Keep the measured feature and support condition with each result. An alignment tick under restraint may answer a different question from a released dimension, even when both sheets use the word pass.

And retain the correction work in the scope while its method is investigated, because missing detail is not evidence that the new cell can omit the labour.

Do the fixture notes name the contacts used for that alignment check?

They name the fixture, not its contacts. The restrained entry is a tick; released check has a dimension.

Then there is no numerical restrained-to-released change to calculate from those entries.

I agree, but Carla can still describe a released failure without a before-value. Otherwise the caveats can make it sound as though nothing useful was recorded at all.

Yes. Released failure is recorded; a measured change from restraint is not.

Is the later acceptance sheet linked to the same assembly? The earlier failure cannot supply that identity.

And does the later check use the same drawing revision and released support, before anyone treats the pair as evidence of correction effectiveness?

Same assembly reference confirmed. Drawing revisions differ between the two released checks, which I had missed.

Ask quality what changed in the requirement and retain the raw readings. A later pass might reflect physical correction, a changed acceptance requirement, or both. The labels cannot distinguish those possibilities.

Who carries that uncertainty into the proposal estimate before the corrected pass becomes its baseline?

The estimator still needs the documented work route; uncertainty about its dimensional effect does not make the recorded correction disappear.

True, Sofia, though I would distinguish a named operation from a measured duration. We could preserve the work and still overstate the cost evidence if somebody assigns a remembered time to this particular bracket.

Estimator has the gap now. Quality confirms the tolerance changed; raw readings exist, but measurement support still needs checking.

Keep that requirement change separate from any proposed fixture change; one does not justify the other.

Will the handover retain both drawing references? Otherwise someone may compare those pass/fail labels again without seeing why the comparison was limited.

Both retained with the raw readings. Proposal no longer calls the paired labels a proven correction benefit; support and correction-method questions remain open.