Kubjika and the Paschimamnaya: four origins

A goddess asks her own name

Early in the Kubjikāmata, the scripture at the head of the Western Transmission, the goddess interrupts to ask a question about herself. Her name is Kubjikā, the bent one, the hunchbacked one. She wants to know why.

kathaṃ me kubjikā nāma kiṃ khañjī pūrva sūcitā

“How is my name Kubjikā? Am I the lame one, as was indicated earlier?” The question sits at 3.36, and the text does not answer it there. It makes her wait thirteen chapters, and she has to put it again at 16.15 before the answer comes.

The answer, when it comes at 16.23, is not the answer most readers expect:

bṛhatkāyo yadā kaścit svalpe vyācarate gṛhe kuñcitāṅgo viśed yasmāt tadvad eṣā maheśvarī

“When one of huge body moves about in a small house, he must enter with his limbs bent. So too this Great Goddess.”

The derivation closes four verses later, at 16.27: tena tvaṃ kubjikā proktā, therefore you are called Kubjikā.

She is not deformed. She is not ashamed. She is bent because she is too large for what she is entering. The verses on either side say what the small house is. Just before, at 16.22, all this is pervaded by the Goddess, gross and subtle alike. Two after, the vessel in question is the yogic body. A second derivation a chapter later puns on a different root, kūj-, to resonate or to coo, and makes her the one who sounds within the submarine fire.

This matters more than a point of etymology, because almost every popular account of Kubjikā you will find supplies a different story, usually an erotic one, and omits the explanation her own scripture gives. It is a small case of a larger problem. The Western Transmission is one of the worst-served subjects in English on the internet, and what little exists is mostly wrong in ways that are checkable.

The transmission that calls itself western

Kaula tantra divided into four streams. Alexis Sanderson sets the scheme out as the Four Transmissions, the āmnāyas, or the Transmissions of the Four Lodges, each with its own deities, mantras, maṇḍalas, mythical saints and myths of origin. The eastern is the Kaula Trika, Abhinavagupta’s tradition. The northern is the Kālī stream, including the Krama. The southern is the cult of Kāmeśvarī, which grew into Śrīvidyā. The western, the paścimāmnāya, is the cult of Kubjikā.

The text uses the term of itself. At 2.22 it calls its own path paścimāmnāyamārgo ‘yaṃ, this path of the Western Transmission.

It is worth being careful about how this scheme relates to the older division of the tantric canon into the two seats, the Mantrapīṭha and the Vidyāpīṭha. They are not rival maps of the same thing. The seats divide the Bhairava-tantra canon; the transmissions divide the Kaula systems that sit above that canon and draw on it. There is no third seat here.

One correction while we are on the subject, since it circulates widely. The English Wikipedia entry on the āmnāyas lists the Kubjikāmata among the texts that set out the four-directional scheme. It does not. The Kubjikāmata’s own fourfold classification, at 19.107, is not directional at all: Island, Goddess, Seat and Siddha. And the text plainly does know the directional vocabulary. It called its own path the western one at 2.22, as we have seen; it names the eastern transmission at 19.104; and at 12.22 it sets a southern transmission alongside a northern path inside a single half-verse. Three of the four quarters are named as transmissions, and the north appears as a path rather than a transmission. What it does not do is build its own four-square scheme out of the directions. The directional map is a systematisation applied across the Kaula cults from outside, and a cult’s self-label could long precede the full four-square scheme.

Four origins, and they do not agree

Here is where the subject becomes genuinely interesting, and where nearly everything written about it in English goes wrong.

Ask where this tradition came from and you get four answers. Two of them are made by the root text itself, in its own colophons, and they are not the same answer.

The first chapter closes candradvīpāvatāro nāma prathamaḥ paṭalaḥ: the first chapter, called the descent in Candradvīpa, Moon Island.

That is the colophon’s claim, and it has to be stated as the colophon’s rather than the chapter’s. Chapter one is not set on the island. Its frame is Himalayan: 1.2 puts the scene on the back of Himavat, at Santānapura, below the Trikūṭa peak, and the mountain himself is the interlocutor from 1.9 to 1.27. The word candradvīpa does not occur in the chapter’s body at all. It first appears at 2.21, and the island is actually described in chapter 20, which carries the identical chapter-title. So the title is a claim the text makes about where the teaching came down, attached to a chapter set somewhere else.

The whole work closes somewhere else entirely:

iti caturviṃśatisāhasre sārāt sārataraṃ śrīkubjikāmnāyaṃ śrīoḍiyānapīṭhavinirgataṃ pañcaviṃśatipaṭalaṃ sārdhatrīṇi sahasrāṇi śrīmataṃ parisamāptam

“The Śrīmata, essence of the essence within the Twenty-four Thousand, the Kubjikā teaching issued from the seat of Oḍiyāna, of twenty-five chapters and three and a half thousand verses, is complete.”

So the text titles its opening a descent on an island and ends by claiming one of the four great tantric seats. Neither is the answer usually given.

The answer usually given is the Konkan, the western coast of India, and it does not come from this text. It comes from the Ṣaṭsāhasrasaṃhitā, the expanded recension, which at 43.27 to 28 places a city called Candrapura on the shore of the western sea, in the great land called Koṅkaṇa. The verses themselves name no teacher. Dyczkowski, who quotes them, reads the passage as the advent there of Siddhanātha, a figure the texts also call Oḍḍīsanātha, Tuṣṇīnātha and Kūrmanātha. A few verses further on the same text names the city’s king Candraprabha as born of the Kadamba line, which is what ties the account to the Goan Candrapura rather than to any other.

I checked the root text for the word. Koṅkaṇa occurs exactly twice in the whole Kubjikāmata: at 5.66, in a list of trades and regions whose disparagement requires expiation, which 5.68 makes explicit, and at 7.18, in the epithet Koṅkaṇeśānī, Mistress of Koṅkaṇa. Neither is an origin story.

It would be easy to conclude from that the Konkan claim is a late accretion, and that conclusion would be wrong. Sanderson arrives at the same part of India by a different route, and a partly lexical one: he reads the goddess-names of the tradition’s own texts and finds Kannaḍa feminine forms among them. His judgement is that the tradition is probably a product of the Deccan, and that the Deccan here should be taken to include the Konkan. He goes further than this article needs, extending the inference from the text to the whole tradition. He also hedges it, calling it probable rather than demonstrated. We should hedge it too.

Four claims, then, and a fifth thing that is not a claim at all.

What the reference books say

Open a handbook and Kubjikā will usually be introduced as a Nepalese goddess.

That is right about one thing and wrong about another, and the distinction is the whole point. Nepal is where this cult survived. It is not where it began, and nothing in its earliest scripture is Nepalese.

The measure of the survival is startling, and Sanderson gives it exactly. Of the sixty-six manuscripts the editors of the critical edition registered, sixty-five are Nepalese, and those that carry dates run from the eleventh century to the eighteenth. One manuscript out of sixty-six comes from anywhere else.

That is a fact about custody, not about composition. Dyczkowski states the methodological point as plainly as it can be put: these dates give us a measure of the time of entry of the Kubjikā cult into the Kathmandu Valley, but not, of course, when these texts were redacted.

And the cult was never only Nepalese. Sanderson puts the sophisticated Kashmiri exegesis of the Śaiva tantras between the ninth century and the thirteenth, and records that it became the standard frame for the cult of Kubjikā in the Tamil south. The tradition’s masculine parallel, the Śāmbhava system, travelled further still, into a body of South Indian post-scriptural literature that includes Umākānta’s Ṣaḍanvayaśāmbhavakrama. The Valley is where the manuscripts are because the Valley is where the tradition was still being copied when it had stopped being copied elsewhere.

The oldest manuscript is not what you would guess

There is a trap here that is easy to fall into and worth naming, because it changes what the dates mean.

The oldest dated witness of this tradition is not a manuscript of the Kubjikāmata. It is a manuscript of the Laghvikāmnāya, a shorter redaction, dated 1037 or 1038 of the Common Era. The earliest witness of the middle-length redaction dates itself by the reign of the Pāla king Rāmapāla, who ruled from about 1072 to 1126. A manuscript that ends up counted among the Nepalese sixty-five, reckoning its own date by a Bengali king, is the kind of detail that should make anyone slow to read custody as origin.

The relationship between the short and the long versions runs the opposite way to the obvious guess. The Laghvikāmnāya lacks six of the Kubjikāmata’s chapters, and on a comparison Schoterman published in 1990, which Dyczkowski reports and endorses, four of those six were drawn from the Tantrasadbhāva. So the longer text is an expansion of the shorter one, not an abridgement of it, despite the short version’s colophons calling it the Kubjikāmatottara, a title that reads as a sequel to the Kubjikāmata rather than as its source.

On when any of it was composed, two serious scholars diverge and the honest thing is to say so rather than pick. Dyczkowski runs a chain of citations backwards, from the Jayadrathayāmala through the Kālī cult of Jñānanetra, and lands the Kubjikāmata, or some early version of it, at the beginning of the ninth century. Sanderson, in The Śaiva Age, gives a relative sequence only, placing this corpus well downstream of the Niśvāsa and the early Saiddhāntika layer, placing the Tantrasadbhāva before the Kubjikāmata and the Jayadrathayāmala after it, and declines an absolute date in that passage. We have not been able to obtain the later article where he may give one, and so we do not attribute a century to him. The confident ninth-or-tenth century figure that circulates online traces to an encyclopedia entry that cites nothing.

A text assembled from another tradition’s chapters

The Kubjikāmata did not begin from nothing, and Sanderson’s statement of its dependence is blunt enough to be worth quoting rather than paraphrasing. The Kubjikā cult is intimately connected with the Trika, he writes, and its dependence is revealed by the fact that much of its principal and earliest scripture consists of chapters and other passages taken, with minor overwriting, from the scriptural corpus of that tradition.

Taken with minor overwriting. That is a strong claim about a foundational text, from the scholar best placed to make it.

It also inherited the machinery. The Kubjikāmata is one of the principal witnesses for the second Sanskrit alphabet the tantras use, the mālinī, and for the coded extraction of mantras that the tradition calls mantroddhāra: it announces the extraction of the mālinī sequence at 4.81 and carries it out in the verses that follow, and it announces one of its own mantras as written in reverse at 10.33.

It does not make the tradition derivative in the sense that matters. What the Western Transmission did with the inherited material is its own: the Trika’s own triad of goddesses is demoted out of the devotional foreground into the ritual and yogic substructure, and something else is installed above it. The cult shares with Śrīvidyā, and with no earlier Kaula tradition, a system of ascending centres in the body and a particular line of Siddhas. Where the Trika and the Kālī cults trace themselves to Macchanda and his circle, Kubjikā’s tradition traces itself to Mitranātha and his.

A caution on the body-centres, because two numbers are in play and both of them are scholarly. The standing monograph on the subject, by Dory Heilijgers-Seelen, is titled The System of Five Cakras in Kubjikāmatatantra 14–16, and five is what chapters 14 to 16 give her. Sanderson, a page on from his statement of the Trika dependence, describes a system of six power-centres as characteristic of the Kubjikāmata’s yogic ritual, and enumerates them: ādhāra or mūlādhāra, svādhiṣṭhāna, maṇipūra, anāhata, viśuddhi, ājñā. It is of that six-fold scheme that he makes the structural claim worth keeping. The six centres later became so universal in Indian yoga that their absence is the surprise, and they are absent from every tantric tradition except this one and the cult of Tripurasundarī. They are also where the Śāmbhava system mentioned above locates its Six Orders, its ṣaḍanvaya.

So the popular figure of six is not an error. It is the right number for a different question. Five is the count for the three chapters Heilijgers-Seelen edited; six is the count for the yogic ritual Sanderson describes. Anyone quoting a single number for the tradition should say which of the two they mean.

The number that is not a number

The gap in English coverage has produced one figure that circulates more than any other: that this is a tradition with a corpus of twenty-four thousand verses. It is a good example of how a number can be true and still be wrong.

Twenty-four thousand is in the colophon quoted above. But read what it is doing there. The Kubjikāmata describes itself as the essence of the essence within the Twenty-four Thousand, and then gives its own measure: twenty-five chapters, three and a half thousand verses. The Sanskrit elides the unit, as colophons do, and simply says three and a half thousand; verses is the understood noun.

The chapter count is exactly right, which is checkable in a minute against the electronic text. The verse figure is not. Dyczkowski takes three and a half thousand to be the text’s actual size. Count the numbered verses in the electronic text of the edition and you get fewer than three thousand. So the scripture’s own smaller number looks like an ideal round as well, just a more modest one than the figure it is nested inside.

The Twenty-four Thousand is the Manthānabhairavatantra, which calls itself the Caturviṃśatisāhasrasaṃhitā, the Scripture of Twenty-Four Thousand Verses, and which sits in a graded series of ideal magnitudes descending from a scripture of a billion verses. These are the tradition’s statements about the scale of its own revelation, not an inventory of surviving manuscripts. The sixty-four Śākta tantras are famous in the same way and for the same reason.

A tradition edited in fragments

Which brings us to the thing that actually limits what anyone can say about the Western Transmission, and it is not a shortage of interest.

The Kubjikāmata has a critical edition, published by Brill in 1988 as the thirtieth volume of the Orientalia Rheno-Traiectina, edited by Teun Goudriaan and Jan Schoterman. It was announced as a project at Utrecht in 1973. It took fifteen years.

The Ṣaṭsāhasrasaṃhitā has fifty chapters. Five have ever been critically edited, by Schoterman in 1982, as his doctoral thesis. The remaining forty-five had no critical edition when Dyczkowski wrote in 2009, which he gave as the reason a detailed study of the recensions would have to wait, and we have found none published since.

The Manthānabhairava is in three sections. Dyczkowski edited one of them, the Kumārikākhaṇḍa, in a set published in 2009 in fourteen bindings and running past five thousand pages. The other two sections are summarised, not edited.

So a reader who wants to know this tradition can consult one complete scripture, one tenth of a second, one third of a third, and a great many manuscripts in Kathmandu that nobody has published.

What the edition cost, and what it is worth

The 1988 edition is the standard one. It is also the subject of twenty-four pages of corrections by Sanderson, published in the Indo-Iranian Journal in 2002, and anyone citing it should know that.

His assessment opens generously, and the generosity is not perfunctory:

This, then, is a text of considerable interest to the student of Śaivism and we owe a great debt of gratitude to the editors for providing us with the editio princeps, especially since it is one of the first editions of a Śaiva Tantric text that has been prepared on critical principles.

Then the criticism, and it is structural rather than a list of slips. He holds that the editors’ preference for two particular manuscripts is unjustified, that they failed to correct obvious errors against the Tantrasadbhāva, a source he had himself communicated to them in 1984, and that they did not obelize, so a reader cannot tell which readings the editors meant and which are corruption they left standing.

Both things are true at once. It is the edition that made the field possible, and it is an edition that needs using with care.

The man who edited five chapters out of fifty

Jan Anthony Schoterman was born in Amersfoort on 27 March 1948 and died in Amersfoort on 29 June 1989. He was forty-one.

He read Indo-Iranian and Dravidian at Utrecht from 1968, where Jan Gonda was among his teachers, and took his doctorate there in 1982 under H. W. Bodewitz. He taught at Utrecht and at Tilburg, held a lectureship at the Kern Institute in Leiden, and held a research post at the Royal Netherlands Institute there. Klaus Karttunen’s Who Was Who in Indology records, in six words, that he died before finding a permanent position.

It would be easy to make too much of this and the record does not support the full romantic version. He published three Sanskrit editions in eight years: The Yoni Tantra in 1980, the five chapters of the Ṣaṭsāhasrasaṃhitā in 1982, and the Kubjikāmata in 1988. He was not shut out of his field. Whether a man of his generation could normally expect a chair by forty-one is not something we have the evidence to say, and we will not say it. How he died is not recorded anywhere we can find, and we will not speculate.

What can be said is narrower and sharper. The Ṣaṭsāhasrasaṃhitā has fifty chapters, and the only five that have ever been critically edited are the five he edited, and they are still the only five. The Kubjikāmata appeared in 1988, the year before he died, so that edition was finished. One book was not: The Kubjikā Upaniṣad, which he and Goudriaan had been working on, appeared in 1994, five years after his death, completed by Goudriaan alone.

The field’s most exacting critic of their joint work added, at the close of his opening tribute and before a word of the criticism, that in the light of the value of this work Schoterman’s early death seems all the more tragic. That is a better epitaph than anything written here, and it was written by the man who would spend the next twenty pages finding fault with his commas.

The goddess who is still worshipped, and still secret

Most of this would be an exercise in philology if the tradition were extinct. It is not.

Kubjikā is the lineage deity of Newar families in the Kathmandu Valley entitled to the sacred thread: the Rājopādhyāya Brahmins who serve as priests at the Taleju temple, and the Karmācārya and Josī castes. Dyczkowski, whose fieldwork is the documentation of record and who calls his own book on it a preliminary survey, describes her as the secret identity of the public deities, known only to initiates. The chain he traces runs through Taleju, the royal goddess of the Malla kings: Taleju is Siddhalakṣmī, and Siddhalakṣmī is identified with Kubjikā in her form as Siddhikubjī. We have not been able to obtain that volume, so everything in this section is reported as his rather than asserted on our own authority.

She is transmitted by initiation, worshipped in household shrines, and her thirty-two-syllable vidyā functions as a standard element in Newar Śākta initiation generally. The length is corroborated inside the scripture, at 17.28.

This is a continuously initiated Kaula lineage rather than a revival, which makes it genuinely unusual. The Kashmiri traditions that produced Abhinavagupta are reconstructed from manuscripts. This one has a priesthood.

The Kubjikāmata is a Kaula scripture, and the ritual system it describes includes consort practice; one chapter is devoted to the ritual partner, and a chapter colophon names sexual union among its contents. This is part of the record and there is no honest way to describe the tradition without saying so. It is also, in large part, why the system was transmissible only through initiation and why its custodians still treat it as secret. What popular accounts do with that material is the failure worth naming: Western ones tend to make it the headline and even to manufacture it into the etymology of the goddess’s name, and Indian devotional ones tend to omit it entirely and leave only the yoga. The philologists do neither, and neither will we. The spine of this tradition’s history is textual and institutional, and it needs none of it.

What the Western Transmission is not

Four errors are worth correcting directly, because they are what a reader will meet first.

It is not a Nepalese cult by origin. Nepal is where it survived, and the sixty-five manuscripts out of sixty-six measure preservation, not birth.

It is not marginal. It is documented in Kashmir, the Deccan, Nepal and South India, it stands behind the royal goddess of the Malla kings, and its scripture is one of the most substantial Śākta tantras to survive.

It is not reducible to Kuṇḍalinī yoga. The identification of Kubjikā with the coiled power is the tradition’s own, and stated in its own scripture, so correcting it would be the opposite error. The flattening is what to avoid: a scripture of three and a half thousand verses, a pantheon, an initiatory lineage and a living priesthood are not a metaphor for a yogic process.

And its goddess is not bent out of shame. The text asked that question of itself and gave its own answer, thirteen chapters later, and the answer is that something very large is trying to get into something very small.

The shape of what we know

A tradition that names itself after a direction, claims descent from a seat in the far northwest, titles its opening chapter a descent on an island, has its later recension place its first teaching on the western coast, is thought by the best modern reading to have come out of the Deccan, and survives almost entirely in the manuscripts of a Himalayan valley.

The honest summary is that we do not know where it began. We know where it ended up, because we can count the manuscripts.

Five chapters out of fifty are edited. The man who edited them did not reach forty-two. The rest is in Kathmandu, filmed, catalogued, and waiting.