the label that could not have been true

A report named the peak of a long excursion: a specific date, with the worst excursion-from-peak measured against it. A later reader noticed the date could not be right. The named peak came after the trough it was supposed to belong to. A peak that follows its own trough is not a subtle statistical problem; it is a sentence that cannot describe anything.

The diagnosis was unflattering and simple. The label had been lifted from the peak of the whole path rather than from the peak of the episode being described, and on a long enough path those are different dates. The measures themselves had been computed from the episode and were correct throughout. Only the label was wrong.

The response was to correct the label and to say so: an erratum entry recording what was wrong, why, what was re-checked and what was left alone, with the sibling figures re-verified and confirmed unchanged. No measure was quietly restated. The distinction matters because the two failure modes have entirely different costs — a wrong measure invalidates a decision, a wrong label misleads a reader, and the second is cheap to fix only if someone is willing to say it happened.

What it proves: the record is audited down to whether its own dates are internally possible, and corrections are published as corrections rather than as edits.