The battle is going badly. Vīrabhadra has broken the line, the gaṇas have driven the devas off the field, and only Indra and the guardians of the worlds are still standing.
So they do the sensible thing. They gather, they bow, and they ask the professional:
O guru! O Bṛhaspati! O father! O immensely wise one! O ocean of compassion! Please tell us quickly. How can we possibly be victorious?1
Bṛhaspati is the preceptor of the gods. He is the ritual and strategic authority for exactly this class of problem, and the question put to him is narrow and practical: how do we win this fight.
And before he answers, the text says he does something odd:
Hearing their words, Bṛhaspati made efforts to remember Śambhu.1
The gods' own guru, mid-battle against Śiva's forces, has to reach for Śiva before he can speak.
What comes back is not a battle plan. It is a lecture on epistemology, and it opens by conceding the fight before discussing it:
Everything that Viṣṇu spoke about earlier has come true now. O Indra! I will expand on that. Listen attentively.1
Then the doctrine, and then — only at the very end — the tactical answer, which is you cannot win.
O Indra! As a result of your childishness, always accompanied by the other guardians of the worlds, you came to Dakṣa's sacrifice. What will you do with this prowess? These extremely angry gaṇas have Rudra as their aide… There is no means to counter it in any way. There will be impediments and the sacrifice. This is the truth. I am telling you the truth.1
What will you do with this prowess. The guru's assessment of his own side's strength is that it is beside the point — not insufficient, irrelevant.
That is the frame the doctrine arrives inside, and it matters: this is not a devotional aside dropped into a war scene. It is offered as the reason the war cannot be won.
The doctrine itself is the sharpest anti-ritualist statement in the Satī Khaṇḍa, and it is worth quoting at length because the list is the argument.
There is a lord of sacrifices who bestows the fruits of every act. The lord does not do anything himself. He gets it done through the apparent agent. All the mantras, herbs, abhicāras, ordinary tasks, karma, Vedas, mīmāṃsās and many other sacred texts based on the Vedas are incapable of knowing Īśa. All the ancient ones have said this. Mahesa himself cannot be known through any of these texts associated with the Vedas. He can only be known if one seeks refuge in bhakti. That is the great learning.1
Count what is on that list.
Mantras. Herbs. Abhicāra — ritual sorcery. Ordinary action. Karma — the whole sacrificial economy. The Vedas themselves. And mīmāṃsā, which Debroy glosses as "examination and inquiry. A school of philosophy."2
That last item is the one that gives the passage its edge. Mīmāṃsā is the darśana whose entire business is the correct interpretation of Vedic ritual injunction. The discipline of getting the Veda right is on the list of things that cannot reach him.
And the speaker is Bṛhaspati. Not a renunciate, not a Śaiva partisan — the gods' own ritual preceptor, the figure who exists to administer precisely this apparatus, declaring it insufficient in the middle of a battle his side is losing.
One clause in that passage is doing more work than the rest, and it connects the doctrine to the plot.
The lord does not do anything himself. He gets it done through the apparent agent.1
The apparent agent. Not an agent — the agency is the appearance.
That is the volume's standing doctrine, stated more baldly here than anywhere else in the Khaṇḍa. It is the same claim the created-being formula compresses into six words every time a new being opens its mouth, and the same claim a householder is instructed to say each morning: I do whatever you engage me in.3
But placed here, in this chapter, it does something specific. It is the doctrinal licence for everything that follows.
Vīrabhadra will shortly execute a commission and be asked "what have you done here?" by the being who gave it. Śiva will deny causing the destruction he ordered. Viṣṇu will apologise for a fight he requested.4
Bṛhaspati supplies the principle under which all three of those are coherent — if the lord acts through apparent agents, then the apparent agent did not act and the lord did not either, and the question who did this has no addressee.
The volume never joins the doctrine to its uses. It states it in chapter 88 and relies on it in chapters 89 and 94.
This is the third complete framework for devotion the volume has produced, and they do not agree.
| where | shape | status | |
|---|---|---|---|
| Three means | ch 10–11 | ordered, śravaṇa first | bhakti as the means |
| Nine limbs | ch 75 | unranked, manana absent | "no difference between bhakti and jñāna" |
| Refuge in bhakti | ch 88 | no enumeration at all | bhakti as the only access |
The third is the most radical and the least equipped. It has no count, no ordering, no practice list — just a negative inventory of everything that fails and one word that works.
And it is the only one of the three spoken by an outsider to the tradition it recommends. The three means and the nine limbs are Śaiva teaching delivered inside Śaiva frames. This is the devas' preceptor, under duress, telling his clients that their entire method is the wrong kind of thing.
That placement is worth weighing rather than dismissing as convenient. A sectarian text putting its strongest claim in a rival authority's mouth is a rhetorical move — the vault has already caught this volume doing it once, when Nandi's curse of the brahmanas turned out to be assembled from Gītā 2.42–43.5 Bṛhaspati saying the Vedas cannot reach him is the same manoeuvre at doctrinal scale: the case against Vedic sufficiency is more valuable coming from the Vedic side.
[n=2 SUGGESTIVE] — two instances of the volume borrowing a rival authority's voice for its sharpest claims. The decisive test: a third instance where a non-Śaiva figure delivers a Śaiva doctrinal position the text could have stated itself.
Your side is losing and you ask the expert how to win.
Watch for the answer that reframes the question, because it arrives in a specific order and the order tells you whether to trust it.
A real reframe concedes first. Bṛhaspati opens with "everything that Viṣṇu spoke about earlier has come true now" — he ratifies the forecast he is about to explain, before explaining it. An expert who reframes without first conceding the immediate facts is usually changing the subject rather than deepening it.
Then it names what class of problem this is not. What will you do with this prowess? He is not saying your force is too small. He is saying force is the wrong instrument — and that is a much stronger claim, because it cannot be answered by more force. Notice whether the reframe rules out a quantity or a category. Only the second is worth restructuring around.
And it should cost the speaker something. This is the tell. Bṛhaspati's answer indicts his own discipline — mantras, Vedas, mīmāṃsā, the whole apparatus he is the custodian of. An expert whose reframe conveniently arrives at "you need more of what I sell" has told you about their incentives. One whose reframe undercuts their own office has told you about the problem.
The residue for practice is small and durable. When the honest answer is there is no means to counter it, say that, and say it before the theory. Bṛhaspati does the opposite — doctrine first, then the verdict — and the gods act on the verdict rather than the doctrine. They stop fighting and, four chapters later, go to Kailāsa and prostrate themselves.6 The theory was correct and the sentence that changed their behaviour was the flat one.
There is a bleak comedy in the chapter that should not be smoothed away.
Bṛhaspati delivers a complete doctrine of divine unknowability, and the gods' response is not conversion. It is "filled with thoughts" — and then Vīrabhadra walks over and shoots them.
Thus, the residents of heaven heard Bṛhaspati's words. Along with Vāsava, all the guardians of the worlds were filled with thoughts.1
Immediately after which Vīrabhadra addresses them with the chapter's cruellest joke — "Expecting to receive a gift, all of you arrived here. I will now give you a return gift… I will grant you avadāna and satisfy you now" — and endnote 871 supplies the pun: *"Dāna is a gift… Vīrabhadra promised them avadāna, the word means a heroic act. But it also means to cut into pieces."*7
So the sequence is: the guru explains that the Veda cannot reach Śiva, the gods think about it, and Śiva's agent offers them a dāna that means dismemberment.
The doctrine and the joke are the same argument in two registers. The gods came to a sacrifice expecting a share, and the volume's position is that the share-economy does not touch the being they excluded. Bṛhaspati says it in philosophy; Vīrabhadra says it in a pun; and the pun is the one that lands.
The last thing to record is what this does to a standing check.
The build has been carrying a question since Cluster H-iii: which bhakti framework recurs in the later chapters — the three means of ch 10–11, or the nine limbs of ch 75? The reasoning was that whichever recurs dates the surrounding material.
Neither recurs. A third appears.
That is a more interesting answer than either branch of the original question, and it weakens the instrument the question was built on. Three complete, mutually non-referring accounts of devotion in one volume — ordered-three, unranked-nine, and unenumerated-refuge — is not a text with a doctrine and some variation. It is a text carrying three deposits that were never reconciled.
[n=3 PATTERN] — three complete frameworks, none acknowledging the others, at chapters 10–11, 75 and 88.
And note what the third one lacks that the other two have: a practice. The three means tell you what to do. The nine limbs tell you what to do. Seek refuge in bhakti tells you what kind of thing will work and gives you no instruction at all — which is exactly what you would expect from a passage whose function is to explain a defeat rather than to train a practitioner.
⚑ STANDING CHECK 4 CORRECTED, 2026-08-11 — the nine-limb framework DOES recur, at Volume 2 chapter 241.
Volume 1 closed this check with "neither recurs, a third appears" and stated [n=3 PATTERN] for three frameworks none of which acknowledges the others. The first half of that is now falsified.
At ch 241, describing the fruits of a hymn: "It bestows the nine kinds of devotion. It is divine and develops internal intelligence."N1
And endnote 781 glosses it: "Shravanam, kirtanam, smaranam, sevanam, dasyam, archanam, vandanam, sakhyam and atmarpanam."N1 That is word for word the list Volume 1's endnote 810 supplied for chapter 75 — same nine members, same order, same translator.
Three things follow, and the second is the one worth keeping.
The framework is standing doctrine, not a local coinage. Sixty-five chapters and a whole volume apart, in different saṃhitās, the same nine in the same sequence.
⚑ But the body at ch 241 does not list them. It says "the nine kinds of devotion" and moves on — the enumeration exists only in the translator's note. The text assumes its reader already knows the nine, which is stronger evidence that the framework was current than a second full enumeration would have been. A list repeated is a list copied; a list referred to by its count is a list in circulation.
And what does not recur is the "three means" framework. Neither the ch 10–11 triad nor the ch 88 refuge-formula appears again in Volume 2's range. The corrected finding is therefore one framework of three recurs, and it is the one with a countable practice attached — which is what you would expect if what survives transmission is what a practitioner can hold. Do not cite the "neither recurs" version forward.
Strongest support is the list itself — mantras, herbs, abhicāra, karma, Vedas, mīmāṃsā, "and many other sacred texts based on the Vedas" — declared incapable of knowing Īśa, by the preceptor of the gods, in answer to a tactical question.
Tension one, and it is the page's own. The volume's sharpest anti-Vedic claim is placed in the mouth of the Vedic side's own authority. That is rhetorically valuable and epistemically suspect, and the volume does the same thing with Nandi's curse. [n=2 SUGGESTIVE]
Tension two. "He gets it done through the apparent agent" is stated here as doctrine and relied on twice in the next six chapters to make an unattributable destruction coherent. The text never connects the statement to its uses.
Tension three. This is the volume's third complete framework for devotion and the only one with no practice attached. The standing check asked which of the first two recurs; neither does.
Tension four. Bṛhaspati has to "remember Śambhu" before he can answer. The gods' own guru accesses the doctrine by the method the doctrine recommends, mid-battle, against Śiva's army — and nobody in the chapter remarks on it.
Open questions. Does a fourth devotional framework appear in ch 96–150, or do any of the three recur? And is "the apparent agent" phrasing used anywhere else in Volume 1, which would be mechanically checkable?
The Nine Limbs of Bhakti and The Three Kinds of Bhakti hold the other two frameworks. This page makes them a set of three rather than a pair, and retires the standing check that asked which of the two recurs — the answer is neither.
Nandi's Curse of the Brāhmaṇas is the volume's other borrowed-authority move: a Śaiva gaṇa attacking Vedic ritualists in Kṛṣṇa's words. Set beside this, the technique looks like a habit rather than an accident — but two instances is a candidate, not a finding.
I Did Not Cause the Breaking Up rests on the doctrine stated here. "The lord does not do anything himself" is the principle that makes an unattributable destruction sayable, and Bṛhaspati supplies it six chapters early, to the losing side.
Give Me a Task holds the same agency doctrine compressed into a formula. Bṛhaspati states in a sentence what the formula performs in six words, and neither passage refers to the other.
Dadhīci Walks Out made the same argument in ritual-validity terms — without Śiva this is no longer a sacrifice — and was told never to raise it again. Nine chapters later the gods' own guru says a stronger version and they listen, because by then they are losing.
One handshake. A second was drafted — bhakti as relational rather than textual knowing, against the vault's role-model material — and cut: it restated the source's own claim in other vocabulary rather than producing anything the Purāṇa does not already say. Under the revised standard, one load-bearing connection is a complete section.
Hallucination and Confidence in Language Models (business). The argument there is structural rather than moral: a model trained to predict the next token was never trained to say I don't know. Every example in its training data showed a human supplying an answer — true, false, outdated, but always answer-shaped. So the system produces answer-shaped output at full confidence regardless of whether the answer is reachable from what it has.
Bṛhaspati's list is that failure mode described from inside a ritual system. Mantras, herbs, abhicāra, karma, the Vedas and mīmāṃsā — the discipline of correct interpretation itself — all produce knowledge-shaped output, and his claim is that none of them can reach Īśa. Not that they are wrong. That they are the wrong kind of instrument, and cannot report their own insufficiency.
The insight neither produces alone: a system with no representation of its own limits will generate confident output right up to the boundary and past it, and the most dangerous component is the one whose job is verification. The technical account locates the flaw in training data; this locates it in completeness — mīmāṃsā exists to check whether a ritual claim is correctly derived, and Bṛhaspati puts it on the list of things that cannot know. A verification layer built from the same material as the thing it verifies inherits its blind spot, which is why the apparatus cannot flag its own ceiling from within. That gives a sharper reading than "the Veda is insufficient": the Veda's auditor is insufficient in exactly the same way, and a tradition that could detect this from inside would not have needed a rival's preceptor to say it.
Sharpest implication. Losing the battle, the gods ask their own preceptor the narrow question — "how can we possibly be victorious?" — and Bṛhaspati first has to "remember Śambhu" before he can answer, which nobody in the chapter remarks on. Then he does not answer it. He concedes the forecast ("everything that Viṣṇu spoke about earlier has come true"), reframes the category (*"what will you do with this prowess?"* — not your force is too small but force is the wrong instrument), and delivers the Khaṇḍa's sharpest anti-ritualist claim: mantras, herbs, abhicāra, ordinary action, karma, the Vedas, and mīmāṃsā are "incapable of knowing Īśa… he can only be known if one seeks refuge in bhakti." Mīmāṃsā is the discipline of getting the Veda right, and it is on the list of things that cannot reach him — spoken by the devas' ritual preceptor, the custodian of that entire apparatus, mid-battle. Inside it sits the clause the next six chapters quietly run on: "The lord does not do anything himself. He gets it done through the apparent agent" — the principle under which Vīrabhadra can be asked what have you done here, Śiva can deny a commission he gave, and Viṣṇu can apologise for a fight he requested. And this is the volume's third complete framework for devotion — after the ordered three means of ch 10–11 and the unranked nine limbs of ch 75 — none of which acknowledges the others [n=3 PATTERN]. It is also the only one with no practice attached, which is what a doctrine looks like when its job is to explain a defeat rather than to train anyone.
Generative questions.