A concordance of the content

We have briefly compared three translations, two in Chinese (T. 813, T. 814) and one in Tibetan (Q. 873), and found that T. 814 and Q. 873 are the closest of all the versions in terms of vocabulary, especially technical terms. However, because of the different grammar and styles of Chinese and Tibetan, there are still various differences in detail between these three versions. In examining the content of this text, it is clear that healing power derived from the dhāraṇī taught by the Buddha that is embedded in this text was important to the propagation of this text. This practical spiritual technology became meaningful to adherents of Chinese Buddhism as it spread throughout northern China. The text was translated into Chinese at least twice, which strongly suggests that it was highly valued in Chinese Buddhist communities. There was one fragment of this text belonging to Dharmarakṣa's translation (T. 813) discovered in the Cave library in Dunhuang, now labelled manuscript S. 4645, and preserved in the Stein collection in the British Library. Yet among the Dunhuang manuscripts, none of the fragments are identical to the section from Dharmamitra's version, presenting instead other sections of the text.

In Table 1 we list a more detailed comparison.

Table 1. A comparison of the Chinese (T. 813, T. 814) and Tibetan (Q. 873) translations

4 According to T. 813, this sūtra is called Hastyupamasūtra.

5 According to T. 813, the Buddha replied to Mañjuśrī's question of how the bodhisattvas simultaneously teach sentient beings in different languages.

Through a comparison of the three translations, we find that T. 814 and the Tibetan translation share the same content and context order, with the exception of the order of section 6 in the Tibetan text.

As we know, early Buddhist texts like the Āgama/Nikāya editions present various titles for the same sūtra (e.g. Norman Reference Norman1995: xxvii; Anālayo Reference Anālayo2007: 15). This sūtra also acquired a number of different titles during its transmission and translation. Everyone who dealt with this text – compilers, editors, translators or copiers – could choose a phrase from the context and make it the title, so long as he thought it represented the core idea of the sūtra.

On a newly published fragment in Sanskrit

The Sanskrit fragment of the Hastikakṣyasūtra discussed here is now officially numbered Or. 15009/672 in the British Library. Its photograph can be accessed in the International Dunhuang Project online database (idp.bl.uk). This fragment was originally part of the Hoernle Collection. The Japanese scholar Watanabe Kaigroku (1872–1933) worked on the Sanskrit manuscripts in the Hoernle Collection. In the archive labelled “Dr Watanabe's papers”, Watanabe numbered this fragment H. 150. vii. 18 and determined that it was from the Hastikakṣyasūtra (Sims-Williams Reference Sims-Williams, Karashima and Wille2006: 69, n. 38; Chen Reference Chen2012: 276). However, this fragment did not receive further study until Ursula Sims-Williams brought it to our attention.

With respect to material and format, the Sanskrit fragment is of a sheet written in Indian pustaka format, and on each side there are five lines extant. Both the left and right edges are damaged. The entire fragment is c. 16 × 8 cm with a single string hole remaining on the left side. The fragment is written in later Gupta script, which was commonly dated to around the fourth–sixth century ce.Footnote ⁶ Thus the Sanskrit manuscript postdates Dharmarakṣa's Chinese translation but appeared at almost the same time as Dharmamitra's Chinese translation. The latter, as noted above, was finished in 441. The Khotanese and Tibetan versions were created at a later date than the two Chinese versions and this Sanskrit fragment. As we will see later, the Sanskrit fragment has a section that corresponds closely to a section in Dharmamitra's translation, which suggests that the latter was based on a Sanskrit version similar to this fragment, if not from the same text itself.

We offer below our transliteration and reconstruction of this Sanskrit fragment.

The title and the metaphor

This section will shed new light on how the text was titled and how the two Chinese and one Tibetan recensions of the text were understood. The Chinese and Tibetan fragments all have titles, while these are no longer extant for the Khotanese fragment and perhaps the Sanskrit fragment. The Tibetan and Chinese Hastikakṣyasūtra fragments offer a striking number of variants in how the sūtra was titled and the types of phrases used for this process.

We will begin with the idea of the “armpit of an elephant”. The Chinese word, Xiangye 象腋, means literarily “The armpit of an elephant”. This title is at first attested in the Catalogue of Nie Daozhen (see below), and has appeared in both Tibetan and Chinese versions. In the Chinese translation by Dharmamitra that bears this phrase as its title, it explains why this text was called The Sūtra of Armpit of an Elephant as follows: “The Buddha told Mañjuśrī: ‘In the past, in this Gṛdhrakūṭa Mountain there were ten thousand Buddhas who have spoken about The Sūtra of Armpit of an Elephant ”. (佛告文殊師利, 過去於此闍崛山中, 有十千佛說象腋經, see T. 814: 782b20 ff.). However, throughout the text, there is no direct explanation as to why this text used this title, focusing on the armpit of an elephant. Since Dharmamitra's Chinese translation was very well known and popular in medieval Chinese Buddhism, the title bearing this phrase has been widely cited in Chinese Buddhist literature, such as the Foshuo foming jing 佛說佛名經 (T. 441, vol. 14, p. 231c) and Ru Lengqiejing 入愣伽經 (T. 671, vol. 16, p. 564b).

Moreover, what was probably the corresponding Sanskrit title can be determined from the opening of the Tibetan version and reconstructed in Sanskrit as the following: “In Indian language: ‘The Mahāyanasūtra named Hastikakṣya’” (rgya gar skad du | hastikakṣyanāmamahāyānasūtra | see (Q 873: 99b8). Kakṣyā Footnote ¹⁰ is derived from an archaic Vedic word, kakṣa, which has two original meanings; one is “armpit”.Footnote ¹¹Kakṣyā is equivalent merely to kákṣa- ¹, and within whose branch there has been a secondary meaning, “Anstrengung”, only since one of Hemancandra's (eleventh and twelfth centuries) lexigraphies, Anekārthasaṅgraha.Footnote ¹²

In the Sanskrit fragment, there is only one akṣara, cchaṃ, preserved. However, it is still possible to reconstruct the sūtra title from this single akṣara: Hastikaccha, if we consider its primitive form, hastikakṣa Footnote ¹³ that is, the Buddhist Hybrid Sanskrit form (BHSD, s.v. kaccha).Footnote ¹⁴ Several centuries later hastikaccha had been further Sanskritized not into -kakṣa, but -kakṣya. Thus the compiler could express the meaning of the title unambiguously, namely the meaning of kákṣa- ¹.

Now let us turn to the second phrase for the title, found in the Tibetan version, glaṅ bo rtsal, which means “the effort of an Elephant”, indicating its Sanskrit form *hastyākrama Footnote ¹⁵ or *hastivikrama Footnote ¹⁶. The title bearing this phrase is attested at the beginning of the Tibetan version, following its possibly original Sanskrit title: in Tibetan language, “The Mahāyānasūtra named rtsal of glaṅ po” (bod skad du | glaṅ po'i rtsal ces bya ba theg pa chen po'i mdo | see Q. 873: 99b8), however, glaṅ bo rtsal should not be regarded as a correct interpretation of the Sanskrit word hastikakṣya.Footnote ¹⁷ And this phrase also appeared in a Tibetan sentence: “bcom ldan 'das kyis 'jam dpal gźon nur gyur pa la 'di dkad ces bka' rtsal to | 'jam dpal saṅs rgyas khri rnams kyis bya rgod kyi phuṅ po'i ri 'di la glaṅ po'i rtsal lta bu źas bya ba'i chos kyi rnam graṅs bśad do |” (Q. 873: 102b2 ff.), which is parallel to the Chinese version (same as T. 814: 782b20 ff., as we have translated above). It is worth noting that glaṅ po'i rtsal lta bu can be reconstructed as *Hastikakṣyopama, if we regard rtsal as a direct translation of the Skt. word kakṣya and lta bu as a meaningful clue to a Skt. origin (see below).

In the context of Q. 873 and T. 814, the effort of an elephant was definitely emphasized. The Tibetan version has the following:

Dharmarakṣa's Chinese translation includes the following phrasing:

Likewise, all sentient beings who could understand this sūtra are also like this.

According to the sentence in these two versions, we can reconstruct the phrase, “effort of an elephant” into a Buddhist Sanskrit cliché, hastivikrama,Footnote ¹⁹ and be convinced of its existence in the Sanskrit exemplar, which the Tibetan translator had for his translation. Since there is no trace in the content related to “the armpit of an elephant”, possibly for the translator the meaning of the sūtra title was not comprehensible, but this phrase, “the effort of elephant” may more accurately reflect what the translator had in mind. Therefore he accepted this kind of interpretation to decipher the puzzling title.

The third phrase in the Chinese title, yuxiang “喻象” (the metaphor of elephant), is worth noting. It was used for translating the Sanskrit title *Hastyupama. Despite an abbreviation of *Hastikakṣyopama, it probably indicates that the power of this text is equivalent to the effort of an elephant (*hastyākramopama or *hastivikramopama) or the stride of an elephant (*hastigatyupama), which would have been understood as powerful and amazing. This phrase appears in the title of the text on the same occasion when the Buddha explained the title of this sūtra to Mañjuśrī in the early Chinese translation T. 813:

This phrase and its content remind us of an older sūtra title in the sūtra-piṭaka, viz. in Āgama/Nikāya, Hatthipadopamasutta (象跡喻經, Skt.: Hastipadopamasūtra), “The simile of an elephant footprint”. Under this title there are two Pāli sūtras: MN. 27 Cūḷahatthipadopamasutta and MN. 28 Mahāhatthipadopamasutta, which correspond to two identical Chinese translations T. 26. 146 and T. 26. 30 象跡喻經 respectively. Although the former mentioned the analogy of a big elephant's footprint, it seems more plausible to link our sūtra to the latter one, in which the analogy of the elephant's footprint is mentioned and Śāriputra appears as the preacher of the sūtra. Furthermore, the main idea of this sūtra, the contemplation of the four elements and the dependent origination of the five aggregates, could indicate the relationship between it and our text. It is noteworthy that in the Buddhist tradition the more recently composed sūtra would often introduce an older sūtra title within the framework of an embedded legend, which claims that the Buddha had preached the older one from the Āgama/Nikāya in ancient times. In this way the authors enhanced the authority of the later text, by stating that it was intrinsically linked to the earlier one, or even the repetition of the earlier one, whose authority had been acknowledged through all kinds of Buddhist traditions.Footnote ²⁰

The fourth phrase used in one of the titles of this text is Xiangbu “象步” (Stride of elephant), which refers to the Sanskrit term *hastigati or *hastivikrama. In Dharmarakṣa's version, it says that:

Even though elephant and nāga are listed together, neither the effort nor the strides of nāga were used in the title of this text. Furthermore, in Dharmarakṣa's version, the stride of a lion was also mentioned, as follows:

Another sentence in Dharmamitra's version shows the same metaphor:

Interestingly the metaphor of a lion is also attested in the Tibetan version:

At the end of the sūtra, the Buddha told a previous Buddha who also preached this sūtra that he had given different names in each version – all are related to the stride of a lion: the first is Leshizibu 樂師子步 (Skt. *siṃhagatyadhimukti) “being fond of the stride of the lion” (T. 813: 781a12); the second is Shiziyoubu 師子遊步 (Skt. *siṃhagati) “the walking stride of the lion” (T. 814: 786c18); and the third is a Tibetan phrase seṅ ge'i stabs su 'gro ba (Skt. *siṃ havikr āntagati) “Going in the manner of a lion” (Q. 873: 114a6). These names for the previous Buddha show that “the stride of lion” was essential for demonstrating the power of a Buddha. The stride of lion also appeared in another text entitled Rāṣṭrapāla translated by Dharmarakṣa (Boucher Reference Boucher2008: 98–9).

Since vikrama (viz. Tib. rtsal) means both “effort” and “stride”, the Tibetan title glaṅ bo rtsal can also be interpreted as “Stride of an elephant”. However, it must be added that there is another variant that does not make reference to an elephant. The title of Dharmarakṣa's Chinese translation (T. 813), entitled Wuxiwang 無希望 “Without desire” (Skt. *akāṅkṣa), has no demonstrable connection with the previous four phrases (titles?). Since the sūtra belongs to the Perfection of Wisdom literature, “without desire” indicates a state of mind that the absolute truth is transcendent, that all phenomena are unobtainable, and there is nothing to be desired. In the context of Dharmarakṣa's translation, however, this word xiwang 希望 is only attested in the phrase yu zhongyou wu xiwang 於眾祐無希望, “having no desire for the blessing from the Buddha”, which appears three times.Footnote ²¹ This phrase is concerned with one of the elements, in particular the penultimate in the causal chain, which leads to the core idea of this sūtra, anutpattikadharmakṣānti “intellectual receptivity to the truth that states of existence have no origination”, which the Buddha explained to MañjuśrīFootnote ²² and then in more detail to Śāriputra (T. 813: 777c18–778c27; T. 814: 784a4–c24; Q. 873: 107a2–110a2). Both in Dharmamitra's Chinese version (T. 814) and in the Tibetan version (Q. 873) this phrase can be understood as “having no ripening of the fruit of donation”, reconstructed as *dānāvipāka and should be recognized as the original expression in their Sanskrit exemplar. The translation of this phrase in the Tibetan version (Q. 813) is much more literary; however, it could be derived from a Sanskrit compound, *vipākāpratikāṃkṣaṇa, or the Tibetan phrase re ba med in the Tibetan sentence rnam par smin la re ba med.

Despite these connections there is still not sufficient evidence to explain why Dharmarakṣa's Chinese version (T. 813) used this title. Here we suggest a new hypothesis that the second part of the best-known title, Hastikakṣya, kakṣya must have been mistaken for kāmkṣa “desire”, and the first part, hasti, for some form of negation, like a, nāsti, etc.

Given that the exemplar Dharmarakṣa had for his translation could have been in Gāndhārī written in Kharoṣṭhī script, or in Buddhist (Hybrid) Sanskrit written in Kharoṣṭhī or Brāhmī script,Footnote ²³ there are two possibilities that arise from misreading or miscopying. The first is *nāstikāṃkṣa (Skt.) and *nastikakṣa (Gāndhārī) “Nonexistence of desire” for hastikakṣa in both languages.Footnote ²⁴ If the exemplar was written in Brāhmī, ha Footnote ²⁵ could be confused with nā through the incorrect placing of its final stroke, ka with kāṃ through forgetting a point for anusvāra, however, ha and nā in our fragment, which is written in the later Gupta script and unlikely to have been read by Dharmarakṣa, can be more easily exchangeable.Footnote ²⁶ If the manuscript was written in Kharoṣṭhī, all the akṣaras of both words are the same except for ha and na, which are hard to misread. Another possibility is the Gāndhārī *asakakṣa (Skt.: *asakāṃkṣa) instead of hastikakṣa, since a and ha are confusable,Footnote ²⁷ and sti is confusable with sa due to the elision of two strokes for -ti.Footnote ²⁸ In the third century, Buddhist texts preserved in Gāndhārī and in Kharoṣṭhī script were gradually transposed into BHS and Brāhmī script,Footnote ²⁹ and miscopying and misreading must have been inevitable.Footnote ³⁰

To summarize with respect to titles, we have the contemporaneous appearance of three titles: Yuxiang (*Hastyupma) and Wuxiwang (*Nāstikāṅkṣa, *Akāṅkṣa or *Asakāṅkṣa) in T. 813 and, according to the Catalogue, a Xiangbu (*Hastigati or *Hastivikrama). Then after more than a century, we have a Xiangye (Hastikakṣa or Hastikaccha in its BHS form, Hastikakṣya in its developed Skt. form) in T. 814, with its Skt. correspondence, Hastikaccha in the Skt. fragment and Hastikakṣya in the Tibetan translation. Finally, the word Hastikakṣya is transmitted into Glaṅ-po'i rtsal and Glaṅ po'i rtsal lta bu (*Hastikakṣyopama) in Tibetan.

If we rely on the work of other scholars who have theorized that in Central Asia and north-west India Buddhist texts were transmitted from “a Middle Indic”, i.e. Gāndhārī, some of which are derived from central India and in a Mi. other than Gāndhārī,Footnote ³¹ in the third century alongside Gāndhārī, the texts were also preserved in BHSFootnote ³² until their complete Sanskritization in the fifth century,Footnote ³³ then we can consider that with the exception of the title Wuxiwang, all the others can possibly be affiliated with a Sanskritization process from Middle Indic via Gāndhārī or BHS.Footnote ³⁴ When we accept the hypothesis that our Mahāyāna sūtra had borrowed the authority of Hatthipadopamasutta from Āgama/Nikāya during its compilation, the process must start with Yuxiang. There was probably another title in Mi. *Hatthigaccupama (Skt. *Hastigatyupama, for Skt. sta > Mi. ttha, see Pischel Reference Pischel1900: §307, von Hinüber Reference Von Hinüber2001: §229; for Skt. ty > Mi. cc, see Pischel Reference Pischel1900: §280, von Hinüber Reference Von Hinüber2001: §247, Oberlies Reference Oberlies2001: §16.1), whose abbreviations are Yuxiang (*Hastyupama) and Xiangbu (*Hastigati). This Mi. title could be written as *Hatthigacchupama (von Hinüber Reference Von Hinüber2001: §192) too, and as a compound it could have been incorrectly deciphered as *Hatthigaccha-upama and transmitted into Gāndhārī *Hastikakṣa[-uvamu]Footnote ³⁵ viz. BHS *Hastikaccha[-upama] (for Skt. k > Mi. g, see Pischel Reference Pischel1900: §202, Brough Reference Brough1961: 42 and §31, von Hinüber Reference Von Hinüber2001: §174–6; for Skt. kṣ > Mi. ccha, see Pischel Reference Pischel1900: §317–22, von Hinüber Reference Von Hinüber2001: §232–5, Oberlies Reference Oberlies2001: §18.2), then further Sanskritized into *Hastikakṣya[-upama], again Xiangye (Tib. mchan khuṅ) and the convincing reading in the Skt. fragment – Hastikaccha. Finally *Hastivikrama, which literally corresponds to Tib. Glaṅ-po'i rtsal, could be an invention of the translator based on the language context.

The last question we need to address is why the two contemporaneous translators, Dharmarakṣa and Nie Daozhen, read two titles of the same sūtra yet no single title overlapped with the other? The first possibility is that there was a Mi. title attested in the colophon of that manuscript, which is hypothesized in the paragraph above, *Hatthigacc(h)a and has two Skt. correspondences, *Hastigaty[upama] and *Hastikakṣa. Nie Daozhen absorbed both Skt. correspondences into his catalogue, while Dharmarakṣa was convinced by the possibly wrongly copied *Nāstikāṅkṣa or *Akāṅkṣa in the text and replaced the one in the colophon. The second possibility is that they got information on the title from different sources, namely from two or more manuscripts.

Consequently, Hastyupamasūtra can be regarded as an abbreviated form of not only the title of the canonical sūtra, Hastipadopamasūtra, but also the manifold appellations discussed above. Nevertheless, the sūtra circulated with the best-known name, Hastikakṣya.

Concluding remarks

We can draw some basic conclusions to sum up our discussions. First, of the five versions of this sūtra in four languages, the earliest Chinese version, by Dharmarakṣa, and the Sanskrit version share the same or a very similar transmission source, an earlier source text, while Dharmamitra's Chinese translation, the Tibetan and the Khotanese versions shared a later source text. The first group of texts existed no later than the third century according to the dating of T. 813, but may have disappeared no earlier than the fifth century due to the dating of the Sanskrit fragment. The second variant appeared no later than the fourth century according to the dating of T. 814, but should have existed until at least the eighth century due to the dating of the Khotanese fragment. We find that the form of the second text was more stable and had a much longer circulation period, while the first was quite short.

Second, the metaphor of the power of elephant was the focus of this sūtra, though the titles indicated this through different phraseology. The metaphor of the elephant symbolized both the power of this text and the power to be gained by sentient beings who could understand and spread this text.

Third, interestingly, the Hastikakṣyasūtra employed a long gāthā for retelling the story of how the Buddha taught this sūtra and how this sūtra could benefit all sentient beings. It shows a typical early form of a Mahāyāna Buddhist text.

Abbreviations

BHS

= Buddhist Hybrid Sanskrit

BHSD

= Franklin Edgerton, Buddhist Hybrid Sanskrit Grammar and Dictionary, vol. 2: Dictionary. New Haven: Yale University Press, 1953.

CPD

= A Critical Pāli Dictionary, begun by V. Trenckner, ed. D. Andersen et al., vol. I, 1924–48, vol. II (fasc. 1 ff.). Copenhagen: Royal Danish Academy of Sciences and Letters, 1960.

EWAia

= Manfred Mayrhofer, Etymologisches Wörterbuch des Altindoarischen. Heidelberg: Carl Winter, 1986–2001.

Mi.

= Middle Indic.

MN

= Majjhimanikāya. V. Trenckner and R. Chalmers (ed.), The Majjhima-Nikāya. London: Pali Text Society, 1888–99.

Mvy

= Mahāvyutpatti, R. Sakaki, 2 vols. Tokyo: Shingonshū Kyōto Daigaku, 1925–36.

'Phaṅ-thaṅ ma

= Bod ljoṅs rten rdzas bśams mdzod khaṅ (ed.), Bod ljoṅs rdzas bśams mdzod khaṅ gi rtsa che'i dpe rñiṅ gces bsgrigs dpe tshogs las. Beijing: Mi rigs dpe skrun khaṅ, 2003, pp. 1–67.

PTSD

= The Pali Text Society's Pali–English Dictionary, ed. T. W. Rhys Davids and W. Stede. London: Pali Text Society, 1921–25.

PW

= Otto Böhtlingk and Rudolph Roth, Sanskrit-Wörterbuch, 7 Bde. Saint Petersburg: Buchdruckerei der Kaiserlichen Akademie der Wissenschaften, 1855–75.

Q.

= (Qianlong) Peking bsTan 'gyur: The Tibetan Tripiṭaka. Peking Edition, ed. Daisetz T. Suzuki 鈴木大拙. Kyoto: Otani University, 1955–61.

T.

=Taishō Shinshū Daizōkyō, ed. Takakusu Junjiro 高楠順次郎 and Watanabe Kaigyoku 渡邊海旭. Tokyo: Taishō Issaikyō Kankokai, 1924–32.

T. 813

= Foshuo wuxiwang jing (Sūtra Spoken by the Buddha on No Desire), trans. Zhu Fahu 竺法護 (Dharmarakṣa) in the third century.

T. 814

= Xiangye jing (Sūtra on the Armpits of Elephant), trans. Tanmomiduo 曇摩蜜多 (Dharmamitra) in 441.

ZHDCD

= Zanghai Dacidian/Bod-rgya tshig-mdzod chen-mo, ed. Zhang Yisun 张怡荪. Beijing: Mi rigs dpe skrun khaṅ, 1984.