One concrete data point for the thread. One practical note: write down what you did and when, before you need it. Reconstructing a timeline from memory three months later is how people end up unable to answer the one question that would have resolved it.

I would rather be corrected than agreed with, if it comes to it.

Following on from Dr.NephBHM_UK — and this may be the naive question:

Was that from a primary source or from a summary of one?

jason_paloalto said:

One practical note: write down what you did and when, before you need it.

Filing a mild objection. Mild because I might be wrong; an objection because nobody has addressed the case that does not fit. The objection nobody has made: this only works if the measurement is doing what we assume it is doing, and that assumption has not been tested here.

Happy to go further on any of that.

jason_paloalto said:

One practical note: write down what you did and when, before you need it.

That is right, and it stops being right at the edges. The general case is well behaved; the interesting cases in this thread are all at the boundary where the general case breaks.

Reporting back.

Following up because I said I would. I changed one thing, left everything else alone, and waited the full interval before looking. It resolved, and I cannot prove which half of that did it.