Śiva does not do the disreputable things in this corpus. Somebody else does them, and afterwards the text says he ordered it.
That is the finding, and for a long while in this build it was stated one step too confidently. The correction is small and it changes everything, so here it is up front:
In every one of Volume 2's eight commissioned operations against a third party, Śiva is never quoted issuing the instruction. Not once. Every instance where he is named as the author is a sentence by the narrator, added after the act, in the third person: following the wishes in Maheshvara's mind, urged by the lord, followed the instructions of Shiva, urged by Shiva, following Shiva's command.
⚑ That claim has a boundary and it is worth having up front, because a first draft of this page stated it without one and was wrong. Volume 2 does contain Śiva issuing a quoted commission — at ch 224, to Kṛṣṇa: "Therefore, follow my command. Use jrimbhanastra to cause yawning in me. Use that opportunity to do what you wish."9 The exception is exact and it is the section below.
Meanwhile the volume contains three commissions that are spoken aloud on the page, with the words given. All three belong to somebody else — Pārvatī once, and the devas twice.
So the corpus runs two different things that look alike and are not: characters commissioning, and a narrator attributing. Reading them as one mechanism, which is what a fast read produces, loses the only part of this that is actually strange.
Take the exception first, because it defines the rule.
At ch 224 Śiva has promised Bāṇa protection and cursed Bāṇa's arms, and Kṛṣṇa has arrived to carry out the curse. Kṛṣṇa states the position — "Following your command, I arrived here to slice off Bana's hands… Withdraw from this battle. Otherwise, your curse will fail" — and Śiva agrees in his own voice, on the page, twice:
"You have spoken the truth. I did curse the king of daityas. It is a result of my command that you came here, to sever Bana's arms… How can I possibly witness the slicing of Bana's arms? Therefore, follow my command. Use jrimbhanastra to cause yawning in me. Use that opportunity to do what you wish."9
That is a quoted instruction, unambiguous, with a named agent, a named weapon and a named purpose. Kṛṣṇa fires it and "Deluded by jrimbhanastra, Girisha yawned."
Now notice what is being commissioned. Śiva is the object of the act. He is not ordering an operation against a wife, a city, a devotee or a demon. He is ordering his own incapacitation, in front of the other party, so that a promise he cannot keep and a curse he cannot withdraw can both be honoured.
And it is the one operation in the volume after which nobody is billed. Not Śiva for failing to protect, not Kṛṣṇa for disabling a god, not Bāṇa.
So the pattern is not that this god cannot be quoted commissioning anything. It is narrower and much stranger: when the act falls on somebody else, no instruction is ever shown; when the act falls on him, the instruction is quoted in full.
Three times in Volume 2 an instruction is quoted.
Chapter 192 is the cleanest in the corpus. Pārvatī, to Viṣṇu, with the target and the method both named:
"You should understand that he has himself shown the path. Follow my command and destroy his wife's dharma of faithfulness to her husband. O Rama's lord! There is no other way."1
There is no ambiguity about who authored that, what was ordered, or who is to do it. The goddess names the act, names the agent, and calls it her command.
Chapter 299 is the devas, and it is unmistakably a hiring. Frightened of the boon Rāvaṇa has just received, "Dejected, they invited Narada and asked him" — and their brief is stated: "Devise a means to free devas of their misery." Nārada accepts on the spot: "O devas! Cast aside your sorrow. I will do what is needed."2
Chapter 188 is also the devas — and this one has been miscategorised in this build's own planning notes, which listed it under Śiva's spoken commissions on the strength of "he summoned Narada and sent him."
Read the chapter and that phrase is a narrated act, not a quoted instruction. No words of Śiva appear. What is quoted is the devas' request, delivered after Nārada arrives among them: "O immensely intelligent one! Please make efforts to ensure Jalandhara's destruction." And Nārada's acceptance: "According to my capacity, there is no doubt that I will accomplish your objective."3
Note the scope of that brief. The devas commission an outcome — destroy Jalandhara. Nothing is said about how. The method that follows, advertising Pārvatī's beauty to a demon in order to make him covet another god's wife, is Nārada's own.
Against those three, five instances in which Śiva is named as the author and no instruction exists.
Chapter 170. Nārada tells Kārttikeya, truthfully, that his parents deceived him — and the narrator attaches: "Following the wishes in Maheshvara's mind, you, Narada, addressed Kumara in these words and were silent."4 Wishes in his mind. Not a word spoken.
Chapters 173 and 176. Śiva refuses to kill the Tripurans and refers: "Nevertheless, go and tell Vishnu about this reason." Viṣṇu tries force and fails. Viṣṇu then conceives the corruption in his own first-person deliberation — "Therefore, I will use my own maya to create impediments in their pursuit of dharma" — and only afterwards does the narrator gloss it, twice: "following Shiva's command."5
Chapter 188. Beyond the devas' spoken brief, Śiva's role is two narratorial phrases: he "summoned Narada and sent him", and later, mid-scene, "urged by the lord, he spoke to Jalandhara."3
Chapters 210–211. The second impersonation: "Hari followed the instructions of Shiva and the mother of the universe", and "Also sent by Shambhu, Hari took away Shankhachuda's excellent kavacha."6 Both principals named, neither shown speaking.
Chapter 229. "Devarshi Narada was urged by Shiva. He went to their house and spoke about Shivaa's beauty."7
Five instances, five different formulas, and no instruction anywhere. [n≥3 PATTERN] for the narratorial register in Volume 2.
And a correction to this build's own notes about the last of them. One more correction, and it matters for how strong the pattern is allowed to be.
This build's planning notes record ch 229 as ch 188 "repeated verbatim", on Vidala and Utpala. It is not.
Chapter 188 gives Nārada a long quoted speech: the journey to Kailāsa, the inventory of Jalandhara's treasures item by item, the withheld praise, the naming of Pārvatī, the description of her, and the closing needle — "You are not like him. You are not the lord of every kind of jewel."3
Chapter 229 gives the entire operation nine words: "He went to their house and spoke about Shivaa's beauty."7 No speech is quoted at all.
Same device, radically compressed — which is a real observation about how this text handles a repeated move, and a much weaker claim than verbatim repetition. It should not be counted alongside the volume's actual doublets.
And ch 229 contains one detail that pushes the register point to its limit. At the moment of the killing, Śiva does not speak to Pārvatī either. He signals:
"Through a glance of his eyes, the divinity indicated to Durga… From the signs in her husband's eyes, the one who is half the body of the omniscient one, understood."7
The volume's last commission is transmitted by looking.
The registers meet once, and they contradict each other in one chapter.
At ch 176 Śiva gives his own account of what happened at Tripura, and he names other agents:
"Consequent to the strength of the maya used by Vishnu and the intelligent Narada, the daityas have started to follow adharma."5
He also tries to hand off the killing — "perhaps Vishnu or someone else can kill them" — before being talked into it.
And in the same chapter Brahmā, delivering the opinion that clears him, says the opposite:
"It was because of your command that they were sent to delude them."5
Śiva does not deny it. He does not confirm it. He kills them.
So inside one chapter: the narrator says Śiva commanded it, Brahmā says Śiva commanded it, Viṣṇu's own deliberation is on the page saying he decided it himself, and Śiva says it was Viṣṇu and Nārada.
There is no reading of ch 173–176 in sequence that resolves that, and the resolution a first pass produces — quietly settling it in Śiva's favour, because that makes the better pattern — is exactly what the vault's rules forbid. The disagreement is the finding. A text that has one register would not be able to disagree with itself about this; a text that has two can, and here it does.
Underneath the register split sits a second regularity, and it is the one with the sharpest edge.
In the narratorial cases, the agent chooses the method, and the choosing is shown.
Viṣṇu at ch 173: force first, from his own initiative; failure; then "I will use my own maya", in his own mouth, as his own decision.5 Nārada at ch 188: given an outcome by the devas, he invents the means — the tour of the treasury, the itemised praise, and then the missing jewel.3 Nārada at ch 299: given a general brief, he invents the flattery, the extraction, the doubt about Śiva's reliability, and the test.2
None of those methods is specified by anyone. In each case the record afterwards says the act followed Śiva's command — and in each case what the reader watched was somebody else thinking.
That is a more interesting mechanism than a god who issues secret orders. It is a text in which initiative flows upward in the reporting. The agent decides, and the credit — or the liability — is assigned to the principal by the narration, without anyone lying, because the narrator is not a character and cannot be cross-examined.
You are trying to work out who is responsible for something, and the answer keeps sliding.
Here is the one move that does most of the work, and it costs nothing.
Separate what was said from what was reported.
Take every sentence you have about the decision and sort it into two piles. Pile one: somebody's words, quoted, with a speaker. Follow my command and destroy her dharma of faithfulness. Pile two: a description of intent, written by an observer. He was acting on instructions. It came from the top. Following the wishes in his mind.
Then count the piles.
If pile one is empty, you do not have an order. You have an attribution — and attributions are made by people with their own reasons for making them. In this corpus every attribution to Śiva is pile two, and every quoted instruction belongs to somebody else.
Now the second move, which is harder and more useful. Ask who chose the method.
Responsibility for an outcome and responsibility for a method are different things and they routinely sit with different people. The devas asked Nārada to solve the Jalandhara problem. They did not ask him to advertise another god's wife to a demon. He came up with that. If you only ask who wanted this to happen, you will miss that the worst part of what happened was designed by the person carrying it out — and that person's name is not in anybody's account of who ordered it.
And the third, for when you are the one being described: watch for the sentence that gives you an intention you did not state. He was fine with it. She wanted this. Following the wishes in his mind. It is generous, it is usually well-meant, and it converts you from a person who can be asked into a person who has already answered. The correction is one sentence — I didn't say that, and I'm not saying it now — and it has to be made before the account sets.
The register split is a fact about Volume 2, and it is worth being precise about how far it reaches, because Volume 1 does not behave the same way.
At Volume 1 ch 120, Śiva commissions the slander of Nārada — his own messenger — and the instruction is quoted, in his mouth: *"In every kind of way, use deceitful words and be harsh."*8
That is pile one. Śiva speaking, naming the manner if not every detail.
So the honest statement is narrow and it is more interesting than the broad one:
[n≥3 PATTERN] within that range.[n=1 OBSERVED] — this build has read that chapter's decisive line and not the rest of Volume 1's commission material, so no claim is made about Volume 1's overall behaviour.What that leaves is a real question rather than a resolved pattern: whether the shift from a god who gives orders to a god whose orders are inferred is a compositional difference between the two volumes, a development in how the text wants him seen, or an artefact of which episodes each volume happens to carry. Nothing in the English decides it, and this page does not.
The strongest evidence is the asymmetry itself, and it is checkable: three quoted commissions in Volume 2, none of them Śiva's; five attributions to Śiva, none of them quoted.1234567
Tension 1 — ch 176 has the text disagreeing with itself about authorship, and it is left open.5 Śiva names Viṣṇu and Nārada; Brahmā and the narrator name Śiva; Viṣṇu's own deliberation is on the page.
Tension 2 — the narratorial register cannot be tested. A character can be contradicted; a narrator's gloss cannot. Whether the volume's Śiva-attributions are true is not a question the text makes answerable, and the page does not treat their frequency as evidence of their accuracy.
Tension 3 — the method is the agent's in every narratorial case.352 Whether that makes the agents the real authors or merely the designers is not settled by the text.
Tension 4 — the two commissioners who speak are Pārvatī and the devas. Volume 1's spine was Śiva's deniability; Volume 2's spoken commissions belong to a goddess and a committee.
Open question. If a commission can be transmitted by a glance (ch 229), by wishes in a mind (ch 170) and by a narrator's phrase (ch 173), what would it take for this text to show a commission that did not happen — and is that distinguishable from any of these?
Against The Slander That Śiva Wrote: that page holds the Volume 1 instance, and it is the one that anchors the whole pattern in quoted speech. It should be read as the exception to Volume 2's practice rather than as its template — and this page's claim is narrower than that page's for exactly that reason.
Against Tripura, the Hardest Case: the ch 173–176 material is worked there as an episode and here as a mechanism, and the two must not drift apart. Both hold the disagreement open; neither settles who authored the corruption.
Against The Two Wives Who Were the Weak Point: the Vrindā and Tulasī operations are the register split at its clearest, one of each, twenty-five chapters apart, on identical crimes. Pārvatī speaks the first; the narrator attributes the second.
Against Kumāra, and the Son Who Left and Nine Heads for a Trick: Nārada's two clearest operations bracket the volume, and they differ in exactly the way this page describes — ch 170 attributed to wishes in a mind, ch 299 commissioned out loud by a committee of frightened gods.
Against Argument Never Wins Here: there is a connection worth naming and not overstating. The corpus never lets an argument decide anything, and it never shows the decisive instruction either. What it shows are speeches that fail and outcomes that arrive, with the causal link supplied by narration in both cases.
Faxian: Witness vs Hearsay — that page tracks a discipline: a traveller who marks the difference between what he saw and what he was told, and keeps the two categories apart in his own record. The value of the account is not that Faxian was unusually truthful. It is that his text lets a later reader sort it.
This corpus does not offer that service, and sorting it anyway is what produces this page's finding. Once the commissions are separated into quoted and reported, a pattern appears that is invisible while they sit in one pile — and it is the reverse of the obvious one. The god everyone reads as the great commissioner of this text never commissions anything on the page in Volume 2. His authorship is entirely hearsay, in Faxian's sense: reported, unattributed to any moment of speech, and supplied by a voice that no character can question.
Two things follow, and the second is the useful one.
The first is that the sort is doable on any narrative, and it is cheap. Two piles, one question each.
The second is about what the ratio tells you. Faxian's discipline is usually taught as a way of grading reliability — trust the witnessed parts more. But a source in which the central claim exists only in the hearsay pile is not simply less reliable about that claim. It is telling you what kind of claim it is. Śiva's authorship of the corruption at Tripura is not a badly evidenced fact. It is a position the text is taking, and it is taking it in the one register that cannot be argued with, while simultaneously letting Śiva deny it in the register that can.
Plausible Deniability as Escalation Framework — that page's claim is that deniability is not damage control bolted on afterwards. It is the architecture that permits escalation in the first place: without an alternative narrative available in advance, each step meets resistance; with one, the sequence can continue.
The Purāṇa fits that and then supplies a component the political version usually has to construct at expense. In the state cases, the alternative narrative has to be manufactured — controlled media, a cover story, a plausible accident. It is work, and it can fail.
Here it is free, because the alternative narrative and the official narrative are produced by the same voice. The narrator says Viṣṇu acted on Śiva's command; Śiva says it was Viṣṇu and Nārada; both statements stand in the same chapter, and the text never adjudicates. Nobody had to build the cover. The text's ordinary machinery for telling a story generates it as a by-product of having a narrator at all.
Which suggests a sharpening worth carrying back. Deniability is easiest wherever the account and the actors are narrated by the same authority — a house history, an internal investigation, a founder's memoir, minutes written by a participant. The expensive kind of deniability is what you need when someone else controls the record. The cheap kind is what you get by controlling the telling, and it does not look like a cover-up from the inside, because nobody involved has to say anything untrue.
Sharpest implication. The corpus's spine is not a god who gives secret orders. It is a text in which initiative flows upward in the reporting. Every disreputable operation in Volume 2 is designed by the person carrying it out — Viṣṇu conceiving the corruption after force fails, Nārada inventing the advertisement and the flattery and the test — and every one is glossed afterwards as Śiva's command by a voice with no name and no accountability. Where an instruction is actually spoken, the speaker is Pārvatī or a committee of frightened gods. And at chapter 176 the two registers meet and openly disagree, with Śiva naming his agents while Brahmā names him. The volume does not resolve that, and neither should anything downstream of it.
Generative questions.