CHAPTER 10 DEALT WITH TANTRIC GENERATIVE PROcesses that were formally designed and executed and that gradually evolved into possession as the reconstruction was effected and the symbolism became pervasive. This chapter addresses a different mode of tantric possession, one that, as described in the previous chapter, is “positive,” but is much more dramatic, cutting through deeper cultural and sociological layers.1 The term we adopt for this hitherto unrecognized form of oracular possession, taken from a few Sanskrit texts discussed below, is svasthāvesa (literally “possession of one who is in a good state of [mental and physical] health).2
Two preliminary observations are in order. First, this term itself indicates “positive” possession, the opposite of opportunistic possession of one who is ill, which, as we see in Chapter 12, falls under the āyurvedic category of āgantuka (pathology induced from without by demonic grahas [seizers]). Svasthāvesa is, however, much more specific than the literal meaning of the term indicates. Second, the textual evidence presented here demonstrates that oracular possession akin to that which fills the pages of modern ethnographies, described in Chapters 2 and 4 above, may be historicized much more than scholars have recognized. That svasthāvesa has been barely mythologized and narrativized lends weight to this historicization and allows us to trace with a greater degree of confidence its origins and diffusion beyond the borders of India and South Asia.
Although the term svasthāvesa was, apparently, never widely used, it is attested in many Sanskrit texts beginning in the second half of the first millennium C.E.. We examine these occurrences shortly. It is prudent, however, first, to provide a summary description of this possession. In svasthāvesa, a medium causes a spirit or deity to descend into any one of several reflective objects or into the body of a young boy or girl, after which the medium or youth answers questions from a client regarding events of the past, present, or future. Most of the Sanskrit texts in which svasthāveśa is mentioned by name or described without using this term are unedited tantric compendia from northern and eastern India that may be dated from the ninth to twelfth centuries. A few texts that mention svasthāvesa or its distinctive components predate this material, while others describing this practice without identifying it by name are, with rare exception, much later manuscript fragments found in libraries in South India.
To the best of my knowledge, the name svasthāvesa is mentioned in only four published texts: the Harṣacarita, a prose hagiography of King Harṣa composed by Bāṇabhaṭṭa in the mid-seventh century; the Kathāsaritsāgara, composed by Somadeva in the eleventh century; a tantric compendium called Īśānaśivagurudevapaddhati (ĪŚP) by the eponymous Īśānaśivagurudevamiśra, composed in Kerala in the eleventh century; and the anonymous tantric digest titled Tantrarāja of the sixteenth century. Except the ISP, all of these were composed in North India. I discuss these texts below, but more important are the unpublished manuscripts in which this term appears. Of central concern is the transmission of this ritual, at least in its textual redactions, from its probable point of origin in North India, eastward across the Himalayas into China and southward into the Deccan and the presentday states of Andhra Pradesh and Tamilnadu.
I discuss the condition and extent of the Indic material, a revealing intermediate tantric text called Mantramahodadhi, the key ritual and ontological terms, the ritual participants, and the immediate historical roots of the practice. I cannot here enter into a discussion of the deeper historical roots of svasthāveśa, which, I suspect, are enmeshed in shamanistic practices that are unrecoverable today. I mention this simply to acknowledge a question that will arise for many readers, not because I am in a position to answer it. The available evidence simply does not permit this. However, I examine comparable and largely derivative Chinese texts, before finally turning my attention more fully to the South Indian manuscripts. It is also important to note the historical events that facilitated the transmission and dissemination of this unique tantric material. Because svasthāvesa involves as its most striking project the inducement of deity possession in children, I conclude this chapter by speculating on the reasons for this.
The Himalayan Śaiva and Buddhist Tantras that mention svasthāvesa are primarily dedicated to descriptions of the worship of Śiva and various goddesses. The Śaiva texts fall within a class called Siddhānta, rather than under the better-known but highly suspect designation “Kashmir Saivism.”3 Thus, the texts are not the commonly cited ones (by Abhinavagupta, Kṣemarāja, Utpaladeva, and their predecessors), but the lesser-known Jayadrathayāmala, Tantrasadbhāva, Sekoddeśa, Niśvāsaguhya, Bṛhatkālottara, and Cakrasaṃvarapiṇḍārtha.4 None of these except the Sekoddesa has been edited or published;5 the remaining ones exist in manuscript form only and have been examined almost solely by Alexis Sanderson.6 These texts undoubtedly represent but a few of what must have been a much greater number of Sanskrit texts that mentioned this and other similar apotropaic rituals current at the time. What is striking about svasthāvesa, and what occupies us here in part, is the remarkable continuity of these texts with certain early Chinese Buddhist tantric texts cited by Edward L. Davis in his recent volume Society and the Supernatural in Song China and Michel Strickmann in Mantras et Mandarins and, much more extensively, in Chinese Magical Medicine,7 as well as in later manuscripts written in Sanskrit and mixed Sanskrit and Telugu in South India dating to perhaps seven or eight centuries later.
This broad conformity sparks several questions. Were specific possession cults transnational? Was Asian, especially Indian, religion organized along more microscopic definitions of lineage than I had hitherto believed? And if so, what sorts of identifiable historical forces could account for this organization? These were just some of the questions I began asking of this material. I cannot claim to provide fully satisfactory answers to all these questions, but I hope here to initiate the process of determining them.
Among my most frustrating, though not unexpected, discoveries in the course of working on this project is that Sanskrit texts dedicated wholly to deity or spirit possession are relatively few in number. Given the ritual nature of most possession, I optimistically hoped to find practical guides (prayogas, paddhatis) on ritual invocation of possession. These, I anticipated, would resemble other nontheoretical ritual texts in form and composition. Although it appears that such texts were rarely composed, they were surely written in greater numbers than the available evidence from manuscript catalogues suggests. A perusal of the available volumes of the New Catalogus Catalogorum (NCC) reveals just four manuscripts with the word āveśa in the title. All these manuscripts are housed in Chennai, in the Adyar Library and the Government Oriental Manuscripts Library (GOML).8 Three of these, including two of which are very revealing and fairly complete, are in Telugu script. The other is in Grantha, a variety of Tamil script that better reflects Sanskrit. To this must be added eight manuscripts from the recently published Descriptive Catalogue of Sanskrit Manuscripts. Volume XVI: Tantra Mantra Sāstram of the Oriental Research Institute (ORI) in Mysore. All of the Chennai and Mysore manuscripts are prayogas, though most of them are fragmentary or incomplete. In the Mysore collection, four single-folio palm leaf manuscripts in Nāgarī script appear to be copies of a prototype. Another, slightly longer, is in Tamil script, while the three longest and most detailed are in Telugu script.9
This small number, given the vast quantity of tantric and agamic ritual ephemera, suggests that most of these manuscripts did not survive precisely because they were “ephemeral.” They probably existed only in locally produced “chapbooks” that were not, on the whole, deemed sufficiently important or respectable to hand over to manuscript libraries.10 It is likely, however, that a fair amount of material on possession, and probably svasthāvesa, remains buried in collections of unexamined tantric manuscripts housed in personal and institutional libraries in various parts of India, especially in the south. Most of the recognized (and published) tantric literature in South India is dakṣiṇācāra (literature of the “right-handed path”). This “path” and its literature eventually became dominant in Tamilnadu, Karnataka, and much of southern and coastal Andhra Pradesh because of its support by the Śaṇkarācāryas, who authorized the domestication and transformation of “left-handed” practices involving Tripurā and other goddesses, assigning them an advaitic and, therefore, “right-handed” trajectory.”11 Nevertheless, an appreciable quantity of vāmācāra material was permitted within rather wide latitudes of secret teachings. Among these culturally and textually subcutaneous practices were strategies for inducing oracular possession, which was regarded by the brahmanical orthodoxies as “folk” or “popular” practice, even when brahmans participated in them as fully as others. Notable among these “others” were women and people of lower social rank, whose social and religious orbit brahmans supposedly did not frequent. Many of these practices were considered indelicate and potentially polluting, even menacing, by certain segments of the brahmanical orthodoxy. Thus, their social locus encouraged their secrecy. This marginalization included manuscripts as well, which were rarely distributed outside small communities of practitioners. I know from my own experience and that of Indian scholars with whom I have spoken that certain individuals allegedly holding manuscripts on possession and other vāmācāra topics have refused to hand them over to libraries or have even destroyed them after they have learned that scholars have suspected them of owning such manuscripts. They have been kept secret because their contents are regarded as dangerous, polluting, and unbrahmanical. This has occurred in South India, including Andhra Pradesh.
I mention this state because the most complete Sanskrit manuscripts on āveśa that I have thus far been able to locate, in the NCC and the Mysore catalogue, are in Telugu script and thus from Andhra Pradesh.12 The other relatively complete and seemingly intact manuscripts are in Grantha and Tamil scripts, thus from Tamilnadu. The most important manuscript I am aware of is a collection of āveśa sections in Telugu script culled from various Tantras.13 One section is titled Āveśabhairavamantraḥ and another is called Bhairavāveśaḥ. Neither have colophons, thus we do not know their provenance. Another section is described as the tenth paṭala of the Dakṣiṇakālikāgama, while another, titled Āveśakālikāmantraḥ, is recorded as the fifth paṭala of the Vetālatantra. An apparent addendum to this, possibly also from the Vetālatantra, is titled Bhūtāveśakramaḥ.14 The very existence of this manuscript suggests that a paṇḍit or tantric practitioner (or both) took an interest in this topic, at least to the extent of collecting various accounts of it, almost certainly for performative purposes.
One of the Telugu manuscripts that I have examined, housed in the GOML, states in its colophon that it is the third chapter (paṭala) of a text called Sudarśanasaṃhitā. Complete manuscripts of this text (in Telugu script) are housed in the ORI in Mysore,15 the Academy of Sanskrit Research in Melkote, and Sampūrṇānanda Saṃskṛta Viśvavidyālaya in Varanasi,16 while scattered chapters of it are found in the libraries of the University of Pennsylvania,17 the GOML, and (very likely) in many private libraries. I have examined the Pennsylvania manuscript, which is troublesome and corrupt18 and have found certain affinities between it and the Chennai manuscript. Although the Pennsylvania manuscript is in Nāgarī script and may be from Maharashtra, the compositional style, diction, liberal sprinkling of bija mantras, and subject matter dealing with an obscure aspect of vāmācāra Tantra (worship of Kārtavīryārjuna) reveal a respectable similarity to the Chennai manuscript. The colophon in the Pennsylvania manuscript states that it is the twelfth paṭala of the Sudarśanasaṃhitā,19 suggesting that this text is a loose compilation of tantric ephemera.20 Of the remaining manuscripts that I am currently examining, the ORI catalogue lists three composed on palm leaf, one in Grantha, and two in Telugu script. An extract from one of the Telugu manuscripts appears in the appendix to the catalogue. It is called Hanumaddīpavidhiḥ (Injunctions on the Light of Hanumān) and is identified, according to the colophon, as the twentieth paṭala of the Sudarśanasaṃhitā.21 An apparently related fragment from the GOML, discussed below, is identified as the third paṭala of this text, called Āñjaneya Āveśavidhiḥ (Injunctions to Induce Possession of Āñjaneya).22 It is important to mention this here, because Hanumān has emerged as one of the primary deities for possession in India, especially in his five-faced (pañcamukhi) form.23
Of the two primary divisions of possession, involuntary and voluntary, the former found textual bases principally in the specialized fields of āyurvedic and tantric exorcistic literature.24 However, voluntary possession, particularly that with an oracular component, appears to have been regarded as more suitable material for textuality. Two reasons may be cited for this. First, because it could be practiced, hence prescribed—and, in general, prescription was formalized through literary production. Precedents for the commission of such material to the written word are found in the literature on omens and portents, dating back to the Vedic vidhāna literature,25 scattered passages from the Sanskrit epics, and the landmark works of the astrologer and mathematician Varāhamihira around 550 C.E.26 In addition, possession was experienced as a symptom of tantric initiation, as we have seen, and for this reason it took on greater soteriological, hence literary, urgency.27 Its commission to textuality was part of an ongoing process of domesticating the ever-receding fringe, part of the expansion of Sanskritic culture into new areas of discourse that extended from the periphery to the center. This is likely a case of Sanskritization in which popular practice was brought into the realm of brahmanical orthodoxy, albeit to a place on the fringe of such orthodoxy. Sanskritization in this case might have also mitigated the threat possession presented to more normative brahmanical orthodoxy.28 This threat was of a total and unrecognizable transcendence that undermined orthodox notions of purity and an independent and impermeable ātman or self. Transcendence achieved during a state of full possession, known to many but experienced by relatively few,29 not only eroded confidence in a stable self, which now hosted other voices, but also annihilated the wellformed and complex personality, the carefully, if culturally, crafted person that served as custodian to the self. Thus, possession negated both the metaphysical and social entities, the self and the person.30
The two key Sanskrit terms that are given momentum in the Śaiva texts and define our present investigation reflect this Sanskritization, this movement from periphery to center, are svasthāvesa and prasenā. The former, as mentioned, is positive, oracular possession, a state of health (svasthā) that by its very designation must be distinguished from negative, disease-producing possession. Svasthāvesa thus indicates voluntary or invited possession of deities or middle-level spirits for the purpose of divination, while negative possession, for which there also exists no single descriptive or denotative term,31 is involuntary possession brought on by the independent agency of malevolent spirits or lower ranking semidivine beings for destructive purposes. Because an extended discussion of early Indic demonologies (bhūtavidyā) appears in Chapter 12’s discussion of āyurvedic texts, it is necessary to mention here only that the taxonomy of spirits and deities migrated to China with material trade and Buddhism.32
Of particular interest, however, is a spirit not noted in any āyurvedic text: the prasenā, a spirit mentioned, as far as I am aware, only in these tantric texts, and in Tibetan and Chinese Esoteric Buddhist texts. Neither the word prasenā nor its variant pratisenā, found in a few Sanskrit Buddhist texts, is listed in this sense in any Sanskrit dictionary.33 The word prasena (with a short, rather than long, final a) appears in the Sanskrit lexicons only as a proper name; never does it appear in its literal meaning, in spite of its clear martial associations (Skt. senā [army]).34 However, prasenā is found in its Middle Indo-Aryan (MIA) variant, pasiṇa, in the Pāiasaddamahaṇṇavo (Prākṛtaśabda-mahārṇavaḥ, PSM), a Prakrit-Hindi dictionary compiled in the 1920s. In this entry, it is defined as a minor female deity who, through metonymy, becomes identified with the divinatory practice in which she was the primary oracle. In this way, the prasenā largely escaped the purview of mainstream Sanskritic culture.
What is striking is that the Sanskrit prasenā appears to be derived from the MIA pasiṇa, the Prakrit form of the Sanskrit prasna [question]. Prasenā, then, is a back formation, derived from a non-Sanskrit word that has been subsequently Sanskritized. I am indebted to Alexis Sanderson for this insight and for forwarding to me references to some of the texts through which this can be proved.35 This is confirmed by the PSM, which cites Jaina texts that predate or are contemporaneous with the Kashmiri and other northern material and provide a meaning for pasiṇa that is very close to that of prasenā in the svasthāvesa ritual. The PSM glosses pasiṇa as “calling a deity into a mirror, etc.; a special kind of mantric knowledge,”36 citing the Jaina Āyaraṃgasutta (Ācārāṅga Sūtra) and Ṭhānaṃgasutta (Sthānāṅga Sūtra).37 The former had a long history of textual composition and accretion and contains some of the earliest Jaina writings, dating to the last centuries B.C.E. However, the later portions consist of separable texts that include exegetical commentary on the earlier parts as well as exhaustive accounts of monastic behavior.
Among these later portions is the Niśītha Cūrṇi, composed in the mid-seventh century by the prolific Jaina scholar Jinadāsagaṇi Mahattara.38 This is likely the section of the Āyaraṃgasutta that the PSM cites.39 The earlier parts are much too early to include any mention of the word prasenā. The PSM also cites the term pasiṇāpasiṇā, employing the feminine pasiṇā, attested here as well as in the Jaina Pravacanasārodhāra and the Bṛhatkalpabhāṣya (available in manuscript only). This term, which could be rendered in Sanskrit as prasenā-prasna (the art of resolving questions through the mediation of a prasenā), is glossed by the PSM as “calling a deity in dream, etc., through the power of mantric knowledge, after which it relates the fruit, both auspicious and inauspicious.”40
According to the Niśītha Cūrṇi, a question is asked of a pasiṇā, which had entered one’s thumbnail (aṃguṭṭha-pasiṇā) or arm, the leftovers after eating a sweet called kaṃsāra, a piece of cloth, a mirror, a sword blade, water, or a wall. Pasiṇāpasiṇā, according to this text, is also a kind of divination in which a question is answered by a pasiṇā who appears in a dream, thus confirming that the PSM drew its definition from this late section of the Āyaraṃgasutta. This pasiṇā is called “dream deity” or “dream-divination” (suviṇā-pasiṇā).41 It must be noted that the northern and eastern (and subsequently the Tibetan and Chinese) use of this kind of divination may well have been adopted, at least partially and indirectly, from these Jaina sects.
The word praśna, which lies at the root of this divinatory practice, means not only “question” (it is the primary word for “question” in Sanskrit) but also “horary astrology,” a usage traced at least to Varāhamihira, who uses it in this sense in the Bṛhatsaṃhitā.42 This sense persists into modern astrological practice (jyotiṣa). The two primary tasks of Indian astrology are prediction and fixing times and places for auspicious events. This is no less the case in horary astrology, in which the client asks the astrologer a question about the past, present, or future, and the astrologer answers based on the current position of the planets, rather than on the client’s natal horoscope. Because of the difficulty of determining specific birth times before the modern era of precise timekeeping, most premodern astrology in India was praśna. Eventually, it appears, praśna/pasiṇa became a more generally employed word for “divinatory practice”; it was subsequently feminized as pasiṇā or prasenā and identified as a spirit or minor deity that could be invited into the body of the practitioner or a child, or into an inanimate object, and reveal answers to questions about the past, present, and future.43
The word prasena is used by Bhaṭṭotpala (or Utpala) in his commentary (Vivṛti) on Bṛhatsaṃhitā 2.15. Varahamihira’s verse translates: “One who is instructed through magic [kuhaka-], possession [āvesa-], or by any concealed being, or from hearing [advice whispered] in the ear should never be consulted; such a person is not an astrologer [daivavit].”44 Utpala expands these exclusionary categories. He glosses kuhaka- as indrajāla- (Indra’s net), a more common term for “sorcery” or “magic.” He adds the category prasenā (prasenādikena). Āveśa, he says, means “entrance of a body by a deity, etc.” (devatādidehapraveśena).45 Concealed beings (pihita-) are those whose bodies are covered and unseen. All of these, he says, speak with disembodied voices after they situate themselves in cavities of rocks, in walls, and so on. An entity that whispers in the ear, he says, is a karṇapiśāca (ear demon).46 Those who control karṇapiśācas are well known in the world, Utpala adds. After they master certain mantras, the karṇapiśāca will reveal to them anything they want.47 Utpala offers a second interpretation. He says that those who have questions should sit around a child in a circle. The child’s father then subjects him to a certain ritual and listens to the clients’ issues. The child will then articulate the answers to the questions. My impression is that Utpala had limited knowledge of this practice and appears to have regarded it as possession; perhaps he observed such a session. It is more likely that he merely heard about it from others, as his account has a definite secondhand texture to it.
That Bhaṭṭotpala was Kashmiri must not be overlooked.48 It was in northern and eastern India that the practice of svasthāvesa through employing a prasenā to speak through the mouth of a child received its greatest impetus. We shall discuss this material momentarily. But what both Varāhamihira, who lived in Ujjain in the sixth century and who declined to grant authority to āveśa in any form, and Bhaṭṭotpala, who expanded Varahamihira’s exclusionary categories, pointedly argue is that knowledge that is neither scientifically verifiable nor systematically obtained, which is to say through a recognized lineage, is valid. Varāhamihira understood that the nature of astrological and other forms of classifiable but fluidly realized knowledge, such as lore associated with omens and portents, invited frauds, amateurs, and charlatans.49 Thus, in the second chapter of the Bṛhatsaṃhitā he attempted to delineate professional standards. This was obviously a delicate topic that required firmness by both the author and the commentator. What it also demonstrates, above and beyond the situating of arcana as a sastra or legitimate body of knowledge, is the division between “folk” and “classical” that was readily perceived by the bearers of Sanskritic culture in the first millennium. Varahamihira’s exposition also illumines his efforts to encroach upon the margins of folk culture by incorporating certain elements of it into the Sanskritic mainstream.50 Despite the apparent ease with which Sanskrit literati co-opted or classicized folk elements, the division between folk and classical always remained sharp, even if the contours of the division proved variable or fluid. Indeed, the awareness of inside and outside, and the readiness to blur and amalgamate alterity, was in India one of the keys to the surprising duration of the control by the Sanskrit literati of a variety of discourse modes. This control, as well as the ever-shifting distinction between folk and classical, was often presented as little more than an exercise by the literati to uphold the epistemological requirements of Sanskritic knowledge systems.51 Indeed, what most people, including brahmans, did in practice was surely quite different. This is the situation today, and there is sufficient evidence in Sanskrit texts of all periods to demonstrate that it was the case in antiquity. Thus, the discourse models adopted by Varāhamihira, Bhaṭṭotpala, and most Sanskritic philosophers and systematizers, omitted the inconsistencies that we find, for example, in similar literature in China, as the work of Strickmann and Davis testifies.
An example of the Sanskritization to which we refer, and the challenge to accepted Sanskritic epistemological agendas, is the employment of children as central in Sanskritic ritual. This certainly represented an encroachment of popular practice into the corpus of Sanskritic culture, as there is no evidence for it in earlier vedic, epic, or purāṇic ritual. Nor is there evidence that it is an innovation of mid- to late first-millennium tantric systematizers. The engagement of children as spiritual factotums in adult rituals thus subverts ritual boundaries that are, for the most part, set by the Vedic ritual theorists, who do not grant children adhikāra for Vedic ritual performance. Indeed, even in the Tantras that are key to our understanding of both the ritual use of children and the incumbent epistemological process, this subject is mentioned only sporadically. Nevertheless, these are the passages that we must examine here. Children are as capable as adults of understanding and responding to ritual,52 and, apropos of our current project, it is the Chinese rather than the Indians who appear to have recognized and legitimated this capability in texts that offer thicker descriptions of the responses of children to such ritual. Regardless of whose texts may be more revealing, the question of the use of children is an important one. I speculate on this below; but it may be said now that inducing possession in another person, whether in an adult or a child, rather than in an inanimate object or in oneself, not only was good theater but also gave the impression that the responsibility of agency and the temptation to manipulate results were reduced, which doubtless conferred a greater sense of authority and authenticity upon both the oracle or medium and the intermediary ritual officiant.
As in the Jaina Prakrit and Kashmiri Sanskrit texts, in Tibet and East Asia the terms cognate to pasiṇa/prasenā denoted both the tantric ritual and the deity that mediated the oracular experience. In a few esoteric Buddhist texts of the Kālacakra tradition, notably the Sekoddeśa, which appears to have been a section of the “root” Kalacakra Tantra (the Paramādibuddha or Kālacakramūla Tantra),53 and Kalkin Puṇḍarīka’s Vimalaprabhā on the Laghukālacakra Tantra, the term pra phab-pa, a translation of the Sanskrit pratisenā (prognostic image), is attested in the Tibetan versions.54 This is surely a variant of prasenā. It also appears in chapter 43 of the Cakrasaṃvara-Tantra, in a verse that reads, “Having repeated the mantra over a sword, water, one’s thumb, a lamp or a mirror, one will cause the descent of the divinatory image [prasenā] by means of the yoga of oneself [as the deity].”55 This is clearly drawn from earlier Indic material.
Giacomella Orofino has written an important and informative article on the Tibetan art of mirror divination, in which pratisenā occupies a major place.56 Orofino understands pratisenā as “a hyper-Sanskritization of the word prasenā, a term of uncertain etymon.”57 We have here dealt with the etymology, though Orofino is incorrect in labeling pratisenā a hyper-Sanskritization.58 The commentator on the Tibetan version of the Sekoddeśa, Nāropā, cites a text called the Pratisenāvatāratantra (Tib. pra dbab pa la ‘jug pa’i rgyud), now lost, which, he says, enumerates eight kinds of prognostication, all of which, it seems, we may regard as varieties of svasthāvesa.
We now turn to the Himalayan and eastern rite of svasthāvesa, not losing sight of our objective of comparing it to the manuscript material from China of the same period and southern India of later epochs. Although the relevant Tantras are themselves undated, Sanderson suggests that one of the earliest, the Nisvāsaguhya, could be a product of the sixth century, though it can be dated with certainty only to a single ninth-century Nepalese manuscript. Thus, Sanderson states that, though the Nisvāsaguhya “seems to be one of the very earliest Tantras of the Siddhānta,” it is nevertheless representative of “a tradition for which we have epigraphical evidence from the sixth century onwards.”59 Sanderson’s remaining texts, which describe svasthāvesa in greater complexity and variation, were composed at various times up to the twelfth century. The epigraphical evidence to which Sanderson refers is important because it helps establish a general dateline for the diffusion of svasthāvesa to China. This evidence exposes a Śaiva tradition from Central and South India that appears to have entered Kashmir only in the first few decades of the eighth century with the ascendancy of Lalitāditya, the wealthiest and most successful of Kashmir’s imperial monarchs. It is worth noting that Lalitāditya and his successors patronized both brahmanical sects and Buddhism, establishing a climate conducive for intellectual exchange. Given the strong tantric presence in Kashmir at that time, this active exchange of ideas quickly engendered a common tantric foundation for Hindu and Buddhist sects.
Sanderson and others have extensively explored the epigraphical evidence for the background of northern Tantrism. Among the most important epigraphia are the following:60 (1) A hoard of nine copperplate inscriptions of a local monarch named Mahāsivagupta Bālārjuna discovered in Sirpur (Śrīpura, the ancient capital of Dakṣiṇa Kosala; Raipur District, Chattisgarh), which includes records of grants to Śaiva ācāryas. This is shown by their initiation names, which confirms that the Śaiva Siddhānta was established by the second half of the sixth century.61 (2) The Senakapāṭ stone slab inscription issued in the fifty-seventh year of the reign of Śivagupta (c. 647).62 This records a grant to an ascetic named Sadāsivācārya, who, the inscription implies, is the lineage successor to an ascetic named Sadyaḥśivācārya who hailed from a hermitage (tapovana) called Amardaka.63 The latter is also a name of Kalabhairava, a form of Śiva dominant in this tradition, who, as we see below, is closely affiliated with the possession deities in some of our texts.64 It is also important that this hermitage is the parent institution of most of the Saiddhāntika lineages recorded in inscriptions, lineages that eventually became ascendant in Kashmir. (3) Inscriptions recording the Śaiva initiations to several important southern monarchs in the latter half of the seventh century.65 Although this is evidence for the broad dissemination of Śaiva lineages, the inscriptions do not take note of specific practices. Nevertheless, linguistic and cultural evidence permit us to speculate on the range of practices that constituted the divinatory milieu of svasthāvesa.
It is striking that the year 647 is mentioned; this coincides with the conclusion of Harṣavardhana’s reign in northern India (r. 607–647). It was in approximately this year that his court poet Bāṇa wrote the Harṣacarita (The Deeds of Harṣa), a hagiographical description of Harṣa and his reign. This well-known and exemplary text includes the earliest attested mention of the term svasthāvesa, referring in one passage to a kind of powder used to induce it.66 This is noteworthy because the later Sanskrit manuscripts of South India, to be discussed below, require scented and powdered ash (bhasma) from an offertory fire ritual (homa) as one of the ingredients. Together with the Jaina evidence, this suggests that svasthāvesa had by this time entered into the general currency of cross-disciplinary esoteric practice,67 even if textual mention of it was rare. This, however, is not surprising, as Harṣa was known for his tolerance toward both heterodox and orthodox sects and took an active role in bringing them together. Indeed, along with twenty kings, the most eminent śramaṇas and brāhmaṇa are said to have attended a great assembly he once convened in his capital city, Kanyākubja. It is at convocations such as this that exchange of knowledge that contributed directly or indirectly to the formation of svasthāveśa ritual could have occurred.
With this in mind, then, let us now turn again, briefly, to the Śaiva and Buddhist Tantras and to the ritual employment of the prasenā. The Niśvāsaguhya, the Tantrasadbhāva, the Cakrasaṃvarapiṇḍārtha, the Sekoddeśa, the Bṛhatkālottara, and, most importantly, the Jayadrathayāmala (JY), the latest and most compendious of these texts, present a complex intertextuality.68 So, it is fair to present a composite account of the ritual of svasthāveśa.
The texts state that the prasenā is a divinatory form assumed by a female deity (vidya) that appears before the practitioner during a ritual performed at a site sacred to Maheśvara (rudrasthāna) or in a temple to the goddess (mātṛgṛham). The practitioner, called mantrin, māntrika, or simply sādhaka, repeats (japa) relevant mantras one hundred, one hundred and eight, one hundred thousand, or five hundred thousand times. This empowers him to undertake the āvesa. In these Tantras this ritual is always presented by Śiva, sometimes called Bhairava, as a great and wondrous secret, in response to queries from his dialogic counterpart, the Goddess, who is given a large number of names. As a result of these mantra repetitions, a conjunction (saṃyoga) is forged with a prasenā, classified as a Vidyā or Mahāvidyā goddess. This assumes a kind of communication, in which the mantrin either becomes fully identified with the prasenā or gains control of her. The JY declares that the mantrin possesses a special reverence, which is a result of meditation properly performed on this highest śakti (feminine energy). In a statement reminiscent of later Tibetan esoteric Buddhist initiatory practice, the JY states that this śakti arises in one’s belly on the disk of the moon, and it is here that the mantrin accesses her, in waves.69 The practitioner becomes established in a divine body (divyakāyasthitaḥ) and is able to perceive the expanded significance of a single point (bindumandiram).
There is apparently not just one standard prasenā but, potentially, many. These prasenās prevent the ill effects of poisoned food, repel all manner of black magic (the well-known ṣaṭkarmāṇi are listed), prevent capture by serpents, and counter destruction by the weapons of enemies. Through a seamless connection with such a divinity, the progress of ensorcelled boats, carts, and machines (yantra) is halted. Through a hundred thousand mantra repetitions, says the JY, one achieves the goal of this ritual, freedom from adversity. The goddess abiding or being reflected in various media is seen directly. She shows the past and the future that the sādhaka himself sees.
The loci of the prasenā are many. She appears in highly insubstantial and unstable form as a shadow or apparition. Her appearance does not come out of thin air, however, but
on the surface of water, a metal pot, a sword-blade, in the flame of a lamp, in a mirror, the eye of a girl, the sun, the moon, his own thumb smeared with oil, or within his body in the point of light between his brows, and there reveals the answer he seeks. The answer may take the form of apparitional writing or a disembodied voice, or it may be uttered by a young boy or girl placed in a trance for this purpose (svasthāveśaḥ); or it may appear to such a medium in one of the aforesaid substrates; or it may arise in the sādhaka’s mind when he awakens after a night spent in a temple of the deity.70
The Buddhist practice of pratisenā, noted by Orofino, requires fairly extensive preparatory ritual, including the propitiation of spirits, the consecration of the ritual ground by covering it with a fresh layer of cow dung, the construction and disposition of yantras, the recitation of mantras, offerings to various divinities, and the employment of a youth to gaze at a specified reflective surface. This may be a mirror, a sword, a thumb, a lamp, the moon, the sun, water, and the eye, any of which can become a medium through which a pratisenā appears.71 Citing the Tibetan version of the Subāhuparipṛcchānāmatantra, (Questions of Subāhu), Orofino notes that the youth must be washed and dressed in clean white clothes, and possess certain physical characteristics, including “perfect limbs, no prominent veins, bones or muscle joints, a beautiful shape, clear and slanting eyes.” Furthermore:
He or she should sit on a rug of kusa grass facing eastward, incense should be burnt, and the secret mantras should be recollected. After having removed the dust from the mirror, one [the officiant] should recite the mantra, mentally concentrating on the boy (or the girl), seven, eight or ten times.… [Then] the young medium, facing westward, looking through the eye of the divinity, will see past, present, and future events.72
The South Asian texts, including the Tantras in question73 and the South Indian manuscripts, omit much of this detail. Some of it may be assumed, however, including the physical prerequisites for the youth, as these were common features of Indian ritual prescription. Orofino’s description gives the impression that this type of divination was not uncommon, that it was part of the known complex of shamanistic practice in Tibet.
We have seen from the mid- to late first-millennium Jaina texts that the ritual of svasthāvesa was multidenominational in South Asia, that this was shared across sect and lineage boundaries. A further example of this, though not explicitly labeled svasthāvesa, is found in Hemacandra’s Yogaśāstra (5.173–176). The context here is very specific, the determination of lifespan. In this passage, the deity is said to enter into an inanimate reflective medium, such as a mirror, after which a young girl picks up and passes on messages transmitted by the deity through the reflective surface. One might argue that this is not possession of the girl; rather, it is an allied divinatory practice. However, in South Asia, people, especially women, are considered possessed if they transmit such messages in trance states. There is every reason to believe that this was also the case in the first millennium. Hemacandra says:
Upon being queried, a deity [devatā], who has been made to descend into a mirror, a thumb, a wall, or a sword through a rite involving mantra repetition, announces her verdict regarding time [of death]. The mantra to be recited is oṃ naravīre svāhā, and is perfected after 10,008 repetitions of it during a solar or lunar eclipse. After that, whenever such a question is asked, the ritualist need repeat it only 1,008 times at that moment. The deity then becomes absorbed in the mirror, etc., following which a young girl [kanyā] announces the verdict. In this way, the deity, attracted by the virtues of a good sādhaka, herself speaks decisively on topics regarding the past, present, and future [trikālaviṣayam].74
This variety of media that might become possessed appears to be retained in the comparable Chinese texts, though the later Sanskrit manuscript material focuses on inducing āveśa only in young boys. It is interesting to consider that the earliest Sanskrit suggestion of this procedure is found in one of the foundational texts of Āyurveda, the Susruta-Saṃhitā, composed in approximately the second century C.E. This text includes the following verse: “As a reflection is to a mirror or other similar surface, as cold and heat are to living beings, as a sun’s ray is to one’s gemstone, and as the one sustaining the body is to the body, in the same way ‘seizers’ [grahāḥ] enter an embodied one but are not seen.”75 This raises an important question: Do we see in this statement a movement from metaphor to metaphysics as the idea proceeds from āyurvedic texts to tantric texts? Can we assume that what began as a trope ended up assuming a concretized cosmological and ritual locus? Or does Susruta take an older known divinatory practice and turn it into a trope? Based on the present state of evidence, it is impossible to answer these questions confidently.
The question arises as to whether we can justly assume that the primary Indian source materials are in fact the basis of the practice of svasthāvesa found in China or whether China had its own indigenous sources quite apart from the Indian ones. To take this one step further, we can ask whether the advent of svasthāvesa and similar divinatory practices in India might be explained as borrowings from China, which, it is well known, had a rich history of such practices. Although I believe that we may preemptively reject this, the question arises because Davis cites Tang texts of the seventh and eighth centuries that agree almost verbatim with the Indian Śaiva and Buddhist texts that may be slightly later in composition. Furthermore, the epigraphical evidence cited above, though clearly linking the northern Saiddhāntika tradition with similar traditions farther south, fails to specify practices, as we have seen. An objection might then be raised concerning the provenance of these practices, especially because the earliest reference to svasthāvesa in South Asia, in the absence of other data, appears to be in the Harṣacarita, a text almost precisely contemporaneous with the earliest Chinese material. With this in mind, we should settle any lingering doubts regarding the matter of the direction of transmission.
Possession ritual is documented in China from the mid-first millennium B.C.E. onward, and scholars point to artistic and epigraphical evidence that might push that date back another millennium.76 Most of this speculation revolves around the issue of shamanism, which, because of cultural variation and lack of perspicacious definitions, falls prey to the same sort of amorphous characterization (and caricature) as befalls Tantra.77 Indeed, in South and Central Asia, what may have been more strictly speaking shamanistic became identified and classified as tantric, a problem that continues to find no agreeable solution among either scholars or indigenous participants. An example of this definitional vertigo may be found in modern-day South Asia, where spirit healers who might more accurately be described as shamans are widely regarded by both their clientele and scholars as tantrics, a label that they stridently eschew.78
In partial answer to our question, it must be noted that not only do the sinologists Strickmann and Davis acknowledge the Indian origins of this kind of divination (references to Ucchuṣma,79 Nāgārjuna, Gaṇeśa, and other figures traceable to Sanskrit abound), but they do not entertain the possibility of influence flowing in the opposite direction. This view is enhanced by the studies of Iyanaga, Stein, Snellgrove, Davidson, and others who have addressed the Buddhacization of Maheśvara/Rudra/Śiva and his journey to the east. Iyanaga recounts the submission and conversion of Maheśvara by Trailokyavijaya according to the Chinese and Japanese sources.80 Stein tells of the conversion of Maheśvara to Buddhism by Vajrapāṇi, “the wrathful suppressor of all evil ones,” using several Tibetan tantric sources,81 while Snellgrove describes the rise to prominence of Vajrapāṇi and his defeat of the demon Maheśvara according to the Mañjuśrīmūlakalpa.82 Davidson employs Indic and Tibetan (Sa-skya-pa) sources to relate the tale of Maheśvara’s subjugation and rebirth as Heruka, “the cosmic policeman,” and analyze historical, literary, and doctrinal functions.83 Maheśvara, says Davidson, “became one of the great scapegoats of Buddhist Mahāyāna literature.”84 As in Davidson’s passage from the Sarvatathāgatatattvasaṃgraha, Snellgrove’s from the Mañjuśrīmūlakalpa has Vajrapaṇi crushing Maheśvara with his foot. Following this, the Buddha himself “entered the body of Maheśvara and he pronounced this verse: ‘Oho! The peerless wisdom of all the Buddhas. Even a body which is dead returns to the sphere of the living!’”85 Vajrapaṇi then uttered the mantra oṃ vajra muḥ and formed a mudrā with his fingers, after which Maheśvara rose from the dead, imbued with the spirit of the Buddha, and made the world safe for mankind.
Most important, however, is the work of Robert Mayer on the assimilation of Maheśvara in Tibetan Buddhism.86 Greatly influenced by Sanderson’s work on early Śaiva Tantras, Mayer concludes that the figure of Maheśvara in Tibetan Vajrayāna (and onward into Chinese esoteric Buddhism) was a reconfiguration of this deity from the sectarian Saivism of the Indian Himalayas, specifically of the kāpālika sect.87 From there, Davidson notes, it was a short step to the transformation of Maheśvara into a powerful spirit, a Heruka under the control of Buddhist deities. It is this Maheśvara, much more prominent in India, who, in his many guises and aspects, particularly as (Baṭuka-)Bhairava and Hanumān, was possessed in the ritual of svasthāvesa and assumed a major role in other divinatory practices.88 In China (though not quite in Tibet) these guises fell away, leaving Maheśvara as a divinatory deity in his own right. The early epigraphical evidence in India testifies to the existence of the kāpālikas and other transgressive Śaiva cults89 and lends credence to the argument for a literature, now missing, that linked these epigraphical sources to the later Śaiva siddhānta texts that contained sections on svasthāvesa (and other forms of āveśa). These texts, then, must have been transmitted to Tibet and China, serving as the textual basis for oracular possession along the lines of svasthāvesa.
Sanderson’s continuing work on Śaiva and Sākta literature and traditions that derive from this demonstrates that this mountainous region was home to a large number of sects from roughly the sixth to fourteenth centuries that worshipped Śiva/Rudra/Maheśvara and a large number of goddesses, many of whom are not described elsewhere in Indic literature. This veneration was largely apotropaic and divinatory, utilizing yantra and mantra, and much of it was better preserved in Tibet (and subsequently China) than it was in India, at least in northern India. These practices found a much more agreeable home in Tibet, where Buddhism, in its “co-option of Śaiva kāpālika tantrism,”90 tamed the wild god, as the studies of Mayer et al., indicate. This was due in some measure to the influence of certain shamanistic practices in Central Asia. Regardless of the reason(s), however, Mayer tracks the taming process.91 As suggested, his primary focus is on the manner in which Buddhist mythic narrative transformed Śiva, subverting his raw, menacing, transgressive power to the aims of Vajrayāna Buddhism, assigning him a secondary place in Buddhist maṇḍalas, and so on. Two forms of this conversion appeared in Tibet, “one Mahāyoga concerning the converting of Rudra, and one Anuttarayoga, describing the converting of Bhairava.”92 It is striking that in southern India of later periods, as well as Tibet, the deities invoked in svasthāveśa were Maheśvara and Bhairava (though in India Hanumān was later added to the staff). In sum, in spite of contemporaneous, or perhaps earlier, Chinese sources, we must assume on linguistic and cultural grounds that Indic texts preceded both the Chinese and surviving Kashmiri texts. The linguistic evidence, which we now examine more closely, consists of the appearance in the Chinese texts of cognates for the Sanskrit āveśa and Maheśvara, as well as descriptions of the practices of ritual that are indisputably borrowings from India.93
Strickmann and Davis cite several texts that contain material strongly reminiscent of Indic and Tibetan āveśa and svasthāvesa. Indeed, the Chinese employ the word aweishe a direct transcription of āvesa, “to designate possession rites in which a spirit was invoked into the living body of a medium. The term might also apply to procedures in which the spirit of a living person was co-opted, so to speak, into the pantheon.”94 Strickmann and Davis cite a large number of passages, including one from the Amoghapāśasūtra,95 which strikingly resemble Orofino’s description of childhood possession in the Tibetan Subāhuparipṛcchānāmatantra. According to Strickmann, the Amoghapāśasūtra, which was translated from Sanskrit into Chinese around the end of the seventh century, was the first Chinese Buddhist text to give instructions for inducing deity possession. It was at this point, says Strickmann, that a “new Tantric synthesis was about to become known in China”96 The purpose of this aweishe ritual, which invokes Guanyin (=Avalokitesvara), was therapeutic, to heal an individual suffering from spirit-induced illness. The passage is not an isolated ritual option, but occurs in a more general discussion of ritual healing. The entire passage is worth citing:
Or again, there is this method. If it is desired to enchant a person, the spell-possessor should bathe himself and put on fresh garments. Next he should recite the spirit-spell to protect his own person. Then he is to construct a ritual area using cow dung, making it square and painting it in the appropriate colors, strewing assorted flowers, and setting out various white-colored food-offerings. Next he should take a virgin boy or girl, bathe the child, and imbue its body with fine fragrances. He should clothe it in a pure white garment and adorn it with all manner of ornaments. He should then have the child sit cross-legged in the ritual area; he recites the spell bandha (“bind”) and he plaits the child’s hair. When he is done reciting the spell and plaiting the hair, he takes more flowers and fills the child’s hands with them. In addition, he takes fine quality incense, crushes and scatters it. Then, additionally, he recites a spell over uncooked rice, which he sprinkles, together with flowers and water, within the ritual area. Next he should burn sandalwood incense and recite Kuan-yin’s spirit-spell; he should recite it three times over the flowers and then cast them in the child’s face. Then the child’s body will begin to tremble. If you wish it to speak, pronounce another spell [given in the text] over pure water and sprinkle it in his face. As you recite the spell, be sure that your hand does not touch the child. When you have recited in this manner, the child will speak. If you ask about good or evil things in the past, future, or present, it will be able to answer all your questions. If the spell-holder wishes to send away the spirit who has lodged in the child, there is another spell given which he should recite.97
In fact, the phenomenon of child possession receives much more attention in Chinese religious texts than in South Asian or Tibetan ones. Whether this indicates greater occurrence of this kind of svasthāvesa in China is an open question. Although I suspect that this was the case, it is also possible that the Chinese texts were simply more richly descriptive of such activity and more amenable to including the experiences of non-elite groups, whereas the South Asian prescriptive material was more vigilantly or fastidiously spare and brahmanical. There is little point in reviewing all the examples set forth in Strickmann’s books. He does, however, translate and comment on the Chinese translation of the Subāhuparipṛcchā, where Posseu-na, Sseu-na or the god Po reveals desired knowledge after being caused to descend into objects similar to those on the list in the Kashmiri Sanskrit material.98
Strickmann comments on some of the terms discussed here. He recognizes the importance of the word prasenā, admitting that its origin is obscure. He suggests, incorrectly it seems, that it might be related to the name Prasenajit.99 However, the Bukkyōgo daijiten, a Japanese dictionary of Buddhist terms, gives the transliteration of prasena (here, contrary to Strickmann’s texts) in Chinese as poluosai a kind of chess game based on military formations that use elephants and horses, also called xiangqi 100 This is based on Fazang’s (643–712) commentary on the Chinese translation of the Brahmajāla sūtra (Fanwang jing ), the Fanwang jing shu What this suggests is that the name Prasenajit, which in any case is much too early to indicate one who is victorious over a spirit called prasenā, probably means “one who is victorious over a ruler whose military strategy included formations of elephants and horses.”101 As for the Tibetan pra pha[b]-pa, he rightly translates this as “to make a pra spirit descend.”102 Both Strickmann and Davis suggest that the entry abhisha (the Japanese transcription of the Sanskrit āveśa) in Hōbōgirin (I:7ab) indicates that hashina, the Japanese term for the āveśa ceremony, may also be derived from the Sanskrit praśna, again from the questions that are put to the medium by the performer of the ritual or his client.103
Davis’s work also demonstrates that during the seventh and eighth centuries the terminology and character of certain varieties of divination took on a decidedly South Asian character. During these centuries, the “aweishe rite” was introduced through esoteric Mahāyāna Buddhism. Davis discusses at length this rite and the general question of the use of prepubescent mediums. He says, in part:
All in all, the āveśa rituals seem to have been not too distant from the most popular kind of séance performed by the Tang-dynasty spirit-medium for officials and members of the aristocracy. In these séances the spirit-medium summoned a divinity either into himself or into some other localized place. Then he either answered himself, or communicated the divinity’s responses to, specific questions posed by the supplicant about “good and evil fortune,” “matters past, present, and future,” or “things yet to come.” What has changed in the āveśa rites is the fact that the spirit-medium’s trance was officiated by a Tantric master and resulted from the power conferred on the master by his own identification with a Buddhist deity. In The Rules of Āveśa as Explained by the Deva Maheśvara, a text in one short scroll submitted to the throne by Amoghavajra, the officiant or “practicant,” as he is called, must first transform himself into the eponymous Maheśvara, thereby empowering the mantras and mudrās he then employs to compel an “emissary of Maheśvara” to possess the children. The text notes that after the appropriate spell has been recited seven times, “the boys and girls will tremble violently, causing one to realize that the “Holy One” (shengzhe) had entered their bodies.104
Davis indicates that the children were placed on yantras, though he does not describe them. He informs us, however, that “the Tantric master recites various incantations to summon one or more Buddhist deities, sometimes identified, sometimes not, into either a luminous, reflective object (water, mirror, jewel, pearl), the image or icon of the divinity, or the body of a child.”105
From two to ten boys or girls between the ages of eight and fourteen were possessed. They were subjected to elaborate purifications after their eligibility had been confirmed as a result of meeting certain requirements in physical appearance. The Chinese texts state that “when the divinity has descended into the children, they will speak of ‘all matters past, present, and future,’ of ‘things yet to come,’ or of ‘good or evil fortune,’ or that they will be able to answer any question asked of the god.”106 Two of the texts require the children “to gaze into a mirror, which had either been purified with ash from a homa offering or otherwise empowered by the divinity, and then to answer all manner of questions based on what they saw.”107 With respect to the emissaries of Maheśvara mentioned above, whatever their original identity, it is certain that in later periods, in India at least, Hanumān and Bhairava were regarded as emissaries of Śiva, identical to the eponymous Maheśvara.
Who were these children? Davis excerpts part of a Buddhist tantric text that records the possession of a boy by Ucchuṣma. In Sanskritic mythology Ucchuṣma, “the Crackler,” is an aspect of Agni. However, in esoteric Buddhism Ucchuṣma, glossed as “the Vajra-Being of Impure Traces,” commands a considerable host of spirit-soldiers. The possessed boy was called “Holy One” (daozhe) by the monk who wrote the text, either because he was considered to have dedicated himself to a religious life as a follower of “the Way” or because he was transparent to the divine, the “Venerable Spirit,”108 who may have been Ucchuṣma. The possession here is described vividly, but not paradigmatically. That the boy is said to have jumped and flung himself about, grabbed a sword, run out of the temple gate until he reached a pile of cow dung, struck the pile three times with his sword, then collapsed on the ground,109 leads Davis to conclude that in this episode:
we are far from the highly controlled, rarefied, and even claustrophobic atmosphere of the Buddhist āveśa rites, in which two or more virginal and purified children stood passively before the master amidst lighted incense and strewn flowers; in which the descent of the divinity and the onset of trance were distinguishable only by the most subtle of signs—the cessation of breathing, unblinking eyes, and a slight reddish tint around the pupils; and in which the children had in essence become living icons, as luminescent, but also as confined, as the pearl or crystal for which they were substitutes.110
The “basic structure of the Tang rituals of āvesa,” says Davis, consists of “the controlled possession of a boy by a cultic divinity and his subsequent clairvoyance.”111 The archive of āveśa material that Davis and Strickmann introduce demonstrates that possession ritual in Chinese religion, whether identified as tantric Buddhist, Daoist, Daoist-influenced esoteric Buddhist or vice versa, as well as rites of Daoist- and Buddhist-influenced independent spirit exorcists, changed radically during the Tang dynasty (618–907).112 This, it appears, was due to Indic (and Tibetan) influence. The general tenor of the change was that late- and post-Tang possession ritual was more extreme than that of the non-tantric ritual whose practitioners frequented the early Tang imperial court.
The vast majority of cases of possession described by Davis and Strickmann share the essential elements of the ritual of svasthāvesa found in the earlier tantric texts and the later South Indian manuscripts. Other Chinese aweishe and Japanese abisha materials that they cite support this ritual kinship. The most important common features of these rituals are the following: (1) the imposition of the spirit into the medium or into a compatible third party, which is, characteristically, a child, though the texts add that a divinity “could also be compelled to enter a reflective object or hollow icon,” such as a mirror;113 (2) mention of “matters past, present, and future,” which indicates an oracular function to the ritual; (3) the engagement of Maheśvara, which directly indicates a link, first, to the deities of the Indic ritual of svasthāvesa, which include Maheśvara and, second, to Bhairava and Hanumān, both aspects of Śiva, of whom Maheśvara is a multiform; and (4) the use of mantras and mudrās, which help induce the possession. Davis states: “The unity of practitioner and divinity is a defining feature of Esoteric Buddhism and a mark of the extent to which even Daoist therapeutic rituals had become ‘tantracized’ in the Song.”114 This is a feature not only of Esoteric Buddhism, but also of Indian Śaiva theology and practice.
To take this one step further, this is a feature not only of Himalayan Śaiva and Sakta practice, but of subcontinental Indian devotionalism (bhakti), a fact readily evident in both mid-first-millennium Tamil devotional poetry and approximately contemporaneous Sanskrit counterparts. It is generally agreed that Tamil devotionalism exerted critical influence on the Sanskrit texts, helping to establish both the fundamental patterns of devotional practice and the emotional or evocational parameters of the devotional constitution.115 This Sanskrit devotionalism subsequently influenced the Kashmiri Śaiva and Sākta practice and textuality, where elements of tantric urgency and severity, including worship of Mahāvidyās and other sanguinary goddesses,116 were compounded with this otherwise zealous devotionalism. This was then replicated in the devotional fervor characteristic of Esoteric Buddhism. In short, the devotional impulse (bhāva) expressed most decisively in the Vaiṣṇava literature of the subcontinent (e.g., the Bhāgavata Purāṇa) is heavily implicated in the development of Esoteric Buddhism, a link that has been insufficiently explored.117
We must now briefly examine a few Sanskrit āveśa manuscripts found in the South Indian libraries. These manuscripts are quite ordinary in that they prescribe the construction of easily inscribed yantras or maṇḍalas that were likely in common use.118 Other normative features are the recitation of bija mantras found in nearly all mantra collections (though they have their own unique order here), and brief statements of purpose and fruits to be gained (saṇkalpa). Although they have a standard format and structure, these manuscripts have unique content, as shown by a composite summary.
The ritual of āveśa should be performed secretly, in the privacy of one’s house or, better, at an isolated Śiva temple. An officiant (māntrika)119 oversees the construction of one or sometimes two maṇḍalas inscribed on the ground with colored powder. The maṇḍalas are described in minute detail; indeed, these descriptions occupy the greater part of the manuscripts. They are oversized yantras, geometric representations of the deities for whom they are intended, in the present cases Hanumān and (Vaṭuka-) Bhairava.120 The latter is, in a manuscript titled “Āveśabhairavaṃ Śarabhakalpe,” identified with the sarabha, the ghastly eight-legged celestial beast brought into the realm of the gods by Śiva.121 After the maṇḍala is constructed, an eight-year-old boy with good qualities, who has bathed and is pure, is seated on it in lotus pose.122 Then the māntrika recites a hundred times the long and intricate mantras designed to bring about possession. After this, cooled, powdered, and scented ash (bhasma, vibhūti) from a havana (ritual fire) is applied to the boy’s forehead. The boy should then become possessed by Hanumān or Vaṭukabhairava as the māntrika calls out āvesaya āveśaya (let him become possessed, let him become possessed), after which the boy gains the ability to communicate knowledge of the past, present, and future, including auspicious or inauspicious fruit that may be reaped by the client in a future birth.123 The manuscripts do not state whether this information is given only upon questioning or whether the boy is understood to speak at the independent instigation of the deity. The possession is lifted after the ash has been applied twenty-one times, accompanied by another long mantra.
A partial model for the Hanumān āveśa understood in these manuscripts occurs in the Rāmāyaṇa. The manuscript called “Avesa Hanumantam”124 states that the boy becomes identified with Āñjaneya after mantras for Hanumān as the messenger of Rama (ramaduta) and Lakṣmaṇa are recited. The exorcist should utter the words āveśaya āveśaya ehi ehi praveśaya praveśaya (Possess! Possess! Come! Come! Enter! Enter!), followed by an exhortation to follow the example of the Rāmāyaṇa, “O Rāma and Lakṣmaṇa, possess the violated palace in Laṅkā” (laṅkāprāsādabhañjanam). Hanumān is referred to here by the appellations Jagatprāṇatanūbhava (“Son of the Wind,” equivalent to Vāyuputra, a similar name in more common use) and Prāṇavāyusvarūpa (One Whose Form Is Vital Wind), both suggesting the qualities of pervasiveness, invasiveness, and penetration, as well as, more subtly, indicating the importance of prāṇa in “tantric” yogic practice. Hanumān is also one of the two archetypal devotees in Hinduism, the other being Prahlāda, the son of the demon king Hiraṇykaśipu, who maintained his devotion to Viṣṇu against all odds and successfully invoked him to appear personally and defeat his own father. This he did in the form of Narasiṃha, Viṣṇu’s Man-Lion avatara, emerging from a pillar to eviscerate Hiraṇyakaśipu. It is no accident that the healing and exorcism center at Bālājī, in Rajasthan, is a Hanumān temple pilgrimage site and that Prahlāda is also represented there as a deity, which in my experience is unique. Thus, it is consistent with recognized practice that notions of ethereality, devotion, incarnation, and violence, all attendant during possession, are brought out in these manuscripts.
The procedures for employing children for divinatory purposes in India were not limited to a shadowy presence in a few obscure, regionally specific texts. This is similar to the situation in China, where it was relatively widespread, as Davis’s extensive documentation shows. However, even if the practice was not as widespread in India as in China, corroborative evidence from other Tantras shows that it spread beyond the confines of a few local cults. At any rate, knowledge of it appears to have entered into more mainstream Tantra sastra.125 The Mantramahodadhi (MM), a popular and diffuse compendium of both dakṣiṇācāra and vāmācāra tantric practice composed in the sixteenth century by Mahīdhara,126 includes several verses on this practice or on an adaptation of it.
First, the sādhaka should properly worship the goddess (specific goddesses are not named here), thoroughly divinizing both himself and the maṇḍalas or yantras that he is using to replicate the goddess. This he should do by observing certain vows, divinizing his body through performance of nyāsa and enlivening the maṇḍala or yantra with other mantras, some of them complex and extremely long, totaling hundreds of syllables. He should then “mount a corpse either at a crossroad or a cremation ground [smasāna], abandon shame and fear, and keep his mind concentrated on the mantra (vidyā).127 Then, in the middle of the night, concentrated on his japa, he should hear these distant words: ‘Become transcendent and obtain complete perfection in all mantras’” (pārago bhava vidyānāṃ sarvāṃ siddhim avāpnuhi).128 The text continues: “He should seat two eight-year-old children born into learned families before him and place his hands on their heads. He should then perform japa of this mantra. Both of these children will then become capable of expounding on Vedānta and Nyāya. One who is curious about this ought to see for himself the wonders of this Vidyā.”129 Although the instructions for this ritual do not mention possession, we must assume that it is implicit, as important components of svasthāvesa or other related practices of oracular possession are evident here.
This description is replete with marks of brahmanical orthodoxy and represents a major transformation compared to the earlier material on child mediums. The evidence from the earlier Tantras as well as the brief and highly specified manuscripts from southern India suggests that the tantric practices on which this description was based are here adapted to “higher” goals of brahmanical orthodoxy. The prasenā is not mentioned, and apparently had a short life, at least for a deity (or even a spirit), and disappeared by the sixteenth century, her shadowy station having been usurped by a Mahāvidyā130 or by Bhairava or Hanumān, aspects of Śiva. Knowledge of the past, present, and future as the goal of this sādhana was replaced by the more brahmanical goal of gaining knowledge of Vedānta and Nyāya. Similarly, the strings of bija mantras employed to invoke the Goddess and so on seem to have disappeared.
Concerning the acquisition of knowledge of Vedānta and Nyāya, it is important to note that these verses occur in a brief subsection (5.83–90) prescribing three tantric rituals to be performed on children, of which this is the third. The first states that after cutting the umbilical cord, the sādhaka (the father?, a hired tāntrika?) should inscribe a mantra for Vācaspati, the Lord of Speech, on the tongue of the newborn child with a sharp blade of dūrva grass that had been dipped in gorocanā, an extrusion from the biliary tract of a cow (either a large gallstone or a bezoar), which is used for tantric and alchemical purposes in India.131 Upon reaching the age of eight, the child will then become proficient in all sastras (sarvaśāstrajñatā). The second ritual prescribes mantras and offerings (balidāna) that enable the child to become a great poet (kavitvakṛt). Neither of these mentions possession; but it is difficult in the tantric culture of these texts to imagine that gorocanā or balidāna by themselves confer extraordinary knowledge on a child. The purpose of these offerings, certainly, was to attract spirits or minor deities to enter the child, either permanently or temporarily, thus enabling the child to recite sastras or compose poetry.
As for the uncharacteristic absence of mantras, the text of the MM is replete with strings of such mantras; indeed, this fifth chapter (taraṇga) is particularly congested with them. Thus, it may have been assumed that the aspirant has practiced certain sādhanas and mastered the complex mantras before settling in the smasāna and reciting this mantra of intent. This is supported by the earlier tantric texts and the South Indian manuscripts, all of which prescribe strings of bija mantras for this purpose. It is likely that Mahīdhara was well aware of the tantric environment and background of this practice. To give him the benefit of the doubt, it may not be so much that for a popular handbook he decided to opt for the sanguine over the sanguinary (there is plenty of sex and violence in this Tantra), but that he was vigilant about not losing the sometimes tenuous connection between tantric and other, more public, brahmanical practice.
The term svasthāvesa is attested in the sixteenth-century tantric compendium Tantrarāja (TR), brought to public notice by Sir John Woodroffe in 1918.132 This text presents a fairly close, if highly condensed, depiction of at least some of the material found in the early tantric materials and the South Indian manuscripts. The TR describes two yantras of the goddess and a ritual (TR 9.61–101), the most important part of which is the act of situating on one of the yantras a young girl who is to be guided into and out of svasthāveśa. The text speaks of the “performance” of svasthāveśa, an admission that should not be overlooked. The relevant verses state, “one should perform svasthāveśa” (9.76–78) on the first of the two yantras, because it is over this that the officiant recites the relevant possession mantras. The girl should then become identified with the deity, after which the officiant should burn a kind of incense called sarjjarasa. When she enters into the possession, the girl is worshipped reverently and with appropriate ceremony. Afterward, the client should ask for whichever objects are desired, upon which the girl will always reply fittingly. After being worshipped suitably, the deity becomes her, and after rites, such as sprinkling water, are performed on the first yantra the officiant causes the girls to disembark from their condition in which they are controlled by others.133 The Tantrarāja does not specify who these “others” are, and the commentary states, rather aimlessly, “piśāca, etc.,” apparently to qualify his previous assertion that it is a devatā.
The TR does not in itself shed any light on the South Indian manuscripts. No mantras are given, nor are deities named, whether Hanumān, Bhairava, or a goddess. Moreover, the commentator, Subhagānandanātha, reacts to this rite with an unfamiliarity reminiscent of Bhaṭṭotpala’s in his comments on Varāhamihira’s reference to prasenā in the Bṛhatsaṃhitā, which is to say that he responds as a tantric paṇḍit rather than a tantric ritualist. In any case, this reference is noteworthy by its very presence. It demonstrates that the term svasthāvesa was current, at least in some tantric circles, in the sixteenth century, a period that approximates the composition of the South Indian manuscripts (though the latter could be a century or two later).
To complete the comparative aspect of this topic, an examination of svasthāveśa not only as a term for nonpathological possession but, more specifically, as oracular possession of children, we must offer more examples of this phenomenon in Chinese literature. Only then can the extent to which the Chinese material is derivative of the Indian be determined.
Some compilations of Tang dynasty stories tell of children possessed by the spirits (shen) of learned men. Glen Dudbridge explicates several such accounts from a genre called zhiguai ( tales of the marvelous), one exemplar of which is the eighth-century Guangyi ji ( The Book of Marvels). In Dudbridge’s first story, a girl named Wang Fazhi from the town of Tonglu serves the spirit of a young man called Teng Quanyin, with whom she had an affinity in a previous birth; she begins to experience regular possession of his spirit before the age of five. In frequent meetings with the county magistrate, writers and poets, and Buddhist monks, Quanyin, speaking through Fazhi, demonstrates his literary, scholarly, and religious erudition, composing poetry extemporaneously.134
If the story is to gain any credibility, it must be assumed that the girl was subjected to ritually induced possession. Even if the story is to be treated as a member of another genre—for example, didactic tales, fiction, folklore, or an intermediate genre—we must still ask how, even in the imagination, a five-year-old girl is able to “serve the spirit of a young man.” My sense is that this “service” must have been ritual, and ritually induced, under the supervision of a learned master in this art. Even if the story is regarded as pure fiction, this element of cultural background must be assumed. This inference is based on the presence of frequent descriptions (and tales) of oracular possession weighted toward either their narrative or ritual components, when, in fact, in any “real” or “imagined” event, both are assumed to be equally present.135 In the case at hand, the description of the ritual must have been suppressed in favor of the narrative. As in the Mantramahodadhi, this story does not state or imply that the erudition revealed during possession was maintained by the girl outside the mediumistic act. This too argues for ritually induced possession. Oracular possession is rarely reported as a spontaneous experience; it nearly always adheres to known, effective, and ritually adumbrated models, as mentioned several times earlier. It is, as noted elsewhere, publicly performed, even if that public is very small. In general, it requires an expectant and knowing audience.
Strickmann, in Chinese Magical Medicine, relates an aweishe tale in which the Chinese emperor’s twenty-fifth daughter was at the point of death. The tantric master Vajrabodhi attempted to intercede by sending the souls of two seven-year-old girls from the palace as emissaries to Yama, in a bid to save her. The strategy was to coerce them to deliver to Yama an edict that was written for this purpose but was then ritually burned while Vajrabodhi recited an incantation over it. As a result of this ritual, the girls, whose faces Vajrabodhi wrapped in red silk, after which they lay on the ground, were able to recite the edict perfectly. “Vajrabodhi then entered into samādhi. With inconceivable force he sent the girls to King Yama with the edict.” King Yama relented, but only briefly. He sent the soul of the princess back with the girls, but after half a day she died. In that brief period, she delivered a message to the emperor: “It is very hard to alter destiny as fixed in the other world. King Yama has sent me back to see you only for a short while.”136
As in the previous story, the aweishe ritual is implied more than it is stated. Here at least we are told some of the outline of the ritual: the girls’ heads are wrapped in red silk, they are laid on the ground, an edict is written and burned, Vajrabodhi recites an incantation, he then enters samādhi. This is not prescriptive literature, however, and the narrative element dominates: the spirits of the girls travel to Yama’s world with the burnt essence of the edict and deliver it to Yama.137 The latter, no doubt out of deference to the emperor (as well as to Vajrabodhi), sends the princess’s spirit back with the two girls. In the end, though, it is a temporary measure; once dead, the princess cannot remain alive in the human world for more than an additional half a day. In this story the aweishe empowered the girls to engage in shamanic travel. Patently, this aweishe does not match the primary deployment of āveśa in Sanskrit, where it most often indicates deity or spirit possession instigated by the person being possessed.
In what ways, however, does this differ? It is definitely aweishe, because it is labeled as such in the text. It is also clearly oracular. In addition, the description of the girls in aweishe is reminiscent of that found in tantric texts. They collapse on the ground in a trance as a result of what is, by any other name, ritual initiation, or at least ritual inducement.138 The observed symptoms are consistent with possession. This is, it appears, sufficient to confer upon it the label aweishe. The text does not identify, or even suggest, a possessing entity; yet, for it to be aweishe, the assumption of one must have been present. This suggests a gray area in indigenous identification of possession: Sometimes, for the sake of clarity, we want possession to conform to its simplest notion. But this is not always the case in India or China, where an experience is labeled āveśa or aweishe according to its intensity, rather than to other criteria, such as elision of personality. The girls, young as they are, have had a powerful experience of physical collapse, induced by the weaving of hypnotic spells by a ritual master, with an incumbent interpretation that the spells attract a deity who then possesses the girls and carries the message inscribed on the burned edict to Yama. There is no indication, however, that their personalities have shifted to those of others. We are, then, bound by the word of the text, which calls this aweishe, even if it is not explicit how exactly the spirits or the possession manifest.
Strickmann also notes the use of the word āveśa when exploring the link between dream and divination among children. Although both dream and memory are regarded as unreliable for establishing philosophical truth, because they are regarded as fragmentary, they are to a great extent rehabilitated in religious and historical narrative.139 In Mantras et Mandarins, Strickmann discusses the nexus of dream, divination, and possession, showing their convergence within the realm of Tantra.140 This is an attractive association, and Strickmann works it into a discussion of a ritual performed on children that produces initiatory dreams. In this ritual a mantra is recited 108 times, a mudrā is formed with the fingers, the child falls into a kind of sleep or reverie, possession descends (it is here that Strickmann uses the word āvesa), and the child then will relate matters good and bad.141 Much of this, it appears, is derived from rites of svasthāvesa. Most pertinent for our purposes is the point that all three—dream, divination, and possession—are regarded as authoritative when emanating from children.
This introduces the possibility that any kind of oracular divination, with even the slightest hint of possession, could qualify as āvesa. If so, would this dilute the entire category and render it nearly meaningless? My answer is no, it would not. In the final analysis, the oracular experiences under discussion were āvesa, conceptually even svasthāvesa. This would still be the case even if these experiences lacked the characteristic behavioral shifts accompanied by specifically identified possessing entities. As these texts demonstrate, oracular divinatory practice, especially when enacted through other media, most notably children, is a central feature of healthy (svasthatā), nonpathological possession (āvesa). Indeed, this alone would qualify the practice as svasthāvesa, even if other practices might also qualify.
These stories are sufficient to demonstrate the similarities found in the Indian and Chinese practices of divinatory possession. The evidence is conclusive that much, though by no means all, of the similarity is due to influence from the India and possibly Tibet. The presence of Indian terms and names, from āvesa/aweishe to Mahādeva, render it certain that both the Indian tantric textuality and the transmission of esoteric Buddhist narrative substantially influenced the Chinese practices in question, at least to the extent that their texts reveal these practices (though it is possible that, like Indian texts, the Chinese texts provide us with a normative, sanitized version of them). The results of oracular possession as a primary component in divinatory practice—knowledge of the past, present, and future, displays of miraculous learning, healing secrets and other prognosticatory or revealed knowledge—are attested in many cultures; they are not limited to areas of Sino-Indian cultural contact. With respect to these regions, however, sporadic examples of such oracular or divinatory possession are quite ancient, as we have seen. Recall, for example, the tales of Bhujyu Lāhyāyani in the Bṛhadāraṇyaka Upaniṣad that exhibit important resonances with the material discussed here, though they do not contain the same degree of certainty with respect to retrievable knowledge as asserted in either the Chinese stories or the MM.
Despite these relatively parallel attestations, it nevertheless appears that the Chinese ritual texts address therapeutic or exorcistic practice to a greater degree, and with more subtle distinctions and nuances, than does the Indic material. For example, Davis writes of this practice in China: “The style of these exorcisms appears to reflect many things at once—the late Tang rituals of āveśa, the Daoist rites of kaozhao [kaozhao fa, “Rite of Summoning for Investigation”], and especially the characteristic practices of village spirit-mediums in the Song.”142 The Indic texts domesticate and Sanskritize practices that appear to be derived from village spirit-mediums, women, or others of lower social rank. The problem with the Indian material, to state the obvious, is that the textuality, particularly in Sanskrit, reflects the interests of the literate brahmanical and ruling classes, whereas in China, at least during the Song (960–1279) and Ming (1368–1644) dynasties, the textuality was more likely to include direct accounts and interests of a greater range of social classes. We know, for example, from modern ethnographic accounts in India (including my own observations) that a strong and widespread tradition of therapeutic possession exists that is startlingly similar to the “Rite of Summoning for Investigation,” though I have no evidence that the Chinese practice influenced the Indian one or vice versa. In India, this practice is found largely among non-brahman communities that are underrepresented in Sanskrit accounts of exorcistic practice.143 In spite of these sociological and anthropological resonances, the Chinese and Indian practices appear at present to be wholly independent developments.
In lieu of decisive conclusions on this vast material, I instead pose a question: Why was possession induced in prepubescent youths? I cannot produce a confident reply; the answers lie in addressing both the nature of childhood and its various Indian and Chinese cultural constructions; in other words, both nature and culture. My initial impulse is to tread lightly on questions of culturally constituted psychology and the applicability of psychoanalytic theory, an impulse that should be evident from my earlier methodological explorations. I believe that, in spite of the efforts of Kakar and others, these areas have been insufficiently explored and I am particularly unqualified to offer a mature opinion on them.
Answers to this question, as to so many others, risk being naively obvious or incomprehensibly complex. My first answer is that youths may have been used because they were regarded as pure, as embodiments of moral neutrality, and because youth itself was regarded as a natural restorative. The single scrap of supporting evidence for this is found in Kalkin Puṇḍarīka’s Vimalaprabhā on the Laghukālacakratantra, which specifies that a virgin’s (kumārikā) success in this ritual, which enables her to predict events of the past, present, and future based on visions seen in an oracular mirror (pratisenādarśe), is due only in part to the grace of the guru or presiding ācārya. Equally important is the fact that she has not yet experienced sexual union. Puṇḍarīka rejects the view that it is the acarya’s grace (ācāryaprasādaḥ) alone that causes the virgin to be empowered by the deity of the mantra. He suggests, instead, that if the ācārya has the ability to empower the girl, he ought to be able to empower himself as well, thus gaining the ability to answer questions as an oracle. But this does not occur, notes Puṇḍarīka: The ācārya is not able to generate the visions that produce in himself oracular skill. In fact, says Puṇḍarīka, the empowerment is only effective in a virgin who “has not experienced the pleasure of ejaculation that comes about through the friction of the two [sex] organs [dvīndriya].”144 Pressing this further, one can say that this explanation could also be associated with notions of undiluted and unambiguous sexual and gender identity, which would become too complicated to withstand the temptations of this divination after sexual maturity and activity have been achieved.
A further indication of the vulnerability of children to possession may be seen in the evidence amassed in Chapter 12, as well as in Chapter 6, to possession of children, even fetuses, by destructive spirits called “childsnatchers” (bālagraha). It is surely the purity, innocence, and emotional unguardedness of children that was believed to make them easy targets for such spirits. It is likely that these same qualities made them attractive to ritualists who administered induced childhood possession. This, however, differed from possession by bālagrahas in that it was “positive” or oracular rather than “negative” or disease-producing and was characterized by the guided and protective presence of the experienced ritualist.
A second suggestion is that the mantrin must be only that; he can officiate, but he has neither the innate ability nor the extrinsic eligibility to perform the possession. It is essential that he immerse himself fully in his officiating duties and that, conversely, he not open himself up to the risk of losing control of the ritual process. Such a loss of control or power would implicate him in loss of control not only over the boy or girl in possession but also over the forces of destiny that are being exposed and manipulated under his priestly control. This is a matter of masculine brahmanical power, of control over personality and identity where complete eclipse or transcendence of self and person is an accepted fact of successful possession. Although answers to certain questions (praśna, pasiṇa, prasena, etc.) might be sought by anyone in a crystal ball as it were, for example, in water, a mirror, or an oil-smeared thumb, a person in an inferior social and intellectual position was preferred because he or she could articulate divinatory or oracular answers under the watchful and practiced eye of the mantrin, who would, according to the conventions of hierarchy, retain the right to censor or reinterpret the words of the oracle if they were to appear immature, wild, or irresponsible.
Svasthāvesa is still performed in India, though the only example of it that I have seen was at the Bylakuppe refugee colony near Mysore by an elderly Tibetan woman, who is highly regarded in the community and uses an “oracular mirror.” The woman, called Ama Ta Bap (Plate 9), a name indicating that she divines by plucking a hair and “reading” it (though I observed no cases in which she used a hair), lives in a small house in a remote corner of this 35,000-acre colony, far from the Gelugpa and Nyingmapa monasteries that dominate the Bylakuppe landscape. Central to her practice is a shiny brass surface with an abstract pattern lightly etched on it into which she stares, which serves as the backdrop of her puja altar. The client sits on a chair in front of her and off to her right as she sits cross-legged on an elevated cushion before her altar. After she hears and acknowledges the question, she makes a few offerings with rice, water, and other items, stares into the brass plating, and answers the questions. In my experience her answers are vividly descriptive of visions she has just experienced. This, of course, does not in itself imply possession, and she did not provide me with that level of detail. Although the nexus of possession and reported visionary experience is well attested, it is also sometimes denied.145 I cannot, therefore, say with certainty that Ama Ta Bap believed that she was possessed or that the question arose for her clients; I can say only that the setting and proceedings resemble classical accounts in which possession is a possibility.146
The likelihood that her oracular trances are the result of possession, however, is increased by Diemberger’s ethnographies of female oracles in Tibet. The usual practice among Tibetan oracles is to undergo an initiation or empowerment in which certain “energy-channels” (rtsa) are opened. This psychic opening permits the individual to become possessed by any number of minor deities or spirits. Diemberger writes,
The popular perception is that impurities in the energy-channel are responsible for aberrant behaviour. Once these are ritually purified, possession is considered to be under control and confers upon the oracle an extraordinary competence in helping the other living beings.147
With the deity or spirit in control, the oracle then resorts to mirror divination, which “allows the gods to express themselves, with the oracle acting as an intermediary who can also provide the relevant explanation while in a lucid state.”148
I have heard reliable reports that a few monks in this Tibetan colony perform similar divination, confirming it as a respected practice among diaspora Tibetans. The textual evidence noted herein similarly confirms that it was practiced in India in the past. Much of this history is encapsulated in a Mughal-era painting (Plate 10), a late sixteenth-century projection of the use of an oracular mirror by Alexander the Great. The painting, dating to the year 1597 and ascribed to a Hindu artist named Dharmadāsa in the court of Akbar, was a visual interpretation of part of a long Persian poem by Amir Khusraw of Delhi, called “Ā’Īnah-i Sikandarī” (Mirror of Alexander), composed in the year 1299.149 What is remarkable for our purposes is that this painting by a Hindu in a late sixteenth-century Muslim court, an interpretation of a Persian poem of three centuries earlier, portrays the use of an oracular mirror by a legendary Greek conqueror of fifteen centuries earlier. In this way, the life of magical or oracular mirrors is stretched seamlessly through thousands of years and numerous cultural, religious, and political boundaries, without apparent discrimination.150
A few additional examples should shed a bit of light on a few variants of the practice of childhood possession, if not exactly on its diffusion. One is the divinatory process employed to discover a tulku (sprul-sku) or reincarnation of certain recently deceased lamas. This process often has been described and needs no recounting here,151 but Orofino writes that the “pra divination has been, over the centuries, one of the arts used to foresee the place where the Dalai Lamas would be born.”152 In Taiwan, child mediums, who are required to be illiterate, may still be found. They are called jitong (divination lad), a term that implies that they are both male and young. However, some jitong are older, and others are girls. A practice that appears to have evolved from this is still observed. Certain adult mediums in Taiwan wear bibs designed for children in their oracular practice. This appears to be commemorial, a relic from earlier times when children acted as mediums.153 This mild attempt at disguise or impersonation was probably intended both to incite the medium to assume a childlike state and to inspire the local deity to possess the medium. An explanation for the use of children as mediums in China, which conforms with the likely intent of this practice, is that their hun (soul) is less stable and therefore more displaceable.154 This notion is not found in South Asia, though it may be implied by the very fact of the use of child mediums. That said, however, we must enquire still further into the texts. Unfortunately, I have found no Indic text that addresses child mediumship or reflects on it in this manner. As expected, the texts are an amalgam of myth and prescription, mostly the latter. Much of the prescriptive aspect, as also should be expected, implicates paradigmatic figures. Perhaps primary among these is Hanumān, a widely invoked exemplar, who, we have seen, is the archetypal devotee and “Son of the Wind,” a name that evokes his ethereal ancestry. He is also celebrated for his strength; he is the patron deity of wrestlers.155 This is not unimportant, because strength is a natural attribute and symbol for victory—and possession is always an ordeal of sorts, in both its performance and its aspirations. Furthermore, Hanumān is forever youthful. His personality is opposite that of other monkeys: He is celibate and therefore retains one of the characteristics of youth. He is also the embodiment of devotion, a requisite of the teacher-disciple relationship that is close to the heart of svasthāvesa, as it is to the power of spirit healing at Bālājī. His youth, strength, and virility, channeled into his master’s service, render him the embodiment of truthfulness and dharma. Thus, he is characterized as a dharmic and therefore (marginally) Sanskritic, an exemplar for those who require such a figure to legitimize their possession practices.
There is no question that children are held in high regard in India, as potential bearers of crucial esoteric knowledge.156 In both South and East Asia, it appears that their supposedly natural talents for innocence and purity, as well as an apparent moral neutrality that distinguishes them from adults, rendered them more trustworthy in delivering unbiased answers to questions of weighty and biased human (which is to say, adult) concerns. This was even more important when considering the ambivalent position of the mantrin. On this ambivalence, I note Stanley Tambiah’s observation of Thai Buddhist exorcists as “both a caricature and an inversion of the orthodox Buddhist monk. He uses Buddhist sacred words for purposes diametrically opposed to those of the monk; the latter chants sacred words in order to teach morality and to transfer merit and blessings, whereas the exorcist uses the sacred words to frighten spirits and drive them away.”157 The mantrin (Mandarin) or māntrika in tantric ritual was similarly afflicted: He was often an adept, but not always trusted or regarded as a moral exemplar by the general public, even by his own clientele. He therefore had to turn to his opposite to bring about the goals of tantric divinatory ritual.
In conclusion, we have seen here wide-ranging data demonstrating that the practice of svasthāvesa, regardless of whether the practice was identified by that name, had a broad distribution in South and East Asia. It appears to have begun in northern India among tāntrikas (whether or not identified by that name) affiliated with Buddhist, Hindu, or even Jain lineages, who adapted and textualized practices that predated them (cf. the citation from Susruta). It then spread into Tibet and China through transmission of tantric lineages beginning in the seventh century. It is impossible, in general, to know for certain how the geographic distribution of a particular practice coincides with the distribution of its textuality, especially if that textuality is sporadic, geographically widespread, and stretched over more than a millennium. This is especially the case where literacy is low and popular practices are legitimated (in this case, through Sanskritization) on an ad hoc basis, and where the practice in question has all the earmarks of a nonexclusive, non-brahmanical production. What first surfaced in north India in the fifth to seventh centuries as a variety of praśna or pasiṇa became prasenā farther north and east a few centuries later. Then it appeared half a millennium later in tantric texts and an assortment of prayogas in South India that strikingly resemble the Chinese texts. The description of the ritual, or at least clear references to it, most prominently its use of children in oracular possession, found its way at various points into Indian and Chinese narrative literature (cf. Kathāsaritsāgara and Guangyi zhi). It is a very colorful history and one that raises significant question about notions of personal agency and epistemology, the nature of childhood in India and China, and the transmission of ideas, practices, and textuality across lineage, class, and geographic boundaries.
NOTES
1. I am particular grateful to Alexis Sanderson, who brought the unpublished Śaiva texts to my attention, shared with me his copious transcriptions and notes on them, and eventually went through a draft of this chapter, offering many positive suggestions, some of which are worthy of specific attribution. I am also indebted to E. Muralidhara Rao and H. V. Nagaraja Rao for help in transcribing a number of Sanskrit manuscripts from Telugu script into Nāgarī; to the participants of a conference on Tantra and Daoism held at Boston University in April 2002, especially John Lagerwey, David Germano, and Ronald Davidson; and to Wendi Adamek for explicating the Chinese and patiently answering many questions that must have seemed basic to her.
2. Subhagānandanātha’s commentary, called Manoramā, on the Tantrarāja Tantra 9.76, where he glosses svasthāveśaḥ as svasthasya devatāvesanam.
3. Suspect because it is limited neither to Kashmir nor to the worship of Śiva. Even in its most generous specifications, the term “Kashmir Saivism” could never include the works we consider here. “Kashmir Saivism,” instead, includes texts of the Trika, Krama, or Bhairava traditions. On the place of each of them in Śaiva sects and lineages, see Sanderson 1988. See also p. 406 n. 2.
4. Sanderson has exploited one or more of these texts in virtually all his published work. The manuscripts that Sanderson used, parts of which he has shared with me, are the following: Jayadrathayāmala National Archives, Kathmandu (NAK) 5–4650; Nisvasaguhya, NAK 1–277; Tantrasadbhāva, NAK 5–1985, and the Nepal-German Manuscript Preservation Project (NGMPP) A188/22; Bṛhatkālottara, NAK 1–273. In Sanderson’s estimation, the provenance of the two Buddhist texts among them, the Cakrasaṃvarapiṇḍārtha and the Sekoddesa, is probably eastern India.
5. The Tibetan translation appears in full, while only the first chapter of the Sanskrit text has been thus far edited; cf. Orofino 1994b.
6. Sanderson has been utilizing manuscripts of these texts in his scholarship for two decades; see, e.g., 1986, 2001.
7. Davis 2001; Strickmann 2002.
8. The NCC is not yet complete. Eleven volumes have been published to date, and it appears to be about half completed. Thus, more relevant manuscripts may yet be located. However, I have also checked the more likely catalogues from which the NCC is drawn and have come up empty.
9. Rajagopalachar 1990:1:50–53, catalogue numbers 48062–48069. Extracts from three of these appear in the appendix (which has separate pagination), pp. 57–59. #48063 (4868/22), in Telugu script, describes possession of Avesabhairava (asya śrī āveśābhairava mahāmantrasya agastya ṛṣiḥ). It states, however, that the devatā of the ritual is named Svasthāveśabhairavaḥ (Bhairava [viz. Vīrabhadra] who induces svasthāvesa). #48067 (10024/12), one of the brief palm leaf Nagari manuscripts, appears to be no more than mantras, as the title indicates (Āveśamantraḥ). I have examined #48062 (548/7) and #48069 (548/3), both in Telugu script. Although these differ significantly from the Chennai manuscripts, and probably come from different (though related) lineages, they do not add much to our knowledge of the procedures for svasthāvesa. However, #48067 (10024/12) has an unusually complete colophon: iti śrī āgamarahasye atharvaṇatantrakhaṇḍe umāmaheśvarasaṃvāde āveśamantrayantrapūjādividhir nāma prathamaḥ paṭalaḥ | (Thus ends the first paṭala called “Instructions on mantra, yantra, and puja to bring about possession,” appearing in a discussion between Umā and Maheśvara in the section on the Tantra of the Atharvaṇas in the illustrious Āgamarahasya). However, I do not find this in the published edition of the Agamarahasya (see Chapter 10, n. 5). This title, meaning “Secret of the Sacred Doctrine,” could well be generic; thus, more than one text of this name could exist in manuscript.
10. On the range and local production of such ephemera, see Diehl 1956:42–65; equally important on the origins, history, contents, production, and distribution of chapbooks is Hanaway and Nasir 1996.
11. The term dakṣiṇācāra is to be distinguished from the directional āmnaya transmission known as the dakṣiṇāmnāya. See Sanderson’s guide to medieval Śaiva lineages, 1988:681–690ff., for a description of the directional āmnāya transmissions. For a discussion of Sākta Tantrism under the influence of the Śaṇkarācāryas, see Brooks 1992:43ff.; also Kavirāj 1963. For an account of Srividyā practice, see Khanna 1986.
12. On the proximity of the Sanskrit vedic ritual traditions and both brahmanical and nonbrahmanical possession in Andhra Pradesh, see the various articles by Knipe, esp. 2004, which compares the two traditions.
13. ORI ms. #548 (see accession numbers 48062, 48069).
14. This manuscript appears to be only about 150 years old, but is in very poor condition, crumbling to dust upon mere touch. I am not familiar with the Dakṣiṇakālikāgama or the Vetālatantra.
15. Rajagopalachar 1990: 2:276–277.
16. I have not yet examined the manuscripts at the latter two institutions.
17. Cf. Poleman 1938:#4658, 14 ff. 9.5 x 4.1. 12–13 lines; University of Pennsylvania manuscript #578.
18. As modern (rather than nineteenth-century) Sanskritists, we are hesitant to comment too negatively on the language and diction of our materials. Thus it is striking (and encouraging) to see Sanderson refer to a text written in “barbarous Sanskrit” (referring to the Yonigahvara); cf. 2001:15n14.
19. iti sudarśanasaṃhitāyām uttarakhaṇḍe śrīkārttavīryārjunadīpakalpakathanaṃ nāma dvādaśapaṭalaḥ I am in the process of preparing an edition of this text, which appears to deal primarily with corasāstra, the art of thievery, as the presence of the semidivine character Kārtavīryārjuna might indicate. One text of this peculiar sastra, called Śaṇmukhakalpa, was edited by George (1991), published in 1966 and revised twenty-five years later.
20. A quick perusal of the Sudarśamasaṃhitā manuscripts in Mysore reveals that most of it consists of rituals (vidhāna) for Hanumān.
21. Ms. #51670. The colophon reads: iti śrīsudarśanasaṃhitāyāṃ hanumaddīpavidhir nāma viṃśatiḥ paṭalaḥ. The section dealing directly with possession of Hanumān, which we address below, is the third paṭala, called Āñjaneyāveśavidhiḥ.
22. Manuscript number D 7763: iti śrīsudarśanasaṃhitāyām ānjaneya āveśavidhir nāma tṛtīyo paṭalaḥ || (Āñjaneya is another name of Hanumān).
23. Thanks for this information are due to Philip Lutgendorf, who sees Hanumān as occasionally shamanic. Cf. Lutgendorf 1991:48 on Hanumān’s “crude strength and occasional destructiveness,” attributes which are sometimes invoked in possession. The pañcamukhī Hanumān has Hanumān in the center, flanked on its four sides by Narasiṃha, Varāha, Garuḍa, and Hayagrīva; cf. Lutgendorf 1994. Although this is Vaiṣṇava imagery, Hanumān has also been associated with Śiva since antiquity, as one of the eleven Rudras. I have seen many large and very recent images of this form of Hanumān in Andhra Pradesh, where we also find a good deal of possession of Hanumān.
24. We investigate this in Chapter 12. It is sufficient for now to note that one of the most prominent appearances of this is Īśānaśivagurudevapaddhati" 41–44. The character of these chapters as mantravidhana is unquestioned. Tantric healing practices derivative from the ISP survive in Kerala, though in Malayalam medium rather than Sanskrit, in both the aṣṭavaidya tradition of the Nambudiris and other less brahman-dominated Ayurvedic traditions. On the former, still insufficiently researched, see Zimmermann 1987:213.
25. Cf. Bahulkar 1994, Bhat 1987, and Caland 1908.
26. His monumental Bṛhatsaṃhitā served as the basis for all later works on the subject. After the Atharvaveda, the medical literature, where we might expect to find systematic accounts of omens and portents, turns self-consciously scientific and rational; see Roçu 1986. On the dating of Varāhamihīra, see Pingree 1994:570.
27. Especially in the semantics of samāveśa; see Chapter 10.
28. Sanderson 1985 also speaks of this, though rather differently.
29. See the discussions above on the MBh and the Bhagavadajjukā.
30. The voluntary or involuntary status of possession during public or semipublic ritual is often problematic. Participants in festival possession often report that their possession is spontaneous and involuntary. In these situations, however, it is almost always learned, suggested, and socially advantageous. For one such report, see Neff 1995:194 ff.
31. See Chapter 4, which discusses words and terms used for negative possession in Sanskrit and different regional languages.
32. On the antiquity of cultural traffic across the Himalayas, dating to the third millennium B.C.E., see Xu 1991. Xu’s research is archaeological and focuses on Karuo, near Qamdo on the eastern edge of the Tibetan plateau. The physical evidence reveals considerable cultural affinity between Burzahom in the Kashmir valley and Karuo. Xu speculates on routes across Tibet by tracking the evidence from Neolithic cultures of the period. See also Samuel 2000. While Samuel is not able to comfortably posit “significant influence from the Indus Valley cultural tradition” (p. 653) into Tibet from the mid-third to mid-second millennium B.C.E., he points out enough evidence for early contact “to suggest that the Indus Valley should be taken seriously as a possible source for early Tibetan cultures, and that its neglect in the literature so far is unjustified” (p. 666). He speculates that perhaps “a reflection of the Indus Valley civilization” may be found “in the accounts of the semi-mythical Bonpo homeland ’Ol-mo lung-ring, the place of the original teacher of Bon, ston-pa gShen-rab” (ibid.); cf. Martin 1995.
33. This includes Sanskrit-English, Sanskrit-French, Sanskrit-German, Sanskrit-Hindi, and Sanskrit-Sanskrit.
34. This curious lacuna includes the massive scriptorium of the Deccan College Sanskrit dictionary project, which, like the other lexicons, lists it only as the name of various royal personages. This should reflect not on the choice of texts selected for use in compiling this dictionary but on the difficulties incumbent on any attempt to compile a complete Sanskrit lexicon. The earliest reference to this name is in the Maitrāyaṇi Saṃhitā (3.1.9) and appears often in the Harivaṃśa, sporadically in the earlier Purāṇas, and once in the Jaina Mahāpurāṇa of the ninth century (3.150). The best known of these, however, is Prasenajit [Conqueror of Prasena], a king of Srāvasti (in the northeastern part of present-day Uttar Pradesh) in the time of the Buddha, who was converted to Buddhism by his wife, Mallikā; cf. Hirakawa 1990:33.
35. E-mail communication, September 13, 2001.
36. darpaṇa ādi meṃ devatā kā āhvān, mantravidyāviśeṣ |
37. The PSM also cites the compound pasiṇavijjā, viz. Skt. praśnavidyā, from the same texts. I am unable to find these terms in the texts; thus, it is likely that they are from the commentaries. The commentary on the Āyaraṇgasutta was written by Śīlāṇka in 876, a date that is more consistent with the remainder of the evidence. It is important that, among the sixty-four baneful vidyas listed in Sūtrakṛtāṇga Sūtra 2.2.13(26–27), a text that probably predates almost entirely the period that concerns us here, does not list pasiṇa, etc. See Jacob 1895:xl for cautionary words on dating this text, and 366f. for a translation of this passage. It may be mentioned here that the Sanskrit word vyantara, a general term for all manner of spirit-beings, is also attested much more in Jaina than in brahmanical texts.
38. Cf. Jain 1999:8; more detailed is Sen 1975:6–9.
39. Cf. Sen 1975:317. This is also mentioned by Jain 1984:264–65. See also Nandi 1994:395 and passim for an account of a detailed Jain tantric text called Jvālinī Kalpa. This text remains unpublished, and exists in only one manuscript housed in the Jaina Siddhanta Bhavan Library, in Arrah, Bihar. I have not examined it.
40. mantravidyā ke bal se svapan ādi meṃ devatā he āhvān dvārā jānā huā subhāsubh phal kā kathan |
41. Niśītha Cūrṇi 3, p. 383: suviṇayavijja kahiyaṃ kathiṃtassa pasiṇāpasiṇaṃ bhavati; cf. Sen 1975:317.
42. Bṛhatsaṃhitā (BṛS) 2.5, which reads in part: tātkālikapraśnaśubhāśubhanimittāni, predicting “good and bad omens upon timely inquiry.” This section of BṛS addresses the qualities required of an astrologer, much of which concerns the ability to answer questions. I refer here to the 1895 edition of the BṛS, ed. by Sudhākara Dvivedī. Other works attributed to Varāhamihira whose titles suggest a prior history of this term are the Prasnamahodadhi and Prasnacandrikā. These, however, are lost. BṛS 50.6 gives the best direction and time for prasna. Thus, it is likely that the astrological art of prasna was known to Varāhamihira.
43. I have encountered no instances of the Sanskrit prasna/prasnā in this sense.
44. kuhakāveśapihitaiḥ karṇopaśrutihetubhiḥ | kṛtādeśo na sarvatra praṣṭavyo na sa daivavit ||
45. This is something of a reversal, as āveśa is usually interpreted as a deity entering the body of the sādhaka or practitioner. Utpala, it appears, had little knowledge of these practices.
46. The Vidhānamālā, p. 26, defines karṇagraha and prescribes a brief exorcism. One who hears the voice of a relative, one’s guru or guru’s wife, or hissing sounds (ciṭciṭākāram) is possessed by a karṇagraha (taṃ karṇagraho gṛhṇāti). The exorcism is quite simple. Both pure offerings of fruit, flowers, sesame seed paste, etc., and impure offerings consisting of the five, liquor, meat, etc., are offered as bali at a crossroad at dawn, and a simple petitionary mantra is recited: gṛhāṇemaṃ baliṃ deva karṇāriṣṭakara graha | āturasya sukhaṃ siddhiṃ prayaccha tvaṃ mahābala ||. See the Madamamahārṇava section in Chapter 12 for more on this kind of exorcism.
47. See Kalkin Puṇḍarīka on Laghukālacakratantra, vol. 3, p. 89, that a karṇapiśācikā can relate knowledge gained in astrology and other mathematically based divination. I have heard certain tāntrikas in present-day India employ the term karṇapiśāca in this same sense.
48. On Bhaṭṭotpala and his works, see Bhat 1981:1:xxxvi–xlv; also Shastri 1962, who argues that Bhaṭṭotpala completed his commentary in 830 C.E. Bhat argues for a much later date, 966 C.E. Although Bhat knows the Vivṛti well, he does not follow it in many instances where he should. In addition, Bhat never states which edition of the Sanskrit text he is following; indeed, he gives no critical apparatus whatsoever. Thus, for example, in his translation our present verse is numbered 2.33, without explanation.
49. Strickmann incorrectly cites the use of the word prasena by attributing it to Utpala’s tenth-century commentary on Varāhamihira’s Bṛhajjātaka. In fact, it does not occur in this text, which is dedicated to the formalities of predictive astrology. Strickmann cites Böhtlingk’s 1879 revisions and additions to the earlier St. Petersburg lexicon of Böhtlingk and Roth (1855–1875) (Böhtlingk has the citation right), which glosses Utpala’s meaning as “eine Art Gauklerei.” Strickmann adds, “c’est–à-dire une sorte de prestidigitation, d’illusionisme, de charlatanisme” (1996:226). Strickmann’s error reappears in his posthumously published Chinese Magical Medicine (2002:215). In his Sanskrit-English dictionary, Monier-Williams (1899:698) defines prasenā as “a kind of jugglery,” citing the “scholiast” (surely Utpala) on Varahamihira’s BṛS (this is in addition to references to several kings named Prasena).
50. This is borne out in the course of the next several centuries, as astrology becomes an indispensable element in priestly education. On the history of this movement, see Inden 1992.
51. I maintain this position, though I also agree with Sanderson (e-mail communication, September 27, 2001) that it was a literary and stylistic issue. The latter, in my view, are at least partly an outcome of these epistemological exigencies. Sanderson defends the Indian mode of presentation as follows:
The Indian sources give us straightforward prescriptions of kanyāvesa and similar rites involving children. These naturally tend to be a little thin on the incidentals that would animate rich descriptions, but I see no evidence that they are epistemologically inhibited. A fairly detailed “Sanskritic” prescription of the prasena ritual is seen in, for example, the Indian Buddhist Subāhuparipṛcchāsūtra. A copy of this text was in the possession of the Chinese monk Wuxing in 674. and it was translated into Chinese by Śubhākarasiṃha in 726 (T. 895), and by Fatian (T. 896) in the late tenth century. It is listed with other Tantras of the Kriyātantra class in the Tibetan Ldan dkar library inventory of the early 9th century. The fact that the ritual is mentioned only sporadically in Sanskrit texts is not surprising. This is true for most kinds of special Tantric ritual, since the Tantras are by convention overwhelmingly concentrated on setting out the obligatory rituals (initiation, regular post-initiatory and pratiṣṭhā) of this or that Mantra system and, incidentally, on the kāmya inflections of the regular rather than on such special performances.
I deal with the Subāhuparipṛcchāsūtra shortly. Sanderson is doubtless correct about the “convention” of the Tantras in addressing certain kinds of ritual. Nevertheless, these accounts rarely exhibit the narrative intimacy of the Chinese texts. In this sense, the Subāhuparipṛcchāsūtra could be the exception that proves the rule.
52. On ritual as a learned response, see Chapter 2, n. 100.
53. See Orofino 1994b:11ff, for a lengthy discussion of the place of this text in the Kālacakra textuality.
54. Cf. Newman 1988:133. For the Sekoddeśa passage, see Orofino 1994b:134.
55. Translated from the Tibetan by Gray (2006, in press). The Sanskrit of this chapter is lost. Gray provides a translation of Bhavabhaṭṭa’s commentary on this verse: “Having repeated [the mantra] one hundred and eight [times] over a sword, etc., one will cause the descent of the luminous divinatory image.” The name of the mantra that causes the divinatory image to descend into the object is upahṛdaya.
56. Orofino 1994a. See also Bellezza 2005:23, 68f., 437f., plates 42, 43, and passim, on different kinds of divinatory mirrors and their use by Bön shamans in the largely unexplored and undocumented regions of northern Tibet.
57. Ibid.:614. Orofino suggests (pp. 615–616) that the word pratisenā is an intentional recasting of the word prasenā, thus turning it into a metaphor that conveys the idea of reflection. It would then have been modeled after words with which it is associated that also express reflection, in particular pratibhāsa (see n. 128). I have no alternative proposal.
58. This can occur only if the word from which it is derived is not Sanskrit, and prasenā is Sanskrit.
59. Personal communication, September 13, 2001. Sanderson further notes, “Tantric Saivism of this relatively public and strongly soteriological variety was not merely present in the seventh century but well established” (2001:11).
60. This account is based on a long message sent to me by Sanderson (March 10, 2002) and references he has suggested, as well as on his 2001 article.
61. Cf. Bakker 2000:15f. n42.
62. Cf. Dikshit and Sircar 1955–56; Shastri 1995: 1:169–172, 2:154–159.
63. āsītsadyaḥśivācāryaḥśrīmān varyastapovatāṃ (-tām) |śrīmadāmarddakakhyātitapovanavinirggataḥ ||
64. Dikshit and Sircar state: “Amardaka, which is the name of Kāla-Bhairava, a form of Śiva, was probably derived from the locality where Bhairava was worshipped” (1955–56:34–35).
65. See Sanderson 2001:8–10n6: These kings are the Cālukya Vikramāditya I of Bādāmī, the Eastern Gaṇgā Devendravarman, and the Pallava Narasiṃhavarman of Kāñcī.
66. svasthāveśacūrṇam ivendriyāṇām asantoṣam iva kautukasya; cf. Harṣacarita, ucchvāsa 1, Führer’s edition, pp. 37–38 (I thank Professor Sanderson for this reference); cf. also the Harṣacaritam, ed. by Musalgaonkar and Musalgaonkar, p. 61. Musalgaonkar states in his Hindi vyākhyā, “indriyoṃ ko niścal karanevāle cūrṇ ke sadṛś.” See also the edition by Kane, text p. 10, note p. 196. In this note Kane states, “he was as though a powder that influences the senses to be comfortable. The idea is: in his presence, the senses of the spectators were lulled into pleasure.” Both Kane and the Musalgaonkars, great paṇḍitas and scholars though they were, missed the point here.
Another, later, mention of svasthāveśa is in the Kathāsaritsāgara, dated 1064 C.E.: so ’haṃ kadāpy akaraṇaṃ svasthāveśaṃ prasaṅgataḥ | śubhalakṣaṇam āsādya kaṃcit kṣatrakumārakam || (70.56) (Once I apprehended a certain ksatriya boy with good characteristics and, through perseverance, instigated him to enter a state of svasthāvesa without any apparent cause). The purpose of this svasthāvesa was to enable the instigator of the possession, an ascetic (tapasvin) student of Śuddhakīrti, to find magical herbs and ultimately the subterranean palace of a serpent king (nāgendra-bhavanam) in which he would obtain a magical sword that would enable him to become Lord of Siddhas and roam about unconquered (sa siddhādhipatir bhūtvā vicaraty aparājitaḥ || 70.61d). Recall the reference to “good characteristics” (śubhalakṣaṇaṃ) noted above by Orofino in the Subāhuparipṛcchānāmamantra.
67. I remain cautious about the proliferation of the word Tantra in contexts such as this. Thus, I avoid saying that svasthāvesa “entered into general Tantric currency.”
68. Regarding these texts, Sanderson writes:
The Nisvāsaguhya survives in a ninth-century Nepalese ms., but seems to be one of the very earliest Tantras of the Siddhānta, a tradition for which we have epigraphical evidence from the sixth century onwards. The Tantrasadbhāva is known to Abhinavagupta and there are Nepalese mss. from around the same time. The Bṛhatkālottara, also called the ṣaṭsāhasra in reference to its length, is not cited by the famous abundantly quoting Saiddhāntikas of the period 975–1050. I know of no citation in a work that predates the twelfth century. By then it was a very popular, widely known source. The Jayadrathayāmala, also known as the Tantrarāja, was widely known by the twelfth century. Kṣemarāja quotes from the first of its four ṣaṭkas (sections of six thousand verses) and Abhinavagupta quotes from the Mādhavakula, a work that is part of the fourth. Vimalaprabodha, Rājaguru of King Arimalla of Kathmandu, who ruled from 1200–1216 C.E., knows the whole work. The Buddhist Tantric texts belong to the tenth and eleventh centuries. (personal communication, September 13, 2001)
For more on the character of these texts, see Sanderson 2001:31–32n33. For a brief discussion on the date of the JY, which was cited by Kśemarāja in the first half of the eleventh century, see Sanderson 2002:1–2.
69. kṛtayogāñjaliḥ samyak candracakrodaroditām | cintayet paramāṃ śaktiṃ taraṅgakulavigrahāṃ |
70. Sanderson, personal communication, September 13, 2001.
71. pratisenāvatāratantre kila darpaṇakhaḍgāṅguṣṭhapradīpacandrasūryodakanetreṣv avastuṣu [read: aṣṭasu] pratisenāvatāra uktaḥ; cf. Newman 1988:133.
73. Sanderson has examined the relevant Śaiva texts that have survived in manuscripts. However, given the large number of such works that are cited in the extant texts and commentaries, I would be surprised if these details, and much more, were not mentioned in some of these missing texts. For our present knowledge of the names of these texts, see Sanderson 2001:3–4n1.
74. The syllables ṭha ṭha are code for svāhā. See now the translation by Qvarnstrøm (2000:600), that in part misunderstands this passage. This chapter of the Yogaśāstra is very much in the tradition of omens and portents established by Varāhamihira.
75. SuSaṃ 6.60.19: darpaṇādīn yathāchāyāśītoṣṇaṃ prāṇino yathā | svamaṇiṃ bhāskarasyosrā yathā dehaṃ ca dehadhṛk | viśanti ca na dṛśyante grahās tadvac charīriṇam ||
76. For example, Paper 1995. Though this is not the most reliable source for commentary on such material, Paper does present the evidence; for my views on this book, see F. Smith 2001.
77. Samuel writes of “the multiple misunderstandings that seem inevitably caused by any use of the term ‘shamanic’ in scholarly discourse, however carefully defined” (2000:663).
78. See Chapter 4, in which the exorcistic healers at Bālājī deny any tantric pedigree.
79. Contra Iyanaga 1985:693f., followed by Davis 2001:129. The derivation of this word can hardly be from Sanskrit ud√śiṣ, ucchiṣṭa (rejected [from the mouth]). Ucchuṣma as “the Crackler,” or, better, “the Desiccator,” is unproblematically derived from ud√śuṣ (to dry up). As an aspect of Agni, Ucchuṣma crackles or desiccates as he dries up the offerings presented to him.
82. Snellgrove 1987:134–141, esp. 139. On Vajrapāṇi, Snellgrove remarks, “I confess to finding him by far the most interesting divine being throughout the whole history of Buddhism, for he has a personal history and considerable personal character” (p. 134).
83. Davidson (1991) rightly notes that the functions of myth in Buddhist practice have been overlooked. Moreover, his critique of Stein and Iyanaga must be seriously considered (p. 214ff.).
87. This is described at length in Mayer 1996:115–128.
88. In his essay titled “Born to be Wild: Siddhas, Outcastes, Shabaras, and Local Esoterism,” Davidson counters Sanderson’s argument. He states:
Heruka was developed in imitation of Maheśvara, and is thus the appropriation of Śaiva/Kāpālika systems for Buddhist purposes. Unfortunately, this excessively reifies the potential sources, and when we examine the record on Heruka, it appears much more problematic. While Heruka is formed in imitation of Maheśvara in the myth contained in the mid to late eighth century version of the Sarvatathāgata-tattvasaṃgraha, the 726 C.E. translation of the Subāhupariprcchā contains an apparently earlier reference to Heruka, there depicted as a local demon like a ghost (piśāca). This is in close consonance with the probable origin of Heruka, as stated in the Kālikā-purāṇa, which identifies Heruka as the divinity of a cremation ground (smasana), possibly where human sacrifice was to be performed. The cemetery was close by the Kāmākhya-pīṭha, and Heruka was provided a specific iconography. (p. 9)
Sanderson argues this often, perhaps most appositely providing details on the “iconography of subjection” as a tool employed by both Buddhist and Hindu tantric sects (1994).
89. See Bakker 2000 for Pāsupata evidence.
91. Some of what Mayer proposes does not strike me as right or reasonable within the extended South Asian religious complex. For example, he suggests that “the Buddhist tradition in India would have found the facts of its debts to Saivism embarrassing” (ibid.:305). For more on these Central Asian shamanistic practices, see Bellezza 2005.
92. Mayer 1996: 116. See also Granoff 1979, which employs art historical evidence to date the Buddhist Śiva to the sixth century.
93. To the question addressed in this section, Sanderson responds: “We know only that we do not know for sure that Śaiva sources teaching this practice are earlier than the Buddhist materials transmitted to China. It is only extremely probable that there were such sources for the Buddhists to draw on, and the fact that Bāṇa can refer to svasthāveśa in a literary composition strongly suggests that this was so. Certainly the descriptions of Bhairava tantric practice found in his writing are based on exactly the class of Śaiva sources that is richest in references to prasenāvidhi/svasthāveśa” (e-mail communication, September 27, 2002).
94. Strickmann 2002:208. I employ pinyin romanization unless citing a secondary source that uses the Wade-Giles system. Strickmann points out that āveśa is “sometimes translated into Chinese as ‘total entering’ (pien-ju)” (p. 207). This fits the semantic profile of āveśa as “immersion,” “absorption into,” etc.
95. Strickmann 2002: 204f., citing T. 1097, Bukong juansuo tuoluoni Zizaiwang zhou jing, the Chinese translation of Amoghapāśasūtra.
97. Ibid.:204–205. Strickmann translates Amoghapāsa, “He whose Noose is Unerring,” the name given to the twelve-armed form of Avalokitesvara, with the melodramatic “Slipknot That Never Misses Its Target” (p. 326n17). The Sanskrit text of the Amoghapāśasūtra is only now in the course of being edited and published. This is in the form of a roman transcription of a codex unicus that has come to light in Tibet; cf. Kimura et al. So far, a bit more than one-third of the text is published (66 folios out of 162). However, the section translated by Strickmann has yet to appear, so we eagerly await its publication. Before this work, the only description of the text, indeed what is probably a mere fragment of it, has been Meisezahl 1962. For an interesting reference to the healing practices described in this text, see Granoff 1998b: 218n1. This work includes two edited Tibetan versions, one from Dunhuang, the other a later Kanjur version, and an edition of a fragmentary Nepalese palm leaf manuscript in Sanskrit (University Library Cambridge, Oriental Ms. 152). This dhāraṇi text sheds no light on the material dealt with here, except that many unusual words in it overlap with what we have found, appositely, kākhorda (pp. 291, 315), a kind of evil spirit (cf. Edgerton 1953:2:175, who cites this as an Iranian loanword). See also entry 3826, khakhōrda, in Turner 1969:201. Sanderson has collected a large number of attestations of this word not noted in these dictionaries. He informs me that this word “occurs widely in Mahāyāna texts and in the Rājataraṅgiṇī and in Śaiva texts that may well have been composed or redacted in Kashmir, in whose vernacular alone it has survived through to NIA” (e-mail communication, October 10, 2002). One interesting usage occurs in the Netratantra 19.132, where khārkhoda- appears as a kind of malevolent being, though Kṣemarāja, the commentator, glosses it khārkhodāḥ paraprayuktā yantrāḥ (yantras that may be used against others).
98. Cf. Strickmann 1996:226f., and in greater detail in 2002:211–218.
99. Strickmann 1996:227. The Sanskritic philological speculations in which he indulges in his long note 43 on pp. 327–328 of Chinese Magical Medicine should be summarily discarded. Causing spirits to descend is otherwise known in Tibet and South Asia. Cf. Allen 1986,Aziz 1976,Macdonald 1975, and other references mentioned in Chapter 4. The notion of the spirit or jīva of a teacher passing into the body of the lineage successor is also found in Sikhism, where it became an article of belief that the spirit of one guru passed to his successor “as one lamp lights another.” This notion gained acceptance not just in the Tibetan institution of tulkus but by the convention adopted in India by the ten Sikh gurus, who all used the signature of the first one, “Nānak,” in their compositions.
100. See Nakamura 1975:2:1092c.
101. The sole support I could find for this in Sanskrit literature—and it is vague, at best—occurs in the Hastyāyurveda, a text on the care of elephants: svapnaśīlaś ca bhavati mandakram aviceṣṭitaḥ | prasenaṃ kurute kāmaṃ gambhīraṃ vedayaty api || (2.61, p. 306). This verse occurs in a section on minor diseases (kṣudraroga-), in this case a ślaiṣmika or kapha-based condition that slows elephants down but nevertheless leaves them deeply lustful. The word prasenam is by no means clear, but it could refer to a formation of elephants. Neither Strickmann nor Davis suggest the possibility that the name Prasenajit here could refer to the hermit scholar of that name in eastern India with whom the Chinese Buddhist scholar and traveler Xuanzang reputedly studied at the end of his sojourn in India (he began his return journey to China in 643). On this Prasenajit, see Lusthaus 2002:408–412.
102. “faire déscendre le pra.”
103. Cf. Davis 2001:280f. n27 lists both Chinese and Japanese āveśa texts, as well as secondary-source references, many of them in Japanese. For the Hōbōgirin, see Lévi et al. 1929–83. On the āvesa ritual in esoteric Buddhism in Japan, a descendant of that practiced in China and Tibet, see McCullough 1973.
104. Davis 2001:124–25. I have left out Davis’s many notes to this paragraph, but see pp. 280–282nn17–27 for identification of āveśa texts and comparisons with the cognate rite in Japan.
110. Davis takes this description directly from a Chinese Buddhist text (ibid.: 128).
112. The names of the Chinese works cited by Strickmann and Davis that contain sections on aweishe reveal culturally sinified Buddhist titles: e.g., Buding shizhe tuoluoni binifa (Secret Rites of the Spells of the Divine Emissary, the Immovable One, T. 1202), in which the Immovable one is the tantric Buddhist deity Acala; Jingangfeng louge yiqie youzhi jing (the Yogin’s Book of All the Yogas of the Diamond-Pinnacle Pavilion,, T. 867), translated by Vajrabodhi (662–732 C.E.); Yaoshi rulai guanxing yigui fa (The Medicine Buddha Contemplation Ritual, T. 923), also translated by Vajrabodhi. Strickmann (2002:218–227) discusses treatments of negative, disease-producing possession, viz. bālagrahas, that reveal Indic influence from āyurvedic sources (e.g., Caraka 7.5, Suśruta ch. 60, see Chapter 12 for explication of these sections) or even the MBh (3.216–219). The Chinese works are: Hu zhu tongzi tuoluoni jing, or Tongzi jing (Book of the Dhāraṇi for Protecting Children, T. 1028a), translated into Chinese during the first half of the sixth century, and the Luofonu shuo jiuliao xiao’er jibing jing (Book of Rāvaṇa’s Explanations of How to Cure the Ailments of Children, T. 1330). On the latter, see Filliozat 1935, 1937. For a medically detailed account of bālagrahas, see the section from the Kaśyapasaṃhitā translated by Wujastyk (1998:212–230).
115. See, most forcefully, Hardy 1983. Though this long and detailed study has received substantial criticism over the years, its principal theories regarding the influence on North Indian Sanskritic devotional culture by Tamil religious culture and sectarian poetry remain intact. Another excellent study that came out at nearly the same time is Yocum 1982, esp. pp. 187–194.
116. For example, Sanderson (2001:6n4) notes a goddess named Bahumāṃsā ([She who possesses] much flesh) in a ninth-century Nepalese manuscript of an Ur-Skandapurāṇa.
117. I have been looking into this for another current project on the development and distribution of the terminology for different types of memory in Sanskritic culture and discourse, viz. the distinctions between smṛti, smaraṇa, and smara, all derivatives from the Sanskrit root √smṛ (related, of course, to the English word “memory”). Smṛti denotes a class of legalistic socioreligious texts, cultural or religious tradition, and the function of memory; smaraṇa indicates meditation, repetition of divine name, and memorialization; smara is another word for “love” or “the god of love” (Kāmadeva, etc.).
118. The use of yantras for bhūtnivārak (warding off bhuts) is still in common practice in India. See, e.g., Hanumān-Upasanā, by Rādhā Kṛṣṇa Śrīmālī (pp. 34ff.). Among the yantras he discusses are the bhūt-pret-nivārak yantra, the bhūt-nivārak yantra, and the mūṭh-nivārak yantra. In conjunction with this, a peacock-feather fan is employed to “sweep away” spirits, especially from small children (p. 56).
119. This term is widely attested in South Asia for “exorcist.” See Mumford 1989:148n12 and Chapter 4, n. 194.
120. Mysore ms. #48062 states that the procedure described therein can be used for possession by any deity: brahmaviṣṇumaheśādi devatāvesaniscayam | This manuscript is unusual in that it does not specify that the possession occurs in any locus other than the ritualist. Recall the association of Bhairava with possession in the Prabandhacintāmaṇi of the late first millennium.
121. Unfortunately, this manuscript is incomplete and lacks a colophon. On the link between Bhairava and the sarabha in coastal Andhra Pradesh, where both are regarded as avatāras of Śiva, see Knipe 1989:142. He adds: “The sarabha rupa (monster form) of Śiva appears to be conflated with the bhairava rūpa (Vīrabhadra) in the Rajahmundry cult and its symbols, as indeed it is in the Śiva-Purāṇa” (p. 153n11). For this and other sources, see Gonda 1970b:106f. See also the translation from the Kālikā Purāṇa in O’Flaherty 1975:193–197, an account that contains several episodes of possession.
122. GOML D7763 reads: brahmacārī samabhyarcya aṣṭahāyanamātrakaḥ | yantropari ca saṃsthāpya bhasma mūrdhni vinikṣipet | … evaṃ yantraṃ samālikhya baṭukaṃ ca tathopari padmāsanena saṃsthāpya bhasmamantreṇa nikṣipet |. Mysore #48069 reads: abhyaṇganaṃ ca kṛ tu brahmacārī tathopari | svastikāsanasaṃyuktaṃ pūrvābhimukhataś śuciḥ | aṣṭahāyanamadhyasthaṃ kṛaṃ śokavivarjitam |
123. Mysore #48062: etanmantroccāraṇena pratyakṣāveśam āpnuyāt | atītānāgataṃ caiva janmāntaraphalaṃ tathā || bhaviṣyaj janma karma ca śubhāśubaphalaṃ tathā | On the link between Bhairava or Vīrabhadra and svasthāvesa, see note 9; and on the identity of Vīrabhadra in Andhra Pradesh, see Chapter 8 and Knipe 2001.
125. I use the word sastra guardedly in this context. Because of the amorphous nature of Tantra, it is not possible to locate a single pedagogy or set of systematics that usually qualifies the word śāstra.
126. There is no evidence that this is the same Mahīdhara who composed a commentary on the Mādhyandina recension of the Śuklayajurvedasaṃhitā.
127. The word vidyā is a synonym of mantra.
128. Compare this to Īśānaśivagurudevapaddhati 1.25.75: svaikyataḥ śūlinīṃ dhyātvā svasthāveso bhavej japāt (After meditating on the goddess who holds a stake [śūlinī] and realizing one’s unity with her, one should gain svasthāvesa through the performance of japa).
129. Mantramahodadhi 5.87–90. Cf. GOML D7763: evaṃ yantraṃ samālikhya baṭukaṃ ca tathopari / padmāsanena saṃsthāpya bhasmamantreṇa nikṣipet / śatavāraṃ mantritaṃ ca āveśaṃ ca bhavedhṛvam / [corr.: bhaved dhruvam].
130. In the earlier Śaiva and Sākta texts Vidyās served as embodiments of female divinity and existed alongside the prasenā. Mahāvidyās were “specific to the late East Indian Tantric literature and its pan-Indian development” (Sanderson, e-mail communication, October 10, 2002).
131. See Strickmann 2002 for the use of an “ox bezoar,” either alone or mixed with cinnabar. When rubbed on a certain kind of seal, it “will not only summon all manner of spirits but will work large-scale geological, hydraulic, and cosmological changes as well. These include shifting mountains, draining seas, causing lightning and thunder, causing darkness in the heavens and over the earth, making the sun and moon fall from the sky, and bringing clouds and rains” (p. 323n102). On the consumption of the human gorocana, said to be located in the heart, and the powers derived from this peculiar anthropophagy, discussed in certain Yoginītantras, see Gray 2005:55f., and for more on its identity, ibid.:47-48n8; for its use as an artistic symbol in Tibetan art, see Beer 1999:188.
132. See Taylor 2001:217ff. for evidence that Woodroffe knew no Sanskrit at all.
133. Śrī Subhagānandanātha’s Manoramā glosses svasthāveśam in 9.76a as svasthasya devatāvesanam. On this peculiar incense (9.76d), he says simply sarjjaraso nāma dhūpadravyaviśeṣaḥ. It appears to have been unknown to him.
134. Dudbridge 1995:1ff., 190–191 (#84–85). Dudbridge’s interpretations are not as convincing as his translations. He relies much too closely on interpretative methodologies of possession formulated in the 1930s and 1950s, which now appear seriously dated, not least because they were based on prevailing psychoanalytic and Christian viewpoints of the day. However, compare this story with Diemberger’s brief ethnography of a teenage girl who experienced an initiatory dream, after which she became possessed, giving her the ability to recite major sections of the Gesar of Ling (2005:154). See also Baptandier 1996, which discusses the rise of the cult of Lady Linshui in Fujian province and how it was based on mediumistic activity, in which the founder, a girl named Chen Jinggu, who lived a very short life during the Tang dynasty, appears in possession. This possession often occurs in male mediums called fashi who dress as women on this occasion and address calamitous issues. See Chapter 4 note 41 on this term. It appears that most cults in China have been based on spirit-mediumship. I am increasingly convinced that this is also the case in South Asia.
135. See my review article of ten books on possession, F. Smith 2001.
136. Strickmann 2002:208.
137. Here we have a mixture of Buddhist and Daoist ritual. Strickmann writes: “We already know that Taoist written documents were burned in the flame of the incense burner, and that they make their way heavenward in the company of special emissaries, spirit-messengers transformed by priests from the breaths of their own bodies” (ibid.:210).
138. Recall the symptoms of tantric saktipāta discussed in Chapter 10 (cf. Tantrāloka 29.207–208).
139. On the unreliability of dream and memory, see, e.g., Yogasūtras 1.5–11 and commentaries. Halbwachs 1992:41ff. argues that memory is a social category while dream is individually constituted and unreliable.
140. Strickmann 1996:291ff. “Le rituel est l’essence du bouddhisme tantrique, et ses traditions oniriques font partie intégrante du processus rituel” (p. 295).
141. “Dans le livre du yogin de tous les yoga du pavillon au faîte de diamant [T. 867], les instructions recommandent de réciter le mantra d’Acala cent huit fois, de former mudrā-épée avec les doigts, et ensuite de s’endormir immédiatement. La divinité viendra alors ‘prendre possession de vous’ (āvesa), et vous verrez en rêve toutes sortes de choses concernant le bien et le mal” (ibid.:298).The possessing spirit is not identified. I suspect, however, that it is not a prasenā, which he would certainly mention if it were the invader.
143. As discussed earlier, the clientele at Bālājī and other centers of exorcistic healing in India today includes members of all classes, castes, and religious affiliations.
144. Compare this to Orofino (1994a:613), who adds: “In the same way the yogin can see the images in the ether through the power of meditation on the supreme immovable bliss once he has abandoned the pleasure of the emission of his bindu” (ibid.). The text of the Sarnath edition reads, in part: tatkāraṇam asti yena kāraṇena kumārikāyāṃ dvīndriyasaṃgharṣaṇāc cyutisukhopalabdhir nāsti | (3:88). Sanderson notes, “the expression dvīndriya, for the sex organs is standard Mantrayāna terminology” (e-mail communication, October 10, 2002).
145. Tachikawa (2000:234) points out that Tibetans distinguish visualization from possession, just as Japanese shamanic healers of the Shugendo sect do. The latter is a synthesis of Buddhism and Shinto.
146. Knecht discusses this as a general problem in deciphering shamanism. Consultations with tutelary spirits—a phenomenon just short of possession, though distinguished from it—are another possibility (2003:16).
147. Diemberger 2005:132. See ibid.:134 for a photo of a female oracle holding a divination mirror.
148. Ibid.:135. Similarly, Bellezza’s extensive ethnographic evidence (2005) of a connection between mirror divination and possession in Northern Tibet all but confirms it in the case of Ama Ta Bap.
149. For the remarkable Islamic account of the exploits of Alexander the Great, see Seyllor 2001:19–20.
150. The revolving mirror, cast under a magical spell, is mounted atop the high tower at the top of the painting. According to Khusraw, the tower was erected at the edge of the Sea of Rum (the Mediterranean Sea), which is depicted here. For a description of the scene, see ibid.:84. For another use of mirror divination in Tibet, see Yang 1993. In this practice, certain mirror divinization chanters (pra-ba) are able to “see” the written forms of the Tibetan epic Gesar of gLing, which they then copy and chant. On the possibility of the export of magical mirrors from India, or at least their concept, to the Near East at undetermined points in antiquity, see Allen Thrasher’s inquiry on “Bhattah mirrors” on the Indology-listserv dated February 22, 2006 (http://listserv.liv.ac.uk/cgi-bin/wa?A2=ind0602&L=INDOLOGY&P=R1807&D=1&F=P&H=0&I=-3&O=D&T=1), and his references to Deveney 1997.
151. See, e.g., Avedon 1984:4ff.; Aziz 1976.
152. Orofino 1994a:616. On the general phenomenon of incarnation as possession in Tibetan Buddhism, see Aziz, who writes, “Tibetan Buddhism, as it is practiced in India and Nepal, allows for a host of diviners, spirit-mediums, oracular devices, monks and yogins, each of whom deals in his own way with malevolent as well as friendly forces encountered in the normal course of people’s lives” (1976:357). On the use of the word avatāra for possession in Buddhist texts (viz. the Mañjuśrimūlakalpa) see below, ch. 12 note 141. On the use of mirrors in Central Asian Islamic divination, see Levin 1996:245.
153. I am grateful to Urs App for this information. He has filmed some of these mediums (all female) in an ongoing project designed to collect some of the remaining shamanistic traditions of Taiwan on video.
154. I am grateful to John Lagerwey for this point.
156. One need look no further than the stories of the baby Kṛṣṇa for attestation of this. Examples could be easily multiplied.