TANTUthe weave of the Sri Yantra, measured — ritual, canon and folklore, with receipts

Aakar × Fable · Newark, California · July 12, 2026 · sequel to MARMA

MARMA solved the figure: nine triangles, one closure equation, every coordinate exact in a single quadratic field ℚ(√N), N a 27-digit integer. But a Sri Yantra is not just a figure. It is walked — the navāvaraṇa pūjā visits its nine enclosures in strict order. It is worn — every ring is mapped to a mantra syllable, a station of the body, a named goddess. And it is argued over — which of the infinitely many valid figures is the one?

Each of those systems makes claims. TANTU spent one day asking, of each: is this mathematically forced, or is it a convention — and if convention, exactly how many bits of choice does it carry? Same rules as last time. Exact or it didn't happen. Falsifiers pre-registered before the runs. Refutations published with the same pride as theorems. And this time a second, unrelated model — GPT 5.6, call-sign Sol — re-derived every headline from scratch and attacked the rest; one of our encodings did not survive, and you will read the corrected version below, marked as such.

Finding IThe figure has 64 rooms, not 43

Every book says the nine triangles cut the circle into 43 small triangles. We built the exact arrangement — every intersection point certified in ℚ(√N) — and simply counted.

The exact Classical figure has 64 cells: 47 triangles, 15 quadrilaterals, 2 pentagons. No slivers, no numerical debris — the smallest cell is a third of a percent of the star. The community's Optimal figure, solved to sixty digits, has exactly the same census. The traditional 43 is a painter's idealization: each named "triangle" is a band-shaped region that quietly ignores one or two lines passing through it. The quadrilaterals are real. You can see them in any honest drawing, including the copper-plate yantras of Varanasi — nobody counts them, and now we know exactly which cells they are.

Two more things fell out of the counting. First, the four-parameter family is combinatorially chambered: across 12,246 valid configurations, only about a third share the Classical's 64-cell anatomy; neighboring chambers have 60 or 68 cells, separated by walls where a corner slides across a line while every closure constraint still holds. Second — and we did not expect this — Huet's own published Classical lives in a different chamber than the exact one. His parameters were rounded to three decimals in 1990; the rounding moved the figure across a wall. MARMA's published junction audit (18 crossings, 7 apex-on-base, 6 circle contacts) reproduces verbatim at the rounded point; the exact figure's census is 16 pure crossings + 4 corner-on-line contacts + 7 + 6, and the difference is precisely that wall.

Finding IIThe ritual is a graph algorithm

The navāvaraṇa pūjā walks the enclosures outside-in — bhūpura first, bindu last. The tradition calls this samhāra-krama, the order of absorption. We asked a blunt question: does the ritual's order live in the figure's combinatorics?

Build the dual graph: one node per cell, an edge wherever two cells share a wall, one extra node for the outside. Run breadth-first search from the outside node. The shells come out:

14 · 10 · 10 · 8 · (8 · 8 · 5 · 1)

The first four shells are exactly the populations of the four triangle-āvaraṇas — caturdaśāra 14, bahirdaśāra 10, antardaśāra 10, aṣṭakoṇa 8 — in ritual order. And the peel is canonical: the outside node is the unique maximum-degree node of the graph (degree 14 against an interior maximum of 4), so the samhāra structure is recoverable from pure adjacency, with every coordinate deleted. A millennium of practice walks the figure in the order a graph traversal would.

We had pre-registered the naive version of this claim — BFS from the bindu reproduces (1, 8, 10, 10, 14) — and that version is false: the bindu-out shells are 1, 3, 6, 10, 10, 10, 14, 4, 6. The ritual's own direction is the one that works. Below the aṣṭakoṇa the graph sees finer than the ritual: three more shells (8, 8, 5, 1) that tradition compresses into "trikoṇa and bindu." And one detail we can't resist reporting: the deepest room in the labyrinth is not the room with the bindu. It is the small quadrilateral hiding directly behind it — one step further from the world than the still point itself.

The graph also confesses what it cannot know: it has 8 automorphisms, not 2. Besides the mirror, the up-pointing and down-pointing tip cells at each lateral point of the star are graph-twins with identical neighborhoods. Adjacency knows the ritual's order; it cannot tell up from down.

Finding IIIThe measure ledger

With the arrangement exact, every area is a number of the form a + b√N — not an approximation. The full table (all 64 cells, all nine drawn triangles, every band) ships with the lab. The headlines:

Shakti and Shiva are nearly in balance. The five downward triangles, summed as drawn, carry 1.03320× the area of the four upward ones — the goddess holds a certified 3.3% advantage. But measure coverage — the region lying under at least one triangle of each family — and the ratio drops to 1.01078: less than 1.1%. Both numbers are exact, and both are certified irrational.

And the golden-ratio folklore is over, not because we searched and failed, but because it is impossible. Every ratio in this figure lives in ℚ(√N). The golden ratio lives in ℚ(√5), √2 in ℚ(√2), √3 in ℚ(√3) — and one quadratic field contains another only if the product of their radicands is a perfect square. We checked: N, 2N, 3N and 5N are all non-square.

φ, √2 and √3 cannot equal any ratio of areas or lengths of the Sri Yantra. Not to fifty digits — ever. The refutation is structural.

Finding IVWhy the "Optimal" isn't optimal

The community's canonical figure — the Optimal Sri Yantra, pinned by MARMA to 300 digits — satisfies three elegant closures: extra circle contacts, an equilateral central triangle, centered on the circle's center. We asked whether it also extremizes anything natural, and swept five functionals over all 12,246 valid configurations: cell-area equity, total ink, worst crossing angle, extreme-cell ratio, sanctum share.

It extremizes none of them. We pre-registered that falsifier and it fired. Worse for the canon: on both equity and robustness the plain Classical beats the Optimal (variance 2.87e-5 vs 3.41e-5; worst angle 31.9° vs 19.8°), and the equity optimum lands closer to the Classical than to the Optimal. The Optimal's true character is what its discoverers said it was — contact elegance — and nothing more. The full map of all five landscapes is published in the lab.

Finding VThe π/5 yantra

But one functional found something better than a verdict. Ask for the most robust Sri Yantra — the one that maximizes its worst crossing angle, the figure a shaky hand distorts least — and the answer locks into an exact value.

The binding angles form a three-way standoff: the innermost triangle's slant against the horizontals (θ_b), the crossing of the c-lines at their apex (180° − 2θ_c), and the c-against-b crossing (θ_c − θ_b). Suppose all three exceeded 36°: the first and third force θ_c > 72°, the second forces θ_c < 72°. Contradiction. So 36° is a hard ceiling — and it is attained. A sixty-digit witness on the tie manifold is a valid yantra whose complete junction audit shows every other angle family standing clear (next lowest 37.0°), its minimum equal to π/5 to thirty-eight digits, verified independently three times.

the most robust Sri Yantra has minimum marma angle exactly π/5

Its innermost triangle is a golden gnomon. The optimum is not a point but a two-parameter plateau, and neither the Classical (31.9°) nor the Optimal (19.8°) lies on it. We spent the morning evicting the golden ratio from the yantra's folklore, and the afternoon finding pentagonal geometry waiting at the bottom of a variational question nobody had asked. The figure repays honesty.

Finding VIHow many bits does revelation carry?

The tradition maps the yantra onto the fifteen-syllable mantra, the subtle body, and the nine retinues of goddesses, and calls the whole correspondence revealed. We formalized each mapping from primary sources — Yoginīhṛdaya with Amṛtānanda's commentary, the Ṣaṭcakranirūpaṇa, Kāmakalāvilāsa, Khaḍgamālā, Bhāvanopaniṣad, verse by verse — encoded only the constraints the texts themselves state, and counted the bijections that survive. One survivor means "revealed, not chosen" earns a combinatorial receipt. R survivors mean the choice carries log₂R bits.

Honesty note, with the adversary's fingerprints on it: our first encoding of the top row claimed R = 1 from counts alone. Sol showed the count-axiom we used was shaped by the answer — symmetric counting leaves the la/sa swap genuinely open — and the corrected receipt above leans on the iconography, which is independently attested wherever maṇḍalas are drawn. The row survives; the shortcut did not.

And the tradition is not rigid where books pretend it is: across paddhatis we recorded at least 8 bits of live variance — two incompatible letter-layouts for the central triangle, two coexisting 50-letter partitions used by the same commentator, three names for one yoginī class, disputed populations for the eighth enclosure. One planned test was gated rather than fudged: no consulted source places the fifteen syllables at fifteen geometric points of the drawing, so that test was not run, and the gate is on record.

Finding VIIThree audits, three numbers

"Chanting OM produces the Sri Yantra." The legend traces to tonoscope films of the 1960s–70s. Physics first: the nodal lines of a vibrating disk are concentric circles and diameters — no slanted chord, hence no yantra triangle side, can ever be a nodal line of any disk mode. We computed anyway, generously: the best of ~78 disk modes, rotation-optimized, still misses the drawn lines by 13% of the diameter in Hausdorff distance; the best of 24 eigenmodes of the star-shaped domain itself misses by 17%. The yantra is not a vibration pattern, even of its own shape. Refuted, with numbers attached.

21,600 breaths a day = 360 × 60 arc-minutes. True — because the texts stipulate fifteen breaths a minute on a day of sixty ghaṭikās. Astronomy and physiology were rounded until they met. Design, not discovery.

108. The sun's distance, measured in sun-diameters, sweeps 105.6–109.2 over a year; the moon's, in moon-diameters, 102.6–117.0 over a month. So 108 sits genuinely inside both windows — true on certain days, not on average (the means are 107.4 and 110.6), and soft as a design constant: several integers would pass. Subhash Kak published the core observation two decades ago; the range-audit is what we add. Suśruta counts 107 marmas, not 108. The nakṣatra arithmetic 27 × 4 = 108 is exact, and was always just arithmetic.

Finding VIIIThe rational yantra resists — with structure now

MARMA's open problem: does a Sri Yantra with all-rational coordinates exist? MARMA searched 37,422 anchor pairs to denominator 48 and found nothing. TANTU turned the hunt into number theory. Fix the two free rationals at the Classical's values, parametrize the circle anchors by Pythagorean points, and the existence question over each lower anchor becomes rational points on a curve c² = F(a). At every one of 54 fibers we certified: F has essential degree 18 — a hyperelliptic curve of genus 8. By Faltings' theorem each of those fibers carries at most finitely many rational points. (We went in expecting degree 8 and genus 3; the truth is stronger, and a square discriminant is necessary, not yet sufficient — both qualifications are the adversary's, and they stand.)

The problem stands, but it now has a shape: a genus, a bad set, and a degree-14 core waiting for a Chabauty-style attack.

The LedgerSettled and standing

The WhyWhat it was made for

The Śrīcakra enters the written record around the ninth to eleventh centuries, inside Śrīvidyā, with the Vāmakeśvara corpus — the Nityāṣoḍaśikārṇava and the Yoginīhṛdaya, the same texts our sourcing worked from — as its charter documents. And those texts are not coy about purpose. The figure is a machine for a contemplative practice: a diagram of cosmogenesis — the universe emanating from the bindu through nine enclosures, five triangles of the goddess interlocked with four of the god — built so that a practitioner can run the process in reverse. That is what the navāvaraṇa pūjā is: samhāra-krama, the order of reabsorption, walking creation backward from the world's edge to the still point. The word yantra itself means instrument, from the root yam, to harness. It was made as a harness for attention.

The texts bind figure, mantra, body and goddess into one object so that a single drawing could carry the whole curriculum — the Yoginīhṛdaya says outright that cakra, vidyā, devatā and cosmos are one thing in four notations. What is not recoverable: who first drew it, when exactly, and whether its beautiful over-determination was designed from the start or discovered by generations of refinement and then canonized. The traditional attributions are devotional, not evidenced. On provenance, honesty ends there.

The VerdictDeep sense, not hidden theorems

Does any of this mean the makers possessed hidden mathematics? The disciplined answer our receipts force: yes to a deep sense of mathematics, no to hidden theorems.

The hidden-knowledge reading is refuted item by item. The Classical figure provably contains no golden ratio — absence certified, not merely unfound. The π/5 gnomon lives at a point of the family no tradition ever visited, answering a question none ever posed. Tradition counts 43 where there are exactly 64 — the painted figure idealizes the quadrilaterals away rather than knowing them. And 108 and 21,600 are round-number design. If there were concealed constants in the figure, the day they would have surfaced was this one. They didn't. Whoever now claims the ancients encoded φ in the Śrīcakra argues against a certificate.

What the evidence establishes instead is, we think, more impressive than the myth. Whoever fixed this figure felt an over-determined constraint system without any algebra to name it — the closure lives in a 27-digit quadratic field, and hand practitioners still converged near valid configurations for centuries. Better: the tradition built error-detection into the transmission. The marma rule — the triples must meet exactly — is a checksum; a wrong drawing announces itself. The ritual's ring counts track the figure's true adjacency, which is why the pūjā order fell out of a graph traversal. This is procedural mathematics: exact structure carried through ritual, verse and hand-craft, with a fidelity loss we can now measure — the 43, the eight bits of variance.

And the claim that classical India had deep mathematics never needed this figure as evidence — Pāṇini's grammar, Piṅgala's binary prosody, the śulba constructions, Brahmagupta's quadratics, Mādhava's series for π two centuries before Europe: explicit, published, derived. The yantra tradition is the other half of that genius — not theorem-provers but structure-preservers. Nothing was hidden in the figure. What was hidden — until now — was the equation under it, carried faithfully for a thousand years by people who never needed to see it.

The BoundaryWhat can and cannot be verified

"Is the Sri Yantra's significance real?" is four claims wearing one word, and they do not share a fate.

Claims about the figure — verifiable, and now verified: the geometry is real, hard, exactly solvable, and subtler than the tradition knew. Claims about the system — verifiable, and now verified: the fourfold lattice is not word-salad; its central mapping is combinatorially forced, its ritual encodes a true invariant, its conventions are finite and counted. As a knowledge-architecture, the object demonstrably works — whoever built this built carefully. Claims about effects on practitioners — verifiable in principle, unverified in fact: whether meditating on this figure does something particular to attention is an empirical question, and the existing studies are few and weak. That is the open experimental frontier; someone should do for the practice what this day did for the geometry. Claims about reality — that the yantra is the goddess, that worship liberates, that the cosmos emanates this way — not verifiable in either direction: they admit no falsifier, and this method lives and dies by falsifiers.

So the boundary of verification runs through the object like this: we can verify the loom; we can verify the weave; we cannot verify the goddess. Read closely, the tradition half-agrees — it names its deepest layer rahasya, the secret, and never claimed it demonstrable from outside. One more thing is verified beyond doubt, though it needs no mathematics: a thousand years of unbroken practice, art and argument. Significance to humans is a historical fact. Significance about reality is the thread the method leaves in the reader's hands — where, perhaps, it was always meant to be.

The MeaningWhat was revealed, what was chosen

We set out to measure revelation against convention, and the weave came apart cleanly into both.

More is forced than the skeptic expects. The mantra-to-figure map survives exhaustive enumeration with exactly one candidate. The ritual's walking order is a breadth-first search the figure itself dictates — the pūjā is, among other things, a correct algorithm, executed by hand for a thousand years. And a variational question no tradition ever posed answers in the pentagon's exact angle.

And more is chosen than the devotee expects. The 43 is a painter's idealization of a 64-room truth. The golden ratio is not merely absent but impossible. The canon's "Optimal" optimizes nothing we could name. And where the tradition chose, we can now say exactly how many bits it chose — and point to the verses where it admits as much.

The thread holds. It is human where it is human and mathematical where it is mathematical, and for the first time the seam between the two is drawn on the figure in exact coordinates.

Written in one day, the day after the figure itself gave in — same father in Newark, same model called Fable, this time with a second model across the table checking every claim and winning one argument. For Shivarth, again: first the drawing, now the dance around it. Both have receipts now. ⛵