Fixing Problems in the Sanskrit Heart Sūtra

There's an old IT saying: "the good thing about standards is that there are so many to choose from." Standardisation does help to facilitate interactivity. These words are encoded in the English language, written in a script deriving from Roman writing. On my computer, they become encoded at one level as HTML rendered on your screen according to agreed protocols; at a lower level in terms of TCP/IP packets sent to your computer from a server; and at a lower level still as short bursts of voltage changes on a wire. If the parameters of these voltages, packets or markup languages were not agreed upon then the internet would cease to work.

In India, from about the beginning of the common-era, Sanskrit became a kind of standard for religious discourse. Even Buddhists began producing texts in Sanskrit, or Buddhist Hybrid Sanskrit, from around this time, despite the apparent prohibition on using Sanskrit contained in the early texts (Vin v.33.1). Some of the first Sanskrit texts were the early Prajñāpāramitā sūtras, especially the Aṣṭasāhasikā Prajñāpāramitā Sūtra and the Ratnaguṇasamcayagāthā. However, Buddhist Hybrid Sanskrit is distinct from Classical Sanskrit because it includes many Prakrit forms (to the point where some dialects are more like Prakrit than Sanskrit). In some out-of-the-way places other languages were important: Pāli became the "church language" in Sri Lanka; Gāndhārī was used in the Northwest Frontier and many Gāndhārī texts were translated into Chinese (especially the Āgamas or counterparts to the Pāli Nikāyas). Several Central Asian languages of the Iranian family (e.g., Tocharian and Khotanese) were also important scriptural languages. But most Mahāyāna texts were preserved in a variety of Sanskrit.

Another form of standardisation is the construction of critical editions from manuscript sources. The assumption is that a text that now exists in a variety of versions originated from a single written version, which is obviously not always true in a place like India that favours oral composition. An editor will gather all the existing editions of a text and try to determine a text, an ur-text, that is a plausible ancestor to them all. To do this they note scribal errors, any lines or phrases out of place, broken metre, etc., and try to fix them. Then, when obvious errors are fixed, they look for other ways in which texts evolve: for example, interpolations or other changes by previous editors. The resulting text may be different from any of the surviving manuscripts, as is the case for the Heart Sutra.

In the case of the Heart Sutra, we have known for some time that the core of the text derives from the Pañcaviṃśati Prajñāpāramitā Sūtra and Jan Nattier has shown that it specifically comes from the Chinese version by Kumārajīva (T 223). Further investigation tells us that the stemma codicum most closely resembles the Chinese version ascribed to Xuánzàng (T 251), though it is not a perfect match. T 251 is largely in the idiom of Kumārajīva with a few of Xuánzàng's terms over-laid. Though a version is attributed to Kumārajīva (T 250) who lived two centuries earlier, both attribution and date are plausibly disputed. Nattier argues that T 250 draws on T 25.1509 大智度論 Dàzhìdù lùn (Mahāprajñāpāramitāśastra), a commentary on Pañcaviṃśati attributed to Nāgārjuna and also translated by Kumārajīva, rather than directly from T 223, suggesting it has been edited by someone familiar with the work of Kumārajīva. T 250 also contains two passages, one of 37 characters, which do not occur in T 251. 

In the previous three essays we rehearsed Nattier's arguments that the Sanskrit Heart Sutra is a translation from the Chinese, focussing in the process on a number of infelicitous passages, the conclusion being that the Sanskrit text is, indeed, a translation from Chinese, produced by someone with Chinese as a mother tongue. If we were concerned to produce a better reading on our way to proposing a stemma codicum, some of these infelicities were easily fixed. In the case of the phrase, na cakṣuḥśrotraghrānajihvākāyamanāṃsi, we simply add the negative particle and a case ending to each word to arrive at idiomatic Sanskrit: na cakṣuḥ na śrotraṃ na ghrānaṃ jihvā na kāyo na manaḥ. Thus, the Gilgit ms. of Pañcaviṃśati and, as it happens, also quite a few of Conze's sources, e.g., Ne, Nh, Nk, Jb, Ce, and Cg (once again, Conze misses the opportunity). But some of the other problems run deeper. They would require us to first better understand the Chinese idiom and then make an informed decision about how to render that idea into Sanskrit.

This essay will look, in particular, at two phrases identified by Nattier as working well in Chinese, but becoming clumsy in Sanskrit and in English translations from Sanskrit.

Satyam amithyatvāt

The Chinese characters are 真實不虛 zhēn shí bù xū. Now, the characters 真 and 實 are used in the translation of yathābhūta-jñānadarśana (knowing and seeing things as they are), viz, 見如實、知如真, literally ‘seeing as real, knowing as true’. Where 真, zhēn, means 'real' and 實, shí, means 'true'. Hence, the sense is 'really true' which can be rendered as 'genuine' or 'authentic'. However, according the Digital Dictionary of Buddhism: 真實 has been used to translate a bewildering variety of Sanskrit terms:

akṛtrima, avitatha, avitathatā, aviparīta, ātmaka, ārjava, kalyāṇa, tattvârtha, tatva, tathātra, dravya, dharma-tattva, naya, niyata, nūnam, parama, paramârthatā, paramârtha-sat, paramârthena, pariniṣpatti, pariniṣpanna, pāramārthika, bhūtatā, thūti, maula, yathābhūta, yathāvat, *vāstavikatā, śuddhā, śubha, saṃsevana, sat, satya-kāra, satyatā, sad-bhāva, samyaktva, sāra, sāratā, sva-tantra, sva-naya, svanaya-pratyavasthāna

Choosing which of these was intended is difficult without more context. However, the second part of the phrase is more straight-forward and gives us a point of reference (note the contrast with the difficulty of this part, amithyatvād, in Sanskrit). 虛, xū, 'false', is also used for a variety of terms including śūnya; ākāśa; mṛṣā, mithyā abhūta, but these are all part of one broad semantic field concerned with lack of substance, either literally (śūnya 'empty') or metaphorically (mṛṣā 'false').

Though we find the Sanskrit satyam amithyatvāt unsatisfactory, there are a number of other possibilities that take in the contrast between truth and falsity. One of the main problems with  satyam amithyatvāt is that satyam is not usually contrasted with mithyā. Satyam is contrasted with asatya or, sometimes, with anṛta or mṛṣā. The Vajracchedikā Nāma Triśatikā Prajñāpāramitā or Diamond Sutra (section 14) contrasts satya/mṛṣā: na tatra satyaṁ na mṛṣā 'there is no truth and no lie'. However, this pair is not found in the Pañcaviṃśati and mṛṣā is only used once there, in the compound mṛṣāvādaḥ, 'false speech' (cf. Pāli musāvāda). Mithyā, on the other hand, is usually contrasted with samyañc. Of the various possibilities, samyañc/mithyā seems the more likely pair.

Let us begin with 虛, xū, and take it to convey the Sanskrit word mithyā, 'false'. Thus, 不虛 bù xū ought to be: na mithyā or amithyā, literally, 'not false, non-false', or 'true'. These negative forms are common and important in Buddhist Sanskrit vocabulary (and perhaps also in wider Indian literature). There is a special emphasis in saying that something is "not false" as compared to saying that it is "true". Right down to the present, Buddhists have an anxiety about taking the wrong path, or being given false teachings that do not lead to nirvāṇa. In this phrase, the non-falsity of prajñāpāramitā is almost as relevant as its truth.

The opposite of mithyā is usually samyañc (which becomes samyak/samyag in use, since ñc is not a permitted final). For example, we usually contrast samyagdṛṣti, 'right view', 正見, or 'perfect view' with mithyādṛṣṭi, 'false view', 邪見, xiéjiàn. And so on for all of the Eightfold Path. Though note that the character used here is 邪, xié, rather than 虛, xū.

In Pāli, we sometimes find other juxtapositions of samyañc and mithyā. At SN v.17-8 and DN iii.254 the abstract nouns micchatta/sammatta (Sanskrit mithyātva/samyaktva) are contrasted in terms of the items of the Eightfold Path. At DN i.8 we find that Gotama refrains from arguments of the type 'you are proceeding falsely and I am proceeding correctly' (micchā paṭipanno tvamasi, ahamasmi sammā paṭipanno). And at DN iii.128 a contrast is made between understanding the meaning and the words of any given doctrine, either of which can be micchā or sammā: e.g., ‘ayaṃ kho āyasmā atthañhi kho micchā gaṇhāti byañjanāni sammā ropetī’ti. (grasping a wrong meaning while having a right sense of the words).

From amongst the many possible translations of 真實 given by the DDB we see an abstract noun formed from samyañc, i.e., samyaktva 'completeness, wholeness; truthful', though this word is seldom used in Buddhist Sanskrit (searching across the whole of the Digital Sanskrit Buddhist canon). It's entirely possible for the same Chinese character to be used to translate both samyañc and samyaktva.

So we might have expected the contrast in the Sanskrit Heart Sutra to be along the lines: samyag na mithyā; or samyak ca amithyā ca; or samyagamithyā. Or, if the abstract was preferred, samyaktva na mithyātva etc. In the critical edition of the Aṣṭa by Vaidya we find this contrast between samyak and mithyā used as adjectives:

Saced evaṃ pariṇāmayati, samyak pariṇāmayati, na mithyā pariṇāmayati. Evaṃ ca bodhisattvena mahāsattvena pariṇāmayitavyam. (72) 

If he transforms this way, he transforms truthfully, he does not transform. And thus the bodhisattva mahāsattva should transform.

In a fragment of the Aṣṭa found in central Asia we find reference to a particular samādhi named 'devouring all truthhood and falsehood': samyaktva-mithyātva-sarva-saṃgrasanaḥ nāma samādhiḥ (AṣṭaK line 13). In another fragment (AṣṭaB) we find this explained as:

tatra katama samyaktvamithyātva-sarvasaṃgrasanaḥ nāma samādhir yatra samādhau sthitvā sarvasamādhīnāṃ saṃyuktvamithyātvaṃ na samanupaśyaty ayam ucyate samyaktvamithyātva-sarvasaṃgrasanaḥ samādhiḥ

There is the best of integrated states called "devouring all truthhood and falsehood", remaining in that state he does not perceive the truthhood and falsehood of all integrated states - this is called the integrated state of devouring all truthhood and falsehood.

The same idea occurs in the Pañcaviṃśati (Dutt 1.203)

tatra katamaḥ sarvasamyaktvamithyātvasaṃgraho nāma samādhiḥ yatra samādhau sthitvā samādhīnāṃ samyaktvamithyātvāni na samanupaśyati tenocyate sarvasamyaktvamithyātvasaṃgraho nāma samādhiḥ

Here the wording is almost identical, except that, in the name of the samādhi or integrated state, samyaktva-mithyātva-sarva-saṃgrasanaḥ 'all devouring' has been substituted with sarva-samyaktva-mithyātva-saṃgraho 'compendium of all truthhood and falsehood'. Kimura (1-1.184) has 'there is an integrated state called compendium of truthhood and falsehood' (asti samyaktva-mithyātva-saṃgraho nāma samādhiḥ) and then later (1-2. 65) sarva-samyaktva-mithyātva-saṃgraho as per Dutt, with the same explanation (1-2: 74). Dutt also has (1.143) 'there is an integrated state named compendium of truthhood and non-falsehood' (asti samyak-amithyātva-saṃgraho nāma samādhiḥ).

Elsewhere in the Pañcaviṃśati samyaktva tends to only be used in a compound with -niyato 'connected with, established in, or disciplined by'

bhagavān āha: na mayā subhūte 'nuttarāṃ samyaksaṃbodhim abhisaṃbudhya kathaṃcid api sattva upalabdhaḥ, samyaktvaniyato vā mithyātvaniyato vāniyato vā (Kimura 5:120)

The Bhagavan said, Subhuti, I don't perceive a being anywhere having attained supreme perfect awakening, connected with truthhood, or connected with falsehood, or unconnected. 

Thus we have a precedent in the Perfection of Wisdom literature for the contrast samyak na mithya and for samyaktva-mithyātva. The Heart Sutra is trying to convey that the efficacy of prajñāpāramitā is down to the features of both truthfulness and non-falseness. The ablative case ending indicates from what a verb proceeds, either spatially or more abstractly for what reason the action happens. Prajñāpāramitā is a great spell, etc., is the allayer of all disappointment because 真實不虛, i.e., because it is true/truthhood and because it is not-false/not-falsehood (it is difficult to find matching abstract nouns in English). We might combine the two factors into a dvandvā compound: samyaktvāmithyātvāt. 

Having done all this comparative/deductive work, if we now look again at Conze's critical edition we note that there were a few variant readings of this expression:

C^ae: samyaktvaṃ na mithyatvaṃ

N^e: samyaktva amithyātvā

N^b: samyaktvamithyatvat (not noted in Conze's edition)

Thus, the very readings (with some minor scribal errors) which would make sense in the context were, in fact, available to Conze in his mss., but he rejected them in favour of something which was not good Sanskrit and did not really make sense. Also, the lacuna in Conze's list of the alternate readings here is not the first I have found after examining the manuscripts.

Unfortunately, this undermines Nattiers argument that this passage is a back-translation. Other passages withstand scrutiny better, but here the simpler explanation is that we are mislead by Conze's critical edition. There was and is a better translation of this phrase.

Niṣṭhānirvaṇa

This term is more consistent in the mss. and our job here is not identifying a better reading from the extant mss. because there isn't one. The job here is to look more broadly at how Kumārajīva, in particular, might have used this phrase to translate Sanskrit. Since the passage this term appears in has not yet been identified with a counterpart in other Buddhist texts, we must cast a broader net. However, I think we can assume that the general style of Kumārajīva is likely to be a reference point, because where we have found exact correspondences to date they are to Kumārajīva's translations. We are fortunate to have Seishi KARASHIMA's detailed glossary of Kumārajīva's translation of the Saddharmapuṇḍarīka Sutra (T. 9.262), which shows us where and how each phrase was used and links it to a Sanskrit edition.

The Chinese is 究竟涅槃, jiùjìng nièpán. The last two characters render nirvāṇa while the first two mean 'finally attain' or 'ultimately'. The phrase is usually supplemented in both Sanskrit and Tibetan with the verb pra√āp, 'attains', as a past participle, prāpta. This choice is ironic because the text earlier says na prāptiḥ. Conze tolerates this:

"[Niṣṭhānirvāṇaprāthaḥ) obviously contradicts [na prāpti]. It is just because he seeks no attainment, it is just because attainment is quite impossible, that the Bodhisattva attains or wins Nirvana." (1975: 97-98), 

Conze seems to relish the contradictions sometimes found in Perfection of Wisdom texts, but I've already identified at least two example of how this predilection for nonsense has led Conze astray in editing the Sanskrit text. Generally speaking, when our text is nonsense, we have to ask if we have made a mistake. So we have to ask, is the contradiction part and parcel of the text or simply a mistake? What we want here is something that means 'culminating in nirvāṇa'. The bodhisattva, in a state of non-attaining, relies on perfect wisdom and has no mental obstructions (cittāvaraṇa), and thus they overcome wrong views and attain/reach nirvāṇa. So we can see the temptation to supply a verb like prāpnoti 'to attain' even though the text rules it out. We saw the verb ā√rādh 'to succeed' used in Pāli in the last essay.

Now the characters 究竟 are used to translate niṣṭhā, 'state, condition; conclusion, termination'; but they are also used to translate atyanta, 'ultimate, culmination; arrive, reach', and sometimes atyantaniṣṭhā (pointed out by Dan Luthaus on Buddha-L). It would seem that atyanta is a better choice, here. The terms atyantaśūnyatā, 'ultimate emptiness', and atyantaviśuddhitām, 'ultimate purity', are found quite frequently in Pañcaviṃśati. The compound atyantaniṣṭhā, however, still begs the addition of a verb or verbal form, so in this sense it does not solve our problem. 

Consulting Karashima's glossary we find some extra possibilities. Karashima has identified a number of uses for this Chinese phrase in translating the Sanskrit Saddharmapuṇḍarīka or Lotus Sūtra. But one in particular stands out.

為求聲聞者說應四諦法，度生老病死，究竟涅槃 (3c17)

Wèi qiú shēng wén zhě shuō yīng sìdì fǎ, dù shēnglǎobìngsǐ, jiùjìng nièpán

The parallel in Vaidya's Sanskrit Ed. is

yad uta śrāvakāṇāṃ caturāryasatya-saṃprayuktaṃ pratītyasamutpāda-pravṛttaṃ dharmaṃ deśayati sma jāti-jarāvyādhimaraṇaśoka-paridevaduḥkha-daurmanasyopāyāsānāṃ samatikramāya nirvāṇaparyavasānam | (12)

Here 究竟涅槃 corresponds to samatikramāya nirvāṇaparyavasānam, 'going beyond [suffering] to the conclusion of nirvāṇa'. Samatikrama (sam+ati+ √kram) means ‘going entirely over or beyond’; while paryavasāna (pari+ava+√so) means ‘end, conclusion’ or 'ending, concluding'. Kumārajīva also translates nirvāṇaparyavasāna (without samatikramāya) with 究竟涅槃 at 19c4, 50c4, 50c7. Additionally, he used these characters to translate: parinirvāṇa (7c2) and samavasaraṇa (12b5) which overlap semantically. 

Given the context in the Heart Sutra we're looking for a word or phrase that indicates that the bodhisattva's path culminates in nirvāṇa (which is not an attainment, but rather the extinction of the fires of greed, hatred and delusion). Contra Conze (1975), I see no reason to construct this as a paradox. That the goal is a liberation from something, rather than an attaining to something, is not so difficult to grasp. As a compound, nirvāṇaparyavasānam can mean exactly 'culminating in nirvāṇa', because paryavasāna is a verbal noun. As such, it is probably the best candidate for what was written as 究竟涅槃 from amongst the choices identified. The paraphrasing effect of going from Sanskrit to Chinese to Sanskrit might have produced the sequence:

nirvāṇa-paryavasānam → 究竟涅槃 → niṣṭhā-nirvāṇa

I suggest, then, that nirvāṇaparyavasānam is a better reading for 究竟涅槃 in the Heart Sutra than niṣṭhānirvāṇa, and were I editing the text would propose this substitution to create a readable text.

The assumption here is that the Chinese text was inspired by Sanskrit texts throughout. This is an assumption that requires further investigation, though I see preliminary evidence that even the parts not clearly associated with the Large Perfection of Wisdom Text drew on Chinese idioms of Kumārajīva's translations of Buddhist texts. In other words, the text has been composed to conform to Buddhist idioms, probably by somebody familiar with Kumārajīva's translations.

Conclusion

In this essay and the previous one, I have proposed two additional changes to the wording of the Sanskrit Prajñāpāramitāhṛdaya, to go with the revision of the first paragraph proposed last year, and the stylistic observations made by Jan Nattier also discussed in my last essay. The two latest suggestions are:

1. satyaṃ amithyatvāt  → samyaktvāmithyātvāt.
2. niṣṭhā-nirvāṇa(-praptaḥ) → nirvāṇaparyavasānam

The first is supported by extant mss. readings, though the second is not. In the second case some mss. attempt to solve the problem of the unreadability of niṣṭhānirvāṇa by adding the past participle prāpta, though this creates readable nonsense. The case for the second change, then, is based on readability and an attempt to establish alternatives by tracing how 究竟涅槃 was used to translate Sanskrit terms by Kumārajīva.

Jan Nattier argued that in both cases we have evidence for a back translation from Chinese. I have shown that in the first case this is incorrect, as it seems to be a problem with Conze's critical edition. However, the second does seem likely to an artefact of a phrase moving from Sanskrit to Chinese and back to Sanskrit.

On investigation, we find an accumulation of errors and infelicities in the critical edition of the Sanskrit Heart Sutra, along with a series of suggestions for how to improve the text. A new critical edition, and one which pays much close attention to alternate readings, is now more than desirable, it is urgent. In my next essay I'll propose a new edition of the Sanskrit Heart Sutra which incorporates the changes suggested so far.

~~oOo~~

Bibliography

• 御製大乘妙法蓮華經序 [Saddharmapuṇḍarīka Sūtra] CBETA Ed. Online: http://tripitaka.cbeta.org/T09n0262_001

• Bidyabinod, B. B. (1927) "Fragment of a Prajnaparamita Manuscript from Central Asia", Memoirs of the Archaeological Survey of India 32: 1-11. = AṣṭāB. Online: Gretil Archive: http://fiindolo.sub.uni-goettingen.de/gretil/1_sanskr/4_rellit/buddh/ppfrag_u.htm

• Conze, Edward. (1975) Buddhist Wisdom Books: The Diamond Sutra and the Heart Sutra. George Allen & Unwin. 

• Konow, Sten (1942) 'Central Asian fragments of the Ashṭādaśasāhasrikā Prajñāpāramitā and of an unidentified text'. Memoirs of the Archaeological Survey of India, 69. = AṣṭāK, unidentified = Praj(U1). Online: Gretil Archive. http://fiindolo.sub.uni-goettingen.de/gretil/1_sanskr/4_rellit/buddh/ppfrag_u.htm

• Seishi KARASHIMA [辛嶋靜志] (2013) A Glossary of Kumārajīva's Translation Of The Lotus Sutra [妙法蓮華經詞典]. Dharma Drum Buddhist College. Online: http://buddhistinformatics.ddbc.edu.tw/glossaries/files/kumarajiva.ddbc.pdf

• Vaidya, P.L., Ed. (1960) Saddharmapundarikasutra. (Buddhist Sanskrit Texts, 6) Darbhanga: The Mithila Institute. Online: http://fiindolo.sub.uni-goettingen.de/gretil/1_sanskr/4_rellit/buddh/bsu036_u.htm