Corrections, sources, collaborations

A message that names a page, quotes a sentence and attaches a source is a claim about what the record shows, and claims like that get checked against the cited literature regardless of how the message is phrased. A message that asks the archive to soften a term, drop a comparison, or lead with a different number without offering a source for why is a request about tone, not a claim about fact, and it is answered as such — usually with a pointer to the methodology page, which already explains why the lower-bound convention exists and is not revisited case by case.

This distinction is not a way of dismissing difficult mail; testimony in particular often contains no citation at all and is still taken seriously, because a firsthand account is itself a kind of source, evaluated on its own terms rather than measured against a footnote standard built for secondary scholarship. What the distinction protects is consistency: two readers making the same kind of request should get the same kind of answer, whether the request is easy to agree with or not.

Three requests recur often enough to name plainly. The archive will not write a paragraph to a correspondent's specification, however well-intentioned the ask, because editorial judgment here is made openly on the methodology page and defended there, not negotiated privately by email. It will not remove a sourced claim because someone finds it uncomfortable, embarrassing to a national narrative, or inconvenient to a family history — removal on request would make the archive's presence or absence of a fact a function of who complained loudest, which is the opposite of what a sourced record is for. And it will not share who wrote in, what they asked, or what they disclosed, with anyone outside the small editorial process that handles the inbox.

None of that forecloses correction: if a request comes with a source showing the existing text is wrong, it is corrected regardless of who sent it or why they cared. The refusal is specifically to tone requests, removal requests unaccompanied by a sourcing argument, and any request to treat correspondence as public or shareable. Those three lines are held the same way for a government archive, a private citizen, and a journalist, because the point of stating them here is that they do not bend to who is asking.

Non-commercial reuse with attribution is the default answer to translation offers, classroom requests and small-scale republication, because the project's purpose is read, not owned — a teacher excerpting a chapter for a lesson or a volunteer translating it into a language not yet covered is doing exactly what the archive exists to make possible. Larger requests, such as translating an entire section or republishing at scale, are answered the same way in principle but benefit from writing in first, since the archive can then supply a clean source file and the current wording of any figure that is under active revision, rather than have a translator work from a page mid-update.

What is not granted, in any version of a reuse request, is commercial use without a separate conversation, or reuse that strips the sourcing apparatus out of a chapter while keeping its conclusions — a translation that keeps the number but drops the footnote defeats the purpose of publishing the footnote in the first place. Anyone translating or excerpting is asked to carry the citations across, not just the prose, which is also why a sample passage is a useful thing to send before starting a larger translation.

Is there a difference between writing about an error and writing about a disagreement with how something is framed?

Yes, and the archive treats them differently on purpose. An error is a mismatch between the page and a cited source, and it gets corrected. A framing disagreement is a request to change an editorial choice the archive has made openly and stands behind, and it gets a considered answer rather than a change.

What should someone send if they want to submit a source the archive has not used?

Name the document, dataset or study specifically enough to be found and checked — a title, author, date and, where possible, a page or table reference — rather than describing it in general terms. A source that can be verified quickly is far more likely to end up cited than one that has to be tracked down first.

Will testimony from a survivor or a descendant be published without asking first?

No. Testimony sent to the archive is read carefully and treated as sensitive by default; nothing from it is published, quoted or attributed to a named person without that person's explicit, separate agreement to be identified and quoted.

Can someone ask the archive to take a chapter down entirely rather than correct it?

That request is heard, but a sourced chapter is not removed because a reader would prefer it did not exist; if a specific claim within it is shown to be wrong or unsupported, that claim is corrected or removed, which is a narrower and different outcome from taking the chapter down.

Does offering to translate a chapter mean the offer will automatically be used?

No. Every translation, whether offered by a volunteer or produced in-house, is checked against the source text for accuracy before publication, because a mistranslation would misrepresent both the archive and the scholarship it is drawing on.