Hiraṇyākṣa is fighting the boar and Brahmā arrives to tell Viṣṇu to get on with it. The instruction ends with a time:
O lord! Do not play with him, like a child with a snake that has presented itself. O god! O Achutya! Before his own terrible time arrives and he resorts to his own maya, kill this evil one. O lord! This terrible evening is approaching. O one who is everyone's atman! Ensure victory for the gods. This is the auspicious conjunction known as Abhijit and it lasts for two muhurtas.1
Ninety-six minutes, and the reason is given in the same breath: evening is coming, and evening is when the asura's power operates.
So this is not a ritual nicety. It is a tactical window with a stated closing condition, and Brahmā is giving Viṣṇu a deadline.
Debroy annotates the figure, and the annotation does not agree with what it is annotating:
Abhijit is usually taken to be one muhurta, not two. There are thirty muhurtas in a span of twenty-four hours, fifteen during the day and fifteen during the night. Some muhurtas are auspicious, others are not and Abhijit muhurta is the most auspicious of the fifteen muhurtas during the day. Literally, Abhijit is a muhurta that brings victory. Today, Abhijit muhurta, depending on the place, will be around noon.2
Usually taken to be one muhurta, not two.
The text says two. The tradition says one. The note reports both, marks the disagreement, and leaves it standing.
It is worth being precise about the translator's move here, because it is careful.
He does not emend the text to one. He does not silently translate the discrepancy away. He does not assert that the Bhāgavata is wrong.
He says usually taken to be — which locates the one-muhūrta figure in received practice rather than in a rule — and then supplies the framework that makes the disagreement legible: thirty muhūrtas a day, fifteen by daylight, Abhijit the most auspicious of the fifteen, literally the one that brings victory, falling around noon.
Everything needed to see that the numbers conflict, and no adjudication. That is the same posture as the note on the Buddha identification, where he writes though this is by no means obvious, and the note on jina, where he offers two readings and adopts neither.3
This one is not merely a bookkeeping difference, and working out why takes a step.
Abhijit falls around noon. Brahmā's stated reason for urgency is that this terrible evening is approaching. If Abhijit is one muhūrta at midday, then the window and the threat are hours apart, and the urgency in Brahmā's speech has nothing to do with the conjunction — he is naming an auspicious moment and separately warning about nightfall.
If it is two muhūrtas, the window is twice as wide, but the structural problem is the same: Abhijit's position is fixed at noon by the framework the note supplies, and evening is not adjacent to it.
So the passage bundles two different kinds of timing. Kill him during Abhijit is a ritual-astrological instruction about the most auspicious moment. Before evening arrives is a tactical instruction about when the enemy gets stronger. They are stated as one urgency and they are not one thing.
Register: this reading depends on the noon placement, which comes from Debroy's note rather than from the text. The passage itself supplies no time of day for Abhijit. What can be shown is that the text says two muhūrtas, that the note says the usual figure is one, and that the note places Abhijit around noon while the text's stated threat is the evening.
The volume gives no reason and it is worth marking the plausible ones without choosing.
It could be a variant reckoning — traditions do differ on the extent of auspicious conjunctions, and usually taken to be concedes that a usual figure implies unusual ones.
It could be that the composer wanted a wider window for narrative reasons, since a fight is being conducted inside it.
It could be an error that entered the transmission and was never corrected, which is the least interesting and entirely possible.
What the passage does not do is defend the figure, and the note does not attribute it. Two competing numbers are placed in front of the reader, one in the text and one in the apparatus, and the reader is left holding both.
The count-that-does-not-hold is by now this volume's most reliable behaviour.
Seven sheaths at 2(1) and eight at 2(10).4 Four kinds of ignorance at 3(12) and five at 3(20).5 Nine creations that continue past nine, and gods of eight types followed by nine names.6 Two avatāra rosters, one numbered and one not.7
What makes this instance different is that the disagreement is not internal. The Bhāgavata does not say one muhūrta anywhere; the conflict is between the text and the tradition surrounding it, and the only place the conflict exists is the endnote.
Which means this discrepancy would be invisible without the apparatus. A reader of the Sanskrit alone sees two muhūrtas and has no occasion to doubt it. The note is what creates the problem, and the note is also the only reason anyone would know there was one.
Worth stepping back to what the passage is actually doing with the timing, because the content is stranger than the arithmetic.
Brahmā is telling the supreme deity to hurry up and kill someone within a specific auspicious window, because otherwise the enemy's own hour arrives.
That places both parties inside a shared temporal order that neither controls. The asura has a time when his powers work. The god has a conjunction that favours victory. Neither can simply act; both are waiting for or racing against a schedule that exists independently of them.
And the deity complies. He kills Hiraṇyākṣa in the following chapter, and nobody suggests he could have done it whenever he liked.
Which is a genuine constraint on omnipotence, stated administratively, and never remarked on.
You have a figure from a primary document and the people who work with that material use a different one — a spec that says one thing and an industry that assumes another, a founding document and current practice, a standard and its universal deviation.
Debroy's handling is the model and it has three parts.
Report both, and attribute each. The text says two. It is usually taken to be one. Not the correct figure is — the source of each number is part of the information, and collapsing them destroys the most useful thing you have.
Supply the framework that makes the conflict legible. His note gives the thirty-muhūrta day, the fifteen daylight divisions, the meaning of the name, and where it falls. None of that adjudicates, and all of it lets a reader work out what turns on the answer. A bare sources differ is not the same service.
And leave it standing if you cannot settle it. The temptation is to pick, because an unresolved number looks like unfinished work. But a discrepancy you have documented is an asset for the next person, and a discrepancy you have quietly resolved is a decision they cannot see you made.
The check on whether it matters: work out what the number is load-bearing for. Here it is the width of a tactical window in a fight, which is why the conflict is worth flagging and why nothing much depends on it. Elsewhere in this volume a count differs by a factor of ten in the size of the cosmos, and that one matters.
The strongest evidence that the translator is documenting rather than correcting is the phrase usually taken to be. It locates the rival figure in practice rather than in authority, and it does not claim the text is in error.
Tension — the text says Abhijit lasts two muhūrtas and the received figure is one. Neither is adjudicated. Preserved unresolved.
Tension — the passage bundles an auspicious-conjunction instruction with a before-nightfall warning, and the note places Abhijit around noon. The two urgencies are not the same kind of timing and the text treats them as one. Preserved.
Open question. Is there a reckoning tradition in which Abhijit spans two muhūrtas, or is the Bhāgavata's figure isolated? The note does not say, and the answer determines whether this is a variant or a slip.
This vault's material on Purāṇic astrology treats muhūrta selection as a settled technical practice — fixed divisions, standard attributions, and calculations that different schools perform the same way even when they disagree about interpretation.
A text giving Abhijit at double its standard extent is therefore not a variant reading of an unclear matter; it is a different number for something usually computed identically. That is why the note is worth having and why it is worth not resolving.
The Arithmetic That Does Not Fit documents the same apparatus reporting a numerical check that fails and then routing around it. The two together show the translator's consistent practice on numbers: state the problem, supply the framework, decline to fix.
With Legge's Translator Discipline: Inter-Translator Disagreements. That page works the practice of putting disagreement on the page — where a reading diverges from another's, the divergence is printed, so the reader can see a choice was made.
Debroy does this here with a difference: the disagreement is not between translators but between the text and the tradition that uses it.
What the pairing produces: Legge's discipline addresses a rivalry between readers of one source, and its value is that it exposes interpretive choice. This case has no rival reader — nobody disputes what the Sanskrit says. The conflict is between what a document states and what the community operating that document does, and those come apart constantly in ways that inter-translator disagreement never captures. A specification and its universal workaround, a rule and its accepted breach, a founding text and the practice of its adherents: in each, both parties are correct about different things, and there is no reading to adjudicate. Which suggests a second discipline alongside Legge's, and it is rarer: not where do the experts differ on what this says, but where does what this says differ from what is done, with both recorded and neither corrected. Debroy performs it in one clause and does not name it.
Sharpest implication. The discrepancy exists only in the apparatus. A reader of the text alone has a coherent passage with a number in it; the note is what makes the number a problem — and it is also the only reason anyone would know to look. The translator has manufactured the difficulty and supplied the tools for it in the same sentence, which is what careful annotation does and why annotation is never neutral.
Generative questions.