A goddess is produced from another goddess's body, and the volume explains her name twice.
Ch 361-6(47), in the Umā Saṃhitā, while the gods are in the middle of a hymn:
"Since the destroyer of the asura Shumbha directly emerged from the sheaths in the body, she is praised under the name of Koushiki."1
Ch 412-8.1(24), in the Vāyavīya, fifty-one chapters later and in a different Saṃhitā, during an austerity undertaken to change complexion:
"Devi… quickly cast aside her outer skin and became Gouri. The one she created from her own skin got the name of Koushiki."2
Different substance. Different occasion. Different audience. Different Saṃhitā. Same name, same figure, same enemy — she kills Śumbha and Niśumbha in both.
Unflagged.
Set side by side:12
| ch 361-6(47) (Umā) | ch 412-8.1(24) (Vāyavīya) | |
|---|---|---|
| Occasion | the devas are praising Gaurī; she asks "Who is praising me?" | Brahmā requests it, mid-austerity, to solve a boon problem |
| Mechanism | "directly emerged from the sheaths in the body" | "cast aside her outer skin" — and the new being is made "from her own skin" |
| Prior motive | none stated | cosmetic — the austerity's stated aim is to shed a dark complexion |
| First appearance | manifests as Kumārikā from Gaurī's body, and speaks | produced as a Śakti and handed to Brahmā |
| Other names conferred | Ugratārikā, Mahogratārikā, Mātaṅgī | Kālī — "Since she had the complexion of a dark cloud" |
| Task | *"I will accomplish your task. I am independent and do not need any support."* | "Their destruction can only come about through you" — assigned |
⚑ The last row is the substantive one. In the Umā account she declares her own independence and vanishes to do the job. In the Vāyavīya account she is commissioned, handed over, and posted to a mountain.
The obvious harmonisation is that "sheaths" and "outer skin" are the same thing described twice, and it does not survive contact with either passage.
Ch 361's is an emergence. Something comes out of the sheaths — the body is the container and the goddess is what was inside it. Nothing is discarded and Gaurī is not changed by it.
Ch 412's is a shedding. The outer layer is cast aside and the discarded layer is the material the new being is made from. Gaurī is changed by it — that is the whole point of the austerity, and the name Gaurī, the fair one, is conferred by the act.
⚑ One account has her produce a daughter from within. The other has her produce a daughter from what she has removed. The second is the only one in which the mother is altered, and it is the only one in which the product is also Kālī — "Since she had the complexion of a dark cloud."
⛔ Both passages carry a translator's note at the derivation — 1475 at ch 361 and 1864 at ch 412 — and neither note refers to the other account.12 [n=1 OBSERVED] as a doublet; the collation is exact and both passages are quoted in full.
The Vāyavīya version has a stated motive and it is worth quoting because of what it makes the goddess.
Devī's austerity is undertaken for a reason she gives herself: *"There is a dark complexion in my body. I wish to use appropriate means to get rid of it and become Gouri."*2
Brahmā arrives "as if he did not know" the reason, treats the motive as trivial — "For this reason, what was the need to perform these fierce austerities? Your own will is sufficient for that" — and immediately repurposes the operation, stating the real problem outright:
*"I have bestowed boons on two daityas named Shumbha and Nishumbha. They are insolent and are impeding devas. Their destruction can only come about through you."*2
⚑ Three things arrive in that exchange.
The boon-economy admits its own failure, in the granter's mouth, as the reason the whole operation is necessary. He cannot revoke what he gave; he needs a third party to undo it.
Another tester who already knows. "As if he did not know" puts this with the volume's other pre-informed examiners — Śantanu's deliberately false piṇḍa instruction, Śiva arriving "so as to test their powers."3
And the discarded thing becomes the weapon. What she wanted rid of is made into the being who wins the war. ⚑ The Umā version has no equivalent of any of this — no motive, no boon problem, no repurposing, and nothing discarded.
A detail that matters for how this doublet is classified.
The Vāyavīya narrates the Kauśikī production twice — once at ch 412 as it happens, and again at ch 414, where Devī describes it to Śiva after the fact:
"O divinity! Did you not see that I created Koushiki? There has been no such maiden in the world, nor will there ever be. Brahma will tell you about her valour and strength, her residence on Vindhya and her victory. In a battle, she will kill Shumbha and Nishumbha."4
The two Vāyavīya accounts agree. Same creator, same commission, same posting, same enemies.
⚑⚑ So the Saṃhitā that tells the story twice is internally consistent, and the disagreement is entirely across the Saṃhitā boundary. That is the same behaviour the build found in the Kṛṣṇa–Upamanyu material, where the Vāyavīya back-references itself accurately and does not know the Umā Saṃhitā's version exists.5
⚑ Which is a stronger kind of evidence than a bare doublet: not two careless accounts, but two internally coherent traditions running in parallel with working cross-references that never cross.
Both accounts spend their energy on naming, and the two name-sets barely overlap.
Ch 361 confers Kauśikī, Ugratārikā, Mahogratārikā and Mātaṅgī — the last "Given the way she manifested herself" — in one paragraph. It also, in the same chapter, produces Dhūmāvatī, named for the smoke of the enemy she incinerated.1
Ch 412 confers Kauśikī and Kālī, the second from complexion.2
⚑ Only Kauśikī is common to both, and it is derived differently on each side — from what she came out of in one, from what was taken off in the other.
⚑ And this belongs to a habit rather than an accident: roughly seventeen name-etiologies in this volume, fifteen of them from an injury, a killing, a crime or a catastrophe, with only two exceptions.6 Three of the seventeen are in ch 361 alone.
Worth stating, because the text owns a doctrine that would cover this exactly and does not use it.
Ch 400-8.1(12) states, as a general hermeneutic, that accounts disagree because each sage reports a different kalpa — "Based on what happened in one particular kalpa, maharshis speak about their relative power and control and about how they originated from each other."7
That would licence two origins for Kauśikī without strain. It is in the same Saṃhitā as one of the two accounts, twelve chapters earlier.
⛔ It is not applied. Nor is back-reference-and-decline, which the Vāyavīya uses correctly five times — including, in this very chapter, "Since this has been narrated elsewhere, there is no need to describe it in detail" about the Śumbha–Niśumbha battle.2
⚑⚑ So ch 412 declines to re-narrate a battle on the grounds that it has been told elsewhere, and in the same breath supplies an origin incompatible with the one told elsewhere. The machinery is in the chapter, working, pointed at the wrong object.
You have two accounts of the same thing's origin and want to know whether you are looking at carelessness or at two traditions.
Check whether each side is internally consistent. ⚑ This is the test that distinguishes them and it is cheap. Corruption scatters — a careless corpus gives you three half-versions that disagree with each other and with themselves. Two traditions give you two coherent accounts, each stable across its own retellings. Here the Vāyavīya tells it twice and agrees with itself both times, and disagrees with the other Saṃhitā on every point.
Then look at what each version needs the origin for. One account produces her while the gods are singing and has her declare her own independence. The other produces her because a granter cannot revoke a boon and needs an instrument. Same being, two jobs — and the origin in each case is shaped to the job, which is usually the more informative half.
Look for a discard. ⚑ When one version has something removed — a skin, a layer, a component — and the removed thing becomes the product, that version is doing more work than the other and is likelier to be the local elaboration. The plainer emergence-from-within is the cheaper story.
And check whether the corpus's own reconciliation tools were available. Here a general doctrine covering exactly this case sits twelve chapters away in the same section, and a decline-to-repeat clause is used correctly in the same chapter, about a different object. A tool present, working, and pointed elsewhere is a much sharper finding than a tool absent.
Evidence. Both origin passages read in full in place and collated row by row;12 the ch 414 restatement read in place and checked against ch 412;4 endnotes 1475 and 1864 confirmed to attach at the two derivations and to make no cross-reference;12 the kalpa-variance doctrine and the in-chapter decline-to-repeat clause located.72
Tension 1 — a determined harmoniser can sequence them. Nothing forbids reading ch 412's shedding as the event and ch 361's "emerged from the sheaths" as a compressed allusion to it. ⚑ The text does not do this, the occasions differ (a hymn against an austerity), the initiating parties differ (the devas against Brahmā), and ch 361 has her emerge as Kumārikā and speak, which is not a summary of anything in ch 412.
Tension 2 — ch 361 sits in the volume's import zone. The Umā Saṃhitā runs forty-six per cent over the shloka budget the text publishes for itself and carries every import the build has located, including the Devī Māhātmya material two chapters earlier. ⚑ So the Umā account may not be this text's composition at all, which would make the doublet a join rather than an inconsistency. [PLAUSIBLE — needs corroboration]; the collation has not been run.8
Tension 3 — "unflagged" is bounded by the census. The complete endnote layer was censused against ten intervention markers and no note connects the two accounts. A note that merely mentioned both without flagging a problem would not necessarily have been caught.
Open question. Ch 361 gives her Mātaṅgī and ch 363 gives a Daśamahāvidyā list containing Mātaṅgī as a separate entry from Kālī. Are the two lists in this volume consistent about which of these names belong to which goddess? The build has recorded that ch 361 and ch 363 diverge — Dhūmāvatī against Dhūmrā, Tripurasundarī once against Śrīvidyā and Śrīmatī-Tripurasundarī as two — and has not collated the whole set.
Against The Kailāsa Is a Different Book: the same signature in narrative rather than technical material. That page's two Saṃhitās each state a tattva count twice and disagree with each other — "corruption produces scatter; this produces two coherent positions." ⚑ Here the Vāyavīya states an origin twice and disagrees with the Umā. Same evidential structure, different subject matter, and the two together are the build's strongest case that this volume is assembled from sections rather than drafted.
Against One Story, Two Gods: the two doublets are neighbours and behave differently. There, one episode is told twice with the deity swapped; here, one figure is given two origins with the deity constant. ⚑ Both are in the same six-chapter Devī block on the Umā side, which is either a strong argument for that block being imported or a strong argument for it being the most heavily worked material in the volume.
Against Only What a Character Asks: fifty-one chapters and a Saṃhitā boundary, so no listener is present at both. ⚑ And this is the case where the text comes closest to catching itself — ch 412 declines to re-narrate a battle because it has been told elsewhere, in the same chapter where it supplies an incompatible origin. The awareness that there is an "elsewhere" is right there, and it does not extend to the origin.
Against Damage Becomes Institution: ch 412's version is that page's purest instance in this volume. What she wanted rid of becomes the army — a cosmetic complaint, an austerity, a shed skin, and the shed skin kills Śumbha and Niśumbha.
Zero handshakes, stated rather than filled. The page's transferable content — that internal consistency on each side distinguishes two traditions from one careless one — is carried in full by the Implementation Workflow, and is already the load-bearing move on this build's Kailāsa page, where it earns a cross-domain connection on much larger evidence. Reaching for a second domain here would either duplicate that page's handshake or manufacture one on a single doublet, and the vault's rule is that a page with one real connection is finished and a page with a decorative one is a defect.
Sharpest implication. One goddess, one name, two origins. *Ch 361-6(47), while the devas are mid-hymn: she "directly emerged from the sheaths in the body", manifests as Kumārikā, declares "I am independent and do not need any support", and vanishes to do the job. Ch 412-8.1(24), fifty-one chapters later in another Saṃhitā: Devī undertakes an austerity to shed a dark complexion, Brahmā arrives "as if he did not know" and repurposes it — "I have bestowed boons on two daityas named Shumbha and Nishumbha… Their destruction can only come about through you" — and she "cast aside her outer skin", the discarded skin becoming Kauśikī, who is also Kālī.* Emergence from within against production from what was removed; independent against commissioned; no motive against a cosmetic one. Unflagged, with a translator's note at each derivation and neither referring to the other. ⚑ And the Vāyavīya tells its version twice and agrees with itself, so this is not scatter — it is two internally coherent accounts running in parallel across a Saṃhitā boundary. ⚑⚑ Sharpest of all: *ch 412 declines to re-narrate the Śumbha–Niśumbha battle on the grounds that "this has been narrated elsewhere", and in the same chapter supplies an origin incompatible with the one told elsewhere.* The machinery is in the room, working, aimed at the wrong object.
Generative questions.