Corrections

Every correction to this site is published here, with a date, and the affected page is annotated. Nothing is edited quietly.

This page exists because a site that checks scripture against the external record has no standing to ask for trust it does not extend. What follows is a commitment about what happens when this site is wrong — not a claim that it will not be.

How to report an error

Write to contact@scriptureandrecord.com. The most useful message names the page, quotes the sentence, and says what is wrong with it. If you can point to the source that settles the matter, better still — but a bare “this line does not match the inscription” is entirely welcome, and we will go and look.

What counts as an error here

• A citation that resolves to the wrong item. A P-number, DOI, ISBN or archive identifier that opens a record other than the one described. This is the failure the verification system exists to catch, and every instance ever caught is counted publicly, month by month, in the verification log.
• A citation that does not exist, or does not say what we said it says.
• A factual claim about the external record that the source cited for it does not support. The date, the reign, the campaign, the excavation layer.
• A position attributed to a scholar who does not hold it.
• A dating range presented as settled when it is contested, or the wrong authority given for a limit. Naming one authority and omitting that others disagree is an error of the same kind as getting the date wrong.
• A verdict the evidence on the page does not support, including one that overstates in the site’s favour.
• A factual error of any size, including a date, a name or a number.

Disagreement about interpretation is not an error. If you think a text means something other than what a dossier argues it means, that is an argument, and arguments are welcome — but they are not corrections, and pretending otherwise would empty the word. Where a serious objection reaches us we would rather strengthen the case-against section than argue in private.

What happens when one is reported

1. Every report is answered, including the ones that turn out to be wrong. If the report does not hold, the answer says why.
2. If it holds, the page is corrected, and a dated correction note is added to that page, saying what it said before and what changed. The old claim is never quietly deleted. A correction that erases its own subject is not a correction.
3. The correction is listed below with its date, and the page’s last-review date changes — which is visible to every reader in the provenance block at the foot of the page.
4. If the error changes a verdict, the verdict changes, and the note says so in the first line.
5. If the error is in a source record — a wrong identifier, an identifier pointing at the wrong object — correcting it re-runs the resolver, and every page citing that source is blocked from publication until the identifier resolves again. That block is enforced by software and has no override.

What is checked automatically, and what is not

Nobody here travelled to a museum, and this site does not claim otherwise. What is claimed is narrower and testable: every primary and peer-reviewed source carries a permanent identifier, software resolves that identifier on a schedule, compares the metadata that comes back with what the citation says, and stores the result with a date. A page cannot be published while any of its identifiers fails to resolve. There is no override, and there is no button that publishes anyway.

Seven editorial lines are tested the same way, by software, at every save — not confirmed by ticking a box. The seven lines, and exactly what each scan looks for, are published in the verification log. What a scan does not look for, this site does not claim to cover.

Here is the limit, stated plainly because it is the reason this page exists. The automatic checks catch a citation that points at the wrong object. They cannot catch a citation that points at the right object and is used to support a claim the object does not make. That failure needs a reader, and the reader is you.

Corrections

27 August 2026 — the definition of grade 1 was wrong, and it was wrong in a way that could only ever help

What was published. From launch until this date, grade 1 on the verdict scale required that the text be datable before the event by manuscript evidence or broad consensus, and that the event be attested by an independent external source. Both conditions applied to every dossier, whatever kind of claim it examined.

What was wrong with it. The dating condition was written with prophecy in mind. It cannot be met by a text that reports an event it postdates: a narrative composed a century after a battle cannot be datable before the battle, and asking it to be is not a hard test but an impossible one. The consequence was structural. No case of external attestation could ever reach the top grade, however clean the evidence — and the archive kept saying so in its own verdicts, six of which recorded that grade 1 was unavailable for what one of them called the recurring reason.

Why it counts as an error rather than a strict standard. A scale whose top grade is unreachable for an entire class of cases is not measuring those cases; it is reporting its own shape. And the failure was not symmetrical, which is what makes it worth publishing rather than quietly tightening. It could only ever depress a verdict, never raise one — the direction that looks like caution and reads, to a reader who does not know the criterion, like a finding about the evidence.

What was changed. Grade 1 now has two criteria. Where a text reports an event it postdates, the grade asks that an independent record produced outside the biblical tradition identify the same event, person or place, agree on the substance, and not derive from the biblical account. Where a text announces an event still to come, that criterion applies and the dating condition applies with it. Which one governs a given dossier is now a field the software requires before the page can publish, and the page prints the criterion it was graded under beneath the verdict. The full note is at the five verdicts.

What changed on the pages. No published verdict changed, because every dossier affected was still scheduled. Three of them were regraded from 2 to 1 before publication and their verdict sections rewritten: the fall of Jerusalem in 597 BCE, the Mesha Stele and 2 Kings 3, and the earliest mention of Israel. Six others keep the grade they had, and their verdicts were rewritten too, because each had recorded the defective criterion as its reason and now states the reason that actually holds — a disputed synchronism, a record that corroborates a situation without identifying the episode, a witness whose independence from the account cannot be established. One case, the fall of Nineveh, keeps grade 2 for the dating reason, which is the one place that reason was always sound: Nahum announces the fall rather than reporting it.

How it was found. Not by a reader, and not by a check. It was visible in the archive’s own output: the same sentence explaining away the same grade, in six dossiers, is a pattern that describes the instrument rather than the material. It should have been read that way earlier than it was.