One inspector measured the bracket while it was still clamped

Only that entry with certainty. Some sheets say bench, others say check OK. The spreadsheet made them all look much more comparable than the paperwork does.

Then mark the unknown stages explicitly and rebuild the comparison from entries you can actually identify; don't fill the gaps by assuming one inspector always used the same method.

Keep the drawing revision too. Are all those brackets even the same variant?

Same thickness and drawing revision, thankfully. The main confusion seems to be measurement stage, although I still need quality to confirm the bench supports used.

Were any bench readings taken after straightening? That would be another stage hidden under the same heading.

Two sheets mention correction before the final reading. I'll keep those separate. Our useful set is shrinking, but at least I can explain why.

Small honest set beats a large muddle.

You may still need a fresh trial. We lost a week trying to squeeze an answer from old records that never contained it. Ask quality what minimum new measurements would settle the fixture question.

Whether to add restraint to the upright. We have withdrawn that recommendation based on this graph and asked quality and welding to define a current-part trial with identified measurement stages.