Jagannātha and Viśvāvasu

There is a temple on the coast of Odisha where the god is made of wood, and where, at intervals, the wood is replaced. The rite is Navakalevara, the new body. It was last performed in 2015.

At that moment the brahmin priests step back, and a different body of temple servants takes charge of the images. They are the Daitas. Their right to handle the god at his most vulnerable is explained, in the telling everybody knows, by a descent.

The Daitas are said to come from a forest man.

He had the god first. That is the whole of the claim, and it is why a chapter about a temple sits inside a book about mantras. Before any king, before any brahmin, before any temple, a Śabara is said to have worshipped the deity in secret in the forest, and a king had to send an envoy into the hills to find him. In a book that has spent chapter after chapter watching the word Śabara worn as a costume and used as a slur, this is the one place where a Śabara stands at the centre of a great tradition rather than at its margin.

So the chapter does to the story what this book has done to everything else. Find the text. Date the text. Ask what the record says before that date, and outside it.

What comes back is awkward in a particular way, and the shape of it is worth having in advance. The forest man really is in a Sanskrit text, and this chapter can give you the verses. The Sanskrit is late. And the older Purāṇic telling of the same legend, the one everything downstream is built on, has no forest man in it at all.

None of that touches the Daitas. They are living people with a real and unbroken role in a functioning temple, and their standing does not depend on the antiquity of a legend. The dispute here is about a historical claim, and only that.

It is also a dispute in which serious Indian scholarship sits on both sides, which is why this chapter ends without a verdict. What it can do is set the documents in order and show the reader where the seam runs.

The story everyone knows

Tell it properly first. It is a good story, and millions of people hold it as their own.

King Indradyumna hears of a deity worshipped in secret in the forest, an image called Nīlamādhava, the blue Mādhava. The worshipper is a Śabara, a forest man, named Viśvāvasu. The king sends a brahmin, Vidyāpati, who finds him, in most tellings marries into his family, and is led blindfolded to the shrine. Then the image vanishes. What returns is wood: a log on the tide, from which the images of Jagannātha, Balabhadra and Subhadrā are carved. And the descendants of Viśvāvasu remain in the temple as the Daitas, the servants who take charge of the images at Navakalevara, the renewal in which new images are made.

This is the most widely told instance of tribal ritual authority inside mainstream Hinduism, invoked in schoolbooks, in Odia cultural nationalism, in serious scholarship, and in almost every popular account of Śābara mantras reaching for a respectable precedent. If Śabara ritual knowledge is real, foundational and appropriated anywhere, it is here.

Begin with the clock, because this cult has one, and it is unusually precise.

A king changes the way he signs

In the first weeks of 1230, at Bhubaneswar, a king changed the way he signed his grants.

Anaṅgabhīma III of the Gaṅga dynasty began issuing inscriptions in which he was no longer quite the owner of his kingdom. He called himself the deputy of Viṣṇu, and the son of Puruṣottama, and through 1230 a series of dated donations laid the Orissan empire at the feet of its deity. Kulke’s reconstruction of that year in the Eschmann, Kulke and Tripathi volume works through the documents one at a time, with the dates on them. The ninth of January is the deputy-title inscription at Bhubaneswar. The twenty-third of February gives land to Puruṣottama and to brahmins after a bath in the Mahānadī. The twentieth of March is not the king’s grant at all: it records donations made by his wife, at Kanchipuram, to Viṣṇu-Aṭṭiyūrnātha, by the order, ādeśāt, of Lord Puruṣottama. The fourteenth of May is a pilgrimage to Puri with land to brahmins. Later in the same year, on a date the record does not preserve, Anaṅgabhīma consecrated a Puruṣottama temple in his newly founded capital at Cuttack, the abhinava Vārāṇasī, the new Benares.

A state cult founded in daylight, in stone, with a calendar attached.

Note where the March document was issued. The queen was donating in the Tamil country, a thousand miles from Puri, and the authority she invoked was the order of the deity at Puri. The ideology travelled fast, and it travelled with the household.

Hold on to how much documentation that is. Kings quarrelled over this cult in writing. Land moved. Endowments were engraved. From the twelfth century to the fifteenth there is an epigraphic corpus a historian can walk through, and this chapter will come back to what two volumes of it turn out to contain.

What is in the text

Viśvāvasu is real as a textual figure, and the reference is precise.

Skanda Purāṇa, Vaiṣṇava-khaṇḍa, Puruṣottama-kṣetra-māhātmya, chapter 7, verses 85 to 88. A māhātmya is a glorification, the genre that establishes a sacred place: what happened here, why it matters, what accrues to the pilgrim. Chapter 7 is the story of Indradyumna, ninety-nine verses of it, and inside it a Śabara called Viśvāvasu comes out of the middle of the mountain, grey-haired, his worship of Hari already finished. Nīlamādhava is named a few verses later, at 93 to 99. The Śabara settlement, Śabaradīpaka, sits just before, at 82 to 84.

Be exact about where on the page. This treatise reads the chapter in G. V. Tagare’s English translation for the Ancient Indian Tradition and Mythology series, and Tagare renders verses 85 to 88 as one continuous block. Verse 85 taken alone is a different sentence altogether, about a brahmin bowing his head with palms joined. The reference is therefore to the block and never to the single verse.

The date of that page is the next thing to establish, and the answer is uncomfortably late.

The text is the one known in Odisha as the Utkalakhaṇḍa, and Stietencron, in the 1978 volume, dates it to the early fourteenth century, listing it alongside the Gītagovinda of the twelfth. That is seventy years or so after Anaṅgabhīma’s reconstitution of the cult. And it stands almost exactly alongside the first appearances of the name Jagannātha itself in the epigraphic record: the Śrīkūrmam inscription of 1309 and the Simhāchalam inscription of 1319.

Get the citation for those two inscriptions right, because there is a trap in the footnotes it comes from. Kulke’s own note, number 22 of his chapter, gives the two references with no editor named at all: Śrīkūrmam inscription no. III (1309 A.D.), line 11, Ep. Ind. V, 1898, p. 35; and the Simhāchalam inscription of 1319, SII vol. VI, no. 714. D. C. Sircar appears in the two footnotes immediately above it, once for a 1957 study and once as the editor of the Puri copper-plate of Bhānudeva II in the Journal of the Asiatic Society of Bengal XVIII.1 (1951), p. 25. Two neighbouring notes fuse into one citation very easily, and the fused version is impossible on its face: Sircar was born in 1907, and an 1898 volume of Epigraphia Indica cannot carry a reading of his. The inscriptions and their dates stand on their own.

Sit with the arrangement even so. State religion in 1230. The name in inscriptions from 1309. And the Sanskrit text that tells you a Śabara found the deity first belonging to the same generation. That is the profile of a foundation narrative composed at the moment an institution needs one.

Which is an inference, and is flagged as one. A text can preserve something far older than itself. Purāṇic material routinely does. The date alone settles nothing.

What sharpens it is the version that came before.

The Purāṇa with no Śabara in it

This is the strongest philological fact on the subject, and the reason this chapter is not a shrug.

The Indradyumna legend does not begin in the Puri māhātmya. There is an earlier Purāṇic telling, and it runs across Brahma Purāṇa 43 to 48. In the Schreiner and Söhnen-Thieme electronic text used here, chapter 42 is a Devī chapter, and the cycle opens at 43.1, with a king of Kṛtayuga whose valour equalled Indra’s. Stietencron, working from the Veṅkaṭeśvara edition with its different chapter division, cites the same legend as Brahma Purāṇa 42.16 to 84. The numbering depends on the edition. The content does not.

The king is there. The temple is there. The Śabara is not.

No Viśvāvasu. No Nīlamādhava. No forest worshipper at all. In the opening chapter of the cycle Indradyumna simply resolves to worship Hari, surveys the earth, goes straight to Puruṣottama-kṣetra, performs an aśvamedha, builds a temple and installs Saṃkarṣaṇa, Kṛṣṇa and Subhadrā. There is no search. No envoy. No forest, and no hidden image to find in it.

And the negative is wider than those six chapters. A whole-text search of the Brahma Purāṇa, all 246 chapters of it, returns śabara zero times, nīlamādhava zero and vidyāpati zero. Viśvāvasu occurs six times, across three passages: 39.10, then 132.3, 132.6 and 132.8, then 171.14 and 171.20. All six are the gandharva of that name. 132.3 gives his sister Pippalā, 171.14 calls him the great lord among gandharvas, 171.20 names his son Citrasena. Not one of the six stands anywhere near the Indradyumna material.

A fourth zero is available on the same page and it is a false one. Pulinda is in the Purāṇa. It stands once, at 27.56, in the inflected form pulindāś, inside a geographical roll of peoples that has nothing to do with Indradyumna, and an exact-string search for pulinda does not find pulindāś. That is the failure chapter 10 reports about the spelling आण, where one orthographic form was searched and the silence was read as a fact about the language.

The same trap sits under the Viśvāvasu count. An exact-string search for viśvāvasu returns three of the six, and it loses the three that stand in the genitive viśvāvasoḥ, where the stem vowel shifts and the string stops matching. One mechanism, two words, in one passage of one Purāṇa.

Every count above is a stem search, and so is every other term in the source note. The Brahma Purāṇa does name forest peoples elsewhere in its own text, and it names the gandharva Viśvāvasu six times. It still does not name a Śabara anywhere near this legend, and its Viśvāvasu is never a forest priest.

Read that twice, because it does something a date on its own cannot. We have the same legend without him, then the same legend with him, and we can see the seam.

The story acquired a tribal originator that it did not originally have.

The absence has been checked against a Tier 1 electronic text of the whole Purāṇa rather than taken on report, and the finding stands at full strength. Geib, Die Indradyumna-Legende (1975), is the critical study of the narrative versions and the one work that would settle the stratigraphy: which recension carries what, in what order, on what manuscript evidence. G. C. Tripathi’s own footnote in the 1978 volume says Geib examined the legend in all its versions, Sanskrit and Oriya alike.

What two volumes of inscriptions actually contain

Now the second finding, and it has to be stated at exactly its size, because the tempting version of it is a claim about a record nobody here has read.

The tempting version is that no inscription of any date mentions Śabaras, Viśvāvasu or the Daitas in connection with the Puri cult. That is a claim about every inscription in Odisha. Scoping it to the scholarship consulted rescues nothing, because a reading list is not the epigraphic record.

So two volumes of the record have been opened.

S. N. Rajaguru’s Inscriptions of Orissa is the standard Odishan corpus, published from Bhubaneswar by the Orissa State Museum and the Orissa Sāhitya Akademi. Volume I part 2, covering 300 to 700 CE, and volume II, covering roughly 600 to 1100 CE, were obtained in full text and searched. The result: Viśvāvasu zero, Nīlamādhava zero, Daita zero, Jagannātha zero across both. Puruṣottama occurs once, in volume I. And Śabara or Sāvara occurs only as a personal name and a place name: a witness called Sāvara Nandiśarman, styled Mahattara and Mahāmahattara, three times in volume II by its own index, and a donee village called Sāvarabheṇḍaka in volume I.

That is a real result and it should be stated at exactly its size. In Rajaguru’s volumes I and II, covering 300 to 1100 CE, Śabara appears as somebody’s name and somewhere’s name, and never in connection with a cult; Viśvāvasu, Nīlamādhava and the Daitas do not appear at all. Anything wider needs volumes III to V, which are freely available and were not searched here. Volume I part 2 and volume II stop before Anaṅgabhīma III, so they do not touch the Puri corpus at all. The negative that matters most is still unsearched.

D. C. Sircar is the right authority for the reading of individual inscriptions and Rajaguru for the corpus, and the distinction matters here. Sircar’s Select Inscriptions Bearing on Indian History and Civilization is cited by K. C. Mishra. Sircar should not be asked to carry a universal negative in either direction.

And Śabaras are in the stone, once you look for them outside the Puri cult. K. C. Mishra reads the Sirpur inscription, found by Beglar and noticed by Cunningham in 1884, in which Indravala is the son of Udayana, of the Śabara lineage; he raises the possibility that this Udayana is the Śabara king captured and released by the Pallava Nandivarman, and connects him with Śabarāditya, defeated by Kāmārṇava, the first Gaṅga king, on the peak of the Mahendra. N. K. Sahu, in History of Orissa volume I (Utkal University, 1964), has Śabaras and Pulindas as the aboriginal population of early Kaliṅga, the Gaṅga occupation of Kaliṅga after the defeat of Śabarāditya, and the territory of Ptolemy’s Sabarai on the Mahānadī near Hirakud.

So the honest shape of the silence is this. Śabaras appear in inscriptions as kings, lineages, witnesses and villages. What is missing, in the two volumes searched and in the scholarship read, is any inscription connecting them to the deity at Puri.

And there is a counter-position on the record. Giridhar Gamang argued the historical case for the Śabaras, and Rousseleau reports that his book worked from literary Indian sources and from inscriptions, holding that Śabaras appear regularly through regional history as civilised allies of the main kingdom. The book is Sabara Tribe in Indian History, New Delhi, 2008, and its title page carries the spelling Gamang, which is the spelling used here.

That book is open. A complete scan of the 2008 Kitab Mahal edition, 106 pages, sits on the Internet Archive and is downloadable by anybody. Its optical reading is a Devanagari model run over English and is worthless, so the pages have to be read as images, and nobody on this project has read them. So the position of this chapter is not that its conclusion has survived Gamang. Its conclusion has never been tested against him. Whether his inscriptions concern the Puri cult specifically or Śabaras in Odishan history generally cannot be determined from a report of a book, and a negative asserted against a man whose book you have not opened is not an argument.

Be honest about what a silence can prove even when it is properly scoped, which is less than a positive attestation. Inscriptions record what donors wanted recorded. They are not a census of belief. A servant caste’s origin myth is not what an endowment deed has occasion to mention, and arguments from epigraphic silence have been overturned before.

The affirmers, read at page level

The affirmers’ evidence turns out to be documentary, and it has to be on the table before anything is concluded against them.

K. C. Mishra, The Cult of Jagannātha, Firma K. L. Mukhopadhyay, Calcutta, 1971, is the affirmers’ book. He prints an extract from a palm-leaf manuscript called the Jarāsabar Vaṃśa Bibaraṇī, the dynastic account of the Jarāsabaras, secured by S. N. Rajaguru from the village of Dimirijholā in the Parlakhemundi taluk. The extract has twelve Śabara families on the Mahendra mountain, the Bāra-Ghariās, worshipping and protecting Nīlamādhava installed as a Dārubrahma, a divine log, in the keeping of a Śabara named Satruśāla; then Indradyumna, the capture and release, the divine voice, the temple at Nīlakandara. Mishra holds that it proves the deity was originally worshipped by Śabaras on the Mahendra.

That is an Odia manuscript source, and it is reported here at second hand through Mishra’s transcript. A dynastic account collected in the twentieth century is not thereby old. What it is, certainly, is evidence the sceptical side has to answer.

The rest of Mishra is equally concrete. He reports the Skanda Purāṇa presenting Puruṣottama as the Śabara Devatā, worshipped by Śabaras under Viśvāvasu, where the Padma Purāṇa has Bhils. He works through Sāraḷā Dāsa, the fifteenth-century Odia poet: Vasu Śabara worshipping the stone Mādhava in secret at the deity’s own command because of the Kali age, the deity also called Śabarīnārāyaṇa, and Jarā’s descendants to be known as the Daitas while Vasu’s become the brahmin priests. He sets beside it a different Odia recension, the Deuḷatoḷā of Nīlāmbara Dāsa, in which Vidyāpati is helped by Lalitā, daughter of Viśvāvasu, and Indradyumna meets Viśvāvasu at Banki in Cuttack district. And he describes the Daitāpatis of Śabara origin taking over the fortnightly Anavasara worship after the bathing festival, when the brahmins do not worship.

N. K. Sahu is the honest disappointment, and he has to be described accurately rather than kept on a list. History of Orissa volume I was searched in full. On the Jagannātha and Viśvāvasu question specifically it yields nothing: Viśvāvasu zero, Daita zero, one incidental mention of a Nīlamādhava temple. He belongs in this chapter as a source for Śabaras in Odishan political history and not as an affirmer of the Puri origin story.

The book is not a German book

One structural point, and it is easy to get backwards.

It would be natural to take The Cult of Jagannath and the Regional Tradition of Orissa, Delhi 1978, for the German side of the argument. Read its contents pages and it is nothing of the kind. Seven chapters and the closing essay are by Indian scholars. G. C. Tripathi wrote three of the chapters and the closing essay, G. N. Dash three, including both of those titled “The Evolution of Priestly Power,” and P. Mukherjee one. The volume’s Odia primary apparatus is entirely theirs.

It matters for more than a ledger. This chapter’s central observation about routing is that every citation for the Viśvāvasu and Daita tradition goes back to the Mādaḷā Pāñji, the Puri temple chronicle, which is a narrative source and not an inscription. The man doing the routing is G. N. Dash, an Odia historian working on an Odia chronicle. Dash reconstructs the Somavaṃśī connection with Puri directly from the chronicle’s Rājabhoga section: the recovery and re-installation of Jagannātha by Yayāti Keśarī 146 years after the Raktabāhu invasion, the deity hidden near Sonepur, Yayāti Keśarī identified with Yayāti Mahāśivagupta I of roughly 922 to 955 CE; and a second tradition in which Śiva-Bhuvaneśvara orders Vāsudeva Bāhinīpati to fetch Coḍagaṅga from the south and make him king. His footnotes on that page cite Rajaguru’s introduction of 1971 and P. Mukherji of 1940.

G. C. Tripathi, elsewhere in the same volume, prints four verses of Sāraḷā Dāsa’s Odia Mahābhārata in transliteration, from Madhyaparva twice, Sabhāparva and Vanaparva, identifying Jagannātha as the Buddha incarnation of Viṣṇu; and he summarises Sāraḷā’s story that Kṛṣṇa’s body, after he was shot by a Śabara hunter, would not burn, floated to Puri, remained in the custody of the Śabara, and became the tree from which the image was carved.

And the chronicle itself has now been opened. The Mādaḷā Pāñjī, Prāchī edition, edited by Ārtta Ballabha Mahānti, Principal of Cuttack Government College, published by the Prāchī Samiti, Cuttack, 1940, reprinted by Utkal University in 1969, is the edition the 1978 volume’s own bibliography uses. A full scan was obtained and the Daitāpatis are present in the narrative text. Nothing from it is quoted here.

The scholar who does not say what she is said to say

Here the popular literature goes wrong in the most citable way possible: by naming the right authority for the opposite conclusion.

Anncharlott Eschmann is the scholar most often invoked for the tribal thesis. Her work on the Hinduisation of autochthonous deities in Odisha founds the whole modern discussion, and her chapter in the 1978 volume is titled “Prototypes of the Navakalevara Ritual and their Relation to the Jagannātha Cult.” That is the chapter people reach for when they say Eschmann proved the renewal ritual is tribal.

Read the chapter and the parallel she draws is Khond.

Her argument runs on wooden posts. On her account the Saora village shrine of south Odisha consists of a post with a pot hung on it, and what is relevant to the worship is the pot rather than the pole. From that she draws the consequence that repairs or even replacement of the post are not an important ritual, and she reports the remark an old Saora made to her about it. Nine words. The whole of their content is that one would simply go and get a fresh piece of wood.

Then comes the sentence that is easiest to harden into a proof, and it has to be given its own shape. She does not say that renewal ritual is tribal and Khond and therefore not Saora. She writes that one might venture the hypothesis that rituals of renewal can occur only where the post itself, rather than a more complicated structure that could be repaired, is taken as the symbol of the deity. She then says the hypothesis would explain why such a ritual is found mainly in central and western Odisha, and she inserts, in her own parenthesis, the qualification that this is as far as she could find out up to now. That is a region where the lady of the post, Stambheśvarī, has been worshipped since the sixth century, and where most of the tribal cults concerned are Khond.

Nothing from the 1978 volume is reproduced verbatim in this chapter. The nine-word informant line and Stietencron’s clause in his own footnote were both verified against the page image, and both are reported in paraphrase rather than quoted, so that the apparatus matches the one this treatise applies to Dasgupta, Dvivedī, Elwin and Goudriaan and Gupta. Eschmann’s own two hedges, the ventured hypothesis and the provisional scope, are paraphrased above for the same reason.

Say plainly what this does and does not do. Eschmann’s whole case is that the Puri renewal has a tribal antecedent, and her position sits on the historical side of the ledger. What her argument does not support is the identification of that antecedent with the Saora, and that identification is the only thing making the Jagannātha story usable as evidence about Śabara ritual knowledge. Her own finding about the Saora post is that its replacement is not an important ritual. It would be easy to turn that into ritual irrelevance, and then into field testimony that ends the argument. It is one scholar’s observation, one informant’s remark, and one hypothesis she marks as ventured.

That is still enough to break the popular use of her name.

The name as a claim

Stietencron, in the same volume, makes the move explicit in a footnote most readers never reach.

Writing on the early temples of Jagannātha, he notes that the Khonds and communities related to them are the only groups practising a ritual renewal of wooden posts today, as Eschmann has shown, and then adds the sentence that reframes the question. He writes that the very description of the ritual experts as Śabaras may itself be the outcome of an attempt to attach them to the Śabaras of the older Indradyumna legend.

He offers it as a possibility. Look at the direction it runs.

In the popular telling the Daitas are called Śabaras because they descend from Śabaras, and the legend records the descent. In Stietencron’s reading the arrow reverses. The legend exists first. The ritual experts, whoever they were, are attached to it afterwards, and the ethnonym is the attachment.

The name becomes a claim made later.

G. N. Dash comes at the same material from another side. His two chapters are both titled “The Evolution of Priestly Power,” and the title is the interpretation. Dash reads the versions that come down through Sāraḷā Dāsa as a charter for rank: who stands where inside the temple establishment, which servants displace which others, at which moments. A story about precedence, produced by people with reason to care about precedence. Dash himself takes the tribal element as substantially historical. His reading is still one in which the story does work in the present of its composition.

A narrative about origins, told inside an institution, is always also a document about the institution.

The gazetteer trap

One more thing, and it is the error most often committed in the secondary literature, including by careful people.

O’Malley’s Puri Gazetteer of 1908. In his caste notes he describes the Daitas: a small caste found only in Puri, who displace the brahmin priests at certain festivals, who do the work when old images are replaced by new, who hold rent-free villages for their temple service, and who, he writes, “are believed to be of Savar origin.”

It is quoted constantly as independent confirmation: a colonial officer, an official document, apparently outside the temple and its interests.

Look at the verb. Believed. O’Malley is not reporting a finding. He is recording what people in Puri told him they held to be true, which is his job, and he marks it as such.

So citing him alongside the māhātmya does not give you two witnesses. It gives you the same temple tradition twice, the second time in an official binding.

Double-counting a source is not corroboration.

Where this actually stands

Now the honest map, without a verdict pretending to be on it.

Taking the tribal origin as substantially historical: K. C. Mishra, read here at page level and arguing from a palm-leaf manuscript, the Sirpur inscription and two Odia recensions; G. N. Dash, who reads the tradition as a priestly charter while accepting its tribal element; Padhi; Giridhar Gamang, unread; and Eschmann, heavily qualified and pointing at the Khond rather than the Saora. N. K. Sahu should come off this list, because his volume does not carry the claim.

Sceptical: Stietencron; Kulke; Rousseleau (2022); P. K. Nayak; and G. C. Tripathi, who has to be qualified as carefully as Eschmann does, because filing him flatly among the sceptics misreads him. In the 1978 volume he writes, in a sentence paraphrased rather than quoted for the copyright reason above, that the Śabaras may well have been worshipping the original Jagannātha in a forest, and that what is inconsequent is the further claim that the object of that worship was a finished stone Nīlamādhava made to the rules of Hindu iconography, since the Śabaras are not known to have possessed images of that kind. His scepticism is aimed at the Nīlamādhava element.

Serious people on both sides, working from the same texts and the same stones and reading them differently. The tribal origin of the Puri cult is a live dispute, and this chapter does not close it and could not. What can be said without adjudicating is that the popular telling presents the contested side as settled.

The reading behind this chapter, stated plainly, because a reader is owed it. Stietencron, Kulke, Tripathi, Eschmann and Dash have been read at page level in the 1978 volume. Mishra 1971 and Sahu 1964 have been read at page level. The Mādaḷā Pāñjī edition is open and is not readable at page level here. Rajaguru’s whole run is open on the Internet Archive, volume III in its two parts of 1960 and 1961, volume IV of 1966 and volume V in its three parts of 1975 and 1976, and volumes III to V are unread. Sircar’s Select Inscriptions volume I is open in both the 1942 and the 1965 editions, which is its own small problem, because Mishra never says which of the two he used and they paginate differently, so his pp. 35-36 cannot be repeated until an edition is attached to it. Gamang is open and unread.

So the instruction to a reader is narrow and specific. Discount the epigraphic negative, which rests on two volumes of a five-volume corpus. Discount everything this chapter concludes against the affirmers, because the one affirmer writing from inside the name has not been read at all.

The costume, again

Here is why this chapter belongs in a book about mantras, and it is not because Puri has anything to do with them.

Chapter 4 went to the year 779, to a Jain Prakrit romance in which the practitioners of the Mahāsābarī vidyā, the great Śābara knowledge, turn out to be Vidyādharas in Śabara costume. Dressed as Śabaras.

Chapter 2 found the lexicons doing something adjacent: a named people flattened into a term for the savage, then for barbarous speech, then for sorcery, each sense borrowing prestige from the last.

And here, in the one place where Śabara ritual authority is supposed to be documented rather than merely invoked, a legend circulates without a Śabara in it, an institution consolidates, a text is composed, and the legend now has a Śabara in it, and a name, and a lineage of servants attached to that name, and eight hundred years later it is the thing everybody knows.

That is a reading, and Mishra’s palm leaf is the reason it is only a reading. If the Jarāsabar Vaṃśa Bibaraṇī carries anything old, the seam this chapter thinks it can see is somewhere else.

What can be said is that the story cannot carry the weight the mantra literature puts on it. It is contested at its foundation, late in the surviving Sanskrit text, absent from the earlier Purāṇic telling, and the version that names the Saora is the version its own most-cited authority declines to endorse.

The next chapter goes looking for the same word in another tradition entirely: a Buddhist adept called Śavaripa, who turns out, on Indian scholarship of 1946 and 1950, to stand one generation above the founder of the Nāth line.

The name keeps arriving.

This time, so do the sources.