The strongest disclosure in Volume 1 is a footnote of eleven words:
To make it clear, we have taken a few liberties with the text.1
Not this shloka is difficult. Not the meaning is uncertain. Not translations usually say. A statement that the English departs from the Sanskrit, made in the first person plural, with a stated motive.
It is endnote 1230, and it is the only note of its kind in 1,575.
The placement is the finding, and it is not a neutral passage.
The note attaches to Satī's condemnation of her father, at 4(4) — the speech she delivers in the sacrificial arena before raising udāna from her navel to her brow and burning.2
Specifically, it sits at the line about how good people handle other people's faults:
Good people play down small faults and magnify qualities, while the greatest ignore the faults. However, you only find faults.3
So the one place in the volume where the translator says the English is not tracking the Sanskrit is inside the volume's most emotionally loaded speech, at the sentence about the ethics of finding fault.
That is the most quotable moment in the Dakṣa cycle, delivered by its most sympathetic figure, at the point where she is prosecuting her father — and the reader is told, in a footnote, that liberties were taken.
The note does not say, and this is worth being precise about.
A few liberties could mean word order, could mean supplying an implied subject, could mean compressing a construction that does not survive into English, could mean a fuller intervention. The disclosure tells you that something was done and not what.
Which is more than most translations offer and less than the reader needs.
And the stated motive — to make it clear — locates the intervention in service of the English reader rather than of any doctrinal position. He is not smoothing a difficulty of belief; he is smoothing a difficulty of expression, or says he is.
The honest summary is: an intervention is declared, its extent is not specified, and its direction is toward readability.
The consequence for this vault is concrete and it constrains other pages.
Satī's speech is the source for several findings in the Dakṣa cluster — her two-akṣara claim about Śiva's name, her pravṛtti/nivṛtti argument, her escalating rule about how to respond to hearing the lord vilified, and the stated reason she gives up the body, which is the patronymic her husband uses.4
All of those rest on the fine wording of a speech the translator has flagged.
Which does not invalidate them. The disclosure attaches to one line, not to the whole speech, and the notes elsewhere in the speech are ordinary glosses.
But it sets a ceiling on how hard any of it can be pushed. An argument that turns on the precise phrasing of a sentence in this speech carries the qualification, and the pages that use it say so.
That is the practical value of a disclosure like this: it is not a warning that the translation is bad. It is a boundary marker telling you where the evidence thins.
The vault holds a matching note from the same translator's Śiva Purāṇa, recorded in that build: we have deviated from the text… as in our translation.5
Same species, two books. [n=2 SUGGESTIVE], and the vault's constraint on register applies — two instances is not a pattern and this page does not call it one.
What can be said at n=2 is narrow and still worth having: this translator does sometimes say outright that he has departed from the text, in a footnote, in the first person plural. Most translators do not, and the ones who do not are not thereby more faithful — they are less legible.
The comparison basis matters for a different reason too. This vault has both Purāṇas by the same hand, which is what makes the whole intervention census possible. A single-book reading could not distinguish this translator's habits from translation practice in general.6
This note is one of roughly nineteen disclosed interventions in Volume 1, across 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, overruling other translators, admissions of non-comprehension, and this.6
Against eleven across 150 chapters in Śiva Purāṇa Volume 1.
A higher rate, and one species that is new.
Which is a fact about the corpus rather than about either book, and the honest version of it is: the same hand disclosed more here. Whether that reflects the translator or the text is not settleable from the volumes alone, since the Bhāgavata is the more heavily commented work and offers more occasions for departure.
This note is local. The volume has two others that are not, and the contrast is the reason this one is the strongest.
n. 4 covers all of chapters 1 to 3. n. 1483 covers every shloka in this invocation.7
Those are blanket disclosures — a single statement governing a stretch of text, after which the reader is on notice for the whole range.
They cover more and tell you less. A blanket disclosure says something has been done across all of this and gives you no way to localise it. You cannot mark a passage, because they are all marked; you cannot exempt one, because none is exempt.
n. 1230 does the opposite. One line, one intervention, one motive — which means everything outside that line is, by implication, not covered by it.
And that is what makes it usable. A vault can carry a qualification attached to a specific sentence in a specific speech. It cannot usefully carry chapters 1 to 3 are approximate, because that qualification applies to too much to change any particular claim.
The strongest disclosure in the volume is the narrowest one, and the two widest ones are the least actionable — which is a general property of disclosures and not a criticism of these.
You are relying on a source that has told you, somewhere in its apparatus, that it has altered something.
Treat the disclosure as a boundary, not as a verdict. A translator who says we have taken a few liberties has given you information a silent translator has not. The correct response is to mark the affected passage, not to discount the source — the undisclosed interventions in every other text are the larger problem and you cannot see them.
Find out how localised it is. This note attaches to one line. A disclosure at the head of a chapter, or covering every shloka in this invocation, is a different object and covers different ground.6 The scope is usually stated and usually skipped.
And propagate the qualification to whatever you build on it. The rule that matters is downstream: any claim resting on the fine wording of a flagged passage should carry the flag with it. This is cheap at the point of writing and impossible to reconstruct later, which is why it has to be done at the time.
The strongest evidence that this note is exceptional is its uniqueness. It is the only first-person admission of departure in 1,575 endnotes, and the only one that states a motive for the departure.
Tension — the extent of the intervention is not specified.1 A few liberties is a disclosure without a scope. Preserved.
Tension — it lands in the volume's most quoted speech, at its most quotable line. Whether that is coincidence or a consequence of difficult material clustering in emotionally dense passages is not determinable. Preserved unresolved.
Open question. The Śiva Purāṇa sibling is recorded in this vault from that build rather than read in this one.5 Does a third instance exist anywhere in the Debroy corpus?
The devotional reading of a translated Purāṇa treats the English as transparent — the text is the text, and the apparatus is scaffolding to be read past.
This note makes that reading unavailable at one specific point, and the vault's whole translator-layer cluster exists because of notes like it.
What is worth stating is that the disclosure is a service, not a defect. A translation that never says this is not more faithful; it is less inspectable. The nineteen disclosed interventions in this volume are the ones the reader can account for, and the count of undisclosed ones is unknown and unknowable from inside the English.
The vault holds the translator layer as content precisely because it is the only part of the object that reports on itself.
With Transgression and Concealment. That page holds the operator's calculus around a deviation — that the concealment is usually costlier than the act, that a disclosed transgression converts an exposure into a manageable fact, and that the timing and framing of the admission determine whether it functions as candour or as evidence.
Endnote 1230 is that calculus resolved in favour of disclosure, in a domain where nobody would have caught it. No reader of the English could have detected the liberty, and the translator says so anyway, in a footnote, in the first person plural, with a motive attached.
What the pairing produces: the operator's framing treats disclosure as risk management — you admit because exposure is likely or because the admission buys credibility. Neither applies here. Exposure was near-impossible and the credibility gain is invisible to the ninety-nine per cent of readers who never open the endnotes.
Which identifies a category the calculus does not model: disclosure to an audience that does not exist yet. The note is not addressed to a reader who would have caught it. It is addressed to a reader who is checking — a scholar, a comparer, a build like this one — and it costs nothing with everyone else while doing everything for that reader.
That is a genuinely different instrument from the ones the transgression literature handles, and it has a signature: it appears in the apparatus rather than the body, uses the first person, states a motive, and is never repeated. Its value is entirely to the small population that reads apparatus, and its cost is entirely borne in a place the large population never looks.
Sharpest implication. Eleven words, once in 1,575 notes, in the first person plural with a stated motive — and they land inside Satī's condemnation of her father, at the line about how good people handle other people's faults. Which sets a ceiling on every argument in this vault that turns on the fine wording of that speech, and the pages that use it carry the qualification. A disclosure nobody could have caught, addressed to the small population that reads apparatus.
Generative questions.