Does this corpus reach for the Devī Purāṇa at random, or only at the points where its own Śaiva account has run one-sided?
Seven citations across five sources, over twenty-one months, catalogued on the page. Three of them share a shape, and it is not a subtle one.
And the sentence the whole thread turns on, one line before the third: *"Śiva himself will be as a guru to Mā Pārvatī at that moment — but he forgets that Mā Pārvatī is also a guru to Śiva."*
An eighth data point runs the other way and belongs to the pattern: BD 30:43 names the Devī Purāṇa as one of two texts where a teaching is not written, before routing to a regional tradition. So the corpus uses it as a boundary marker in both directions.
This is a Bhairava corpus with a standing doctrine that the Devī is reached through Bhairava — the adhikāra gate, four sources, one of the build's two cross-domain filings. A passage in which the Devī gives Śiva a guru cuts directly across it, and the corpus never notices.
Which is either sloppiness or structure. The thread's working hypothesis is structure: a compensating source is not a rival authority, it is the same system's other half, consulted when the first half is not sufficient — which would explain why the inversion is neither argued for nor retracted. It is not a competing doctrine; it is what gets said when the standing doctrine runs out.
1. Test the pattern against the remaining citations. AY, AG and MD each name the text once and those three have not been read at their Devī Purāṇa ranges. If they also cluster at points of Śaiva insufficiency, the pattern is at [n≥6]. If they are ordinary narrative citations, it is at [n=3] and should be stated at that strength.
2. Run the same test on the Śiva Purāṇa. It is the corpus's most-cited purāṇa and the obvious control. If it is also invoked selectively — at points of some other insufficiency — then selectivity is a property of the corpus's citation practice rather than a fact about the Devī Purāṇa, and this thread collapses into a much duller finding.
3. Resolve the Bhairavī/Pratyaṅgirā divergence at PB 22:45. The build's own two records disagree: the inventory reads the caption "Vahima" as Bhairavī; the page reads it as Pratyaṅgirā from the Siṃha-hara epithet and flags it as its least secure name. Neither is adopted. The identification changes what the Devī Purāṇa is being asked to carry and should not stay open indefinitely.
4. Follow the dropped gloss. KB 24:19 attaches iḍā, piṅgalā and suṣumnā to the mutual-guru claim — "this is the combination of the connection of Śiva-Śakti svarūpa" — and drops it in the same breath. If the two are mutually guru, the channel joining them is not a route from one to the other but the place the mutuality is enacted. The corpus gestures and moves on. That is either an unbuilt page or an over-reading, and one more instance would decide it.
⛔ No absence claim from a name-grep. This build has recorded five cases of a Sanskrit name vanishing from a transcript that plainly carries it; the Devī Purāṇa's caption forms alone run to "Dvi Purana," "dvi purana," "dvipurana." Ranges must be read, not searched.
⛔ Nothing here is a claim about what the Devī Purāṇa contains. The thread is about what this corpus asks it to carry.
[SPECULATIVE] on the compensation hypothesis. The seven citations and the inversion sentence are sourced.