Chung-Hwa Buddhist Journal第13.2期 (p157-193)： (民國89年)，臺北：中華佛學研究所，http://www.chibs.edu.tw
Chung-Hwa Buddhist Journal, No. 13, (2000)
Taipei: The Chung-Hwa Institute of Buddhist Studies
Paul M. Harrison
Associate Professor, University of Canterbury
While it is clear that the bodhisattva ideal lies at the very heart of Mahāyāna Buddhist doctrine and practice, in many respects its historical development remains obscure. In an attempt to shed some light on this, the following paper examines the notion of the “celestial bodhisattva.” Although this notion enjoys a wide currency in contemporary Buddhist scholarship, it is appropriate to ask whether it is at all useful, or indeed meaningful, and whether it corresponds to any indigenous Buddhist category. Focusing on the first Mahāyāna sūtras translated into Chinese by the Indo-Scythian Lokakṣema in the late 2nd century C.E., the paper explores the details of their portrayal of Mañjuśrī, who may be regarded as a paradigmatic case of a so-called celestial bodhisattva. It turns out that in these texts Mañjuśrī plays a very important part, while Avalokiteśvara is a comparative non-entity. Using Mañjuśrī as a test case, the paper concludes that the concept of the “celestial bodhisattva” is not a useful one, and has no clear indigenous
referent. It also offers some general hypotheses about the early history of the bodhisattva ideal, and about the cult of the great bodhisattvas, which appears on the basis of the evidence reviewed here to have been a later and secondary development.
關鍵詞：1.Mañjuśrī 2.Celestial Bodhisattva 3.Mahāyāna 4.Lokakṣema
Even Buddhist Studies, once characterized by Richard Gombrich as an intellectual backwater, occasionally has its surface ruffled by the winds of academic fashion. Indeed, looking at recent issues of the Journal of the International Association of Buddhist Studies──in particular Vol. 18, No. 2, 1995──or Donald Lopez’s Curators of the Buddha (1995), we can see that the storm of post-modernism has well and truly broken over it, lashing the established notions and the master narratives of the past with its rhetorical blasts and exposing their structural inadequacies to the elements. Of course this drama of deconstruction──to switch metaphors midstream──has been staged whenever a new generation of scholars has succeeded the generation who trained it, with the usual predictable Oedipal turns of plot, but nowadays the cast tends to be tricked out in much fancier theoretical costume, and the chorus is French rather than Greek. Thus Buddhist scholars, like their counterparts in other fields, have been called upon to rethink their perverse attachment to textual sources, their mistaken belief in objectivity, their compromised status as outsiders (or as insiders, when occasion arises), their suspect political motivations, and sundry other factors that supposedly render their enterprise less than straightforward. Even those unrepentant traditionalists like myself who go on doing what they have always done, do so with more than sideways glance at this feast of self-reflexivity, and they too are caught up in the constant revision of the field and its categories. After all, it does sometimes happen that the winds of change──to revert to the original metaphor──bring improvement, especially when they blow in less theoretical and more substantive directions.
One aspect of Buddhist Studies of which this is demonstrably true is the history──especially the early history──of Mahāyāna Buddhism. The well-entrenched theories of the great masters like
Akira Hirakawa and Étienne Lamotte are now being subjected to a searching and extensive critique by contemporary Buddhist scholars, and although one can hardly speak of a new consensus taking shape, it is at least clear that certain features of the old picture are unlikely to form part of the new one. Among these features we might cite the ideas that Mahāyāna Buddhism was lay-centred, that it was predominantly devotional in orientation, that it sprang from a revolt against monastic privilege and self-absorption, that it was organizationally distinct from the Mainstream Nikāyas, that it was one single movement, and that right from the start in India it carried all before it. Recent studies by Schopen, Gombrich, Ray, Silk, Sasaki, Williams, myself and others have started to build up a picture which is more nuanced and pluralistic, more historically cautious, and more inclined to emphasize the monastic or──better still──renunciant side of things. The origins of Mahāyāna Buddhism are thus no longer what they used to be.
Part of this process involves rethinking the bodhisattva paradigm. Whatever else Mahāyāna Buddhism is──or, to be more careful, whatever the various different movements which coalesced into the phenomenon which we now call Mahāyāna Buddhism were──most would still agree that its defining characteristic is the promotion of the ideal of the bodhisattva, i.e., the belief that full Buddhahood or supreme and perfect awakening is the proper goal of human endeavour. The bodhisattva ideal is such a central feature of Mahāyāna Buddhism that it is now taken for granted. However, it too may merit critical re-examination, to see whether we are still entertaining ideas about it which have passed their use-by date.
The bodhisattva ideal is a big topic, so I propose in this paper to confine myself to one aspect of the conventional wisdom, the notion
of the “celestial bodhisattva.” The extent to which this idea has become part of the Buddhist Studies canon can be gauged from the article on “Celestial Buddhas and Bodhisattvas” in the Encyclopaedia of Religion edited by Mircea Eliade. In this lengthy and informative piece David Snellgrove uses the term “celestial bodhisattva,” and others like “divinity,” “celestial being,” “god” and “goddess,” without ever once problematizing them. It is to be noted that he also emphasises the bodhisattva’s function as saviour figure, offering compassionate assistance to others. In approaching the bodhi-sattva like this──as a heavenly saviour──Snellgrove is of course simply continuing a long tradition in Western scholarship, which, whenever it began, certainly received its first authoritative articulation in the work of Har Dayal (1932), whose ideas have since been echoed by such writers as Ling (1976), Basham (1981), Robinson and Johnson (1982) and others.
To a certain extent the two terms──“heavenly” and “saviour” ──go together, and imply each other. That is to say, to characterise bodhisattvas as celestial or godlike is to emphasise their otherness and stress their saviourlike role as objects of cult and sources of assistance from on high. It cannot be denied that the cult of bodhisattvas was eventually an important part of Mahāyāna Buddhism, wherever it was practised, and that the use of the
language of divinity has a certain functional aptness, but the fact remains that such language also obscures other aspects of the bodhisattva ideal and, when not handled carefully, leads to a considerable misreading of the religion. What is more, it is ahistorical, and arguably imposes the later situation on the earlier: what the bodhisattva ideal became is taken to be its original impulse. I have already pointed out (Harrison 1987) what I consider to be a more accurate reading of the bodhisattva ideal, at least in the beginning. Building on that earlier contribution, I would like in this paper to take another look at the figure of Mañjuśrī──who, along with Avalokiteśvara, is a celestial bodhisattva if ever there was one──using only the evidence of the earliest Chinese translations of Mahāyāna sūtras made by Lokakṣema. This evidence will form the basis for certain general hypotheses about early historical developments in this area.
As is usually the case, others have traversed this ground before. Mañjuśrī has already been the subject of an extended article by Étienne Lamotte in the journal T’oung Pao (1960). Lamotte contrasts the rather late attestation of this figure in Buddhist iconography with his early appearance in a host of Mahāyāna
sūtras translated into Chinese──not the only time that a mismatch between the archaeological and literary records has been noted──and then proceeds to furnish us with a plethora of scriptural references and historical accounts──not all of them consistent──relating to Mañjuśrī and his cult, mostly culled from Chinese sources. This wealth of detail is effectively summarized in Birnbaum 1986. A more tightly focussed study is provided by Akira Hirakawa in an article in the Madras Journal of Asian Studies (1983). At the end of a useful (but not exhaustive) review of the lore relating to Mañjuśrī in the early Chinese translations, Hirakawa concludes that these texts indicate the importance of this bodhisattva in the rise of Mahāyāna Buddhism, and hypothesizes that descriptions of him may be based on meditative experiences. Now, one might well think that Lamotte and Hirakawa ought between them to have said the last word on the subject. However, it may still prove worthwhile to take another look at Lokakṣema’s early Chinese translations, and ask ourselves what part Mañjuśrī plays in them.
Mañjuśrī appears in 6 of the 9 translations which we can regard as genuine surviving products of Lokakṣema and his school. These are (in the order in which they will be dealt with in this paper) the Aṣṭasāhasrikā-prajñā-pāramitā-sūtra (AsPP; T.224), the Wenshushili wen pusa shu jing《文殊師利問菩薩署經》(WWP; T.458), the Druma-kinnara-rāja-paripṛcchā-sūtra (DKP; T.624), the
Lokānuvartanā-sūtra (LAn; T.807), the Dousha jing《兜沙經》(DSJ; T.280), and the Ajātaśatru-kaukṛtya-vinodanā-sūtra (AjKV; T.626). Mañjuśrī does not figure in Lokakṣema’s translations of the Pratyutpanna - buddha - saṃmukhāvasthita - samādhi - sūtra (PraS; T.418), the Akṣobhya-tathāgatasya-vyūha (AkTV; T.313), or the Kāsyapa-parivarta (KP; T.350). However, he does appear in one of the two other texts which we have good reason to suppose were translated by Lokakṣema, even though his versions of them have not survived. These two are the Śūraṃgama-samādhi-sūtra (Śgs) and the text known from the single Chinese translation entitled Chengju guangming dingyi jing《成具光明定意經》(CGD; T.630), attributed to Zhi Yao 支曜, a contemporary of Lokakṣema. Mañjuśrī features in the former but not the latter. This means, then, that 7 out of the 11 texts translated or known to have been translated by Lokakṣema make mention of Mañjuśrī
Mañjuśrī’s appearances in these texts range from walk-on parts in some to starring roles in others. To the former category belongs Lokakṣema’s translation of the Aṣṭasāhasrikā-prajñā-pāramitā-sūtra (AsPP), the Daoxing banruo jing《道行般若經》(T.224), the sole version of this work to mention Mañjuśrī as present during the delivery of the discourse (425c6). Here Mañjuśrī is, as it were, plugged into the “frame story” (Rahmengeschichte) of the text, in a way which is not matched in any other version of the AsPP, including the Sanskrit text as we now have it. The AsPP is in fact one of those Mahāyāna sūtras which propounds Mahāyāna teachings in a carefully crafted Mainstream setting, putting the new
dispensation in the mouths of the great representatives of the old. Whether this scriptural camouflage reflects an earlier textual stratum is a moot point (see below). In any case, the insertion of Mañjuśrī’s name into the nidāna of this archaic version has all the appearance of an interpolation made when the text passed through the hands of those who regarded him as an important figure.
Despite the promising title of the Wenshushili wen pusa shu jing (WWP), or the Sūtra of Mañjuśrī’s Questions Concerning the Bodhi-sattva Career, Mañjuśrī is referred to only twice, and plays no active part. Lokakṣema’s translation (T.458) is the only version in existence, and is bedevilled by obscurities, but it is clear enough that it introduces our bodhisattva as a paradigmatic practitioner of the Mahāyāna. Right at the start of the text Śāriputra asks for permission to question the Buddha, to which Śākyamuni replies: “Well done, Śāriputra, well done! You should ask your question. If you have only heard the name of the dharma of the causes and conditions of the tathāgata-career from Mañjuśrī and have not yet fully obtained the thing itself, I shall now expound it to you. Listen carefully, listen carefully” (435b5-7). The text here is not entirely unambiguous, but seems to allude to a previous incident in which Śāriputra is supposed to have heard the title of the following sūtra──here dasājie-shu yinyuan fa ming (怛薩阿竭署因緣法名) ──from Mañjuśrī without having heard the text itself. In any case Mañjuśrī is clearly a possible source of such a teaching. Much later in the text the bhikṣu Rāṣṭrapāla asks the Buddha whether there is anybody in the assembly who is actually practising the Tathāgata-career which is being expounded, and is told that the bodhisattva Mañjuśrī is doing so (437b10-12), after which the exposition is taken up again. Thus, although he is only mentioned briefly twice, Mañjuśrī is implicitly in the assembly, to be held up as an
exemplary practitioner of the Tathāgata-career (i.e., the bodhisattva-caryā), and as a font of teachings on it.
In the Druma-kinnara-rāja-paripṛcchā-sūtra (DKP) the Buddha is on the Gṛdhrakūṭa with a large assembly of bhikṣus and 72,000 bodhisattvas assembled from other worlds. Lokakṣema’s version (T.624) lists 49 by name, the last being Mañjuśrī (349b7). But he plays no further part in the sūtra, although later in the text, during an exchange between the kinnara king Druma and King Ajātaśatru on the subject of the merit generated by bodhisattvas, Druma says “You, sir, have obtained two kalyāṇa-mitras. The first is the Buddha. The second is Mañjuśrī. Through their grace the doubt arising from the unrighteous acts you committed was completely dispelled” (364b12-14). This is a clear allusion to the contents of another scripture, the AjKV (on which see below), and is a rare example of explicit intertextuality in a Mahāyāna sūtra (cf. Harrison 1992: xvi).
Mañjuśrī makes a somewhat more significant appearance in the Lokānuvartanā-sūtra (LAn), in which he plays the part of the Buddha’s interlocutor. The only named member of the assembly of bhikṣus and bodhisattvas, it is he who asks the Buddha to explain the supramundane or transcendental (lokottara) nature of the Awakened Ones. In Lokakṣema’s version (T.807) he is referred to at the beginning and at the end (751b and 753c) simply as the bodhisattva Mañjuśrī (Wenshushili pusa 文殊師利菩薩), in the Tibetan translation in v. 2 as mkhas pa ’Jam dbyangs, i.e., Mañjughoṣa the Wise, in v. 4 as thugs skyes sras gces ’dzin mchog, i.e., (the
Buddha’s) beloved (spiritual) son (Skt. aurasa), and in v. 113 again as ’Jam pa’i dbyangs or Mañjughoṣa, the leader of the Buddha’s sons (sras po). The rest of this short verse sūtra (only 113 stanzas in the Tibetan translation) consists in the Buddha’s reply, so here too, as in the AsPP, Mañjuśrī plays his part only in the frame story. What is intriguing, however, is that the parallel verses in the Mahāvastu are spoken not by the Buddha but form part of an extended eulogy delivered by the Venerable (āyuṣmant) Vāgīsa (see Jones 1949: 129, n. 5).The same figure, this time spelled Vāṇgīśa, re-appears later (ibid., 222-224) to recall a past life in which he was the disciple of the future Śākyamuni. Now, although it is perfectly possible that this is the same person as the Thera Vaṇgīsa, reputed author of numerous verses of praise and renowned for his gift for inspired eloquence or pratibhāna (see Malalasekera 1937, s.v.), what is curious is that Vāgīśvara (Lord of Speech) is also one of the names of Mañjuśrī. Can this be a mere coincidence?
In the Śgs Mañjuśrī is one of the great bodhisattvas listed at the beginning of the text (Lamotte 1965: 119),although the role of chief interlocutor is played by Dṛḍhamati. Later he is given as an example of a bodhisattva who received the prediction to full awakening after having conceived the aspiration to it (ibid., 214), comes to the fore as a seasoned traveller through other worlds (224-226), defines for Dṛḍhamati the expressions puṇya-kṣetra (231-235) and bahu-śruta (236-238), and recalls his own previous fictitious attainment of nirvāṇa many times over as a pratyekabuddha (241-245), his feats in this regard being explained by his mastery of the śūraṃgama-samādhi. In the same way his former appearance as the Buddha Nāgavaṃśāgra is recalled (260-264). Such manifestations, Mañjuśrī points out, are an easy matter for those who understand the true nature of reality as he does. He is thus to the fore as an
exemplary practitioner of salvific magic and teacher of the bodhisattva path.
In the Doushajing (DSJ; T.280), a short text later incorporated in the massive Avataṃsaka-sūtra, Mañjuśrī plays the leading role in a grand cosmic epiphany staged by the Buddha for a huge host of bodhisattvas visiting him directly after his awakening. As the Buddha illuminates distant buddha-kṣetras with his light, their bodhisattvas come flocking to him with their retinues. The first to arrive is Mañjuśrī, from the world lying to the east named *Hiraṇyavarṇa, which belongs to the Buddha *Acalajñāna (445b16-19). When all the bodhisattvas are assembled it is Mañjuśrī who actually gives the sermon which is the centre-piece of the text, expounding the inconceivable variety of appearances and names assumed by the Buddhas in general and by Śākyamuni in particular in response to the different capacities of the sentient beings to be saved (445c26-446a16). Here in the DSJ Mañjuśrī is located in a specific Buddha-field, lying far to the east of Sahā. In fact, the traditions assembled by Lamotte are by no means consistent in this regard, although the east comes up most frequently (see Lamotte 1960: 18-31). But in any case the DSJ is somewhat more specific than the DKP, which has Mañjuśrī among the many bodhisattvas assembled from other buddha-kṣetras in the ten directions, but gives no details. On the other hand the AsPP, the WWP, the Śgs
and the LAn do not make it clear at all that Mañjuśrī is a visitor from another world.
Finally, the most striking text in our small group of sūtras is undoubtedly the Ajātaśatru - kaukṛtya - vinodanā - sūtra (AjKV). Mañjuśrī’s importance and superior understanding of the Dharma is established right at the beginning of this text, when, in a separate location on the Gṛdhrakūṭa from where the Buddha is holding court, he outdoes his 29 companions in a kind of competitive discussion of omniscient cognition and its preconditions. His powers as a teacher are further illustrated by his conjuring up of a phantom Buddha to reveal the nature of phenomena to his audience. Joining the Buddha and his disciples, he continues to teach, with the Buddha’s approval. Then comes a long narrative sequence which is little more than a panegyric to Mañjuśrī, and has in fact has survived as a separate sūtra. Having accused the Buddha of ingratitude if he does not share his almsfood with him──at first blush a shocking breach of etiquette──Mañjuśrī then outdoes the great śrāvakas in a display of magical power, after which he has his primacy vis-a-vis not only them but also the Buddha himself established by Śākyamuni on the basis of an avadāna in which a bhikṣu called jñānarāja inducts a boy by the name of Vimalabāhu into the Mahāyāna. The concluding passage is worth quoting in full here (for convenience the Tibetan version is cited, although the Chinese is quite similar):
Śāriputra, if you should be doubtful, puzzled or uncertain as to the identity of the monk and preacher of Dharma called jñānarāja on that occasion and at that time, then, Śāriputra, you should not be that way. Why is that? Because on that occasion and at that time Mañjuśrī here was the monk and preacher of Dharma called jñānarāja. Śāriputra, if you should be doubtful, puzzled or uncertain
as to the identity of the merchant’s son called Vimalabāhu on that occasion and at that time, then, Śāriputra, you should not be that way. Why is that? Because on that occasion and at that time I was the merchant’s son called Vimalabāhu. Śāriputra, Prince Mañjuśrī caused me to conceive the aspiration for awakening after giving me the almsfood, which was my first aspiration to awakening, and that is the way, Śāriputra, in which this is to be known. One should see that the greatness of a Buddha, the ten powers, the (four types of) assurance, the unhindered cognition and anything else belonging to the Realized One have all come from the instigation of Prince Mañjuśrī. Why is that? Because omniscience has been attained on the basis of that moment of aspiration. Śāriputra, I see in the ten directions innumerable and incalculable Realized Ones who have been established in awakening by Prince Mañjuśrī and who are called Śākyamuni, just like me, as well as those who are called Tiṣya, Puṣya, Śikhin and Dīamkara, and I could go on reciting for an aeon or more than an aeon the names of all those Realized Ones who, after being established in awakening by Prince Mañjuśrī, are now turning the Wheel of the Dharma, and still not come to the end of them──to say nothing of those who are pursuing the course of a bodhisattva, residing in the Tuṣita Realm, taking rebirth, going forth from the household life, practising austerities, or sitting on the Terrace of Awakening! That is the way, Śāriputra, in which this is to be known, that it is about Prince Mañjuśrī himself that they speak and teach who rightly say and teach that he is the mother of the bodhisattvas, their father, the one who shows compassion to them, and their instigator. And that, Śāriputra, is the reason and the cause, why, on account of a former favour, Prince Mañjuśrī charged me with ingratitude.
The insistence that Mañjuśrī takes ritual precedence here in the matter of the almsfood is of course symbolic of the fact that the spiritual achievements of the Buddha, on which the attainments of
the śrāvaka or arhat depend, are themselves premised on the existence of the bodhisattva. Throughout the rest of the AjKV as well, Mañjuśrī plays a leading role as a preacher of the message, being appointed by the Buddha as the only person able to relieve the spiritual anguish of King Ajātaśatru, as a result, indeed, of a karmic connection through countless lifetimes (see 404a-b). In helping Ajātaśatru come to terms with the evil he has wrought, Mañjuśrī has ample opportunity to deliver many teachings on Mahāyāna themes. But as well as that, he uses his magical powers to summon bodhisattvas from another buddha-kṣetra to whom he then gives lengthy discourses on dhāraṇīs, on the bodhisattva-piṭaka, and on the avaivartika-cakra (the wheel that does not turn back). He also performs great miracles of transformation when proceeding triumphally to Ajātaśatru’s palace. Thus not only is he an important character in the AjKV, he is at stage centre throughout most of the sūtra, his status as a virtual Buddha being established explicitly (see 405a). Thus the AjKV is really Mañjuśrī’s sūtra, a fact which is reflected by the title of Dharmarakṣa’s version of it (T.627).
We can see from this brief review that Mañjuśrī looms very large in the Lokakṣema corpus, slightly more in fact than Lamotte suggests, since he omits the DSJ from his discussion, and deals with the AsPP and the Śgs at other points in his paper. But Lamotte’s observation that Mañjuśrī’s name is to be found frequently in the earliest Mahāyāna texts available to us, despite his comparatively late appearance in artistic remains, is amply confirmed by our review, as indeed it was by Hirakawa’s earlier study. Of course there are problems with using a number of texts
to build up a single composite picture of Mañjuśrī’s personality, since these texts may have been composed at different times and in different places, and may therefore reflect a variety of traditions rather than just one. But this methodological caveat hardly affects our conclusion: that the texts translated by Lokakṣema reflect the emergence of Mañjuśrī as an important archetypal bodhisattva figure by the middle of the second century C.E., be it in one milieu or in many.
The same cannot be said──at least on the basis of these materials──for Avalokiteśvara. The translation of the Sukhāvatī-vyūha attributed to Lokakṣema (T.361) and cited by some as the first historical evidence for the cult of Avalokiteśvara is not a genuine work of his. However, it may well be that the version commonly attributed to Zhi Qian (支謙), T.362, could have been done by Lokakṣema, even though certain sections of it are unlikely to have come from his hand. This leaves us with a single clear citation: the name Guanyin (觀音) appears in the list of bodhisattvas
at the beginning of Zhi Yao’s (支曜) translation of the Chengju guangming dingyi jing (CGD), made in 185 C.E. However, Avalokiteśvara plays no further part in that text. And while his name is listed at the beginning of Kumārajīva’s version (T.625) of the DKP, as bodhisattva No. 37 (368a15), and in the Tibetan version (as No. 11, see Harrison 1992: 10), it is not given, as far as I can see, in Lokakṣema’s earlier translation. Furthermore, its appearance at different points in the list in the two later versions is decidedly suspicious. Avalokiteśvara does not appear in any of the other texts we have reviewed, not even in the PraS, where we might have expected him to be mentioned, since this text provides explicit evidence of the cult of Amitābha. On this basis we could say that Avalokiteśvara was virtually unknown to the people who produced the Mahāyāna sūtras translated by Lokakṣema, or was not regarded as important by them. One can only speculate as to the reasons for this. It is possible, for instance, that stories about Avalokiteśvara arose in a different geographical area from that in which the texts under review were put together.
Let us return now to Mañjuśrī. Having established his importance, what can we say of his status as a “celestial bodhisattva”? Is the concept of the celestial bodhisattva actually attested in our sources, or is it an analytical category devised by Western scholarship? To answer this question we must first of all ask if there is any indigenous Sanskrit term corresponding to this expression. According to some scholars such a term does indeed exist. It is the word mahāsattva, usually translated as “great
being” (see, e.g., Robinson and Johnson 1982: 78; Nakamura 1986: 267; Harvey 1990: 124). However, this is a mistake. It is abundantly clear in many passages in our texts that mahāsattva is simply an alternative designation or stock epithet for bodhisattva, as Edgerton maintained (see BHSD, s.v.), and thus while it is certainly applied to highly advanced beings like Mañjuśrī, it is also applied to ordinary people pursuing the bodhisattva path who may be a very long way indeed from reaching their goal, or even from reaching the much vaunted stage of non-regression. In this respect the evidence of later versions of the texts with their fuller wording is unequivocal, but we must look for proof in Lokakṣema’s translations themselves. Here we find that there is often no Chinese equivalent where the Sanskrit has mahāsattva or the Tibetan reads sems dpa’ chen po, a probable sign that the semantic weight of the term was light enough for Lokakṣema, working in his usual elliptical style, to omit it. In other places, however, we find the expression “bodhisattva and mahāsattva” (Chin. pusa mohesa 菩薩摩訶薩) used quite frequently, in contexts which make it clear that being a mahāsattva is of no great moment. One may cite, for instance, PraS §3C (905a23-24), which mentions bodhisattvas and mahāsattvas who do not possess the abhijñās. The AkTV, which uses the term with unusual frequency, awards the title even to someone who has only just conceived the aspiration to awakening (752a29-b6), alludes to Vinaya regulations in describing lay bodhisattvas and mahāsattvas of Abhirati who may not hear the Dharma directly from the Buddha, unlike their renunciant counterparts (758c2-9), and urges bodhisattvas and mahāsattvas in this world who are renunciants (pravrajita) to proclaim the teaching to householder practitioners or to secure copies of it from them at
all costs (763c24-764a10). In the AsPP the term is occasionally used interchangeably with kula-putra and kula-duhitṛ (e.g., 444b25-c8), and even applied to people whose teaching, copying or reception of the Perfection of Insight is deficient in various respects, due to the influence of Māra (446c22-447a5). Further examples could no doubt be hunted down, but the point is sufficiently established. As for the meaning of the term mahāsattva itself, it is hard to improve on the extensive discussion by Kajiyama (1982), who inter alia draws attention to its interpretation as a bahuvrīhi (in line with a similar construe of the term bodhisattva): the mahāsattva, then, is not necessarily a “great being” at all.
Having seen this candidate for the position thus disposed of, one might contend with greater force that the 10th-stage bodhisattva or the bodhisattva who has one more rebirth to go (eka-jāti-pratibaddha) may legitimately be regarded as “celestial,” especially if they remain in a heavenly realm like Tuṣita and merely send emanations to do their work below, as those on the verge of Buddhahood are commonly held to do. Indeed, the AkTV makes explicit mention of the link between the attainment of eka-jāti-pratibaddha status (yi sheng bu chu 一生補處) and rebirth in Tuṣita (754c14ff.; cf. also 763c). In a similar vein the DKP reinterprets the practice of devatānusmṛti in a Mahāyānist way in Section 4I (353c9-10): “Their minds think always of heaven, and then enter
[the state of] one rebirth remaining (yi sheng bu chu). This is then a jewel.” The implication seems to be that focussing on the deities helps one to become a bodhisattva among them. In the AsPP where the Sanskrit text refers to eka-jāti-pratibaddha (Vaidya’s ed., p. 215), Lokakṣema (465c21) has the term aweiyan (阿惟顏), which corresponds to the Sanskrit abhiṣeka. In the AjKV eka-jāti-pratibaddha appears several times with specific reference to Maitreya at 393a18-21, and is once again translated yi sheng bu chu. In the Śgs too the connection between eka-jāti-pratibaddha, the tenth stage and abhiṣeka is established, with Maitreya again presented as the paradigm case (Lamotte 1965: 258). The link between abhiṣeka and the tenth bhūmi is also mentioned elsewhere in the text (166, 265); but it is not entirely clear, as Lamotte claims (280), that the tenth bhūmi is called tathāgata-bhūmi (see p. 137). Although the evidential value of the Śgs is reduced for us by the lapse of time between Lokakṣema’s lost translation of 186 C.E. and Kumārajīva’s extant version made sometime in the first decade of the 5th century, we know from the AsPP passage cited above that Lokakṣema was familiar with the concept of abhiṣeka. The term eka-jāti-pratibaddha is also attested in the DSJ, where it is applied to the audience of bodhisattvas foregathering from other Buddha-fields at the beginning of the text (zhu pusa si (or ci) yi sheng bu chu 諸菩薩賜一生補處, 445a14), and the ten stages are mentioned too
(pusa shi daodi 菩薩十道地 at 445a28, cf. also 445b1). On the basis of these references we can say that this set of concepts had been developed by the middle of the 2nd century, although the precise extent to which the daśa-bhūmi scheme had evolved by this time is not clear. However, these concepts seem rather more precise and limited in scope than the term “celestial bodhisattva,” and what is more, none of them is explicitly applied to Mañjuśrī. But even were they so applied, I think it is fair to say that the notion of the celestial bodhisattva as a distinct type is not strongly supported by the use of these terms, in our texts at least. That Buddhists believed in the existence of great bodhisattvas is another matter, and so is their obvious belief in a continūm or differential scale of spiritual attainments. But such beliefs reflect purely quantitative distinctions, of degree rather than kind, and not a qualitative distinction between two discrete categories of bodhisattva, the mundane and the celestial, between which a clear line can be drawn.
This is, of course, to approach the problem on the level of terminology. There are other ways of coming at it. We have established Mañjuśrī’s presence in these early texts, but can we
speak of a cult of Mañjuśrī in them? After all, the celestial bodhisattvas are commonly held to have been developed as objects of cult. But there is no evidence for it here. Should one ask what evidence of cult would look like, the answer would surely be that it would consist in explicit injunctions to worship Mañjuśrī or his image, to bring him to mind or engage in visualization practice directed towards him, or to call on his name for help and assistance, as, for example, in the Sad-dharma-puṇḍarīka (SP), one is instructed to call on the name of Avalokiteśvara. In fact Lamotte deals with a number of texts in which this kind of practice with respect to Mañjuśrī is recommended, one of the earliest of which appears to have been the so-called Mañjuśrī parinirvāṇa-sūtra (T.463, Wenshushili banniepan jing《文殊師利般涅槃經》), first translated into Chinese by Nie Daozhen towards the end of the 3rd century (Lamotte 1960: 32-39; see esp. p.38). However, there are no instructions of the type sketched above in the Lokakṣema corpus. There is, on the other hand, plenty of such evidence for the cult of the Buddha and of various Buddhas of the present (especially in the PraS and the AkTV, but also in the AsPP ), for the cult of the stupa and, most notably, the cult of the book as investigated by Schopen, and there are scattered but intriguing allusions to other aspects of Mahāyāna Buddhist cult practice, but nowhere is the cult of bodhisattvas recommended. Nor is there any mention of these figures being available to the inhabitants of this world or ready to provide them with protection, with the exception of the bodhisattva Maitreya, who as the future Buddha of this world clearly falls into a special category (see esp. AsPP, AjKV, DKP). 
Indeed, since he is held to be resident in the Tuṣita heaven awaiting his eventual promotion he might indeed merit the title of celestial bodhisattva, as we saw above. Other bodhisattvas like Mañjuśrī are in fact residents of other buddha-kṣetras. While it is a terminological quibble to say that these Buddha-fields are not strictly speaking “heavens” in terms of Buddhist cosmology, what is really important for our purposes is whether these bodhisattvas too are available as objects of supplication and sources of help and inspiration, in the same way that Maitreya is. Although in other, later texts that is the case, it is not so in these earlier translations, where the emphasis is decidedly on the bodhisattva as saving subject, whatever his or her rank. Although there are always problems with an argument from silence, our contention is supported by the fact that a number of Lokakṣema’s translations do advise believers that general protection will be forthcoming from various supernatural agencies (i.e., devas, etc.), but the salvific
intervention of bodhisattvas is not promised at such points (where it would be natural to expect it). The silence in this regard is deafening, especially when one considers that these works are far from backward in promising supernatural protection and worldly benefits of all kinds to their devotees. However, in later versions of the same texts the situation changes. A good example is provided by §§14E-14H of the PraS, where the Sui Chinese (T.416) and Tibetan versions promise the help and assistance of bodhisattvas to those bodhisattvas and mahāsattvas (sic!) who take up the teaching, while Lokakṣema’s translation and the Sanskrit text do not (see Harrison 1990: 116-117, 280-282, 300). A similar case is found in AsPP 431b24-26 (cf. Vaidya ed., pp. 26-27). These instances are prima facie evidence of the influence of the bodhisattva cult in its developed form on later recensions of the sūtras.
With respect to Mañjuśrī, then, and as far as these sources are concerned, the appellation of “celestial bodhisattva” is warranted neither terminologically nor functionally, and what applies to him──at least on the level of function──applies a fortiori to the other bodhisattvas who make an appearance, with the exception of Maitreya. That said, another question presents itself. If Mañjuśrī and the others are not functioning in these early translations as cult objects, then what are they doing? How are we to explain their role in the texts, or indeed account for their existence in the first place? To come up with an answer I think we have to take a
different approach, one which avoids the euhemeristic dead-end rightly rejected by Lamotte (1960: 8-9) but also skirts the banal recourse to Śūnyavāda rhetoric which leads us nowhere either (cf. Lamotte 1960: 9-10, 96). The tentative solution which I propose involves considering the situation from the point of view of the people who produced the Mahāyāna sūtras, and approaching the great bodhisattvas who appear in these texts as literary rather than cultic creations.
When one starts to think about it, one sees that from the outset the writers of Mahāyāna sūtras were put in a difficult position by their Mainstream predecessors. In attempting to redefine the goal of Buddhist practice and legitimate it scripturally their dilemma was as follows. Mainstream Buddhist canonical literature is not just a matter of dry doctrinal exposition, but uses narrative to convey meaning. The truths of Mainstream Buddhism, in all the canonical literature which has come down to us in the Nikāyas or Āagamas and in the Vinayas in their various translations, are illustrated and reinforced with copious references to historical persons, or persons whom we are led to believe were historical. Given the standard personality ideal of Mainstream Buddhism (arhatship and nirvāṇa), an ideal which does not involve personal survival in the normal sense, this kind of historical anchoring produced no problems for the authors and transmitters of Mainstream Buddhist scripture. But in the process of elaborating their traditions they used up the available stock of personalities, nearly all of whom were held to have attained some grade of awakening or liberation during or soon after the Buddha’s own lifetime. In fact the sheer pressure of piety would have necessitated this outcome: the Buddha was naturally such a powerful and effective teacher that he brought virtually everyone he was in contact with, except the irremediably stupid and wicked, to liberation, not only his ordained disciples──out of 500 leading bhikṣus only Āananda hadn’t attained perfection by the time
Gautama died──but also most of his lay followers. As a result of this exemplary hagiographical thoroughness on the part of their forerunners, the Mahāyānists found themselves in a difficult position, since by the logic of their new teaching they could not claim that anybody who had obtained even stream-entry had been in reality a bodhisattva. None of these figures could be cited as role models. And yet the inclusivistic approach typical of Indian religion would necessitate that the old picture (and thus the old literature) must somehow be harmonised or squared with the new, not simply erased and painted over. At the same time, the use of the pre-awakened Gautama as a model bodhisattva would probably have been of limited efficacy, because the Jātakas were shared with Mainstream Buddhists as a whole, and might give an undesirable impression of rarity and difficulty, as well as being unsuitable as vehicles of doctrinal innovation. To glorify and exemplify the new ideal something rather different was required.
There were, as far as I can see, three possible solutions to the impasse. The first──and arguably the earliest──was to subvert the historical record and use arhats as proponents of the new teaching, thus śrāvakas with appropriate leanings were pressed into service, the most significant cases being Mahākāśyapa and Subhūti. We find this approach employed in several of Lokakṣema’s texts, most notably the AsPP and the KP. But this method has obvious limitations: no matter how eloquently they may expound the new dispensation, these well-known arhats can hardly embody it, since they are famed for the successful consummation of the spiritual orientation which the followers of the Mahāyāna would condemn as the inferior way, the Hīnayāna.
The second solution is to hold up as bodhisattvas real persons whose attainments were either unknown or not widely known. The
householder bodhisattva Ugra perhaps falls into this second category. He is the chief protagonist of the Ugra-pariprcchā-sūtra (UgP), also translated into Chinese in the late 2nd century. An Ugra or Ugga appears in the Pāli canon, where we find two householders of this name, Ugga of Hatthigāma and Ugga of Vesālī (see Malalasekera 1937: s.v.), about whom the traditions seem inevitably to have become interwoven and confused. Whether the householder bodhisattva is the same as one or both of these characters is a moot point, since the UgP is set in Śrāvastī, and both Uggas are held in any case to have attained the state of an anāgāmin. It is to be hoped that Jan Nattier’s forthcoming study of this text will clarify the matter. Another less problematic case is the parricide king Ajātaśatru, to whom Theravādin canonical literature, at least, imputes no spiritual attainment. There is thus no traditional impediment to the prediction in the AjKV of his eventual attainment of Buddhahood. His character is saved for the writers of that text by the fact that his celebrated crimes precluded any previous rise to sanctity. If this is not scraping the bottom of the historical barrel, then what is? But in any case, there is no doubt that the supply of figures like this with no known record of spiritual attainment was extremely limited. And if historical personages had failed to achieve anything under the direct influence of the Lord Buddha there were no doubt good reasons for it, which would hardly make them credible representatives of the Mahāyāna.
The third solution was simply to make it up, to invent fictitious, non-historical figures and work them into the pseudo-historical framework, which, along with a host of other devices, was designed to impart the traditional look to the new texts, to dress them in scriptural camouflage. And this eventually became the preferred solution, with new characters cut out of whole cloth, ranging from the rather more mundane types like the householder bodhisattvas Bhadrapala and his friends (PraS) to the more magnificent figures like Mañjuśrī, who has been the subject of this paper. But despite
the staggering array of names, one can hardly speak of differences of personality where these bodhisattvas are concerned, any more than one can in the case of Buddhas. They are little more than cardboard figures, cut from the same template. Where did this template come from? This is perhaps an unanswerable question, but I would suggest that one need not look outside the Buddhist tradition to find the answers. The models are there in the Mainstream canons, in the Āagamas, the Vinayas and the Jātakas, which frequently celebrate the pedagogical skills, the magical powers and the self-sacrificial zeal of the Buddha and his disciples. A change of name here, an adjustment in status there, and what was, say, a story about an arhat becomes the tale of a bodhisattva, to which a slightly different moral could be attached. That there may have been external influences from such diverse sources as Iranian religion, Greek sculpture, Indian theism and so on, as has been postulated, cannot be denied, but it cannot be proved either. Indeed, given the direction in which our evidence points, such influences may well have been more literary than cultic. That is to say, Mahāyāna sūtras may owe more to the literary conventions of the Purāṇas than they do to the ritual cult and iconography of the Hindu gods (which seems to me to invoke euhemerism once removed). But all such debts may have been incurred later in the piece.
To opt for an internal “organic” model of development largely in terms of the Buddhist tradition itself is, of course, to state a methodological preference rather than a historical fact, since the model is little more than a hypothesis in any case. Although there are some indications of a gradual introduction of bodhisattva figures into the texts (which I take as supporting the internal model), we are hampered by what I believe is the relatively late transmission of Mahāyāna sūtras to China, and the loss of evidence relating to earlier stages of development. That is to say, by the time Lokakṣema was at work, the Mahāyāna was in full swing, and
far from being in all cases its first literary products, many of his translations represent its middle period. We can never be entirely certain about what is early and what is late, which makes it harder to be sure about the original impulses behind the appearance of the great bodhisattvas. However, the important thing to see here is that in this material Mañjuśrī and the other bodhisattva figures function as part of a web of what we might call literary strategies of legitimisation. But the challenge of achieving an authentic traditional look was also to be balanced by the inspirational agenda of the literature, an agenda which eventually came to dominate, as increasing success no doubt led to greater boldness. One notes therefore that the Lokakṣema corpus appears to reflect a variety of approaches and thus, perhaps, strata in the historical development of Mahāyāna literature, from the relatively sober promulgation of the new teachings by the old guard in the AsPP, the KP, and the WWP to the full-blown magical cosmological son et lumiere of the DSJ, the DKP and the Ajkv. These last two texts are especially noteworthy for their complex and sophisticated structure and their handling of philosophical issues. But we should guard too against over-interpreting the appearance of a bodhisattva in a text, especially an early one. We quite naturally read such occurrences in the light of
the later tradition, in which the term bodhisattva summons up an image which is iconographically and mythologically loaded, to say the least. Were we to translate the relevant passages a little differently, so that they read “at that point a person aspiring courageously to awakening got up from his seat . . .” then they might not appear quite so different from Mainstream sūtras.
The fact remains that even in those Lokakṣema texts (arguably the later ones) in which Mañjuśrī and other bodhisattvas take the leading roles, the cult of the great bodhisattvas──if we can dispense now with calling them celestial──is completely unattested. It is therefore quite likely that this cult represents an even later development, built on top of the aforementioned literary strategies and not underlying them or preceding them. This would explain──or reduce──the apparent gap between the textual and archaeological records to which we alluded above. While the great bodhisattvas are indeed to be found in the late 2nd-century translations of Lokakṣema, as we have seen, explicit evidence of the bodhisattva cult as such does not start to appear in the Chinese translations until the second half of the 3rd century. It is the gap between these later sources and the first images of bodhisattvas (other than the one who became Śākyamuni) which is significant, and this appears to be not so great. If there is any general conclusion we could draw from this, it would be to make the unremarkable observation that whenever there is a discrepancy between textual
and archaeological sources, we should ask ourselves if we are interpreting one of the sources (or both of them) correctly. As to the specific question of the process whereby the bodhisattva cults evolved, we might remember that even in our own time fictional characters have shown themselves quite capable of becoming the objects of cult with religious or moral overtones──one thinks of Leonard Nimoy’s Spock intoning the Desiderata──but when we speak of cults in this connection we are only speaking metaphorically, and such cases are trivial by comparison. The great bodhisattvas of the Mahāyāna sūtras, who eventually became the objects of real cults, were from the outset figures of high moral and religious significance, conjured up by their authors and set to work in the world, just like the nirmāṇas or emanations they themselves were described as creating within the texts. Their subsequent transmutation into saviours and guides, called upon by real people in real need, is eloquent testimony to the power of life to imitate art, and of visionary imagination to become reality.
1981. “The Evolution of the Concept of the Bodhisattva,” in Leslie Kawamura, ed., The Bodhisattva Doctrine in Buddhism (Waterloo: Wilfrid Laurier University Press), pp. 19-59.
1986. “Mañjuśrī,” in Mircea Eliade, ed., The Encyclopedia of Religion (New York: Macmillan), Vol. 9, pp. 174-175.
1932. The Bodhisattva Doctrine in Buddhist Sanskrit Literature (London: Routledge & Kegan Paul; reprint edition Delhi: Motilal Banarsidass, 1970).
1995. “Mañjuśrī and His Sword,” in K.r. van Kooij & H. van der Veere, eds., Function and Meaning in Buddhist Art: Proceedings of a Seminar held at Leiden University 21-24 October 1991 (Groningen), pp. 135-138.
1990. “Recovering the Buddha’s Message,” in Tadeusz Skorupski, ed., The Buddhist Forum, Vol. I (London: School of Oriental and African Studies), pp. 5-20.
1998. “Organised Bodhisattvas: A Blind Alley in Buddhist Historiography,” in Paul Harrison and Gregory Schopen, eds., Sūryacandrāya: Essays in Honour of Akira Yuyama on the Occasion of His 65th Birthday (Swisttal-Odendorf: Indica et Tibetica Verlag), pp. 43-56.
1982. “Sanskrit Fragments of a Lokottaravādin Tradition,” in L.A. Hercus, et al., eds., Indological and Buddhist Studies, Volume in Honour of Professor J.W. de Jong on his Sixtieth Birthday (Canberra: Faculty of Asian Studies), pp. 211-234.
1987. “Who Gets to Ride in the Great Vehicle? Self-image and Identity among the Followers of the Early Mahāyāna,” JIABS, Vol. X, No. 1, pp. 67-89.
1990. The Samādhi of Direct Encounter with the Buddhas of the Present (Tokyo: The International Institute for Buddhist Studies).
1992. Druma-kinnara-rāja-paripṛcchā-sūtra: A Critical Edition of the Tibetan Text (Recension A) (Studia Philologica Buddhica, Monograph Series, VII) (Tokyo: The International Institute for Buddhist Studies).
1993. “The Earliest Chinese Translations of Mahāyāna Buddhist Sūtras: Some Notes on the Works of Lokakṣema,” Buddhist Studies Review, Vol. 10, No. 2, pp. 135-177.
1995a. “Searching for the origins of the Mahāyāna: What are we looking for?” The Eastern Buddhist, New Series Vol. 28, No. 1, pp. 48-69.
1995b. “Some Reflections on the Personality of the Buddha,” Otani gakuhō, Vol. 74, No. 4, pp. 1-29.
1998. “Women in the Pure Land: Some Reflections on the Textual Sources,” Journal of Indian Philosophy, Vol. 26, No. 6, pp. 553-572.
1990. An Introduction to Buddhism: Teachings, History and Practices (Cambridge: Cambridge University Press).
1983. “Mañjuśrī and the Rise of Mahāyāna Buddhism,” Journal of Asian Studies, Vol. I, No. 1, pp. 12-33.
1981. “The Bodhisattva Idea in Chinese Literature: Typology and Significance,” in Leslie Kawamura, ed., The Bodhisattva Doctrine in Buddhism (Waterloo: Wilfrid Laurier University Press), pp. 125-152.
1949. The Mahāvastu, Vol. I (London: Routledge & Kegan Paul).
1982. “On the Meanings of the Words Bodhisattva and Mahāsattva in Prajñāpāramitā Literature,” in L.A. Hercus, et al., eds., Indological and Buddhist Studies, Volume in Honour of Professor J.W. de Jong on his Sixtieth Birthday (Canberra: Faculty of Asian Studies), pp. 253-270.
1960. “Mañjuśrī,” T’oung Pao, Vol. XLVIII, pp. 1-96.
1965. La concentration de la marche héroïque (Śūraṃgamasamādhi-sūtra) (Brussels: Institut Belge des hautes études chinoises).
De La Vallue Poussin, Louis
1915a. “Bodhisattva,” in James Hastings, ed., Encyclopaedia of Religion and Ethics (Edinburgh: T. & T. Clark), Vol. 2, pp. 739-753.
1915b. “Mañjuśrī,” in James Hastings, ed., Encyclopaedia of Religion and Ethics (Edinburgh: T. & T. Clark), Vol. 8, pp. 405-406.
1976. The Buddha (Harmondsworth: Penguin Books, 1976).
1937. Dictionary of Pāli Proper Names, 2 vols. (London: Pali Text Society).
De Mallmann, Marie-Thérèse
1948. Introduction à l’étude d’Avalokitesvara (Paris: Civilisations du Sud).
1986. “Bodhisattva Path,” in Mircea Eliade, ed., The Encyclopedia of Religion (New York: Macmillan), Vol. 2, pp. 265-269.
Quagliotti, Anna Maria
1990. “Mañjuśrī in Gandharan Art: A New Interpretation of a Relief in the Victoria and Albert Museum,” East and West, Vol. 40, pp. 99-113.
1994. Buddhist Saints in India: A Study in Buddhist Values and Orientations (New York: Oxford University Press).
Robinson, Richard H. & Willard L. Johnson
1982. The Buddhist Religion: A Historical Introduction (3rd ed., Belmont: Wadsworth).
1997. “A Study on the Origin of Mahāyāna Buddhism,” The Eastern Buddhist, New Series Vol. 30, No. 1, pp. 79-113.
1979. “The Mahāyāna in Indian Inscriptions,” Indo-Iranian Journal, Vol. 21, pp. 1-19.
1994. The Origins and Early History of the Mahāratnakūṭa Tradition of Mahāyāna Buddhism with a Study of the Ratnarāśisūtra and Related Materials, Ph.D. dissertation, University of Michigan.
1986. “Celestial Buddhas and Bodhisattvas,” in Mircea Eliade, ed., The Encyclopedia of Religion (New York: Macmillan), Vol. 3, pp. 134-143.
1989. Mahāyāna Buddhism: The Doctrinal Foundations (London: Routledge).
關鍵詞：1.文殊師利 2.神聖菩薩 3.大乘 4.支婁迦讖
 This is a revised version of a paper originally presented at the symposium “The Ambiguity of Avalokiteśvara and Bodhisattvas in Buddhist Traditions,” held at the University of Texas at Austin, 25-27 October 1996. For their kind invitation to the symposium I wish to thank the Department of Art and Art History and the Center for Asian Studies of the University of Texas, while for helpful comments on an even earlier version of the paper I am indebted to Malcolm McLean, Jay Garfield and Kate Blackstone.
 Gombrich 1990:5.
 A representative but not exhaustive listing of scholarship in this area would include Schopen 1979; Williams 1989; Harrison 1987, 1995a, & 1995b; Ray 1994; Silk 1994; Sasaki 1997 and Gombrich 1998.
 Snellgrove (1986: 143) writes: “The whole bodhisattva doctrine represents a remarkable aspect of Buddhist religion, expressing a degree of compassionate concern for others that is either far less developed or lacking altogether in other Indian religious traditions. The distinction between a Buddha who represents an ideal state still to be achieved and a bodhisattva who assists one on the way there remains fairly clear throughout the history of Buddhism. Only rarely can a Buddha become an object of prayer and supplication.”
 To be fair, one assumes that Snellgrove’s article is intended to be read in conjunction with Nakamura’s on the “Bodhisattva Path” (Nakamura 1986), but this way of carving up the topic is hardly an improvement on de La Valléé Poussin’s more integrated approach some 70 years earlier (1915a).
 See also Harrison 1995 for a more recent statement of the possible links between buddhology (i.e., theories about the Buddha’s person) and the emergence of the bodhisattva ideal.
 On these texts and the rationale for their use see Harrison 1993. This paper proceeds on the methodological assumption that only what is attested in the translations of Lokakṣema has evidential value for the reconstruction of the earlier history of Mahāyāna Buddhism. The testimony of the later Chinese and Tibetan translations or the Sanskrit versions where they exist may reflect later historical developments, and must be bracketed accordingly.
 The earliest clear reference given by Lamotte (1960: 4) is to 6th-century Chinese representations of Mañjuśrī as he appears in the Vimalakīrti-nirdeśa. Snellgrove (1986: 142) also notes that there is very little iconographical evidence for most of the great bodhisattvas before the 6th century. Despite more recent attempts to assert the contrary──see, e.g., Quagliotti 1990, but cf. Gail 1995──Mañjuśrī in particular does not appear to be clearly attested in India before that date. Of course, it is up to art-historians to clarify the archaeological record here, and I am happy to leave this task to them. My intention in this paper is to get the textual evidence in better focus.
 Lamotte (1960: 8) also notes Mañjuśrī’s comparatively late and insignificant role in Prajñāpāramitā literature, which is somewhat surprising given that he is commonly regarded as the incarnation of wisdom or insight (prajñā).
 Abbreviations are followed by the Taishō numbers of Lokakṣema’s translations. Unless otherwise noted, all subsequent references will be to his Chinese versions.
 Now extant──except for fragments──only in Kumāraījva’s Chinese translation (T.642) and in Tibetan. For full bibliographical details see Lamotte 1965.
 Maitreya is also said to be present, with countless other unnamed great bodhisattvas. Maitreya is mentioned again as the leader of the bodhisattvas at the end of the Sanskrit version of the text, but Mañjuśrī’s name does not reappear, there or in Lokakṣema’s translation. On Maitreya see further below.
 Assuming that this is the case, it is, of course, rather curious; see above, n. 9.
 Lokakṣema’s version has 73,000 (348b25), but this is almost certainly a scribal error.
 Cf. The Tibetan version in Harrison 1992:12 (§1E).
 For the Tibetan version in section 12G see Harrison 1992:253. Kumārajīva’s version of this section can be found at T.625, 385b10-25 (cited in Lamotte 1960:95). Both these later versions are much more detailed.
 References to the Śgs will be to Lamotte’s French translation.
 Incidentally, Lamotte (1960: 29) mistranslates the corresponding passages from T.278 (418b) and T.279 (58a), since Mañjuśrī and his vast retinue betake themselves to Śākyamuni, not to the Buddha of their home world.
 The names are not certain: Lokakṣema transcribes the Buddha’s name as Ashiduo 阿逝墮 (the last character is decidedly suspect), the buddha-kṣetra’s name as Qilianhuan 訖連桓 (for which Lamotte’s reconstruction Suvarṇavarṇa is most unlikely). However, a great deal of work remains to be done on Lokakṣema’s transcriptions before we can be sure what their Sanskrit or Prakrit referents were.
 This is T.629, the Fo shuo fang bo jing《佛說放鉢經》, on which see Harrison 1993: 155-156.
 A similar point is made by Candrakīrti in the Madhyamakāvatāra, for which see Lamotte 1960: 92-93.
 However, in his discussion of the Lokakṣema corpus, Hirakawa goes into detail only on the AjKV, the Śgs and the DSJ (but see p.27 for a summary statement).
 E.g., by Basham (1980: 31), quoting Mallmann (1948: 21).
 For some brief preliminary comments on this possibility see Harrison 1998: 556-557. I hope soon to produce a more detailed and conclusive study of this question. If T.362 is accepted as a genuine Lokakṣema translation, either in whole or in part, then this would have a significant bearing on the chronology of several aspects of Mahāyāna Buddhism. The relevant passage in T.362 is to be found at 308b11-23 (cf. T.361, 290a14-28: at this point the two texts clearly carry different recensions of the same translation), and asserts, when describing the two great bodhisattvas of Amitābha’s realm, that people in this world who find themselves in dire straits and in terror of officials (xianguan 縣官) need only take personal refuge (ziguiming 自歸命) in Avalokiteśvara and Mahāsthāmaprāpta to be set free. If this does come from the hand of Lokakṣema, then it is indeed the earliest evidence for the cult of bodhisattvas, in the terms understood in this paper. It is curious, however, that this particular claim does not appear in the Sanskrit text of the Longer Sukhāvatī-vyūha, or in the later Chinese translations attributed to Saṅghavarman and Bodhiruci, which raises the possibility that it could be a Chinese interpolation. There are several passages in T.361 and T.362 which are clearly to be explained in this way.
 Cf. Jan 1981: 139.
 The name at the corresponding point in the list (349s27) is Shichuxiji 視處悉吉, which does not look to me like a translation of Avalokiteśvara, although the character shi may render avalokita. Unfortunately Lokakṣema’s characteristic transcriptions have been replaced almost entirely by translations in the recesion of this work which has come down to us.
 For example, throughout the Tibetan version of the PraS Bhadrapāla and his fellow householder practitioners are regularly referred to as bodhisattvas and mahāsattvas.
 This passage (beginning of Chapter 8 in T.224) is ambiguous: it asserts that people designated in one paragraph as kula-putras or kula-duhit.rs and in the next as bodhisattvas and mahāsattvas who have faith in the Perfection of Insight should be regarded as non-regressing (avinivartanīya), but it is not entirely clear in T.224 whether they should be looked upon as if they are non-regressing or seen actually to be non-regressing. The Sanskrit version (Vaidya’s ed., p.104; beginning of Chapter 10) is clearer in this respect but rearranges the terms, viz. the lady or gentleman in question should be treated as a non-regressing bodhisattva and mahāsattva.
 T.224 and the Sanskrit are in agreement here, making it quite clear that it is bodhisattvas and mahāsattvas who are falling into error.
 Cf. Tib. Skye ba gcig gis thogs pa’i byang chub sems dpa’ rjes su dran pa’i phyir / 1ha rjes su dran pa’i sems rin po che, which yields a slightly different sense. As with the other items in this passage, Lokakṣema’s version implies personal realisation of the quality one is focussing upon. Along the same lines Kumārajīva’s version (T.625, 373a21) reads: “the precious thought of commemoration of the gods, in order to be fixed on the level of a bodhisattva with one rebirth to go.”
 Note that in the same passage the Sanskrit text also refers to bodhisattvas and mahāsattvas who have newly set out on the path (prathama-yāna-saṃprasthitānāṃ bodhisattvanāṃ mahāsattvānāṃ), which Lokakṣema (465c19) translates as xin fa yi pusa (新發意菩薩) “bodhisattvas who have newly conceived the aspiration.”
 On this section of the text cf. the brief comments in Jan 1981: 133, where it is suggested that Dousha jing stands for Sanskrit Daśa-sūtra. This is quite plausible. In the nidāna of later versions of the KP (including the Sanskrit), the 16,000 bodhisattvas present are also said to have come from other buddha-kṣetras and to be eka-jāti-pra(ti)baddha. Lokakṣema’s text has neither of these details (and gives the number as 12,000), nor does it mention the bodhisattvas again at the end (the Jin and Qin translations also lack them at the finish).
 Other passages in which stages or──more loosely──key points in the career of the bodhisattva and/or Buddha are dealt with or mentioned occur in WWP (435b19-c11; 437b2-10), AkTV (754c10-755a, 761b4-12), AsPP (432a29-b9; 465c19-22), etc. Strongly emphasised are the bodhicittotpāda and the passing of the point of no-return (avaivartika-bhūmi).
 We should note also that there is no trace in the Lokakṣema corpus of Mañjuśrī’s later title of kumāra-bhūta (cf. Hirakawa 1983: 19-21).
 This is in my view further evidence of the late date of the SP, so often described─without any real justification─as an “early Mahāyāna stūra.” Its first Chinese translation was made by Dharmarakṣa in 286 C.E.
 Although in T.224 he is mentioned in the nidāna, in the Sanskrit text Maitreya pops up suddenly as an interlocutor only at the beginning of Chapter 6 (in T.224 at the start of Chapter 4, 438a12), is then mentioned at the end of Chapter 8 and the start of Chapter 9 (in T.224, 443b23ff, the relevant passage begins Chapter 7; like the Sanskrit it refers to Maitreya as a bodhisattva and mahāsattva), re-appears in Chapter 14 in connection with Tuṣita (451b21-23; on Tuṣita cf. Also 468b19-24), again acts as exponent of the teachings in Chapter 19 (457c2-13), and is listed at the end (Chapter 32). In the penultimate passage in T.224 (457c3) Maitreya is described as someone who danmu dang bu fo chu (旦暮當補佛處) “will fill the place of the Buddha in a short time,” which reflects interestingly on the stock phrase yi sheng bu chu.
 In T.626 Maitreya appears at 393a17ff., 404b9ff., and most importantly, as a future provider of protection to upholders of the text from his position in Tuṣita, at 405c15-23.
 See below.
 We might mention parenthetically that the only beings who truly deserve the title of celestial bodhisattva are the devas who are supposed to have embarked on the bodhisattva path. These divine aspirants to full awakening are mentioned at several points in the AsPP (e.g., 429a19-26; 431a21-23; 435a4-20). However, their undoubted presence in Mahāyāna lore is largely irrelevant to the present discussion.
 That is to say that the bodhisattva’s role as saviour of suffering sentient beings is very much to the fore, but in a subjective sense. The approach throughout is “May I save others!” rather than “May others save me!”
 An apparent exception is found at DKP 15H (367a9-16), where the bodhisattva Maitreya and the bodhisattva Divyamauli are entrusted with the sūtra by Śākyamuni, and undertake to provide assistance to those who uphold the text after the parinirvāṇa of the Buddha. The inclusion of Divyamauli here is the closest any of these texts comes to assigning a “saviour” role to any bodhisattva apart from Maitreya, into whose ambit Divyamauli appears to have been drawn. But it is reasonably clear (especially in the Tibetan version and in Kumārajīva’s translation, T.624, 388b25-4) that the assistance promised relates specifically to the transmission of the text, and not to rescue from fire, flood, sword and the like.
 The choice of great disciples is undoubtedly significant, in ways which are now becoming clear.
 As suggested by Robinson and Johnson (1982: 79): “The strategic function of these bodhisattvas is to serve as Mahāyāna counterparts to the great arhants in the Pāli Sūtras.”
 It is also worth pointing out that in some sūtras the more unusual narrative sequences featuring bodhisattvas occur towards the end of the text, and are thus even more likely to have been later additions. The Sadāprarudita story in the AsPP, which is already found in Lokakṣema’s version, is a well-known example of this. In the WWP the first half of the text features bhikṣus like Śāriputra and Rāṣṭrapāla, while the second half relates the visionary experiences of 26 brāhmaṇas in succession. Lokakṣema’s translation of the KP, on the other hand, lacks entirely those sections (150-165 in von Steäel-Holstein’s edition) in the later Chinese, Tibetan and Sanskrit versions of the text in which the bodhisattva Samantāloka appears.
 Robinson and Johnson (1982: 79) also observe that “strangely, no Sūtra preaches devotion to a celestial bodhisattva until the third century C.E., a full three centuries after these beings entered the literature.” Leaving aside the issue of how one could know with such certainty when anyone or anything entered the literature, one can readily see that this begs the question in a somewhat circular fashion. Snellgrove (1986: 135), on the other hand, asserts more straightforwardly that the “full implications [of the cult of a celestial bodhisattva as a Great Being of heavenly associations] were developed from approximately the first century C.E. onward.”