The sheet says gauge aligned. Final acceptance names a drawing dimension. I have not established that the gauge and inspection use the same reference surfaces.

Find the gauge drawing if it exists. Otherwise aligned can sound much more specific than the evidence deserves.

Keep the gauge revision with the part revision. Both could have changed.

Ask what the correction stage costs to perform while you trace it. Your separate allowance still needs a basis, even if the old dimensions remain incomplete.

The gauge drawing is retained and matches the sheet revision. It is a fit-up check, using different contacts from final inspection. I cannot treat the two entries as paired dimensional readings.

Then the records don't measure movement on release in the way the shorthand suggests. They show two conditions assessed differently. What evidence actually says the first released assembly failed?

Kai, Sam's opening does say failure after release. Are you questioning the failure or the possibility of calculating a change from the two checks?

The latter, and whether the failed condition was recorded before correction. I phrased that poorly. A released failure can be real without making gauge aligned a comparable numerical starting point.

There is a released failure note before correction, but no value against an individual assembly. We have a stage observation, not the paired readings needed to quantify the change.

That distinction should go into the new study brief explicitly.

And don't let the final pass after correction become the predicted first-pass yield in the cost sheet. That would remove the very uncertainty you're trying to measure.

Agreed. The proposal now distinguishes historical final acceptance from the as-released condition. Quality is defining a measurement method for incoming, located and released states on identified current assemblies.

Will the new located measurement use the acceptance datums, or remain a process check? Either can be useful, but they need honest labels.

And defined supports for the released check. Merely unclamped is not a measurement method.

Quality proposes a separate process check while located and a defined drawing-based released check. They are documenting why the results cannot simply be subtracted. The trial has not taken place.

Has someone been assigned to record correction time and whether it was needed for each assembly? That's how the allowance eventually becomes evidence instead of a permanent question mark.

Retain the uncorrected reading before that work, with its identity. Otherwise the new process will produce another fine set of corrected brackets and another missing comparison.

Those records are assigned in the draft study plan. Thank you, Grace; checking the gauge drawing stopped me asking quality for a meaningless subtraction. Acceptance of the new fixture remains open pending the study.