Page #1
--------------------------------------------------------------------------
________________ MISZELLEN ZUR ERKENNTNISTHEORETISCH-LOGISCHEN SCHULE DES BUDDHISMUS By Ernst Steinkellner, Vienna VI. A New Approach towards Improving the Textual Basis for an Understanding of Dignaga's Pramanasamuccayavrtti The deplorable quality of the two Tibetan translations of Dignaga's Pramanasamuccayavrtti has been noted by many scholars and commented on in detail by HATTORI1. What can be done for a proper understanding of Dignaga's arguments has also been stated concisely. by HATTORI: "... to conjecture as far as possible the original Sanskrit form through a comparison of K with V [i.e. the two translations] and with PST [i.e. Jinendrabuddhi's Tika], when this quotes the text. In this connection, we must utilize fully the related Sanskrit materials" 2. The numerous extant Sanskrit fragments are indeed the most important basis for regaining parts of the original and for proper evaluation of the meaning of the Tibetan translations. But one possibility of improving the basic textual material has not been taken into consideration so far: the collection of Tibetan fragments from the Pramanasamuccayavrtti in other works of the school rendered into Tibetan by other, and better translators. A collection of this kind would result in improved knowledge of the Sanskrit original as well as a more valid interpretation of the text in cases where no Sanskrit fragment is extant: First of all, in the latter case we would have a better and more reliable Tibetan version of a certain textual unit to interpret and to compare with the pratika-texts quoted by Jinendrabuddhi. Secondly, even if we have a Sanskrit fragment, this Sanskrit text can better be examined by comparing it with a more reliable Tibetan version than with the two fully extant translations which one would hardly ever dare to take as justification for a textual judgement against an extant Sanskrit fragment. 19 1 Dignaga, On Perception, being the Pratyaksapariccheda of Dignaga's Pramanasamuccaya from the Sanskrit fragments and the Tibetan versions. Translated and annotated by MASAAKI HATTORI, Cambridge, Mass., 1968, 18f. Op. cit., 18. WZKS 33 (1989) 177-181
Page #2
--------------------------------------------------------------------------
________________ 178 E. STEINKELLNER In my review of HATTORI's book? I opposed his attempt to correct the Tibetan of the extant translations because this would result in creating a Tibetan text that never existed", and insisted on presenting these translations as they were transmitted in the Tibetan canon's versions in order not to distort the basis for textual interpretation and reconstitution. But by collecting Tibetan translations of quotations from the Pramanasamuccayavrtti in other works which were translated far better in most cases, we actually have access - at least for a great number of textual units - to a Tibetan translation which is not only superior in quality but is also historically authentic. Of course, the fact that they come from different works implying possibly different treatment of the quoted passages by their respective authors, and that various teams of translators have been at work must be taken into serious consideration. But nevertheless, this would be a very promising new channel for philological efforts devoted to this most important and also difficult of Dignaga's works. One would, in fact, gradually be able to work on the basis of much better Tibetan translations for many passages of import than hitherto possible. It will also be found that these texts correspond much more closely to the pratikas in Jinendrabuddhi's commentary, itself the product of better translators, A simple example should be sufficient to illustrate this approach. In Dharmakirti's commentary on Pramanavarttika III 294, where he states that for those who think that even a case of erroneous senseperception belongs to the mind (manasam), there would be a contradiction with Dignaga's statement, that the sense organs are also the cause for the cognition of blue, two moons etc. (niladvicandradidhiyam hetur aksany apiti), Devendrabuddhi literally quotes the passages referred to from the Pramanasamuccayavitti: slop d pon gyis rTsod par sgrub par bsad par mnon sum du dpyad pa la [1.] "dmigs pa'i don dpyad pa na 'di ltar 'ji ltar snan ba las ses pa skye ba de ltar dmigs pa zes brjod dam | 'on te ji ltar yod pa dag gzan du snan ba'i ses pa'i rgyur 'gyur" zes bya ba skabs su mdzas nas | [2.) "mig la sogs pa'i bar yan dmigs pa nid du thal na de dag kyan don dam par rnam pa gean du yod pa dag ni zla ba gnis par snan ba dan snon po la sogs pa'i snan ba'i ses pa'i rgyur 'gyur ro" zes bsad pa'i gzun yin pa dan 'gal lo || 3 WZKS 15 (1971) 222-224. 4 This is in fact the case with HATTORI's edition of Kanakavarman and Dad pa'i ses rab's translation where the transmitted text in many cases is to be found only in the footnotes.
Page #3
--------------------------------------------------------------------------
________________ Miszellen zur erkenntnistheoretischen Schule des Buddhismus 179 The first fragment is from the Vrtti on PSI 14 (HATTORI K 189, 8-11). The beginning of our fragment: dmigs pa'i don shows the correct text which was proposed by HATTORI on the basis of the Tika and the Sanskrit fragment from NCV 96,7 (cf. HATTORI's note 2.15) against the reading dmigs pa la don du of both the V and K translations. The introductory 'di ltar of our fragment shows that the following alternative interpretations were introduced by an explanatory introduction like *tatha hi and not a word introducing an alternative question. That Tibetan fragments of this kind are never totally reliable is demonstrated by the formulation of the first alternative. For this we have a Sanskrit fragment from NCV 99,23 (cf. 96, 11f.) which corresponds nicely with K's translation but is not well represented in our Tibetan fragment, where tesu and te are not translated, and ji Itar snan ba las can at best be understood as a corruption of *ji ltar snan ba de rnams la The translation of the second alternative again corresponds better to the systematical sequence of the Sanskrit fragment known from NCV 99,29 than the K and V translations, although it does not translate the important api which is well represented in K and V. The original Sanskrit text can be assumed to have had the following form on the basis of the extant fragments, the Tibetan pratikas of the sika, the two Tibetan translations and the Tibetan translation of the quotation in Devendrabuddhi's commentary rupadisv alambanartho vaktavyah. (tatha hi) yadabhasam tesu jnanam utpadyate, tatha ta alambanam (ity ukta atha) yathavidyamana anyabhasasyapi vijnanasya karanam bhavanti. "With reference to colour and the like, one should state what is meant by 'the object of cognition'. Namely: they are called 'object of cognition' [as some hold) in that appearance with which a cognition of them arises, or (as others hold] they become as they are a cause of the cognition, although [these cognitions) present an appearance different from them)." The second fragment is from the Vstti on PSI 15 (HATTORI K 191, 1-4). The Sanskrit original is known from a quotation in Manorathanandin's commentary (PVV 206,26-207,2). Our fragment has three major advantages over the translations K and V: They split the second, causal clause into two, ending the first part with a final particle and thus leaving the second part without subject. They both begin with the 5 As proposed in HATTORI's (33) and JAMBUVIJAYA's translations (cf. his note in NC 96,25). 6 Sanskrit words not found in the fragments are tentatively supplied within pointed brackets.
Page #4
--------------------------------------------------------------------------
________________ 180 E. STEINK ELLNER incorrect dmigs which was emended to mig in K by HATTORI which is to be found in our version. And moreover they read only yan as against the bar yan of our fragment. The latter was the reason why HATTORI understood the Sanskrit fragment as beginning with caksuradinam when in fact it begins with yavaccaksuradinam. But this can only be seen on the basis of the translation of this quotation in Devendrabuddhi's commentary. Thus it is evident that a collection of the Tibetan versions of quotations from the Pramanasamuccayavrtti in the various works of the Dignagean tradition can improve our textual basis considerably, and until a Sanskrit manuscript of the Vrtti is found", this valuable additional source of textual information should not be neglected. VII. Another Note on pramanabhuta In note 6 to the introduction of his edition and translation of the sections on the Buddha and the four noble truths in Dharmakirti's Pramanavarttika, pramanasiddhi-chapter', TILMAN VETTER gives the reason for his translation of the term pramanabhuta from the mangala of Dignaga's Pramanasamuccaya as "who is a means of valid cognition" ("der Erkenntnismittel ist"): the interpretation that the Buddha has become a means of valid cognition (as if the text read pramanibhata) which does not figure at all in Dignaga's own explanation of the verse and can, therefore, hardly be based on the term pramanabhuta, was added by Dharmakirti; and he refers to PV II 7-8 as the source of this idea. Since the compound pramanabhuta does really not have a cvi-formation (cf. Panini 5.4.50), VETTER is certainly right in asking for ant interpretation of 'bhata- as the equivalent of the copula serving the simple adjectivization of the first member. On the other hand Dharmakirti and the tradition following his interpretation interpret "bhataas having the function of excluding what has not become, what is eternal". This is also the interpretation presupposed in Jinendrabuddhi's explanation and analysis of the compound". Jinendrabuddhi 7 A Sanskrit manuscript of Jinendrabuddhi's Tika is said to have been found in the People's Republic of China recently. 1 Der Buddha und seine Lehre in Dharmakirtis Pramanavarttika. Der Abschnitt uber den Buddha und die vier edlen Wahrheiten im PramanasiddhiKapitel. Eingeleitet, ediert und ubersetzt. [WSTB 12]. Wien 1984, 14 with note 6. 2 abhutavinivrttaye bhutoktih PV II 7bc. 3 Cf. E. STEINKELLNER, Some Sanskrit Fragments of Jinendrabuddhi's Visalamalavati. In: A Corpus of Indian Studies. Essays in Honour of Professor Gaurinath Sastri, Calcutta 1980, 96-105.
Page #5
--------------------------------------------------------------------------
________________ Miszellen zur erkenntnistheoretischen Schule des Buddhismus 181 explains in fragment 14: bhuta utpannah ...5. bhutavacanam aprajatasyesvaradeh ... ... nityasya pratisedharthame, and ends with an analysis of the compound as a dvandva: pramanan casau bhutas ceti pramanabhutah. While it is quite clear that Jinendrabuddhi explains the meaning of obhuta under the influence of Dharmakirti's interpretation, his analysis of the compound as a dvandva results in a strange meaning of the term: the Buddha is accordingly said "to be a means of valid cognition, an authority, as well as to have become". In other words, the second member of the compound is no longer understood as an equivalent of the copula - now being understood with the first member of the compound, pramana, - but as a predicate in its own right. Thus the term combines two predicates of the Buddha: he is an authority (pramana) and he has become (bhuta). It is evident that Jinendrabuddhi tried in this rather forced way to reconcile Dharmakirti's interpretation of the word obhuta- with the fact that Dignaga's compound does not show a con-formation. This means that, following VETTER, we indeed have to translate the term pramanabhuta in Dignaga's mangala as "who is a pramana" but as interpreted by Dharmakirti and the following tradition the term must be translated as "who has become a pramana". 4 Ibid., 100. 5 The Tibetan text continues: ... byun ba ces pa'i don to, where byun ba may translate either *pradurbhuta or *prajata. 6 Cf. ibid., note 18 for this sentence.