Page #1
--------------------------------------------------------------------------
________________
PAURANDARASUTRA
Eli Franco
A couple of years ago I had the privilege of reading the Tattvopaplavasimhaa Lokayata text-with Pt. Dalsukh Malvania. It is only appropriate, as a small token of my respect and gratitude to Pt. Malvania, to contribute a paper which is directly connected to what he taught me. I have chosen, therefore, to deal here with a Lokayata fragment as it appears in a Jaina source.
The Jainas, no doubt, have made a most original and important contribution to Indian philosophy. However, the importance of the Jaina sources goes far beyond Jaina philosophy itself. Their importance for the study of all other systems of Indian philosophy cannot be underrated, for in no other system is the purvapakṣa presented with such honesty and thoroughness.
Strangely enough, although modern research on the Lokayata school of thought was done exclusively by scholars who were interested in pramäṇasastra, their studies are mainly concerned with the Bṛhaspatisutra, while the Lokayata logicians are almost completely ignored.
It is well known that ever since Dinnaga, epistemological and logical problems came into the foreground of Indian philosophy. Every philosophical school had to respond to the new discoveries made by Dinnaga and his follower Dharmakirti, and the Lokayata school was no exception to this rule. The old Lokayata arguments which were convincing enough as long as inferences were based on simple analogies, could no longer be applied to the new, well-established inferences, which were based on universal concomitance (vyupti). In order to stick to the old doctrine according to which inference is not a valid means of knowledge, the Lokayatikas developed a new kind of self-destructive logic inferences which prove that no inference is valid.
In what follows I shall try to present three different interpretations for one of these inferences : the Paurandara-sutra. The sutra reads': pramāṇašyāgauṇatvād anumānad artha-niscayo durlabhah. "Determination of object through inference is impossible, because means-of-knowledge is not secondary."
The only one I know of who has tried to explain the Paurandara-sutra is Professor Solomon in her study of Bhatta Udbhata. Her explanation follows closely Vädi Devasuri's Syādvādaratnākara. "The nature of heru (reason), which gives rise to inferences is that it is paksa-dharma, attribute of the paksa (the minor term, or subject of the syllogism). Now, the paksa is of the nature of an aggregate of attributeand-thing: the total entity constituted of the thing and its attributes is called paksa. And if that is not ascertained, how can it be ascertained that the hetu is its attribute;.
Page #2
--------------------------------------------------------------------------
________________
Paurandnrusātra
155
and if the paksa is ascertained, anumana would no longer have any purpose to serve. So in order to establish or justify the usage of pakṣadharma (as defining characteristic of hetu), the word paksa, though conventionally denoting the aggregate of attributesand-thing, has necessarily to be understood in a secondary sense as indicating or meaning the thing (alone), which is a part of the total denotation. Thus the pakṣa is gauna (secondary), and that being so, the he tu also is gauna because it is its attribute. So anumana, which is produced by a gauna cause, is also gauna, and being. gauna it is not a pramāṇa."
In my opinion the above explanation does not make sense, and I will try to explain the sutra differently. However, what follows should by no means be considered as a criticism of E. Solomon who simply explains Vadi Devasūri's interpretation of Purandara, and in so far her explanation is most accurate."
she is probably right in assuming that the above interpretation induced Bhatta Udbhata to give a new interpretation of the sutra. I do not think, however, that Vadi's interpretation conveys the original meaning of the sutra, for it simply makes
no sense.
First note the exact words of the hetu: pramäṇasyägauṇatvät. If Purandara wanted to say what Vädi says he wanted to say, this would be a very oblique way of expressing himself--a gauna argument indeed--for he could simply say pakṣasyagauṇaIvat. But this in itself is not enough to discredit Vädi's interpretation; so let us take a closer look at the content of the argument.
According to Vadi, Purandara's argument is based on the fact that, when applied to inference, the word paksa is used in a secondary or metaphorical meaning (upacāra). Now, the question that arises is: Is there something wrong in using a word, and more specifically the word pakṣa, in a secondary meaning?
The Sanskrit texts provide a definite answer to this question: under certain conditions, there is nothing wrong with using the word paksa in its secondary meaning.
The use of the word paksa in its secondary meaning goes back to Dinnaga (PS, III. 8), and at the beginning of the Hetubindu Dharmakirti enters into a rather detailed polemic against Iśvarasena in order to justify Dinnaga's position. By looking into this polemic we can see when the word may be used in its secondary meaning, and when not.
Iśvarasena's objection is quite clears: prayojanābhāvād anupacāraḥ. Thus the argument presupposes that one should not use a word in a metaphorical sense unless one has a reason or a purpose (prayojana) for doing so. And Dharmakirti's answer shows that he shares Iśvarasena's presupposition, for it consists in showing that Dinnaga had good reasons for the employment of the word paksa in its secondary meaning. Therefore, if Purandara wanted to convince anybody of his argument, one would expect him to give some reason why the metaphorical use is not appropriate
Page #3
--------------------------------------------------------------------------
________________
156
Eli Franco
Not only does he not do so, but, as a matter of fact, he argues (again according to Vadi) in favour of the opposite view : In order to establish the usage of pak şadharma (pakşadharma-vyavahāra-siddhaye), the word paksa has to / must be employed in its secondary meaning to denote the dharmin alone (pak şa-sabdas tad-ekadese dharminyupacaraniyah). But this presupposes that one should never use a word in its secondary meaning, which is nonsense.
Now, if an argument makes no sense it is always better to assume that we do not understand it, or that something went wrong with the transmission of its original meaning, rather than to conclude that the argument really is nonsensical. When we deal with Lokāyata fragments, we have to be even more cautious, because they appear to us not in their original context, but as pūrvapakşas in their opponents' works, and in such a case even a deliberate misrepresentation cannot be excluded. Therefore, before we can reconstruct the original meaning of a fragment, we have to take into account all sources where it is quoted or referred to. If a fragment is always interpreted in them in the same manner, and if this interpretation is acceptable within the general framework of Lokāyata, we can reasonably conclude that its original meaning is not misrepresented or misunderstood. This, however, is not the case with the Paurandarasūtra, for two more interpretations of it are to be found in Prabhācandra's Prameyakamalamārtanda (PKM) p. 473.8-97 yac-cāgaunatvāt pramānas yety uktam tatrānumānasya kuto (gaunatvam,) gaunārtha-vişayatvāt, pratyakşapūrvakatvād vā? "As for the (hetu) stated above, namely because a means of knowledge is not secondary', why does it apply) to inference? Is it because its object is a secondary (i.e. unreal) thing, or is it because it is preceded by perception ?"
According to the first interpretation given by Prabhācandra the word gauna directly refers to inference, and not indirectly through the word paksa as stated by Vādi Deva. In other words, the argument intends to show that inference is not a means of knowledge in the strict sense of the word. When formulated in this way, one immediately recalls that this is precisely the Buddhist position, because every inference involves universals, and universals, according to the Buddhists, are pure mental constructions. As Dharmakirti paradoxically puts it : "[an inferential cognition] is right cognition, inspite of its being false, because it is connected with the object."
Interpreted in this manner, the argument starts to make sense. It becomes convincing, for it shows up one of the weakest points in the Buddhist theory of knowledge. Indeed, the Buddhist logicians never quite managed to explain how an inference which has a fictitious entity for its object, is nevertheless connected with reality, and leads to efficient action.
Attractive though it is, I feel uneasy about this interpretation, because it limits the scope of the argument to Buddhist inference alone, as was also clearly realized by Prabhācandra, for he says (PKM, 473.9-474.1): na tāvad adyo vikalpaḥ, anumānasyāpy adhyaksavad västava-sāmānya-višeşātmakārtha-vişayatvābh yupagamāt. na khalu kalpita-sāmānyārtha-visayam anumānam saugatavaj jainair iştam ......
Page #4
--------------------------------------------------------------------------
________________
Paurandarasūtra
157
“The first alternative (i.e. that inference has an unreal object), to begin with it, is not applicable to our inference], because (we) accept that inference, just like perception, has for its object a real thing the character of which is general as well as particular. Unlike the Buddhists, the Jainas certainly do not accept an inference the object of which is a thing (that is) a mentally constructed universal."
Of course, the argument of the Lokāyata could be directed against a specific conception of anumāna, but the sūtra's formulation looks more like a general objection against inference. If the sūtra is to be undersood as a general argument against inference, then it presupposes a refutation of the universal, and since its original context is unknown to us, this possibility cannot be excluded. I cannot see, therefore, any reason to reject this interpretation.
However, the second interpretation proposed by Prabhācandra is at least equally possible.
According to this interpretation, the argument does not have anything to do with the primary or secondary meaning of words. (Note how the meaning of the word gauna changes according to the different interpretations.) It is inference, the cognitive process itself, which is secondary, and it is secondary to perception.
A mere look at this interpretation immediately recalls the definition of inference given in the Nyāyasūtra 1.1.5 (tat-pūrvakaṁ trividham anumānam...) where tat-refers to pratyak sa of 1.1.4). But unlike the previous interpretation, the present one is certainly not limited to the Nyāya theory of inference alone, for all Indian philosophers agree that inference, at least in the final analysis, is based on perception.
One might ask : So what? we admit that inference is preceded by perception, but why should this fact prevent us from calling it a means of knowledge ?
Taken at its face value, the argument seems to presuppose that a means of knowledge has to be absolutely autonomous, but this would be a rather strange presupposition. The Naiyāyikas who defined inference as preceded by perception were not bothered by this dependence. Neither do I see anything in the dependence relation as such to prevent inference from being a means of knowledge. The dwarf who sits on the giant's shoulder depends on the giant, but neveptheless reaches higher.
I think we have, therefore, to look for a more specific type of dependence relation.
Unfortunately, the extant fragment does not specify which kind of dependence relation Purandara had in mind while formulating his argument, and in the present state of our knowledge we cannot reach any definite conclusion on this point.
Drawing on our knowledge of other Indian philosophers, there are, roughly speaking, two reasons why a means of knowledge is not admitted. Either it is not valid (validity is not to be reduced to truthfulness, it can include other conditions such as having a new object etc.), or it is included in another means of knowledge
Page #5
--------------------------------------------------------------------------
________________
158
Eli Fronco
(for the Buddhists for instance, sabda is not a pramāņa because it is included in anumāna).
As a matter of fact, Prabhācandra considers the second possibility at the very beginning of the Carvāka-pūrvapaksa (PKM, 472.2-3): ye hi pratyakşain ekam eva pramānam ity acakşate na teşām anumānādi-pramāṇāntarasyāntarbhāvah sambhavati tad-vilaksanatvād vibhinna-sāmagri-prabhavatvāc ca. “Those who say that perception is the one and only means of knowledge, for them inference and the other means of knowledge cannot be included [in perception), because they f.e. the other means of knowledge) are defined differently (from perception), and because they arise from different causal complexes."
of course, we cannot conclude that the Cārvākas in fact maintained this position just because the possibility is raised (and rejected).
If, on the other hand, the dependence of inference on perception is to be understood in such a way that it invalidates inference completely (i.e. as a means of knowledge in the technical sense of the word, this does not stand in contradiction to accepting inference as useful for everyday purposes, cf. below), then the dependence has probably to be understood as the dependence on the apprehension of vyāpti.
Now, the apprehension of vyāpti is not due to inference itself, because in order to make an inference, one already has to know a vyāpti. Cf. PKM, 473.3-4 : näpi (vyāpti-grahanım] anumānatch, asya vyāpti-grahana-purassaratvāt, tatrāpy anumānato vyāpti-grahane 'navasthetaretar aśrayatva-dosa-prasangah.
"Nor is (the apprehension of vyāpti] due to inference, because it (i.e. inference) is preceded by the apprehension of vyāpti. If there too (i.e. in the case of the vyāpti which precedes the inference) the apprehension of vyāpti is due to inference, this will result in the fallacies of infinite regress or mutual dependence."
If one cannot apprehend the vyāpti by inference, it has to be apprehended by perception, but the vyāpti cannot be apprehended by perception either, because perception is limited to objects which are directly in contact with one's sense organs, whereas the vyāpti applies to a whole group of things which are beyond the range of one's sense organs. Thus, inference not only depends on perception, but it also depends upon something perception cannot provide, and that is why it is not valid.
There is still another way of interpreting the dependence of inference on perception. Suppose, at least for the sake of argument, we somehow manage to get an apprehension of vyāpti through perception. Now, a vyāpti relation does not obtain between particulars, but between universals alone. Thus, an inference cannot have a particular for its object, for in that case there will be no vyāpti. But if an inference has an universal for its object, then it serves no purpose, for the universal is already known before the inference is undertaken. Thus, inference is secondary to perception. because it apprehends an object already apprehended by perception. Cf, PKM, 472,
Page #6
--------------------------------------------------------------------------
________________
Paurandaras tra
5-6 artha-niscayakam ca jñānom pramāṇam, na ca anumānād artha-niścayo ghatatesamanye siddha-sadhanad viteze anugamābhāvāt.
159
"And a means of knowledge is a cognition which determines the object, but the determination of the object through inference is impossible, because if [the object of the inference is] a universal, it is a proof of what is already proved, [and] if [the object of the inference is] a particular, there is no concomitance."
It is clear from the above that I do not claim that any of these interpretations was actually intended by Purandara. What I have tried to do is to give possible explanation of the sutra within the Lokayata framework, that is, by using arguments which we know for sure to have been advocated by the Lokayatikas. The passages quoted from Prabhacandra follow immediately upon the quotation of the sutra, but we do not even know whether these arguments were formulated by Purandara or by other Lokayatikas and even if they are all Purandara's, they could be independent. arguments which did not have anything to do with the sutra."
Whatever the case may be, there is one more question to be asked: what did Purandara try to accomplish with his argument? He certainly did not try to discard inference altogether, for in a fragment quoted by Kamalaśfla he clearly states the opposite view: 10
Purandaras tv aha "loka-prasiddham anumanam cārvākair aptsyata eva, yat tu kaiścil laukikam mārgam atikramyānumānam ucyate, tan nişedhate" iti.
"Purandara said: 'inference which is well known in the world is certainly accepted by the Carvākas too, but that which is taught as (or: called) inference by certain [would-be philosophers] who transgress [with it] the worldly path (i.e. everyday innocuous inferences), that is rejected."
If we put these two fragments together (they are to the best of my knowledge the only ones of Purandara that have come down to us), the picture becomes clearer, for now we know that the sutra's aim is not to discard inference altogether, but to limit its scope to worldly affairs. By denying inference an independent status, that is, by emphasizing its dependence (gauna) on perception, Purandara wanted to restrict the field of operation (visaya) of inference to that of perceptible objects.
This is probably the most significant contribution of Purandara be the Lokayata school (at least it is the one for which he is remembered), and in order to understand its importance, we have to reconstruct its necessity.
The old position of the school as formulated in the Pṛhaspati-sutra, namely perception alone is a means of knowledge, involved too many internal contradictions. which became apparent when the theory of knowledge became the central preoccupation of Indian philosophers. Dharmakirti, for instance, formulated at the beginning of the Pramanaviniscaya11 three short, but devastating arguments against the Carvāka, and many more could be formulated in a similar vein. Arguments like these be urgently answered, or the Carvaka would have been kicked out of the philosophical scene.
Page #7
--------------------------------------------------------------------------
________________
160
Eli Franco
Purandara's answer consists in complying with this kind of argument without giving up the basic Lokāyata positions. Thus inference was accepted as far as it did not, or better, could not, contradict the Lokāyata positions, but as soon it was to be applied to prove the existence of Soul, God and other non-perceptible objects, which could endanger, if not annihilate, all the Lokāyatikas were fighting for-and I think it was ultimately to found social and political institutions independently of religious dogma-it was to be rejected as a non-valid means of knowledge.
In order to do this Purandara had to admit inference, but only up to a point, which is a rather difficult task, for once inference is accepted, there is no apparent reason why one should use it only as long as it suits the Lokāyata. Thus, the task which presented itself to Purandara was to find a reason why inference should never overlap perception. And his answer probably was : because it depends upon perception.
I argued above that Prabhācandra's interpretations of the sūtra are to be prefered to those of Vadi Deva and the others, because they make more sense. Now, suppose I am right in my premises, and that the sūtra does make better sense that way, does it follow that this was the original intention of the sūtra ?
Let me argue briefly in favour of the opposite view. The sūtra is mentioned and explained several times, and the majority of the sources interpret it in the same manner. Should we not accept the interpretation favoured by the majority ?
Moreover, Jayanta and Bhāsarvajña are roughly two centuries earlier than Parbhācandra (active c. A.D. 1025-1060), which means they are closer to Purandara's own time, and had, therefore, a better chance of first hand acquaintance with his work.
Further, Jayanta and the others give one sole interpretation of the sūtra, whereas Prabhācandra brings up two vikalpas in the PKM and no less than nine in the NKC. From this fact we can conclude that they were sure about the sūtra's meaning, whereas Prabhācandra was not quite certain about it.
Considering all this, we could make the following assumption : Bhāsarvajña and Jayanta were well acquainted with Purandara's thought, whereas Prabhācandra had only a fragmentary knowledge of it, and interpreted a nonsensical argument in an intelligent manner, but one which does not convey the original purport of the argument.
Such a view, however, no matter how well formulated, is self-defeating. There can be no reason why we should accept an illogical, confused interpretation rather than a logical one, for this would involve a contradiction in our presupposition. If we do not want to make our authors talk good sense, we have no business to comment upon them. Besides, any such reasoning can be cancelled by selfreference.
All this does not mean, of course, that what we take to be a nonsensical argument is in fact nonsensical, for we may have misunderstood it. But as long as our
Page #8
--------------------------------------------------------------------------
________________
Paurandarasūtra
161
ignorance is not enlightened, we have to stick to what we deem the more reasonable interpretation.
Now, if the interpretation given by Vādi Deva and the others is untrue, why was it repeated over and over again ?
They may have taken it from each other, or from a source unknown to us, but this is begging the question, for it arises again : Why was there at all an untrue interpretation of the sūtra ?
Frankly, I don't know; I can only suggest a hypothesis : it is due to a juxtaposition of two things which do not belong together. Someone who had DharmakTrti's discussion of the word 'paksa' in mind, concluded wrongly that Purandara was referring to it in his sūtra.
Having rejected the paksa interpretation we are still faced with the problem, which of the other two interpretations is to be accepted.
Here we cannot argue on grounds of method, because both of them make sense, and both of them are compatible with the Lokāyata position. The only way to decide between these interpretations would be to check the context. But dealing as we do with fragments, we have no context. Moreover, for what we know, it may have been both, for sūtras are sometimes deliberately formulated vaguely enough to allow different interpretations.
For the time being, I can see no way but to leave this question open.
NOTES
1. The sūtra is quoted in the following books :
Nyāyabhūşana of Bhāsarvajña, ed., S. YogIndrānanda, Varanasi 1968, p. 210.18. Nyāyamañjari of Jayanta Bhatta, ed. K. S. Varadacarya, Mysore 1969, p. 312.8-9 (Banares 1936 ed. p. 108.10). Pramānavārtikavrttiţikā of Karņakagomin, ed. R. Sāṁkstyāyāna, Allahabad 1943, p. 25.26 Prameyakamalamārtanda of Prabhācandra, ed. M. Sarraph and R. Jain. Vira Jñanodaya Granthamala 23, Delhi vi, ni. sam. 2504, p. 477. 2-3. Sarmatiarktaprakarana of Siddhasena Divākara, ed. with Abhayadevasūri's Vyākhyā by S. Sanghavi and B. Doshi, Gujarātapuratattvamandiragranthāvali, Ahmedabad 1924, pp. 70.27; 72.33; 554.23. Vyomavati of Vyomaśiva, ed. G. Kaviraj, Chowkhamba Sanskrit Series, Benares 1930, p. 575.14 (which reads pramāṇasya gaunatvāt...; this is obviously wrong). The sūtra is also quoted in the Syadvādaratnakara of Vadi Devasūri, Arahatambata Prabhākara ed. p. 261 (?). I was unable to consult the book, but it is quoted by E. Solomon, "Bhatta Udbhata", Annals of the Bhandarkar
Oriental Research Institute, Diamond Jubilee Volume, Poona 1978, p. 985.
Page #9
--------------------------------------------------------------------------
________________
162
Eli Franco
2. Cf. E. Solomon ibid, p. 985. 3. Cf. Syādvādaratnākara p. 261 : tato 'vaśyam paksu-dharma-vyavahāra
siddhaye dharma-dharmi-samudāye rūdho'pi pakşa-śabdas tad-ekadese dharminy-upacaranīyaḥ. tasmad itthaṁ pakşasya gaunatvam. Tad-gaunatve ca hetor-api gaunatā tad-dharmatvalakṣaṇatvād-asya. tasmad gauna-karanajanyatvena gauņam anumānam. Gaunatvāc ca na pramānam. Quoted by E. Solomon, ibid, p. 985 n. 1. ...CL E. Steinkellner, Dharmakirti's Hetubinduh, Wien 1967, vol. II, p. 83 n.5. .5. Cf. E. Steinkellner, ibid, vol. I, p. 31.10-11, 6. Cf. n. 3. 7. References are given to the Delhi ed. (n. 1.). If you use a different edi
tion, all the passages quoted here are from the very beginning of the Pratyakşoddeśa. It is interesting to note that Prabhācandra certainly knew the 'paksa' interpretation, although he does not mention it here. But in the Nyāyakumudacandra he gives no less than nine vikalpas for the word gauna, the seventh of which reads dharmini pakşa-sabdopacārāt. Unfortunately, these vikalpas are dealt with too briefly. I think most of them can be subsumed under the interpretation given in the PKM; it is improbable, however, that all the nine of them were advocated by the Cārvākas, let alone by Purandara himself, as a direct explanation of the sūtra. Cf. Nyāyakumudacandra ed. Mahendrakumar, Mānik Chandra Dig. Jain Series 38, Bombay 1938, vol. I p. 71.1-4; yac ca "agaunatvāt' ity uktam (cf. p. 67.16), tatrā-numā nasya kuto gaunotvam-avisada-svabhāvatvāt, svärtha-niscaya parāpek şatvāt, visamuādakatyst, pratyaksa-pūrvaeatvāt, asthad anutpadyamanatvāt, avastu-visayatvāt, dharmini pakşa-sabdo-pacarat, badhyamānatvāt, sādhya
sādhanayoḥ pratibandha-prasādhaka-pramāņābhāvad va ? 8. Cf. E. Steinkellner, Dharmakirti's Pramānaviniscayah, Zweites Kapitel :
Svārthanumānam, Wien 1973, vol. I p. 25: bhrāntir apy artha-sambandha
tah pramā. 9. The fact that this is not an arbitrary juxta position on our part is corro
borated by the passage quoted in n. 7, where similar interpretations are
directly related to the word gauna. 10. Cf. Tattvasangraha of Shāntaraksita with comm. 'Pañjikā' of Kamala
sila, ed. D. Shastri, Bauddha Bharati Series 1, Benares 1968, ad. v.
1481-82 p. 528.9-10. 11. Cf. T. Vetter, Dharmakirti's Pramānaviniscayah, 1. Kapitel : Pratyakam,
Wien 1966, p. 34 n. 1: pramānetara-sāmānya-sthiter anya-dhiyo gateh l
Page #10
--------------------------------------------------------------------------
________________ Paurandarasurra 163 pramanantara-sadbhavah pratisedhac ca kasyacit il "There is another means of knowledge (viz. inference ), because of the general determination of means of knowledge and non-means of knowledge, because of apprehension of cognition of another [person), and because of negation of anything (whatsoever)." The verse is quoted by Prabhacandra at the end of refutation of the Carvaka-purvapaksa PKM, p. 477.4-5 (unidentified by the editor).