Something unusual happens in six chapters of this volume, and it happens without anybody appearing to arrange it.
Four men are responsible for a cow that is not theirs. In each case the cow ends up dead or wrongly delivered. And the four differ in exactly one respect: what the man meant.
One misdirects a cow by honest mistake. One kills a bull at night, in a storm, having taken it for a tiger. One slaughters a cow deliberately, in the twelfth year of a famine, to feed a starving family. Seven boys kill one deliberately, cover it with a ritual, and lie about it.
That is a controlled experiment. The object is constant, the relationship is constant, the offence is constant, and intention is the only thing that moves.
Then look at the sentences.
The two accidents draw the harshest and most permanent penalties in the set. The two deliberate acts draw verdicts that are split, reversed, or both.
Nṛga, ch 350-6(36). He has promised a cow to one brahmana and, in the translator's phrase, "mistakenly gave it to another."1 No malice, no gain, no concealment — an administrative error in a donation.
"He became a lizard."1
A species change. He is eventually released, but not by anything he does and not for a very long time: Kṛṣṇa lifts him out of it, generations later.
Pṛṣadhra, in the same chapter. He is guarding his guru's cows at night. A storm; something moves among the cattle; he strikes at what he takes for a tiger and kills a bull.
And here the text does something remarkable: it states the mitigation explicitly, in the guru's own perception, and then overrides it in the same sentence.
*"his guru… realizing that he had committed the sin inadvertently and not voluntarily… 'You are not a kshatriya. Become a shudra.'"*1
⚑ Read that twice. The finder of fact establishes the absence of intent as a finding — inadvertently and not voluntarily — and then hands down a permanent demotion out of the man's varṇa, which in this text's own scheme is a punishment ordinarily requiring lifetimes of service to reverse.2
The mitigation is not ignored. It is noticed, articulated, and disregarded.
Ch 352-6(38). A twelve-year drought. Satyavrata is already disinherited and already carrying a grievance the text presents as reasonable: *"The brahmana has caused harm to the lineage. The illustrious Vasishtha did not prevent my father from abandoning me. The sage did not say that I, the son, should be instated."*3
He finds Viśvāmitra's family starving while the sage is away, and he slaughters Vasiṣṭha's wish-fulfilling cow to feed them. Deliberate. Premeditated. Not concealed.
Vasiṣṭha curses him, and the curse is a tally. The name is the count of the offences:
*"'Failing to satisfy your father, slaying a cow that belongs to your guru and intercourse with someone you should not have intercourse with — these are the three sins you have committed.' Because of the three sins, he addressed him as Trishanku."*3
tri-śaṅku — three sins. The name is the sentence.
And then the other brahmana comes home.
Viśvāmitra, whose family ate, grants him a boon, performs his sacrifice, instates him as king, and sends him to heaven in his body — "While devas and Vasishtha looked on."3
⚑ The clause is the whole point and it is the text's own. The two verdicts are not delivered in different chapters by different authorities who never meet. They are delivered in the same episode, and the one who cursed him is a named spectator at the reward. Nobody remarks on it.
Ch 355-6(41), and this is the one where a single narrator delivers both verdicts.
Seven brahmana boys are set to mind their guru's cow. They kill and eat it, and the justification is an explicit ritual loophole, stated by the most pious of them:
*"There are things one should not do. However, offered to the ancestors, it can be done. Therefore, all of you should control yourselves and perform a shraddha ceremony. There is no doubt that the cow can then be received, following dharma. If we follow dharma and worship the ancestors, we will not suffer from adharma."*4
They sprinkle the cow, offer it, eat it — and lie to the guru, telling him "the cow had been killed by a tiger."
Verdict A, from Sanatkumāra: "Because of their cruelty, violence, ignoble behavior towards a guru and excessive violence, the seven brothers were born as the… sons of a hunter."4
Verdict B, later in the same chapter, from the same narrator: *"You will obtain this because of the favours of the ancestors and because of the good deed you did. You did follow dharma, sprinkling the cow with water and offering it to the ancestors."*4
The loophole failed and the loophole worked. Unflagged — notes 1428–1440 in that range cover the names, the species, the frames and the liberties the translator took, and none addresses the split verdict.
Set the four side by side with intention as the axis:5
| Case | Act | Intention | Verdict |
|---|---|---|---|
| Nṛga | cow misdirected in a donation | honest error | species change — becomes a lizard |
| Pṛṣadhra | bull killed, mistaken for a tiger | explicitly found inadvertent | permanent varṇa demotion, mitigation stated and overridden |
| Triśaṅku | cow slaughtered in a famine | deliberate, under duress | cursed and renamed — then given a boon, a sacrifice, a kingdom and bodily ascent |
| Seven boys | cow killed, laundered, lied about | deliberate and premeditated | punished as cruelty and credited as a good deed, one narrator |
The two accidental cases receive the harshest and most permanent penalties. The two deliberate cases receive split or reversed verdicts.
⚑ Four controlled cases, one object, one variable, and the answer inverts. That is a stronger result than any of the four episodes alone, and it is only visible because the volume happens to put them within six chapters of each other.
⛔ What this is not. It is not a doctrine — nothing in the text says intention aggravates or that accident is worse. Nobody in the volume compares any two of these cases. The pattern is a property of the four verdicts read together, which is a thing no character is in a position to do. [n=4 OBSERVED] on the ordering; [n≥3 PATTERN] on the underlying phenomenon, which is that outcome and intention are uncoupled in both directions across this volume.6
⚑ And the honest reading of why is available in the text's own materials rather than in any theory of desert. Three of the four verdicts are delivered by an offended party — a guru, a brahmana whose cow it was — and the fourth, Viśvāmitra's, by a beneficiary. What varies is not the intention. It is who is doing the sentencing and what happened to them. Nṛga's brahmana lost a cow; Pṛṣadhra's guru lost a bull; Vasiṣṭha lost a wish-fulfilling cow; Viśvāmitra's family ate. On that reading the four are perfectly consistent and the consistency is the wrong one.
This complicates a Volume 2 page and should be recorded precisely, because the direction matters.
the-tirtha-launders-anything established that the discharge instrument "reads acts and never intentions" — a premeditated, pre-costed murder clears as cleanly as a vrata performed by accident.7 The mechanism there is indifference: intention is not an input.
Volume 3's four cases do not confirm that. They invert it. Here intention is an input — the guru states it, the text records it, the loophole turns on it — and the input is weighted backwards. Indifference would have produced four similar sentences. What the volume produces is a lizard for a clerical error and a kingdom for a slaughter.
⚑ So the two findings are not the same and neither replaces the other. A system that ignores intention and a system that reads it and inverts it look identical from any single case and completely different across four. That distinction is what this page is for, and it is available nowhere in Volume 2 because Volume 2 never puts four cases of one offence in six chapters.
⛔ Both findings stand. the-tirtha-launders-anything holds on its own evidence, and this page records the complication rather than resolving it.
One further thing runs through three of the four, and it is easy to miss because it looks decorative.
Triśaṅku is named for his three sins.
Pṛṣadhra is renamed for his, and this one is the harder case. Note 1388: *"Vrishaghna — the one who killed a bull, so named because of what he did. His actual name was Prishadhra."* And the body then uses the crime-name throughout.8
⚑ That is not an epithet. It is a displacement. The man who killed a bull by mistake, in the dark, in a storm, is thereafter called Bull-Killer by the text that has already recorded the killing as inadvertent. The permanent record keeps the act and drops the finding.
⚑ It belongs to a much larger habit: roughly seventeen name-etiologies in this volume, fifteen of them derived from an injury, a killing, a crime or a catastrophe, with only two exceptions.9 What the guru's-cow cluster adds is that the naming and the sentencing are the same operation — Vasiṣṭha's curse is the name, and Pṛṣadhra's punishment is administered twice, once by demotion and once by vocabulary.
You are looking at a set of decisions from the same body — disciplinary outcomes, code review verdicts, escalation calls, moderation actions — and you want to know what the body is actually optimising for.
Find a repeated offence, not a representative one. One case tells you nothing; a survey of many different offences tells you almost nothing, because the variation drowns the signal. What you want is the same wrong thing, four or five times, close together — the same object, the same relationship, the same act. Those clusters exist in every organisation and nobody looks at them as a set, because each one was handled by whoever was on duty.
Then hold the act constant and vary one thing. Here it is intention, and the axis was already in the material: the text states one man's inadvertence explicitly. Yours will usually be seniority, or timing, or who complained. You are not looking for whether the outcomes were fair. You are looking for what predicts them.
Expect the stated mitigation to be present and inert. The Pṛṣadhra case is the diagnostic one and it is much more common than outright indifference: the mitigating fact is noticed, articulated in the record, and then not carried into the outcome. A system that never mentions intention is easy to spot and easy to fix. A system that documents it and disregards it looks like careful process, generates a paper trail that reads as fair, and produces the same result.
And check who was harmed by each case. ⚑ When the outcomes make no sense on the stated variable, try the unstated one. Four verdicts that are incoherent on intention are perfectly coherent on whose cow it was — and that reading was available the whole time, in the text, without any theory.
Evidence. All four cases read in full in place: ch 350-6(36) for Nṛga and Pṛṣadhra with note 1387 and note 1388 read at their own lines;18 ch 352-6(38) for Triśaṅku, with the three-sin tally, the grievance, and "While devas and Vasishtha looked on" quoted directly;3 ch 355-6(41) for the seven boys, with both verdicts and the intervening notes checked for any acknowledgement of the split.4 The cluster was found by a scripted sweep for cow-episodes returning exactly four, all inside ch 350–355.5
Tension 1 — three of the four sit in chapters this build originally classified as low-yield genealogy and swept rather than read. They were recovered on a later pass. ⚑ That is worth recording on the page rather than in a method note: an instrument tuned to doctrine finds nothing in a lineage chapter, and the volume's sharpest single argument about desert was sitting in one.
Tension 2 — Triśaṅku's case has a duress term the others lack. A twelve-year famine is not nothing, and a reading in which Viśvāmitra is rewarding necessity rather than the act is available. ⚑ It does not repair the ordering, because the accidental cases have more excuse than duress and get less mercy — but the page does not pretend the four are perfectly matched.
Tension 3 — the seven boys' verdict may be two speakers, not one. The chapter runs five frames deep and the vocatives break three times inside it, all three translator-flagged.10 ⚑ A frame this damaged is a weak place to rest a claim about a single narrator's self-contradiction, and the claim is therefore stated as one chapter delivering both verdicts rather than as one voice.
Open question. Is there a fifth case? The sweep found four in this volume and did not run on Volumes 1 and 2. A fifth accidental case with a light sentence would falsify the ordering outright, and the check is cheap.
Against The Tīrtha Launders Anything: the direct competitor, and the resolution is in the body above. That page's instrument reads acts and never intentions; this cluster shows a system that reads intention and weights it backwards. Indifference and inversion are indistinguishable in any single case and unmistakable across four. Both pages stand.
Against Disabled by Māyā, Then Held Liable: the same shape in a different register, and this cluster is the human-scale version. There the agents are gods driven by Śiva's māyā and punished for what the drivenness caused; here the agents are men whose lack of intent is stated by the finder of fact and disregarded. ⚑ The god-level and the man-level cases agree, which is not something the volume ever says, and it makes the māyā material less a theological special case than the top end of an ordinary distribution.
Against Damage Becomes Institution: Pṛṣadhra is the sharpest instance that page will ever get, because the crime-name displaces the real name and the translator says so outright. Not an epithet acquired alongside an identity — an identity replaced by an offence.
Against Only What a Character Asks: the cluster is a worked demonstration of that page's mechanism. Four verdicts, six chapters, no interlocutor anywhere — no character hears two of them, so no character can compare them, so the incoherence is not available to anybody inside the book. Ch 355 comes closest, holding both verdicts in one chapter, and nobody in the chapter is listening to both.
Moral Agency and Categorical Responsibility — that page's argument is that blame requires a point in a causal chain at which a person becomes responsible, that no such point survives inspection, and that we assign it anyway because a system of desert needs one.
The four cows supply the empirical half that argument usually has to do without: a jurisprudence that locates the point explicitly, records it, and then sentences as if it had not.
Pṛṣadhra's guru does the whole analysis. He establishes the mental state — "realizing that he had committed the sin inadvertently and not voluntarily" — which is precisely the finding on which a responsibility-based system would reduce or discharge the penalty. And he then imposes a permanent demotion out of the man's varṇa, in the same sentence, with no bridging clause. The categorical judgement is not made in ignorance of the causal analysis. It is made after it, on top of it, without contradiction being felt.
⚑ That is a sharper demonstration than the philosophical case, because the philosophical case has to argue that the responsibility-point is illusory. Here it is found, stated in the finder's own words, and then does no work — which suggests the point was never load-bearing at all, and that the causal analysis is something a moral system performs beside its verdict rather than upstream of it.
And the exchange runs both ways. What the psychology page supplies back is why four incoherent verdicts feel unremarkable to everybody in the book: if the responsibility-point is a construction rather than a discovery, there is no fact about which of the four men deserved what, and therefore nothing for a reader inside the text to notice going wrong. What this cluster supplies forward is the observable signature — look for the mitigating finding entered into the record and then absent from the disposition. A system that never asks about intention is arguably more honest than one that asks, answers, writes the answer down, and rules on something else; and the second is far more common, because asking is what makes a process look like justice.
One handshake. The organisational-decision material is carried at page level by the Implementation Workflow rather than reached for twice.
Sharpest implication. In six chapters the volume runs what amounts to a controlled experiment: four men are responsible for a cow that is not theirs, the offence is constant, and intention is the only variable. Nṛga misdirects a cow in a donation by honest error and becomes a lizard. Pṛṣadhra kills a bull at night in a storm, taking it for a tiger, and his guru — explicitly finding that he committed the sin "inadvertently and not voluntarily" — permanently demotes him out of his varṇa and the text thereafter calls him Bull-Killer, a name the translator notes has displaced his real one. Triśaṅku deliberately slaughters a wish-fulfilling cow in a twelve-year famine, is cursed and renamed for three sins, and is then given a boon, a sacrifice, a kingdom and bodily ascent by the brahmana whose family ate — "While devas and Vasishtha looked on." Seven boys kill their guru's cow deliberately, launder it through a śrāddha, lie about it, and are punished as cruelty and credited as a good deed in the same chapter by the same narrator. ⚑ The two accidental cases draw the harshest and most permanent penalties; the two deliberate cases draw split or reversed verdicts. Nobody in the volume compares any two of them, and no note anywhere addresses the split. And the incoherence dissolves on a variable the text never names: what predicts the sentence is not what the man meant, but whose cow it was and what happened to them.
Generative questions.