Avesha: what the tantras did when nothing happened

A grown man is expected to fall over at his own initiation, and the scripture grades him on the direction he falls in.

The text that says so is the Tantrasadbhava, a Shaiva scripture of the Shakta or goddess-centred currents, and in its ninth chapter it describes a consecration. The guru holds a pitcher. The stream is poured. And then, in the verses Judit Törzsök translated and Vikas Malhotra prints, it tells you what you are looking for.

This is the sign of those who have been transformed [by initiation]: they move about, tremble or shake. And when the bonds of the soul are cut, the initiate falls on the ground. He who falls on his face is the best, he who falls on his back is average, and he who falls on his side is taught to be the least good by Shambhu.

Christopher Wallis notes something unusual two verses earlier. The disciple is not the only one shaking. The guru’s hand trembles, and so does the pitcher.

The word for what is happening to them is avesha: entering, being entered, taking possession of. It is the word this article is about, and the first thing to see is that this text made it the gate. Wallis’s formulation for the passage is exact. It takes pata, falling, as the key piece of evidence for the descent of grace, and requires it in order to proceed further.

So the rite turns on an involuntary event. A man may want initiation for twenty years, present himself, pay for it, fast for it, and stand in the hall at the appointed hour. None of that makes his body do anything.

Which raises the question this article follows:

What do you do with the one who does nothing?

What the texts tell the guru to do

The Tantrasadbhava answers first, and shortly. If the disciple shows no sign he should be rejected and not initiated.

The Malinivijayottara, root scripture of the Trika, the Kashmiri lineage that produced the tradition’s greatest exegetes, is harder, and it is worth having in full because the usual summary drops the middle of it. At 11.37 to 11.39:

athava kasya cin nayam aveshah samprajayate tad enam yugapac chaktya sabahyabhyantare dahet taya samdahyamano ‘sau cchinnamula iva drumah patate kashyapiprshthe akshepam va karoty asau yasya tv evam api syan na tam caivopalavat tyajet

Or if this avesha does not come about in someone, then let him burn him with the power, outside and within at once. Burnt by it, he falls on the back of the earth like a tree whose root is cut, or else he convulses. But he for whom even thus it does not come about, him let one abandon like a stone.

The guru does not simply write the man off. He applies more power, with a result specified in advance, and the stone clause is the second failure and not the first. Abhinavagupta thought the rule worth keeping, and versifies it at Tantraloka 29.210, so the instruction travels from scripture into the tradition’s largest work of exegesis.

The Brahmayamala is the most generous. At 2.46cd to 102ab, on Wallis’s provisional numbering, it supplies five alternative observances for when the vows do not produce their aim, across some fifty verses.

Notice what all of these have in common. Reject him, burn him, set him five more austerities: every one is a further thing to do. The tradition’s first instinct, when the body gives nothing, is to apply more procedure to it.

Hold that, because the tradition’s second instinct is to ask whether procedure was ever the point.

The fourth answer: borrow a body

There is a fourth thing you can do, and it answers a different failure by the same reflex.

The Nishvasaguhya sets out a rite at 3.24 to 3.27. Its trigger is not a silent initiation hall: it is the sadhaka, the practitioner who has already been initiated and is working toward power, who sees no auspicious or inauspicious sign in his dream. What he does then is recite om candike krama krama svaha ten thousand times, mix oil with lac, smear his left thumb and empower the oil with mantra, and then wash the faces of a girl and a boy and have the two of them look at the thumb.

The thumb is the practitioner’s own, which is the detail that matters and which the shorter retellings of this rite tend to lose. Somadeva Vasudeva, whose constituted text this is, prints a close parallel from the Brihatkalottara that spells out what the Nishvasaguhya leaves to grammar: there the yogin rubs his own left thumb and then shows it to the children.

His body produced nothing. So a child’s eyes are borrowed, and what they are pointed at is his own hand.

And now the detail that unsettles the usual account of this rite, which Malhotra supplies himself although it cuts against the genealogy he is building. In this earliest version there is no avesha. The word is simply not there. The child is employed to read the oil on the thumb, and not as a medium.

She is not yet a vessel. She is a reader.

The rite had a name, and the name is the plainest thing about it. Prasena, which goes back through the Middle Indo-Aryan pasina to prashna. The question.

It is older than the Shaiva scriptures that prescribe it, and it was disreputable before they took it up. At Dighanikaya I.11 the Buddha’s list of wrong livelihoods already includes the mirror-question, the girl-question and the god-question: the three things the rite does, proscribed by name, centuries earlier. That locus reaches this article at second hand through Malhotra and has not been checked against a Pali edition, which is worth saying, because it is doing real work here.

The instruments had other uses. In Bana’s Harshacarita, in the seventh century, a goddess emerges from the sword Attahasa, legitimising a succession. No question is put and nothing is reported back, so that is not the rite; it is one of the rite’s instruments, at court, doing political work. And in eleventh-century Kashmir, in the same valley and roughly the same generation as Abhinavagupta, the satirist Kshemendra listed the same instruments as a confidence man’s kit: khadge ‘ngushthe salile, at the sword, at the thumb, in the water.

One text often said to carry this material does not. The Svacchandatantra was searched here in two independent witnesses, and prasena occurs in neither; the thirteen near-matches in the first are the ordinary adjective prasanna, clear or gracious.

What the child became

She does not stay a reader.

The Tantrasadbhava carries a fuller version at 9.375 to 9.384, and there the rite is appended to the samaya initiation, the preliminary one that binds the candidate to the rules. The surfaces are listed: sword, mirror, thumb, water, lamp, pot, moon, sun. The child is now one and either sex, darikam vatha darakam, a girl or else a boy, where the Nishvasaguhya wanted two. The oracle’s content is past, present and future. And the invocation has changed person:

svakayam pashyami badham shrinomi svayam jighrami

I see my own body. I hear clearly. I myself smell.

A first-person verb is not by itself proof of whose mouth it is in, and a formula spoken by the officiant would use the same words. But the rite’s own trajectory points one way: by the time of the Jayadrathayamala it has become a full svasthavesha, oracular possession in a waking medium, which Malhotra translates at 3.14.70 to 3.14.76. Those numbers reach him through three intermediaries, and the content of that rite is not reproduced here, because this article has not seen it.

The numbering of the Tantrasadbhava carries its own caution. Wallis records that Bang, Törzsök and Dyczkowski each number the text differently, says his own edition is not to be taken as finished or final, and marks the divergent numbering with sigla. No argument in this article rests on the exact figures.

What happened between the earliest instruction and this one is that the child acquired the capacity the candidate in the hall was being tested for. A body that would produce the sign. The tradition solved the problem of the unresponsive adult by finding a more responsive body and pointing it at his thumb.

That is a procedural solution to a doctrinal problem, and the doctrine was waiting.

A note on the ground here, for anyone who does not have the map. These texts are Shaiva scriptures composed roughly between the sixth and the tenth centuries, and they fall into two broad streams: a dualist one, the Siddhanta, and the goddess-centred Shakta currents, of which the most esoteric is the Kaula and the one the Kashmiri authors belonged to is the Trika. Abhinavagupta and Kshemaraja come later than any of these scriptures and are exegetes rather than scripture. Scripture and commentary are different layers, and keeping them apart is most of the work.

The question a root tantra asks out loud

Everything so far has been a manual telling someone what to do. The Kiranatantra, a scripture of the Siddhanta, stops and asks why there is anything to do.

At 4.2 it puts the thing as a dilemma, and the dilemma is the one this article started with:

anyesham sannikrshtapi bodham sa kurute na kim yogyanam upakari ced ragavan syac chivas tada

Though she is equally near to others, why does she not produce awakening in them? And if she benefits only the fit, then Shiva would be partial.

If the power is everywhere, why is the hall full of men it does not touch? And if it touches only those it finds suitable, the god plays favourites.

The answer is an image, at 4.3:

yatharkarashmisamparkat padmabodhah samo na kim kanicit pratibudhyanti tathanyani na jatucit

Since contact with the sun’s rays is the same, why is the awakening of the lotuses not the same? Some awaken; others never do.

The sun, the verse goes on, has no attraction or aversion, and the Lord likewise has neither. The difference is in what the light falls on.

Bhatta Ramakantha, commenting, turns the image into a term. Where the scripture speaks of the fit, he writes: malasya pakvatvam atra yogyata. Fitness here is the ripeness of impurity. The power bestows awakening pakvamalanam, for those whose mala, whose impurity, is ripe, and no partiality can be inferred from her on that account.

He has a second image for it. A teacher wakes sleeping pupils with his staff; they are all equally in front of him, and he wakes only those fit to be taught. Delusion is the impurity, and its sleep is its incapacity to do its own work, which is what a degree of ripening means.

Then comes the sentence that reaches back to the pitcher. The Kirana says the power illumines the soul’s own nature bodhacihnabalena, by the force of the sign of awakening. Ramakantha glosses the force:

bodhacihnasya yad balam malaparipakah

The force of the sign of awakening is the ripening of impurity.

The sign has whatever force it has because the impurity is ripe. Which also settles what the sign measures, and the answer is not the man’s wanting. Nor, in any text read for this article, does the ripening get graded by the direction of a fall; that pairing would be this article’s invention, and it is not made here.

The absence of the sign is named too, in the same commentary, with its own consequence. Bodhacihnabalabhave, in the absence of the force of the sign of awakening, another mode has been stated, and the candidate is routed to it: to karmasamya, a deadlock in which two karmas come into equality, where the operative fact is not an awakening-sign at all. The man who does nothing is not thereby finished. He is reassigned.

There is a floor, and Ramakantha states it without softening. Where a candidate is utterly dull, like the animals, there is no ascertaining of devotion and the rest, which are themselves the effect of the descent of power, and so there is simply no entitlement to initiation. The descent is the cause; devotion is its symptom; where you cannot read the symptom you cannot infer the cause.

A reader will want to make all this a doctrine of merit, and the Mrigendragama closes that door at Vidyapada 7.17. The ceasing of impurity’s hold, it says, so ‘pi na svata eva syad api yogyasya vastunah, would not come about of itself even in a thing that is fit, since the insentient must always be impelled by the sentient. Impurity cannot ripen itself. The ripening is also the work of the power.

Which is a more interesting position than it first looks, and a weaker one than it sounds. The Siddhanta names a condition on the soul’s side and then concedes that the condition is supplied from the other side. It has not explained why this man and not his neighbour. It has given the unexplained thing a location and a name.

Abhinavagupta refuses the question

The non-dualists of Kashmir noticed that, and said so.

Abhinavagupta’s objection is a regress, and he states it in one breath in the Tantrasara at 11.4. Suppose, with the doctrine Vasudeva traces back to Sadyojyotis, that the descent of power happens on the ripening of impurity. Then what is that ripening, and what is its cause? And by the same objection, he continues, the following are refuted:

vairagyam dharmavisheso vivekah satseva satpraptih devapuja

dispassion, a particular merit, discernment, service of the good, attainment of the good, worship of the deity.

Read that list and notice what it is. Every item is something a candidate could set out to do. Dispassion is cultivated, merit accumulated, service and worship performed. Abhinavagupta is not calling them insufficient. He is calling them the wrong kind of thing to cite, in a chapter about why the power descends on one man and not his neighbour.

Kshemaraja presses the same regress afterwards, asking what causes the ripening, and adding that if the answer is nothing, it should happen to everyone at once. He is doing it, as Somadeva Vasudeva has shown, with an eye on the Buddhist philosopher Dharmakirti, whose critique of initiation the Shaiva authors are answering as much as they are answering each other. The quarrel is not purely domestic.

And then, at Tantrasara 11.6, Abhinavagupta does something more drastic than refuting the answer. He rules the question out.

na ca vacyam kasmat kasmimshcid eva pumsi shaktipata iti

And one must not say: why is there a descent of power in just this one person?

The reason is that the question has no subject. It is the Supreme Lord himself who appears in that way, and this being so, who indeed is this person with reference to whom the objection is raised? The Tantraloka puts the same refusal in verse at 13.205 to 13.206: why not earlier, why precisely then, is not appropriate, and the matter is not open to cross-questioning.

That is a wall, put up by a man who wrote thirty-seven chapters of explanation about everything else.

He does leave something standing. The power presides everywhere, he says at 11.18; only, in these cases, the face she turns is Ghora or Ghoratara, the terrible and the more terrible, where elsewhere it is Jyeshtha. The body that gives no sign is not outside the power’s reach. The power is there and it is working, in its down-dragging aspect. Ramakantha, from the other school, calls the grace-bestowing form by that same name, Jyeshtha.

And Abhinavagupta draws one asymmetry that does real work. Grace, he holds, does not depend on karma. Concealment does. So the non-dualist can say to the man in the hall: that you were graced would not have been explained by anything you did; that you are held where you are, is.

Utpaladeva put it as a complaint. At Shivastotravali 13.11 he tells the god:

shaktipatasamaye vicaranam praptam isha na karoshi karhicit adya mam prati kim agatam yatah svaprakashanavidhau vilambase

At the moment of the descent of power you never make the deliberation that is due, O Lord. What has come over you today, that you tarry in the business of revealing yourself?

Abhinavagupta quotes it and reads it as doctrine. By the words never and due, he says, Utpaladeva stated the Lord’s non-dependence and his freedom from partiality in the matter of the descent of power. The devotee’s grievance that the god does not even stop to consider becomes the proof that grace is undeliberated.

One word, two glosses

The two schools are not arguing past each other. They are glossing the same scriptural word.

The word is in the first chapter of the Malinivijayottara, at 1.42: the tranquil Shaiva power, which bestows the fruit of liberation, becomes connected with the soul at some time yogyatavashat, by force of fitness. Abhinavagupta names that verse as his source himself, at Tantraloka 13.199, having taken up this same meaning as it stands in the Malinimata, which is his name for the Malinivijayottara, and he then reproduces the passage across the verses that follow, interleaved with his own comment. Because he names it, the identification does not depend on Jayaratha.

Fitness, then. One word, in one scripture, which both traditions have to account for.

Ramakantha fills it: malasya pakvatvam, the ripeness of impurity, a condition located on the soul’s side. The school then concedes, through the Mrigendra, that the condition is itself supplied from the other side.

Abhinavagupta declines to fill it, glossing it with something that cannot be inspected, fitness for identity with Shiva, and forbidding the next question.

It is tempting to make that a disagreement about the cause of grace, and it is not quite one. Both schools end in the same place: a descent that nothing on the recipient’s side accounts for. The Siddhanta walks one step further in before stopping and gives the stopping-place a name. Abhinavagupta stops at the door and says the room does not exist. They differ in how much they are willing to say before they stop, which is a difference in temperament as much as in doctrine, and it is the difference that produced three centuries of argument.

What the scripture promises the body that does cooperate

Against all of that, it is worth seeing what a cooperating body actually does, because the answer is not what the word absorption suggests.

The Malinivijayottara has the guru read the signs. At 11.34 to 11.36 it tells him to mark carefully the cihnasanghata, the cluster of signs, beginning with joy: joy, rising up, trembling, sleep, and whirling as the fifth, by the power that has entered, mandativradibhedatah, according to the distinctions of weak and intense and so on. So the inference from a sign to a graded descent of power is scriptural, and does not begin with the commentators. Abhinavagupta turns it into a reading rule at Tantraloka 29.196, where the guru is to discern a descent of power of such and such a kind by the distinctions intense, weak and the rest.

Then, in the twentieth chapter, the same scripture says what the thing being graded does to a man. He has been contemplating the supreme phoneme for seven days, in one of five places in the body beginning with the navel.

evam samsmaratas tasya divasaih saptabhih priye rudrashakti-samaveshah sumahan samprajayate avishto bahuvakyani samskritadini jalpati mahahasyam tatha geyam shivaruditam eva cha karoty avishta-chittas tu na tu janati kinchana

For him who contemplates thus, after seven days, O beloved, a very great absorption into Rudra’s power comes about. Possessed, he babbles many utterances, in Sanskrit and in other tongues; great laughter, and singing, and the cry of a jackal. He does these things with a possessed mind, and he knows nothing at all.

Avishto. The participle is the same root the oracular rite uses. The reward promised to a sanctioned seven-day practice, in the root scripture of the Trika, is behaviour for which the physicians had a diagnostic category.

Two cautions, both load-bearing. The last clause, na tu janati kinchana, he knows nothing at all, reads naturally as an awareness gone out; it can also be read as an absence of discursive thought, which in the non-dualist account is the mark of the highest state and not its loss, and nothing above depends on which is right. And shivaruditam is a compound with a long vowel in the printed Kashmir Series text, which rules out any reading as Shiva’s weeping and leaves the feminine shiva, the jackal, whose howl is a lexicalised ill omen in Sanskrit and which the text itself uses twice more in lists of involuntary noises. The reading taken here is the jackal. These verses have no scholarly edition, translation or discussion; the only prior rendering found is an unpublished typescript of 1956, which reads the jackal too.

And here is the obvious objection to all of it. Seven days of deliberate contemplation, with a stated result: how is that not a man producing absorption by wanting it?

Abhinavagupta has an answer and it costs something to use. Ritual action, he says at Tantraloka 2.8, does not deserve to be called a means at all, because it arises from that and is rather its manifestation; at 2.14 those who would reach that nature by a direct means are dullards who want a firefly in order to see the sun. The practice that works is the practice the descent has already produced. At Tantrasara 2.2 he describes one dridhashaktipataviddhah, pierced by a firm descent of power, who attends to the guru’s word once and needs no mantra, worship, meditation or observance thereafter.

What it costs is this. Chapter 2 of the Tantraloka is the chapter on anupaya, the means-less way, the topmost rung of a scheme of four means whose lowest rung is exactly where a seven-day contemplation of a phoneme in a bodily place belongs. Reading 2.8 against the twentieth chapter of the Malinivijayottara is reading the top of a hierarchy against its bottom, and it is an exegete telling a scripture what it meant, in an article that has just said keeping those layers apart is most of the work. The Siddhanta would say the answer relabels the problem rather than solving it.

So the objection is not closed. What can be said is narrower and survives: in no text read for this article does anyone tell the man in the hall that he should have wanted it more.

The word never sorted by wanting either

If the tradition had wanted a vocabulary that marked the sought off from the suffered, it had one available and declined to build it. Geoffrey Samuel argues the trouble runs the other way, and that the English words are the problem: possession, trance, mediumship and shamanism come from different origins, and all of them presuppose an intrusive external spirit-entity. He qualifies himself at once, since Indic thought has no shortage of such entities, and proposes a patterning of the mind-body complex instead, acquired voluntarily and consciously or involuntarily and unintentionally. The Sanskrit evidence below is what any such proposal has to sit on.

Open Monier-Williams at the entry and the problem is there in a single headword. The 1899 page gives six sense-groups, unnumbered, and two sit side by side: absorption of the faculties in one wish or idea, intentness, devotedness to an object, and then demoniacal frenzy, possession, anger, wrath. A Victorian lexicographer working from the whole corpus put devotedness and demoniacal frenzy next to each other and felt no need to split the entry. The adjacent noun aveshana he glosses, among other things, as possession by devils, which is also, in the same entry, a workshop.

Wallis adds a syntactic observation at page 132. In the tantric sources he finds two usages, split over time as well as in sense. Where avesha means something like possession, in the earliest sources, it construes with an agent in the instrumental. Where samavesha means absorption or immersion, in the mature tradition, it construes with a locative or a genitive. One is entered by something. One is immersed in something. The underlying point, that the root itself carries an active and a passive sense, to enter and to be entered, is also his, and he credits Torella for having made it before him.

Against that stands Frederick Smith’s claim, the one most often repeated. Forms in a-vish are nearly always distinguished from pra-vish: the first indicating friendly, benign or self-motivated possession, the second more often possession generated from outside. None of that is first-hand here; Smith’s 2007 article on the relevant epic chapters exists in no open form, and the wording above is from the book, at page 14, as Kathleen Taylor gives it in her review.

The hedges are the claim. Nearly always and more often are Smith’s own, and the 2007 article, where Geoffrey Samuel quotes it, has a generally in the same place. A tendency in usage is what is asserted, and a tendency survives counter-examples. The scheme is also three-way and not two: Taylor reports that Smith assigns malign, disease-producing possession to a different family altogether, the derivatives of grih, to grasp or seize, and it is that last group, on her summary of his scheme, whose meaning corresponds more closely to the English word possessed. Wallis rejects the ranking at page 135, finding it surprising that Smith should regard avesha as having much less polysemy than pravesha, entering, or grahana, seizure.

The epic bears one look, in the chapters Smith’s abstract names, Mahabharata 14.17 to 14.18. At 14.18.7, the jiva, the living principle that transmigrates, enters the limbs of the embryo part by part and endows them with consciousness, and the verb is avishya, which is Smith’s own example and works exactly as his account predicts. The next verse glosses that same entry with a pravesha noun, under the simile of molten metal taking the shape of its mould. The verse after that reaches for pravishati, fire entering a lump of iron. Three consecutive verses, one transaction, and both roots used of it. That is a counter-example and not a refutation, which is all a hedged tendency invites.

The physicians asked a different question entirely

The division of labour everyone expects is that the tantras handle the possession that was sought and the medical literature handles the possession that afflicted, as illness. The first half is defensible. The second does not survive contact with the texts.

The Caraka Samhita does give the exogenous kind its own slot. The Unmada Nidana, the section on the diagnosis of madness, lists five types, three from the humours, one from their combination, and agantuja, the one that comes from outside, which is said to differ from the others in cause, premonitory signs, symptoms, pain and response to treatment.

The cause the text gives is a moral one. Prajnaparadha, an error of judgement. A person disregards gods, ascetics, ancestors, gandharvas and yakshas and rakshasas and pishachas, the classes of non-human being, along with teachers and elders, and undertakes inauspicious acts; the gods and the rest then cause madness in him because of his own inauspicious activities, and the text records that some authorities attribute it to deeds of a previous life.

The treatments match. Ghee potions, and of hundred-year-old ghee the text says that the sight or touch or smell of it cures all kinds of spirit-possession. Eye-salves, collyria, nasal medications. And in the same chapter, for the same condition: worship, sacrifices, libations, incantations, burnt offerings, prayers, pilgrimages, vows and acts of expiation. Where the possessing being is a god, a sage, an ancestor or a gandharva, the physician is told to avoid harsh collyria and to treat gently. Drugs and mantras, one chapter, one condition.

And the text’s own sub-distinction runs on an axis nobody expects. The beings’ purposes are sorted into three: himsa, harm; rati, affection or pleasure; abhyarcana, worship or obeisance. The prognosis follows the purpose and not the patient. The harm-intent form, in which the patient may walk into fire or water, fall into pits or injure himself, is judged incurable, and a spirit described as desirous of avenging itself marks a case to be left to his fate. The affection and worship forms are curable. Mixed forms are incurable.

The question the medical text asks is why the entity came. Not whether the person wanted it.

So a possession that arrives unbidden and unwelcome can be of the worship kind, and treated by propitiation. A possession procured in a consecrated setting could in principle be of the harm kind. The initiation hall could not make wanting produce the sign, and the physicians did not think wanting was the relevant fact about the patient either.

Liable for wickedness, and for hunger

A tantra reaches a similar place through a list.

A caution first, because it bears on everything in this section. The copy of the Netratantra held for this article is a single scanned text whose character recognition is too corrupt to witness a reading. Everything here comes from Malhotra, who follows David Gordon White, and none of it was checked against the Sanskrit.

The Netratantra, which Malhotra dates to the seventh or eighth century, devotes its nineteenth chapter to counter-rites against seizure. The mechanism is a flaw in the shadow. The goddess asks how the yoginis, the goddess’s female attendants, attack and oppress people through the chayacchidra, the shadow’s crack, and how to avert the drishti-pata, the casting of the gaze. The victims are all creatures, the text says, and children especially.

Then comes the list of who is vulnerable, and it begins where you expect. The wicked, the impure, the ill-conducted. Those who abandon vows, who dishonour their parents, who desire their teacher’s wife, who speak untruth, who are envious or proud. Those who have touched a corpse.

And then, in the same list, without a break: those who have not bathed, those who are deeply grieved, those who are hungry, women who are menstruating, those who are frightened.

Malhotra draws the conclusion the chapter forces. The text and Kshemaraja’s commentary also recognise that possession occurs involuntarily, through unwilled or ethically neutral acts.

You are open to it because you are wicked, or because you are hungry. One list, no partition, and the conduct of the victim turns out to be one entry among several.

The twentieth chapter then supplies a justification for the killing that is, on its own terms, complete. The yoginis do not kill from passion, hatred, confusion or delusion. They kill at the command of the lord, and to supply the sacrifice of the god of gods. Their victims were generated by Bhairava for the purpose, and so their death is his anugraha, his grace: the yoginis cut away mala and what follows is purification and reunion. The text and its commentator place the devoured high in the cosmos, in the realm of pure knowledge. One passage in the nineteenth chapter has three competing readings that Malhotra leaves standing, writing that he will proceed regardless of its precise meaning, and this article prints none of them.

As Malhotra reports it, the argument is airtight. It is also monstrous.

Where the body is not yours to volunteer

In a living tradition the question changes shape again, because there the body that will produce the sign is not chosen by its owner at all.

Rich Freeman spent the field season from August 1987 to August 1988 in northern Kerala, working on temple worship in the Tantrasamuccaya lineage and on teyyattam, the dancing of the gods, the possession worship of castes historically without caste status. The dancers are specialists of the avarnar, those without caste-status, and the role descends hereditarily, generally matrilineally. A dancer takes vows and a limited fast before performing, and for a major festival he lives in partial seclusion in a hut near the shrine for as many as forty-one days.

Nobody in that account volunteers. The body is assigned by birth and prepared by rule.

Freeman never uses the word avesha: not in the article, not in its fifteen endnotes, not in the editor’s appended list of technical terms. The thing transferred is caitanya, conscious power. The act of invocation is avahana. And the state itself has a name:

the Sanskritic word darsana (literally ‘seeing’), which refers in standard Indian religious usage to worshippers passively viewing the enshrined idol of a deity, has the common meaning in Kerala of becoming possessed by the god.

Darshana, everywhere else, is what a devotee goes to a temple to receive. In Kerala it is also the word for the god arriving in your body. The hedge here is firm and it is this article’s own: Freeman translates one modern Malayalam manual and cites the Sanskrit commentaries only by page range, so his silence is evidence about his presentation and not evidence that the Tantrasamuccaya lacks the word.

He does set out how the deity’s arrival is tracked, and the tracking is done in the grammar of the songs. They move through a sequence of deictic shifts: third-person praise and narrative, then second-person praise and the genre of invocation, which is called vara-vili, the call to come, then a song of congealing the god’s presence, and finally the first-person speech of the god, daiva-vakku.

The god is spoken about. Then spoken to. Then speaks.

And from a second living tradition comes the thing none of the textual accounts prepares you for. Arik Moran, reviewing William S. Sax’s fieldwork among the Bhairav oracles of Garhwal, reports that on Sax’s evidence the oracles sometimes try to avoid possession, and that Sax classes them as patients rather than agents. Sax’s book was not read here and this reaches the article through Moran alone.

The appointed medium, in a functioning oracular tradition, trying to get out of it. That is worse for the man in the hall than it first sounds. Wanting will not produce the event, and in Garhwal not wanting will not prevent it.

What the statute had to decide

Modern institutions have the same problem and less room to be philosophical, because they have to decide cases.

Narendra Dabholkar spent eighteen years trying to pass a law in Maharashtra against what his organisation called blind faith. A private member’s bill by A. P. G. Dasturkar in July 1995, carried on the vote and never acted on. Cabinet approval in 2003, at which point the title changed from a law against superstition to a law against black magic. A version passed and filibustered in 2005. A joint committee in 2007 that took public submissions and received fifty-five thousand against and eighty thousand in favour. The bill lapsing with the 2009 elections.

He was shot dead on 20 August 2013. The Governor signed an ordinance on 24 August, and the Act that followed was deemed in force from 26 August under its own section 1(3).

The device that finally got it through was a refusal to define anything. The Act does not say what faith is, or what blind faith is. It attaches a schedule of specific acts, which can be amended, and it says nothing about belief; every occurrence of the word religious in the text as it stands falls inside the exemptions.

Four of the twelve scheduled items concern possession, and they cut in both directions. Item 1 is the long one, criminalising assault under the pretext of expelling a ghost, and its specificity tells you what was actually happening: tying with rope or chain, beating with a stick or whip, making the person drink footwear-soaked water, chilli smoke, hanging from a roof, plucking hair, heated objects applied to the body. Item 5 criminalises claiming possession in order to frighten or defraud. Items 6 and 7 criminalise accusing another person of being possessed, and the assault and public stripping that follow such an accusation. The accusation is its own subject, and in the scholarship on healing, accusation and the village the accused and the afflicted are generally two different people.

So the statute does not outlaw possession. It outlaws particular uses of it, by particular people, on purpose, which is the correct shape for a criminal offence and worth saying plainly: a law that punished the event itself would punish the possessed.

The difficulty is a drafting one. Section 12 exempts a long list of practices, and section 12(1)(5) protects, with no condition whatsoever, the invoking of the spirit and all other religious rituals. Only two of the eight exemptions carry the proviso about causing physical injury or financial loss, and they are numbers 3 and 4: stating and circulating the miracles of deceased saints, and the performance of prayers, upasana, which is devotional service, and religious rites at a home, temple, dargah, gurudwara, pagoda or church. So the proviso is not confined to speech, and one of the two conditioned exemptions is ritual performance, while the exemption covering the invoking of the spirit carries none.

An evening can therefore satisfy Schedule item 5 and section 12(1)(5) at once. Item 5 has its own threshold, fear or fraud, and the Act is not short of a test for whether item 5 is made out. What it lacks is a rule for what happens when the offence and the unconditioned exemption both apply to the same conduct, and no judicial interpretation of these provisions was examined for this article.

The drafting history shows the clause that would have reached the event was there and did not survive. The earlier sixteen-item schedule, which Dabholkar’s organisation still publishes on its own site, was far more possession-centred than the enacted twelve. It criminalised claiming to be possessed by divine or evil power and performing miracles in its name. It criminalised beating mentally ill patients in the belief that they are possessed. And it contained this:

to become possessed by supernatural powers and then pretend to give answers to any questions in this mental state

That is the prasena rite, described in a twenty-first-century bill, and it is not in the Act.

The World Health Organization’s ICD-10, in its 2019 version, defines trance and possession disorders and then restricts the category: include here only trance states that are involuntary or unwanted, occurring outside religious or culturally accepted situations. ICD-11, in force since January 2022, splits the code in two and repeats the clause in both. An oracle possessed in an appointed role at a recognised shrine is unlikely to meet either code, which is less impressive a convergence than it looks, and the observation is this article’s own: the category was drawn so as to exclude sanctioned practice, and it cannot then be produced as independent evidence that the traditions drew the same line. The ICD is the WHO’s general classification of diseases rather than the psychiatric manual proper, both codes additionally require distress or impairment, and neither has any operative role in an Indian criminal court.

As for whether the law works, the answer appears to be that it largely does not. ThePrint, reporting in January 2024, put registrations at around two hundred and fifty cases through 2022 against more than a thousand complaints received, with the conviction rate described by activists as dismal and no figure given. Among the reasons offered: police unaware of the Act or unsure what counts, frequent changes of government, lost funding, and officers who share the beliefs in question.

The law asks what a person meant to do. It is a fair question and the right one for a court. It is also not a question about the event itself, and on that the tradition turns out to have gone further than anything so far in this article suggests.

The initiation that needs no body at all

Everything above has assumed the sign mattered. The texts that demand it say so loudly, and this article has taken them at their word for twelve sections. They are not the whole of the tradition. Wallis finds that the Siddhayogeshvarimata, an early Shakta scripture, does not posit avesha as a requirement in initiation at all. And the plainest evidence sits in the same book as the reading-rule.

Abhinavagupta devotes the twentieth chapter of the Tantraloka to an initiation that produces evidence, and opens it by calling that initiation mudhajanashvasapradayinim, the one that bestows reassurance on the simple-minded. The evidence there is a measurable loss of weight on a balance rather than a convulsion, so it is not the rite of the falling hall; what carries over is his estimate of what such evidence is for.

In the next chapter he initiates the dead.

The twenty-first chapter sets out the rite for those who cannot present themselves: an uninitiated person, a king and such, the indolent, the fallen, the dead, the child, the sick, the woman, the aged. What stands in the hall is an effigy of darbha grass and cow dung, installed as though it were the disciple, and the soul is drawn to it and held there as if asleep. The apparatus that produces signs is dropped as surplus, and said to be dropped: the preliminary consecration, the consecrated food, the site, the couch and the mandala, the enclosed diagram of the deity, are of no use here, Abhinavagupta writes, since they serve the disciple’s own ritual preparation and his dream-vision. Then, at 21.38:

na spandate na janati na vakti na kilecchati tadrishasyaiva samskaran sarvan pragvat prakalpayet

It does not quiver, it does not know, it does not speak, it does not indeed will. For precisely such a one he should perform all the consecrations exactly as before.

Na spandate. It does not quiver. The word is the one the sign-texts use for the trembling in the hall, and hearing the echo is this article’s own reading rather than a commentator’s; Jayaratha on this chapter was not available. What is not in doubt is the instruction. Perform all the consecrations exactly as before. Two verses later the result is stated without qualification. With his bonds released he goes to Shiva.

The Siddhanta arrives nearby by argument. At Kirana 6.22 the text’s own interlocutor raises this article’s question from the other end: why is not even the slightest sign of one released from his bonds to be seen, and why is there nowhere a manifest sign? The scripture replies at 6.23 that where a manifest sign is seen anywhere, there too there is inconstancy, vyabhicaritva. Ramakantha, glossing, adds that such a sign is nowhere stated to belong to one released from his bonds, and belongs instead to one on whom the power has descended. What the text will accept, at 6.25, is a disposition rather than an event: the ending of latent traces, and unhesitating observance of what has been given. These renderings are this article’s own, since Goodall’s published translation of the passage was not obtained.

So the Siddhanta does more than tolerate the candidate who gives nothing. It holds that a manifest sign, wherever one appears, is inconstant, and it will not let such a sign stand as evidence of release.

The ritual literature makes the same provision in the plainest way. The Somashambhupaddhati, a Saiddhantika manual of the later eleventh century, prescribes a nirbija, seedless, initiation for those who cannot undertake the observances, and allows initiation by mere mental operation. The Svacchandatantra, which belongs to another current again, gives the roster of who receives it: the child, the simple-minded, the aged, the woman, the pleasure-bound, the diseased. It is close to the same list the Tantraloka gives for initiating the absent and the dead.

What none of these rites requires is a responsive body. The initiation hall of the Tantrasadbhava is one practice inside a literature that elsewhere initiated the absent, the incapable and the dead, and did it with every consecration performed in full. In those books the gate was not the fall.

No modern study found for this article puts these two bodies of material together, and the search had real limits. Sanderson’s papers on ritual meaning and on the rescue of the dead, Brunner-Lachaux’s edition of the Somashambhupaddhati, Goodall’s English of the Kirana passage, and Ferrario’s dissertation on the grades of the descent were each identified and none obtained; any one of them may have framed this already. What can be said is that the scholarship on the signs and the scholarship on initiating the dead are separate literatures with separate specialists, that the contrast here is assembled from the primary texts, and that Abhinavagupta framed it himself by writing the two chapters back to back.

What the sign was never going to prove

Go back to the hall.

A man is standing where the Tantrasadbhava expects him to fall, and he does not fall. He asked for this. He was accepted for it. He has done what he was told to do for as long as he was told to do it.

The Tantrasadbhava says send him away. The Malinivijayottara says burn him with the power, and if even that yields nothing, abandon him like a stone. The Brahmayamala sets him five more observances. The Nishvasaguhya, for a different failure, oils a thumb and borrows two children’s eyes. When the texts stop giving instructions and start giving reasons, Ramakantha says his impurity is not ripe, and the Siddhanta concedes through the Mrigendra that the ripening was never the man’s to manage, and Abhinavagupta says the question may not be asked, refuting dispassion and merit and discernment and service and worship in one line as he goes.

Four procedures, two doctrines, and in no text read for this article does anyone say he should have wanted it more.

Then there is what the sign was worth to the people who demanded it. Abhinavagupta names the evidential initiation and sources it to the Sadbhava-shasana, the Tantrasadbhava, taught to him by Shambhunatha: the text this article opened with is the one he credits for the rite of reading signs. In the Tantraloka the same collapse can be manufactured: at 29.270 the yogin lashes the disciple’s channel wrapped up within his own and forcibly causes even an adept to fall to the ground, so the fall arrives through somebody else’s body. At 29.239 the same manual warns that where the upper cakra, the highest of the centres on the inner axis, is not reached, the piercing is nothing but possession by a pishacha, a ghoul. The Kirana says that where a manifest sign appears at all, it is inconstant, and will not let one stand as evidence of release. And in his twenty-first chapter Abhinavagupta performs every consecration over a soul that does not quiver, does not know, does not speak and does not will, and sends it to Shiva with its bonds released.

The man face-down on the floor of the initiation hall is not holding a result. He is holding something that still has to be interpreted, in a tradition that spent three centuries arguing about what could be inferred from it and then, in its most careful books, declined to infer much.

The sign was never going to settle what the man intended.

It was never going to settle what had arrived, either.