The Shiva Sutras: the rock, the count, and the commentary that won

The rock

Late in the eighth century, or early in the ninth, on a mountain above the Kashmir valley, a man went looking for a rock.

He had been told where to find it. The telling had come in a dream, and the dream is reported by a commentator who wrote roughly two hundred years later, which is the first thing worth saying plainly about the Shiva Sutras: almost everything we are told about how they arrived comes from people who were not there.

The commentator is Kshemaraja, pupil of Abhinavagupta, and he opens his Shiva Sutra Vimarshini with a scene. There was a teacher in Kashmir, he writes, a great devotee of Maheshvara, whose devotion had opened under the descent of grace. He had refused the teachings of Nagabodhi and other adepts who stood in the lower view. His heart had been purified in the true tradition of the many yoginis and siddhas of the supreme Lord. He lived on the glorious Mahadeva mountain, and his name was Vasugupta.

इह कश्मीरेषु शक्तिपातवशोन्मिषन्माहेश्वरभक्त्यतिशयात् … श्रीमहादेवगिरौ महामाहेश्वरः श्रीमान् वसुगुप्तनामा गुरुरभवत् ।

iha kashmireshu … shrimahadevagirau mahamaheshvarah shriman vasuguptanama gurur abhavat

“Here in Kashmir … on the glorious Mahadeva mountain there was a teacher, a great devotee of Maheshvara, named Vasugupta.”

Kṣemarāja, Śivasūtravimarśinī, p. 2 #

Then, Kshemaraja says, once, Paramashiva favoured him in a dream. The reason given is not private and it is not mystical. It is political, in the way the history of doctrine is political. The secret tradition was about to be cut off in a world all but saturated with the dualist view, and Shiva, intent on giving grace, woke his insight and spoke:

“अत्र महीभृति महति शिलातले रहस्यम् अस्ति तद् अधिगम्य अनुग्रहयोग्येषु प्रकाशय” इति ।

atra mahibhriti mahati shilatale rahasyam asti tad adhigamya anugrahayogyeshu prakashaya

“Here, on this mountain, on a great rock-face, there is a secret. Obtain it, and reveal it to those fit for grace.”

Kṣemarāja, Śivasūtravimarśinī, p. 2 #

He woke and went looking. He found the great rock. And at the mere touch of his hand, Kshemaraja writes, it turned over: karasparshanamatraparivartanatah. The dream was confirmed by sight. On the face of the stone were the sutras, which he calls shivopanishat-samgraha-rupani, a compendium of the secret doctrine of Shiva. He mastered them, taught them to good disciples beginning with Bhatta Kallata, and gathered the same teaching into the Spanda Karikas.

It is a good story. It has a mountain, a stone that moves at a touch, a man alone, and a reason. The question this article is built around is simple, and it has a long answer. How much of it did Vasugupta’s own circle know?

What his own students said

Not all of it.

Kallata was Vasugupta’s disciple, and the Rajatarangini, Kalhana’s chronicle of the kings of Kashmir written around 1148, places him in the reign of Avantivarman, who ruled from 855 to 883. The verse is brief, and it does not call him a Shaiva teacher:

“In the time of Avantivarman, the illustrious Bhatta Kallata and other holy men [siddhas] descended to the earth for the benefit of the people.”

— Rajatarangini V.66, in M. A. Stein’s translation

That is the firmest chronological peg in the story. It is a peg for the student, not the teacher.

Everything said about when Vasugupta lived is reasoning backwards from that line. Buhler concluded in 1877 that he “came forward with his doctrine about the end of the eighth century, or even a little later”. Jaideva Singh put him in the last part of the eighth or the beginning of the ninth. Most current accounts say around 800 to 850. The range is narrow, which makes it look settled. Under it is one verse about somebody else.

Kallata’s own commentary on the Spanda Karikas, the Spandasarvasva, ends with three verses. The second of them tells the origin story in a single sentence:

दृब्धं महादेवगिरौ महेशस्वप्नोपदिष्टाच्छिवसूत्रसिन्धोः । स्पन्दामृतं यद्वसुगुप्तपादैः श्रीकल्लटस्तत्प्रकटीचकार ॥२॥

dribdham mahadevagirau mahesha-svapnopadishtach chhivasutrasindhoh / spandamritam yad vasuguptapadaih shrikallatas tat prakatichakara

“The nectar of Spanda, strung together by the revered Vasugupta out of the ocean of the Shiva Sutras taught in a dream by Mahesha on Mahadeva mountain: that Shri Kallata made manifest.”

Kallaṭa, Spandasarvasva, closing verse 2 #

The mountain is there. The dream is there. The rock is not. There is no stone, no turning, no inscription, no searching. The sutras were taught in a dream, and that is all.

This matters more than it looks. One verse, at the end of one commentary, is the closest thing anyone has to an account from inside Vasugupta’s own circle. Dyczkowski calls Kallata’s the oldest version and probably the most reliable. Whether it is Kallata’s own hand or an accretion of the kind colophons attract is a fair question, and we cannot settle it. Some manuscripts do not carry the verse, which is why Jagadish Chandra Chatterji pointed out that it may not be his. Chatterji edited the Kashmir Series volume of 1916 in which it is printed. The man who put the verse into print is the man who observed it might not be Kallata’s.

What can be said is what is on the page. It stands at the end of Kallata’s commentary. It has the mountain and it has the dream. It does not know about the rock.

Two other commentators tell it differently again, and they do not agree with each other.

Varadaraja, whose Shiva Sutra Varttika was edited by Madhusudan Kaul and printed at Srinagar in 1925, opens with Shiva taking the initiative in a stronger form. In his telling Shiva inscribed the sutras, samalikhya shilatale, on the rock-face on Mahadeva mountain, and instructed Vasugupta in a dream. The rock is back, and now it is Shiva who wrote on it.

Bhaskara, whose Shiva Sutra Varttika was printed in the same series in 1916, drops both the rock and the dream. Jaideva Singh reads a dream back into the verse when he summarises it; the verse does not have one. His third introductory verse reads:

श्रीमन्महादेवगिरौ वसुगुप्तगुरोः पुरा । सिद्धादेशात्प्रादुरासन् शिवसूत्राणि तस्य हि ॥३॥

shriman-mahadevagirau vasuguptaguroh pura / siddhadeshat pradurasan shivasutrani tasya hi

“On the glorious Mahadeva mountain, formerly, to the teacher Vasugupta, by the command of a Siddha, the Shiva Sutras appeared.”

Bhāskara, Śivasūtravārttika, upodghāta 3 #

Siddhadeshat. By a Siddha’s command. No dream, no stone, no voice of Shiva. A human or semi-divine adept gave the order, and the sutras came.

So the mountain is unanimous, and the man’s name, and very little else. Three of the four have the dream. One has the dream and nothing more. One has the dream and a stone that turns at a touch. One has the dream and an inscription. One has an adept’s command and no dream at all.

The obvious move is to sort them by date and watch the story grow. The obvious move is not available.

Kallata is placed, near enough, by Avantivarman’s reign. Kshemaraja is placed, near enough: Abhinavagupta’s pupil, active around 1000 to 1050. Those are the two fixed points, and they are the only two that hold still.

Varadaraja will not. Jaideva Singh puts him at the end of the fifteenth century. His own editor, Madhusudan Kaul, would say only that the date is not positively known, and that he lived before the beginning of the sixteenth. Dyczkowski calls him Kshemaraja’s junior contemporary, which moves him four hundred years the other way.

And Bhaskara, who has the sparest account of the four, cannot be placed at all. Georg Buhler, working through the Kashmiri manuscripts in 1877, put him in the first half of the eleventh century, four generations downstream of Kallata. Jaideva Singh agreed, which makes him later than Kshemaraja. Mark Dyczkowski argued the reverse. Bhaskara cannot be earlier than Utpaladeva, whom he quotes; his spiritual genealogy points to the last quarter of the tenth century; and that would make his commentary the older of the two, an independent line rather than a dissent from an established one.

So the one witness whose position would settle whether the story grew is the one witness nobody can date. What survives is smaller and still worth having. The fullest account is not demonstrably the earliest. And the man who tells it is the same man who fixed how many sutras there are.

The man he turned away from

One name in Kshemaraja’s opening sentence is usually passed over, and it should not be.

Vasugupta, he writes, was anangikritadharadarshanastha-nagabodhyadi-siddhadeshanah: one who did not accept the siddha teaching of Nagabodhi and the others who stand in the lower view. Chatterji, editing the text at Srinagar in 1911, put a two-word footnote under that compound. Nagabodhi bauddhacharyah. Nagabodhi is a Buddhist teacher.

Varadaraja makes the same point with the gloves off. His opening verse describes a world overrun by “siddhas beginning with Nagabodhi, forerunners of the nastikas, deniers of the Self and the Lord.”

Read the origin story again with that in it. A Kashmiri Brahmin on a mountain refuses the instruction of Buddhist adepts. Shiva then gives him a text, and the stated reason is that a rival reading was winning. Kshemaraja says the rival was the dualist exegesis of the Shaiva scriptures. Varadaraja says it was the Buddhists. The two are not the same enemy, and both commentators are writing centuries after the fact, each with his own opponent in view.

What is not in doubt is the shape of the claim. The Shiva Sutras are presented, from the beginning, as a corrective. They are not offered as a new teaching. They are offered as an old one, already written, under a stone, recovered at the moment the tradition was losing an argument.

One detail is worth holding and worth deflating in the same breath. The word Kshemaraja uses for what Vasugupta refused is siddhadeshana, a Siddha’s instruction. The word Bhaskara uses for how the sutras arrived is siddhadeshat, by a Siddha’s command. The same term, doing opposite work. It is tempting to call that a seam. It probably is not one. Siddha, an adept, was ordinary respectful usage in Kashmir for exactly these teachers, and Kalhana’s chronicle applies it to Kallata himself. Kshemaraja is not refusing Siddhas as such. He is refusing a particular set of them, the ones he says stand in the lower view.

Seventy-seven

Ask how many Shiva Sutras there are and you will be told seventy-seven, divided into three unmeshas, “openings,” of twenty-two, ten and forty-five. Nearly every account a reader is likely to meet gives this number. It is on the reference pages, it is in the introductions, and it is correct, in the sense that the edition almost everyone reads has exactly that many.

The edition almost everyone reads is Kshemaraja’s.

His first unmesha runs from चैतन्यमात्मा, chaitanyam atma, “the Self is consciousness,” to महाह्रदानुसंधानान्मन्त्रवीर्यानुभवः, the attainment of the potency of mantra through union with the great lake. Twenty-two. The second, ten. The third, forty-five, closing on bhuyah syat pratimilanam, a last folding back of the world into consciousness. All three Srinagar editions read that sutra the same way. Iyengar’s 1912 translation does not: he prints bhuyah syat pratiniryanam and renders it “there will be relapsation again,” which is a reminder that the text a reader meets in English has its own history of slippage.

Varadaraja’s 1925 edition agrees with him in the first unmesha: twenty-two, the same readings, the same word-divisions, down to the contested sixteenth. Two editions, published from Srinagar fourteen years apart, in agreement. For most purposes that would settle the matter, and for most writers on the subject it has.

There is a third edition.

Seventy-nine

In 1916 Bhaskara’s Shiva Sutra Varttika was published from Srinagar as volumes four and five of the Kashmir Series of Texts and Studies, printed at the Indian Press in Allahabad. The volume opens with a plain list of the sutras, set out in large type, numbered, under the heading अथ शिवसूत्रवार्तिक-सूत्रपाठः, “the sutra-text of the Shiva Sutra Varttika.”

The first division is not called an unmesha. It is called a prakasha, an illumination. And it does not end at twenty-two. It runs to twenty-three, and closes with the editor’s own line: इति प्रथमप्रकाशसूत्राणि, “these are the sutras of the first prakasha.” The third division has forty-six, not forty-five. The total is seventy-nine.

The differences are not a scribe’s drift. Three of them matter here, and each one is a decision.

The first is a reading. Where Kshemaraja and Varadaraja end sutra 1.7 with तुर्याभोगसंभवः, turyabhoga-sambhavah, “the arising of the enjoyment of the fourth state,” Bhaskara ends it तुर्याभोगसंवित्, turyabhoga-samvit, “the awareness of it.” Arising, or awareness. One letter-cluster, and a different claim about what happens in the gap between waking, dream and deep sleep.

The second is a division, and it changes the meaning twice over. Kshemaraja prints one sutra: शुद्धतत्त्वसंधानाद्वाऽपशुशक्तिः. Or, through union with the pure principle, freedom from the power that binds the fettered soul. The pashu, the beast, is the bound self, and pashushakti is what holds it. Varadaraja prints it the same way, as one sutra, numbered sixteen.

Bhaskara cuts it in two. Sixteen becomes शुद्धतत्त्वसंधानाद्वा, “or through union with the pure principle.” Seventeen becomes स्वपदशक्तिः, svapadashaktih, the power of one’s own station. And he glosses svapada as satpada, the true station, which is Shiva. Where Kshemaraja reads a negation of bondage, Bhaskara reads an affirmation of sovereignty.

This is not an artefact of the printed front matter. Deep in the running commentary, on page twenty of the volume, the sutra stands on its own line as a lemma, numbered ॥१७॥, under a running header that reads सू० १७,१८, with Bhaskara’s own verses unfolding beneath it. The split is his.

The third difference is an extra sutra in the third division, at position fifteen, which Kshemaraja’s text does not carry as a sutra at all.

So the count is contested in three places, two of which move the total, and the dissent has been in print since 1916. It is not a discovery. Jaideva Singh notes the split at sixteen in his own commentary and says there is no special point in it; Dyczkowski, who translated Bhaskara, describes the text as some eighty brief statements rather than seventy-seven. What is missing is not the scholarship. It is the headline figure, which goes on being printed as seventy-seven with nothing after it.

The witness who was not a witness

Two editions against one looks decisive until you read how the second one introduces itself.

Varadaraja’s Varttika opens with four verses of homage, and then, at the head of the fifth, a single line that tells you what the book is:

महामाहेश्वरश्रीमत्क्षेमराजमुखोद्भूताम् ।

mahamaheshvara-shrimat-kshemaraja-mukhodbhutam

“arisen from the mouth of the illustrious Kshemaraja, great devotee of Maheshvara.”

Varadarāja, Śivasūtravārttika, 5 #

He is not shy about it. The verse runs straight on into the next, where he says he is making a commentary on the Shiva Sutras वाक्यैरेव तदीरितैः, in the very words spoken by him, set into anushtubh metre. Varadaraja’s book is Kshemaraja’s reading, versified.

Which means the arithmetic changes. There are not two independent editions agreeing against one. There is Kshemaraja, a verse rendering of Kshemaraja that says so on its first page, and Bhaskara. The count stands at one against one, with a cadet branch.

The name he went by is part of the story. Varadaraja was also called Krishnadasa, which is how the 1925 title page gives him and how he signs his own closing verses, where he describes himself as following the path of the exposition Kshemaraja settled. Bhaskararaya, in the eighteenth century, quotes that Varttika more often than he quotes Kshemaraja directly, which is one measure of how thoroughly the versification had absorbed its original.

That is an uncomfortable position for a number everyone repeats without a footnote.

Two lines, and a paper shortage

The editors of the Kashmir Series knew exactly what they had. In 1911, publishing Kshemaraja’s Vimarshini as volume one, they promised that the introduction to the forthcoming Varttika would set the two interpretations side by side and compare them.

In 1916, publishing the Varttika, they explained why it had not happened:

“But, owing to shortness in the stock of the same kind and quality of paper as has been used in the printing of the text of the Varttika and it being thought inadvisable to use paper of a different sort in one and the same volume of a book, the publication of this promised Introduction is held back for the present.”

The comparison of the two traditions of reading the Shiva Sutras, promised in the first volume of the series, was postponed for want of matching paper. The same shortage shaped the rest of the book. The Varttika alone would have made too thin a volume, so the Spanda Karikas with Kallata’s commentary went in behind it, “printed on the same paper.”

That is how Kallata’s closing verse, the one that knows about the dream and not the rock, comes to sit in the same binding as Bhaskara’s siddhadeshat. Not by editorial design. By stationery.

Two of these three editions were not printed where they were published. The 1911 Vimarshini was run off at the Nirnaya-Sagar Press in Bombay, the 1916 Varttika at the Indian Press in Allahabad; only the 1925 volume was printed in Srinagar itself. A department in Kashmir was issuing Sanskrit editions through presses some seven hundred and a thousand miles away.

What caused the shortage the preface does not say, and there is no need to invent a reason for it. The stock that ran short reads like the Allahabad printer’s own. And the preface is signed October 1913, three years before the volume it introduces appeared.

Against that accident, the editors did add one thing deliberately. They printed the anonymous Shiva Sutra Vritti as an appendix, so that readers could make “a comparative study of the two different interpretations of the Shiva Sutras, one given by Kshema Raja and the other as we have it in the Varttika.”

And they said something about that Vritti in passing that is still being argued over. It is, they wrote, “obviously only an extract from the Vimarshini of Kshema Raja.” A digest of the big commentary.

The modern Sanskrit introduction to a later reprint of the same Vimarshini reads the identical evidence the other way. The verbal agreement between the Vritti and the Vimarshini is extensive, its author writes, from which it may be inferred that Kshemaraja built the Vimarshini by taking the older Vritti as his base and expanding it.

Jaideva Singh offered a third answer, which dissolves the question by making it one man’s business: there is a strong presumption, he wrote, that the author of the Vritti was Kshemaraja himself.

One editor says the short text is a boiled-down version of the long one. Another says the long one is a swollen version of the short one. A third says both are his. All are looking at the same overlapping sentences.

Kshemaraja himself supplies one fact that none of them disputes. He opens the Vimarshini by saying why he is writing. Having considered, he says, the incoherence of the commentaries current here, he expounds the Shiva Sutras in accordance with the tradition of the teachers. Iyengar renders the line more sharply, as commentaries “vitiated by ignorance of tradition.” Either way the point stands: whatever the Vritti is, Kshemaraja had predecessors, and he thought they had got it wrong.

What the sutras ask

None of this would matter if the text underneath were thin. It is not, and its architecture is the most interesting thing about it.

The three divisions are usually named for three upayas, means, and read as a descent rather than a course of study. Before that framing is used it needs a caution. It is not the text’s.

It is Kshemaraja’s. His colophons close each division as the illumination of the shambhava, the shakta and the anava means. Bhaskara, dividing the same material into three, titles his sections by something else altogether: the nature of the light of consciousness, the dawn of innate knowledge, the vibration of the divine powers. And Dyczkowski goes further than either of them. There is no reason, he argues, to suppose the aphorisms were divided to match the three means at all. The scheme, on his account, was formalised by Abhinavagupta. Nothing, he says, suggests Vasugupta knew it.

So what follows is a reading, and a particular man’s. It is also the reading every modern reader meets the text through, which is reason enough to set it out.

The first, shambhavopaya, the way of Shambhu, is the way of no method. P. T. Shrinivas Iyengar, who published an abridged English translation of Kshemaraja’s commentary at Allahabad in 1912, described its moment precisely. The Shaktis, he wrote, so urge a man from experience to experience that such an interval cannot be attained; but “by the practice of introspective meditation, it is possible to sense momentary intervals between two consecutive states of consciousness, when the supreme light flashes forth and the man becomes a Bhairava.” The gap is not lying open. It has to be made sensible. What it does not then require is a technique applied inside it.

The second, shaktopaya, works through knowledge, by holding a pure thought-construct until it stops being a construct. “I am Shiva,” held long enough.

The third, anavopaya, the way of the limited self, is where the machinery lives. It carries the breath-work, the postures, the channels, the concentration on the elements. Iyengar sets out its programme in a sentence: meditation accompanied by pranayama, dharana, pratyahara and samadhi, with pranayama leading to the dissolution of prana in the fire of udana in the central channel.

And then he gives the key to the ordering in one sentence: “This gradual progress towards Samadhi is the characteristic of Anavopaya, but Shambhavopaya leads to the same goal without effort.”

The book is built from the top down. It opens with the method that is barely a method and closes with the one that takes a lifetime. A manual would be built the other way. On Kshemaraja’s reading the Shiva Sutras are arranged as a descent through what grace has not already done for you.

Even the top is not quite the top, once the later tradition is in view. Abhinavagupta’s scheme has a fourth term above the three, anupaya, the non-means, where there is nothing to be done at all. It is not in the Shiva Sutras. Jaideva Singh puts the reason plainly: the book is a text on yoga, and it leaves anupaya out because that stage transcends yogic activity and cannot be described. The text has three doors. The tradition that grew from it counted four.

Iyengar drew a blunt conclusion from all that technique. The Shiva Sutra here translated, he wrote in 1912, is “practically but a manual of Yoga.” A century of later scholarship reads it as non-dual metaphysics with a yogic third act. Both readings are looking at the same third division. The disagreement is about what the first twenty-two are for.

Three sutras from the third division show how far the text will push its own logic. The self is an actor, nartaka atma. The inner self is the stage, rango antaratma. The senses are the audience, prekshakanindriyani. The tradition that produced Abhinavagupta’s aesthetics had the theatre in its root text already.

Arguing with Dharmakirti

The origin story begins with a Kashmiri Brahmin refusing a Buddhist teacher. The tradition that grew from it spent the next two centuries in the same argument, and did not always win it by refusing.

Kshemaraja, the man who fixed our text at seventy-seven, wrote a short treatise answering the Buddhist logician Dharmakirti, appended to his commentary on the Svacchanda Tantra. The dispute is about whether initiation can do anything.

Dharmakirti’s objection, in the Pramanasiddhi chapter of his Pramanavarttika, is the kind that is hard to answer because it is so cheap to make. If mantras can stop the seeds of karma from sprouting, then rubbing seeds with oil or burning them should end rebirth too, since those also stop seeds sprouting. And the weighing-initiation, in which the candidate is weighed before and after, is absurd on its face: sin has no form, so it cannot have weight.

Kshemaraja’s answer turns on a distinction. Oil and fire can only obstruct a capacity the seed already has. A mantra, he says, does something else: it adds. And the objection would in any case destroy the neutralising of poison by mantra, which both sides accept. If a mantra can take the death out of venom, the power is conceded already, and what is left to argue about is its reach.

Somadeva Vasudeva, who edited and translated this exchange, notes where Kshemaraja finally rests his case. Mantras have an inconceivable power, and the only real difference between the two sides is that the Buddhists refuse to allow that the power runs as far as liberation. Vasudeva adds that his tone is at times conciliatory, which is an odd thing to be able to say about a polemic.

That is the texture of the whole encounter. Dharmakirti was the Buddhist the Kashmiri Shaivas argued with most, and pieces of his thought were absorbed into the Pratyabhijna philosophy as it took shape. Vasugupta is said to have turned away from Nagabodhi. Kshemaraja, two hundred years later, argued with Dharmakirti in Dharmakirti’s own vocabulary.

Why Kshemaraja won

Put the three questions side by side and the same names keep appearing on the same sides.

On how the sutras arrived, Kshemaraja has the dream and the rock. The other account, a Siddha’s command, is carried by Bhaskara and is reported to have the support of Ramakantha and of the author of the Spandapradipika.

On who composed the Spanda Karikas, Kshemaraja says Vasugupta. Bhaskara and the author of the Spandapradipika are reported to say Kallata.

On how the text divides, Kshemaraja has seventy-seven. Bhaskara has seventy-nine.

The second of those, who wrote the Spanda Karikas, has a detail worth following. Bhaskara’s fourth introductory verse has Kallata transmitting the material चतुःखण्डानि, in four sections, and the four can be accounted for: three nihshyandas in his commentary, and a fourth part named in the verse after, the Tattvarthachintamani. Buhler printed those verses in 1877. He added that he did not quite understand them, and that they appeared to him corrupt.

It is tempting to call this two schools and leave it there. The evidence will not quite allow it. Kallata’s own closing verse, from inside the Spanda tradition, backs Kshemaraja on the dream and on Vasugupta’s authorship. The lines cross.

What is clear is the difference in what each line carried. Counting only the commentaries on the sutras themselves, Kshemaraja’s reading is attached to three: the anonymous Vritti, whatever its direction of dependence; his own Vimarshini; and Varadaraja’s verse rendering, which announces its source in its opening breath. Bhaskara’s is attached to one, and the 1916 editors were blunt about what they had to work from: four manuscripts, of which they said that three were very imperfect. When Georg Buhler went through Kashmir buying manuscripts in the 1870s, the copy of Bhaskara’s Varttika he secured was incomplete.

The usual answer has a name attached to it, and it is worth looking at. Jagadish Chandra Chatterji, introducing the first volume of the series in 1911, wrote that most of the different interpretations of the sutras are now lost, owing no doubt to the influence of the great Abhinavagupta, whose explanation Kshemaraja gives and acknowledges. Then he added the exception, and it is the whole of this article in one clause: the only one left that differs is Bhaskara’s.

Chatterji asserts the causal part rather than arguing it, and it is not obvious how it would be argued. Kshemaraja was Abhinavagupta’s pupil and says so in his own colophon, so the influence is real; whether it is what did the displacing is another matter. What can be seen in the record is simpler. Kshemaraja’s reading is the one the manuscript and printing traditions carried, and Bhaskara’s is the one they barely carried.

No refutation of the losing reading is on record. It was printed in 1916, bound beside its rival, and in 1992 Mark Dyczkowski translated the whole of it into English for a university press. What it never acquired was the thing Kshemaraja’s reading has: a place in the account the tradition gives of itself.

The mountain, and who was still on it

The mountain is still there, and so is the problem with it.

A place in Kashmir is pointed out as where the sutras were found, and in the local speech it carries the name Shankarpal, a worn-down form of Shankaropala, Shankara’s rock. Jaideva Singh locates it precisely: Mahadeva mountain, in the valley of the Harvan stream behind the Shalimar garden near Srinagar. Both of the editions in Kshemaraja’s line took it seriously enough to photograph. The 1911 volume carries a plate captioned Shri Mahadevagiri; the 1925 volume carries one captioned Mahadeva Mountain and another of the boulder itself, Shankara Pal.

So this is better attested than a vague local memory, and it is still not what it would need to be. Dyczkowski, who reports that the faithful venerate the rock, adds that there is now no trace of any inscription on it. Iyengar, translating in 1912, did not know which mountain Mahadevagiri was: he rendered the name and added a parenthesis, “(Kailasha?),” and the question mark is his. A community keeps a site and a name. No inscription, no epigraph, no date. The plates record a devotion, which is a real thing to record, and they do not record a discovery.

What is documented is that the teaching did not leave the valley with the manuscripts. In 1975, and again in 1978, Swami Lakshman Joo taught the Shiva Sutra Vimarshini at his ashram at Ishber, in the foothills of the Zabarwan, above Dal Lake. The sessions were recorded, the first series in English. An English edition drawn from them was published in 2002.

His visitors are, in effect, the modern bibliography. Lilian Silburn of the French national research council first went to India in 1949, spent some five years studying Trika philosophy with him, and returned nearly every year afterwards. Alexis Sanderson and Mark Dyczkowski were among his regular visitors. Paul Reps worked through the Vijnana Bhairava with him in Kashmir and printed the result as the “Centering” section of Zen Flesh, Zen Bones in 1957. Several of the books that now serve as standard references on Kashmir Shaivism were written by people who had sat in that room.

He was born in Srinagar in 1907 and died at Noida in 1991. The transmission he carried was still running in living memory, which is why a reader who wants the Shiva Sutras in English today must either buy a modern translation or go back to 1912. The text is twelve hundred years old and its best modern expositions are recent enough to sit inside copyright. That is an odd thing to have to say about a scripture, and it is the plainest measure of how late this tradition was still being handed over by mouth.

What the stone was for

There is a reading of the origin story that survives everything this article has done to it.

Strip out the rock, which the earliest witness does not have. Strip out the Siddha, which the latest one does not have. Strip out the question of whether a Kashmiri hillside is the right hillside. What remains in every version is a mountain, a man, and a conviction that the thing being recovered was older than the recovery.

That conviction is the text’s own argument about itself. The Shiva Sutras do not claim to be composed. They claim to be found. The first unmesha teaches that the way requiring no method comes first, and the way requiring every method comes last, which is the same claim turned inward: what you are looking for was already there, under something, and the work is the turning over.

Four witnesses could not agree on how the sutras reached the world. Two could not agree on how many there are. A third disagreement, over the last word of the last sutra, turns out on inspection to be a translator’s and not a commentator’s. Two editors could not agree which of two overlapping commentaries came first, and a third dissolved the question by giving both to the same man. The one commentator who might settle whether the story grew cannot be dated well enough to say whether he was dissenting from a tradition or starting one. The arguing has gone on for a thousand years and is not finished.

The stone turned at a touch. Nobody has agreed since about what was written on it.