Somewhere in Skandha 3 the apparatus acquires a new voice. It appears three times and it is unmistakable:
Instead of noble rishis, translations usually say that Vishnu's words were like mantras composed by rishis, that is, Vedic hymns. This seems to be forced and unnecessary.1
Shri is interpreted as Devahuti's ornaments, but there seems to be no reason to do this.2
This is interpreted as the way a caterpillar moves, clinging to a new blade before it lets go off the old blade. Though not impossible, it would be far-fetched to interpret this as metamorphosis into a butterfly.3
Forced and unnecessary. No reason to do this. Far-fetched.
Three verdicts on other people's readings, delivered in a translation whose declared method is to give the simplest sense.
They are not the same move, and the differences are informative.
n. 999 names the field. Translations usually say — a claim about what the standard rendering is, followed by a rejection of it. The reader is told there is a majority position and that it is wrong.
n. 1069 names nobody. Shri is interpreted as — agentless passive, no usually, no attribution. Someone interprets it that way; the note does not say who, and does not say how many.
n. 1560 concedes first. Though not impossible — the only one of the three that grants the rejected reading any standing before dismissing it, and the only one where the alternative is described in full so the reader can judge.
So the species has three registers: contradict the field, dismiss an unattributed reading, and concede-then-reject. All three are in Skandha 3 and later.
This is the finding the vault should keep, and it is checkable.
The build's translator-intervention census across the whole volume catalogues eight species. Blanket disclosures, flagged divergences from the Mahābhārata, arithmetic that does not fit, corrections of the text's own premise, silent normalisation with the receipt left in, admissions of non-comprehension, and one explicit statement that liberties were taken.4
Overruling other translators appears in none of the Skandha 1–2 material and three times from Skandha 3 onward.
[n=3] — enough to name as a species, not enough to explain. Three instances in a volume of 1,575 notes is a rate, and a rate that starts partway through.
What that means is not determinable from this volume. It could be that Skandhas 3 and 4 contain more contested passages; that the translator's confidence grew; that the material he was working against changed; or that three notes is noise wearing a pattern's shape. [SPECULATIVE — no cause is claimed.]
Look at what the rejected readings would have added, because in all three cases they add the same kind of thing.
n. 999 — the rejected reading makes Viṣṇu's words like mantras composed by rishis, that is, Vedic hymns. That is an upgrade: it converts plain speech into liturgy.
n. 1069 — the rejected reading makes Śrī mean Devahūti's ornaments. That is a downgrade in the opposite direction: it converts a goddess into jewellery, presumably to avoid the awkwardness of Śrī appearing in a domestic scene.
n. 1560 — the rejected reading makes the caterpillar simile a metamorphosis into a butterfly. That is the biggest upgrade of the three: it converts a statement about continuity — gripping the next blade before releasing the last — into one about transformation.
In each case the rejected reading is more theologically interesting than the one Debroy keeps.
And in each case what he defends is the flatter option: plain speech, the goddess where the text puts her, and a caterpillar that crawls rather than transforms.
Which is consistent with his declared method and worth stating as the pattern rather than as three separate judgements. The species is not disagreeing with colleagues. It is declining enrichment, and the disagreement is a by-product.
n. 1560 matters more than the other two because the passage it governs is doing real work.
The volume's account of how karma crosses between bodies rests on that simile: a caterpillar does not give up a blade of grass, but moves on. Read as continuity, it forecloses any interval between death and rebirth — the accounting instrument is never put down.5
Read as metamorphosis, it would say something completely different: that the transition is a change of kind, that what emerges is not what entered, and that the intervening state is a chrysalis rather than a grip.
Debroy rejects the second reading as far-fetched, and the doctrine on the page depends on his being right. The volume's transmigration mechanism is a translator's call between two available readings of one image, and the note is where you find that out.
One absence completes the picture, and it is the sharpest thing about the species.
He never names the translators he is overruling. Not once. Translations usually say, is interpreted as, this is interpreted as — three agentless or collective formulations, no citations, no editions, no commentators.
Which is a considerable asymmetry. The reader is told a majority position exists and is wrong, and is given nothing with which to check either half.
The vault has logged the same agentless-citation shape from this translator's Volume 2, where an unattributed formula appears exactly where an authority is about to be demoted.6 Here the demotion is of unnamed peers, and the withholding of the name removes the surface an objection could attach to.
The three notes are not freelance opinions. They are the declared method being applied at the points where it costs something.
The vault holds the method from this translator's own statement of it: the simplest available reading, minimal doctrinal commitment, and no adjudication between schools.7
A policy of simplicity is invisible when the simple reading is also the standard one. It only becomes visible where the field has settled on something more elaborate — and at exactly those points, following the policy means saying that the field is wrong.
Which is what all three notes are. n. 999's forced and unnecessary, n. 1069's no reason to do this, and n. 1560's far-fetched are the same policy encountering three enriched readings and declining all three.
So the species is not a change of temperament partway through the volume. It is the visible surface of a consistent method meeting more contested material.
Which is the more economical explanation for why it appears only from Skandha 3 onward — and it is testable in principle against where the tradition's commentarial weight actually falls in this text, which is not something this build can check.
You encounter a note, a comment or a review that says the standard approach is wrong.
Check whether the rejected version is described well enough to judge. n. 1560 describes the caterpillar reading in full and then rejects it; nn. 999 and 1069 describe the alternative in a clause. Only the first lets you disagree with the correction, which is the difference between a note that informs and a note that instructs.
Ask what the correction is defending, across cases. Three rejections that all decline the more elaborate option are not three judgements; they are one policy applied three times. Policies are more useful to know than verdicts, and they are only visible in aggregate.
And notice when the party being corrected is unnamed. Not because anonymity is dishonest — it is often just economy — but because it changes what you can do next. A named alternative can be consulted. Translations usually say cannot, which means the correction has to be accepted or ignored, and there is no third option available to a reader.
The strongest evidence that this is a species rather than three unrelated notes is the shared direction. All three rejections decline the theologically richer reading and keep the plainer one, which matches the translation's declared method.
Tension — the species appears only from Skandha 3 onward, at [n=3].4 Three instances is enough to name and not enough to explain. Preserved unresolved.
Tension — none of the overruled translators is named, in the one species of note where the reader would most need to check. Preserved.
Open question. The caterpillar note decides the volume's transmigration mechanism. Is there any other endnote in the four Skandhas on which a doctrine so directly depends?
This vault holds the Śiva Purāṇa volumes by the same translator, and the intervention census is built on that comparison: ~19 disclosed interventions across eight species in this volume, against eleven across 150 chapters in Śiva Purāṇa Volume 1.4
That is a substantially higher disclosure rate, and this species is part of why.
What the comparison cannot settle is whether the rate reflects the translator or the text. The Bhāgavata is more heavily commented in the tradition and has more contested passages, which would produce more occasions to disagree without any change in disposition.
The vault holds both volumes and the honest statement is the narrow one: the same translator discloses more here, and one species of disclosure is new.
With Logan's Rebuttal as a Status Power Move. That page holds the mechanism by which a public correction functions as a status transaction — that the content of the rebuttal matters less than the act of issuing it, and that correcting an unnamed collective claims a position above the whole group at no interpersonal cost.
The three notes are that manoeuvre in an unusually clean form, because the medium removes every complicating factor. There is no audience reaction, no reply, no relationship to manage. Translations usually say… this seems to be forced and unnecessary is a rebuttal delivered into a footnote, to nobody, permanently.
What the pairing produces: the status framing predicts that correcting a collective is cheap and effective, which it is. What this case adds is the asymmetry of durability. A spoken rebuttal is answerable in the moment and forgotten; a printed one persists, and the party corrected is a category rather than a person, so no answer is possible at any point.
Which suggests the operative variable is not status at all but the availability of reply. The three notes differ in exactly that respect: n. 1560 describes the rejected reading fully, which restores the reader's ability to disagree, and nn. 999 and 1069 do not. Same speaker, same medium, same species — and one of the three leaves the door open.
That is a distinction worth carrying into any corrective register, because it is invisible from the status frame. The question is not whether the correction elevates the corrector. It is whether a reader who wanted to check could.
Sharpest implication. All three rejections decline the theologically richer reading and keep the flatter one — plain speech rather than liturgy, the goddess rather than her jewellery, a caterpillar that grips rather than transforms — which makes the species a policy rather than three verdicts. And the third is load-bearing: the volume's account of how karma crosses between bodies is a translator's call between two readings of one image, with the alternative described only well enough to be dismissed.
Generative questions.