A sage cannot live in his own hermitage, and the reason is not a rival, a curse or a war.
*"There is a danava… He hides himself inside the earth there, hidden inside the sand… To destroy the worlds, he lies down there and practices terrible austerities. He exhales his breath after an entire year and at those times, the earth, with its mountains, forests and groves, trembles. There are sparks from the coal and red smoke. Because of this, I am incapable of remaining in my hermitage."*1
An annual exhalation. Ground-tremor, sparks, red smoke. Buried in a sand-sea beside a cold desert plain.
The remedy is not a battle. It is excavation: a king and his hundred sons dig a pit in the sand until they find him.
Ninety-seven of the sons burn.
What makes this episode unlike the volume's other demon-killings is that the danger is not an army, a boon or a usurped office. It is a periodic physical event, described by its effects.
⚑ Read the symptom list as it stands: the ground shakes; the mountains, forests and groves shake with it; there are sparks and red smoke; and it recurs on an annual cycle. The text supplies a buried, breathing agent as the cause, and a location — "an ocean that is full of sand" by "the plain of the cold desert", west-facing when uncovered.
⛔ No claim is made here about what the passage is an account of. It is an etiology inside a Purāṇa, offered by a character, and this page reports it as that. What is worth recording is the form: an intermittent natural phenomenon explained by a sleeping being whose respiration is the mechanism, and whose removal "the world will regain its natural state."
⚑ And the framing is domestic rather than cosmic. Utaṅka's stated grievance is that he cannot stay in his hermitage. The world-scale consequence is mentioned; the reason he raises it is that he has been driven out of his house.
The killer becomes Dhundhumāra, Dhundhu-slayer. That is unremarkable in a volume where roughly fifteen of seventeen name-etiologies derive from an injury, a killing, a crime or a catastrophe.2
What is remarkable is when the name arrives.
The chapter first states it as a completed fact, ahead of the whole episode: *"Having killed Dhundhu, that excellent king obtained the name of Dhundhumara."*1 Then it goes back and narrates the events — and the name is announced twice more, in advance, by two further parties:
⚑ Three announcements, all before the killing, one of them from the firmament. The volume's other damage-names are conferred after — Triśaṅku is named for three sins already committed, Vṛṣaghna for a bull already killed, Dhūmāvatī for the smoke of a man already burned.2
⚑⚑ Here the name precedes the act and the act fulfils it. [n=1 OBSERVED] — one instance, and it is the only prospective name-from-damage the build has found in this volume. Which makes the deed a discharge of an identity already conferred rather than the origin of one, and puts the episode in the same family as this corpus's forecasts: announced in advance, by an authority, and carried out.
The cost is stated in one clause and never returned to.
*"His sons dug and discovered Dhundhu hidden inside the sand, covering the western direction. Flames issued from his mouth… The emerging fire burnt down the one hundred sons. O lord among sages! Only three were left."*1
⚑ Three things about that sentence.
The arithmetic is exact and grim. A hundred sons are established earlier — "Kuvalashva had one hundred sons who were excellent archers" — the fire takes them, and three survive. Ninety-seven deaths in a subordinate clause.
They die from the excavation, not the battle. The fire is Dhundhu's reaction to being uncovered; the king's own attack comes afterwards, and he wins it with a fiery arrow that stops the water. The sons are the cost of finding the thing, not of fighting it.
And nothing in the chapter registers it. No grief, no rite, no consequence. The next sentence has the king rushing the demon, and the one after that has him showing Utaṅka the result. ⚑ The episode's emotional register is entirely procedural — a problem located, a cost incurred, an outcome delivered.
⛔ Recorded as what is on the page. The volume is not indifferent to death elsewhere, and no claim is made that this passage represents its attitude generally.
A Śaiva Purāṇa, and the operative energy is not Śiva's.
Utaṅka's offer is explicit: "Vishnu had granted me a great boon earlier in an earlier yuga. Because of that, your own energy will be reinforced by Vishnu's energy." And it is executed: "On Utanka's invocation and for the welfare of the worlds, the lord's, Bhagavan Vishnu's, energy entered his body", and again at the digging — "Narayana's energy pervaded him, and he became extremely energetic and stronger."1
⚑ The brahmana holds a Vaiṣṇava boon, the king is charged with Vaiṣṇava energy, and the danava dies of it. Śiva is not in the chapter.
⛔ Not offered as sectarian evidence. This volume routinely absorbs and reassigns; it takes an Upaniṣadic episode and hands it to two different deities in two Saṃhitās, and it identifies Viṣṇu and Brahmā as names of Śiva at ch 374.3 ⚑ What the passage shows is that the lineage chapters do not appear to have been swept for this — the Vaiṣṇava attribution stands unaltered and unremarked, in a corpus that elsewhere argues at length about who the destroyer-form belongs to.
The frame is worth naming, because it explains the shape of everything above.
Dhundhu is not the chapter's subject. The chapter is a genealogy — the Ikṣvāku line, name begetting name — and the episode is inserted to explain how one king in the list got his name.
⚑ That is the mechanism the whole cluster runs on. A dynastic list needs its names accounted for, and an etiology is the cheapest possible account. Which is why the arithmetic is exact and the grief absent: the passage is not telling a story about a bereaved father. It is discharging a lexical obligation, and once Dhundhumāra is explained the chapter returns to begetting.
⚑ And it is why the episode is one of several in a range this build originally classified as low-yield lineage material and swept with doctrine-tuned instruments, which returned zero. Dhundhu, the seven-boys cluster, Nṛga, Pṛṣadhra, Gālava, Śaśāda and the Śākya entry are all in that range.4
One further thing happens in this chapter, and it is rare enough in the corpus to be worth flagging even though it belongs elsewhere.
Utaṅka does not merely report the demon. He talks the king out of retiring to the forest, and he wins:
*"There is no dharma like that of protecting the subjects.… In this world, it is seen that there is great dharma from protecting the subjects. That cannot be seen in the forest. Therefore, your mind should not turn in that direction."*1
The king is persuaded — and offers his son instead, which is a concession rather than a capitulation.
⚑ In a corpus where accurate testimony reliably fails to move anybody, an argument that changes a decision is a counter-instance and must be recorded as one.5 ⛔ The mechanism here is duty outranking renunciation — which is not the mechanism Volume 2 identified for winning arguments, and is therefore a second exception rather than a confirmation.
You are reading a long list — a register, a chronology, a table of names — and there are anecdotes embedded in it.
Ask what obligation each anecdote discharges. In a genealogy the answer is almost always this name needs explaining. ⚑ Which tells you immediately what the anecdote is optimised for: it must reach the name, and everything else in it — costs, motives, grief — is unpriced. That is why ninety-seven deaths fit in a subordinate clause.
Then notice that unpriced material is the most reliable material in the passage. The name is the thing being argued for and will have been shaped to arrive. The hundred sons, the sand, the annual breathing and the Vaiṣṇava boon are load nobody was carrying, which makes them likelier to be inherited than invented.
And check whether the label precedes the event. ⚑ Here it is announced three times before the killing, once from the sky. A name conferred in advance and then earned is a different object from a name conferred by an act — the first is a prophecy the narrative discharges, the second is an etiology the narrative produces. They look identical in the finished text and they came from opposite directions.
Finally: sweep the boring ranges by reading, not by instrument. A sweep tuned to doctrine returns nothing in a list of kings. Everything on this page was in a chapter an earlier pass had scored zero.
Evidence. Ch 351-6(37) read in full in place: Utaṅka's description of the buried danava and its annual exhalation; the retirement argument; the father's transfer of the commission; the voice from the firmament; the excavation; "The emerging fire burnt down the one hundred sons… Only three were left"; the fiery arrow; and the three Viṣṇu-energy statements.1 The name-etiology census run across the whole body.2
Tension 1 — the prospective naming may be narrative order rather than a claim about names. The chapter states the outcome first and then narrates, which is ordinary Purāṇic practice. ⚑ What is not ordinary is the two in-narrative announcements — the father's and the sky's — both of which occur before the deed and neither of which is a narrator's summary. The finding rests on those two.
Tension 2 — "Śiva is not in the chapter" is an absence claim and is bounded. It is a claim about ch 351-6(37) as read, not about the surrounding range, and it is made only because the chapter positively attributes the operative energy to Viṣṇu three times.
Tension 3 — the geological reading is a description of the symptom list, not an identification. ⛔ The page reports what the passage says and does not claim the passage is about earthquakes, seismicity, or any particular location.
Open question. Utaṅka says Viṣṇu granted him the boon "in an earlier yuga." Does that boon appear anywhere else in the corpus? A sweep for Utaṅka across all three volumes has not been run, and if the boon is unattested elsewhere it belongs with the several other clauses in this volume that reference events the text does not contain.
Against Damage Becomes Institution: Dhundhumāra belongs to that page's family and sits awkwardly in it, which is why it is worth the entry. Every other instance confers the name after the damage. This one confers it three times before. ⚑ Either the mechanism has a prospective variant, or this is a prophecy borrowing the form of an etiology — and the two are indistinguishable in the finished text.
Against The Guru's Cow: same chapter-range, same lineage material, and the same discovery about method. Four of that page's cases and all of this one's material sit inside ch 350–355, a range covered by instrument sweep before it was read, where the sweeps returned nothing.
Against What Makes an Argument Win: Utaṅka's persuasion is a counter-instance and the mechanism is new — duty outranking renunciation, which is not the quotation-the-opponent-already-endorsed mechanism that page carries. ⛔ Recorded as an exception; neither finding is deleted.
Against The Lineage That Swallowed the Śākyas: the two pages are the same operation at two scales. A genealogy explains the names it carries and files the names it acquires, and both behaviours are visible in adjacent chapters of one roll.
Zero handshakes, stated rather than filled. The episode's transferable content — that an anecdote embedded in a register is optimised to reach a name, and that everything unpriced in it is therefore the more reliable evidence — is carried in full by the Implementation Workflow, and it is a philological observation rather than one needing a second domain to license it. The prospective-naming point is genuinely interesting and is one instance; reaching for a cross-domain frame on [n=1 OBSERVED] would produce exactly the decorative connection the vault's rules forbid.
Sharpest implication. A sage cannot live in his own hermitage because something buried in a sand-sea exhales once a year — "the earth, with its mountains, forests and groves, trembles. There are sparks from the coal and red smoke." The remedy is not a battle but an excavation: a king and his hundred sons dig a pit until they uncover him, and ninety-seven of the sons burn in a subordinate clause the chapter never returns to. The kill is then routed entirely through Viṣṇu's energy, three times over, on a boon a brahmana was granted "in an earlier yuga", in a Śaiva Purāṇa where Śiva does not appear in the chapter. And the killer's name — Dhundhumāra — is announced three times before the deed, once by the father handing over the commission and once by a voice from the firmament, in a volume where roughly fifteen of seventeen name-etiologies are conferred after the damage. ⚑ A name conferred in advance and then earned is a prophecy wearing an etiology's clothes, and the two are indistinguishable in the finished text. The whole episode exists because a dynastic list needs one of its names accounted for — which is why the arithmetic is exact and the grief is absent.
Generative questions.