The bracket passes after a correction nobody timed

Mixed finished batch. Two operators remember the difficult ones, not a count.

Don't make a percentage from the memorable ones. Ours were memorable precisely because they were awkward.

Yes. One tick for the batch after the bench work. No individual-part entries.

Our revised fixture plan needed current-part observations for that reason. Could your next ordinary batch distinguish released checks from later correction, with the same part identities through both?

And who puts the tick in, Grace? The person doing correction or somebody closing the paperwork afterwards?

The batch tick at least tells you the operation wasn't imaginary. Keep it as history without asking it to reveal which brackets went through the vice.

Inspection closes the batch route, David. The operator does the correction. We've agreed a current batch with individual entries; the estimator knows the old tick can't supply a rate.

Same drawing and acceptance method for that current batch as the proposed fixture review?

Drawing H. Released check to quality's current method. Fixture designer has both.

Are incoming fit-up differences recorded too, Grace? I wouldn't want the new fixture blamed for a variation that was already there before the bracket was assembled.

Nora, yes, but don't let the observation sheet grow until nobody fills it in. Ask the people doing the work which details they can reliably capture.

Could the operator and inspector try one blank sheet together before the batch starts? That should expose any field each thinks the other person completes.

They did. Correction start was owned by nobody. It now has a named recorder.

Good catch before the trial. Does the entry distinguish active correction work from waiting for inspection?

That split helped us explain cost without pretending the operator spent the whole interval correcting. The waiting still affected delivery, just for a different reason.

Yes, separate now. First current batch recorded; quality is checking the linked readings before we use the count.

Will you post whether the correction frequency was anything like the remembered problem batch? Curious how far the memories and the ordinary work diverged.