What is the Sri Yantra?

The book that is gone

Around the turn of the thirteenth century, in Kashmir, a scholar named Rajanaka Jayaratha sat down to write a commentary on a tantra. He came to the verses describing how the diagram is drawn, and he paused to make a reader’s aside. With the intention of making it easy, he wrote, the drawing is retold elsewhere, in the Shrirasamahodadhi, the Ocean of Essences, chiefly by the use of threads.

Then he quoted the book.

What he quoted is a working recipe. Make an excellent mandala, a ritual diagram, three hands across. Lay the brahma thread through the centre of the pericarp. Set down seven threads running north to south. Take away the fourth, and a middle dividing space appears. From the ends of the first, second and third threads, extend downward and join them on the circle, and six more triangles come out of it, and from those the fourteen-cornered figure and the ten-cornered. Then the eight-cornered, and inside it the single triangle, which is the fifth. Outside everything, two lotuses, one of eight petals and one of sixteen. Then a square, with four portals. Raised lines of white, filled with red powder.

Jayaratha wrote Sanskrit; every rendering of him here is Louise Finn’s, from her 1986 edition, unless another translator is named. A hand is the distance from elbow to fingertip, about eighteen inches. The pericarp is the seed-bed at the centre of the lotus, and the brahma thread is the one laid through it. The band holding the two lotuses is twelve fingers wide. This is not theology. This is a workshop manual.

The Shrirasamahodadhi has not survived. No manuscript of it is known. Jayaratha attributes it to Ishvarashiva, head of the Shrisara monastery, and the teacher-lineage Jayaratha gives would place him in ninth-century Kashmir. Finn does not accept it. Her objection is short and hard to answer: had the eminent Ishvarashiva written on this tantra, Abhinavagupta would have quoted the text. One would expect Abhinavagupta to have at least quoted from the text, she writes, something which he did not do.

The root tantra, the Nityashodashikarnava, also called the Vamakeshvarimata, has its own drawing verses at 1.29 to 1.41, and they are older. What the lost book supplies is something else: the oldest surviving measured recipe, with a thread grid and a unit of length, and it reaches us as a quotation inside a commentary, from a text no living person has read, by an author whose century is disputed. That thread is followed further in the opening chapter of the Sri Vidya treatise.

Which leaves the figure itself. Nine triangles, and a count the manuscripts cannot agree on.

Nine triangles and forty-three

The figure itself is easier to describe than to draw.

Nine large triangles interlock. Four point upward, called vahni, fire. Five point downward, called shakti, power, the feminine principle. Their intersections cut out smaller triangles, arranged in rows of eight, ten, ten and fourteen. Those four numbers come from an enumeration verse credited to the Rudrayamala, on an attribution this article comes back to. The survey that Chiodo follows for the verse and for the construction methods is R. C. Gupta’s, in the Indian Journal of History of Science in 2007, at pages 177 and 180. Eight plus ten plus ten plus fourteen is forty-two. Add the central triangle and the figure contains forty-three. At the exact centre sits a point, the bindu, and whether it belongs to the count is a question the commentators answer differently.

Forty-three is the number T. A. Gopinatha Rao gives at page 330 of his 1914 volume, the number P. H. Pott is reported to give, at second hand, since his book has not been read here, and the number the Tripura Upanishad arrives at, at its third verse, by counting aloud: she was one, then nine, then nineteen, then twenty-nine, then forty-and-three. That is the running total exactly. Two commentators count it differently, and the difference is the whole question. Bhaskararaya reads the opening “one” as the madhya trikona, the central triangle, which becomes nine with the eight of the vasukona, the eight-cornered figure, and then nineteen, twenty-nine and forty-three by the inner ten, the outer ten and the fourteen. On his count the bindu never enters at all. Ramananda, commenting on the same verse in the same printed volume, reads the “one” as the bindu. Forty-three comes out either way. What is in dispute is whether the point is the first term of the series or stands outside it.

The arithmetic is satisfying and the witness is late. The Tripura Upanishad is one of the Shakta Upanishads, the goddess-centred ones, and how late they are is contested. Brooks puts them post ninth or post tenth century and declines to go further. Of the Tripura Upanishad in particular he says the date is not likely to be resolved until more evidence is available, and he objects on principle to the argument from silence, to dating a text late merely because nobody mentions it earlier. Later datings circulate, between the twelfth and fifteenth centuries and after the fifteenth, the second of them a remark in a review of Brooks rather than a study of its own.

Verse 11 of the Saundarya Lahari gives forty-three. Or forty-four. W. Norman Brown’s critical edition, Harvard Oriental Series 43, prints trayash catvarimshat, forty-three, and translates it so. His apparatus lists about a dozen manuscripts carrying catush catvarimshat, forty-four, and they fall on both sides of his own division between the northern and the southern recensions, so the forty-four is a minority reading and not a regional one. The Adyar edition, long the standard printed text in the south, also gives forty-three and sends the forty-four to a footnote. What that footnote says is the reconciliation: those who read forty-four take the angles of her dwelling to include the bindu, so the forty-fourth item is the point and not a triangle. The forty-four circulates in modern writing out of all proportion to its manuscript support, and a number given without a witness is a choice somebody else has made.

Around the triangles: a lotus of eight petals, a lotus of sixteen, a set of three circles, and the bhupura, a triple frame with four square gateways in the four directions. Huet reads that frame as a temple ground-plan, which puts the Sri Yantra in the same family of thought as the Vastu Purusha Mandala.

The nine circuits have names, and the names matter more than the shapes. They are called avaranas. The word is ordinarily translated “enclosures,” and the dictionaries support that: covering, concealing, enclosing. They also give obstruction and interruption, the next sense after enclosing in Monier-Williams and two further on in Apte, and that is the sense Brooks foregrounds when he writes about this diagram, because the ritual requires it. Just as one pierces the six cakras of the yogic body, he writes, so the adept overcomes the obstructions of the mundane world by ascending each of the sub-cakras and acquiring the powers associated with each. The ritual is therefore called navavarana puja, which Brooks renders as the worship which overcomes the nine obstacles. Named goddesses called yoginis occupy the circuits. At Devipuram, Mani Rao gives ninety-eight locations for the goddesses of the Khadgamala, the garland of names that maps them onto the figure, where one practitioner placed his mantra at each in turn, moving up and back down. Published enumerations vary, and Brooks stops to correct one of them, where two of the circuits’ names had been swapped. The nine classes the yoginis belong to run from prakata, explicit, through rahasya, secret, to parapara-rahasya, which Brooks renders as the supreme of the supreme secret. The full liturgical architecture is set out elsewhere.

On Brooks’s reading, each of the nine circuits is a thing in the way, and the worship is the clearing of it.

The text that refuses to draw it

There is a fact about the Sri Yantra’s geometry that is easy to miss.

The Yoginihridaya, the diagram’s own foundational scripture, does not tell you how to draw it.

The two texts are halves of one corpus, and the distinction gets lost because both travel under the same name. The Nityashodashikarnava, also called the Vamakeshvarimata, is the root tantra and carries the drawing verses. The Yoginihridaya is the second half, and it is the one that says what the diagram is.

Andre Padoux, who translated the text, says so flatly. “Rather surprisingly, it does not say how the diagram looks and how to draw it, but describes its apparition, its ‘descent’ (cakravatara), as a divine cosmic process.” The Shricakra in that text “is shown here as a diagrammatic cosmic vision rather than as a ritual diagram.”

What the text gives instead is an ontology. A verse near the opening of its first chapter, 1.8 in the Muktabodha text of the Kaviraja edition, which is the single witness consulted here, is the structural keystone of the whole tradition: panchashakti-caturvahni-samyogac cakrasambhavah. The cakra, the wheel, arises from the conjunction of the five shaktis and the four fires. That rendering is this account’s own. The same proposition turns up inside the lost Shrirasamahodadhi, in a part of the thread recipe not quoted above: the cakra arises from the interlacing of the five shaktis and the four fires. And Jayaratha, glossing the root tantra’s drawing verses, quotes scripture to a related effect: the five shaktis are connected with creation, and the four fires with dissolution. The same formula, then, in the lost book that Jayaratha copied out and in the text that explains the root tantra’s metaphysics.

And the reason the Yoginihridaya does not draw the figure is that, in its own terms, the figure is not a drawing. Bhaskararaya, writing in the eighteenth century, sets out the doctrine: there are three forms of deity, physical (sthula), subtle (sukshma) and supreme (para). The physical is the anthropomorphic goddess of the meditation verses. The subtle is her root mantra. The supreme is the yantra, the diagram itself. On that scheme Lalita in her iconic form is the lowest of the three and the diagram is the highest. Brooks puts it in one line, with Brahman standing for the absolute: if the shrividya is Brahman’s sound then the shricakra is Brahman’s shape.

The thousand names of the Lalita Sahasranama end by walking the same ladder. Name 996 is Shricakraraja-nilaya, she who abides in the royal Sri Chakra, and Bhaskararaya glosses it with a line he quotes from elsewhere, that the Sri Chakra is the body of Shiva and Devi: just as the jiva, the living self, resides in the body, so those two reside in the Sri Chakra. Then Shrimat-Tripurasundari, then Shri Shiva in the feminine, then she whose form is the unity of Shiva and Shakti, then Lalitambika. The hymn’s last five names begin at the diagram’s address and end on the goddess herself.

The popular Western summary of all this is the phrase “yantra, mantra and deity are one.” The real formula in Sri Vidya has five terms, not three. Amritananda, whom Padoux places in the thirteenth or fourteenth century, names them: the Self, the guru, the mantra, the goddess, and the shricakra. The Lalitopakhyana gives an overlapping five at 41.1 and 41.2, where Agastya asks Hayagriva about the goddess, her yantra, her mantra, her teacher and her pupil. It puts the pupil where Amritananda has the Self. Either way the guru is on the list, and the compressed Western version is the one that drops him. That turns out to be the part that cannot be bought.

Two ways to draw it, and each one fails

The classical tradition hands down two construction methods, and the oddity is that they fail on exactly opposite conditions.

Kaivalyashrama’s method, in his Saubhagyavardhani commentary on verse 11 of the Saundarya Lahari, divides the vertical diameter into forty-eight equal units and places the vertices on that grid. It gets the enclosing circle right by construction, because the circle is where you start. It has no mechanism whatsoever for forcing the triangles to meet correctly where they cross.

Lakshmidhara’s method builds outward from the central triangle, row by row, eight then ten then ten then fourteen. Because each new row is laid on the intersections that already exist, every crossing comes out right. And the outer circle will not close. It refuses to pass through the side vertices of two of the triangles.

One method gets the circle and loses the intersections. The other gets the intersections and loses the circle. Gerard Huet’s verdict on the inside-out instructions is that there is “no hope, except by extraordinary luck,” of landing the right points on the circle. C. S. Rao, an associate professor of civil engineering at IIT Bombay, diagnosed why in 1998. The traditional recipes carry redundant measurements, and the redundant ones disagree with each other. The errors at the intersections, he wrote, occur due to the incompatibility of the redundant measurements in the specifications. The construction from traditional literature “is adequate for small objects for worship,” and the error becomes significant for large sizes of the figure. Gupta names both commentators for these two methods. Kulaichev, who had set both out in the same journal in 1984, names neither.

Now notice what those two drawing orders are. Outside-in is samhara krama, dissolution. Inside-out is srishti krama, creation. They are the two directions in which the diagram is ritually traversed. Lakshmidhara, in the sixteenth century, assigned them to the two schools: the Kaulas conceive the figure only by dissolution, the Samayins prefer creation. Bhaskararaya refused to arbitrate and treated both with equal deference. That dispute is a live one, and Lakshmidhara’s own version of the diagram is upside down, with the bindu lifted above the central triangle and the four fire triangles, which in his figure point downward, assigned to Shiva.

So the geometric dilemma and the sectarian dispute are the same dilemma, seen twice. The question of which end you start from is a question about the shape of creation, and it is also the reason the figure will not close.

There is a third direction, usually left out. Sthiti krama, preservation, runs partly from the outside and partly from the centre. Brooks describes it as combining and counterbalancing the other two, and then adds the detail that undoes it: it fails to serve a ritual purpose, and he found that many contemporary practitioners had never heard of it.

Neither method produces the figure the texts specify. Whether an exact one could be drawn with straightedge and compass stayed open into this century.

The question that stayed open until 2021

Roughly twenty conditions have to hold at once. Two of the nine triangles must be inscribed in the same circle. Seven pairs must have an apex sitting exactly on another triangle’s base midpoint. And twelve further conditions each require three lines to pass through a single point, on both sides of the axis, which is twenty-four points.

Alexey Pavlovich Kulaichev, of the biology faculty at Moscow University, measured the problem in 1984 and found that most of the figure’s lines pass through three to six points of intersection of other lines, all of which must coincide. He gave a formula for the nested iterations an accurate hand drawing would need, ad plus bcd plus cd plus d across its four components, and printed no number for it. Put ten steps at each of the four and it comes to 1,210, which is this article’s arithmetic and not his. He was after something harder than either Huet or Chiodo would attempt: a single figure, pinned down by four conditions over and above the ones the texts require. He reduced that to four unknowns and four polynomial equations, and estimated that eliminating them to a single variable might reach degree 12,544, requiring a number representation of four thousand digits. These demands, he wrote, very far exceed the capacities of modern computers.

He was too pessimistic. C. S. Rao solved sets of those constraints numerically in 1998, in the same Indian Journal of History of Science, and in 2014 V.-V. Pustynski showed that Kulaichev’s over-constrained figure has exactly one solution and can be computed to better than a hundredth of a per cent on a pocket calculator, in several days of work.

Kulaichev is also credited with a sentence the internet has repeated ever since, that the origin of the Sri Yantra from the ancient world is still an incomprehensible phenomenon. The published English paper does not contain it. His abstract is milder: a deeper study of this complicated and relatively unknown phenomenon requires the co-operation of specialists from different fields of knowledge. The famous line about the Sri Yantra’s incomprehensibility is itself a thing that got away from its source.

Huet, working in 1990 and publishing in 2002, came at it with a constraint system and Newton approximation, having first measured Zimmer’s frontispiece in a drawing program to get a starting configuration. He found the range of valid solutions so shallow that it was, in his words, “absolutely impossible to optimize” the diagram to the point where the triangle slopes would vary in a monotonous fashion. Push one point too high and another collapses onto a line. The tension between those two pitfalls, he wrote, is the essential difficulty in drawing the right diagram. He also admitted, with some grace, that he never derived the lotus petals mathematically at all. He obtained them by trial-and-error fiddling with Bezier curves.

Huet and Chiodo, working a generation apart and by quite different methods, arrive at the same number. The minimal conditions the figure has to satisfy, as against the measurements the texts supply, leave it four degrees of freedom. Huet stated it as a theorem. The Sri Yantra is an under-determined Euclidean plane geometry problem with four real parameters, admitting an infinity of solutions around the classical form. C. S. Rao reached a compatible conclusion from a third direction, by numerical optimisation: an unlimited number of figures with different proportions of components can be obtained which satisfy the general specifications of Sriyantra from traditional literature.

Which means there is no single correct Sri Yantra in the sources. And Rao was able to say how far off the traditional figure is. The inherited recipe spaces its base lines at 6, 6, 5, 3, 3, 4, 3, 6, 6, 6 units on a circle of radius 24. His computed optimum wants 5.88, 6.22, 5.00, 2.75, 2.74, 4.00, 3.66, 5.84, 5.73, 6.16. Seven of the ten inherited integers are out by a quarter of a unit or less. Of the three that are not, the seventh is the worst, out by two thirds.

From Kulaichev’s paper to Chiodo’s, nobody could say whether an exact figure could be constructed with straightedge and compass, and the sheer difficulty of drawing one was itself treated as the mystery. Kulaichev went further and proposed that the figure was originally spherical and had been flattened by generations of copyists, on the strength of a typology in which only one class of surviving image is geometrically rigid.

In May 2021, Alessandro Chiodo, a mathematician at the Institut de Mathematiques de Jussieu in Paris, published twenty-one pages in Comptes Rendus Mathematique and closed the question.

“To the best of my knowledge,” he wrote, “the natural question of whether the Shri Yantra is constructible by straightedge and compass has remained open until now.”

It is. What he constructs is the entire family of figures satisfying the minimal conditions, rather than one of them. The construction reduces to a classical problem: given a circle, a line and a point, find the circles through the point that touch both. There are exactly four such circles, and the exact drawing of the Sri Yantra is based on one of them. The problem is Apollonius of Perga’s, from the third century before Christ. Apollonius solved it, the solution was lost, Pappus of Alexandria recorded in the fourth century that it had existed, and Francois Viete worked the solutions out again in 1600.

Chiodo’s conclusion takes away one of the sphere’s supports. He puts it carefully. The hypothesis of a three-dimensional origin would have been reinforced had the figure turned out not to be constructible in the plane; it is constructible, so its coordinates lie in the field of constructible numbers. That removes a reason for reaching into three dimensions. It does not touch the evidence Kulaichev actually had, which was a typology in which only one class of surviving image is geometrically rigid.

The result settles something different. Bolton and Macleod in 1977, and Fonseca in 1986, had each proposed a construction built on the golden ratio, and Chiodo’s verdict on both is that they miss the concurrency conditions in a way the eye cannot catch. The field of constructible numbers contains the golden ratio. It also contains an enormous quantity of other things. The geometry does not single out phi.

Where the famous claims came from

Several of the things most commonly said about the Sri Yantra can be traced back to the sentence that started them, and the tracing is the point.

The oldest Sri Yantra is at Shringeri, and it dates to the first millennium. This claim has a paper trail. It goes back to page 70 of Madhu Khanna’s 1979 book, where she writes that one of the earliest surviving specimens of the Sri Yantra is preserved in the Shringeri Matha, established by the philosopher-saint Shankara in the eighth century AD, although its imagery is far older. The hymn from the Atharva Veda, she goes on, contains a description of a yantra-like figure composed, like the Sri Yantra, of nine geometrical permutations.

Read it carefully, because two different claims are doing different jobs. The eighth century belongs to the monastery. The object is “one of the earliest,” which is a ranking and not a date. And the antiquity Khanna does assert is in the second sentence, resting on her note 20, whose route to the Atharvaveda runs through Shamasastry’s rendering of the verse in a book about the origin of an alphabet. That is the next item on this list, and it is the same note.

Kulaichev reproduced her sentence in 1984, close to word for word, with the eighth century still attached to the monastery exactly as she had it. Neither of them dates the object. What went out into the world was the sentence with the ranking turned into a superlative and the monastery’s date transferred to the plate.

What is actually attestable at Shringeri is an engraved plate, photographed as Plate XCVIII of Gopinatha Rao’s Elements of Hindu Iconography in 1914 and captioned, typographical slips and all, as a Shrichakra on a metal plate at the Shringeri Matha. Rao gives it no date.

And the chain may still be running. Chiodo, who settled the constructibility question, reports in 2021 that Kulaichev mentions a Shringeri representation dated to the seventeenth century. This account cannot find that dating in Kulaichev. His Shringeri sentence carries no date for the object, and his only seventeenth-century dating comes eleven pages later, for three metal and stone plates whose location he never states, footnoted to reproductions in Khanna’s book and in Nik Douglas. One can see how the two could be run together. They are still two different claims. And the care is worth taking, because the other two things Chiodo attributes to Kulaichev in that same paragraph, a seventh-century Srivijaya inscription by way of de Casparis and the Atharvaveda hymn, are both on Kulaichev’s page 279 and both reported correctly. Two of the three links hold. It only takes the third. This article carried that error in its own draft, taken from Chiodo, until it was checked against the page. Careful people transmit mistakes. That is how a literature works, and the only defence is to open the page.

The Atharvaveda describes the Sri Yantra. The verse is Atharvaveda 10.2.31, the hymn of the eight-wheeled, nine-doored citadel of the gods: ashtacakra navadvara. Griffith renders it “with circles eight and portals nine.” Whitney has “eight-wheeled, nine-doored.” Monier-Williams gives navadvara as nine-doored and glosses the plural as the nine apertures of the body. Nobody translating the Veda as Veda finds triangles in it.

The reading that puts them there is a late Tantric one. In 1906 R. Shamasastry carried it into English in The Origin of the Devanagari Alphabets, at page 16 of the Varanasi reprint, printing the Sanskrit word in brackets beside his translation so that any reader could see exactly what he had done: the impregnable city of the gods consists of eight circles and nine triangles (dvara). He named his two commentarial authorities in the following paragraph, as Bhaskarananda and Lolla, which are his spellings for two men. The Setubandha he credits to Bhaskarananda is Bhaskararaya’s, which identifies the first; Lolla Lakshmidhara is the standard form of the second. He was not hiding anything.

Khanna quoted Shamasastry’s rendering in her note to page 70. Kulaichev then cited Shamasastry, with Khanna’s verse numbering, Khanna’s reprint date and a fresh error in Shamasastry’s initial, which is how we know he was working from her footnote and not from the book. A Tantric commentator’s reading of a Vedic word, passed along through two notes, is still doing service as evidence that the Sri Yantra is three thousand years old.

The base angles match the Great Pyramid. The pyramid material is in Bolton and Macleod’s 1977 paper, on Huet’s reading of it; the paper itself is paywalled and this account has not read it. The claim was carried into wide circulation by George Gheverghese Joseph’s The Crest of the Peacock, first published in 1991. In every correctly constructed Sri Yantra, he writes, whether it is enclosed in a circle or in a square, the base angle of the largest triangles is about 51 and a half degrees. And the slope of the face to the base of the Great Pyramid, he notes a few lines later, is 51 degrees, 50 minutes and 35 seconds. Huet’s published verdict on the Bolton and Macleod paper was “dubious angular relationships with the Great Pyramid of Cheops.”

On the only angles the sources let one compute, the claim does not come out, and the arithmetic is this article’s own. C. S. Rao prints no base angles at all; the only degrees in his paper label a table of spherical quantities. The two triangles inscribed in the outer circle are the largest in the figure, which is what Joseph’s claim is about. Taking C. S. Rao’s traditional spacings and his computed optimum, and applying Chiodo’s condition that those two triangles share the outer circle, their base angles come out between 52.2 and 53.5 degrees. That is between four tenths of a degree and one and six tenths away from the pyramid, which is not a match at the precision the claim implies. And the two do not agree with each other: one sits near 53.5 degrees, the other near 52.2. There is no single base angle in the figure to compare.

The section survived into the second edition, where it can still be read. Then Joseph withdrew it himself. The preface to the third edition, published by Princeton in 2011, explains that the section on the mathematics inherent in the sriyantra was dropped because of the dubious assumption that it contained hidden geometry. In that edition the word appears exactly once, in the sentence announcing its removal.

The Sri Yantra is the shape of the sound Om. This one can be traced to a single page. It is plate 65 of Khanna’s book, printed on page 116, and the caption reads: “Sri Yantra created in an electronic vibration field, an experiment in the translation of sound into vision.” It ends: “Still from a film by Ronald Nameth.” A film-maker, then, and a film.

The caption does draw an analogy between sound and seeing. It mentions no Om, no tonoscope, no Chladni plate, and it says what it is: a still from a film. Khanna’s book does mention a tonoscope, seventy-two pages earlier, in a passage with nothing to do with this plate, where what the instrument is said to produce is a circle. The plate and the tonoscope meet only in the retelling.

And the physics does not help the claim either. Nodal figures on a vibrating plate are fixed by the plate’s boundary and its symmetry: a circular plate gives concentric circles and straight lines through the centre, a square plate the nodal lines its edges and its symmetry group allow. What no plate gives is nine isosceles triangles in two families, four one way and five the other, meeting at the twenty-odd concurrences this figure requires, with four parameters still free. A vibrating plate does not draw the figure you ask it for. It draws the modes its boundary allows.

Even the enumeration verse, bindu-trikona-vasukona-dashara-yugmam, is worth a provenance note, and it does not come out the way one expects. Gupta calls it a classical verse frequently quoted from the Rudrayamala, and then cites it from the Purashcharyarnava, a digest compiled around 1775, and from the Vachaspatyam, a nineteenth-century lexicon. Brooks prints the same verse in a slightly different reading and attributes it to the frontispiece of a ritual handbook printed in Madras in 1984, making no Rudrayamala claim at all. Neither gives a chapter or a verse. The tantra the verse is credited to may well carry it. Nobody who quotes it appears to have looked.

Against all of which, there is the line from The Tantric Way that both Huet and Chiodo thought worth quoting: the Sri Yantra, in its formal content, is a visual masterpiece of abstraction, and “must have been created through revelation rather than by human ingenuity and craft.”

Chiodo’s paper answers the second half of that. Human ingenuity suffices. It needed Apollonius. Whether revelation also happened is not a question a proof can reach.

How old it actually is

The record is thinner than the claims and more interesting than them.

Brooks holds that the earliest Sanskrit sources devoted specifically to Sri Vidya cannot realistically be dated before the eighth or ninth century. In the front matter of the same book he calls the tradition’s roots traceable to the sixth. Both sentences are his, and the distance between them is a fair measure of how firm the ground is.

Finn, after weighing the evidence, inclines to the eleventh century for the root text, and thinks it was composed outside Kashmir while the Yoginihridaya was probably composed inside it. Anna Golovkova calls the Vamakeshvarimata the earliest extant tantra on the worship of the goddess Tripurasundari, works across the tenth to the fourteenth centuries, and dates the text itself to the early eleventh century at the latest, which is tighter than Finn. Huet, writing as a source critic rather than a dater, observed that he did not know of any published representation of the figure anterior to the seventeenth century, leaving open its date of creation.

Three further things should be set beside those positions.

The tradition’s name is younger than the text that founds it. Golovkova avoids the word Shrividya for the early cult of Tripurasundari on the ground that it is a comparatively recent name for the later, Vedanticised tradition, and notes that the Yoginihridaya does not yet use it for the fifteen-syllable mantra. A fourteenth-century first attestation for the word circulates in the secondary literature. This account could not verify it at source, and says so rather than repeating it.

No Sri Yantra object carrying its own date has come to this account’s notice. Attributed datings do exist in the art-historical literature. Mookerjee and Khanna published a Rajasthan gouache of a classical Sri Yantra in The Tantric Way as about eighteenth century; Khanna published a South Indian copper one in Yantra as about seventeenth; and Kulaichev assigns three metal and stone plates to no later than the seventeenth, without saying where they are. What is missing is a date on the object itself, and an absence is better stated as an absence than quietly filled in.

And the practice spread much later than the text. Elaine Fisher’s work places Sri Vidya initiation spreading “like wildfire, virtually without precedent, through the intellectual circles of Nayaka south India in the early seventeenth century.” Her evidence is elite and Brahmin; she calls Sri Vidya something of an open secret in Tamil Brahmin society. She also holds that the tradition reached the far south long before that, soon after its first zenith in Kashmir, and reads the Tamil verse work the Tirumantiram as showing it, while rejecting its traditional early dating and putting it in the twelfth or thirteenth century. So the seventeenth century is when it became fashionable among the literati, which is a different claim from when it arrived.

The story that Shankaracharya installed the Sri Chakra at the Kamakshi temple in Kanchipuram belongs, on Fisher’s account, to two chronicles of his life that name it, and to a third that gives the same arc without naming it. Anantanandagiri’s Shankaravijaya and Cidvilasa’s Shankaravijayavilasa both place his final ascent at Kanchipuram and both make the establishing of that Sri Chakra his accomplishment there; neither can be dated, beyond having to exist before the mid-eighteenth century. Rajachudamani Dikshita’s Shankarabhyudaya, a refined seventeenth-century life, follows a strikingly similar arc in a wholly different register: its seventh and eighth cantos have Shankara establish the Sarvajnapitha at Kanchipuram and receive from Kamakshi a vision of her true abode, which Rajachudamani sets out across a hundred and eleven verses of the eighth. The authenticity of its closing chapters has been challenged as a later interpolation, and Fisher defends them on stylistic grounds. What the tradition holds and what the record carries are two different things here, and they are worth keeping apart.

The record is thin on when. On the question of who was qualified, it says a great deal more.

Who was allowed near it

Brooks names initiation as perhaps Tantrism’s most distinctive feature, precisely because the established criteria of caste and gender are not the primary tests of qualification. The most distinctive Tantric caste and gender value, he writes, is that non-twice-born persons of both genders may theoretically receive Tantric initiation. Bhaskararaya inverts the usual rule and quotes a text to do it: twice-born persons without initiation into the kula, the lineage, are not qualified. Twice-born means the upper varnas, the ones who take the sacred thread. A Vedic birthright, on Brooks’s gloss, does not necessarily entitle one to learn the Tantric Sri Vidya. What qualifies a person is disposition, and what decides it is the guru.

And then the sentence that immediately follows in Brooks: in practice many conservative Tantric traditions do not admit women, especially women still in childbearing years, as full initiates, due to brahmanical taboos regarding periodic ritual impurity. The restriction is real, and its grounds are brahmanical purity law rather than Tantric doctrine. Those are different things and the difference is the finding.

Brooks also records that Sri Vidya flourishes today in both north and south India among women and men of many castes, and that historically it has always been limited to those who possess ritual literacy in Sanskrit. He states that limit as a fact. The reading offered here, which is this article’s and not his, is that Sanskrit ritual literacy was the filter actually doing the work, and that it tracked caste and gender closely without having to name either. The caste and gender question has its own chapter.

One detail from inside the rite is worth more than the generalisations. Bhaskararaya, glossing name 974, Bindutarpana-samtushta, which Ananthakrishna Sastry renders as pleased by offerings in the bindu, identifies that point with the sarvanandamaya chakra, the circuit made all of bliss, and then specifies what each group offers there: offerings in that chakra by brahmins with milk, by kshatriyas with ghee, by vaishyas with honey and by shudras with mead. The four are the varnas, the caste classes, in their conventional order. Caste is not used there to exclude anyone from the rite. It is used to differentiate what one pours at its centre. And the list takes shudra participation for granted.

Set that against the Vaishnava material. Marion Rastelli, working on the Pancharatra samhitas, the Vaishnava ritual texts, describes an initiation in which how far a person may enter the mandala depends on varna and on sex: a brahmin into the innermost enclosure, a kshatriya into the second, a vaishya into the third, while shudras and women must stay outside it altogether. The Sri Vidya position is a specific and contested stance, and the Pancharatra rule shows what it is a stance among.

Secrecy, meanwhile, is doctrine. The Parashurama Kalpasutra says that the mantras of the Vedas and other teachings are explicit, like a prostitute, and that among all views this one is most secret. Rameshvara’s gloss is sharper: Vedic teaching can be obtained for a price, while this can only be the gift of the guru’s grace, and the knowledge that leads to the realisation of Brahman is not available even for gold.

That is the doctrine. What happens when somebody builds it a building is a modern story.

The diagram you can walk into

Devipuram’s own account has it that in 1983, meditating on a hillock near Anakapalle outside Visakhapatnam, a nuclear physicist felt a heavy object placed in his heart, dug at the site, and found a Meru buried there, the three-dimensional form of the diagram, cast in panchaloha, an alloy of five metals. That is the telling the community holds. What the record carries is the building that followed.

Prahlada Sastry, 1934 to 2015, took the name Amritanandanatha Saraswati, becoming a modern namesake of the medieval Amritananda named earlier, and began building. The Sahasrakshi Meru Temple measures 108 feet by 108 feet and stands 54 feet high, and three levels must be climbed to reach the inner sanctum where Lalita sits among the goddesses of the Khadgamala. Mani Rao, whose 2019 study in Religions is the academic treatment this account relies on, writes that it “is literally a Sriyantra one can walk into.” You enter and leave by different flights of stairs, so that the walk up is itself a circumambulation. C. S. Rao, the engineer, noticed the building from the other direction, and recorded in his 1998 paper that the Meru temple at Devipuram uses the Sriyantra as the plan view for its multi-storeyed structure. The story is told at length in the treatise.

The building’s dates are the institution’s own. Its trust gives 1983 to 1994. Mani Rao’s study does not date the construction, and the figures that circulate for the consecration and the later reconsecration come from the temple and from encyclopedia entries rather than from any independent record. The much-repeated detail that Sastry spent twenty-three years at the Tata Institute of Fundamental Research is in the same position: it is not in Mani Rao, and the sentence carrying it online has no source attached.

Two further things about Devipuram deserve care, because both are widely repeated in forms the sources do not support.

The first is the claim that it offers initiation without regard to caste or gender. The Sri Vidya Trust says, in its own words, that it wants to ensure Sri Vidya is accessible to all without limitations of gender, caste, creed or faith. Mani Rao’s study neither confirms nor denies it. It does not discuss caste at all. What it does stress is guru-transmission: the benefits of a mantra are said to accrue only upon initiation by the guru, and mantras so given are guarded with the utmost secrecy. Devipuram’s own festival document advises beginners not to chant mantras without initiation from a guru, and labels certain practices as requiring diksha, initiation. So the open-access framing is the institution’s, and the academic study of the place is silent on it, which is not the same thing as supporting it.

What Mani Rao does record is something subtler and more useful. The fifteen syllables of the panchadashi mantra, she notes, are no secret, and yet unless the mantra has been given by a guru, its practice is considered ineffective by traditional practitioners. The secret has moved. It is no longer in the text. It is in the transmission.

The second is the matter of authority. Devipuram’s own website announced in 2013, two years before Sastry died, that after his death the goddess’s sandals would be in charge rather than any human being, that there would be no more pithadhipatis, holders of the seat, and that there would be only administrators of the trust. That announcement is the single source for the change, and it reaches most readers at second hand. Taken at face value, it describes the deliberate abolition of an office. A lineage that abolishes its succession does not thereby abolish the problem the succession existed to solve. Where it goes instead, and who carries it, is not something an outsider reading announcements can settle.

Mani Rao records one more thing worth keeping. The placement of the Khadgamala goddesses at the Sahasrakshi Meru Temple, and in the manuals, is based on notes circulated among Sastry’s disciples, and it is ultimately the guru of a community of sadhakas, practitioners, who approves such charts. Which puts the placement where Mani Rao found it: with one teacher’s disciples, rather than in a transregional canon.

The copies of this figure have travelled in a circle.

The circuit

In 1926 Heinrich Zimmer published a book on Indian cult images that, as Gudrun Buhnemann notes, did not target an academic readership, and that influenced Carl Jung in his interpretation of the mandala. From Zimmer through Jung, the Indian diagram became a Western psychological archetype. The reading of the Sri Yantra as a universal map of consciousness descends from that line and from no Indian source.

The pictures in that line are mostly wrong. Zimmer’s 1926 edition carried erroneous descriptions of the figure; the 1984 English translation that corrected them is itself marred by errors in its diagrams, and the one exact diagram in it, the frontispiece Huet measured, was added by the translators and is absent from the original. The “false Sri Yantra” that circulated for a century, including on the cover of the Campbell-edited Myths and Symbols in Indian Art and Civilization, is traced by Huet to the cover of the Prapanchasara Tantra, which Woodroffe published in 1914 as the third volume of his Tantrik Texts, under the name Arthur Avalon. Books on symbolism, Huet observes, usually print the thing incorrectly: Campbell reproducing the false version, Jung its upside-down inverse.

Then, in 1979, Thames and Hudson published Madhu Khanna’s Yantra: The Tantric Symbol of Cosmic Unity. Three years later the same house published Robert Lawlor’s Sacred Geometry: Philosophy and Practice, in its Art and Imagination series: a book built on R. A. Schwaller de Lubicz, whose works Lawlor had translated, written by a man who had been living at Auroville. One imprint, three years apart, two incompatible accounts of what a geometric figure is. An Indologist’s, and a Hermetic geometer’s. That shelf is substantially how the Sri Yantra entered the Western sacred-geometry frame.

Which brings the thing round in a complete circle, and Buhnemann is the one who documents the closing arc. Yantras executed on copperplates, she writes, are copied from drawings in manuscripts, from sketchbooks, from printed books, and “their structures and the mantras inscribed in them often contain errors that go unnoticed due to the ignorance of craftsmen, copyists and practitioners.” She cites Pranavananda, who examined nearly two hundred Sri Chakras from various parts of India and concluded that most of them show major or minor flaws in their designs or other irregularities, and so do not tally with the descriptions in ancient texts. She interviewed a painter in Bhaktapur who had not seen the mandala type she asked about, and who eagerly photocopied her diagrams. And she records the Indigo Gallery in Kathmandu displaying a painted lingatobhadra, one of the geometric diagram types she works on, copied from Khanna’s book.

So the circuit runs: tradition, to Western book, to South Asian craftsman, to object, and back. The authenticity problem therefore has an unexpected address. The manufacture is now partly downstream of Western publishing.

And the tradition’s own objection to a yantra bought over a counter, reconstructed from the sources instead of from indignation, is practical rather than moral. A drawn yantra is inert. It becomes what it is through pranapratishtha, the infusion of life. Gudrun Buhnemann sets out the Tantric form: the deity is brought out of the devotee’s heart by exhaling the breath over a flower, which is then placed on the object that is to be given life. She cites Parashurama Kalpasutra 4.1 with Rameshvara’s commentary, and Mahanirvana Tantra 6.63 and following. Whether the flower goes on the bindu is a detail she does not give. The complaint is about emptiness, and says nothing about commerce. Which is a far quieter claim, and a much harder one to answer.

On 2 July 2026, Sri Sri Vidhushekhara Bharati, the Jagadguru of the Shringeri Sharada Peetham, performed the Sri Chakra Maha Yantra Sthira Pratishtha, the fixed installation of the diagram, in the sanctum of the Jogulamba temple at Alampur in Telangana. The temple is one of the goddess seats on the Ashtadasa list of eighteen Shakti Pithas and the only one of them in Telangana, and the preparatory rites had run through the day before. A diagram out of the tantras, set in the innermost chamber of a Shakti Pitha, in the ordinary course of the religious year.

Most of the figures anyone has examined are drawn wrong. It was never the drawing.

The goddess, the tradition says, is the shape.