Correction policy

The short version

If something here is wrong, please say so. You do not need a title, an affiliation or an introduction, and you do not need to be polite about it. Corrections are the most valuable thing this project receives.

What we need

Enough for someone else to check it: what the record says, what it should say, and why — a source, a measurement, a photograph, a published reference, or your own examination of a comparable sheet. A correction that arrives with a source can be acted on the same week. One that arrives without one becomes an open question instead, which is still useful.

What happens

1. Every submission is acknowledged.
2. It is assessed against the evidence hierarchy — where does this sit relative to what the record currently rests on?
3. If it is accepted, the affected claim is superseded rather than overwritten. The old claim stays in the record, marked as superseded, so the history of the mistake survives.
4. The change is logged with its reason and the name of the person who found it, unless they ask otherwise.
5. If it is not accepted, we say why, in writing. A rejected correction that was reasonable is published in the rejected-hypotheses record, because a plausible lead that turned out to be wrong is worth as much to the next researcher as one that was right.

Target turnaround

Acknowledgement within a week. A decision within a month. If a correction needs the physical sheet to be examined, it may take longer than that, and we will say so rather than let it go quiet.

Disagreement

If a correction is declined and the person who sent it still thinks it is right, that disagreement is published on the affected claim. The project would rather carry a visible dispute than a false consensus.