Abhāvapramāṇa and error in Kumārila's commentators

ABHĀVAPRAMĀNA AND ERROR IN KUMĀRILA’S COMMENTATORS

Elisa FRESCHI

1. Background

The starting point of the following study is a passage of the $\dot{S} V$ (Nirālambanavāda 117cd-118) on erroneous and negative cognitions. Apart from the interest of the subject itself, this passage constitutes an intriguing opportunity to compare some of the main commentators on the $\dot{S} V$ on a controversial issue. After a short introduction on cognition of absence and on error according to the Bhāṭtamīmāṃsā, this study will focus on the interpretations offered by Kumārila’s classical commentators and by his modern interpreters. ${ }^{1}$

1.1 The role of abhāvapramāna

The Bhāṭta school of Pūrvamīmāṃsā is the only philosophical school in classical India which accepts absence as an instrument of knowledge. Keeping in mind the stock-example “There is no pot on the floor” all other schools understand absence either as the sheer perception of something else (so Nyāya, Prābhākaramīmāṃsā, Śāntarakṣita, ${ }^{2}$ etc.), or as an inferential judgement (so the Buddhist

^[1]

epistemological school). ${ }^{3}$ The reasons for the different stance of Bhāṭtamīmāṃsā are still debated.

According to Birgit Kellner, though there are some grounds to argue for the assumption of absence as a distinct instrument of knowledge, none of them is really compulsive (Kellner [1997: 70-72]). Kellner closely considers the many dissimilarities among the instances regarded by Mīmāṃsakas to be applications of absence as an instrument of knowledge and argues for the inconsistency of absence as an instrument of knowledge in the Mīmāṃsā before Kumārila: “Why did Kumārila accept the Non-present-being as a separate instrument of knowledge? The apparent answer would be that such acceptance was at his time already widely admitted - Prabhākara, in fact, calls it after all a prasiddhi”. ${ }^{4}$ She then examines Kumārila’s dealing with this instrument of knowledge and underlines his effort to build out of such dissimilar instances a single theoretical frame which he then applied also to other, similar, problems:

The fact of gathering together such disparate cases under a single epistemological ‘umbrella’ leads to the conclusion that Kumārila clearly updated the range of problems to which one should, in his view, apply the form of abhāvapramāna he inherited. In this way, he drastically widened the sphere of application of the abhāvapramāna. ${ }^{5}$

On the other hand, Taber notices how absence as an instrument of knowledge fits into the general picture of Kumārila’s philosophy, insofar as it constitutes a further instance of trusting cognitions as they appear: “[T]he well-known doctrine of intrinsic validity (svatah prāmānya) […] is conductive to accepting the reality of the things we cognize; for to say that a cognition is true is to say that its object is real. […] Thus it seems natural for Kumārila to want to hold that the cognition that something is not must apprehend a real entity” (Taber

^[1]

[2001: 76-77]). Kumārila, maintains Taber, admits absence as an instrument of knowledge within his epistemology chiefly because it cannot be reduced to either inference or direct perception (Taber [2001: 76, fn 13]). At this point it is worth remembering that Bhāttamīmāṃsā, like Nyāya, acknowledges a category called “absence”, and that this can be grasped only by absence as an instrument of knowledge according to Kumārila’s school (whereas in Nyāya it is grasped by perception). Absence is not an ontologically distinct entity, as this would be paradoxical, it is instead a “category” like, for instance, quality or inherence, that is, a modality of reality. Hence, absence as an instrument of knowledge has no ontologically distinct object although it is the only instrument of knowledge which can grasp objects according to their absent aspect, that is, according to the aspect of what they are not. Perception or inference can know a bare floor, but only absence can know it according to the absent aspect of its non-being a pot.
1.2 A few words on the Bhātta analysis of perceptual error

Perceptual error has been often used in classical Indian philosophy as an argument against realism. Since in the case of perceptual error, it is argued, cognitions are found to have no external object supporting them, why should one assume the existence of an external world according to the way we seem to perceive it? So, Kumārila’s analysis of error primarily attempts to refute the possibility of a support-less cognition. In the words of one of Kumārila’s main commentators, Pārthasārathi Miśra:

In every case [of perceptual error] the connection appears in fact to be non-existing, but the connected elements do exist. ${ }^{6}$

Let us consider one of the standard examples of perceptual error, the mother-of-pearl mistaken as silver. Though they are falsely connected, both connected elements are real. Mother-of-pearl is real, although it is perceived just vaguely as something bright and lustrous. Silver is also real, although not directly perceived in this case. Another commentator on the $\dot{S} V$, Sucarita Miśra, deals with the case of water appearing in a mirage as follows:

^[1]

But that [water] does exist elsewhere. In fact, the real existence of water, which is grasped in the external world, is not invalidated, even by means of the succeeding invalidating cognition [in the form “this is not water (but a mirage)”]. In fact the [succeeding] opinion is not “that water does not exist [at all]”, but “this is not water”. What does [this amount to] say? “Here there is no water”. Therefore, only the connection here, that is, of a certain place, with water, is negated, [and] not water [itself]. ${ }^{7}$

Thus, the content of erroneous cognition is the non-existing relation of two existing items, while that of cognition of absence is the non-existing aspect of an existing item. Apart from this parallel, in the Bhāṭta epistemology there is no direct link between them. In answering those Buddhist objections that use both instances as evidences of the unreliability of direct realism, however, Kumārila has dealt with arguments relating to erroneous cognitions and arguments relating to absence-cognitions close to each other in the same section of the $\dot{S} V$. His commentators have further speculated on this topic, but as far as I know no one of them has clearly defined each other’s boundaries. This could depend on their awareness of the two cases as altogether heterogeneous (and, thus, not in want of any further distinction). In fact, no classical commentary seems to imply that error consists in a misapplication of absence as an instrument of knowledge. On modern scholars’ opinions diverging from this view, see § 5.

2. The root text: Kumārila’s ŚV Nirālambanavāda 117cd-118

In this section I will try to translate Kumārila’s text as neutrally as possible, so that just the questions it raises can be noticed. Their possible solutions will be examined in the next paragraphs, dealing with Kumārila’s commentators. Kumārila’s relevant statement reads as follows:
tasmād yad anyathā santam anyathā pratipadyate II tan nirālambanaṃ jñānam abhāvālambanaṃ ca tat I bhāvāntaram abhāvo 'nyo na kaścid anirūpaṇāt II (ŚV Nirālambanavāda 117cd-118).

^[1]

A preliminary translation will now be given, only for the sake of reference:

Therefore, a cognition that knows something to be other than what it is, is supportless, and that has an abhāva-support.
Absence is just another being, nothing else, because it is not discerned.
The textual problems I will deal with in the next pages are the following ones:
(i) The preceding context

The initial “therefore” should sum up the whole preceding discussion on illusory cognitions, etc.
(ii) The odd mention of nirālambanam

The oddity of Kumārila’s mention of “supportless” (nirālambanam), which should instead be altogether denied by him.
(iii) Do the two tads share the same referent?

Do tad … tad in 118a and 118b refer to the same cognition that is described in 117 cd as cognising something to be other than what it is?
(iv) Is the ca copulative?

Connected with the above is the question about the meaning of $c a$ in 118b; is it copulative or adversative ${ }^{8}$ ?
(v) The meaning of abhāva in 118b

Does it mean “absence” or “non-existing”?
(vi) anirūpanāt

The final anirūpanāt could have more pregnant a meaning than expected.

3. Uṃveka Bhātta’s commentary

Uṃveka is believed by some scholars to be a direct disciple of Kumārila. He must have lived in the 8th century, before Kamalaśīla,

^[1]

who happens to mention him. ${ }^{9}$ His is the oldest extant commentary on the $\dot{S} V$. Nevertheless, this does not mean that it is necessarily the most reliable. As regards, for instance, the crucial topic of the self validity of cognitions, Taber has argued that Uṃveka’s understanding should be rejected in favour of Pārthasārathi’s one (Taber [1992]). More in general, it can be suggested ${ }^{10}$ that Pārthasārathi’s very distance from Kumārila allowed him to reconsider his philosophy with the help of some centuries of reflections and hermeneutic efforts. Moreover, Pārthasārathi’s commentary is a line-to-line one and this ensures for it a privileged position as an introductory tool into the $\dot{S} V$. Uṃveka’s $t i t k \bar{a}$, on the other hand, is often characterised by a terse and complex style.

As for our present subject, let me mention an instance of the gap between Kumārila and his first commentator. One of the main problems in appraising the role of absence as an instrument of knowledge in Kumārila’s epistemology consists in understanding its range of application. Kumārila possibly borrows from the Vaiśeṣika system the distinction among four different kinds of absence (previous, posterior, reciprocal and absolute absence, see $\dot{S} V$ Abhāvavāda 2-411). Strangely enough, the standard example of absence, the absence of a pot on the floor, does not fit properly in this scheme. ${ }^{12}$ In fact, the pot’s absence could be interpreted as previous or posterior absence, but what about a floor where a pot has never been placed? This could not either be though as a case of absolute absence (atyantābhāva), since it does not fit with the stock examples of absolute absence, where, e.g., horns are supposed to be necessarily absent from a hare’s head. The scheme, hence, seems to have been conceived to deal with cases different to the one which later became the standard one. Unfortunately, no intermediate stage in this development can be traced, since Uṃveka already mentions the pot as if it were the obvious example. The following is his commentary on

^[1]

Nirālambanavāda 117cd-118:

Abstract

How then is it that, although the fact of having an external support is common [to all cognitions], dreaming cognitions, etc., ${ }^{13}$ are said to be supportless and having a support which is not [there] (asat’)? In reply to this doubt he said: “Therefore” etc. ( $\check{S} V$ Nirālambanavāda 117cd-118ab). Even you (Buddhist opponent) must admit only this (cognition) in which something appears differently [as it really is] as the “cognition having a support which is not [there]” (abhāva), but not the cognition like “here in this place there is no pot”. Since this is admitted to have an existing thing (bhāva) as support, as it is supported by the surface of the floor devoid of a present pot. This is what he says with the hemistich beginning with “Another being” ( $\check{S} V$ Nirālambanavāda 118cd). ${ }^{14}$

That is, Umveka interprets 117cd-118ab as answering an implicit question, and reads kārikās 117cd-118 as follows (more details below):

Therefore, a cognition that knows something to be other than what it is (,that is, an erroneous or illusory cognition), is supportless, and it has a support which is not [there].
Absence [instead] is another being, nothing else [hence, cognition of absence is not supportless], because it is not discerned [as such, i.e., it is never ascertained independently of what exists].

Let us now look for Umveka’s answers to the questions listed above.
(i) The preceding context

The initial “therefore” is not explicitly commented upon by Umveka, but it fits nicely in his interpretation of 117 cd as answering an implicit question. More in general, Umveka’s mentioning “dream cognitions, etc.” shows he understands these verses as closely related with the preceding ones.
(ii) The odd mention of nirālambanam

Umveka does not deny that dreaming cognitions, hallucinations, error,

^[1]

etc., are “supportless”. In doing so, he possibly follows the Vṛttikāra, who also did not reject the objector’s assumption that dreams are supportless, but rather focused on confuting the possible consequence of all cognitions being devoid of support. ${ }^{15}$ In spite of this similarity, Uṃveka builds his confutation of the Buddhist conclusion that all cognitions have no external object on a different basis. As we will see also in Sucarita Miśra’s and Pārthasārathi Miśra’s commentaries, he understands nirālambanam as a kind of paryudāsapratisedha (implicative negation). In this way, he can state in the same sentence that dreaming cognitions are “supportless” and that they do have an external support. In fact, their being “supportless” does not depend on their not having a support, but rather on their having an improper support: ${ }^{16}$
[Question:] The supportless-ness (nirālambanatva) we have been talking about in regard of dreaming cognitions, etc., is it caused by the absence of an external support, or by the fact that the image [appearing in those cognitions] is [in reality] an empty one (, that is, it does not correspond to the cognition’s external support), or by [its] being without cause?
[Answer:] The supportlessness of dreaming cognitions, etc., is surely not caused by the absence of an external support, since an external support is present in all cases. ${ }^{17}$

15 See Schmithausen [1965: 140, §17b].
16 According to the Indian grammar, the negative particle can have two meanings, prasajyapratiṣedha and paryudāsapratiṣedha. The latter is generally found when the negative particle is compounded with a noun (e.g., abrāhmanam ānaya), whereas the former is generally found when it is used independently with a verbal form (e.g., anṛtaṃ na vaktavyam). Prasajyapratisedha is described as a case where the negative sense predominates (“in any case, one should not say an untruth”), whereas in paryudāsapratisedha the negation is not absolute, but rather qualifies something else. In fact (as explained by Patañjali and Kaiyaṭa, ad MBh 2.2.6) abrāhmanam ānaya does not mean “one should not bring a brahmin”. Rather, what one intends to say is that one should bring someone qualified by his not being a brahmin, namely, a kṣatriya, etc., who is similar to a brahmin though not being one. Apart from similarity, other five senses are listed (see Vasu [1988: 258] ad $\bar{A}$ 2.2.6, and Abhyankar [1977: 213, sub voce nañ]). In our present case, understanding nirālambanam as a paryudāsapratisedha could have the sense of aprāśastya (“having an improper support”). This would not contradict the grammatical understanding of nirālambanam, since according to the main stream in the Pāṇinian school, prefixes such as nis- (in nir-ālambana) qualify the meaning of the subsequent noun or verb and do not have by themselves a fixed meaning. Paryudāsapratisedha and prasajyapratisedha are commonly used as explanatory devises since Patañjali (see, e.g., MBh ad $\bar{A}$ 1.1.43 and ad $\bar{A}$ 2.2.6). In spite of that, the term paryudāsapratisedha is not present in any of the $S V$ 's commentaries ad loc.
17 svapnādipratyayānāṃ nirālambanatvam ucyamānaṃ bāhyārthālambanābhāvād vā pratibhāsaśūnyatvena vā, kāraṇābhāvād vā? na tāvad bāhyārthālambanābhāvena svapnādipratyayānāṃ nirālambanatvam, sarvatra bāhyālambanatvasya vidyamānatvāt (ŚVVTṬ ad Nirālambanavāda 107cd. Obviously, Uṃveka rejects also the third option, see his commentary ad 109cd. The exact meaning of pratibhāsa according to Uṃveka is still to be investigated.

(iii) Do the two tads share the same referent?

The two tads are read by Umveka as referring to the same sort of erroneous or illusory cognitions.
(iv) Is the ca copulative?

Consequently, what follows the $c a$ is read as elaborating on the preceding nirälambanam: “it is supportless and (“that is”) it has a support which is not there”.
(v) The meaning of abhāva in 118b

Umveka interprets abhāvālambanam as “having a support which is not [there]” (asādālambanatva), as highlighted in his introductory remark, where he substitutes abhāvālambana with asadālambana. For this interpretation, he relies on his interpretation of $\check{S} V$ Nirālambanavāda 108ab:

Although the fact of having an external support is common [to all cognitions], dreaming cognitions, etc., are not pieces of knowledge, since there is a succeeding invalidating cognition. Hence says [Kumārila]: “[In every case there is an external support], which [, however, exists] in another place or time”. 18

So, 117cd-118ab refer in Umveka’s interpretation only to erroneous or illusory cognitions (and not to cognitions of absence), as the external object grasped (e.g., silver), though existent somewhere else, is not present at the time such cognitions arise:

Since that, by which [the erroneous cognition] is supported, is not there (abhāva), this (erroneous cognition) is not a piece of knowledge. ${ }^{19}$
(vi) anirūpanāt
anirūpanāt is not commented upon by Umveka. My rendering of it in the above translation of Nirālambanavāda 117cd-118 according to Umveka’s interpretation relies on my general understanding of the verse and of Umveka’s commentary thereon.

The deflationary interpretation of abhāva in 118b as just meaning ‘non present’, and not defining the specific aspect of a knowledge content that can only be known through absence as an instrument of

^[1]

knowledge, makes the further step of kārikā 118 cd less natural. If indeed the abhāva mentioned in 118b has nothing to do with absence as a category, then why is absence mentioned in the succeeding hemistich? Uṃveka must in fact add an adversative particle to make sense of it.

4. Sucarita Miśra’s commentary

Sucarita Miśra’s time is still unsettled, although he is generally held to belong to the 12th century. ${ }^{20}$ However, a terminus ante quem is given by Ratnakīrti, who happens to mention him, and this seems to support an earlier date (something like 900-950). ${ }^{21}$ His commentary resembles in many respects Pārthasārathi’s one, but, their relative chronology remains an open question. As far as the present issue, at least, Sucarita’s Kāśikā is shorter but more philosophically engaged than Pārthasārathi’s Nyāyaratnākara. Unfortunately, Sucarita does not quote fully the verses he is commenting upon. So, one cannot be totally sure about the text he is reading.

The verses $117 \mathrm{~cd}-118$ appear in a section of the $\dot{S} V$ dedicated to the refutation of the Buddhist assumption that, since there is no external referent, all cognitions are in reality supportless. Kumārila ${ }^{22}$ rejects the Buddhist syllogism “all cognitions are devoid of support, because they are cognitions, like dreaming cognitions”, ${ }^{23}$ in various ways. Beginning from $\dot{S} V$ Nirālambanavāda 107 cd he shows how the syllogism’s illustration (drṣṭānta), cannot demonstrate the supportlessness of cognitions, insofar as even dreams, etc., are not devoid of support. In Sucarita’s words:

Since there is no absolute (sarvathā) absence of support (nirālambanatva), even of dreaming cognitions, etc., the example does not include what it ought to demonstrate. […] In regard to whatever cognition, there exists

^[1]

indeed a support, in another place or [time]. The fact of having a non[adequate]-support (anālambanatva) is just this, that one grasps in a different place [or time] what one has seen in a certain place or time. It is not the absolute absence of what one seizes. ${ }^{24}$

In Sucarita’s interpretation, Kumārila first (vv. 107cd-108ab) shows how dreams, that is, purely mental cognitions, are not supportless. Then (vv. 109cd-112a), he turns to erroneous cognitions depending on the external (bāhya) sense faculties, and shows how even the latter are not devoid of support. Among them, Kumārila mentions the case of the cognition of a hare’s horn (vv. 111cd-112a). In this passage, this looks like an instance of erroneous cognition, but the same example is employed by Kumārila in ŚV Abhāvavāda 4 in order to illustrate a sort of absence (namely, atyantābhāva, absolute absence). Indeed, Sucarita, like Pārthasārathi, employs it as the bridge between the two topics of erroneous cognitions and cognitions of something absent. In fact, although the illusory cognition of a hare’s horn may have as support the hare and a horn seen elsewhere, in Sucarita’s commentary a hypothetical objector states that the very cognition “there is no hare’s horn” is supportless. In reply to that, Sucarita explains (ad v. 112bd) that we do not seize the absence of the hare’s horn independently of any existing support. ${ }^{25}$ Similarly, concludes Sucarita, in every cognition emptiness always refers to something, so that the cognition has as support that of which there is emptiness. ${ }^{26}$ In dealing with this topic, his commentary makes it clear (ad vv. 112b-114ab) that those (see infra, §5) referred by Kumārila after v.112a are all instances of absence-cognitions. As examples, he mentions typical instances of absolute absence (atyantābhāva), like the world being empty, that is, devoid, of the son of a barren woman. Thereafter (ad vv. 114cd-116), he deals with the central issue of ŚV Nirālambanavāda 107-118, that is, whether a non-present object can give rise to a cognition. This is, indeed, the

^[1]

main feature common to both absence-cognitions and erroneous cognitions. Sucarita explains that a non-present object cannot have other kinds of causal efficacy (water in a mirage, for instance, cannot calm thirst), but it can arouse cognitions. In fact, explains Sucarita, a non-present rain, that is, a future or past one, can be inferred through the heaviness of the clouds or through the wet ground. Hence, it can be instrumental in the arising of a cognition. ${ }^{27}$ While discussing this issue, he carefully avoids the ambiguity of the term abhāva and speaks instead of non-present (avartamāna) items. Indeed, the riddle of ŚV Nirālambanavāda 114cd lies in the apparent contradiction of stating that an absent/non-existing object can give rise to a cognition of its absence/non-existence. Sucarita’s standpoint reduces the issue to the capacity of an existing, though non-present, object to arouse a cognition. In this way, a remembered piece of silver can stimulate the erroneous cognition of silver (when before one’s eyes there is mother-of-pearl), and the absent pot one expected to see on the floor, can give rise to the cognition of its absence. On the other hand, maintains Sucarita, an absolutely (ekāntam) non-existing item does not produce any cognition. What exactly he means by “absolutely non-existing”, as different to the atyāntābhāva of a hare’s horn, etc., which do raise cognitions, remains open to further enquiry. ${ }^{28}$ For sure, Sucarita tries to meet the Mīmāṃsaka need for accommodating among what is able to raise cognitions “entities” such as the Vedic prescriptions, and, consequently, has to understand “existent” (sat) in a broader sense. ${ }^{29}$

As for the passage 117cd-118, the following is Sucarita’s commentary:
[Objection:] Do you then maintain that there is nothing having no support (anālambana)?
[Reply:] If so, there would not be any distinction between truth and falsity. Thus, he (Kumārila) says "Therefore, that cognition [which grasps something in a way different to how it actually is, is without an [adequate]

^[1]

support]". We call a cognition perceiving what is indeed an external [object] in a way different to how it [actually] is, ‘having no [adequate] support’ (anālambana). What is grasped is not non-existing [instead such a cognition grasps something existing, but in a false way], so this [cognition] is said to be false.
[Objection:] Then, what is the support of a cognition having absence as support? Not existence, as this would be contradictory. Not absence, because this (absence) does not appear independently, i.e., as distinct from existence. Therefore, it ends up necessarily (balād) by having no support (anālambana).
[Reply:] Hence, he (Kumārila) says “[and that cognition] has absence [as support]”. [Then] he states the reason for that: “[Absence is another] being [, nothing else]”. It is true that an independent absence is not ascertained, for how could this (absence), which is a characteristic of a being, be independently extracted from [it] (apanī-)? Hence, because of the very reason mentioned by you (objector), that is, “because it is not” independently “ascertained” (anirūpanāt), the cognition that arises in the case of, e.g., a cow - which is existing according to its true nature (ātman) - through the nature of another absent [thing], 30 that [cognition] is said to have absence as support. This has been stated also before (in 118ab) and is now (in 118cd) ascertained. This is the difference. ${ }^{31}$

As for the questions we asked above,
(i) The preceding context

The initial “therefore” concludes the preceding discussion denying the existence of cognitions with no external support, and introduces the question in this vein: ‘therefore, since all cognitions are proved to have an external support, …’.
(ii) The odd mention of nirālambanam

In his commentary ad vv. 107cd-117ab, Sucarita states that all cognitions have an external support, though some of them have a

30 Sucarita refers here to the standard example of reciprocal absence, that is, the reciprocal absence of a horse in a cow and vice versa.
31 kim idānīm anālambanaṃ nāma, na kiñcit tatrabhavatām, yady evam apahnutaḥ satyamithyātvavibhāgaḥ. ata āha - tasmād iti jñānamantena. bāhyam evānyathāsantam anyathāpratipadyamānam anālambanaṃ jñānam ācakṣmahe. nāsad grāhyam. tad evaṃ mithyocyate iti. abhāvālambanasya tarhi jñānasya kim ālambanaṃ, na tāvad bhāvah, virodhāt. nābhāvah, tasya bhāvād viviktasya svatantrasyānirbhāsāt*. ato balād anālambanam evāpatitam ata āha abhāveti || 117 || atra kāraṇam āha - bhāveti. satyaṃ na svatantro 'bhāvo 'vasīyate, bhāvadharmo hy asau kathaṃ svatantro 'panīyeta. atas tvaduktād eva svatantrānirūpanāt kāraṇāt sadātmanāvasthite gavādau vastvantarābhāvātmanā yaj jñānam utpadyate tad abhāvālambanam ucyata iti, idaṃ ca prāg apy uktam adhunopapāditam iti vivekaḥ. *I am reading viviktasya svatantrasyānirbhāsāt instead of viviktasyāsvatantrasyānirbhāsāt found in the edition. See a few lines below satyaṃ na svatantro 'bhāvo 'vasīyate and svatantrānirūpanāt.

non-present one. Hence, as he explicitly states while commenting ad 117cd-118, the nirālambanatva he admits does not contradict bāhyālambanatva. But how is this possible? Sucarita systematically glosses nirālambana with anālambana. ${ }^{32}$ This might be understood as a paryudāsapratisedha ${ }^{33}$, that is, “having a non-support”, “having a non-[adequate] support”, namely, a support which does not appear in the cognition according to its own form, since such anālambana cognitions are not supported by the same object appearing in the cognition. They do have an external support, since they do not grasp an asat, but they grasp it otherwise, e.g., they grasp mother-of-pearl just as something bright and mistake it for silver. A false cognition is in this way to be distinguished from a true one. Thereafter, Sucarita considers whether the same holds true for cognitions of absence. At first sight, the latter ones could be defined nirālambana, since they are caused by, say, the bare floor, although what appears in the cognition is instead the absence of the pot. The bare floor cannot, in fact, appear in the absence-cognition, since otherwise this would not differ from perception. It is for this reason, according to Sucarita, that Kumārila has stated in v. 118b that “[cognition of absence] has instead absence as support”. The reason is added, in Sucarita’s interpretation, in the next 118 cd . The absence of something is indeed an attribute of another existing item, and hence a cognition of absence is supported by, say, the bare floor, which is grasped by it according to its absent aspect, i.e., the pot. Better, in Sucarita’s words, the cognition of absence of a horse in a cow is supported by the cow, which is grasped through its absent aspect, the horse. In fact, Sucarita (unlike Umveka) does not introduce the pot as the standard example of absence. This may suggest that his interpretative school adhered more closely to the $\check{S} V$ classification of abhāva than Umveka’s one.
(iii) Do the two tads share the same referent?

The reference of tad … tad in 118a and 118b in Sucarita’s interpretation is quite complicated. He possibly reads Kumārila as follows: tasmād yad anyathāsantam anyathā pratipadyate, tan nirālambanaṃ jñānam. [abhāvajñānaṃ] ca abhāvālambanam. Like Pārthasārathi, also Sucarita justifies the shift of subject in v. 118b as an answer, but the objection present in Sucarita’s text is sharper.

^[1]

(iv) Is the ca copulative?
ca is slightly adversative. In fact, it coordinates two different cases, although absence cognitions are implicitly included also in the initial tasmād.
(v) The meaning of abhāva in 118b

As for the meaning of abhāva in 118b, Sucarita links it surely to cognitions of absence, as it is proved by his mentioning the example of the reciprocal absence of a horse in a cow.
(vi) anirūpanāt

As we shall see also in Pārthasārathi’s commentary, Sucarita is convinced that absence cannot be ascertained independently of what exists (in his commentary ad v. 112b he writes: “Although absence is not grasped independently, the existing support is, instead, grasped”. ${ }^{34}$ Accordingly, Sucarita glosses the anirūpanāt as svatantrānirūpanāt, “because it cannot be discerned independently [of something existing]”.

What made Sucarita propose his interpretation? Splitting v. 118ab, is surely not the easier way to interpret it, but Sucarita may have felt the split as not too broad because of the preceding discussions paralleling absence cognitions and erroneous ones. ${ }^{35}$

5. Pārthasārathi Miśra’s commentary

Pārthasārathi Miśra lived some generations after Uṃveka, between the 10th and the 13th century. ${ }^{36}$ His commentary has become the most used one among both middle-age and modern scholars, because of its clarity and because it closely explains Kumārila’s text. Pārthasārathi comments as follows ŚV 117cd-118 (Kumārila’s kārikās are printed in bold):

Then, would not there be any supportless cognition (nirālambanaṃ jñānam) at all? If so, then [this conclusion] is opposed to common belief (loka). Therefore he (Kumārila) said in the following three hemistichs: Therefore that which seizes a certain [object] in another way [different to the real one] || 117 || that is a supportless cognition. Since

^[1]

in this (explained above) way there is in every case an external [support], therefore only the cognition that grasps a certain object, e.g., mother-ofpearl, in a different way, e.g., as silver, is a supportless cognition, not the one which has its own [nature] (be it its existing aspect or its non existing one) ${ }^{37}$ as its support.
[Objection:] Then, the cognition of absence (abhāvajñāna), like “here there is no pot”, “this is not silver” and so on, what support has it? Indeed, a support called “absence” does not exist!
[Reply:] Therefore he (Kumārila) said: and that [cognition of absence] has absence as its support I
The [cognition] which “seizes a certain [object] in another way [different to the real one], that is a supportless cognition”, [and] that alone [out of the two cognition of absence and erronious cognition] has existence as its support (that is, the existence of something else). It has been said: "Every object, indeed, is both existing and not-existing in nature (sad-asad$\bar{a} t m a k a m) " .38$ Among those [erroneous cognitions and cognitions of absence], the cognition, e.g., of mother-of-pearl as silver, which seizes an existent thing (bhāva) called mother-of-pearl - whose real form [should be grasped] as a mother-of-pearl form - through the real silver form of another existent thing, namely silver, is erroneous. On the other hand, the absence-cognition, “this is not silver”, which seizes a thing really existing insofar as [its] mother-of-pearl form through the non-existing form of silver itself, [that cognition] has absence as its support, and is not erroneous. Therefore, both erroneous cognition and cognition of absence seize a certain object in another way. Among them, error seizes something through the existing aspect of something else; cognition of absence, instead, seizes something through the absent aspect of something else ${ }^{39}$. This alone is the difference.
[Objection:] But cognition of absence has non-being only as its support and not another being.
[Reply:] Therefore he said: Absence is another being, nothing else, because it cannot be perceived [independently from another being] II 118 II
The meaning [of the verse] is: That aspect of an existing [object], like silver, called absence is not independent, because it cannot be perceived

^[1]

[independently]. Rather, it is another being, i.e. it is a property of that. ${ }^{40}$ Therefore, also absence-cognition has another being, like mother-of-pearl, as its external support. Thus, it has not its own aspect (i.e. absence) as support. As for “and that [cognition of absence] has absence as its support”, in this case with the word “that” this alone is reflected upon: a cognition, even if it has absence as support, that very cognition has verily an existing external support.
[Objection:] But how can something which appears as an absence have an existing external support?
[Reply:] Therefore it has been said - “[Absence is] another being […]”. 41

40 One cannot perceive absence as such, therefore it is only something existent, e.g., mother-of-pearl, which is endowed with both a being and a non-being character (the being mother-of-pearl aspect and the non-being silver aspect mentioned above).
41 kim idāntṃ nirālambanaṃ jñānaṃ nāsty eva, tathā cel lokaviruddhah syāt. ata āha pādatrayeṇa - tasmād yad anyathāsantam anyathā pratipadyate || $117 \|$ tan nirālambanaṃ jñānam, yasmād evaṃ sarvatra bāhyam asti, tasmād yad eva vijñānam, anyathā śuktikādirūpeṇa santam artham anyathā rajatādirūpena grhnati, tad eva nirālambanaṃ jñānam, na tu yad ātmālambanam iti. yat tarhi iha ghaṭo nāsti, idaṃ rajataṃ na bhavati ityādikam abhāvajñānam, tasya kim ālambanam, na hy abhāvākhyam astīty ata āha - abhāvālambanaṃ ca tat I yad evānyathāsantam anyathā pratipadyate tan nirālambanam, tad eva bhāvāvalambanam*. tad uktaṃ bhavati - sarvaṃ hi vastu sadasadātmakam. tatra śuktikārajatādijñānaṃ śuktikākhyam bhāvaṃ śuktikārūpeṇa sadrūpaṃ bhāvāntarasya rajatasya yat sadrūpaṃ rajatarūpaṃ tena rūpena grhnad bhrāntaṃ* bhavati. nedaṃ rajatam iti tv abhāvajñānaṃ śuktirūpena santam eva bhāvaṃ rajatasya yad asadrūpaṃ tenātmanā grhnad abhāvāvalambanam abhrāntaṃ bhavati. tena bhrāntijñānam abhāvajñānaṃ cobhayam anyathāsantam artham anyathārūpena pratipadyate. tatra bhrāntir anyasya sadrūpenānyat pratipadyate, abhāvajñānaṃ tv anyasyāsadrūpenānyat pratipadyata ity eva višeṣa iti. nanv abhāvamātram evābhāvajñānasyālambanam, na bhāvāntaram ata āha - bhāvāntaram abhāvo 'nyo na kaścid anirūpanāt || $118 \|$ rajatāder bhāvasya yad abhāvākhyaṃ rūpaṃ tan na svatantram, anirūpanāt, kintu bhāvāntaram eva tat taddharma ity arthah. tenābhāvajñānasyāpi śuktikādibhāvāntaraṃ bāhyam evālambanam iti na svāṃsālambanam iti. yad vā “abhāvālambanaṃ ca tat” ity atra tacchabdenaitad eva parāmrśyate - abhāvālambanam api jñānaṃ tādṛ́sam eva bāhyabhāvālambanam eveti. nanv abhāvāvabhāsinaḥ kathaṃ bhāvālambanatvam iti? ata uktam bhāvāntaram iti. (NR ad ŚV Nirālambanavāda, 117-118). *I have been mainly following the 1993 edition of the $N R$, which closely follows the 1978 one. However, the other two editions available to me, that is, the 1979 and the 1898 one, diverge here on two important points. The 1898 and the 1979 editions, in fact, read abhāvālambanam instead of bhāvāvalambanam and bhrāntam instead of abhrāntam. The reading bhrāntam is obviously the correct one, as the context unmistakably demonstrates (I thank Prof. Taber for having suggested me this emendation before knowing about the 1898 and 1979 editions). As for bhāvāvalambanam/abhāvālambanam, instead, I believe the reading of the 1978 edition to be the correct one. The 1979 edition is probably mainly based on the 1898 one (whose graphic peculiarities it thoroughly repeats, see, e.g., the " $=$ " in anyath $\bar{a}=$ rajatādirūpena and anyath $\bar{a}=$ śuktikādirūpena in the above text), though in the prāstavikam (preface) the editor declares he has consulted several other (non better specified) manuscripts (p. [2]). The 1898 edition, on the other hand, has no preface and it is hence difficult to judge how soundly based it is. The 1978 edition is based on the 1898 one, which has however been improved by the editor through an old (prācīna) and a recent (arvācīna) manuscript, both held in Varanasi, Sarasvatī Bhavana Library, and through a comparison with related Mīmāṃsā and non Mīmāṃsā works (see ŚV 1978 p. 27). As for the context, the reading abhāvālambanam seems to me a banal repetition of the content of the $\dot{\bar{S}} \mathrm{V}$ verse. Moreover, Pārthasārathi intends exactly to demonstrate that the verse tan nirālambanaṃ jñānam abhāvālambanaṃ ca tat refers to two different sorts of cognitions. So, the reading abhāvālambanam referring to nirālambana cognitions collides strongly with the context. Other minor differences between the 1978 and the 1898 editions have not been mentioned here.

Thus, Pārthasārathi reads Kumārila’s text as comparing erroneous cognition and cognition of absence. Those have something in common, since they both seize something according to the aspect of something else. If one considers again the example of the mother-of-pearl mistaken for silver, together with the subsequent cognition that “this is not silver”, their difference can be grasped in the following scheme:

Erroneous cognition: Invalid cognition, grasping an (existing) mother-of-pearl, according to its [non-existing] aspect of silver (, which however exists elsewhere).

Cognition of absence: Valid cognition, grasping an (existing) absence of silver, according to its aspect of mother-of-pearl (, which exists in this instance).

But why does one in both cognition of absence and erroneous cognition grasp something according to the aspect of something else? In the case of absence-cognitions it is obvious that, if one were not to look for, say, a pot, one would not grasp anything but the bare floor. This is what Kumārila calls the jighrks $\bar{a}$ element in absence-cognition, that is, the “desire to grasp”, leading the cognitive process (one grasps the absence of the pot, because one desires to grasp the pot and one is stricken by its absence). In the case of erroneous cognitions, this willing aspect is not (as far as I know) explicitly postulated. Nevertheless, one could easily imagine that, at least in case of mother-of-pearl mistaken for silver, one’s desire to find some silver has played a leading role in the cognitive process. Apart from this “desire to grasp”, Kumārila mentions another factor accounting for cognitions of absence, that is, udbhūti, “emergence”. This could explain the succeeding cognition, as one realises that “this is not silver”. This absence-cognition is indeed not lead by any desire to grasp, but by the compelling emergence of the silver’s absence.

So, Pārthasārathi’s answers to the above mentioned questions are as follows:
(i) The preceding context

The initial “therefore” (tasm $\bar{a} t$ ) is glossed thus: “Since (yasmāt) in this (explained above) way there is in every case an external [support], therefore (tasmāt) only the cognition that grasps a certain object, e.g., mother-of-pearl, in a different way, e.g., as silver, is a supportless cognition”. Pārthasārathi refers to the preceding verses; since v. 107cd

Kumārila 42 is trying to deny the opponent’s syllogism “every cognition is devoid of any external support, because it is a cognition, like dreaming cognitions”. On the contrary, maintains Kumārila, even dreaming cognitions do have an external support. Pārthasārathi explains:

An external [support] - which, after having been really experienced in another place or in another time, is presently remembered in the dream is understood, due to a fault, as proximate in time and space. Hence, even in this case there is no lack (abhāva) of external [support]. 43

And so do all other odd types of cognitive experiences, like hallucinations. More in detail, Kumārila refers to dream (vv. 107cd-109ab), to the seeming fire-circle one perceives when a torch is moved rapidly and circularly (v. 109cd), to the illusory buildings one might see in the sky because of the shape of clouds (v. 110ab), to mirage (v. 110cd-111a), to the illusion of seeing a hare endowed with horns (111cd-112a). The latter example is of much interest, insofar as the hare’s horn is the example mentioned in $\check{S} V$ Abhāvavāda 4 as instance of “absolute absence”, atyantābhāva. The connection is further underlined by Pārthasārathi’s terminology; in his commentary on that verse, Pārthasārathi describes absence cognition in terms of the non-existing form (asadrūpa) of the absentee, ${ }^{44}$ and the same term is used also in ad vv. 117-118. Next comes v. 112ab, which Pārthasārathi interprets as referring to the negation “there is no hare’s horn”. Thus, the hare’s horn case could be seen as the bridge between erroneous cognitions and cognitions of absence. Indeed, both erroneous cognition and cognition of absence share the same nature of seizing something according to the nature of something else. In this way, “therefore” in 117c can sum up both verses 109cd-112ab and verses 112 cd -117 ab . The latter are interpreted by Pārthasārathi as instances of cognition of absence. In this way, cognitions of emptiness (in the form “the house is empty of Caitra”, i.e., "Caitra is not in the

^[1]

house"), cognitions of something impossible (“there is a hundred elephants on top of my finger”), and cognitions of something which could never be experienced (like the Sāṅkhya primordial Nature) are all interpreted as having to do with cognitions of absence. In the latter two cases absence-cognitions makes its entrance as a succeeding invalidating cognition, like “there is not a hundred elephants on top of my finger”, and “there is nothing like primordial Nature”. Such a succeeding step is not explicitly mentioned by Pārthasārathi whilst commenting on vv. 113ab-117ab, nevertheless he hints at it whilst dealing with v. 118b (“it is not silver”), and he mentions it in the hare’s horn case (“there is no hare’s horn”). Elsewhere, Kumārila speaks of absence as the only instrument of knowledge which can seize the content of some unacceptable thesis held by an opponent, like the Buddhist emptiness one. He (or Pārthasārathi following him) could have meant the hint at the Sānikhya primordial Nature in the same vein. So, Pārthasārathi’s interpretation of Kumārila’s kārikās is:

Therefore, [among] that which seizes a certain [object] in another way [different to the real one] || 117 ||
that is a supportless cognition, and that has absence as support.

The first line can indeed be read as regarding erroneous cognition, but also absence cognitions, as Pārthasārathi explains in his commentary on 118 b .
(ii) The odd mention of nirālambanam

Nirālambanam is admitted only as regards erroneous cognitions, whereas absence cognitions do have an existing support. Pārthasārathi seems to differentiate between cognitions which grasp an object as it is - i.e., according to its existing aspect if they grasp the object as, say, “a piece of mother-of-pearl”, or according to its absent aspect if they grasp the object as, say, “something which is not silver” -, and cognitions which grasp the object in a way different to the way it is. Thus, Pārthasārathi’s admission of nirālambana is not meant to favor his Buddhist opponent, as he clearly explains that supportlessness is only stated in regard to false cognitions. The target of such an admission could instead be those Naiyāyikas who upheld that all cognitions have a real support, thus postulating a special contact between the sense faculties and the silver appearing in an erroneous

cognition, or between the mind and such silver. ${ }^{45}$
(iii) Do the two tads share the same referent?

The two tads in 118ab refer to two different kinds of cognitions. See (i). This constitutes the main disadvantage of this interpretation, since at first sight it postulates a break within hemistichs a and b of v. 118. The mention of absence as an instrument of knowledge is explained by Pārthasārathi as the answer to an implicit question. However, as a matter of fact, both absence and erroneous cognitions can be, according to this commentary, implied by the definition of v. 117 cd , and by its initial “therefore”. So, following Pārthasārathi, there is no major split in the argument (although a split in 118ab cannot be denied).
(iv) Is the ca copulative?

Consequently, ca is read as an adversative coordinating two different cases.
(v) The meaning of abhāva in 118b

Abhāva is read as referring to absence. Nevertheless, Pārthasārathi deliberately employs the ambiguity of this term in order to disprove the opponent’s thesis. He makes the opponent uphold a-bhavā-lambanatva, which in the opponent world-view means “the fact of non-having an existing support” and replies that, indeed abhāva exists and can play the role of a support (commentary on v. 119).
(vi) anirūpan̄āt

Unfortunately, Pārthasārathi does not comment on anirūpanāt. The context of his commentary seem to implicate an understanding of it as “since [absence] is not [independently] ascertained”.

Pārthasārathi’s interpretation has the disadvantage of postulating many extra words to be added to Kumārila’s verse (see, e.g., the explanation of nirālambanam, and the complex reference of yad and tad, tad in 117c-118b). This point can be rephrased as follows: Is Pārthasārathi interpreting Kumārila or is he offering him the best way-out of a difficult trap? This does not mean that Kumārila has defined by mistake erroneous cognition nirālambanam, thus not considering that he was giving his very opponents a powerful weapon against his arguments. Kumārila’s preceding kārikās in the Nirālambana Chapter rule out such an hypothesis. Nevertheless,

^[1]

Pārthasārathi may have thought that Kumārila’s nirālambanam needed a further specification, and the same may have happened as for the splitting of the verse. Pārthasārathi may have chosen to split the verse in order to eliminate what he thought was a weak point in Kumārila’s argumentation.

6. Contemporary scholars on ŚV Nirālambanavāda 117cd-118

No contemporary study is entirely dedicated to this passage, but several authors have offered a translation of it within their accounts of Kumārila’s theory of absence or of error. As, apart from Gangāāātha Jha’s translation of the $\dot{S} V$, they did not translate but the relevant $k a \bar{r} i k \bar{a} s$, they did not take into account the possible role of “therefore” as connecting them to the preceding verses.
G. Jha’s translation, to begin with, closely follows Pārthasārathi’s commentary in interpreting 117cd-118:

Therefore it is only that (cognition), which comprehends an object otherwise than in the form it exists in, that can be said to be “devoid of substratum”; and that Cognition which has ‘negation’ for its object is, in fact, one that has a real substratum; because this ‘negation’ too is not an independent entity by itself; for it is not so comprehended.

On the other hand, in his illuminating study on the theory of error in Indian philosophy, L. Schmithausen offers an original interpretation of this passage:

So, when in an instance of error a thing appears according to the specific aspect of something else, this actually means that a non-existing [item] that is, an existing [item], which is not concretely identical with the thing actually available - manifests itself. […] Through the remarks given in this paragraph, our hope is that we are adequately interpreting Kumārila’s mind when he says: “Therefore, a cognition which grasps something differently from the way it is, that is objectless, and that is what has a non-existing [item] as its object. [For,] the non existing [item] is the other thing, nothing else, because [another non-existing [item]] cannot be established” (ŚV Nirālambanavāda 117c-118d). 46

^[1]

That is, Schmithausen would answer as follows to the questions asked above:
(ii) The odd mention of nirälambanam

Nirälambanam is attributed to Kumārila and not to the opponent.
(iii) Do the two tads share the same referent?

The two pronouns tad refer to the same entity, namely “cognition”. This makes this interpretation smoother than Pārthasārathi’s and Sucarita’s ones.
(iv) Is the ca copulative?

Schmithausen’s rendering of the $c a$ is probably the easiest one, as this coordinates two statement about the same entity (and is hence copulative). Nevertheless, his comment on this passage presupposes a more complex reading of $c a$ :

Error consists, hence, in the apprehension of something as different than it actually is, […] In every instance [of erroneous cognition] the object [of error] consists (at least also) of something which is not present. Error is also distinct from valid cognition insofar as it can be defined “objectless” in the sense that, as regards the particular element [being the object of error], it has no present object. It can also be defined as “having a non-being as its object”. 47

Here, error is said to be possibly defined as “objectless” and also as “having a non-being as its object”. Does the above statement mean that Schmithausen interprets $c a$ in 117d as meaning rather “also”, that is, a cognition grasping something in a different way is said to be objectless and also having a non-existing [item] as object? Yes, but, the “also” seems to coordinate two different nuances of a plausible definition of erroneous cognition. 48
(v) The meaning of abhāva in 118b

Abhāva is understood as the absence of, say, silver in the case of

^[1]

mother-of-pearl. Erroneous cognition is, hence nirālambana because it grasps an object which is not present. Moreover, Schmithausen interprets differently bhāvāntaram as “the other being” instead of “another being”. Erroneous cognition of silver grasps an object which does not exist, because the absent object is nothing but the other one, that is, the silver which is only present elsewhere.

Explicitly following Schmithausen, Kellner goes a step further in the identification of what is the content of an erroneous cognition. In fact, since an erroneous cognition grasps a non-existing object, she interprets it as an instance of absence-cognition, although invalid:

The erroneous cognition of a mirage is corrected when one determines the heated ground as distinct from water. However, Kumārila describes the erroneous cognition itself as the cognition of a non-being, since it has an object that is not available at the present time and in the present place. Obviously, this cannot be a valid cognition, since it does not determine its content as it actually is. If such cognition were valid, then it would determine its object - water - as non-existing (or it would determine the heated ground as distinct from water) and error would not occur. This means that error can be analyzed as the cognition of a non-existing [item], but not as a valid negative cognition. In this way, an invalid negative cognition - whose invalidity is established by a latter, valid and negative cognition - would be responsible of the arousal of error. ${ }^{49}$

In terms of the silver example, what is grasped is the absence of silver. Hence, erroneous cognition of silver is an instance of invalid absence-cognition. Absence-cognition would in this way cover a range of cognitions from “cognition of an absence” (in the form “there is no yogurt in milk”, as described in ŚV Abhāvavāda 2) until “invalid cognition of something which does not exist”.

In a succeeding article, however, Kellner analyses closely $\dot{S} V$ 117cd-118 and proposes the following translation:

^[1]

Therefore, the cognition which is [as the opponent claims to be] without an objective basis cognizes something that exists in one way as existing in another. And this (cognition) has (indeed) an objective basis which is not real (abhāva); but this unreal object is simply another real object (bhāvāntara), and nothing else, for (any other object) is undifferentiated (anirūpanāt) (Kellner [1996: 155-6, fn 27]).

Here, Kellner would answer as follows to the questions abovementioned:
(ii) The odd mention of nirālambanam

Nirālambanam is attributed to an objector. If I am rightly understanding her, according to Kellner, Kumārila’s statement can be rephrased as follows: “You (Buddhist opponent) think that erroneous cognitions are supportless. They have indeed an unreal support, but such an unreal support is nothing but another real object”.
(iii) Do the two tads share the same referent?

The two pronouns refer to the same cognition.
(iv) Is the ca copulative?

Yes, ca coordinates two statement about the same cognition.
(v) The meaning of abhāva in 118b

Abhāva is understood as just meaning “not real”. Kellner refers in this regard to similar instances of a non-technical use of abhāva, like $\dot{S} V$ Abhāvavāda 48, and concludes: “In this way, the necessity of pressing abhāva into a framework encountered in discussions of cognitive error does not arise and, in turn, need not give rise to any of the confusions mentioned above” (Kellner [1996: 156, fn 27]).
(vi) anirūpanāt

I am afraid I cannot understand Kellner’s rendering of anirūpanāt. So, the addition of an initial adversative particle (“but”) seems to me not justified.

Summing up, I am not fully convinced by Kellner’s leaving apart absence as an instrument of knowledge, against Sucarita and Pārthasārathi and notwithstanding the preceding kārikās. It is possible, however, that this choice is mainly due to her principal aim, that is, showing that Schmithausen is not justified in stating that absence is a relative concept, since Schmithausen’s understanding mainly depends on these verses (see Kellner [1996: 156-157, fn 27]).

In his review article of Kellner [1997], which however does not take into account Kellner [1996], Taber deals with the same verses:

I interpret these difficult lines as follows:
"Therefore, a cognition that knows something to be other that what it is, is ‘without an object’, but that [cognition, in reality] has non-being as its object [i.e. the non-being of silver; so it is not really objectless].

And non-being is just another being [i.e., the non-being of the silver is the being of the mother-of-pearl], nothing else [hence, it is something real], because it is not discerned [as such, i.e., it is never ascertained independently from what exists]."

The idea of 118 cd (bhāvāntaram abhāvah…) must be understood in light of the doctrine of the unitary nature of being […] (Taber [2001: 77 fn 20]).
As for the above questions:
(ii) The odd mention of nirālambanam

Nirālambanam is understood as just a provisional admission, immediately denied by the succeeding abhāvālambanam.
(iii) Do the two tads share the same referent?

Taber’s rendering of the two tads is smooth, as it does not postulate two different referents.
(iv) Is the ca copulative?

Taber’s interpretation relies on his reading of $c a$ as an adversative.
(v) The meaning of abhāva in 118b

Abhāva is interpreted as referring to non-existence as the cognition content to be grasped exclusively by absence as an instrument of knowledge.

So, Taber postulates an initial stage in which one cognises through absence as an instrument of knowledge, although one does not become aware of it, that “this is not silver”. This absence of silver is later mistaken as presence of silver. This complex structure aims, in Taber’s view at supporting Kumārila’s realism:

The reality of non-being, in turn, enables him to develop a realist analysis of perceptual error. When I mistake mother-of-pearl for silver I do not cognize something that doesn’t exist in any sense. Rather, I apprehend the absence of silver in regard to mother-of-pearl - which is real insofar as it is identical with the mother-of-pearl - incorrectly, as the being or presence of silver. (Taber [2001: 77])

Does the instance of mother-of-pearl mistaken as silver fit into Kumārila’s own depiction of absence-cognition? The question cannot be easily answered, since the very definition (if it is a definition) of absence-cognition in Ś $V$ Abhāvavāda 11 is by itself not non-

controversial. ${ }^{50}$ Surely the cognition of silver in case of mother-ofpearl lacks the essential character of an absence-cognition, that is, the form “it is not”. As already hinted at, on the other hand, John Taber implicitly proposes to split the silver-cognition into two and defines absence-cognition only the first step of this cognitive process, that is, the knowledge that there is no silver in mother-of-pearl. Cognitions are not self manifest according to Mīmāṃsā, and thus this first step cannot be rejected just because one is not aware of it. This first passage, however, is not mentioned in any commentary and I am afraid it somehow contradicts the very raison d’être of absence as an instrument of knowledge. If indeed absence is not an ontologically separate entity, then absence as an instrument of knowledge is essentially qualified as an awareness that, say, $x$ is absent in $y$.

7. Conclusions

Although Uṃveka Bhatta has been the first commentator on the $\dot{S} V$, his readings have not been universally accepted. This can be possibly explained by his views on Kumārila, which seem often indebted more to his own philosophical understanding than to a close and word-byword reading of Kumārila. On the other hand, Sucarita Miśra and Pārthasārathi Miśra represent an alternative interpretative school. Although the question of their relative chronology cannot be settled by a limited comparison such as the present one, their differences in the passage under question show that Sucarita deals with the same themes treated by Pārthasārathi by displaying a deeper philosophical interest. ${ }^{51}$ As for the specific issue at stake, all Sanskrit commentators agree in denying supportlessness in the case of cognitions of absence and in accepting it in the case of erroneous cognitions, hallucinations, etc. The inconsistency of such an acceptance could be solved (as with Kellner [1996]) with the assumption that the commentators provisionally borrow the term nirälambana (“supportless”) from their Buddhist opponents. Nonetheless, I rather think that all commentators propose to interpret nirālamabana (“supportless”) as anālambana (“having a non-support”) and anālambana as an

^[1]

instance of implicative negation (“having a non adequate support”). In this way, they show how the opponent’s terminology can be better explained and justified within a Mīmāṃsaka perspective.

Whatever the case, it is difficult to tell who is right in representing Kumārila’s original meaning of 117cd-118, as both Uṃveka and Pārthasārathi/Sucarita have to somehow force the reading of the verse. How this is due mainly to the complexity of the theme is shown by the fact that also contemporary scholars acknowledge that the passage is, to say the least, awkward.

REFERENCES

A. Texts and Translations

Ā: Āṣṭādhyāȳ̄ of Pānini, edited and translated by Śrīśa Chandra Vasu, Benares: Sindhu Charan Bose, 1891 (rpt. 1988).
Kāśikā of Sucarita Miśra, edited with Ślokavārttika of Kumārila Bhaṭta by K. Sāmbaśiva Śāstrī, Trivandrum: Trivandrum Sanskrit Series, 90, 99, 150, 1926, 1929, 1943.
MBh: Mahābhāṣya of Patañjali, edited with the commentary Bhāṣyapradīpa of Kaiyaṭa Upādhyāya and the Supercommentary Bhāṣyapradīpoddyota of Nāgeśa Bhaṭta by Bhargava Sastri Bhikaji Josi, Delhi: Caukhamba, 1992.
NR: Nyāyaratnākara of Pārthasārathi Miśra, edited with Ślokavārttika of Kumārila Bhaṭta by Svāmī Dvārikadāsa Śāstrī, Varanasi: Tara Publications, 1978.
NR: Nyāyaratnākara of Pārthasārathi Miśra, edited with Ślokavārttika of Kumārila Bhaṭta by Ganga Sagar Rai, Varanasi: Ratna Publications, 1993.
NR: Nyāyaratnākara of Pārthasārathi Miśra, edited with Ślokavārttika of Kumārila Bhaṭta (the editor’s name is missing), Benares: Caukhamba, 1898.
NR: Nyāyaratnākara of Pārthasārathi Miśra, edited with Śābarabhāṣya of Śabara Svāmin and Ślokavārttika of Kumārila Bhaṭta by Gajānana Śastrī Musalgaonkar, Varanasi: Bhāratīya Vidyā Prakāśan, 1979.
Śāstradīpikā of Pārthasārathi Miśra, tarkapāda, edited with the commentary Yuktisnehaprapūraṇi and with Gūd̄ārthavivaraṇa by Rāmakṛ̣̣̣a by Dharmadatta Jhā, Bombay: Nirnaya Sagar press, 1915.
ŚV: Ślokavārttika of Kumārila Bhaṭta, translated with extracts from the commentaries Kāśikā of Sucarita Miśra and Nyāyaratnākara of Pārthasārathi Miśra, by G. Jhā, Calcutta: the Asiatic Society, 1900-1908 (rpt. 1983).
ŚV: Ślokavārttika of Kumārila Bhaṭta, see also NR, ŚVTṬ, Sucarita Miśra’s Kāśikā.
ŚVVTṬ: Ślokavārttikavyākhyā Tātpāryaṭikā of Uṃveka Bhaṭta, edited by S.K. Ramanatha Sastri, revised by K. Kunjunni Raja and R. Thangaswamy, Madras: Madras University Sanskrit Series 13, 1971.
TS: Tattvasaṅgraha of Śāntarakṣita, edited with the commentary of Kamalaśīla and with an introduction in Sanskrit by Embar Krishnamacharya, Baroda: Central Library, 1926. (rpt. 1984 and 1988).
B. Studies

Abhyankar, Kashinath Vasudev
1961 A dictionary of Sanskrit grammar, Baroda: Oriental Intitute.
Chakrabarti, Arindam
1997 Denying Existence. The Logic, Epistemology and Pragmatics of Negative Existentials and Fictional Discorse, Dordrecht/Boston/ London: Kluwer (Synthese Library, vol. 261).

Chakrabarti, Kisor Kumar
1978 “The Nyāya-Vaiśeṣika Theory of Negative Entities”, Journal of Indian Philosophy 6: $129-144$.

Freschi, Elisa and Alessandro Graheli
2005 “Bhāttamīmāṃsā and Nyāya on Veda and Tradition”, in F. Squarcini (ed.), Boundaries, dynamics and construction of traditions in South Asia, Firenze/New Delhi: FUP/ Munshiram Manoharlal.
Freschi, Elisa
Forthcoming “Facing the boundaries of epistemology: Kumārila on error and negative cognition”, Journal of Indian Philosophy.
Halbfass, Wilhelm
1986-92 “Observations on the Relationship between Vedic Exegesis and Philosophical Reflection”, Journal of Oriental Research Madras LVI-LXII: 31-40.
Kajiyama Yūichi
1999 The Antarvyāptisamarthana of Ratnākaraśanti, Tokyo: The International Research Institute for Advanced Buddhology.
Kataoka, Kei
2003 “Experience and Reflection: Historical Development of the Argument on the Means of Grasping Invariable Concomitance”, Nanto Bukkyo (or Journal of the Nanto Society for Buddhist Studies) 83: 1-32. (In Japanese)
Kellner, Birgit
1996 “There are no pots in the Ślokavārttika. Kumārila’s definition of the abhāvapramāna and patterns of negative cognition in Indian Philosophy”, Journal of the Oriental Intitute Baroda XLVI(3-4): 143-167.
Kellner, Birgit
1997 Nichts bleibt nichts: Die buddhistische Kritik an Kumārilas abhāvapramāna: Übersetzung und Interpretation von Śāntarakṣitas Tattvasamgraha vv. 1647-1690 mit Kamalaśīkas Pañjikā. Mit einem Anhang zur Geschichte der negativen Erkenntnis in der Indischen Philosophie, Wien: Wiener Studien zur Tibetologie und Buddhismuskunde.
Rao, Srinivasa
1998 Perceptual Error: The Indian Theories, Honolulu: University of Hawaii Press.
Ruegg, David Seyfort
1981 The Literature of the Madhyamaka School of Philosophy in India, Wiesbaden: Harrassowitz.
Schmithausen, Lambert
1965 Maṇdanamīśra’s Vibhramavivekah: Mit einer Studie zur Entwicklung der indischen Irrtumslehre, Wien: Hermann Böhlaus Nachf.
Taber, John
1992 “What did Kumārila Bhaṭ̣a mean by svatah prāmānya?”, Journal of the American Oriental Society 112: 204-221.
Taber, John
2001 “Much ado about nothing: Kumārila, Śāntarakṣita, and Dharmakīrti on the cognition of non-being”, Journal of the American Oriental Society 121(1): 72-88.
Thrasher, Allen Wright
1979 “The dates of Maṇḍana Miśra and Śaṃkara”, Wiener Zeitschrift für die Kunde Südasiens 23: 117-139.
Vasu, Ś.C., see $\bar{A}$.
Verpoorten, Jean-Marie
1987 Mīmāṃsā Literature, Wiesbaden: Harrassowitz.

Universita “Sapienza” (Could you write this in English?)
Roma