose that direct perception is a conjunction (saṃnikarśa) of a sense with its object are wrong in their supposition, because any conjunction is the result of some action. Conjunction is a coming together (saṃyoga) and that is stated in their school to be produced by action. What produces a thing is proof (pramāṇa) of it; what is produced is the result of the producer (and not a proof).

(Opponent) Though it is a result, still as regards knowledge, the conjunction is proof.

(Answer) Not so. For when the topic is the (enlightening) proof which makes a thing known (jñāpaka pramāṇa), it is not proper to bring in a (creating) proof which produces a thing (kāraka pramāṇa). As your conjunction is, on your showing, the cause of producing knowledge (jñānasya utpādaka), it is not what makes that knowledge known. Direct perception and so on are brought forward to explain cognition of the knowable object (prameya), not as a cause for the rise of knowledge. The latter would introduce something irrelevant into the context.

(Opponent) But knowledge is itself something to be known (prameya), and when it arises, it is known; so the conjunction which makes it rise (utpattikāraṇa) is what makes it known (jñāpaka).

(Answer) No, for you yourself admit that knowledge is known only by knowledge, and a conjunction is not itself knowledge, for you have stated it to be what causes the rise of knowledge.

(Opponent) As providing the occasion for it, the conjunction is a cause of the knowledge of knowledge, so it is right knowledge (pramāṇa).

(Answer) Then the opening of the eyes and so on, and the inscrutable will of the Lord (as destiny), could not be refused the status of right knowledge, because they too provide the occasion for knowledge. If knowledge is to be itself known, the only way to avoid an infinite regress is to accept consciousness as the very nature of self (ātman).

(Opponent) The knownness of knowledge is simply as being an attribute inherent in the self, so no other knowledge which would involve a regress is needed.

(Answer) You still need a further knowledge, a direct perception, of the inherence of the knowledge in the self, because that fact is not accessible to the senses. Moreover, since your conjunction is supposed to be what produces knowledge as its result, the production will take time, so that knowledge would not be immediate.

If the knowledge is to be immediate, it can not be the result of a conjunction. And we have already pointed out that such ideas as rejecting are not results of simple knowledge of forms and so on.

Also there is the uncertainty. As when there arises in a number of men the same knowledge of the body of a beautiful woman, in one of them there is the idea of possessing her, in another the idea that she should be avoided, and in yet another neither of these ideas. There is no certainty whether the idea of possessing or the idea of avoiding will arise, though in both cases the knowledge is the same; and in the one who is indifferent, neither of the ideas appears. What is uncertain in its relation to another, cannot be the result of it.

Again, the indifference of the third man is simply a negative, inasmuch as it is no more than absence of avoiding and possessing, and the result of a positive right knowledge should not be a negative. The pair of ideas, taking and rejecting, are from the pair love and hate. Following the maxim ‘no effect without a cause’, without the latter pair there will be no idea of avoiding or taking; but there is still the consciousness of right knowledge, and that is what is called indifference. Since therefore there is this uncertainty about them, the ideas of avoiding and so on cannot be the result of right knowledge.

Furthermore, the ideas of taking and so on being the results of love or hate or absence of either, it would follow that love and hate or indifference would themselves be right knowledge.

(That would be a contradiction) but there is no contradiction when we take the result to be awareness by Puruṣa of the mental process.

(Opponent) It has not been shown how the mental process can itself be a knowable, for what is an instrument cannot be the object – the sickle is not cut by itself.

(Answer) The explanation is, that an action is of a different character from the instrumentalities which bring it to fulfilment. The instrumentalities, united for the completion of the action, operate reciprocally as object and subject for each other. But where a so-called action causes the result by its mere existence, its instrumentality is only figurative and not like that of a true agent. For to say that acting is the same as causing to act would be making the ‘action’ (by mere existence) into an instrumentality, and what is past is not to be described as future.

Further it entails a regress. If action were an instrument, there ought to be another action by which it would be impelled. Without this other action, no instrumentality would be possible for it. This other action would also be the instrument of still another action, and that of another, and so on. Since the chain of actions would never cease, it would never accomplish its result. If it is said that a final one could cause the result without needing any further action, then why not the first one?

But it is not objectionable that the mental process should be knowable, because of the very fact that it is a revealer.

We see that revealers, such as light, are instruments just because of their inherent knowability, and so it is with the mental process too.

(Opponent) The eye and other sense-organs also are revealers, but they are unreliable, because (sometimes) they do not perceive.

(Answer) The reason is, that there is a mingling of two things of the same kind. The senses are physical, and since there is the light of the eye which appears simultaneously with the external light which is of a physically similar kind, there is confusion as to which it is that is being perceived. When the light of several lamps appears simultaneously, it cannot be made out which is the light of which. And so with the ears and other senses.

(Opponent) Then in the case of the inner sense, which is not physical, it is a revealer of the same kind as the mental process, since that too is a revealer. So the mental process will not be perceived.

(Answer) Not so. Thanks to the consciousness (caitanya) with which it is endowed, it is not separate or distanced (from consciousness). Self (ātman) is all-pervading, and being subtle and extremely pure, has perception of the mental process because the mind is not distanced from it. Therefore it is established that Puruṣa, taking on the form of that very mental process of right knowledge, is aware of it. It is that being known which is the result.

Some maintain that a mental process is not directly knowable and there is no means to perceive any result (of right knowledge). For a result to be perceived, some unknown (means of) knowledge must be presumed. (They have to say that) no result is perceived, because they accept no proof that could operate to perceive it. And if one were accepted, there would be a regress. A right knowledge assumed to perceive the result (of right knowledge) will itself have a result, for which another right knowledge will be needed in proof, and there will be no end of the chain.

On their view, since there is no right knowledge of it, it must follow that this result does not exist, because there is no proof. Where means of proof which would reveal some existence do not go into operation, we assume that the thing is not there.

(Opponent) There could still be a right knowledge of the result though not itself perceived.

(Answer) No, because that would go against its character of being a revealer. Though seen to be revealing, your revealer would not (itself) be perceived. But a torch, even on the peak of a mountain, never loses its innate character (of being visible).

(Opponent) The right knowledge, being simply a revealer, reveals its result also.

(Answer) That confirms what we are maintaining. For one has perceptions of love and hate as particular mental processes. ‘I love’ and ‘I hate’ are each a particular process of the I-idea (aham-pratyaya); the perception of the mental process of love and so on is by perceiving oneself as having or not having them. Otherwise there would indeed be necessary some other means of knowledge to know them, as if they were external forms. And then there would be a regress.

So refusal to recognize that right knowledge has a result (i.e. being witnessed by Puruṣa) is arrogant obstinacy and should be disregarded as vain argument.

What is to be inferred, all about what one seeks to know; inclusion, as being like them in characteristics and nature, and exclusion, since it is not found there. Relation: what is this relation, and what two things does it relate? For a relation does not relate without two things to be related, as is clear from the sentence beginning with the relative pronoun which.

(Opponent) This inference is well known to be simply a relation between an indicatory mark and its possessor.

(Answer) No, for that would not suffice for drawing a conclusion (gamaka). A mere relation does not in itself lead to a conclusion. The relation with the mark has to be one of exclusion or inclusion. This is a question of the mark; it is no more than having or not having the mark, and that does lead to a conclusion.

(Opponent) But the mark alone is not thus conclusive without reference to a possessor of it. If a mark is to be taken as conclusive without reference to a relation between it and its possessor, then anything at all would be conclusive; so the relation should have been specified as relating these two.

(Answer) He has specified it, in the words which has as its field inclusion among things of a like class to what is to be inferred. Unless it were connected with what is to be inferred, namely the possessor of the mark, it would not be included among things of a like class, or excluded from things of an opposite class. So the clause beginning which confirms the generally accepted relation.

A mental process having as its field the things related and their relation, specified positively and negatively (anvaya-vyatireka), mainly concerned with determining a universal, a universal possessing the indicatory mark.

The matter is clear merely from the definition. But to give instances and counter-instances demonstrates very plainly that there are actual examples of it. When it is said that the moon and stars have motion, it means that having motion is to be inferred in them. A corresponding (universal, namely) getting to another place, appears in things of like class which have motion, such as the man Caitra, and this excludes the stationary things of a different class, such as the Vindhya range. The relation is certain, because having motion is constant (in the examples of getting to another place).

So a relation, invariably in the particular class, can inform one of another thing related, as in the case of cat and mouse (vadhya-ghātaka, hunter and prey) in causal relation.

(Opponent) Allowing that between things which are not mutually exclusive there can be a permanent relation such as getting to another place, which leads to a conclusion, but there is no relation between cat and mouse, which are incompatibles, so how can you say there is anything to lead to a conclusion?

(Answer) The argument holds because there is a permanent relation between them (though invisible). Even in the case of the heavenly bodies, the feature of ‘getting there’ is not directly manifest to us (being too slow): if it were, the conclusion that they have motion would not be a case of inference (sāmānyato dṛṣṭa) at all. It is not in the moon and stars that the indicatory relation of getting there is directly perceived (though it is perceived in the case of the man).

(Opponent) If a relation can be conclusive even though not directly perceived, then everything would be conclusive as to everything, for there is nothing which is not related to everything else, since they are all in universal space.

(Answer) That is why the relation, in the field of inference, is only what suffices to cause the rise of the idea bringing the desired conclusion (gamya-gamaka). With the cat and mouse, from seeing one of them, one concludes the absence of the other, which is a pure inference. From seeing that a thing is not in its usual place, it is concluded that it is somewhere else, just as from the sight of the cat one infers absence of the mouse. There is the familiar example, ‘when Caitra, a living man, is seen not to be at home, he is out.’

(The discussion now becomes technical. The opponent wishes to show that besides the three mentioned in the sūtra there are other means of right knowledge. One of these is cognition of absence, and a classical example referred to several times is this: a man looking for a jar glances into a bare room and concludes, ‘the jar is not here’. The opponent calls this direct cognition of absence of the jar. Śaṅkara brings it under an inference from the sight of the bare floor, making it an idea and not a direct perception.)

(Opponent) Surely these are (the proofs postulated by some schools called) cognition of absence (a-bhāva) and presumption (arthāpatti), and not inference, for they do not depend on indicatory features like the horns of a cow for instance.

(Answer) No, because they do have indicatory features, namely the ones stated. The relation between cat and mouse has the feature of mutual exclusion. Wherever the cat may be, the mouse is never there – this is the invariable connection. The presence of the cat invariably means absence of the mouse, so also presence of the mouse means absence of the cat; it is like the presence of smoke and the inferred presence of fire.

(Opponent) How does the idea of the absence of the mouse from its usual haunts come to be known from the mere sight of the cat there? The operation of the means of right knowledge of existent things is in regard to the cat alone, and not to the mouse.

(Answer) From the mere right knowledge operating in regard to the cat’s presence, the idea of the absence of the mouse arises, as from the mere right knowledge operating in regard to the presence of smoke there arises an idea of the presence of fire. Right knowledge operating as regards the specific characteristics of the cat gives rise to the idea of the absence of the specific characteristics of the mouse, just as the idea of the specific characteristics of fire arises from right knowledge operating in regard to the specific characteristics of smoke.

But for the one who argues for absence as a right knowledge, that knowledge of absence which he claims could never be specific; following up the sight of one of the incompatibles (like the cat) does not give actual sight of the absence of the other one (the mouse). Whereas to follow up a particular indicatory mark is as it were sight of the particular possessor of the mark. There is nothing specific about absence of the operation of right knowledge capable of revealing existent things, through which an idea could be formed of anything specific.

(Opponent) There is a specific thing which one wishes to know (for instance, whether the jar is there or not, and one knows it) from absence of any right knowledge of it.

(Answer) No; a non-existent can have nothing to do with associations of memory or will. What is non-existent like a hare’s horn is never found producing an effect by association.

(Opponent) We find it now.

(Answer) No; the two things are on quite a different footing and it is not an example.

(Opponent) Well, there is a similarity.

(Answer) No, because in the case of the jar we see that it is real.

Nor is it reasonable that absence, which does not function because it is not a thing, should exist as a means of right knowledge. For in the world, what has no function is never found bringing about an effect. Therefore the idea of the absence of a knowable object, arising from a right knowledge capable of revealing it, is a knowable object like the idea of a jar; and because it is a fact, it causes some action or cessation of action. A counter-example is a non-existent like a hare’s horn. A right knowledge of absence, supposed to be produced by absence, would not in fact come into existence at all, for we see from the examples given that a thing is like what is known to be its cause.

By the fact of being present, the sight of the cat is the indicatory mark for the idea of the absence of the mouse, and like the idea of smoke which causes the idea of fire, it dispels any doubt. Or one could say that the idea of the absence of the mouse, arising from the indicatory mark of the cat there, has a quality of certainty, like the idea of fire (from the indicatory mark of smoke).

(Opponent) The assertion ‘the jar is not here’ is not an inference, because it is not arrived at from an inferential relation of indicatory mark and its possessor. For the idea of fire is a following-up of an indication like smoke, but when it is said ‘the jar is not here’, there is no following-up of any such indication, because the idea is already there without it.

(Answer) Our position is not wrong. It is an idea in the form ‘here’ that is the indicatory mark. Following up the indication does not have to be from a previous idea of it. The mental impression in the form ‘here’ is the indicatory mark for the idea of the absence of the jar, the affirmative impression being excluded. Just as with the idea of the cat, here too the relation is one of exclusion. Where the things that cause the impression are not mutually exclusive, there is the idea ‘it is’. It is because of this invariable relation that we say that what is called presumption (arthāpatti) is in fact merely inference.

In the case where the absence of a living man, Devadatta, in the house, means his presence outside, there is an inference both ways: when he is in, then he is not out, and when he is seen outside, this is his not being in – that is the pattern. The cause of the perception of absence is thus explained as the fact that something which could impress the mind is not impressing it.

The knowledge of powers like burning in fire, is mere inference. For in experience, we find in ourselves an unvarying relation of result and what causes it. It is seen in our own self: ‘I can effect this result’ or ‘I cannot’, which is a cognition of power specific to oneself as possessed by the self (ātman). The result is the exercise of the power of burning (or the latent powers in oneself), and this is certainly an inference.

(Opponent) Some give this instance of direct perception of something not there: Suppose that an object lit by some light is being viewed with the help of a mirror (placed behind it), though not of course as a unitary whole. Now what is seen (in the mirror) is not actually there, so that the same illuminating direct perception of what is there (the front of the object) also illumines what is not there.

They give this case also: someone is told, ‘Fetch the robe which has no pattern on it.’ On going to the store-room he finds a robe with a pattern next to one with no pattern. Seeing the patterned one, he knows by direct perception the robe distinguished by not having a pattern, and brings it. In this case there is no other means of right cognition except absence (through which he could know the robe).

(Answer) Not so; that would be to deny that objects are known through the mind’s taking on their form. And it is generally accepted that whatever the mind takes the form of, that is known as it is by direct perception. In the case of a blue thing’s being yellow (to vision affected by jaundice), the idea of the relation of the characteristic and its possessor turns out to be illusory (mithyā). In the same way, when it is said that the jar is not here, the idea of a relation of non-existence of the jar to the location Here is an idea of an unreal relation; no relation either acquired or inherent is conceivable between the absence of the jar and the location Here. Whereas when it is said that Devadatta is here, there is a relation. Therefore the idea of absence, like the blue thing’s looking yellow, must be illusory as an idea, even though there is a notion of a relation of characteristic and its possessor.

If (you still say that) there is a relation between the absence of the jar and the location Here, then with a thing which is not blue, you will have a relation of non-existence of blueness, referring to an idea of a relation of blueness between the two things. But a relation is something that actually exists, and it is a contradiction to have this ambiguity about it, for you admit that absence is not an actual thing.

(Opponent) It is said by the grammarian, ‘One hundred and one are the meanings of the genitive case’ (as in absence of the jar, for instance).

(Answer) But then relationships could not be only of the two kinds called inherence and conjunction (as your school holds).

(Opponent) Let us say that the hundred and one relations of the genitive (including absence of the jar) can be subsumed under these two.

(Answer) No, for in the case of absence, it cannot belong to either.

Further, when it is said ‘Here in the sky there is no flower’, we get a relation of the sky which is invisible, and the absence of the sky-flower which is also invisible. It is all invisibility alone, and like the absence of flowers falling from the sky, so in the location Here on earth, the absence of the jar is invisible. It has to be said how it would be distinguished.

(Opponent) (The cause of the anomaly is) because space is not visible (whereas earth is).

(Answer) Not so. For if a flower does fall in space, we see that it exists. So the reason cannot be that space is not visible. Just as the invisibility of space cannot be a cause of the real existence of a falling flower, so it cannot be a cause of perception of the opposite of real existence.

(Opponent) The cause is simply the invisibility of space.

(Answer) Then the invisibility of anything concealed, though unrelated, would also have to be a cause.

(Opponent) The absence of sky-flowers is directly seen.

(Answer) That would mean that the connection of sense with its object would be unnecessary (for perception), and if so, it would follow that everyone would have extra-sensory perception and be omniscient.

(Opponent) Well, we admit that absence of sky-flowers is a case of inference (but not absence of the jar).

(Answer) As with the absence of the sky-flower, there is no distinguishing feature of ‘absence of a jar here’; so it is certain that it is a matter of inference. Even for the proponent of absence as a means of right knowledge, the absence, for instance, of fire in waters concealed or remote must be a matter of inference; so the absence of a jar here, being of the same class, is knowable only by inference. This is the reasonable conclusion.

Another case of it is, that the Vindhya range has no movement because it does not get anywhere. The assertion is, that the Vindhya has no motion; the reason is, because it does not get anywhere. Taken with the case of Caitra, inference is shown working both ways, like the one-eyed crow (traditionally able to look all round). The relation is positive in the case of the moon and the stars having motion, and negative in the case of the Vindhya’s not getting anywhere. It is thus demonstrated by both agreement in presence (anvaya) and agreement in absence (vyatireka), and the definition is complete.

Again, these applications show that illusion is ruled out (from the definition), and demonstrate that the definition has no defect because it applies equally to a different mark and its possessor.

Someone competent motivated by goodwill to the other free from any defect (doṣa), his object being to tell something seen or inferred by him, intending to convey to the other to that particular hearer his own knowledge his own experience; by word: this here means a sentence, from which the particular meaning emerges. It is from the sentence that it is conveyed. The mental process whose object is the meaning of the sentence is mainly concerned with determining a universal, as in the case of inference.

Authority for the hearer: this sort of knowledge has been preceded by the original knowledge of the thing possessed by the speaker, for whom it is not authority, since he has in mind something seen or inferred by himself, so that his own knowledge has come through the senses or from some indication. Authority is not direct perception because it is not something within the field of the senses, and it is not inference because it does not depend on any such relation of indication and indicated as described above. For the knowledge arising from the meaning of the (scriptural) words ‘heaven’, ‘apūrva’ (the invisible power of the sacrifice which produced a later effect), and ‘divinity’ (of the sacrifice) does not come from some previously established relation of indication and indicated.

But as always with right knowledge, the result is an awareness of the mental process by Puruṣa.

When a speaker says questionable things: when from his utterance, the mental process called authority arises, the result brought about by the operative statement is as before. The reference is to what is merely lacking in proof (i.e. not seen or inferred by the speaker himself), but that is not meant to exclude what is actually untrue, the operation of the statement being the same in both cases.

The authority of what is said by a speaker of something questionable is, however, to be distinguished. What is this something questionable which he speaks? It is something neither seen nor inferred by that speaker – for instance a Buddha or Arhat, since their teaching is other than what is well known to informed people (śiṣṭa). For a mind which is not infected with the taint of passion and so on will not describe otherwise what is well known (as true). This is fallacious authority being only a semblance (ābhāsa).

Though he has not spoken in so many words of semblance in the cases of direct perception and inference, by now pointing out that authority can in some cases be only a semblance, he implies that semblance is to be recognised as possible also in the previous cases of direct perception and inference, when their presentations are contrary to fact.

But if an original speaker has seen or inferred it: if it derives from the Lord as the first speaker, there is no reason to doubt, for his authority as a speaker of truth is unquestionable.

(Now the discussion turns to analogy (upamāna) which in some schools is rated as an independent means of knowledge. Śaṅkara denies this, to follow the sūtra which does not mention it.)

Analogy (upamāna, also comparison and resemblance) presupposes words, and therefore is not a separate means of right knowledge but comes under authority.

It occurs when having heard from a forest man ‘a buffalo is like a cow’, one sees a buffalo in the forest. Analogy is simply the memory of the similarity learnt from the previously heard sentence, and nothing more. The fact of similarity must have been previously learned from the words of the forest man.

Analogy is not simply learning the relation of name and named, though it too is grasped only from some sentence about it. One might see a buffalo in the forest and think ‘it is a cow’, because until he has come to know the word buffalo, he is not aware of the relation of name and named so as to think ‘this is a buffalo’. The very word buffalo must refer back to some sentence about it: (in the case of analogy now discussed, the sentence was) ‘a buffalo is like a cow.’ It is clear to him only when he has previously heard a sentence about it.

(Opponent) But he has had no previous idea of what a buffalo is.

(Answer) You seem to be suffering from an inability to grasp the thrust of the argument – cornered like some wretched dog. (This sentence has been heavily amended by the editors and is perhaps corrupt – Tr.)

To insist that there must have been some previous experience would entail many different cases of resemblance, differing by closer or remoter association with various times and places and circumstances.

(Opponent) Let it be so.

(Answer) A previous idea of the thing is not the mass of associations formerly perceived; they are not the thing itself. For when the word ‘cow’ is spoken, one can always understand the meaning from the word itself without specifications of place and time and so on. The mere knowledge of the relation of name and named is not analogy, for analogy is something by which things are compared, being a knowledge of similarity. In comprehending the relation of name and named, no comparison of anything is being made; that comes afterwards. It is after hearing about the resemblance (from the sentence of the forest man) that he too makes the judgement, ‘this is a buffalo’. So this (suggestion that analogy is nothing but the relation of name and named) fails.

(Opponent) Analogy could be an independent means of right knowledge, not consequent on words; the townsman, on seeing a buffalo in the forest, has the idea ‘this is like a cow’.

(Answer) Not so, because there is no object which it is sought to know (prameya). It is when something is being sought that the appropriate means of right knowledge go into operation to reveal something existent. But there is in fact no one similarity covering the points of resemblance and difference of cow and buffalo.

(Opponent) There is the idea of resemblance, and this must have some basis.

(Answer) True, and it is here that a distinction has to be made: (a) is this like the notion (pratyaya) of a collection, such as a forest? or (b) is it like the idea (buddhi) of a thing like a jar?

Suppose it is like the notion of a forest, where the individual trees making up the pattern of growth are the causes of the notion ‘forest’, which does not exist apart from them. Just so the members (of the patterns) ‘cow’ and ‘buffalo’, through memories of the separate herds that have been seen, are the cause of determining a similarity, which does not exist apart from them. Therefore we arrive at something unreal.

Then let it be that there is here a factuality and demonstrability like the notion of a jar. It is now to be asked: what sort of fact this is. (There are these possibilities:)

Is it a single similarity, conforming to cow and buffalo, but distinct from cow-ness and buffalo-ness?

Or if they are distinct from each other

If (there are two similarities) distinct from each other, located in cow and buffalo but apart from them, are these qualified or not qualified by each other?

Is it that an individual has characteristics of the other class?

Is it that a class has characteristics of an individual of the other side?

Or is (the similarity) a class with characteristics of individuals of both sides?

Again, still further relations between them ought to be postulated with either one being the qualified and the other one the qualifier.

Of these cases, if there is a single similarity applying to both sides, then there is nothing to be compared, because it is just a single thing, as if one should say ‘a cow is like a cow’.

(Opponent) By the similarity, one of two individuals may be compared.

(Answer) Not so: that would be referring to itself, as in ‘like a cow because of cow-ness’.

(Opponent) One individual may be compared with another like it.

(Answer) Not that either, for cow-ness is not a basis for likening a young calf with half horns and a hornless cow.

(Opponent) By considering where their characteristics differ, the similarity can be made a basis for the idea of comparison.

(Answer) Then the similarity perishes. If a differing characteristic is to be the basis of the idea of comparison, you have proposed a destroyer of the similarity.

Furthermore there would be alternatives, such as whether a characteristic under consideration is single, having similarity to both sides, or divided, and there is just the same impossibility as before.

(Opponent) Let us say that the similarity is divided.

(Answer) The very assumption of a division refutes it: a horse is not like a cow. If things are taken one by one, the innumerable considerations become unmanageable. The previously proposed relations between class and individual are refuted in the same way.

Now take the hypothesis that similarity is an imaginary abstraction consisting of bringing out one idea by ignoring any opposing ideas. In that case, how is it that the notion of the spaces in the forest, in front of the trees and to their right and left, is not taken – by ignoring everything else – as a thing (as the trees are taken as a forest, in disregard of the spaces)?

(Opponent) Let it be that the idea of the forest is merely the trees which cause the idea of it.

(Answer) If so, ‘a cow is similar to a buffalo’ is merely an illusory notion (mithyā-pratyaya), as the idea of a forest is.

(Opponent) Well, if as you have said opposites cannot be compared, how is it that a horse is likened to a cow (as a domestic animal, for instance)?

(Answer) Both sides have parts, opposite or corresponding, in themselves which are causes of the idea of analogy, and it comes to the same thing as the notion of a forest.

The point need not be laboured too much: these many parts observed on both sides are the cause of the idea of similarity, as the trees cause the idea of a forest, and so it is established that there is no independent means of right knowledge called analogy (but it is included under authority).

(Reverting to the general point of authority as a means of right knowledge:) Nor should we imagine that since what is knowable is of just two kinds (perceptible or imperceptible), the means of right knowledge must be of just two kinds also (perception and inference, as Buddhists hold). Why not? Because an object is known as determined by the right knowledge, and it is not to be supposed that the object determines the right knowledge. If each object determined the knowledge of it, then everyone would be omniscient. In (the sacred declaration) ‘From worship in the temple comes heaven’, the idea of worship as the cause of attaining heaven is not in the field of perception or of inference, and it would be illusory unless one supposed a separate right knowledge for it.

(Opponent) This idea could be a corollary of an inference.

(Answer) If it is only a corollary, it would not itself be in the class of a right knowledge; in that case, the same would apply to the idea of what is referred to in the (sacred Buddhist) Three Refuges and so on.

In any case, the purpose here is not to analyse right knowledge. The main thing is to point out the accepted means of right knowledge, and other mental processes, with a view to inhibit them. The author of the sūtra-s, therefore, did not make further sūtra-s on them, and the commentator merely added something to confirm what is generally accepted. What objection can there be to following him?