Ten Bhairavas. Ten Mahāvidyās. One paired with one.
That is the organising claim of this entire body of teaching — the thing that makes it a system rather than a set of episodes about individual deities. Every form the corpus teaches has a position in the ten and a goddess attached, and the pairing is what the teacher returns to at the start of episode after episode.
It is also, as this build has assembled it position by position, the least stable structure in the corpus.
Not because any single claim is obviously wrong. Because the positions were given across six recordings over sixteen months, and when you put them on one page they do not compose.
Here is every position this build recovered, with its source and date.
| Position | Form | Consort | Source |
|---|---|---|---|
| 4th | Kabandha | Chinnamastā | KB, June 2025 |
| 5th | Dhūmavanta | Dhūmāvatī | DH, June 2025 |
| 6th | Ekavaktra | Bagalāmukhī | EK, July 2025 |
| 9th | Kāla | — | GT, August 2025 |
| unnumbered | Pañcavaktra | Bālā Tripurasundarī | PV, August 2025 |
Four numbered positions out of ten, from four separate episodes.1
⚑ The other six are not given anywhere in the 34 transcripts. That is an absence claim and it is made only for the range read — every episode in the corpus was read for the cluster A build, and no episode enumerates the roster as a whole.
Which is the first finding, and it is structural rather than clerical: the corpus never lists the ten. It refers to the roster constantly, assigns individual forms to positions in it, and does not once set it out.
Now the collision, and it is at the centre.
Kabandha is the fourth of the ten Bhairavas.2
Chinnamastā is the seventh of the ten Mahāvidyās — stated in the same episode, eleven minutes later.3
And Kabandha and Chinnamastā are the pair.
A fourth cannot positionally correspond to a seventh. So whatever "one-to-one" means here, it is not that position n in one list pairs with position n in the other.
⚑ The episode states both facts and never notices. And the consequence runs through everything: if the correspondence is by character rather than by position, then the numbering is doing no work in the pairing at all — it is a separate fact about each form, and the two rosters are matched by what the deities are like rather than by where they sit.
Which is a perfectly reasonable way for a tradition to pair two lists. It is simply not what "the ten-to-ten table" implies, and the corpus's own framing — ten Bhairavas paired one-to-one with ten Mahāvidyās — invites exactly the reading its numbers refuse.
Then a claim from GT that the corpus's own signal-list does not include, and it makes things worse before it makes them clearer.
The Daśa Avatāra of Viṣṇu, the source says, was supported by the Daśa Mahābhairavas.4
So there are three tens in play, not two. Ten Bhairavas, ten Mahāvidyās, ten avatāras — with the Bhairavas pairing to both of the others.
⚑ Recorded because it is unlisted. The PRD's four correspondence architectures do not include it, and this build found it while reading GT for a different row.5
And it clarifies what kind of object the roster is. A list that can pair one-to-one with two different other lists is not encoding a specific relationship. It is encoding the number ten — and ten is a number this material reaches for the way it reaches for five.
(Added on a return pass, from the eighteen seconds GT spends on it immediately afterward. The claim above was recorded without its mechanism; the mechanism changes what the claim is.)
⚑ GT does not leave supported by undefined. It answers the question in the next breath, and the answer is a single word: fury.
How? When he was in his anger — or whenever his avatāra came into existence, his furious nature of it. Maybe like Narasiṃha avatāra, Vāmana avatāra, or even if you see all the avatāras of Viṣṇu, especially even the one which is going to come, the ⚠️[kulkatara] — the furious nature of Viṣṇu is actually Bhairava.19
⚠️ The tenth avatāra's name is broken in the caption and is not reconstructed here. ⚑ The two that survive intact are Narasiṃha and Vāmana — and the pair is well chosen, because one is the tradition's canonical wrath-form and the other is not. The claim is being made to cover the mild cases as well as the obvious one.
Then the generalisation:
Bhairava is not a person. Bhairava is a consciousness. Bhairava is a tattva — or whatever we call it as sāttvika, tāmasika, rājasika guṇa. In this triguṇa there is an above state, which is called ⚠️[twa].20
⚠️ The state named above the triguṇa is unrecoverable from the caption and is left alone.
⚑ This mechanism does not add a fourth ten. It dissolves one — and it is the strongest available answer to this page's arithmetic problem.
Read the section above again with it in hand. The difficulty was that the Bhairava roster pairs cleanly with two unrelated tens, which looked like evidence that the pairing carried no content. On GT's account it isn't a list at all. Bhairava names the furious aspect of a form — a relation, not a member. Anything with ten forms therefore has ten Bhairavas automatically, because each form has a wrath-aspect and that aspect is what the word points at.
⚑ So the daśa-Bhairava roster may be a shadow cast by the daśa-avatāra roster rather than an independent enumeration, and its ten is inherited rather than asserted. That is why it pairs with two tens without strain: a function applied to a list of ten returns ten of something, whichever list you feed it.
⚠️ This is this build's construction, not the corpus's. GT states the fury mechanism and states that Bhairava is a tattva rather than a person; it does not draw the inference about the roster, and nothing in the 34 transcripts draws it.
⚑ Two things about how the corpus holds the claim, both from the following thirty seconds.
It names its authority, and the authority is a person. "This is what actually my guru has told me."21 No text is cited for the fury mechanism anywhere in the passage.
And it flags its own claim as non-standard before anyone else can. "People may not be accepting that kind of information which I'm trying to share."20 ⚑ That is unusual and worth crediting: the corpus knows this reading is not the received one and says so in the same breath as the reading.
Put the corpus's numbers side by side and they do not belong to one scheme.
Kabandha is 4th of ten.6 Dhūmavanta is 5th of ten.7 Ekavaktra is 6th of ten.8 Kāla is 9th of ten.9 Mahādeva is 46th of 108.10 Aghora is 46th of 108.11 Pātāla is 104th of 108.12
Two different enumerations — a ten and a hundred-and-eight — and forms are placed in one or the other, never both. Nothing anywhere says how the ten relates to the 108. Is the ten a subset of the 108, the top ten of them? A different cut entirely? The corpus does not say, and the two are used interchangeably as though the question does not arise.
And inside the 108, two forms hold the same seat.13
⚑ Which is where this page's honest conclusion sits. The numbers in this corpus are not coordinates. They are honorifics. Saying a form is the forty-sixth or the ninth is a way of saying it is established and located within tradition — it functions as a claim to legitimacy, not as an index into a list anyone could enumerate.
That reading explains everything the positional reading cannot. It explains why nobody lists the ten. It explains why two forms can be forty-sixth without embarrassment. It explains why a fourth pairs with a seventh. And it explains why the ten can pair with two different other tens: the number is a form of standing, and standing is transferable in a way an index is not.
Strip the numbering out and what remains is stronger.
Kabandha and Chinnamastā — a headless god and a self-beheaded goddess. The pairing is thematically exact and each page in this build found it doing real work: her instruction is offer your head as I did, and he is the one who has none.14
Dhūmavanta and Dhūmāvatī — smoke and smoke. She arises from him; he is the flame she is the shadow of; and the corpus's central corrective is that she is not a widow because he exists.15
Ekavaktra and Bagalāmukhī — the one-faced stillness and the goddess of arrest. Their union produces Pītāmbara Vidyā, and the yellow runs through both.16
Every one of those is a genuine structural correspondence, and none of them needs a number.
Which is the finding underneath the finding: the corpus's pairings are sound and its enumeration is not. A tradition that had invented the pairings to fill a table would have got the numbers right and the themes thin. This has it the other way round — which is what you would expect if real pairings were later fitted into a numbered frame.
You inherit a system at work. A grading scheme, a set of tiers, a categorisation everyone uses.
Before you build on it, run the test this page ran.
Ask someone to enumerate it. Not describe it — list it, all the way through, without looking anything up.
What happens, remarkably often, is that they can name four or five items, know there are ten, and cannot produce the rest. The scheme is referred to constantly and has never been written out.
Then the second test: find two items that occupy the same position, or one item that appears in two schemes at different ranks. If the system has been in use for a while and nobody has audited it, you will usually find one within an hour.
Neither finding means the system is worthless. What it means is that the numbers are social rather than structural — they encode standing, seniority, or how long something has been taken seriously, and everyone has been using them as though they encoded position.
The practical consequence is narrow and important: do not build anything that depends on the index. Build on the items, which are usually real, and treat the numbering as commentary.
Five episodes across six months, none of which sets out the roster: KB (June 2025), DH (June 2025), EK (July 2025), GT (August 2025), PV (August 2025).17
Every position on this page was assembled by reading those episodes separately and putting the numbers together — an operation the corpus does not perform and, as far as the 34 transcripts show, has never performed.
⚑ This page is therefore partly a construction. The individual claims are the corpus's; the table is this build's, and the collisions it surfaces are visible only because the material was gathered. That is worth stating plainly: the corpus does not contradict itself in any single episode, and it contradicts itself as soon as you write it down.
What's strong. Four numbered positions attested across four independent episodes, each with a consort, each thematically coherent. The pairings are the most durable thing in the corpus.
Tension, preserved and central. Kabandha is fourth; Chinnamastā is seventh; they are the pair. The correspondence cannot be positional.
Tension, second. Two enumerations — ten and 108 — used interchangeably with no stated relation, and two forms sharing the forty-sixth seat.
Tension, third. A third ten-to-ten pairing, with Viṣṇu's avatāras, unlisted among the corpus's own correspondence architectures.
Absence, stated for the range read. No episode among the 34 enumerates the ten. Six of the ten positions are never assigned.
Open question. If the ten is a subset of the 108, which ten — and by what principle? If it is a separate cut, what makes a form eligible for it? The corpus uses both counts fluently and has no account of their relation.
The vault's Sanderson-register Śaiva material treats rosters as datable artefacts. A list of forms belongs to a recension; it can be compared across manuscripts; its variations are evidence about transmission history, and the whole method assumes an enumeration is a thing you can hold still and count.
Come here and the roster cannot be held still, and the reason is not that this tradition is careless. It is that the two registers want different things from a list.
Scholarship wants an enumeration fixed, because fixity is what makes comparison possible. A practitioner tradition wants a form placed — located within an inherited order that confers standing on it — and placement is a claim about legitimacy that does not require anyone to be able to recite the whole sequence.
Which means the corpus is not doing scholarship badly. It is doing something else, using the same vocabulary — and the collisions this page assembles are artefacts of reading one practice with the other's expectations.
That does not dissolve the problem. A tradition using numbers as honorifics while calling them a one-to-one correspondence will mislead anyone who takes it at its word, including its own students. But it locates the failure correctly: not in arithmetic, in genre.
History — the list that gets tidier as it gets later. The vault holds the founder-problem material: traditions retroject coherent origins, and the demand for a clean beginning reliably produces one whether or not the evidence supports it.18
Numbered rosters are that mechanism applied to structure rather than to persons, and the pairing gives a diagnostic the historical literature does not state directly.
Both cases involve a tradition supplying an order that its own materials do not contain. A founder is retrojected because a movement needs an origin; a roster is regularised because a body of practice needs to look like a system. And in both cases the supplied structure is more coherent than the underlying material, which is exactly why it is suspect.
The insight neither produces alone: the coherence of an inherited structure is inversely related to its age, because every generation that transmits it smooths one more inconsistency — so the version that finally reaches you, complete and symmetrical, is the most edited one. Which yields a rule with real diagnostic force: the ragged version is usually older.
Applied here, it cuts a specific way. The corpus's pairings are ragged — a fourth with a seventh, two forms at forty-six, six positions never assigned — and its framing is smooth: ten to ten, one to one. The framing is the late layer. What the tradition is actually carrying is a set of genuine thematic correspondences that somebody, later, told it were a table.
Sharpest implication. The numbers in this corpus are honorifics rather than coordinates — which is why no episode can list the ten, why two forms hold the forty-sixth seat without embarrassment, and why the ten can pair one-to-one with two different other tens.
Generative questions.