Śavaripa the hunter

Four places above the founder

The tradition keeps eighty-four perfected ones, and it keeps them in order. One to eighty-four, numbered, each man with a place on the roll, and the place is itself a claim. It says who stood where. It says who taught whom. It says, more plainly than anything else on the list says it, what kind of person the tradition believed was capable of this.

Read down the Tibetan roll and the first names land where a reader of Indian religious history would expect to find them. Number one is Lohi-pā, whom the tradition generally identifies with Matsyendranātha, the man the Nāths call their first human teacher. Number nine is Gorakṣa. Number sixteen is Nāgārjuna. Number seventeen is Kṛṣṇacārī. Number forty-six is Jālandhari, which is also a place on a map.

Number five is a hunter.

The name set against him is Śabara, given also as Śabarī. It is the word this treatise is named for, and away from these lists it is not a compliment. Chapter 2 gathered what the literate traditions did with the people who carry it. In the Mahābhārata the Śabaras come out of the dung and urine of Vasiṣṭha’s cow, and go on to live by murder and by looting. In Hemacandra’s Triṣaṣṭiśalākāpuruṣacarita Śabara is a subdivision of the mlecchas, and the mlecchas are people who do not know even the word dharma. A Buddhist treatise lists Śavara among the low places of birth.

And here he is at five. Four places above the founder of a lineage, on a list that both traditions read.

That is the fact this chapter has to hold still, and it is an easy fact to inflate. So take the inflation off the table first, and take it off using the same scholar the chapter is about to lean on.

Dasgupta himself supplies the control, and it belongs here rather than three sections down. He notes that eighty-four is a mystical number in Indian religious arithmetic, attached to the āsanas, to the lakhs of births, to the Śiva-liṅgas of the Skanda Purāṇa, and he says this justifies doubt about the historical nature of the whole tradition of the eighty-four. The recensions also differ. The Tibetan list, the Haṭhayogapradīpikā list and the Varṇaratnākara list are three different lists. What an ordinal establishes is a position inside a tradition’s memory of itself. It is not a date.

So the ordinal proves one thing only. A tradition that counted its own perfected ones carefully put a man named for a forest people at number five, wrote him down there, and left him there for as long as the counting went on. Whether such a man lived is a separate question and this chapter does not answer it. What this chapter is about is who noticed the position, and when, and what they found standing next to him.

Start with a number, and with the man who printed it.

Shashibhusan Dasgupta submitted Obscure Religious Cults as Background of Bengali Literature to the University of Calcutta as a doctoral thesis, and Calcutta published it in 1946. Part III of it is on the Nāth cult, and it is there that he sets out the traditions of the eighty-four Siddhas and prints the list as it comes through the Tibetan sources, in Grünwedel’s compilation, all eighty-four of them, numbered. Lohi-pā at number one he glosses on the same line as Matsyendra, and the same man appears elsewhere in his book as Luī-pā and in Dvivedī as Lūhipā. Kṛṣṇacārī at seventeen is the Kāṇha-pāda whose songs come back at the end of this chapter.

Two other lists sit beside that one in his pages, and they matter for a reason that is almost comic. Dasgupta gives the roll of Mahāsiddhas from the Haṭhayogapradīpikā I.5 to 9, in Iyangar’s edition, and it runs Ādinātha, Matsyendra, Śābara, Ānandabhairava, Cauraṅgī, Mīna, Gorakṣa, and onward. He gives the Varṇaratnākara’s list of Nātha siddhas as well, in which Savara stands forty-third.

The Haṭhayogapradīpikā entry is the one this treatise has met before, and met in mangled form. In the copy of Briggs’s 1938 study of the Gorakhnāthīs that this project was able to consult, the same name in the same roll reads Sahara.

No charge is laid against Briggs for that, and the copy is the reason. On the same line of the same scan, Nātha reads Nfitha and Cauraṅgī reads Gaurahgi: an optical reader turning ā into fi and ṅ into h. A reader that does that to two names on one line cannot be trusted to have kept the b in a third. Whether Briggs’s printed page reads Śābara or Sahara is not something this project has seen, and until a clean page is in front of it, it does not know.

What can be said is narrower, and it is all this chapter needs. Dasgupta gives the name correctly, from the Sanskrit, in 1946. Anybody who has met it only through a poor scan has met it wrong, and the corruption is in the scan until somebody proves it is in the book.

The teacher of Luī-pā

Now the sentence that changes the chapter.

Dasgupta, in a footnote to his discussion of the Tibetan synonyms for Matsyendranātha, or Luī-pā, with whom he says Matsyendra is generally identified, reports from the Tibetan chronicle Pag sam jon zang that Luī-pā is described as a Buddhist sage sprung from the fisherman caste who was a disciple of Śavari-pā; and that Machendra, in the same work, is the incarnation of Mahādeva as a fisherman, at Kāmarūpa. He cites P. C. Bagchi’s introduction to the Kaulajñānanirṇaya for both.

Dvivedī reaches the same conclusion from Hindi and Tibetan material four years later. In Nāthasampradāya, in his numbered list of siddhas, item nineteen is Sabara, and his entry says there have been two siddhas of that name: one the guru of Lūhipā, the kāyastha of King Dharmapāla, whose reign he gives as 769 to 809; and one a siddha of the tenth century. He adds that the two have been mixed up with each other, and that many works written by Sabara survive in Tibetan translation.

Two Indian scholars, working in two languages from two source traditions, four years apart, put the Śabara-named siddha in the same position: teacher of the man the Nāth lineage remembers as its founder.

That is the chronological argument this chapter makes, and it is made out of Dasgupta and Dvivedī. Anybody who wants to check it will find both men’s pages set down in the sources at the foot of this chapter.

State it at its exact size. If Luī-pā is Matsyendranātha, and if the Śabara siddha taught him, then a figure carrying the word śabara stands one generation above the beginning of the Nāth line, and the standard story in which Gorakhnāth invents śābara mantra arrives several centuries and several generations late. Chapter 6 shows how thin the Gorakhnāth attribution is on its own evidence. This chapter shows what was standing there before it.

Three conditions ride on that sentence and none of them is small. The Luī-pā and Matsyendra identification is what Dasgupta calls a general one, which is a report of consensus and not a demonstration. Dvivedī’s own entry says two Śabaras have been conflated, so which of them is Luī-pā’s teacher is a question inside the source itself. And the eighty-four are a tradition about lineage rather than a record of it. What would close all three is Abhayadatta’s Caturaśītisiddhapravṛtti at the Śavaripa life, and Bagchi’s introduction, neither of which was opened here.

The songs, and the mountain

The popular retellings of the Śavaripa life all narrate the same story: a hunter in the Vindhyas, two wives, a teacher, an illusory herd of deer that would not fall. Not one element of it has been reached at primary level here, so none of it is narrated here as though it had a source.

What is attested is smaller. Songs attributed to Savara-pāda survive in the Caryāpada corpus, the Old Bengali and Apabhraṃśa songs of the Buddhist siddhas, and Dasgupta gives them in English. Their imagery is a high mountain with a Śabarī girl seated on it, wearing peacock plumage and a string of the small scarlet guñjā seeds at her throat, addressed by a voice that calls the Śavara exhilarated and mad; trees whose branches touch the sky; the goddess Nairātmā; a bed spread in great bliss. Dasgupta reads the mountain and its summit as the yogic body and its centres, which is how the tradition’s own commentaries read it.

Look at what that does to the hagiography. The peacock feathers, the mountain, the forest, the hunter: the props that the popular retellings of the Śavaripa life garble are the props inside the songs the tradition puts in his own mouth. The hagiography and the poetry are made of the same material.

The songs are described above in paraphrase and not quoted. The description is kept well clear of Dasgupta’s own English rendering, and the guñjā is named by its Indian name rather than by the English one he uses. Nothing here is a translation and nothing here is a practice.

Dvivedī works the same corpus from the Hindi side. He discusses a Caryāpada of Śabarapāda, the one that begins with a lone Śabarī wandering in the forest wearing the ear-ring of the vajra-holder, cited to Haraprasād Śāstrī’s Bauddha Gān o Dohā; and he infers from the ear-ring that Śabarapāda either wore the karṇakuṇḍala himself or lived among yogins who did, which is a Nāth marker.

Then, in the same discussion, the item that replaces a rumour.

There is a claim in circulation that Śavara is the deity or guardian of the Jālandhara pīṭha, one of the seats of the Goddess. It arrives through an unattributed reference-compilation layer with no text and no scholar behind it. Nothing is asserted from it here, and nobody should assert anything from it. Something better stands in its place.

What Dvivedī reports is that Kṛṣṇapāda, the Kāṇhapā of the same siddha lists, recalls Śabarapāda with great reverence, and that in one of his dohās, speaking of the supreme station, the abode of mahāsukha, he identifies it as the crest-lotus at the summit of the great mountain called Jālandhara, where Śabarapāda himself dwelt.

That is a different claim from the rumour, and it is a better one. It is not that a Śabara presides over a Śākta seat. It is that a named siddha, in a song, places his own predecessor’s dwelling on the summit of Jālandhara, which is a place-name that also belongs to a siddha on both lists.

A date, a teacher, and a province

Dvivedī gives three more things: a date, a teacher and a province.

He reports Dr. Benoytosh Bhattacharyya’s inference, drawn from the Sādhanamālā, that the first propagator of the Kurukullā cult was the siddha named Śabarapāda, that his time is the middle of the seventh century of the Common Era, and that he was a pupil of Nāgārjuna. Dvivedī adds that Śabarapāda is also named as the original propagator of the Chinnamastā cult, and notes the consequence for sequence: Kṛṣṇācārya comes after Śabarapāda.

That is a scholar’s proposal reported by another scholar, and it is offered here as exactly that. It also sits awkwardly against Dvivedī’s own siddha entry, which brackets a Śabara with Dharmapāla’s reign of 769 to 809. A mid-seventh-century date and a late-eighth-century date are not the same date, which is presumably part of why Dvivedī says two men of the name have been conflated. This treatise does not choose between them.

And in that same entry Dvivedī reports Pandit Haraprasād Śāstrī’s inference, made on the basis of a work attributed to Sabara, that he belonged to the circle of King Indrabhūti and Lakṣmīṅkarā, who spread Vajrayāna widely in Odisha.

Odisha is where chapter 14 is. Odisha is where chapter 15 is. This treatise has been handling three appearances of one word as three unconnected accidents, and Indian scholarship of 1950 puts two of them in the same province.

That is a convergence and it is not a connection, and Dvivedī is the reason to say so. In the very next lines of the same entry he asks whether Uḍḍiyāna really is Odisha at all, and refers the reader back to his own earlier discussion of the question. The identification of Uḍḍiyāna with Odisha is contested, and the scholar this chapter is relying on contests it on the page it is relying on. Nothing here shows that the siddha of the eighty-four and the people of the Ganjam hills have anything to do with each other. What can be said is that the question is a question rather than an absence.

The rival that lives inside this chapter’s own tradition

There is one fact in this treatise more relevant to this chapter than anything else in it, and it works against the chapter’s own title.

Chapter 3 records Goudriaan suggesting, in the paragraph that defines the Śābara Tantras at all, that the term may be connected with the generic name of the Saṃvara or Śambara Tantras of Buddhist Tantrism. Chapter 3 neither adopts that derivation nor dismisses it, and instructs the reader to hold two candidate etymologies open.

This chapter is about a Buddhist siddha, in the Yoginī-tantra world where the Cakrasaṃvara literature lives, and it argues from the identity of a word across traditions. If Goudriaan’s suggestion holds, the word in the corpus’s title is a borrowing out of exactly that Buddhist literature, and it says nothing about speech, forests or the Śabara people. That would not touch Dasgupta’s list or Dvivedī’s lineage, because those concern a personal name, and Dvivedī and Dasgupta both write the man’s name as Śabara or Śabarapāda in Sanskrit and Apabhraṃśa contexts where the ethnonym is doing the work. It would cut the connection this chapter’s title implies, between the siddha’s name and the corpus’s name.

The honest position is that this chapter’s material survives the rival and this chapter’s framing may not. A reader who takes the Saṃvara derivation should read what follows as being about a siddha called Śabara and about the productivity of that word, and should discount everything the chapter implies about the corpus that shares it.

What White gives, and why he stays

David Gordon White’s The Alchemical Body, 1996, is not load-bearing anywhere in this chapter. He stays, in one role, and it is worth saying why.

Three references have been checked against page images rather than against a search index: his Śābara-pā, his Śabara Tantra, and his Śabaracintāmaṇi. Those three are set down with their pages in the sources. Nothing else in White is drawn on here, and nothing else should be added on this treatise’s authority until somebody has looked at the page it would come from.

Counts in White have to be taken from the book and never from a preview. A search for Śabara across a twelve per cent preview of it returns nothing, which is a fact about the preview. Across the whole book that name stands four times and the title Śabaracintāmaṇi three, which is as thin as it sounds, and not every appearance of the name in his pages belongs to this subject at all: one is Sabarakantha, a district in Gujarat, and one is Śabarasvāmin the Mīmāṃsaka, who is chapter 7’s subject. Even that is a floor rather than a total, because text extraction drops table content, and White’s Śābara-pā sits inside his Siddha-list tables.

And the structural caution stands. White never writes the name Śavaripa. Śavari, Śavaripa and Savari return nothing across his book. His index entry is Śābara-pā, and Śābara-pā does not appear in the numbered selection from the eighty-four that he prints, which carries Luī-pā and Matsyendra, Gorakṣa, Nāgārjuna and Jālandhara-pā at the same numbers Dasgupta’s roll gives them. The equation of White’s Śābara-pā with a hunter called Śavaripa is this treatise’s conjecture and belongs to nobody else. It carries nothing, either: White’s Śābara-pā and Dasgupta’s and Dvivedī’s Śabarapāda are the same name in two transliterations, on lists with the same neighbours in the same order, which is a connection by name rather than by conjecture.

So why keep him. Because no Indian source reached in this research supplies the frame he does. Dasgupta gives the Sahajiyā and Nāth background, Dvivedī gives the lineage and the dating, Bagchi gives the Matsyendra material. None of them treats the Siddha traditions as an alchemical enterprise, with mercury, transmutation and the perfected body running through the Sanskrit rasa literature, which is White’s whole subject and the setting in which a Śabara Tantra and a Śabaracintāmaṇi appear together on his shelves. On that specific point he is the best source available and there is no substitute for him. On his ascription of the Śabaracintāmaṇi to Nityanātha Siddha, note that Goudriaan gives a different one, and chapter 3 leaves the divergence open.

Two lists that overlap

Set the eighty-four Mahāsiddhas beside the Nāth siddhas and something obvious happens. Nāgārjuna is on both. Jālandhara is on both. Gorakṣa is on both. The Buddhist lists carry versions of the same figures, sometimes with the same stories attached.

The overlap has a text behind it, and this treatise has been citing that text through the wrong door.

The Gorakṣasiddhāntasaṃgraha, a Nāth compilation, quotes the Śābaratantra by name and reproduces a roll of twelve Kāpālikas headed by Ādinātha and twelve mārga-pravartakas headed by Nāgārjuna, among whom stand Gorakṣa and Jālandhara. Dasgupta prints both lists in English in 1946, with a footnote attributing them to the Śābaratantra as quoted in the Gorakṣasiddhāntasaṃgraha. Dvivedī prints them in 1950, and adds the compiler’s own framing: the Nāth position is the Avadhūta doctrine, and the Kāpālika doctrine too was revealed by the Nātha. The Sanskrit itself is held locally for this treatise and is Tier 0 and quotable. It is the primary, and chapters 3, 5 and 6 should lead with it.

Goudriaan and Gupta report the same material in their 1981 handbook, through an eighteenth-century intermediary: Kāśīnātha Bhaṭṭa, in the Kāpālikamatavyavasthā, quoting freely from the Śābaratantras in an attempt to establish a historical relation between those texts and the Kāpālikas, including the twelve and the twelve who are said to have proclaimed the Śābara mantras. Goudriaan and Gupta then push back at the Kāpālika attribution and prefer a different affiliation for the figures.

Goudriaan and Gupta 1981 is in copyright and is paraphrased above, not quoted. One word in their sentence cannot be settled from the scan available: the affiliation they prefer is read as Kāṭha in the copy consulted, where other chapters of this treatise report it as Nāth. The printed page must decide it. The load-bearing point is unaffected, because on either reading they reject the Kāpālika ancestry.

So the figure named as the first proclaimer of this corpus, in the only attempt anybody made to give it a Kāpālika ancestry, carries a name that also stands high on the Buddhist list of eighty-four. Handle that carefully. Nāgārjuna is among the most reused names in Indian religious history, worn by a second-century philosopher, an alchemist, a Mahāsiddha and probably several men between, and the reuse is the phenomenon rather than the evidence.

The corpora share more than names. They share a landscape, which is the forest and the hill country rather than the city and the monastery. They share an idiom, in which the hunter, the fisherman and the outcaste carry knowledge the ordained do not. And they share a claim, made in different theological vocabularies and amounting to the same refusal: that liberation is available outside the brahminical apparatus, without the ritual grammar, without the initiations a birth qualifies you for, without the language.

Two traditions, arguing separately, reach for the same figure to argue with.

Born from dung, and enlightened

Now go back to where this chapter started, and hold the two judgements together.

The cow’s dung. The mleccha who does not know the word dharma. The low place of birth.

And, in the Mahāsiddha literature, fifth among the perfected.

The same social position is, in one body of writing, excreted by a cow, and in another, the teacher of the teacher of the tradition. Both judgements were made by literate men, in Sanskrit and Prakrit and Apabhraṃśa and Tibetan. Neither was made by a Śabara.

The costume changes, the wearer does not

And now the turn, which is the same turn chapter 4 reached from the other end.

The elevation is still an outsider’s use of the figure. The songs Dasgupta prints are songs about a Śabarī girl on a mountain, composed in a literary Apabhraṃśa by men whose commentators read the mountain as the spinal column. The forest in them is a setting and the peacock feathers are a sign, legible to a literate audience, of a kind of person that audience did not belong to. In the Kuvalayamālā of 779, chapter 4’s Vidyādharas put on Śabara costume to practise the Śābara knowledge, and the same berries and bark appear there six hundred years earlier.

The costume is a stable literary convention across Jain Prakrit and Buddhist Apabhraṃśa. Neither instance of it is ethnography.

So we have two literatures doing two opposite things with one body, and structurally they are the same operation. Admiration and contempt are both things done to the Śabara by somebody else, in a language he did not write, for an audience that was not him.

The costume changes. The wearer does not.

What is not claimed here, and what is still shut

No historical connection between this siddha and the Śābara mantra corpus is claimed. Nothing here establishes a line of transmission from the Buddhist adept to the vernacular incantation literature, and no text has been produced that puts them in the same room. What is established is narrower: that the same word generated a named siddha with a lineage, an ordinal, a body of songs and a proposed date inside Buddhism, several centuries before the vernacular Nāth corpus that is usually credited with the whole business, and that Indian scholarship documented this in 1946 and 1950 while the popular literature was reciting a story about Gorakhnāth.

Abhayadatta’s Caturaśītisiddhapravṛtti remains unopened. It is the received collection of the lives of the eighty-four and the right door for the hagiographic narrative specifically: the teacher’s name, the sequence, the wording, and the variation between the Sanskrit and Tibetan lines. White dates it to the eleventh or twelfth century, notes the mid-fourteenth-century Tibetan version, and records from Keith Dowman’s Masters of Mahāmudrā (SUNY, 1985) that the Acts were narrated by Abhayadatta rather than written by him.

Two Indian sources behind the two this chapter uses are fixed at page level and not read. P. C. Bagchi’s introduction to the Kaulajñānanirṇaya (Calcutta Sanskrit Series III, 1934) is the authority Dasgupta cites for the Luī-pā and Śavari-pā relation, and for the Candra-dvīpa question. The 1934 Calcutta Sanskrit Series edition is open on the Internet Archive, its text layer reads the Roman introduction correctly, and Savari, Lui and Candradvipa all answer a search in it. It is available, it is searchable, and it is unread here. Rāhul Sāṅkṛtyāyan’s siddha essays, in the archaeology number of the Hindi journal Gaṅgā, are cited by Dvivedī eight separate times and stand behind much of his dating; they are cited here only through Dvivedī and are marked as cited-through. Sāṅkṛtyāyan is the one of the two that is genuinely absent from archive.org.

That is a short list of missing things beside a long list of present ones. The argument stands or falls on material a specialist can check in an afternoon, and this treatise would rather be corrected than believed.

One thing follows whether or not the reading is ever done. If Śabara-ness is a costume, somebody is wearing it and somebody is selling the effect. In the Vindhya songs the wearer is a Buddhist poet. A thousand years later he is a Bombay publisher with a steam press, and the audience is anybody with the price of a pamphlet and the postage. The next chapter follows the costume into the shop.

Two Indian scholars had this man’s lineage in print by 1950.

Nobody selling his mantras has read either of them.