A man buys a hen. He wants the eggs but resents the feeding, so he proposes a division: he will keep the half that lays and dispose of the half that eats. The proposal has an obvious internal logic. It is also, in the most literal sense, going to kill the hen — because the half he wants and the half he refuses are the same animal, and his cleverness consists entirely in not noticing that.
The image is old and it is not really about poultry. It describes a move people make constantly in argument: reaching into a body of testimony, taking the part that helps, and discarding the part that costs — while continuing to lean on the authority of the whole. The trick is that the authority and the inconvenience arrive together. You cannot keep one and butcher the other. The hen does not survive the division, and neither does the argument.
Sanskrit intellectual culture names this move and files it. Ardha-kukkuṭī-nyāya — the maxim of the half-hen — belongs to the working stock of nyāyas, the short comparative maxims that Indian commentators use the way a modern logician uses named fallacies: as compressed diagnostics that can be invoked mid-argument without re-derivation.1
The maxim's specific charge is not inconsistency in general. It is selective acceptance of a source whose authority you have already conceded. That precision matters. Someone who rejects a text entirely commits no half-hen error; they have simply declined the hen. The error belongs to the person who has already eaten from it.
Pt. Madhavacharya Shastri turns the maxim on two opposed camps at once, which is what makes his use of it worth recording.2
The first camp holds that Hanuman was an ordinary animal, a monkey later inflated by devotion. The second holds that he never existed at all — a pure literary invention. These positions look like opposites. Madhavacharya's move is to ask both groups the same question: on what evidence do you say anything at all about Hanuman? Both answer with the Rāmāyaṇa. There is no other substantial witness.
Having established that both skeptics have conceded the text as their evidential base, he applies the maxim. If the Rāmāyaṇa is what tells you Hanuman existed, it is also what tells you what he was like — and what it says is not compatible with the ordinary-monkey reading. You may not draw existence from the text and character from your priors.
The structural feature worth extracting is the order of concession. The half-hen error is set up before it is committed, and it is set up by an admission that felt costless at the time.
A skeptic who begins by granting "well, the Rāmāyaṇa is where the story comes from" thinks they are making a bibliographic remark. They are in fact accepting a witness. Every subsequent selective refusal — that part is embellishment, this part is historical residue — now requires a criterion for the division, supplied from outside the text. The maxim's force is that it demands the criterion be stated. Usually it cannot be, because the real criterion was prior plausibility, which is what the argument was supposed to establish rather than assume.
Madhavacharya's sharpest instance is a single verse. At Rāma's first meeting with Hanuman, Vālmīki has Rāma tell Lakṣmaṇa that this speaker must have mastered the whole discipline of grammar, since across a long and elaborate speech not one malformed word escaped him.3
The choice of evidence is precise. Rāma does not praise Hanuman's strength, which a devotee might be expected to exaggerate, or his loyalty, which had not yet been demonstrated. He praises morphological accuracy over sustained extempore speech — a technical judgement, of the kind a listener either can or cannot make, and one with no devotional payoff.
The case study value is that this is the least likely detail to be a later devotional interpolation, and it is flatly incompatible with the ordinary-animal reading. The skeptic must either accept it, which sinks the position, or excise it, which is the half-hen. And the excision cannot be principled, because the same book, the same speaker, the same scene supplies both the existence the skeptic wants and the eloquence they do not.
Against the second camp the maxim inverts, and the inversion is instructive. To the reader who says the whole account is fiction, Madhavacharya's reply is not a proof of historicity. It is a question about effort: if the text is invention throughout, why work so hard to force its central figure into a shape the text itself never gives him?2
The observation is that total skepticism and selective skepticism are doing different things. Total skepticism has no stake in what the Rāmāyaṇa says about Hanuman's grammar, because it has no stake in the Rāmāyaṇa. The moment a critic starts arguing about what Hanuman really was, they have stopped being a total skeptic and joined the first camp, where the maxim applies. The essay's own resolution — that Vālmīki holds both truths at once, that Hanuman is a supreme scholar and statesman and is repeatedly, explicitly tailed and hairy-bodied — is offered as the reading that keeps the whole hen.
You are reading a paper that opens by citing a survey, and the citation is doing real work — the paper's whole premise rests on the survey's finding that a behaviour is widespread. Four pages later, the same survey reports that the behaviour has no measurable effect, and this time the paper explains that the instrument was crude. You stop. You go back and check whether the crudeness was mentioned when the first number was borrowed. It was not. You have found the hen.
Now you are in an argument about a text you care about. Someone says the tradition is late invention, then cites a passage from it to establish a date. You do not accuse them of anything. You ask the narrow question the maxim is built to ask: what rule are you using to decide which parts of this text testify and which do not? You wait. The value of asking is that the criterion, once spoken aloud, is either general — in which case you can apply it to the passages they kept — or it is not general, in which case everyone present can hear that.
And now you are drafting, and you notice yourself doing it: you have quoted a source for the claim you needed and stepped around the paragraph beneath it. You mark the spot. Either the objection gets answered in the text, or the citation gets dropped. Keeping both would be comfortable and would be the same error, committed quietly and in your own favour, which is the only way anyone ever commits it.
The move rarely announces itself. It shows up as:
The strongest support for the maxim's application here is the concession structure: Madhavacharya does not need to prove the Rāmāyaṇa reliable, only to observe that his opponents have already treated it as evidence. That is a genuinely economical argument and it holds independent of whether one shares his conclusions.
The tension the essay does not close. The maxim proves too much if taken as a general rule. Ordinary historical method is selective — historians routinely accept that a chronicle records a battle while rejecting its account of the omens preceding it, and they do so on stated criteria about genre, interest, and corroboration. Madhavacharya writes as though any division is illegitimate. It is not; undisclosed and unprincipled division is. The essay does not distinguish these, and a reader who adopts the maxim in its strong form would be unable to do source criticism at all.
The second tension is that the essay's own positive reading is itself a selection. It keeps Vālmīki's grammar verse and Vālmīki's tail, which is admirably whole-hen — but the same compendium elsewhere reads Purāṇic conception narratives non-literally via arthavāda, which is a division, and a principled one. The compendium therefore holds both a prohibition on selection and a licensed technique for it, and no essay reconciles them. Preserved, not resolved.
Open questions. What criterion would Madhavacharya accept for distinguishing testimony from embellishment within a conceded source? Is the maxim's real target selection, or only undeclared selection? And does the Aṅka's use of arthavāda count, by the maxim's own standard, as keeping half the hen?
Madhavacharya's use of the maxim converges with Five Flagged Contradictions and One Denied on the discipline of naming an inconsistency rather than smoothing it, and diverges on what follows: that page's tradition flags and holds, where Madhavacharya flags in order to win.
It sits uneasily beside Literal Reading of Six Kaula Terms, which argues that a tradition's own literalism is frequently the historically accurate reading against later sanitising interpretation. Madhavacharya would agree in the Hanuman case and would have to disagree where the Aṅka's arthavāda essays read conception narratives as radiance rather than matter. Same compendium, opposite instincts, unreconciled.
To Confirmation Bias as Ancient Problem. The parallel is exact in mechanism and inverted in framing. Confirmation bias is described in the psychological literature as a processing tendency — something that happens to a reasoner below the level of intention. Ardha-kukkuṭī-nyāya describes the identical output as a rhetorical move, something a person does and can be held accountable for. Neither framing alone gives you what you need. The bias literature explains why the error is universal and largely invisible from inside, but leaves you no way to challenge it in another person without an accusation about their psychology, which is unanswerable and therefore useless in argument. The maxim supplies exactly what the bias account lacks: a public, checkable demand — state the criterion by which you divided your source — that puts the burden on the argument rather than the arguer's motives.
The insight neither produces alone: selective citation is best treated as a bias when you find it in yourself and as a fallacy when you find it in someone else, and the two vocabularies exist because they do different jobs. A tradition that has only the fallacy will moralise about opponents and never audit itself. A tradition that has only the bias will audit itself and be defenceless in public argument. This compendium has the fallacy and, in its arthavāda essays, something close to the audit — which is why its interpretive machinery is worth extracting even by a reader who rejects its conclusions.
Sharpest implication. The half-hen error is not created at the moment of selection; it is created at the moment of concession, when a source is granted evidential standing for a cheap reason. Whoever concedes a witness first loses the right to be selective about it later — which means the decisive move in an argument over a contested text is usually made before anyone notices an argument has begun.
Generative questions.