I. The negated qualifier avinirmuktakleśakośa and the unnegated qualifier *vinirmuktakleśakośa in variant Tibetan translations of a definition of tathāgatagarbha in relation to dharmakāya in the Śrīmālādevīsiṃhanādasūtra

In E.H. Johnston's edition of the commentary (the RGVV) on Ratnagotravibhāga (RGV) i.12 published at Patna in 1950 (p. 12.14–5) we read: ayam eva ca bhagavaṃs tathāgatadharmakāyo ’vinirmuktakleśakośas tathāgatagarbhaḥ sūcyate [ity ucyate?], “Blessed one, the dharmakāya of the Tathāgata which is not free from the covering of defilements is called tathāgatagarbha”. In this basic commentary on the RGV the sentence forms part of a quotation from the Śrīmālādevīsiṃhanādasūtra (ŚMDSS), an important sūtra source on the tathāgatagarbha doctrine included in the Ratnakūṭa collection, the original Sanskrit text of which is now available only in fragments.

In rṄog Blo ldan šes rab's (1059–1109) Tibetan translation of this quotation contained in the bsTan ’gyur, the corresponding text reads in Z. Nakamura's edition (Tokyo, 1967: 21): bcom ldan ’das de bžin gšegs pa'i chos kyi sku ’di ñid ñon moṅs pa'i sbubs las ma grol ba ni de bžin gšegs pa'i sñiṅ po žes bgyi'o. This reading with the negative particle ma is found in the bsTan ’gyur from sDe dge (and in the derived one from Co ne); and it is supported by rGyal tshab Dar ma Rin chen's (1364–1432) commentary on the RGV(V) (Theg pa chen po rgyud bla ma'i ṭīkā = Dar ṭīk (lHa sa ed.)), f. 42b3). But in Ye šes sde's (c. 800) Tibetan translation of the ŚMDSS, the corresponding passage omits the negative and reads (in Tsukinowa's edition (Kyōto, 1940), p. 130): bcom ldan ’das de ñid la de bžin gšegs pa'i sñiṅ po chos kyi sku ñon moṅs pa'i sbubs nas ṅes par grol ba žes bgyi'o. This reading, where the qualifier is unnegated, is attested in bKa’ ’gyur editions from Beijing, sNar thaṅ, sDe dge, and lHa sa, and also in the De bžin gšegs pa'i sñiṅ po gsal žiṅ mdzes par byed pa'i rgyan (mDzes rgyan, lHa sa ed., f. 31a–b) by Bu ston Rin chen grub (1290–1364), an early compiler-editor of the bsTan ’gyur. Equally remarkably, the negative particle ma is also omitted from this passage as quoted in the translation of the RGVV in the bsTan ’gyur editions from sNar thaṅ and Beijing.Footnote ¹

A crucial definition of tathāgatagarbha as attested in different portions of the Tibetan canonical tradition thus varies by either including or omitting the negative particle ma before the translation of (a)vinirmuktakleśakośa in the bKa’ ’gyur translation of the Śrīmālāsūtra and its citation in the bsTan ’gyur translation of the RGVV. The interpreter of the tathāgatagarbha doctrine, as well as the scholar who might wish to restore the original Sanskrit from the Tibetan version of the text, is therefore confronted with a major problem that is both philological and hermeneutical.

A prima facie solution might be provided by conjecturing that, in sentence saṃdhi, an avagraha (usually transliterated by the apostrophe ’) representing the Sanskrit privative particle a ^o either was not written by a scribe (something not unknown in manuscripts) or was mistakenly overlooked by a translator (something that might fairly easily occur in this case). Indeed, according to a standard formulation of the classical doctrine, on the level of ordinary sentient beings the tathāgatagarbha possessed by all these beings without exception is not (yet) free from all the coverings of defilements (kleśakośa).

Yet to conclude that the Tibetan reading lacking the negative particle ma is nothing but an error in translation or textual transmission would probably be to oversimplify what has been a crucial hermeneutical and philosophical issue. For, in one formulation of the doctrine, tathāgatagarbha does appear to have been regarded as (actually) free from all the coverings of defilements and thus to be (already directly) identifiable with dharmakāya. This last view will no doubt be correct proleptically, outside time as it were. Indeed, if this were not so, tathāgatagarbha would be conceived of as subject to change in the course of being made free of the adventitious (āgantuka = glo bur ba) defilements through spiritual practice. And, instead of being unconditioned, it would then be conditioned (samskṛta). But this may not be doctrinally coherent and acceptable.

Concerning this issue of central importance, reference can be made to the sūtra-quotation traceable to the Anūnatvāpūrṇatvanirdeśaparivarta and found in RGVV i.48 (p. 41.2–4: dharmakāyaḥ sarvakleśakośaparimuktaḥ sarvaduḥkhātikrāntaḥ sarvopakleśamalāpagataḥ śuddho viśuddhaḥ paramapariśuddhadharmatāyāṃ sthitaḥ sarvasattvālokanīyaṃ bhūmim āruddhaḥ ... tathāgato ’rhan samyaksambuddha ity ucyate); to RGVV i.24 (samalā tathatā yo dhātur avinirmuktakleśakośas tathāgatagarbha ity ucyate); to RGVV i.27–8 (samāsatas trividhenārthena sadā sarvasattvās tathāgatagarbhā ity uktaṃ bhagavatā / yad uta sarvasattveṣu tathāgatadharmakāyaparispharaṇārthena tathāgatatathatāvyatibhedārthena tathāgatagotrasambhavārthena ca); to RGVV i.149–52 (p. 73.9–11, with some variant readings in the Tibetan translation), also on the connection between tathāgatagarbha and dharmakāya, tathatā and gotra; and to RGVV ii.1 (yo ’sau dhātur avinirmuktakleśakośas tathāgatagarbha ity ukto bhagavatā ...). In RGVV i.25, moreover, the tathatā that is stained is stated to be, simultaneously, both very pure and affected, a state described as inconceivable (p. 21.17–8: samalā tathatā yugapad ekakālaṃ viśuddhā ca saṃkliṣṭā cety acintyam etat sthānam). The Anūnatvāpūrṇatvanirdeśaparivarta, as quoted in the RGVV i.1, points to a closely related issue when it declares both that (the name) “sattvadhātu” is an expression for tathāgatagarbha and that (the name) “tathāgatagarbha” is an expression for dharmakāya (p. 2.12–3: sattvadhātur iti ... tathāgatagarbhasyaitad adhivacanam/tathāgatagarbha iti ... dharmakāyasyaitad adhivacanam). And the same sūtra teaches the non-duality in reality of sattvadhātu and dharmakāya, the difference being only verbal (quoted at RGVV i.50, p. 41.15–6: nānyaḥ sattvadhātur nānyo dharmakāyaḥ/sattvadhātur eva dharmakāyaḥ, dharmakāyaḥ eva sattvadhātuḥ/advayaṃ tad arthena, vyañjanamātrabhedaḥ). Following RGVV i.1, this is the fourth vajrapada or “adamantine topic” of the RGV.

Concerning the twofold aspect of the trans-cosmic dhātu manifesting as both tathāgatagarbha and dharmakāya, but which is ultimately advaya “without duality”, see also below.

Instead of a mere error of translation or textual transmission affecting some Tibetan texts of the Śrīmālāsūtra and the RGVV, we find here a striking ambivalence – even a deliberate paradox – integral to the very formulation of the buddha-nature doctrine in the Buddhist traditions. The problem will then not be purely text-historical, it will also be philosophical (and meditational). Inevitably, this has important implications for the subsequent history of the doctrine.Footnote ²

It is no simple matter to determine the extent to which the reading “freed from the covering of kleśas” might have contributed to the doctrine of tathāgatagarbha being considered intentional (dgoṅs pa can = ābhiprāyika) and of provisional meaning (draṅ don = neyārtha) by some of its interpreters, such as Sa skya paṇḍi ta Kun dga’ rgyal mtshan (1182–1251) and Bu ston Rin chen grub (1290–1364). This interpretation of tathāgatagarbha was current for other reasons also – in particular because of the position taken by the great Indian Mādhyamika Candrakīrti – and it was probably only reinforced by the textual variant found in certain sources where the absence of the negative particle in the description of tathāgatagarbha in relation to its surrounding cover of kleśas made it definitionally indistinguishable from the definition of dharmakāya.Footnote ³

II. The qualifiers (a)muktajña and (a)muktajñāna and the problem of their various renderings in Tibetan versions of the Śrīmālādevīsiṃhanādasūtra and the commentary on the Ratnagotravibhāga

In two passages from the Śrīmālādevīsiṃhanādasūtra quoted in the Sanskrit commentary on the Ratnagotravibhāga (RGVV i.86 and 154–5), the term amuktajña is attested as a qualifier of the buddha-qualities (buddhadharma) with which the tathāgatagarbha is inseparably endowed. And it is specified (i.154–5) that the tathāgatagarbha is not Empty (aśūnya) of these imponderable/inconceivable (acintya) properties, which are accordingly described as inseparable (avinirbhāga) and amuktajña. The same sūtra states further that the tathāgatagarbha is Empty (śūnya) of all the coverings of defilements (sarvakleśakośa), which are then described as separable (vinirbhāga) and muktajña. The relevant passage of the RGVV i.154–5 reads (p. 76.8–10): śūnyas tathāgatagarbho vinirbhāgaiṛ muktajñaiḥ sarvakleśakośaiḥ/ /aśūnyo gaṅgānadīvālikāvyativṛttair avinirbhāgair amuktajñair acintyair buddhadharmaiḥ. The linguistic form and value of the compound muktajña and its negative are not, however, perfectly plain at first sight.

Happily, two closely related expressions, amuktajñāna and muktajñāna, are attested in RGVV i.149–52, where another passage from the Śrīmālāsūtra is quoted. This second citation reads (p. 73.2–4): tasmād bhagavaṃs tathāgatagarbho niśraya ādhāraḥ pratiṣṭhā sambaddhānām avinirbhāgānām amuktajñānānām asaṃskṛtadharmāṇām/ /asambaddhānām api ... vinirbhāgadharmāṇāṃ muktajñānānāṃ saṃskṛtānāṃ dharmāṇāṃ niśraya ādhāraḥ pratiṣṭhā tathāgatagarbhaḥ. Here too the precise meanings of the qualifiers (a)muktajñāna are perhaps not entirely plain; but ^ojñāna- in the second pair of qualifiers is apparently equivalent to ^ojña- in the first pair.

In this passage of the sūtra, it has been declared that tathāgatagarbha is the base, ground and support for the unconditioned dharmas – which are conjoined, inseparable and amuktajñāna – and also the base, ground and support of the conditioned dharmas – which are unconjoined, separable and muktajñāna. This sūtra passage thus points to the twofold aspect of the tathāgatagarbha as partaking of both the conditioned and the unconditioned. Another facet of this twofold aspect, but ultimate non-duality (advaya), is stated in a verse (sometimes said to be from a Mahāyānist Abhidharmasūtra) quoted in RGVV i.149–52 (p. 72), where it is termed dhātu:

Now, instead of definitively clarifying the exact meanings of the qualifiers amuktajña and amuktajñāna, the Tibetan translation of the ŚMDSS by Ye šes sde and his collaborators and the translation of the RGVV by rṄog Blo ldan šes rab and Sajjana complicate the matter.

In rṄog's Tibetan version of the RGVV (ed. Z. Nakamura), the translation-equivalent of both amuktajña and amuktajñāna is bral mi šes pa, and that of their antonyms muktajña and muktajñāna is bral šes pa. But, very remarkably, in the formulaic set of qualifiers including avinirbhāga, amuktajña ... (to which aśūnya may be joined), the Tibetan translation of the ŚMDSS (ed. Tsukinowa, 130–2) ascribed to Ye šes sde reads tha dad du gnas pa ma grol bas šes pa. And in the set vinirbhāga, muktajña ... (to which śūnya may be joined) this translation gives saṅs rgyas kyi chos tha dad du mi gnas pa grol bas/bar šes pa. This passage is very similar to, but not identical with, the quotation from the ŚMDSS in RGVV i.154–5 (p. 76.8–10) cited above. The negative particle in the Tibetan translation-equivalents thus appears inverted, so that Ye šes sde's magrol bas šes pa corresponds to rṄog's bral šes pa = muktajña, and Ye šes sde's grol bas/bar šes pa corresponds to rṄog's bral mi šes pa = amuktajña.

The situation is similar in the parallel passage of the ŚMDSS quoted in RGVV i.149–52 (p. 73.2–3, cited above) where, instead of (a)muktajña, the qualifiers (a)muktajñāna are attested. Here Ye šes sde's translation reads (p. 146) tha dad du mi gnas šiṅ ’brel la sbubs nas grol ba'i šes pa can dag ... ’brel ba ma mchis šiṅ tha dad du gnas la šes pa grol ba ma lags pa ... . In the corresponding passage rṄog's translation has ’brel ba rnam par dbye ba med pa bral mi šes pa ... ma ’brel ba rnam par dbye ba daṅ bcas pa(’i chos) bral šes pa, that is, the rendering that it employs also for (a)muktajña.Footnote ⁴ In other words, Ye šes sde's translation offers for (a)muktajñāna a rendering different from the one it provides for amuktajña; attested in two closely related passages separated by only a few folios in his translation of the ŚMDSS (pp. 130–2 and p. 146 in Tsukinowa's ed.), both of Ye šes sde's renderings are problematic.

In their contexts, however, the sense intended in the relevant passages of the ŚMDSS as quoted in the Sanskrit commentary on the RGV seems clear enough. For muktajña qualifying the kleśakośas and muktajñāna qualifying the samskṛtadharmas, the meaning appears to be “having knowledge separate”; and for amuktajña qualifying the buddhadharmas and amuktajñāna qualifying the asaṃskṛtadharmas, the meaning appears to be “having knowledge not separate” (or: “not having knowledge separate”). The qualifiers in question are possessive adjectival, i.e. bahuvrīhi, compounds. It is difficult to account for the discrepancy between Ye šes sde's and rṄog's Tibetan translations of the sūtra passages; and in rṄog's translation it is no simple matter to determine exactly what he intended the Tibetan translation-equivalents to convey.Footnote ⁵ If the renderings of the Sanskrit offered above are correct – and supposing that the two Tibetan translators’ original Sanskrit texts were identical in their wording – it would appear that rṄog and Ye šes sde understood the pair of Sanskrit qualifiers in different ways.Footnote ⁶

The terms amuktajña and amuktajñāna are elucidated further by the related expression avinirmuktajñānaguṇa qualifying dharmakāya in the Anūnatvāpūrṇatvanirdeśaparivarta as quoted in Johnston's edition of the RGVV i.1. There the Sanskrit text reads (p. 3.4–5): yo ’yaṃ śāriputra tathāgatanirdiṣṭo dharmakāyaḥ so ’yam avinirbhāgadharmā, avinirmuktajñānaguṇo yad uta gaṅgānadīvālikāvyatikrāntais tathāgatadharmaiḥ. This sūtra quotation is found once more in RGVV i.44 (p. 39.7–8) in the form: tathāgatanirdiṣṭo dharmakāyo ’vinirbhāgadharmâvinirmuktajñānaguṇo yad uta gaṅgānadīvālikāvyativṛttais tathāgatadharmaiḥ. In Johnston's edition, the word avinirmuktajñānaguṇa qualifies dharmakāya in both citations. rṄog has translated by ma bral ba'i ye šes kyi yon tan can in both places (ed. Nakamura, 3 and 75). The meaning is “having the quality of [a buddha's] Gnosis not separate”.

Inseparability is thus exemplified in the Anūnatvāpūrṇatvanirdeśa by comparison with a light (āloka) and a precious gem (maṇi). It constitutes the sixth vajrapada of the RGV known as yog(ārth)a following RGVV i.1 and i.42–4; and in the latter passage, which develops an expanded and elaborate explanation of this topic, inseparability is illustrated by the examples (dṛṣṭānta) of the lamp ([pra]dīpa) and the precious gem. Briefly stated, just as a lamp is inseparable from its light and warmth, and just as a precious gemstone is inseparable from its brilliance and hue, so dharmakāya is inseparable from the tathāgata-qualities, surpassing in number the sands of the Ganga. The exposition in this manner of important points of doctrine, and in particular of the tathāgatagarbha doctrine, by means of dṛṣṭāntas, characterizes the RGV and its sūtra sources such as the Tathāgatagarbhasūtra. Footnote ⁷

The formulaic sequences containing the qualifiers (a)muktajñ(ān)a represent a way of expressing the buddha-nature doctrine according to which, in relation to tathāgatagarbha, the kleśas, and indeed all that is conditioned, are āgantuka = glo bur ba “adventitious”. And the excellent qualities are inseparable from it. Hence the interpretation in the context of these qualifiers, and also of avinirmuktajñānaguṇa, raises no major doctrinal difficulty for the buddha-nature teaching.

A question may, however, still be entertained as to whether the inversion of the negative particle between the two qualifiers in Ye šes sde's Tibetan version of the ŚMDSS compared with rṄog's version of this sūtra as quoted in the RGVV, together with the inconsistency between Ye šes sde's renderings of the related expression amuktajñ(ān)a in adjacent passages of the ŚMDSS (ed. Tsukinowa, 130–2 and 146), might possibly have had an influence on the subsequent history of the śūnyatā and tathāgatagarbha doctrines in Tibet and the continuing debates that took place there between gŽan stoṅ pas and Raṅ stoṅ pas (see below).

The problems of philology and translation posed by the qualifiers discussed above do not appear to have been directly caused by the variant reading ñon moṅs pa'i sbubs nas ṅes par grol ba (*vinirmuktakleśakośa) qualifying tathāgatagarbha found in the Tibetan textual tradition and considered in the first section of the present paper. But some of the crucial terminology – in particular avinirbhāga and acintyā buddhadharmāḥ – is found in the passage of the ŚMDSS as quoted in RGVV i.12 (p. 12.10–4). And Ye šes sde's bKa’ ’gyur version of this ŚMDSS passage (ed. Tsukinowa, 128–32) includes in addition the expressions (ma) grol bas šes pa and the variant definition of tathāgatagarbha in relation to dharmakāya examined in the first section of the present article. The sūtra's line of thought culminated in the distinctive idea of inclusion through non-duality expressed in the words (p. 130) de bžin gšegs pa'i sñiṅ po chos kyi sku ñon moṅs pa'i sbubs nas ṅes par grol ba “tathāgatagarbha-dharmakāya ‘free from the covering of defilements’”. The same sūtra passage also explains (pp. 130–2) what it calls the twofold tathāgatagarbhajñāna, the form described as ma grol bas šes pa defining the tathāgatagarbha Empty (śūnya) of all non-constitutive factors and the form described as grol ba šes pa defining the tathāgatagarbha non-Empty (aśūnya) of all integrally constitutive factors. This is the theme of the following section.

III. The expression tathāgatagarbhaśūnyatārthanaya and its Tibetan translation

The RGV commentary on verse i.154–5 (avataraṇikā) contains the words tathāgatagarbhaśūnyatārthanaya ucyate. In rṄog's rendering of the RGVV, the Tibetan equivalent is stoṅ pa ñid kyi tshul du brjod pa de bžin gšegs pa'i sñiṅ po; here Skt. artha is not translated. (This compound may be compared with the expression de bžin gšegs pa'i sñiṅ po chos kyi sku ñon moṅs pa'i sbubs nas ṅes par grol ba cited from the Tibetan translation of the ŚMDSS at the end of the preceding section of this paper, and where dharmakāya instead of śūnyatā figures beside tathāgatagarbha.) This complex Sanskrit compound expresses what would more usually be stated analytically in a separate clause or sentence. In form, it recalls the long compounds found in sūtras such as the Laṅkāvatāra, the sense of which can be fully determined only in terms of the doctrinal contexts in which they appear.

In commenting on the previous verse, the RGVV i.153 has stated that bodhisattvas newly set out in their Vehicle (navayānasamprasthita = theg pa la gsar du žugs pa) – confused as they are in their minds with respect to Emptiness (śūnyatāvikṣiptacitta = stoṅ pa ñid la/las sems rnam par g-yeṅs pa) – are as it were damaged (vipranaṣṭa = ñams pa) concerning the way in which tathāgatagarbha has the sense of śūnyatā. It would appear, then, that not only the modern reader but also early interpreters of the tathāgatagarbha doctrine in relation to śūnyatā have found themselves perplexed.Footnote ⁸

In the reading found in the Sanskrit text, this compound expression refers to the manner in which tathāgatagarbha has the “sense” or “import” (Skt. artha) of śūnyatā. This declaration goes back to the statement found in the ŚMDSS: “Precisely, knowledge of tathāgatagarbha is the Tathāgatas’ Gnosis of Emptiness” (p. 130: de bžin gšegs pa'i sñiṅ po[i] šes pa ñid ni de bžin gšegs pa rnams kyi stoṅ pa ñid kyi ye šes lags te). The corresponding Sanskrit is quoted in the RGVV (i.154–5, p. 76.15): tathāgatagarbhajñānam eva tathāgatānāṃ śūnyatājñānam (rendered by rṄog as: de bžin gšegs pa'i sñiṅ po'i yešes ñid ni de bžin gšegs pa rnams kyi stoṅ pa ñid kyi ye šes yin la, using the honorific ye šes instead of šes pa as the Tibetan equivalent for the first ^ojñāna).Footnote ⁹

Now, as seen in the preceding section of this paper, central to the teaching of the RGV(V) in this place is the question: of just what is tathāgatagarbha Empty (śūnya) and of just what is it non-Empty (aśūnya)?

The issue addressed will raise a fundamental question concerning the relationship of tathāgatagarbha – taught in sūtras belonging to the third and last cycle of the Dharma (chos kyi ’khor lo tha ma, which includes the various Tathāgatagarbhasūtras or sñiṅ po'i mdo) – to śūnyatā – which has been expounded in sūtras belonging to the second or middle cycle of the Dharma (chos kyi ’khor lo bar pa, such as the Prajñāpāramitā Sūtras) and commented on in the writings of the Madhyamaka school. A further question arises as to the precise sense in which tathāgatagarbha may be said to be Empty of other – i.e. non-intrinsic or non-constitutive and heterogeneous – factors, while being at the same time not Empty of what is constitutive of it. It is accordingly linked with the concept of Emptiness endowed with all excellent modes (sarvākāravaropetā śūnyatā = rnam pa thams cad kyi mchog daṅ ldan pa'i stoṅ pa ñid, in RGV i.88–92 where a comparison is made with artists collaborating with each other in executing a painting but with each specializing in one particular painterly activity; this is the citrakaradṛṣṭānta).Footnote ¹⁰

This matter engages, in another form, the question addressed above regarding the relation of tathāgatagarbha with dharmakāya and tathatā.

In Tibet, this crucial complex of issues was to lead to prolonged and extensive discussions and debates between gŽan stoṅ pas and Raṅ stoṅ pas, the latter holding that tathāgatagarbha is (svabhāva)śūnyatā “Emptiness (of self-existence)” – like all things without exception according to the Prajñāpāramitāsūtras and the Madhyamaka – whilst the former held that there is Emptiness only of heterogeneous, non-constitutive, factors (gžan stoṅ), given that the tathāgatagarbha is not Empty of its own nature (i.e. is not raṅ stoṅ). Put another way, the gŽan stoṅ pa advocates a view according to which ultimate reality is positively constituted and its description is accordingly stated in cataphatic or constructivist terms. But in the view of the Raṅ stoṅ pa, ultimate reality is negatively conceived, in terms of svabhāvaśūnyatā, and its description is therefore given in apophatic, apagogic and deconstructive non-substantialist terms.Footnote ¹¹ In the opinion of gŽan stoṅ pas, their opponents advocated a view of Emptiness that is only partial (ñi tshe ba = prādeśika) and relative (gcig la gcig med pa'i stoṅ pa ñid, gcig gis gcig stoṅ pa = itaretaraśūnya).Footnote ¹² Hence, according to gŽan stoṅ pas, their opponents wrongly held the purely deconstructive view of the Empty through negation (dgag stoṅ) and through destruction (chad stoṅ); and their raṅ stoṅ view will then amount to holding that the jar is empty of the jar (i.e. of itself, bum pa bum pas stoṅ pa; cf. rGyal tshab Dar ma rin chen (1364–1432), Theg pa chen po rgyud bla ma'i ṭīkā = Dar ṭīk, f. 164a–b).

The gŽan stoṅ pa's view of tathāgatagarbha in relation to śūnyatā, founded in great part on the portion of the RGV(V) just discussed, is accordingly distinct from that of the Raṅ stoṅ pa. The latter understands the tathāgatagarbha doctrine – expounded in the final cycle of the Buddhist Dharma containing the Tathāgatagarbhasūtra, etc. – to be “co-referential” (don gcig pa “one in sense”) with the doctrine of the Empty (śūnya[tā]) – expounded in the second or middle cycle comprising the Prajñāpāramitāsūtras. A raṅ stoṅ view of the matter has been explained by rGyal tshab Dar ma rin chen in his treatise on the RGV(V) (Dar ṭīk, ff. 166b–167a). It is briefly summed up there in the words: “Understanding the [Tathāgata-]Element ([tathāgata-]dhātu) very pure in nature (prakṛtipariśuddha, ^oviśuddha), which is taught in the last cycle, to be one in sense (ekārtha) with the Empty (śūnya), which is taught in the middle cycle, it will be known that the former [cycle] and the latter [cycle] are not in conflict (f. 168a: ’khor ba tha mar bstan pa'i khams raṅ bžin rnam dag de ñid/ ’khor lo bar par bšad pa'i stoṅ pa daṅ don gcig par rtogs nas sṅa phyi mi ’gal bar khoṅ du chud par ’gyur ro).

The hermeneutics of tathāgatagarbha in relation to the doctrine of śūnyatā adopted by some Raṅ stoṅ pas was finally to be developed through three arguments: (1) that the reference in RGV i.156–7 to the existence of motives (dgos pa = prayojana) for teaching the tathāgatagarbha does not necessarily imply that this teaching was “intentional” (dgoṅs pa can = ābhiprāyika) and hence of only provisional meaning (draṅ don = neyārtha); (2) that there is no real contradiction or incompatibility (’gal ba = virodha) between the two doctrines, which belong respectively to the third and the second dharma-cycles; and (3) that repetition in sūtras may be well-motivated and does not necessarily constitute redundancy (zlos pa = punarukta). This interpretation has been explained in some detail by Guṅ thaṅ dKon mchog bstan pa'i sgron me (1762–1823).Footnote ¹³

This matter might perhaps be summed up in the following manner. In Mahāyāna Buddhist thought, ultimate reality – i.e. what is termed (niṣparyāya-)paramārtha and what belongs to its level – is, as such, considered to be inexpressible conceptually and verbally (anabhilāpya, avācya). Therefore, in order to be able to advert at all to this level through conceptual thought and verbalization, certain instruments have to be brought to bear. These instruments would relate to the discursive philosophical concepts designated by the terms paramārthānukūla “concordant with the ultimate”, (sa)paryāya-paramārtha “modal ultimate” (namely the conceptually notional and the linguistically notational), and udbhāvanāsaṃvṛti “surface level of indication”. In the case of the deconstructivist/apagogic/apophatic vs. the constructivist/cataphatic descriptions of reality with which we have been concerned here, a conceptual tension and a verbal conflict arise between propositions such as those propounded in the two above-mentioned dharma-cycles. But, invoking certain widely accepted hermeneutical principles, Buddhist exegetes and hermeneuticians have pointed out that these contrasting descriptions do not in fact involve mutual contradiction.

Having recourse to further analysis, it might be reasoned that, for true contradiction to arise between them, descriptions must belong to one and the same universe of discourse. But the two descriptions of reality in question are not assignable to exactly the same universe of discourse. They can, therefore, be regarded not as contradictory or incompatible but, rather, as complementary, and also as incommensurable.Footnote ¹⁴ Following this line of analysis, there will exist no contradiction between the doctrines of śūnyatā and tathāgatagarbha. But since, belonging as they do to two distinct universes of discourse, they may still be said to represent discursively two different approaches to reality. And for this reason the one does not merely duplicate the other, so that, between them, there exists no redundancy. In other words, from the doctrinal and cognitive point of view of the illocutionary, the concepts and language employed to expound the tathāgatagarbha teaching are indeed distinct from those used in the Prajñāpāramitāsūtras. Furthermore, from the standpoint of pragmatics and the perlocutionary, this doctrine is clearly set off by the motives (prayojana = dgos pa) for teaching it that have been listed in RGV i.156–7.

The issues at stake and their implications were of the greatest significance for Mahāyāna philosophy, exegesis and hermeneutics. It cannot be concluded that the two contrasting interpretations in question above are attributable only to problems in translation and textual transmission: the roots of the difference lay in philosophical and meditative distinctions made by two important currents within Mahāyāna Buddhist thought.

The third example taken above from the RGVV – that of tathāgatagarbhaśūnyatārthanaya – illustrates how a translation of a crucial textual passage from Sanskrit into Tibetan may not always be entirely perspicuous and unambiguous, a situation that may have contributed to important divisions and debates within Buddhist traditions without, however, having actually been their sole or main cause. In the particular case of the expression under consideration here, rṄog lo tsā ba did not disambiguate the complex compound he was translating through his Tibetan rendering. His choice of translation indeed appears to have allowed space for what was to become a salient and much-discussed issue in the history of Buddhist thought in Tibet.

That the philosophical problem at issue may have already been of significance at the time of rṄog Blo ldan šes rab (1059–1109) is suggested by the fact that a gžan stoṅ line of interpretation is reported to go back to bTsan Kha bo che – a Tibetan contemporary in Kashmir of rṄog lo tsā ba and like the latter a pupil of Sajjana – and to gZu dGa’ ba'i rdo rje (Ānandavajra).Footnote ¹⁵

Concluding summary

The first example considered in this paper concerned a crucial definition of tathāgatagarbha in relation to dharmakāya which our sources have transmitted in two divergent formulations. This variation could result either from an event in the process of translation or from one in textual transmission. The negative particle, represented in Sanskrit saṃdhi by an avagraha after a final vowel at the end of the preceding word, might have been left unrepresented in a source since it is not always written in Sanskrit manuscripts. And the negative particle might then have been omitted in the process of translation into Tibetan and then restored in just one portion of the textual tradition on the basis of parallel definitions. Or the negative particle might have been translated but then omitted in the course of transmission in one portion of the Tibetan textual tradition but not in the other. However this may be, what makes this case interesting and very significant is the fact that these two divergent branches of the textual tradition can be cited in support of two distinct views of the tathāgatagarbha doctrine which are both actually represented in existing hermeneutical traditions.

The second example related to a very remarkable variation in the Tibetan renderings of an important set of Sanskrit terms between readings in the bKa’ ’gyur translation of the Śrīmālādevīsiṃhanādasūtra and corresponding passages from this sūtra as quoted in the Tibetan translation of the Ratnagotravibhāga commentary in the bsTan ’gyur. This set of terms was to form a basis for the hermeneutics of tathāgatagarbha free from all adventitious properties as śūnya “Empty”, and finally as śūnyatā “Emptiness”.

The third example considered also bore on this matter and concerned a complex compound, the original Sanskrit and the slightly divergent Tibetan translation of which are perhaps less than entirely plain in their full import. The compound is, nevertheless, explained in the Sanskrit commentary on the Ratnagotravibhāga, which sets out in some detail the way in which tathāgatagarbha is to be understood as śūnya(tā). This theme lies behind two distinct ways in which the Empty has been understood by Tibetan hermeneuticians.

If the textual issues pointed out in the second and third examples given above do not entirely account for the divergent interpretations mentioned, they contribute very appreciably to supporting them. As for the textual divergence pointed out in the first example, it supports two distinct ideas of the relation of tathāgatagarbha to dharmakāya known to our sources and to bear on the crucial question as to whether this doctrine is of definitive meaning (nītārtha) or of only provisional meaning (neyārtha), i.e. whether it is to be regarded as “intentional” (ābhiprāyika) in its purport.

Each of these three points thus raises major issues in Mahāyāna Buddhist thought in respect of the understanding of buddha-nature and non-duality (advaya, as distinct from monism, advaita).

In their different ways, the examples examined illustrate the fact that Tibetan translations do not in all cases provide standardized and entirely unambiguous renderings of every important term or phrase of the original Indian texts. It is for this reason, among others, that a completely accurate restoration of a no longer available Indian original is hardly possible on the basis of its Tibetan translation alone. But reconstruction of a text remains a possibility when there survive sufficient fragments of it, testimonia such as quotations from it in other texts, and closely related parallel texts. Such reconstruction may be compared with anastylosis in archaeology and conservation work. As for retrotranslation from Tibetan back into Sanskrit, it is feasible and can prove useful provided that it is kept in mind that it cannot be expected invariably to yield a text that is identical in its wording with a no-longer extant Sanskrit original. Given the very nature of translation, this fact is scarcely surprising. And it can be verified when, by a stroke of good fortune, an Indian original is recovered after such retrotranslation has been attempted, however carefully this may have been executed. Retrotranslation from Tibetan into Sanskrit may still prove helpful not only for a reader unfamiliar with Tibetan but also for scholars wishing to form as clear an idea as possible of how texts now accessible only in translation may have been expressed in Sanskrit.Footnote ¹⁶

In any event, the main Sanskrit technical vocabulary – a terminology that was registered in the Mahāvyutpatti and the sGra sbyor bam po gñis pa both compiled c. 800 – has for the most part been carefully rendered by Tibetan translators. Additional Tibetan–Sanskrit word-lists including technical terms can be compiled when original Indian texts survive along with their Tibetan translations. For a number of sūtras and śāstras, invaluable bilingual or multilingual indices serving this purpose are already in existence, an example that should be emulated for other still unindexed texts.