The Self Possessed

This chapter is organized around the general theme of bhūtavidyā, the vidyā (science) of bhūtas (existent beings),⁶ most of which are invisible or assumed to be inhabiting other beings, the most important of which is humans, and are believed to cause various diseases, including certain forms of mental illness. This is a vast subject in India, but the present discussion examines only a few of the many texts within a surprising number of genres in which this is delineated. These include the Atharvaveda (AV) and its ancillary texts, the first millennium C.E. canonical āyurvedic texs, certain Tantras and dharmaśāstra texts, and texts otherwise regarded as philosophical. In Chapter 6, we examined a bhūtavidyā section from the Mahābhārata (MBh, 3.216–219). This is augmented in the present material. After discussions and translations of the key passages in the early āyurvedic texts and a complementary passage in the Sāṃkhyakārikā, we discuss certain modern interpretations of the āyurvedic passages, employing an important early Tibetan medical text as a counterpoint. Then we expand the discussion to include material on bhūtavidyā from the tantric and dharmasastra texts, as well as from a few contemporary ethnographies. The primary purpose of this examination, as of much of the remaining parts of the chapter, is to illustrate that India has a long history of viewing disease-producing spirit possession in moral terms: that behavior, viewed within a framework of moral concern, is a major component in the diagnosis of mental illness, more specifically in the diagnosis of spirit possession.⁷ Along the way it is necessary to discuss at length terms used for certain important possessors, in particular the graha and the piśāca. The chapter closes with discussions of diagnosis and treatment, drawn from both classical āyurvedic texts and contemporary practice, which do not always coincide. It would be a mistake to assume that the āyurvedic texts and traditions are invariable, unchanging, in agreement, or applied consistently in practice. Indeed, in response to changing cultural and intellectual practices, these traditions increasingly embraced elements of Tantra, dharmasāstra, and unclassified local practices, producing a mosaic around bhūtavidyā that, like so much of Indian praxis, leaves everything different, yet somehow the same.

Because possession was widely regarded as an affliction, whether for good or ill, it was medicalized very early. Āyurvedic texts of all periods locate possession within a category of pathology called unmāda (madness), though it is commonly known in āyurvedic circles more sanguinely as mānasika-roga (mental illness). Within that category, possession is regarded as an affliction that enters a person from outside, a condition with “exogenous” (āgantuka) causes,⁸ beyond the scope of identifiable medical or social pathology. In this way, it was regarded as suffering of divine origin (adhidaivika duḥkha), “effected by providential causes or acts of the gods, that is, factors that are beyond human control.”⁹ This view is expressed consistently from the early compendia of Caraka, Susruta, and Vāgbhaṭa, through the later commentaries and independent āyurvedic treatises of the mid- to late second millennium.

The history of bhūtavidyā can be extended back to the early vedic period, in which, in the Ṛgveda (ṚV), the gods (deva, sura) were distinguished from their archenemies (asura, dānava), and in which the battle between the gods and demons was eternally fought, often to a standstill, though with a slight, if temporary, advantage to the gods. The ṚV identifies other categories of ethereal beings as well, who opposed humanity rather than the gods. In form they were anthropomorphic but, like the gods, insubstantial or, more accurately, endowed with a different substantiality. Among the most important of these beings were the gandharva, rakṣas, yātu or yātudhāna, and, later, the piśāca.¹⁰

The term bhūtavidyā first appears in Chāndogya Upaniṣad (ChU) 7.1.2, a text of the seventh century B.C.E. or perhaps a century or two later, where it is listed along with other vidyās including the Vedas, Purāṇas, divine knowledge (devavidyā), and understanding of brahman (brahmavidya).¹¹ It is probable that at this date the term bhūta indicated any type of being, animate or otherwise, visible or otherwise, as certain texts describing domestic ritual (Gṛhyasūtras) prescribe offerings of water to be made to all beings, including deities, heaven and earth, days and nights, the year and its divisions, lunar asterisms, the spatial midregion, the syllable om, numbers, oceans, rivers, mountains, trees, serpents, and birds. In addition, the following celestial, semi-celestial, and other bhūtas are to receive similar offerings: apsaras, gandharva, nāga, siddha, sādhya, vipra (viz. brahmans), yakṣa, and rakṣas, as well as cows, ancestors, and teachers, both living and long deceased.¹² Thus, by the mid-first millennium B.C.E. the word bhūta was applied to all manner of perceived ontological entities, including “spirits.” Bhūta may also have signified beings allied with an “element” (also indicated by the word bhūta), hence used to “personify the elemental fragments of creation, infinite in number.”¹³

One hymn of the ṚV (10.162) makes it clear that both the female body and fetuses were considered to be at great risk of possession by an exotic being, a rákṣas further identified as an ámīva, a flesh eater (kravydam, 10.162.2) that separates the thighs and enters the womb while the husband and wife are sleeping together, in order to lick the inside of the womb (10.162.4) and, presumably, destroy the fetus.¹⁴ This was likely a precedent to the florid demonology in the MBh section on skandagarahas (child-snatchers) examined earlier, and similar pediatric sections of āyurvedic texts we examine more closely here. The ṚV, AV, and Yajurveda (YV) saṃhitās speak not only of spirits but of diseases, demons, and the punishments of Varuṇa and other deities pervading or overpowering the individual. As several scholars have pointed out, deities and spirits do not necessarily work against each other, nor do gods uniformly represent order while spirits represent resistance. As Shail Mayaram notes, spirit possession can “reproduce hegemony and hierarchy, [while] gods and goddesses, on the other hand, can sometimes be countercultural.”¹⁵ This is certainly the case in ancient and classical India, where reversals are as much the rule as the exception¹⁶ and is nowhere more transparent than in the āyurvedic texts, in which deity possession is medicalized in the same way as spirit possession.

This feminine grasper is further mentioned in the Saunaka recension of the AV (AVŚ 16.5.1), where Sleep (svápna) is identified as her son. This is reminiscent of later analyses of possession, for example in tantric texts, where sleep and other forms of extinguished memory or obliterated personality, such as fainting, are described as symptoms of possession. Finally, one passage from the vedic Śrautasūtras recommends that if a person who is possessed (“touched”) by a spirit (bhūtopaspṛṣṭa) speaks during the performance of the pravargya, an important rite preceding the main ritual of the soma sacrifice,¹⁸ the adhvaryu or chief officiant should offer a stick lit at both ends into the āhavanīya (eastern fire), reciting mantras first to Agni (Taittirīya Āraṇyaka [TĀ] 4.28), then directing the words of the piśāca-possessed person to follow (and) possess (anvāvisya) the enemies of the sacrificer, urging death upon them (TĀ 4.21).¹⁹

The AV, as must be expected of a vedic saṃhitā, does not systematically discuss either bhūtavidyā or possession. Nevertheless, bhūtavidyā became more prominent in the medical sections of the AV, while the text mentions possession and remedial measures only casually. In spite of this, the allied sciences of Āyurveda, developed much later, always regarded the AV as their source text, as the inaugural text on Āyurveda.²⁰ In several instances (AVŚ 1.16, 2.4, 3.9), the text provides mantras designed to empower amulets that are to be used to overcome víṣkandhas (disorders or disturbances caused by invasive and malevolent rákṣases [demonic beings in general] or piśācás [flesh-eating demons]).²¹ Other hymns (e.g. AVŚ 4.20, 5.29) reveal an awareness of malevolent ethereal beings, including piśācás and yātudhnas (wandering ethereal sorcerers). Another hymn (AVŚ 4.37) discusses gandharvás and their apsarás wives, beings with malevolent intent, though seemingly not equal in negative force to the rákṣas or piśācá.²² The AVŚ (2.25) identifies an embryo-eating (garbhādám) being called a káṇva, extending the notion from the ṚV that pregnancy had to be protected with great care.²³ The most common word for such invasive disease-causing spirits in Āyurveda is grhī, a term inherited from the ṚV and AV. An example of a ritual invocation for the removal of such spirits occurs in AVŚ 6.112.1: “O Agni, foreknowing, loosen the fetters of the grhi.”²⁴ Thus, upon diagnosis of disease-causing possession, ritual was a prescribed therapeutic adjunct to medical procedures like herbal treatments or surgery.²⁵ This multifaceted approach to healing has continued undiminished in India. For example, the Tantrarāja Tantra (31.27–29, 67) prescribes that a mannequin (puttalī) of clay mixed with earth from a cremation ground, ashes, salt, ginger, garlic, asafoetida, and so on, representing an enemy, be stabbed with a thorn or nail if, due to injury, one is possessed by a piśāca (piśācāviṣṭa), thus dispatching the enemy to Yama’s world.²⁶ This kind of “folk medicine” may be found today in India, as K. P. Shukla discovered while conducting his study of traditional healers near Varanasi in 1969–71.²⁷ Later we turn to further examples of this from the Īśāṉaśivagurudevapaddhati and some derivative Keralan traditions.

The hymn reads as follows:

In examining this passage, it is important to note that if Zysk’s interpretation is correct, the division it describes foreshadows the later āyurvedic distinction between unmāda caused by accountable pathological factors and unmāda brought on by unaccountable invasive entities. Although I do not see in verse 2 evidence for the assertion that “insanity, like death, was considered to be a state when the mind leaves the body,” nevertheless the deployment of únmaditam and únmattam in verse 3 is striking, and the use of the stronger, more direct, únmattam for madness caused by possession is more notable still.³⁰ However, in the absence of evidence for systematic usage of these words in the AV it is questionable whether this is more than a fortuitous effect of the poetic form. In any event, it does appear that at this early stage a distinction was seen between pathological madness and madness caused by possession.

It is useful to consider the deployment of gods (and, we can easily speculate, demons) as arbiters of moral action in the AV, and compare this to the much later Tantras, which also describe the openness of individuals whose moral transgressions made them vulnerable to possession by disease-causing spirits, a topic we discuss at length below. Sanderson observes that, according to the Netra Tantra and Jayaratha’s Viveka on Abhinavagupta’s Tantrāloka, “Yoginīs, Śākinīs and other female obsessors suck life from their victims into themselves as an offering to their regent Mahābhairava enthroned in their hearts.”³¹ Sanderson demonstrates that, according to brahmanical thought, such possession was due to the relaxation of both self-control and conformity to one’s dharma. Thus, “these ever alert and terrible powers of the excluded could enter and possess, distorting his identity and devouring his vital impurities, his physical essences.… Possession, therefore, was doubly irrational: it obliterated the purity of self-control and contradicted the metaphysics of autonomy and responsibility.”³²

In his Tantrāloka (15.595–99), Abhinavagupta cites the authority of the Kulagahvaratantra and Niśisaṃcāratantra as predecessors to his own symbolic interpretation of demonic possessors (graha). Sanderson states that these grahas “conceal the true self (autonomous, unitary consciousness) beneath a phantasmagoric pseudo-identity, contaminating and impoverishing it with categories unrelated to its essence.”³³ These grahas are “obsession with caste (jātigraha), vedic learning (vidyā), the social standing of one’s family (kula), with orthodox conduct (ācāra), with one’s body (deha), one’s country (desa), and material prosperity (artha),”³⁴ in other words, the demons of personal identity. We might compare this sense of possession as identification with aspects of the ordinary human condition to Kausikasutra 28.12, an early ritual commentary on the AV, in which “infringement of certain divine mores or taboos” is illustrated through the prescription of an Atharvavedic mantra (AVŚ 5.1.7) as an antidote for amatigṛhīta, possession by “cluelessness” (amati), another ordinary human condition. Dārila, the commentator on the KausSū explains this term as “one with no idea about dharma, artha, and kāma” (trivargaśūnyabuddhiḥ).³⁵ This, however, does not go as far as Abhinavagupta in bringing symbolic weight to a possessing entity. Indeed, Abhinavagupta and his predecessors are the only figures in Sanskrit literature to provide an explicitly symbolic reading of possession, except, of course, if one considers the entities in the āyurvedic categories purely symbolic. Although the latter is doubtful, as we now investigate, the line between the symbolic and the ontologically actual in these texts is nearly erased.

Possession in the āyurvedic texts has been addressed before, notably by Mitchell Weiss,³⁶ who analyzes Indian categories of unmāda and exogenous disease-causing spirits with reference to Freudian and other psychoanalytic categories.³⁷ In addition, Jan Meulenbeld has succinctly described views and categories of bhūtavidyā set forth in different āyurvedic texts. These textual differences have been considerable and should be kept in mind in the ensuing discussions. Among the areas of disagreement were the nature of bhūtas, their ontological reality, their number and identity, their cultural origins, and what constitutes the discipline of bhūtavidyā, including knowing not just the names and descriptions of bhūtas but the diagnostics, pharmacology, and ritual treatment for possession by bhūtas.³⁸

Weiss’s results, in my view, are not particularly helpful in understanding the āyurvedic categories and, like many comparative studies, make little attempt to bridge the cultural and philosophical gap between the loci of comparison, in this case India of two thousand years ago and Vienna of the early twentieth century. The psychoanalytic prism has the singular advantage of offering an explanation of possession as represented in these texts to a reader with interests in Freudian psychoanalysis, but it neither sheds much light on what these texts attempt to illustrate nor offers an effective model with which to evaluate notions of selfhood that arise from a consideration of possession as defined in these texts. In addition, by failing to consider the emic gravitas of possession, an interpretation that constantly looks over its shoulder at Freudian psychoanalysis unwittingly abets Western cultural and scientific hegemony and authority. As the foregoing discussion should make clear, the etiology and identification of bhūtas cannot be adequately examined in terms of any single theory.

Meulenbeld’s article is of quite a different nature. It provides insight into the complexity and contentiousness of first-millennium āyurvedic debate over the nature of unmāda, which are absent from Weiss’s study. It appears, in short, that the early compendia of Caraka and Susruta modernized the study of medicine in the first few centuries of the Common Era. One topic with which they were forced to grapple was the notion inherited from the AV and other earlier texts that nearly every disease could be personified or anthropomorphized and attributed to possession by spirits and other divine beings. In their attempt to make medicine more empirical, Caraka and Susruta placed distinct limits and conditions on this notion. They were not always in agreement, and this sense of debate persisted through the next millennium or so of āyurvedic textuality. Their fundamental solution to bhūtavidyā, however, was similar: They relegated it to a subcategory of unmāda and conditionally of apasmāra, both of which they believed to have a number of possible causes, including those that were inherent (nija) or explicable.

Caraka (Cikitsāsthāna [CaCi] 3) treats four categories of fever (jvara) brought about by external factors: (1) abhighātaja (those resulting from injury); (2) abhiṣaṅgaja (those resulting from afflictions by evil spirits, as well as emotional overload, including uncontrollable passion, grief, fear, and anger); (3) abhicāraja (those arising from sorcery, including hostile ritual, mantras and oblations); and (4) abhisāpaja (fever brought on by the curse of a spiritually advanced being). CaCi 3.123, addresses abhiṣaṅgaja fevers: “Abundant tears [are observed] in [fever caused by] grief; when fever is due to fear, terror becomes prominent; extreme agitation [is evident] when [fever] arises out of anger; nonhuman [symptoms are observed] in [fever arising from] possession by spirits [bhūtāveśe].”³⁹

The foundational texts of Āyurveda, Caraka Saṃhitā (6.9.16ff.), Suśruta Saṃhitā (6.60), and Aṣṭāṅgahṛdaya Saṃhitā (ch. 6), discuss the “epidemiology of possession,”⁴⁰ in this case bhūtatantra (doctrine of bhūtas), a category that includes bhūtonmāda (insanity produced by the influence of bhūtas).⁴¹ Although these enumerations are regarded as different categories of graha or bhūta, they share fundamental territory with categories of personality. This resonance is borne out in complementary sections of Caraka and Susruta, as well as in the most important texts belonging to the second chronological level of āyurvedic textuality, the Kāsyapa Saṃhitā and the texts of Vāgbhaṭa, which postdate the earlier compendia by four or five centuries.⁴²

These sections describe sattvas—essences, personality types, or mental states⁴³—as part of a discourse on embryology, the human body in its development from conception to birth. Many of these sattvas correspond in both personality and name to the various grahas listed in the bhūtavidyā sections.⁴⁴ From this, we must assume that grahas or possessing entities may be viewed as substantialized or reified collocations of personality attributes. Nevertheless, it appears that the authors of these foundational texts maintained the distinction between native personality and grahas that independently personified these attributes. This is consistent throughout the canonical āyurvedic literature. Bhūtavidyā consists, formally at least, of two divisions (graha or bhūta), though in much of the literature they are used interchangeably, with the word graha attested more frequently. Nevertheless, when a distinction is required, the category of graha is used to denote bālagrahas, the mostly female possessors of pregnant women and children. But bhūtas are almost entirely male grahas that are known by their specified characteristics. Bhūtas may be regarded as subspecies under which no further taxonomic breakdown is required. Thus, the category bālagraha may include sixteen, eighteen, or more varieties, while grahas as subcategories of bhūta may be gandharvas, yakṣas, pitṛs, niṣādas, and so on, without further subspecies. The latter division consists of bhūtas called grahas, which the brahmanically constituted āyurvedic texts have drawn from vedic and other more “orthodox” sources.

This categorical and taxonomic arrangement is represented fairly consistently from the MBh through the āyurvedic literature and in modern folklore. These two divisions appear to emerge from quite distinct local traditions and probably from different strata of society as well. Their odd integration into the Vanaparvan of the MBh and the pediatrics sections of the āyurvedic texts suggests that bālagrahas existed outside the realm of Sanskritic culture and discourse until they were incorporated into the fold, Sanskritized, gaps filled in, and systematized, one for every year, and so on, as we saw in the MBh and see again later in this chapter. The more brahmanically inclined (if not always more benign or better-behaved) grahas were taken from an assortment of vedic texts, reconceptualized in a way that brought them into a closer, even familial, relationship, and presented in a coherent and systematic discourse. Another difference between them is that pregnant mothers and children possessed by bālagrahas appeared to maintain their “own” personalities, while those possessed by more orthodox bhūtagrahas appear to have been consistently overwhelmed by the personality of the possessing agent.

Subsequently, the project of bhūtavidyā, including taxonomies, diagnostics and treatment, provided the āyurvedic texts with an exegesis on, and to some extent a solution to, the problem of deviant personality management. Indeed, through the very organization of these passages, the texts have called into question the notion of personality. In a chapter titled unmādacikitsā (treatment for madness), Caraka (9.20–21) describes in detail the unmāda produced by possession (abhidharṣayanti; lit. “overpowering”) of eight kinds of divine and semidivine beings, as well as the characteristics of people who might be possessed by each one, a topic we address in detail shortly.⁴⁵ It must be noted that all the canonical texts kept the bhūtavidyā sections distinct from the sections on bālagrahas. Although I am here including them both within the broader parameters of bhūtavidyā (knowledge of [nonhuman] entities), the texts regarded bālagrahas as a part of pediatric medicine, rather than of mental illness.

In a chapter of similar content, titled amānuṣopasargapratiṣedha (repulsion of contact by the nonhuman), Susruta (6.60.8–18) divides grahabalas into nine classes. First, however, he defines graha. “A graha,” he says, “is well-known as an unstable quarrelsome entity that engages in nonhuman activity and has knowledge of what is hidden or lies in the future. They afflict the unclean, [and] fracture [social] boundaries in whole or in part, even if the purpose of such affliction is [not explicitly malevolent, but] merely frivolous or to satisfy a religious obligation” (6.60.4–5). These are distinguished from bhūtas, a distinction that yields little practical difference, as Susruta himself observes (see Suśruta [Su] 6.60.27). Half a millennium later, in the Aṣṭāṇgahṛdaya Saṃhitā’s equivalent chapter, bhūtavijñānīyabhūtapratiṣedha (entities to be known and repulsed), this number grows to twenty (AH 6.4.9–44).⁴⁶ This suggests that the canon of bhūtas, so to speak, was open, a fact verified by the introduction of grahas with non-Sanskritic names into the AH, a process that has continued to the present. This transition from folk to classical was discussed in Chapter 4, while commenting on descriptions of the phenomenology of possession as represented in modern ethnographies. Here, too, first-millennium chroniclers of mental illness not only did not standardize the list of grahas, which perhaps might be expected of a developing scientific enterprise, but encouraged an expansion of the list. The dynamics of this early interaction between folk and classical are clarified below, in the discussion of enumerations of grahas outside the primary āyurvedic literature.

Convulsions, or seizures of any kind, constitute another condition believed to be caused by malevolent spirits, as well as by planets, also called grahas, that are malevolent and badly placed in one’s horoscope. The general word in Sanskrit for “convulsion” is apasmāra (literally, “forgetfulness”), which is often wrongly assumed to signify epilepsy alone.⁴⁷ The early medical texts did not consider that apasmāra could have an organic cause. This was only realized much later in Europe by Jean Martin Charcot (1825–1893), the “Napoleon of Neuroses,” an early founder of neurology and one of Freud’s teachers. Charcot developed the concept of hysteria and distinguished it from epilepsy, the former of which had psychological causes and the latter organic causes. One definition of apasmāra—striking because it is not in a medical text—is found in the Daśarūpa, a treatise on dramatic theory, which regards seizure as a transitory state (vyabhicāribhāva) of giddiness to be replicated in dramatic performance. The Dasarūpa states, “Apasmāra is āveśa induced, properly speaking, by the influence of the planets, by misfortune, etc. Its results are falling on the ground, trembling, heavy sweating, and slobbering, frothing at the mouth, etc.”⁴⁸ Two synonyms of āvesa, given by Hemacandra, are apasmārarogaḥ (seizures as disease) and ahaṃkāraviśeṣaḥ (a variety of self-identity).⁴⁹ We have confirmation of the latter in the sections from the Tantras cited by Sanderson above, which are roughly contemporaneous with Hemacandra, as well as from the much earlier KausSū. Apasmāra is also found as the name of a certain bhūta or disembodied spirit in South Indian mythology and appears most notably as the demon on whom Śiva Naṭarāja treads in the great temple in Chidambaram.⁵⁰

The primary āyurvedic texts enter into significant detail on possessing entities. These passages shed light on the manner in which possession and possessing entities rearrange personality and behavior. They also reveal to some extent the constitution of selfhood as conceived by these texts. This makes it useful to translate several lengthy āyurvedic text passages in full.⁵¹ This is equally essential if we understand personality and behavior as paramount and grahas and bhūtas as materializations of these behaviors. Before translating these passages, however, we must first describe the nature of grahas.

As Susruta’s definitions of graha and bhūta clarify, the bhūtavidyā passages employ the word graha to describe certain kinds of bhūta. A bhūta, in its very etymology, is any existent ontological entity, while a graha is an entity further delineated by its ability to “grasp” or “hold” (√gṛh). The archetypal “grasper,” which the authors of these texts perhaps had in mind, is the alligator, which in Sanskrit is indeed called grāha. It is perhaps no accident, then, that a grāha is also an idea or notion, and a grāhaka is a perceiver, one who grasps an idea or notion. Consistent with the passage from ChU 7.1.2 cited earlier, if considerably pared down, later Sanskrit literature limits the term to include only these exotic beings and the planets.

In Bṛhadāraṇyaka Upaniṣad (BĀU) 3.2, Jāratkārava Ārtabhāga asks Yājñavalkya how many graspers there are, and how many overgraspers (katigrahāḥ katy atigrahā iti). There are eight of each, replies Yājñavalkya, and they come in pairs, a grasper that is overcome by its paired overgrasper (atigrāheṇa gṛhītaḥ): the inbreath (prāṇa) which is overcome by the outbreath (apāna), speech (vag) by name (nāman),⁵² tongue (jihvā) by taste (rasa), eye (cakṣus) by form (rūpa), ear (srotram) by word (sabda), mind (manas) by desire (kāma), the two hands (hastau) by action (karman), and skin (tvac) by touch (sparsa). Taking this as a guide, a graha or personality-laden ethereal grasper would assume the role of the overgrasper during possession, taking control of the sense organs and sensory functions of the individual, who would in a healthy individual be the grasper of these organs and functions as well as their perceiver, a complex person consisting of both grāha and grāhaka.

One of the common meanings of the word graha in Sanskrit is “planet,” of which nine are counted, the seven after which the days of the week are named, plus the two nodes of the moon, Rāhu and Ketu, which are held to be responsible for eclipses.⁵³ According to the astrological literature, planets are called grabas because they grip or hold people in their power. The Bṛhatparāśarahoraśāstra (3.4) says, in what is surely a very old idea, “those which grasp [gṛhṇanti], constantly moving in the sky, are called planets [grahāḥ].”⁵⁴ Indeed, one of the late books of the AV, possibly dating to 1000 B.C.E., uses the word graha to refer to planets, at least as they are represented in the Indian system. The planet named here is Rāhu, not actually a planet but, rather, the northern node of the lunar path, which W. Caland identifies, at least in this hymn, with Vṛtra, the cosmic serpent, thus a “grasper” par excellence.⁵⁵ An eclipse is also called grahaṇa (bhāskaragrahaṇam [solar eclipse]) in Varāhamihira’s Bṛhat Saṃhitā.⁵⁶ The grahas Rāhu and (in later texts) Ketu, the southern node of the lunar path and thus not “planets” at all, seize the sun or moon, a celestial act of general affliction that requires ritual expiation.⁵⁷ Whether human or cosmic, the result is the same: The sun and moon during an eclipse, as well as individuals who are “afflicted” by grahas, require remedial procedures often resembling those employed in cases of possession as described in medical texts. Thus the influence of planets resembles possession by bālagrahas in Āyurveda as well as in instances of pra√viś, discussed earlier. This influence from the planets is invasive, pervasive, and motivated from without.⁵⁸ Consequently, the remedial or expiatory procedures recommended in astrological treatises are called grahaśānti (pacification of a planet).⁵⁹

Two other related words that share many of the semantic attributes of possession are vigraha (an embodied image) and anugraha (grace). Both can attract the attention of a person and overwhelm one’s sense of identity. The primary meaning of vigraha, following its etymology, is “separation, isolation, conflict.” One of its most common occurrences is in Sanskrit grammar, where dissolution of compounds into their constituent parts is called (pada-)vigraha. Supporting this notion of division into individual units is its meaning in the srauta ritual, where vigraha refers to a “portion,” for example of soma or ghee, which has been removed or separated from the main vessel.⁶⁰ Thus, in its early applications it conveyed a sense of individualizing, approximately the same sense it obtained in the MBh and the Purāṇas, where it indicated the form or image of a deity. In the only study of this word, G. B. Palsule states that vigraha “is something which is taken up or assumed. It is assumed by the jīva; it is something external, something shell-like, which the jīva assumes (vigṛhṇāti) temporarily. An ‘avatāra’ or incarnation is a vigraha, i.e. a form of the God which He has assumed.”⁶¹ After examining its occurrences in the epics and certain late Upaniṣads, Palsule concludes that vigraha appears in two contexts: “(1) assumption of a form, false impersonation or disguise; and (2) personification (real or imaginary).”⁶² Furthermore, because it connotes an assumed form, it never becomes a synonym of deha or śarīra, the usual words for “body.” Philip Lutgendorf records the contemporary North Indian notion among Vaiṣṇavas that “the Lord has four fundamental aspects [vigrah]; name, form, acts, and abode [nam, rūp, līlā, and dhām]—catch hold of any one of these and you’ll be saved!”⁶³ This doubtless reflects a much earlier view, in which an embodied form (rūp) is a “portion” of a deity. This helps explain, linguistically at least, the Vaiṣṇava view that an image of Kṛṣṇa is not a “representation” of the Lord, but the Lord Himself (svarūpa).⁶⁴ Thus, vigraha, like grāhi and graha, indicates an external presence appropriating or “seizing” form.

Anugraha (lit. “favor, kindness, assistance” in early literature) becomes a synonym of kṛpā (grace) in devotional literature.⁶⁵ The Taittirīya Saṃhitā (1.7.2.3) says, in a passage that is fairly typical of anu√gṛh in early Indic usage, “[I have invoked] her [Iḍā] who favors with kindness [anugṛhṇti, i.e., extends her grace to] people in distress and nourishes [gṛhṇti, takes] them as they recover.”⁶⁶ Anugraha does not appear in Upaniṣads until at least the mid- or even late first millennium C.E., in a few Saṃnyāsa Upaniṣads and sectarian Vaiṣṇava Upaniṣads, such as the Nārāyaṇapūrvatāpanīya and Gopicandana.⁶⁷ A typical usage is found in the Kathāsaritsāgara, “realizing that the grace [anugraham] of Kumāra toward them had blossomed.”⁶⁸ In all cases the meaning is mild compared to the “seizing” characteristic of other forms of √gṛh. The prefix anu- (following along) softens the sense of the verbal root, rendering the seizer’s engagement with the seized emphatically, if curiously, benevolent. Although anugraha never, to my knowledge, indicates destructiveness, disease, or sheer power, a further distinction is made in Śaiva Siddhānta: “God’s grace in its tender aspect is called anugraha and in its tough or punitive aspect, nigraha. It will be seen that even nigraha [restraint] is really anugraha because punishment is meted out only to correct and reform the erring individual.”⁶⁹

The Sāṃkhya school of philosophy was rarely (if ever) followed in India as an independent soteriological path, as were most other schools. However, it had a major, even pervasive, influence on the “higher” schools of yoga and Vedānta, on the Mahābhārata, as well as on many other systems of thought, including Āyurveda. Without Sāṃkhya, Āyurveda would look very different; indeed, it is Sāṃkhya that weaves the disparate threads of Āyurveda into a coherent system. The evolutionary principles (tattva-) that constitute Sāṃkhya’s main doctrine are the building blocks of the āyurvedic body, mind, and cosmos.⁷⁰ Within a set of fourteen types of sentient (bhautikasarga) that hierarchizes beings “from Brahmā down to a blade of grass,” the Sāṃkhyakārikā (SK, 53–54), which predates Caraka and Susruta, presents a list of divine beings that was almost entirely accepted by the latter two āyurvedic authorities. This is not to assert, however, that the SK was the sole influence on the early āyurvedic texts; such ethereal beings were, we might say, very much in the air at that time.

More specifically, SK 53–54 speaks of divine (daiva) realms, the human realm, and lower realms of animals, plants, and immovables, though it does not list these realms by name. This is left to the commentaries, particularly the Suvarṇasaptati (on kārikā 53), which lists the daiva realms as Brahmā, Prajāpati, Indra, Gandharva, Asura, Yakṣa, Rakṣas, and Piśāca. However, the Sāṃkhyasaptativṛtti on kārikā 53 lists Brahmā, Prajāpati, Indra, Pitṛ, Gandharva, Yakṣa, Rakṣas, and Pisāca; thus it eliminates Asura and places Pitṛ before Gandharva.⁷¹ The Sāṃkhykārikābhāṣya of Gauḍapāda on the same kārikā lists the divine realms as Brāhma, Prājāpatya, Saumya, Aindra, Gāndharva, Yākṣa, Rākṣasa, and Paiśāca. These are realms named after the specific divine or semidivine beings believed to characterize or inhabit them. In this list Soma, the lunar realm, is indexed before Indra’s realm, in place of either the Pitṛ or Asura realms. The anonymous commentary on the SK called Jayamaṇgalā presents the following list: Brahmā, Prajāpati, Sūrya, Asura, Gandharva, Yakṣa, Rakṣas, and Piśāca. In this list Sūrya replaces Indra, and Asura occurs once again.⁷² The Yuktidīpikā, an important early Sāṃkhya text, lists Nāga in the place of Yakṣa. These lokas (realms) assuredly represent domains in which beings that possess humans reside. These beings have cosmological roles and functions other than possessing innocent people; they comprise a balance of living beings that necessarily populate the existent worlds. Therefore, they may be regarded as archetypal ingredients in a cosmic soup that blends and separates in different consistencies according to proximity, locational strength or weakness, and so on, functioning much like elements of a multivalent, multidimensional, and complex personality. This is the Sāṃkhya cosmos and the Sāṃkhya individual, the thousand-headed, thousand-armed being, ever-shifting within itself. It is a classical expression of fluidity, which is so apparent in the modern ethnographies. The list of divine or semidivine beings in the SK may be briefly compared to similar lists in Caraka and Susruta.⁷³

A good general statement introducing the textuality of bhūtavidyā is Caraka’s succinct description of unmāda:

Aṣṭāṅgahṛdayasaṃhitā 6.4.1–44:

6–8. These weaknesses include undertaking a transgressive act, the ripening of an undesirable action, residing alone in an empty house, or spending the nights in burning grounds and other similar places;⁹¹ public nudity, maligning one’s guru, indulgence in forbidden pleasures, worship of an impure deity, contact with a woman who has just given birth,⁹² and disorder with respect to tantric or Purāṇic fire offerings [homa], the use of mantras, sacrificial offerings not involving fire [ bali], vedic sacrificial offerings [ijya], and positive actions or rites that counter negative ones [parikarma]; as well as a composite neglect of prescribed conduct in the form of daily routine and so on.

30–34b. One whose thoughts are unhealthy, who runs around, not remaining in one place, who is fond of leftovers, dancing, gandharvas, laughter, wine, and meat, who becomes depressed when rebuked, who cries without reason, who scratches himself with his nails, whose body is rough and voice trails off, who trumpets his many miseries, whose speech freely associates what is relevant and irrelevant, who suffers from memory loss, who enjoys nothing, who is fickle and goes around desolate and dirty, wearing clothes meant for the road, adorned with a garland of grass, who climbs on piles of sticks and rocks⁹³ as well as on top of rubbish heaps, and who eats a lot is understood to be inhabited [adhiṣṭhitam] by a piśāca.

I must now evaluate two important critiques of the material on bhūtavidyā presented in the āyurvedic texts: the first by Mitchell Weiss, the second by Terry Clifford. Weiss, as we have seen, deals with the data presented in the texts translated above. Clifford discusses material presented in a Tibetan text called the Gyu-zhi (rGyud-bzhi) (The Four Tantras), “the most famous and fundamental work in Tibetan medical literature.”⁹⁸

Let us first consider Weiss’s study, which summarizes the symptoms of unmāda according to the type of graha possessing the individual. Occasionally he supplements these with psychoanalytic interpretations. For example, with respect to unmāda caused by devagrahas he states that the devas “cause one to see, possibly indicating a paranoid delusion or hallucination, or perhaps an autoscopic hallucination.”⁹⁹ Furthermore: “In the deva and guru et al. types hyperpious, obsessive-compulsive, and paranoid symptomatology are characteristic suggesting rigid internalized controls, preoccupation with the cultural values, and a psychodynamically significant role for guilt. There are hallucinations, and paranoid schizophrenia has been suggested.”¹⁰⁰ Weiss continues, cogently, that the arrangement of the types of graha, in Caraka at least, “suggests a successively decreasing capacity for organized social functioning and a decreasing influence of internalized control mechanisms.”¹⁰¹ To this statement he adds that the sequentiality of the indigenous categories reveals “the psychodynamic significance of obsessive-compulsive, anal personality traits which yield to oral and then phallic narcissistic psychodynamics.”¹⁰² These observations, in my view, fail to consider local context, especially for the deva and guru types, namely, possession as devotional practice. In this context, the “decreasing capacity for organized social functioning and a decreasing influence of internalized control mechanisms” deviate from the norm, as he states, but the incipient context sanctions such emotionally laden behavior as derivative of devotional practice. Thus, to label this particular set of psychodynamics obsessive-compulsive, phallic narcissistic, and so on, is to fix both a taxonomy and a set of values on an experience, in this case the devagraha experience, which misstates, ignores, or downplays its cultural locus and significance as deviant behavior. Similarly, such preemptive classification ignores complementary passages in the āyurvedic texts that describe equivalent personality types not regarded as the product of possession.

Let us turn briefly to some of these passages, which present a de facto argument for bhūtas as substantialized collocations of personality attributes. The primary passage in question is from Caraka’s Śārīrasthāna (CaSā 4.36–40),¹⁰³ the section that describes embryology. This passage details the generation, construction, and development of the body (śarīra), describes different types of sattvas, essences or intelligences, as the final ingredient in the construction of the body. It is important to note that the canonical āyurvedic compendia are inclusive texts: they contain everything from creation mythology to diagnostics to anatomy (as in the present case) to descriptions of innumerable physical conditions to surgery (in Susruta only) to a broad range of therapies, including ritual, dietary, and pharmaceutical.¹⁰⁴

I refer to sattvas (states of being) as “essences,” or, following Meulenbeld, “personality types.”¹⁰⁵ CaŚā 4.36 states that “sattvas are of three types: pure [suddham], dynamic [rājasam], and slothful [tāmasam].¹⁰⁶ Among these, the pure (personality type) is said to be faultless [adoṣam], bearing a degree of auspiciousness. The dynamic [type] is said to be defective [sadoṣam], because it bears some anger. The slothful [type] is said to be defective, because it possesses a certain degree of delusion.” Caraka continues that these categories reveal dominant modes, within which individuality is manifested by unique mixing or texturing of the three. In some cases the personality follows the body’s tendency toward any one of these three, while in others the reverse is found—the body responds to the dominant tendency of the personality. Caraka then subdivides these categories into seven suddha types, six rājasa types, and three tāmasa types.

The suddha types have a clear affinity for divinity. These include: (1) the brāhma type, with the qualities of brahmā, is given to purity, passionlessness, tolerance, good memory, and so on; (2) the ārṣa, with the qualities of ṛṣis, is given to ritual performance and keeping vows and are hospitable, eloquent, free from anger, and so forth; (3) the aindra, with the traits of Indra, is brave, regular in performing rituals, authoritative in speech, generous, virtuous, and so on; (4) the yāmya, one with the qualities of Yama, are economical in action, self-motivated, is free from attachment and negative qualities; (5) the vāruṇa, one with the qualities of Varuṇa, is brave, patient, pure, religious, generous, fond of water, and proper in exhibition of anger and pleasure; (6) the kaubera, one with the qualities of Kubera, exhibit prestige, wealth, philanthropy, purity, and recreation; (7) the gāndharva personality type, who, like a gandharva, is fond of song and dance, poetry and epic narration, scents, garlands, women, and passion.

The six rājasa and three tāmasa personality types are closer to the accounts of grahas given in the texts. The rājasa types include: (1) the asura, with the qualities of an asura, is brave, powerful, cruel, lordly, and intolerant; (2) the rākṣasa, resembling a rākṣasa, is intolerant, angry, violent, envious, and fond of sleep, indolence, and nonvegetarian food; (3) the paisāca, one with the qualities of a piśāca, is gluttonous, lustful, dirty, cowardly, prone to poor dietary habits; (4) the sārpa, one with the qualities of a sarpa or snake, is brave when wrathful but otherwise cowardly, indolent, but with sharp reactions, and fearful when eating and walking; (5) the praita, with the traits of a preta or spirit of the dead, is always hungry, exhibits extreme suffering, is envious, indiscriminate, and excessively greedy; (6) the sākuna, one with the habits of a sakuni or bird of ill omen, is attached to passion and overeating, and is unsteady and ruthless, yet is nonacquisitive.

The three tāmasa essences or personalities include: (1) the pāsava, whose nature is like that of a beast (pasu), is one whose actions are wasteful, is unintelligent, has disgusting habits and diet, and is excessively interested in sleep and sex; (2) the mātsya, who resembles a fish (matsya), is cowardly, unintelligent, hungry, unsteady, passionate, wrathful, fond of water and constant movement; (3) the vānaspatya, whose personality resembles a vegetable (vanaspati), is indolent, always hungry, devoid of intelligence or any of its attributes.

Thus, the texts describe personality patterns as both products of birth and forms of madness. In both cases the characteristic behavior must be manifested as habitual, spontaneous, and nonritual. The only way to tell one from the other, apparently, is that in the latter madness diagnosed as exogenous, the characteristic behavior appears as a sudden and marked change that then becomes habitual, spontaneous, and nonritual. In addition, what confers on the possessed individual the psychophysiological identity of a graha is not just antisocial behavior in the case of a few of these grahas but an āyurvedic diagnosis of humoral imbalance that distinguishes their current condition from their earlier psychophysiological norm (prakṛti). In this case the unmāda may be diagnosed as exogenous (āgantuka) or inherent (nija), in the latter case attributable to identifiable causes other than possession, rendering it medically treatable.

The descriptions of grahas and sattvas, possessing entities and personality types, are noteworthy because of their close resemblance. They elucidate a differentiation not only in order to expose various abnormalities but also to express a sense of variation within the field of personality. The identification of these sattvas as independently or exogenously materialized grahas, which appears to have inspired in the authors of the āyurvedic texts a greater need to describe them graphically, is a kind of cultural phenomenology that in part harks back to the associative thinking of the vedic Brāhmaṇas and Upaniṣads. The linking of the infestation of certain grahas with days of the lunar month in the AH and elsewhere, and the systematic stigmatization of certain personality types as possessed who suffer from social aggravations, such as disrespect for brahmans, are also indicative of this. Whether one is possessed or otherwise, whether natural or exogenously acquired personality, one can well imagine engaging in normal social relations with individuals of many of these categories, including the deva, gandharva, rākṣasa, uraga, yakṣa, and several others.

As suggested, it is the intensity of the experience as well as its suddenness, which according to Āyurveda distinguishes an identifiable pathology from possession by a graha. In moderation, then, many of these behaviors reflect psychological conditions (as distinguished from sociological ones) that were not particularly aberrant within norms that were deemed culturally acceptable in classical India, but that nonetheless signaled danger that in extreme forms were regarded as pathological. It is for this reason that I question the impulse to unhesitatingly resort to Western psychoanalytic labeling. It is by no means certain that classical Indian sociocultural contexts would consistently yield to this sort of labeling. Thus, I am uncomfortable with Weiss’s full-scale application of psychoanalytic categories because he does not demonstrate their necessity in the sociocultural milieu of classical India. His method might have been more convincing if he had attempted to demonstrate the viability and applicability of a few carefully selected Freudian categories based on a cultural study of Dharmasāstras and Sanskrit literature.¹⁰⁷

Weiss might legitimately counter that the fact of possession is not the point; rather, the point is that the descriptions conform with Freudian psychoanalytic categories. My rebuttal, however, would then be twofold: first, in the Indian context they do not necessarily conform with Freudian psychoanalytic categories; second, and conversely, these categories shed little light on unmāda or possession from an Indian or āyurvedic perspective. Simply put, the success of any methodological or interpretative framework in cross-cultural analysis depends on its addressing discourse modes and epistemological issues on both sides of the divide. One thing that neither Weiss nor any other Freudian interpreter of Indic culture has attempted to understand on its own terms is the Indian notion of the dynamics of the interpenetration of the world of spirit with the world of matter. The usual strategy has been to wave it away with a Freudian wand, which, quite simply, is inadequate.¹⁰⁸ Therefore, I would not agree with Weiss’s acceptance of Jane Murphy’s conclusion in which she disputes the argument that cultural relativism as a factor in evaluating deviance is sufficient to undermine the notion that there are typologies of “affliction shared by virtually all mankind.”¹⁰⁹

Now, let us turn to Terry Clifford’s inquiry into the “psychiatric” aspect of Tibetan medicine. It is contemporaneous with Weiss’s study, but is not at all in the same vein. Although Clifford’s book appeared in 1984, she conducted her research in 1976–77,¹¹⁰ a fact worth noting because it appears that the application of Freudian psychodynamics to Asian medical systems peaked at about that time.¹¹¹ Clifford, a psychiatric nurse who took such a great interest in Tibetan medicine that she traveled to Nepal to study with Tibetan doctors, based her extensive study on the Gyu-zhi. This work was originally written in Sanskrit by Candranandana, who lived in the eighth century and also composed a commentary on the AH.¹¹² Indeed, the Gyu-zhi is based to a great extent on the AH, a fact that Clifford notes, though, curiously, she did not utilize the AH in her study, much to the detriment of her work.¹¹³ Candranandana’s work was probably translated into Tibetan shortly after its composition. Chapters 77–79 of the third Tantra of this text are devoted to demonology.¹¹⁴ Clifford’s project is designed to elucidate the text in the framework of Tibetan Buddhist psychological principles, but she also occasionally enters into discussions of its resonances with Freudian and other psychoanalytical theories.

Clifford’s discourse is strongly influenced by Mahāyāna assertions of the self-reflexivity of “demons.” For example, demons are said to be linked to “the devil of the aggregates [skandha-]—frailty of body and mind; the devil of the kleshas [klesa-]—the devastating power of the afflictive emotions; the devil of pleasure—the alluring trap of comfort; and the devil of death.”¹¹⁵ “In a psychoanalytic sense,” she adds, “these demons are in the role of the id trying to obstruct the super-ego’s higher promptings.”¹¹⁶ The legitimacy of such lightly contextualized statements may be questioned, as they are voiced in an attempt to expose, with a demonstration of Western moorings, the Mahāyāna declaration of the nonexistence of demons (or of any independent principles). To this end, she states that demons (Tib. gDon) are, externally at least, “forces whose existence is a coagulation as a direct result of bad and particularly poisonous karmic vibrations … that on some levels they are the embodied forms we give to our negative projections …, the imprints of mental habits and thought patterns whose unconscious hold is so strong that they are projected, unawares, onto the world.”¹¹⁷

Nevertheless, as a point of comparison, Clifford seriously considers the views of the early Protestant psychiatrist Johann Christian August Heinroth (1773–1843) that mental disturbances were caused by sin (cf. AH 6.4.4 above, prajnāparādhaḥ), by transgressive behavior, meaning “selfishness, the instinctual and intellectual self-aggrandizing ego.”¹¹⁸ This “sin,” in Clifford’s understanding of Heinroth, results from a sense of guilt and inner moral conflict, from knowingly transgressing cultural codes. These mental or psychological substantialities (which, once again, may be linked to Marriott’s “substance codes”) are “demons” that overtake and possess an individual.

Clifford also considers Freud’s declaration “that the Oedipal complex was at the core of every neurosis and that prototypical anxiety was generated by our earliest separation experiences. He managed to place sexuality and the anxiety surrounding it smack at the center of psychopathology.”¹¹⁹ The Tibetan Buddhist tradition does not disregard this, she says, pointing out that “desire, lust, and craving are the basic mental-emotional feelings that affect the inner humoural winds, and the stability of the mind is directly related to the condition of the winds.”¹²⁰ Clifford defends the Buddhist stance, arguing that sexual intercourse was occasionally prescribed as therapy and that disciplining the sex drive was part of spiritual practice, rather than “repression in the sense that we know it.”¹²¹ The defensibility of such statements is suspect, as is Clifford’s attempt to demonstrate that the Gyu-zhi recognizes something akin to the Oedipal complex in its assertion that the family into which an individual is reborn is dependent on the “attraction to one parent and aversion to the other.” A few randomly selected statements like this in Asian medical texts, while interesting and tantalizing, do not add up to a “complex” in psychoanalytical thinking, in part because they do not share the same analytical discourse or scientific paradigm.

With respect to exorcism (bgegs-bzlog-chog), Clifford maintains the standard Mahāyāna position, that the afflicted individual must understand through Buddhist practice that these demons do not “really” exist, even if exorcism is otherwise mediated through a variety of ritual procedures, supplemented by appropriate pharmacology. Although the trained lama who serves as the ritual officiant “has compassion for ghosts and wishes to liberate them,”¹²² this is quite different from what Vogt found in Sri Lanka, where the objective was “bringing the demons under the control of Dhamma.”¹²³

With respect to bhūtavidyā, afflicting spirits are often said to be victims of suicide, murder, or other forms of sudden death (akālamṛtyu) and are thus afflicted by extreme clinging and panic. Other kinds of demons or negative entities are the projections created by sorcerers, advanced psychics, or black magicians. Regardless of their point of origin or immediate symptomatology, the afflictions themselves are “the result of obscuration of mind.”¹²⁴ The sheer number of these geks (bgegs) or obstructing entities is staggering: 84,000 (a standard South Asian large number) of which 1,080 are deemed sentient and malignant spirits (gDon), broken down as 360 said to arise from ignorance, 360 from desire, and 360 from hatred. They are also classified as arising from above, below, and in-between. Others are “rulers of planetary forces” (gza-yi gdon), comparable to the grahas that are the possessing energies of the planets in Indian astrology. These are very difficult to treat medicinally, but require therapy in the form of mantra, meditation, and ritual (daivavyapāśrayacikitsā). In a passage that probably does not have an Indian cognate, Clifford states (without textual attribution): “One Tantric medicine used for these diseases is the fresh blood of a murdered person on the tip of a knife. If it is immediately touched to the lip of the patient, he will recover (dried blood will do if necessary).”¹²⁵ Clifford writes at length of the entities from above, below, and in-between, of diagnosis by pulse, skin tone and texture, and urinalysis—information distilled from the practice of modern Tibetan doctors as well as the classical texts.

Chapters 77–79 of the Gyu-zhi describe three types of possessing entities. Chapter 77 contains a list of eighteen “elemental spirits” (’byung-po’i gdon) that in name largely resembles the list from the AH and appears to be drawn from it with several carefully considered emendations. The descriptions of these spirits, however, can be quite different. Three examples should suffice. On gandharvas (Tib. dri-za [scent eaters]), the Gyu-zhi says: “They come from the zone of scents. They are attracted to sweet smelling things, but also to dung hills and other foul smelling places.”¹²⁶ The effect that these beings have on those whom they afflict is that “they are graceful and delight in fragrant smells. They like to sing, dance, and play. They love to wear nice clothes and are attracted towards red ornaments.”¹²⁷ Compare this to AH 6.4.18–19b: “One whose behavior is good, who has a sweet scent, who is blissful, who engages in song and dance, who takes delight in bathing and gardens, who wears red clothes and garlands, and adorns himself with red oils, who enjoys erotic play, is said to be inhabited by a gandharva.” Nāgas (Tib. glu), the Gyu-zhi says, “have serpent bodies with human heads. They live in water and trees. The deities among them guard great Dharma treasures beneath the ocean. They are extremely powerful and dangerous when angry.”¹²⁸ Their effect on those whom they afflict is that they “have radiant faces and red bloodshot eyes with straight piercing stares. They desire the whites [curds, milk, and butter] and the reds [presumably blood]. They flick their tongues and sleep face downwards.”¹²⁹ Compare this to AH 6.4.19c–21b: “One whose eyes are red, who is angry, whose gaze is fixed, whose gait is crooked, and is unstable, whose breathing is incessantly short, whose tongue dangles and shakes, licking the corners of his mouth, who likes milk, jaggery, bathing, and who sleeps face down, is regarded as inhabited by serpent demons [uraga], being fearful of sunlight as well.” The Gyu-zhi describes ro-langs [zombies] as: “A dead body infused with an evil spirit. Some cannot bend from the waist; they have enormous strength.”¹³⁰ In fact, this is the usual interpretation of vetāla from Sanskrit literature, especially the “dead body infused with an evil spirit,” from the Vetālapañcaviṃśati. Those who are afflicted by them “speak the straight truth, sleep much, like ornaments, and have shaking bodies,”¹³¹ This is similar to the AH’s definition of a vetāla (6.4.40): “One who is fond of fragrant garlands, speaks the truth, shakes, and sleeps a lot is believed to be under the control of a vetāla.”

It is important to remember that formally, if perhaps not in practice, the Tibetan text recognizes, through use of the word gDon for these demons, that what possesses an individual is the effect or negative force of each of these, not the entity itself.

As an addendum to the translations, Clifford provides “A Few Parallels to Modern Psychiatry.”¹³³ They are just that—parallels, and, to her credit, Clifford does not attempt to draw the entire enterprise into a psychoanalytic model. In fact, like the Indian texts, many of the categories in chapter 77 of the Gyu-zhi resemble personality profiles rather than pathologies. Clifford states that the effect of these “elemental spirits—the sudden onset of behavior of a consistently alien nature with little physical disturbance and no perceptible organic causes—corresponds to what the Western model … classifies as functional psychosis.” This statement is of limited scope, though she continues that the effects “suggest schizophrenia and manic depressive illness.”¹³⁴ In sum, though Clifford correlates certain categories of the Gyu-zhi with Western, notably Freudian, categories of psychopathology, it is not an all-embracing attempt. And this is one of two elements that confer on it slightly more credibility than Weiss’s study; the other is that she does not predicate her study on the assumption of correspondence.¹³⁵

Since Murphy, Weiss, and Clifford conducted their research in the 1970s, a number of studies have argued forcefully for the “social construction of emotion,”¹³⁶ a powerful notion dealt with here in Chapter 8 that must be considered carefully before generalizing psychological affliction across cultures. This is not the place to enter into a full-scale study of such construction or such emotion, except as they bear on the topic of possession. Clearly, both the Indian and Tibetan classical medical texts articulate emotionally thick symptomatologies in their discussions of bhūtavidyā. But these pools of emotional complexity and density are not always regarded as negatively afflictive, even if, taken to extremes, they are sufficiently transformative to become realized as possession. There appear to be trigger points after which a personality configuration emerges as either affliction or possession. These points are not articulated in the texts; they do not codify medically identifiable points at which any of these bhūtas may be conclusively diagnosed in an individual as possession. Yet the assumption is implicit that at certain points this boundary is crossed. Determining these points requires a more complete analysis of the sociocultural aesthetics of classical India (and Tibet) than medically oriented researchers with agendas that depend on comparative taxonomies have hitherto conducted.

One approach to this may be to recognize that one of the key constructive elements of emotion from culture to culture, if indeed we are warranted in such a generalization, is the manner and degree to which emotion is somaticized. By this I mean that different cultural archetypes, worldviews, political formations, and practices of everyday life contribute to the manner in which emotion is embodied, understood, and recognized within a cultural system.¹³⁷ An emotion that remains hidden to the outside world reflects a certain conditioning that encloses and embodies it uniquely, while the transmutation and activation of an emotion or emotional state into a bhūta of any definition reflects different cultural parameters of conditioning, recognition, and expressiveness. In this sense, anxieties that express acute emotional distress, such as fear (bhaya), apprehension (cintā), envy (īrṣyā), sexual excitement (kāma), anger (krodha), or grief (soka), are somaticized as psychogenic or physical symptoms that express themselves in different ways in different societies. But in India they emerge in a unified sensory manner as possession when they achieve a certain critical mass that is specifically defined yet remains elusive and is impossible to quantify. Thus, it is unsafe and unfair to guess the psychophysical coordinates of the threshold beyond which emotion is somaticized into possession or debilitating affliction.

This is further complicated upon recognizing that the processes of emotional somaticization in India are not referenced in medical literature, except as the texts state that an extreme emotion may result in a debilitated humor. At best one can gain a tentative understanding of these processes through a study of allied or corresponding cultural processes and the normativization of these processes through religious doctrine and practice. Thus, one must be guarded in essentializing such broad and ill-defined processes as the somaticization of emotion. It is perhaps safer to avoid the issue entirely, at least in the present context, and limit our comments to merely recognizing that the cultural configurations of such somaticization serve as the basis for descriptions not only of bhūtas but also of unmāda in the āyurvedic and tantric texts. As we have noted, most Western thinking on the topic of demons and possession has been shaped by a century of Freudian opinion, in which demons and associated phenomena are conceived as repressed traumas and developmental conflicts. This conviction, supported by a wealth of scientific data, is based on a single virtually unquestioned paradigm of scientific understanding. The results are then projected by psychologists, historians, anthropologists, literary critics, and many others onto alien cultures, whether ancient or modern, with little regard for the indigenous dimensions, gestalt, or characteristics of emotional embodiment, or of conflicting scientific paradigms that may be employed in these cultures (e.g., humoral diagnostics in Greek and āyurvedic medicine, circulation of qi, or functional energy associated with breath and movement, in Chinese medicine along a system of meridians not recognized by Western science). In this way, the absence of reference to repressed trauma and developmental conflict in Āyurveda and Tantra is barely considered by Western-style academics. To do so would be to admit the possibility of a different and little understood sphere of psychodynamics, as well as of a different scientific paradigm, in which embodiment is accorded a more active and transmutable status. Nevertheless, both Western and South Asian images of psychodynamic processes reflect rich symbol systems, behaviors pregnant with hidden meanings, and an axiomatic acceptance that self and person are not limited to the reflections of our conscious mind.

Perhaps unexpectedly, the ethnographies of oracular possession, and to some extent of disease-producing possession, support the āyurvedic and tantric viewpoints, as well as the viewpoints of formal aesthetic theory (alaṃkāraśāstra), in a programmatic pursuit of emotional definition. What is revealed is a nakedness of emotional expression, a clarity borne of a quest for emotional purity and conciseness that is somaticized in a relatively uncomplicated manner. This means that the quest for emotional definition is not too deeply enshrouded and caught in repressed trauma and unresolved conflict, but reveals a “natural” momentum that can be identified and defined. This does not mean that emotions generated through possession are unadorned in their nakedness. Indeed, emotions that are envisioned as part of secondary constructions—and possession definitely falls in that category—are identified by theoreticians from Bharata to Abhinavagupta to bhakti philosophers, such as Vallabhācārya and Rūpa Gosvāmī, by their adornments: the gestures, garb, and language that help express it. Alaṃkāraśāstra specifies adornments in poetry, drama, and dance; yet these are not entirely the constructions of Sanskrit paṇḍits. They are based on observation of emotional expression around them, which is to say on local context. In this way, both unmāda and bhūtavidyā must be regarded primarily as social and intellectual constructions that are primarily referential to local contexts. This is why I hesitate to transfer onto either Indian textuality or the sociocultural practices of emotional somaticization on which the texts are based a body of psychodynamic perceptions based on a scientific paradigm alien to Indian local context.¹³⁸

I have limited the above critique to medical texts because Weiss and Clifford have offered the most important theoretical contributions to the consideration of negative possession in South Asia available to date, and they both do so in the context of classical medical texts. However, bhūtavidyā has a broad panorama in South Asian culture, and much of it lies in areas peripheral to these texts. Therefore, in the interest of providing a more complete portrait of negative possession in South Asia, let us consider a few other demonologies, from both texts and ethnographies.

The large number of demonologies in Indian literature and culture vary greatly in content, by region, and according to genre. And, as is the case with demonologies found in first-millennium texts, the lists cannot be called fixed or stable. It is certain that different informants from the same area, cultural, or textual tradition offer different lists. It is outside the scope of this volume, indeed it would be impossible, to attempt to relate them all systematically.¹³⁹ However, a brief précis of this ever-shifting terrain is relevant to illustrating “traditions” of bhūtavidyā. This is best managed by presenting several examples out of what are unquestionably dozens of demonologies, or even hundreds if one considers the idiosyncratic views of one or two texts or religious authorities.¹⁴⁰ Thus, we briefly examine five demonologies, presented chronologically: (1) the list contained in the Bower manuscript; (2) the list presented in the Isanasivagurudevapaddhati; (3) a list found in the Madanamahārṇava, a text on karmavipāka (karmic ripening); (4) a list presented in an ethnographic account of healers in Varanasi; and (5) an account of a demonology found among the Divehi of the Maldive islands.¹⁴¹

The Bower manuscript is so-called because it was sold to a British lieutenant named Hamilton Bower while he was staying at Kuqa, an old Silk Road trading stop on the northern rim of the Taklamakan desert in northwest China, in 1890, while engaged in a desperate search for the murderer of another British soldier. The manuscript consists of seven fragments, including three on āyurvedic medicine, two on divination with dice, and two on incantations to combat snakebite, all dating from the late fourth or early fifth centuries C.E.¹⁴² This is later than Caraka and Susruta, and, in one of the incantations against snakebite, it provides a list of twenty-one types of seizure (grahāto) caused by various ethereal beings. The list appears as part of a long mantra called mahāmāyūrī (the great peacock); indeed, the peacock has long been regarded as a consumer of poison. In this Buddhist text, the Blessed One (bhagavaṃ) recites the mahāmāyūrī for the benefit of his disciple Ānanda, who is instructed to use it to cure the mendicant Svāti, who had been bitten by a large black snake (mahatā kṛṣṇasarpena). In addition to serving as an antidote to snakebite, this mantra mentions other uses for it, including against seizures.¹⁴³ These are “seizure [grahāto] by a deity [deva], a serpent [or chthonic] spirit [naga], a titan [asura], a god or spirit of the wind [maruta], a sky spirit [garuḍa],¹⁴⁴ a celestial musician [gandharva], a horse-headed spirit [often under the control of Kubera] [kinnara], the spirit of a great serpent [mahoraga], a semidivine protective spirit [yakṣa], a demon [rākṣasa], a spirit of the dead that has not received full [or any] post-mortem rituals or offerings [preta], a flesh-eating demon [piśāca], a ghost [bhūta], demons with jar-shaped testicles [kuṃbhāṇḍa], a foul smell or foul-smelling spirit [pūtana],¹⁴⁵ the spirit of a negligent kṣatriya [kaṭapūtana], a childsnatcher [skanda], madness [unmāda], a shadow spirit perhaps causing nightmares [ chāyā],¹⁴⁶ convulsions [apasmāra], and the evil eye [ostāraka].”¹⁴⁷

Among these, the maruta, garuḍa, kinnara, kuṃbhāṇḍa, pūtana, kaṭaputana, chāyā, ostāraka, as well as the unmāda and apasmāra, are themselves kinds of grahas and are not found in the earlier lists, though certainly most of these were known from other contexts. Maruta and garuḍa are included because they explicitly suggest airborne grahas;¹⁴⁸ kuṃbhāṇḍa is likely a local attempt to make sense of the words kuṣmāṇḍa or khārkhōda; kinnara and pūtana are known in Purāṇic lists, the MBh, and elsewhere; kaṭapūtana is mentioned in Manu (12.71);¹⁴⁹ and the latter four, chāyā, ostāraka, as well as unmāda and apasmāra, are clearly reifications of their literal significations. This testifies to the geographic range of Indic demonologies, which were shared by Buddhists and the medical traditions of Central Asia and the Silk Road.

The twelfth-century ĪŚP contains four chapters (paṭalas 41–44) on disease-inducing possession, specifically on childsnatchers (bālagrahas), behavior of the possessed (grahaceṣṭā), exorcism (bhūtabādha),¹⁵⁰ and seizure disorder (apasmāra), respectively.¹⁵¹ As mentioned earlier, this text was quite popular in certain circles of medical and tantric practitioners in Kerala. The first thirty-six verses of paṭala 42 succinctly identify eighteen specific types of graha. This paṭala also links other grahas with the four varṇa-groupings, provides succinct descriptions of humoral imbalances complicit in possession, and offers a description of apasmāra. Because of its brevity and usefulness as a twelfth-century tantric point of comparison with the language of possession and identity of grahas found in classical āyurvedic texts, it is relevant to translate these verses.

1–3. Eighteen kinds of graha take hold of people: (1) nisteja, (2) deva, (3) asura, (4) nāga, (5) yakṣa, (6) gandharva, (7) rakṣas, (8) pitṛ, (9) heḍhra, (10) bhasma, (11) yonyudbhava, (12) kasmalaka, (13) pralāpa, (14) vighna, (15) kuśa, (16) antya, (17) piśāca, and (18) bhūta. All of these, as well as brahman and other [varṇa-based] rākṣasas that are complicit in madness [unmāda] and convulsions [apasmāra], and that desire sacrifice [balikāmāḥ], sex [ratikāmāḥ], or killing [hantukāmāḥ] grab hold of men.¹⁵² They may be divided into two categories: gentle [saumya] and fiery [āgneya].

5b–8b. They afflict [pīdayanti] people who are angry, excited, laughing, afraid, alone at night; are either in a state of impurity, ostentatiously attired, or gain or lose money at night; those who are separated [from their loved ones] or about to die; women who are naked, pregnant, ill-behaved, in the course of having their menstrual period, have bathed immediately after the cessation of their period [thus preparing for sexual intercourse], have just given birth, are afflicted by lust, are drinkers of liquor or eaters of parched grain, who stand at a crossroads at sunrise or sunset, who have not before experienced [sexual] enjoyment, have just been oiled [as in massage], or are disrespected.

This section from the ĪŚP deserves much more comment than can be accommodated here. Nevertheless, I must offer a few words. First, it provides more detail on the vulnerability of women to spirit possession than is found in the earlier canonical āyurvedic texts. It is not that this supposition of increased vulnerability necessarily intensified over the centuries; it may simply be that the ĪŚP was more prolix in negotiating its various dimensions. In fact, as we have seen in the BĀU, references to women’s possession, hence their vulnerability, may be dated to the mid-first millennium B.C.E. or earlier. The somewhat later discussions of bālagrahas in the MBh and the early medical literature also reveal that women were viewed by the later centuries B.C.E. as highly susceptible to possession. The standard attributed causes of this were their monthly release of blood, regarded as attractive to spirits, and their childbearing capacity, in which protective measures to guard newborns against threats both visible and invisible were extreme. The latter was doubtless due to comparatively high infant and childhood mortality rates. Regardless, the disadvantaged condition of women worsened, or at least hardened, over the course of the first millennium C.E. to the point that they were equated with sūdras in much of the Dharmasāstra literature of the day. Thus, it is nearly irresistible to argue that in a patriarchal society that was increasingly oppressive toward women, their stigmatization was more pronounced in the early second millennium, the date of the ĪŚP, than it was earlier. This would encourage us to project back by at least a millennium the arguments of Lewis and others regarding the correlation of oppression of women and spirit possession, which, as discussed earlier, have a good measure of validity in modern-day India.¹⁵⁵

Second, although the presence of varṇa-based grahas is much more evident in the ĪŚP than in the medical texts (note the brahmarākṣasa, kṣatriya-graha, vanig-graha, vṛṣala-graha, antyaja-graha), the influence of the AH and Susruta is nevertheless clear. Third, the variation in clearly defined psychophysical disability is more pronounced. Thus, the heḍhra(ga)-graha, bhasma-graha, and yonija-graha, not mentioned in the earlier medical texts, look less like personality variants than different forms of congenital or neurological disease, or mental retardation.

The Madanamahārṇava (MM), written by Viśveśvara Bhaṭṭa in the fourteenth century,¹⁵⁶ is one of the two most important works of a little-studied genre that falls between Āyurveda and dharmasāstra known as karmavipāka (the fruition of karma). ¹⁵⁷ This literature correlates diseases with actions (karman). Sometimes these are actions said to have been performed in previous births, but by no means is this always the case. The remedies for these conditions consist of ritual expiation (prāyascitta) rather than medicinal preparations, leading, the texts say, to full recovery (niṣkṛti).¹⁵⁸ The fortieth and last chapter (taraṇga) of the MM describes symptoms of possession by thirty-one different grahas, all the result of ill-begotten past karma, most—if not all—of it accumulated in the present birth. In addition to describing symptoms, the text speculates on their causes and prescribes ritual remedies.

The names of many of these grahas are unusual. They are unlike the names of grahas in the other texts and ethnographies examined here, which are largely the toponyms of general categories such as rākṣasa or piśāca, ruthless feminine grāhīs such as Vinatā or Mukhamaṇḍikā, or the frightening bhūtas of Varanasi (see below). The names of the possessors here often do not have obvious connections to the disease or affliction described. Rather, they suggest a brahmanical normativizing, as their names, hence identities, appear to have been sanitized, blunting some of the impact of the afflictions described. Furthermore, the classifications betray a pedantry that foregrounds orthodox morality.¹⁵⁹ This has the effect of conferring on the ritual treatments a sense of formulaic brahmanical efficacy. Because of this framework, the only one I know of that consciously applies principles of sastra to possession, the possibility of endogenous pathology is never considered as a cause of the affliction being discussed. The unyielding agenda, then, is to suggest that questionable moral judgment, contiguous with the appearance of suspect symptomatology, plays a crucial role in weakening the individual, encouraging the hostile takeover of the individual by an untoward spirit.

As we have seen repeatedly from the vedic period to the present, in India no less than elsewhere, adherence to normative social and moral codes programmatically sustains cultural foundations, providing areas of comfort and control for cultural elites and ruling classes. In India, perhaps more than elsewhere, however, socioreligious law (dharma) has been self-regulating, requiring marginally less external control. Nevertheless, it has not always been a procrustean bed on which people bind themselves, as often depicted by scholars and cultural advocates alike. There have always been mechanisms of escape, and these have either been recognized by socioreligious codes or these codes have turned a blind eye to them, knowing that they are necessary for the maintenance of cultural harmony. Not everyone fits into the system. So, rather than forcibly displacing or suppressing those who are ill-fitting, the system has served up and sanctioned these escape routes. Specifically, transgressiveness in India has found culturally acceptable norms and mechanisms in the forms of wandering mendicancy and forest dwelling through religiously inspired renunciation, transgressive modes of sexuality or ritualizing through tantric practice, or, appositely, locating niches for abnormal or resistant personalities through an edifice of multiple or dissociative personality–inspired bhūtavidyā. Oracular possession, as we have seen, takes advantage of these well-articulated loopholes in a positive religiously acceptable manner. However, negative possession does not and cannot always locate or negotiate these loopholes. It is, therefore, dealt with by texts that address abnormal bahavior. These include not just āyurvedic texts but more programmatic texts on the margins of Āyurveda, including the MM. The usual result in the event of such abnormality is an unwitting victim in need of exorcism and healing. We see this in a range of texts and examples, from the paradigmatic possession of Nala by Kali, a child possessed by a bālagraha because of a lapse by the mother or another relative, or an enemy hiring a tantric to curse an individual with an evil spirit, a belief that Dwyer has shown holds sway over large numbers of victims of possession at Bālājī. In all these cases, even if unintended, there is the construal of dharmic or moral lapse. This, then, backgrounds the “ripening of karma” that requires ritual treatment.

The fortieth taraṇga of the MM consists largely of a prose commentary on material from two earlier karmavipaka works, now lost, the Karmavipākasaṇgraha and the Karmavipākasamuccaya. The taraṇga begins, typically, with general rules (paribhāṣā) for ritual exorcism or expiation, include fasting, chanting (japa) brief mantras and names of deities, reciting simple vedic hymns or verses including the Puruṣa-sūkta (ṚV 10.90) and the gāyatrī mantra (ṚV 3.62.10), though non-vedic mantras are also prescribed. The MMprescribes a general expiatory mantra, unattributed to any other text, in which the name of the graha is replaceable: pragṛhīṣva baliṃ cemaṃ dāharūpa mahājvara | āturasya sukhaṃ siddhiṃ prayaccha tvaṃ mahāgraha || (O, Great Fever [graha] in the Form of Fire, possess this offering [bali]; bestow, O Great Graha, comfort and fulfillment on this afflicted person). The second line remains unchanged throughout, while the names of the different grahas are substituted in the second half of the first line, for example, pragṛhīṣya balim cemaṃ aikāhika mahājvara, pragṛhīṣva baliṃ cemaṃ karṇākhya tvaṃ mahāgraha.

The principle of transfer, found throughout Indian religious texts and analyzed in Chapter 5 as “transfer of essence,” is also seen here. The graha is neither killed nor dissolved nor destroyed, but is transferred into the offering material used in a small localized sacrificial offering called bali.¹⁶⁰ This recommended bali consists of offerings of fragrant red flowers, oil, incense, and various other material, placed in a new brass pot (kāṃsyapātra, bell-metal), which is then left in a temple (devalaye) or at a crossroad (catuṣpathe). In addition, homa is often prescribed with offerings of cooked rice (caru), black sesame seeds, and parched grain (laja). On some occasions, well-known homas, such as the kūṣmāṇḍa,¹⁶¹ are recommended, as is the gifting of gold and other valuable objects to brahmans. These rules are set in the context of exorcistic expiatory ritual for the first graha on the list, the prajagraha (offspring seizer). This graha possesses a man who eats leftover food or has sex with his wife while she is having her menstrual period. It is noteworthy that many of the grahas on the list in the MM possess their victim for this offense. The victim then develops a fever, diarrhea (atisara), skin disease (kuṣṭha), and palsied limbs. It is not clear why this graha is called prajagraha.

The victim of a kālanāyakagraha suffers from sudden dryness in the mouth and his sounds become harsh. To alleviate this condition and exorcise this graha, one should offer a mixture of three sweet substances—sugar, ghee, and honey—along with mango sprouts 1,008 times while chanting the jātavedasa mantra (ṚV 1.99). He should also offer as bali scented red flowers and other normal material, as well as powdered parched grains, oil cake (piṇyāka), black sesame seeds, and as much gold as he can afford, in a brass pot at a crossroad. While offering this bali he should recite the appropriate variant on the usual mantra: pragṛhiṣva baliṃ cemaṃ lokanātha mahāgraha | āturasya sukhaṃ siddhiṃ prayaccha tvaṃ mahāgraha || (O, Lord of the World, Great Graha, possess this offering [bali]; bestow, O Great Graha, comfort and fulfillment on this afflicted person).

The āpastambagraha is explained as a spirit (graha) that possesses an unworthy person who sits in a row of brahmans and, because of excessive pride, takes for himself the best plate and consumes special food. This graha is capable of afflicting the transgressor with heart and stomach problems, as well as with consumption. The afflicted person can turn back this possession by performing the cāndrāyaṇa fast and making an offering (balidāna) at a crossroad with the accompaniment of mantras. This, however, does not explain why the graha is called āpastamba. To the best of my knowledge, this is the only occurrence of the word āpastamba that does not refer to the ritual school of that name belonging to the Taittirīya śākhā of the Kṛṣṇayajurveda.¹⁶³ It is not out of the question, however, that pejoratively it does refer to that ritual school, so-called because a brahman of that school once behaved in this way in a community of non-Āpastambins. The Āpastamba school is the dominant vedic sub-sākhā in Andhra Pradesh, Tamilnadu, and many areas of Karnataka and Maharashtra. The ritual school of the reputed author of this text, Viśveśvara Bhaṭṭa, is not given in the colophon or commentary, but in all likelihood he was not an Āpastambin, because he was from North India.¹⁶⁴ Thus, he was probably a member of a different school and had a negative experience with a ritualist of the Āpastamba school, who might have accepted gifts in a manner Viśveśvara Bhaṭṭa judged arrogant.

The victim of a vṛtragraha, an affliction that apparently harks back to the archetypal serpent demon Vṛtra (cf. ṚV 1.32), suffers from dryness of the buttocks, a possible reference to a sexually transmitted disease.¹⁶⁵ As exorcistic expiation, he must perform the prājāpatya fast, offer a homa of black sesame seeds with the gāyatrī mantra, recite the gāyatrī an additional ten thousand times, and give away as much gold as is possible. These are very ordinary expiations, particularly compared to the often-exotic prescriptions in the Āyurveda texts and the rich variety of practices documented by ethnographies.

The victim of a kumbhakagraha (also called jambhakagraha) suffers from fever, incoherent speech, and symptoms of asthma. The exorcism consists of the cāndrāyaṇa fast plus offerings of cooked rice with ghee (carusarṣipī) into a fire while reciting the Puruṣa-sūkta with the vyāhṛti mantras (the syllables bhur bhuvaḥ svaḥ).¹⁶⁶ This and similar passages suggest that offerings into the fire, supplemented by fasting, are in fact āyurvedic remedies. Nearly all possession is classified as vāta disorder in āyurvedic texts, and certainly the uncontrollable body movements and mental disconnectedness that partially characterize possession support this diagnosis. Kapha-producing (and therefore grounding) food such as cooked rice with ghee is often prescribed to combat vāta debilitation. It is not unreasonable to suggest that the exorcisms prescribed here constitute a kind of daiva Ayurveda, which is still practiced in a few places in India. More specifically, these practices could easily fall within the category of daivavyapāsrayacikitsā (spiritual therapy). Although I have not seen this stated in the primary texts, the implicit āyurvedic argument is that fasting combats severe vāta debiltation, while the offerings of rice with ghee, consistent with the ideology of the balidāna given in the beginning of the fortieth taraṇga of the MM, is designed to attract and capture the offending spirit.

The karṇagraha has a long history, some of which we have recounted earlier (see Chapter 11). This graha is not depicted as disease-producing by Varāhamihira or the commentator Bhaṭṭotpala. However, in the MM it is made to conform with the disease-producing norm, stripped of its earlier significance as an oracular spirit. Here, on the authority of Kātyāyana, the MM says that a karṇagraha possesses one who consumes temple property, causing sudden deafness, a freezing of the tongue, and great itching on the limb (= penis?). The antidote is the cāndrāyaṇa fast, homa with the vyāhṛti mantras, and recitation of a hymn from the ṚV (6.24) to Indra 1,008 times. This hymn has little obvious relevance to the exorcism of a karṇagraha, but is prescribed in the srauta ritual as a hymn to be recited in for slaying of the demon Vṛtra.¹⁶⁷ The balidāna to be offered to the graha consists of ground black sesame, wine (surā), meat, rice, parched grain, yogurt, and (probably) onion (kanda). This should all be placed at a crossroad along with a banner, scented flowers, and incense, over which a tantric ritualist (mantravit) should recite the relevant mantras.¹⁶⁸

Another challenging figure is the avatolagraha. The MM quotes the Karmavipākasaṇgraha as its main authority on this. This graha, according to the MM, possesses a person who is disagreeable and a social problem. The texts do not help us with the word avatola, however. This neologism might indicate deceitfulness, more literally “weighing down” or fixing the scales when weighing and selling any substance (perhaps a prohibited one).¹⁶⁹ The physical debilities produced by an avatolagraha include coughing, chest problems, aching limbs, and decreased appetite. The three-step exorcism is a mix of the usual prescriptions: recitations of the Viṣṇu- and Puruṣa-sūktas, performance of the kūṣmāṇḍahoma, offering of a thousand pots for bathing an image of a deity (sahasrakalaśasnānam), and recitation (japet) of a mantra (ṚV 4.31.1) 1,008 times. Like most of the expiations, this one assumes that the victim is a person of means, as these rituals require large expenditures.

Another ethereal enemy of moral lassitude is the śiśagraha. This graha afflicts its victim with fever, diarrhea, eczema (āsyaśośī), and shaking of the hands and feet. The name of this graha is striking because in both the ethnographic record and at modern spirit healing centers such as Bālājī the possessors are often identified by the families of the victim and the ritual authorities alike as children who have died young, their souls (jīva) tortured and unappeased, having not undergone the standard rites of passage (saṃskāra). Thus, an echo of popular possession is probably present here, though it is not stated. Once again, then, we are left with a name with no obvious link to the offense at hand. The victim will achieve peace if he offers as bali a pot filled with parched grain, liquid milk sweet (pāyasa), chicken, lamb, and gold. Along with this, he should hire a priest to offer bali of red cloth on a banyan root (vaṭamūle) with the usual mantra. Very few of these grahas take nonvegetarian food, but for some reason this one does. It is possible that this passage reflects a local practice of offering meat to spirits of deceased babies and children. Thus, in spite of a nearly crushing brahmanical palimpsest, the MM appears to reflect a propinquity of Sanskritic and local practices.

An important contemporary source of bhūtavidyā is K. M. Shukla’s ethnographic account from the environs of Varanasi of the exorcism of eleven kinds of piśāca responsible for different illnesses.¹⁷⁰ Two features of this account are immediately noteworthy. The first is that these bhūtas are extremely malevolent, more evidently gruesome than those in the lists examined above. The second is that Shukla’s study is deeply contextualized; most of the bhūtas are easily located in the prevailing cultural environment—which raises questions about the cultural environment that surrounded the composition of the earlier lists in the Sanskrit medical and tantric literature. These texts, of course, were not multifaceted anthropological tracts, but specialized texts safely within the disciplinary boundaries established at the time. However, it is this feature of broad-based cultural understanding that separates Shukla’s account from earlier lists of nineteenth- and early twentieth-century “orientalist” scholars (particularly those who also happened to be missionaries), which tend to be thoroughly decontextualized or only randomly contextualized.¹⁷¹

The piśācas are the following: (1) a cuṛail, a bloodsucking witch with feet turned backward, found in bamboo groves, bel trees (Skt. bilva [wood-apple, a tree sacred to Śiva]), or banana trees. Sheldon Pollock refers to this demonic creature as “the succubus of the Indian male’s nightmare world, who threatens him with death through sexual depletion and must therefore be suppressed”;¹⁷² (2) a savat or spirit of a deceased first wife, hence the rival of a second, who attacks the latter when she is not properly regarded; (3) a daitya, a nocturnal spirit that dwells in a pipal tree; the word daitya is commonly attested, and not, apparently, a relic of its usage in AH 6.4.16–17 cited above; (4) a brahma, or spirit of a brahman who dies an unnatural or untimely death (akālamṛtyu); (5) a haḍikaś brahma, the spirit of an unwanted newborn that is kept in a basket (haḍi) and buried; (6) a budnua,¹⁷³ the spirit of one who has drowned and attacks an individual who comes to bathe in that pond, tank, and so on; (7) a jin or jinnad, a Muslim spirit who resides in a palm or pipal tree or in a graveyard near a mosque; (8) a sahid, or Muslim martyr; (9) a bhavānī, an unmarried female who dies an unnatural death, then troubles children and other family members if she is not properly worshipped; (10) a marū,¹⁷⁴ the spirit of a person killed because of the wrath of the goddess (also bhavānī, but not identical to the spirit of that designation), who dwells in a house and causes trouble in the family; (11) a ḍīh, a minor spirit who fails to receive promised offerings. The offering of choice to a ḍīh is gāñjā (cannabis), promised to it after a cure is enacted.¹⁷⁵

The Divehi are an indigenous people of the Maldive islands, far to the southwest of the subcontinent, but well within the Indo-Islamic cultural orbit. Clarence Maloney, who conducted close and lingustically informed field-work there in the 1970s, states: “Within the [Divehi] village, jinnis cause spontaneous abortion, barrenness, sickness, epidemics, fainting, unexplained noises or any other sudden or unexplained phenomenon.”¹⁷⁶ Indeed, the general word there for bhūta is jinni. Maloney records forty-two varieties of Divehi jinni, many derived from Tamil or Simhala words, many others seemingly indigenous Divehi words, but eleven probably derived from Sanskrit. These are (1) devi (< Skt. devi), “a generic name for a personified spirit or godling, now applied to male or female spirits which were pre-Muslim deities”; (2) hanḍi (< Skt. caṇḍa, cruel; cf. Caṇḍī), a “generic name for jinni, especially smaller ones; usually but not always female; one may fly through the air or along the ground; it may make a sound like a tongue click, and has a ‘strong spirit’”; (3) furēta (< Skt. preta), “any large or fearful spirit; male or female; esp. an apparition of light over the ocean at night; bad people might become one; it may walk with a shroud”; (4) ferēta (< Skt. preta), “lights on the sea; like a ship full of lights; it may have an eye on the top of its head, so it has to bend forward to see; may catch people; as lights on the sea, often frightens fishermen”; (5) hāmunḍi (< Skt. Cāmuṇḍa, Cāmuṇḍī), which “may take any form, lives in village; disturbs domestic fowl; may be seen as a hen with chicks about it”; (6) buddevi (< Skt. bhudevi),¹⁷⁷ “a devi of a budu (idol)”; (7) dēyō, diyō (< Skt. devī), a “spiritual queen, a devi”; (8) kandā hanḍi (< Skt. Skanda), “a female; troubles men; makes sexual dreams and shows intercourse; it may have children by men, and makes men unconscious”; (9) avaṭeri (< Skt. avatāra),¹⁷⁸ “a kind of hanḍi that lives in the jungle; female; she has long hair and old clothes; if people leave out supplies in the kitchen at night she will grind the condiments, scrape the coconut, cut up the fish and fix the fire”; (10) furāna (< Skt. purāṇa), a “ghost of the dead;” and (11) ravo (< Skt. Rāvaṇa), “a devi that causes disease to children; makes them lean and troublesome”;¹⁷⁹ Maloney notes that jinnis “do not usually possess people as Hindu patron deities may.”¹⁸⁰ By this, he means that there is no evident oracular possession. This indicates that the perceived spirit presences are largely infestations; sources of trouble rather than ecstasy. However, further work must be done on this.

These multiforms of bhūtavidyā demonstrate that it is an enduring feature of the South Asian cultural area, both in time and space, from the vedic period to the present, from the Silk Road and Tibet to the Maldives. These extravagant, even baroque, demonologies nearly always suggest possession, as affliction, harassment, and influence by a bhūta or graha indicate this likelihood in the South Asian psychophysical and cultural systems. The vedic demonologies, like those of modern Varanasi or the Maldives, are integral aspects of folk traditions. The former, however, were captured and canonized by early Sanskritic culture and, therefore, left deep and long-lasting influences on the images and typologies of bhūtavidyā in all strata and nearly all regions of South Asia.

The more systematized demonologies, in the sense of being directly associated with an extended literary tradition, such as are found in BĀU 3.2 and the āyurvedic texts, are consequently concerned primarily with formal epistemologies. In the āyurvedic texts these are more reminiscent of human personality, at least of powerful elements of its relatively normative dark side, than of uncontrolled, irrepressible, and dangerous elements of an invisible nonhuman wilderness, in spite of the exotic labeling in these texts. In other words, the Sanskritic demonologies appear to participate more in the domestic than the wild. In a cultural sense, then, they bespeak the domestication of possession. It is as if relegating them to the realm of literature both sanitizes and scientizes them, which is why they appear both better defined and less dangerous than, for example, the bhūtas of Varanasi. The grahas in the bhūtavidyā lists in the āyurvedic texts may also be compared to the child-snatchers, who were systematized in the MBh and in other sections of the āyurvedic texts. The latter were clearly regarded as horrific and relentlessly dangerous and were, in this sense, closer to the demons of present-day Varanasi. In practice, however, medical practitioners did not rely strictly on the designations from the āyurvedic texts. There always has been, I believe, a spillover from popular culture to text, which may be seen in the increasing numbers of possessing beings from Caraka to Vāgbhaṭa, as well as a spillover from āyurvedic texts into popular culture. The latter may be seen in the list of childsnatchers in the MBh and tantric texts, as well as in some of the more recognizable bhūtas of Varanasi and the Maldive islands.

Because of the ubiquity of the piśāca, a few words must be said about it. An illuminating passage appears in the Nīlamata Purāṇa (NP), a text from Kashmir that depicts a social and spiritual universe densely populated with ethereal beings of nearly unimaginable variety, which readily interact with one another as well as with humans. These beings comprise devas, gandharvas, apsarases, yakṣas, guhyakas, kinnaras, nāgas, rākṣasas, piśācas, and millions of others.¹⁸¹ Pisācas enjoy an honored place in this pantheon. They roam the heavens and the earth, including the rivers, mountains, towns, and forests of Kashmir. Indeed, the geography of Kashmir is vividly portrayed, rendering this Purāṇa a valuable source of historical knowledge. It is also a handbook of popular festivals and astrologically based rituals. Although recent efforts have been made to historicize the NP, much work remains to be done.

The NP depicts gargantuan battles between hordes of these beings. NP 210ff. tells of piśācas in the employ of the gods. These piśācas are under the direct command of the piśāca lord Nikumbha, who has been hired by the god Kubera, himself commanded by Viṣṇu. These piśācas, numbering fifty million, are hired guns for the good guys and, therefore, themselves in some sense good, because they are allied with Viṣṇu.¹⁸² Their job is to battle an equally voluminous number of malevolent piśācas on calendrically regulated dates. In spite of their affiliation with Kubera and Viṣṇu, however, they remain dangerous to humans, thus deserving of their offerings. In one passage the goddess asks the sage Kaśyapa why she should attend to Kashmir, since it constitutes her very body, when there is more important work elsewhere. Kaśyapa replies that in Kashmir humans are always associated with piśācas, so she needs to see that their minds do not become perverted because of this association. Despite the watchful eye of the goddess, the NP (397–410) describes a highly transgressive festival in which humans become possessed by these good piśācas and appear to enjoy themselves immensely. Whether this constitutes defilement or perversion, or whether this matters, is undetermined.

The text of the NP has no clear divisions or properly marked narrative transitions. It moves in stream-of-consciousness fashion from the middle of one topic to the middle of another. Nevertheless, with respect to our story, we may surmise that, beginning on the full moon of the month of Asvina (mid-October), the celebrants should worship brahmans, honor their own ātman (probably indicating some form of meditation, mantra recitation, or brief ritual practice), then enjoy a vegetarian banquet and an evening concert or dramatic performance. On the following two days, they should bathe as usual in the morning, then attire themselves finely and perform appropriate fire rituals. The remainder of these two days should be spent relaxing in the company of friends. After the morning rituals, in contravention of the usual dictum not to engage in sexual activity during the daytime, the celebrants, followers (anuyāyinaḥ) of the good piśāca Nikumbha, should smear their bodies and those of their friends with mud, then engage in revelry and well-turned lascivious speech that is both sexually enticing and arousing (talliṇgārthaprabodhakaiḥ [brings about an erection]). They should shout vulgar words and cavort about. At that time, they become possessed by frightful-looking piśācas—the ones who work for Kubera, we later learn. In the late afternoon or evening, after the celebrants bathe once again, these piśācas exit the bodies of those who have properly engaged in this ritual and enter and curse those who have not. Then those followers of Nikumbha should worship Kesava. Afterward, they should keep an oil lamp lit outside their houses for one month, until the full moon of the month of Kārttika and observe certain ritual vows for six months.

It is not clear how much of this is protest, redemption, escape, or just good clean sybaritic and festive fun. It is possible that the sectarian and political history of Kashmir had an impact on this ritual, but this has not been researched. It is also not clear to what degree this mud-wrestling possession orgy has been reframed or fictionalized by the NP. I know of no equivalent ritual in modern-day India, though it is possible, even probable, that other sexual rituals, such as the tantric cakrapūjā, at one time had variants in which the dramatis personae became possessed by the goddesses of the cakra before engaging in ritual sex.¹⁸³ We can also posit that the possession in this ritual was protective as well as transgressive, the ritual possession being the safety net that thwarted adharma. As a ritual mechanism, this possession shifted the onus of transgressive sex from dharmically conscientious citizens to liminally relocated and reconstituted personalities. Although the sexual aspect does not, to the best of my knowledge, appear in modern possession ritual, group possession ritual is often documented in India, for example, in the Draupadī festival in Tamilnadu and the Siri festival in the South Kanara District of Karnataka.

Pollock’s remarks on the genesis and role of the rākṣasa are apposite here. He cites the Rāmāyaṇa to the effect that the concerns of rākṣasas are “to master the sports of lovemaking and hold crowded fairs and festivals.”¹⁸⁴ Their pursuit of lovemaking can be traced back to the Śatapatha Brāhmaṇa and their powers of transformation to the ṚV.¹⁸⁵ Pollock argues that rākṣasas are “the imaginative product of the confrontation of traditional Indians with their particular forms of desire—in its two primary forms, libidinal and aggressive—representing all that traditional Indians most desired and most feared.… [I]n their libidinized forms, they enact the deepest sexual urges—total abandonment to pleasure, as well as absolute autonomy and power in gratifying lust.”¹⁸⁶ This verdict applies as well to the piśācas in the NP, where ritual both domesticates and liberates them, providing their hosts an opportunity to release their pent-up sexual aggression in a context that not only does not threaten to corrupt or overturn established cultural standards but, through that context—namely, the ritual—reinforces them. This, then, is a ritual of reversal—in this case, reversing an established expectation of monogamy—and is rather like certain other rituals of reversal in North Indian festivals. What comes immediately to mind is the day during the spring festival of Holi on which the women of certain villages in Braj (e.g., Nandagaon, Gokul) beat their husbands and other men, immune, that day, from reciprocation.¹⁸⁷

The piśācas of the NP, at least the good ones, differ radically from the malevolent piśācas depicted in the āyurvedic and tantric texts cited above, where they are described as impure, dirty, desolate, restless, incapable of enjoyment, and so on.¹⁸⁸ The latter piśācas are ever threatening and require ritual expiation. A specific venue for such expiation, the Skanda Purāṇa informs us, is a certain Pāsupata Śaiva temple called Pisācamocana in Varanasi.¹⁸⁹ It is here, so the story goes, that once long ago, during the tretayuga, a great-souled ascetic named Vālmīki performed brahmanical rituals before the Kapardīśa liṇga, then sat for meditation on the banks of a nearby lake. He saw there a ghastly and miserable rākṣasa who described to him how he attained that sorry state. While alive he was a brahman temple priest in far-off Pratiṣṭhāna on the banks of the Godavari river, and while performing his priestly duties he accepted gifts from pilgrims (tīrtha-pratigraha), but kept them all, failing to donate any of them to the temple. For this he was condemned to exist as a piśāca. One day, while he was floating as a piśāca in Pratiṣṭhāna, a young brahman passed by. He failed to perform his sandhyā rites and did not purify himself properly after urinating and defecating. Like Kali possessing Nala, this piśāca passed into the body of this brahman boy (taccharīre ’haṃ saṃkrāntaḥ). Eventually that brahman made his way to Varanasi, for business rather than pilgrimage, but the ethereal guardians of the holy city refused to allow him into the inner shrine area (antargeha) without expunging him of his sins, along with the invasive piśāca, at least temporarily, both of which would reenter him when he left. So, the piśāca was stranded in the outer neighborhoods without a host. Just as it was about to seize and eat a healthy pilgrim, the latter uttered the name of Śiva, purifying the piśāca, which then followed the pilgrim to the Kapardīśa liṇga. There he was noticed by the ascetic Vālmīki, who gave him a small quantity of sanctified ash (bhasma, vibhūti) from his fire to apply to his forehead. This was sufficient to purify the piśāca to the point that it could be admitted for bathing in the sacred pond. As a result, the piśāca lost its frightful qualities and attained a divine body (divyadeham).

The Pisācmocan temple continues as a venue for exorcism. Two types of officiant serve those in need: the bhagat, who invokes the deity in charge of the exorcism, and the ojhā, who actually performs the exorcism. The important elements of the exorcism are, first, ritual procedure and, second, speech, which at Pisācmocan and elsewhere in Varanasi includes mantra and secret language. Among the substances employed are cloves, which, given to the client, serve to transport the pret away from the body, while lemons and sometimes rice-flour balls (piṇḍa) are used for bali or simple sacrificial offerings to the deity. Thus, the spirit, often that of an ancestor (pitṛ), is pacified, after which it is seated (baiṭhnā) in the temple, where the deity is deemed strong enough to hold it forever. Occasionally, the client will drive a nail into the trunk of a large tree on the temple grounds, fixing the offending spirit to it. It is important to note that the prevailing viewpoint is that the spirit rides or sits on the individual, not inside the person.

Many ojhās who frequent Pisācmocan are from or are connected with outlying villages. They bring their clients to Pisācmocan after working with them for lengthy periods, sometimes for years. Many of the clients are women who have suffered a miscarriage, influenced by the notion that an evil spirit has been sent by an enemy. Others are men—often brothers—enmeshed in legal disputes, believing that they have been cursed by spirits sent by the opposing party through hired tantrics. Many more are relatives of people who have died suddenly, a condition that is said to render the subtle body of the deceased vulnerable to possession by nasty spirits.¹⁹⁰ Most of the traffic at Pisācmocan, however, is for fairly ordinary postmortem rites (srāddha, piṇḍadāna). This is performed throughout the year, though preponderantly during pitṛpakṣa, the fortnight of the waning moon in the month of Bhādrapad (usually mid- to late September). This serves the dual purpose of feeding the ancestors, the usual intent, and exorcising them of evil spirits, just in case.

A section from the MBh (3.216–219) on the birth of Skanda and the origin of bālagrahas or childsnatchers was explored in Chapter 6. That passage contains much of the essential information found in the āyurvedic texts, including specialized works on pediatrics, descriptive lists of bālagrahas, diseases attributed to them, and remedies. Perhaps the most comprehensive of these passages appears in the Kāsyapa Saṃhitā (KS) of the seventh century, a text that now survives in only two fragmentary manuscripts, testifying to the fact that it was soon eclipsed by other texts, particularly the AH, despite sections like this one that are more comprehensive than the corresponding sections of the later texts. The passage in question is intact and has been translated well by Dominik Wujastyk.¹⁹² For this reason it is not necessary to enter into a full translation of it or corresponding texts.¹⁹³ The important point is that the fundamental interpretation of the action of grahas, including bālagrahas, at least in the classical āyurvedic texts (we have seen that the perspective in the Tibetan medical texts can be different), is that they are agents of possession. Other symbolic interpretations of the meaning of grahas and verbs used to describe their actions (such as abhigharṣayati [attack]) may be valid as well,¹⁹⁴ but they cannot override the centrality of possession or of moral missteps (even if inadvertent) or weaknesses that are consistently seen to lie at the basis of disease-causing possession. The KS describes many conditions, moral lapses all, ideologically consistent with what is found elsewhere, in which the archetypal childsnatcher, Jātahāriṇī, invades a pregnant woman and terminates her pregnancy. As expected, “only dharma can turn her away,” says the KS.¹⁹⁵

Rather than discussing the canonical āyurvedic texts, it will better serve our interests to examine paṭala 41 of the Isānasivagurudevapaddhati. With respect to the relationship between the ĪŚP and the āyurvedic texts (as well as with the MBh), it is notable that the former is consistent with the latter in rendering semidivine grahas called “mothers” (mātṛ) as bālagrahas that afflict children below the age of sixteen, especially fetuses, while, as we saw above, the ĪŚP is also consistent with the āyurvedic texts in regarding yakṣas, gandharvas, rākṣasas, and so on, as possessors that bequeath insanity to adults.¹⁹⁶

The author of the ĪŚP proposes that, out of the many traditions of treatment of balagrahas, he discusses two of them in brief: the procedures described in the Khaḍgarāvaṇa and in the Nārāyaṇīya.¹⁹⁷ The procedure described in the Khaḍgarāvaṇa enumerates several general procedures, then prescribes modifications depending on the age of the child and the identity of the mātṛkā (Little Mother), the offending spirit. The ritualist should take earth from both banks of a river and construct an effigy (puttalīm) from it in the shape of a child. This will then substitute for the child in the ritual, though it appears that the child should also be present. Then the ritualist should make a packet of cooked white rice along with fresh rice grains, fish, meat, and liquor, together with flowers, yogurt, milk, and black sesame seeds. The wrapper should probably be a strong leaf (this is not mentioned, but is usually the case) and should, according to the text, be tied tightly with yellow thread. The intent seems to be to combine pure and impure offerings, presumably for Gaṇeśa and Vīrabhadra and the ill-intentioned spirits. After the ground for the ritual is cleaned, it should be embellished with thirteen svastikas, eight ghee lamps, and different kinds of fruit, including bananas. One should then make a large vessel, in the middle of which one should place the effigy, then set lamps and other ritual items around it.

Then, at the time of the midday sandhyā, Gaṇeśa should be worshipped with incense and other appropriate offerings. Brāhmī and the other mātṛkās should be installed as the deities of the primary directions around it, and Vārāhī and other goddesses, including Caṇḍikā, should be installed and worshipped in the directional corners. After the tantric ritualist invokes and worships Vīrabhadra and Gaṇeśa at the center, the substances for sacrifice (bali) should be divided in half, with four parts each for Vīrabhadra and Gaṇeśa, and four parts for the goddesses, placed on pipal leaves. The prescribed mantra should be recited for each deity separately, and peaceful (sāttvika) offerings should be given to both Khaḍgarāvaṇa and whichever goddess is to be named, in this case Nandā as the possessor of children in their first year. The mantra is: o namo bhagavate khaḍgarāvaṇāya dīrghadarśanāya hana gṛhṇa muñca bālakaṃṭhaṭha (O, salutations to the illustrious Lord Khaḍgarāvaṇa whose vision is far-reaching. Kill! Possess! Liberate the child! Svāhā.).¹⁹⁸ Variants of this mantra are given for each of the offending mātṛkās. After these offerings are made and the child is exposed to them, one may take the child away.

The text then prescribes certain fumigants or demonifuges to be used as direct therapy on the victim. The substances that comprise these fumigants should be collected and burned in the presence of the victim in order to create an environment unpleasant to the spirit, hence encouraging its departure.¹⁹⁹ Fumigation therapy is still practiced in India, including in Kerala, Varanasi, Rajasthan, and Pune.²⁰⁰ I was informed by a well-known vaidya in Kerala who still manufactures and prescribes such fumigants that they should not be burned for more than fifteen minutes three times a day.²⁰¹ It is doubtful that the recipes used for fumigants today anywhere in India are identical to those listed in the AH, Caraka, Susruta, or other classical āyurvedic texts, or even in classical Tantras, such as the ĪŚP. In Kerala, the ingredients and processes employed in their manufacture are secrets closely held by the families of aṣṭavaidya Nambudiris who have been handed down these ancient medical traditions.²⁰² These physicians have studied the appropriate sections of the AH (but probably not Caraka and Susruta) and have a rough familiarity with the ĪŚP, though it is no longer formally studied or used as a ritual handbook in any of the surviving Kerala āyurvedic traditions.

The fumigation rite for an infant prescribes igniting neem leaves and ghee mixed with mustard seeds left over from offerings made to Śiva.²⁰³ This smoldering concoction should be carried around the infant clockwise, probably in an earthen bowl (though this is not mentioned). Neem leaves have long been regarded as a natural antiseptic,²⁰⁴ and, along with mustard seeds, they give off an intense smoke when burned. The ghee is prescribed because it helps ignite fire and encourages smoke to become dense when the flames die down. This combination of substances is apparently inhospitable to bhūtas, as it surely is to humans. At the same time, the ritualist should place sprouts germinated from nyagrodha, udumbara, aśvattha, bilva, and pālāsa seeds in a pot, rinse or sprinkle them with water for three days, and wash them thoroughly on the fourth day. The reason for this is not stated, but these are all “auspicious” plants, recommended for use in sacrifice (medhya) or as offerings to deities, and are regarded as conducive to the growth and renewal of the child.²⁰⁵ We may assume from this that the ritual should be repeated for three or four days.

The procedures and substances are identical if the affliction, this time by Sunandā, occurs in the child’s second year. The instructions for the third year’s exorcism state that ajaśṛṇga (this should probably read ajaśṛṇgī) should be added to the fumigant. This is a malodorous shrub whose fruit is said to be sharp as a horn, recommended as early as the AV (AVŚ 4.37; AVP 13.4) as a remedy for sore eyes as well as for destroying evil beings.²⁰⁶ For the fifth- to twelfth-year ritual, neem leaves should be burned (dhūpayet) mixed with ghee and guggulu, an aromatic resin prescribed to strengthen bones and connective tissue but also commonly used in homa because it ignites easily and burns continuously. In all cases, after the fourth day the ritualist should take a purificatory bath. According to the Khaḍgarāvaṇa, grahas can take hold of children up to the twelfth year. Īśānaśivagurudeva adds, however, that it can occur up to the sixteenth year, though others say bālagrahas can attack up to age seventeen. The text does not state what one should do with the puttalī after the exorcism is concluded, but, following traditional practice, it is likely that it was to be discarded in a river, temple tank, or other body of water considered sacred (tīrtha).

The ĪŚP does not give the rationale for the puttalī, though it is consistent with both earlier textuality of healing and current practice to infer that the disease, as well as the grāhī, will leave the patient or victim and enter the puttalī. The Kauśikasūtra, for example, prescribes a treatment for hepatitis in which a yellow bird should be placed in a bowl on the floor beneath the head of a person lying on a bed with his head extended over the end of the bed. While a certain hymn from the AVŚ (1.22) is being recited along with other mantras specific to the Kauśikasūtra, water should be poured over the person’s head so that it flows onto the bird.²⁰⁷ The disease then exits the patient and enters the bird. The notion that diseases might disappear but exogenous disease-causing agents continue to circulate is also attested at Bālājī, where possessing spirits may be temporarily transferred to a ritualist or clinician, as described earlier, and eventually be retrained as good spirits in the phauj or army of Hanumān. In that case, they permanently leave the body (and mind) of the client, whose psychosocial dysfunction gradually disappears as the grip (grahaṇa) of the bhūt-pret attenuates.

The (Viṣa-) Nārāyaṇīya, so-called because its first ten chapters deal with poisons (viṣa, primarily from snakes) and their treatment,²⁰⁸ mentions grāhīs not only for specific years of the child’s life but for the first ten days of life and the first twelve months as well. It does not mention a puttalī, but does prescribe the construction of yantras, the placement of divinities within them, and the offerings and primary mantras to be employed for them, some of the latter being quite long and complicated. After noting the name of the grāhī, Pāpinī, which might afflict a child on the day it is born, the text states that a square sacrificial altar (balipīṭham) should be consecrated. This should be performed in the home, though the alternative is given of a crossroad (catvari), a ritually potent locale that is prescribed as a site for balidana in the Madanamahāṛṇava. Vīrabhadra should be installed in the center and the goddesses (mātṛḥ) in the corners and primary directions. Gaṇeśa (vināyakam) and Cāmuṇḍī should be installed in the northeast and worshipped with grain cakes, fish, meat, cooked rice, sesame, powdered spices, and other similar items, including liquor, representing, as in the case of substances prescribed by the Khaḍgarāvaṇa, the spectrum from pure to impure, cooling to heating, right-handed to left-handed.

As before, fumigants or demonifuges should be burned, but here the ingredients may consist of more exotic substances. A simple demonifuge is prescribed for the first day: neem leaves dipped in ghee, then mixed with shoots from the uśīra, (apa)marga, and pipal trees. More difficult is the one prescribed for the second day. After rubbing an ointment of sandalwood, uśīra leaf, mayura leaf, and goat’s urine on the baby, a demonifuge of cow’s teeth, horn, and hair should be burned. More difficult still is the one prescribed for the eighth day. After an ointment of goat’s urine mixed with vacā, a “hot” root often recommended for bestowing mental clarity or sharpness, and kuṣṭha, a very bitter plant, is applied to the baby, a demonifuge of tiger’s nails is burned. On the tenth day, monkey’s hair and nails are burned, and the ritual for the second month consists of, among other things, the burning of a fumigant of neem leaves and garlic. The procedures are often lengthy, thus I cannot enter into a full-scale treatment of them. It must suffice to provide the names of the goddesses and discuss some of the mantras.

The goddess for the first day is Pāpanī (Transgressive Lady), for the second day Bhāṣiṇī (Lady of Speech), for the third Jhaṇṭālī (Misty Lady[?]), for the fourth Kākolī (Raven Lady), for the fifth Siṃhikā (Lion Lady),²⁰⁹ for the sixth Phaṭkārī (Lady Who Exclaims phaṭ!),²¹⁰ for the seventh Mukakeśī (Lady Whose Hair Smells Like Cowdung), for the eighth Daṇḍinī (Lady Who Holds the Staff), for the ninth Mahāmahiṣī (Great Buffalo Lady), and for the tenth Rodanī (Tearful Lady). The grāhī goddess for the first month of the first year is Putanā (Stinking Lady), for the second month Makuṭā (Crested Lady), for the third Gomukhī (Cow-Mouthed Lady), for the fourth Piṇgalā (Tawny Lady), for the fifth Haṃsikā (Goosey Lady), for the sixth Paṇkajā (Lotus Lady), for the seventh Śītalā (Cool Lady), for the eighth Yamunā (Goddess of the Yamunā River), for the ninth Kumbhakarṇī (Pot-Eared Lady), for the tenth Tāpasī (Lady of Austerities), for the eleventh Rākṣasī (Demoness), and for the twelfth Capalā (Trembling Lady). The second year the grāhī is Yātanā (Vengeful Lady), the third year Rodinī (Weeping Lady), the fourth Caṭakā (Sparrow Lady), the fifth Cañcalā (Fickle Lady), the sixth Dhāvanī (Loping Lady), the seventh Yamunā, the eighth Hāyanī (Relinquising Lady), the ninth Kālinī (Lady of Time), the tenth Kalahaṃsī (Supreme Goose Lady), the eleventh Devadūtī (Lady Emissary of the Gods), the twelfth Palitā (Crone), the thirteenth Vāyavī (Lady of the Wind), the fourteenth Yakṣiṇī, the fifteenth Muñjakamuñcī (Wicked Lady of the Bullrushes [?]), the sixteenth Vānarī (Monkey Lady), the seventeenth Bandhavatī (Lady Who Binds), and the eighteenth Kumārī (Princess).

For all of these, sesame is offered into a fire with the mantra o muñca paca daha āgaccha bālike ṭhaṭha (Om, Liberate! Cook! Burn! Come! Young lady, svāhā). After these bali offerings are made, japa (simple repetition of a mantra) is performed while the ritual officiant touches the child. The japa mantras are long, for example, namaḥ sarvamātṝṇāṃ hṛdayamoṭaka mañja kaḍhḍha sphoṭaya sphura gṛhṇa ākarṣaya troṭaya evaṃ jñāpayati hara nirdoṣaṃ kuru bālakaṃ ṭhaṭha (Salutations, O remedy in the form of a heart-pill for all the Little Mothers, cleanse! kaḍhḍha[?]! Flash! Spring forth! Possess! Attract! Break apart! Thus one should announce. O Hara, make this child faultless, svāha).²¹¹ Other, longer mantras follow, which are said to lead to the liberation of the spirit (grahamukti). A number of protective verse mantras should then be recited. These resemble versified nyasas, invoking an abundance of male divinities to protect the directions and the parts of the body. In these texts, male deities are protective and liberating, especially Gaṇeśa and Vīrabhadra, while goddesses either play a subservient role or are themselves the possessing grāhīs.

AṢṬĀṄGAHṚDAYA	GYU-ZHI

deva	lha (deva)
daitya	lha-min (jealous gods; Skt. asura)
gandharva	dri-za (scent eaters; Skt. gandharva)
uraga	kLu (serpent spirits; Skt. nāga)
yakṣa	gNod-sbyin (harm-givers; Skt. yakṣa)
brahmarākṣasa	tshangs-pa (pervasive spirits; Skt. brahma)
rākṣasa	srin-po (cannibal spirits; Skt. rākṣasa)
piśāca	sha-za (flesh-eater spirits; Skt. piśāca)
preta	yi-dag (hungry ghosts; Skt. preta)
kuṣmāṇḍa	grul-brum (vampire ghouls)
niṣāda	byad-stems (evil-curse ghosts)
aukiraṇa	yeng-ched (mental agitators)
vetāla	ro-langs (zombies; Skt. vetāla)
pitṛ	mTshun-lha (ancestor gods; Skt. pitṛ)
guru	bla-ma (lama, guru)
vṛddha (elder)	drang-srong (sage)
ṛṣi	rGan-po (respected elder)
siddha	Grub-pa (magical emanation)