The Prophets among the authorities

Mohammadā, in a list of hero-spirits

Somewhere inside a fifty-rupee Delhi paperback of Śābara mantras, printed in a new edition in 2016, whose commonest name by a long distance is Gorakhnāth, stands a name in Devanāgarī. मोहम्मदा. Mohammadā.

It is there four times, and every time in the same character: as a vīr, a hero-spirit, one of the powers a speaker reaches for. Once it stands inside a numbered sequence of vīrs, in company with Kāl Bhairav, Mahāvīr and Āgyā Vetāl.

Nobody arranged that as a statement. There is no preface announcing an intention, no note explaining the presence, nothing anywhere in the book that treats it as worth a remark. The name is in the stock, in the same list, in the same register, doing the same work Gorakhnāth does.

It is not alone in there either. Khwāja Khiḍr, the water-saint of North Indian and Bengali popular devotion, is in the book. Hasan and Husain are in it, as a pair. The Pāñc Pīr, the Five Saints of North Indian village practice, appear by name. The Seventy Pīrs are carried in the standing enumerated stock in at least four places, in exactly the same breath as the Yoginīs and the Bhairavs. The Qur’ān is named twice. Khudāī, sāheb, fakīr and autār stand once each.

A reader could go looking for all of that and come back reporting that almost none of it is there. It depends entirely on how it is looked for.

Somebody wanting to know whether the prophets figure in this material can type पैगम्बरों into a search box, get one occurrence back, and draw the inference that one occurrence invites: an attestation, a curiosity at the edge of the material, nothing anybody could build on. The inference does not hold. One inflected form of one word, searched once, produces a silence that is a fact about the search rather than a fact about the material, and chapter 10 states the rule that follows from it.

Counted as a lemma set instead, the Islamic vocabulary in that one paperback runs to roughly two dozen occurrences across at least eight distinct terms.

So the finding in this chapter was produced by a regular expression, and the first thing to say about it is what a regular expression can and cannot deliver.

There is no manuscript here, no palm leaf, no colophon, no dated edition read in a library. There is a text file, a search pattern, and a count. The file is Durlabh Śābar Mantra, presented by Tāntrik Bahal, Rājā Pocket Books, Delhi, new edition 2016, read from an archive.org scan whose optical character recognition is poor. There are now eleven such files, eight of them independent Śābara compilations and three of them controls, and chapter 10 publishes the extraction, the search strings, the normalisation rules, the rate table and the roster of names. This chapter uses them and does not re-derive them.

The counts around the naming construction have to be read with their search strings attached, and chapter 10 attaches them. What comes out of that chapter and matters here is one finding, and it can be stated in words. Read by hand across every occurrence of the appeal word duhāī in this book, nearly all of them stand with a named authority beside them. That is one reader’s hand-check of a poor scan rather than a machine count, it is approximate on purpose, and it wants a second reader. How the elements stand in relation to one another is not reported in that chapter or in this one.

So much for the counting. The register itself is the larger matter.

The Islamic register, recounted

Here is the whole of that lemma set, itemised, because a claim of this kind is worth nothing without its list.

पीर, pīr, seventeen occurrences of the token and eighteen of the string. पैगम्बर, paigambar, twice, in two different forms, only one of which a search for the plural oblique returns. मोहम्मदा, four times, always as a vīr, a hero-spirit, and once inside a numbered sequence of vīrs alongside Kāl Bhairav, Mahāvīr and Āgyā Vetāl. ख्वाजा and खिदर once each, together forming Khwāja Khiḍr, the water-saint of North Indian and Bengali popular devotion. हसन and हसैन once, as a pair. कुरान twice. खुदाई once. साहेब once. फकीर once. औतार once.

Two named cults are in there. The Pāñc Pīr, the Five Saints of North Indian village practice, appear by name. So does Khwāja Khiḍr. And the enumerated stock that recurs through the book, the standing list of sixty-four Yoginīs and fifty-two Vīrs and their companions, carries the Seventy Pīrs in at least four places, in exactly the same breath as the Yoginīs and the Bhairavs. The frequency is reported; the phrase itself is not printed, here or anywhere in these chapters, because it is a construction and this treatise prints terms.

Those are token counts on bad optical character recognition, folded across the spelling variants this research could identify, by one reader. The folding rule was visual similarity plus contextual identity and it was not applied blind. A second reader and a better scan would change some of these by one or two. They would not change the shape.

The same trap opens over and over. A reader searches a single form of a word, finds almost nothing, and concludes that the thing itself is marginal, when what is marginal is the spelling. Chapter 10 meets that trap again over आण and states the rule that follows from it: a paperback read through a single search pattern will not carry a claim about a corpus.

So the precise claim about one book is this. In Durlabh Śābar Mantra, Islamic figures are named across at least eight distinct terms, in the same enumerated lists as the Nāth and Śaiva names, and twice in the appeal construction itself.

That claim has been tested against seven more books, and it is too small.

The register, on seven more witnesses

Chapter 10 has run the identical count over eight independent printed compilations from six houses in Delhi, Vārāṇasī and Prayāg, with two occult books from the same commercial shelf as genre controls. Two rows of that table belong to this chapter, and both of them run well above the Delhi paperback.

Rates per 100,000 Devanāgarī tokens, absolute counts in brackets, all token counts.

Two occurrences of paigambar became thirty-three.

Two occurrences of the word for prophet in one Delhi paperback was the thinnest evidential base anywhere in this treatise. Widening the base to eight independently compiled books did not weaken the claim. It strengthened it. Paigambar stands thirty-three times in the largest of the eight, and four of the seven books added to the comparison carry it at rates above the paperback’s. The rates are in the table above and chapter 10 sets out how they were arrived at.

Pīr moves the same way and further. Two hundred and twenty-one occurrences of the word in the largest of the eight, at 100.2 per 100,000 tokens against the Delhi paperback’s 29.0.

And in the two genre controls, one occurrence of pīr and no prophets at all.

So the two occurrences of paigambar in the Delhi paperback understate the register by a factor of between four and ten. The Islamic register is a feature of the printed popular Śābara corpus on eight independent witnesses, it is denser in the scholarly Banaras imprint than in the Delhi pocket book, and it is absent from the neighbouring bazaar genres that share the same shelf and the same customers.

Four cautions on that table, and the first is large enough to govern the whole of it. Every pīr figure above is an upper bound and the next section explains why. The paigambar fold covers seven spellings and would miss an eighth. Three of the eight books carry no paigambar at all, so the register is dense across the corpus without being universal in it, and a claim that every Śābara compilation names the prophets would be false on this table. And the Delhi paperback’s pīr figure here and chapter 10’s roster figure for the same book were reached by different rules, so that they land as close together as they do is a coincidence and should not be read as a confirmation.

Pīr, and which pīr

That count has a problem inside it, and the problem is easy to walk past.

Pīr in Hindī and Urdu is the Sufi title, the spiritual master. Pīr in Braj and in the Hindī of charm literature is also an ordinary noun meaning pain. The two are homographs. A count that reports eighteen pīrs without saying which is not reporting anything.

So the eighteen have to be read one at a time, and this is the adjudication the chapter owes.

Ten of them stand in the honorific sense, and the itemisation reaches ten, because a chapter whose whole point is that a published total can be half a different word cannot print a total its own list does not reach. Four are the Seventy Pīrs inside the enumerated invocation stock, where the surrounding names are all beings. One is the Pāñc Pīr. One stands beside autār and paigambar in a list of powers whose grace is being invoked. Two fall inside a single item in which the honorific and nāth both stand, which is a small and interesting collision on its own, and that item carries the word twice rather than once. One is a pīr in a duhāī. The last is paired with guru.

Six or seven of them are pain. They occur in charms against scorpion sting, against toothache, against a child’s sickness, and in one proverb about one’s own pain and another’s. Nobody invoking a Sufi saint says the thing these lines say.

One is unclear and is left unclear.

That adjudication is the chapter’s actual contribution to the pīr question, and it cuts both ways. The Islamic register is real and is larger than reported. It is also smaller than a naive string count would make it, and anybody who runs this search and publishes eighteen has published a number that is half a different word.

That adjudication has not been done for the other seven books, and it will not be pretended that it has.

Two hundred and twenty-one occurrences of pīr cannot be adjudicated by hand on a poor scan without producing a number worse than the one it replaced. So the corpus-wide pīr figures in the table above count a homograph and nothing finer. They are upper bounds on the Sufi sense, and no proportion is claimed for either sense in any book except the one this chapter read line by line.

The caveat bites hardest exactly where the numbers look most interesting, and that has to be said plainly. The two books with the highest pīr rates relative to their size, after the Chaukhambā compilation, are the Manoj volume and one of the Prayāg parts, and both of those are heavily organised around ailments and their remedies. A book about pain will use the word for pain. There is no way to tell from a rate table how much of 86.8 per 100,000 is a Sufi master and how much is a toothache.

What survives the caveat is the paigambar column, which has no homograph problem at all, and the direction of the pīr rise in the largest witness. A jump from 29.0 to 100.2 is large enough that homograph noise is unlikely to account for all of it. How much of it, this chapter cannot say, and it says so rather than estimating.

A list, and nothing holding it together

The headline numbers are the least interesting part of the result.

A pantheon has a structure: relations of descent, of consort, of hierarchy, of function, and a theology that says why these beings and not others. Any of the Purāṇic god-lists is a pantheon in that sense. So is the āvaraṇa scheme of a Śrī Yantra, where every deity has an assigned enclosure and the enclosures are ordered.

What chapter 10’s roster shows is not that. Gorakhnāth at 379. Hanumān at 107. Īśvar at 80. Gaurā and Pārvatī together at 58. Bhairõ at 36. Mahādev at 33. Then a long thin tail: the Yoginīs, the Nau Nāth and the Caurāsī Siddh, Kāmākṣā, Rāmcandra, Lonā Camārī, Kālkā, the Seventy Pīrs, Mohammadā, Bhaĩsāsur, Mansā, the Prophets, Viṣahārī, Khwāja Khiḍr.

That roster comes with chapter 10’s caution attached, and the caution travels with it. It is one reader’s hand-check of a bad optical character recognition pass, sixteen name-families were folded on a rule of visual similarity plus contextual identity, the folding was not applied blind, and it wants a second reader and a better scan. A second reader would move several of these figures by one or two. The argument below turns on the ordering and the shape of the tail, and neither of those is within reach of that margin.

The corpus has since supplied a machine measure of the same thing, and chapter 10 publishes it and keeps it apart from the roster, for a reason that governs how it can be used here. The roster counts referents, with every epithet and every spelling of one being folded by hand into one line. The machine counts forms, so that every spelling, every inflection, every epithet and every piece of optical noise stands on its own. The machine’s figures come out far larger than the roster’s for that reason and no other, and setting the two side by side without saying so would manufacture a contradiction out of arithmetic.

No name behind either figure is printed in this treatise. The machine’s list for the Chaukhambā compilation exists and stays unpublished, because a description of the frame and a run of its fillers on the same page is assembly by instalment.

What the larger figure does show, and it is the only thing it shows, is that the slot grows with the book. Three and a half times the tokens, four times the distinct forms. Nothing closes the list.

There is no scheme in that. Nobody has worked out how Mahādev stands to a Prophet.

What it is, is a list of people worth appealing to. Assembled by whoever was speaking. Drawn from whatever was locally powerful. Added to when something became powerful enough in that district to be worth naming, and dropped when it stopped being. There is no committee behind a list like this and no doctrine underneath it. There is a working practitioner with a client in front of him and a stock of names that have carried weight within living memory.

That is a sociological document. It is not a theology, and reading it as one produces immediate nonsense.

Which is why the roster is more valuable as a map than as a creed. A book that appeals to Mahādev, to Gorakhnāth and to the Prophets in the same register was not produced inside a sealed Hindu ritual system. It was produced in a mixed North Indian world, where a healer’s clientele crossed the confessional line and so, of necessity, did the healer’s stock of authority.

And here the discipline has to hold, because this is the point at which the material could be made to say more than it does. The evidence is an invocation list in one printed book. It does not show that Muslims used these formulae. It does not show that a Muslim practitioner composed any of them. It does not show a single Muslim reader, buyer or client. It shows that whoever assembled this compilation, or the material behind it, kept these names in the stock. Everything past that is inference and is labelled as inference every time it is made.

The phoneme and the appeal

Set this against chapter 7 and the shape of the treatise comes into view.

The Mīmāṃsā position, built there in its strongest form because everything after is a refusal of it, locates the power of sacred speech in the phoneme. The relation between word and meaning is unmade and beginningless. The Veda has no author. The sound is efficacious because it is that sound, correctly produced, in the correct sequence, by a speaker qualified to produce it. Alter the sound and you have not made a variant. You have made a different thing, and it does not work.

A doctrine of that shape has no room at all for an appeal to a Prophet. It has no room because the question does not arise in its vocabulary. There is nothing to appeal to. Nobody grants the mantra’s effect, so nobody can be asked.

The Śābara register has room for everything. Chapter 10’s roster of this one book names twenty distinct beings or groups of them, on the folding rule that chapter publishes. Nothing about that figure is a ceiling. Take a book three and a half times the size and the names standing in the authority slot multiply several times over. Nothing structural stops the list growing, because the efficacy has been relocated. It sits in the act of appealing, and an appeal takes a name.

There is an Indian statement of exactly that principle, in a tradition this treatise has barely touched, and it is older and better than anything the popular literature offers. Shashibhusan Dasgupta, in the first edition of Obscure Religious Cults (University of Calcutta, 1946), reports from the Vimalaprabhā, the great commentary on the Laghukālacakratantra, that the doctrines of the pāramitā way are set down wholly in Sanskrit, while those of the mantra way are explained in Sanskrit, in Prakrit, in Apabhraṃśa and in non-Sanskritic speech, and that the Śabaras are named as the example. Dasgupta read that in manuscript. A printed edition of the Vimalaprabhā now exists at Sarnath and this treatise has not opened it, so the passage is held at one remove through Dasgupta and is reported as his.

If that holds, vernacularity in mantra is a stated principle of an eleventh-century Buddhist Tantric commentary, from a different tradition entirely, naming the same people. The openness of the Śābara list stops being an oddity of bazaar practice and becomes an instance of something the learned tradition had already articulated.

This openness is not tolerance. Nobody in this book decided to be inclusive, and no ethical achievement is being recorded. It is a structural consequence. Locate power in the correctly produced phoneme and the boundary is absolute and policed by phonetics. Locate power in the appeal and the boundary gives way on its own, because an appeal has an addressee, and addressees can be added.

What it descends from, and what is not known

Now the honest part. This chapter names an Islamic presence in one printed book and cannot say what it descends from.

Two authorities are named in the research file and neither has been opened: Simon Digby on jogīs in Sufi hagiography, and Carl Ernst on the Amṛtakuṇḍa. Nothing is attributed to either here. What can be said without opening them is what a reader would go to them for. Digby is the standing reference for the encounter between Sufi shaikhs and Nāth or Nāth-adjacent yogīs in North Indian hagiography. Ernst is the standing reference for the Amṛtakuṇḍa, the Sanskrit yogic text that travelled into Arabic and Persian and circulated in Islamic esoteric literature. Whether either supports what this chapter has counted, this treatise does not know, and no proposition from either is asserted anywhere in it.

That is a gap, it is named as a gap, and closing it is the highest-value unopened item in this section.

The volume that would settle it

There is a second gap, and it is the one that matters most to this chapter.

The Śābaramantrasāgara, the Chaukhambā compilation that carries the highest Islamic-register counts in the whole comparison, is a pūrvabhāga. A first part. Its compiler, S. N. Khaṇḍelvāl, announces in his preface that a second division will follow, and says where he went to gather it: to faqīrs and ʿāmils at Ajmer Sharif and at Deva Sharif, and to Jain informants in Karnataka and Rajasthan.

The announced second division is a public document, and its shape can be stated exactly. It is openly downloadable, a full public scan, on the same platform that yielded all eleven of the books counted here. Its title page carries चौखम्बा सुरभारती ग्रन्थमाला 568 and शाबरमन्त्रसागर (उत्तर भाग) over Khaṇḍelvāl’s name: the same compiler, the same Chaukhambā Surabhāratī house and the same series as the pūrvabhāga every count above was taken from. The archive record for it carries almost no metadata, which is the likeliest reason none of the seven search strings chapter 10 publishes reaches it, and that is a guess about a search rather than a finding about the book. Its own preface gives its scope as the sects current today, हिन्दू, मुस्लिम, सिख and the rest, which is Hindu, Muslim and Sikh, and not as Muslim and Jain material alone. Across the whole volume जैन stands twice, against ten occurrences of मुस्लिम. Nothing in it has been read here, and no count in this chapter has been re-run on it.

That volume is precisely the one this chapter’s argument predicts. A compiler who has already put thirty-three occurrences of paigambar and two hundred and twenty-one of pīr into his first part, and who then goes to two of the largest Sufi shrines in north India and to Jain informants in two states to gather a second, is doing in 2013 what this chapter says the corpus has been doing for a century: assembling whatever authority is locally available and printing it in one register. What that volume contains would confirm the reading, complicate it, or overturn it. It is the single most valuable unread item in this treatise, it is obtainable by anybody, and the test is one somebody can run.

Two limits on what the preface can be used for, and they are the same limits chapter 10 states for every preface in this set. It is not evidence that any of the material in it came from where he says it came from. It is not evidence about Muslim or Jain practice. It is evidence that a compiler in Vārāṇasī, preparing a book for a Sanskrit scholarly press, announced that intention in print in 2013 and gave the places by name.

The question of where

The count is Hindī-belt evidence. The book is a Hindī book, printed for a Hindī readership, and chapter 17 traces the printing circuit to Bombay and Banaras. Every heavily recurring name in it carries weight in North India. That is the limit of what the count licenses, and it produces a genuine problem for this treatise.

Chapter 3 builds the corpus’s charter from the Śābaracintāmaṇi entry, which Goudriaan and Gupta report and which this treatise paraphrases: Ādinātha, named there as the teacher of Matsyendranātha, receives the mantras through Īśvara’s favour and gives them out to five pupils, in five languages, those of Gauḍa, Kerala, Karṇāṭa, Andhra and Gurjara. Those five are not the pañca-drāviḍa, and a reader with any Purāṇic background will see why at once. Pañca-drāviḍa is a fixed term for the five southern brahmin divisions, conventionally Drāviḍa, Karṇāṭa, Gurjara, Mahārāṣṭra and Tailaṅga. Gauḍa heads the complementary pañca-gauḍa, the northern five.

And the mismatch between the title and the list is itself evidence. The text’s own title carries Pañcadrāviḍa, and the list of five languages inside it includes Gauḍa, which does not belong to that pentad. A title claiming the southern five, over a list that reaches into the north, is a text claiming more ground than its own enumeration covers, which is exactly the point this chapter is making about pan-Indian claim against North Indian practice.

Two things about the sourcing of that charter. The Ādinātha material itself is available in Indian scholarship earlier and in more detail than in the Western catalogue this treatise routed it through: Hazārīprasād Dvivedī gives it in Hindī in Nāthasampradāya (1950), with both lists of twelve in his own note, and Dasgupta gives the two lists in English in Obscure Religious Cults, sourced to the Śābara-tantra as quoted in the Gorakṣasiddhāntasaṃgraha. Chapters 3 and 6 should lead with those. The second thing is the edition behind the page numbers. Every reference to Dasgupta in this treatise is to the 1946 first edition, and the numbers are that edition’s own running heads. The later Firma K. L. Mukhopadhyay edition paginates differently by some thirty pages, so that in the 1946 book its pp. 206 to 207 are on Sufi and Baul love-mysticism and not on this material at all. A reader following a citation into the wrong edition finds nothing, and that is worse than a vague reference. Anyone checking this passage against Lorenzen’s footnote or against Goudriaan and Gupta’s note should know that both of those use the other edition’s pagination. What Goudriaan and Gupta supply, and what no Indian source found here supplies, is the catalogue entry for the (Pañcadrāviḍa)śābaracintāmaṇi and its five named vernaculars. That single sentence is theirs, it is their own bibliographical work on a manuscript catalogue, and it is kept for that reason and no other.

Nothing in the present research shows a living Śābara practice in Kerala, in Karṇāṭa or in Andhra. What the charter may record is a claim to pan-Indian scope made by a North Indian text, which is a familiar and respectable kind of claim and a different thing from a report.

There is a second geography this treatise has not engaged. Kāmrūp, western Assam, functions in North Indian and Persianate imagination as the country of enchantment itself, and the trope is old and durable and turns up wherever this material is discussed. The book counted here reaches for it: Kāmrū and Kāmākṣā recur in it. A Routledge volume of 2023 associated with Sajal Nag, Black Magic, Witchcraft and Occultism: Secret Cultural Practices in India, is the obvious thing to open, and this treatise has not opened it. Catalogue records disagree about whether Nag is its editor or its author, and its scope is India-wide rather than specific to Kāmrūp. It is listed here as an open item and it is not used for anything.

Against the comfortable reading

There is a version of this chapter that would be much easier to write, and it is wrong.

In that version the roster becomes a heartening fact about Indian pluralism. Hindu and Muslim authority side by side in one humble formula, the syncretic genius of the subcontinent, the folk wisdom that never needed to be taught tolerance. It writes itself. It would be quoted.

What the count shows is more specific and less flattering: a practical assembly of whatever authority happened to be available, made by people who needed a result, for clients who needed a result. Nobody in that room was celebrating plurality, and nothing here belongs to the devotional history the Bhakti movement records. The Prophets are in the list for the same reason Gorakhnāth is in the list. Somebody believed that name would carry, in that district, with that person, on that day.

That is a poorer story about intentions and a much better one about how people actually live next to each other. Syncretism as a virtue requires somebody to have chosen it. Syncretism as a working assumption requires only that the wall was never built high enough to notice.

One number stands behind that sentence now, and it is worth ending on because of where it comes from. The Islamic vocabulary counted here is thickest in the most respectable book in the set: a Sanskrit scholarly imprint at Vārāṇasī, with a series number and an ISBN and a price of ₹625, whose compiler doubts in his own preface that any of this material has anything to do with Gorakhnāth. Two occurrences of the word for prophet in a ₹50 Delhi paperback. Thirty-three in the Banaras book.

The wall is lowest where a reader would expect it to be highest.

That is a remark about one imprint, made on one word, and it is offered as an observation rather than a finding. What would test it is the second volume, which is downloadable and has not been read here.

The next chapter stays inside the language and asks the harder question this material raises about meaning itself: why the syllables are called ill-matched, what the Hindī commentators actually did with the phrase, and why Frits Staal turns out to be the wrong authority to settle it.

They did not build a pantheon.

They kept a list of who answers.