Which face did our released bracket check actually use?

Only the copied sketch, Luca; the entries say height with no face named, and the inspector is available to reconstruct the setup

A reconstruction may explain it. It can't recreate readings nobody recorded. Ask which parts are still identifiable before spending hours arguing over the handwriting.

Amy's right about the limit. I meant use it to understand the method, not to backfill the old sheets from memory.

Inspector demonstrated the correct datum setup on an identified retained bracket; copied sketch belongs to an earlier component, and quality has withdrawn it from the proposal

Did the retained bracket fail before correction, or is it one that had already been put right?

Already corrected, so that check explains the method but gives us no first released result for this part; new observation starts before correction

Make the first released check happen where the work happens. If the bracket reaches the correction bench first, you'll collect another perfectly tidy set of answers to the wrong question.

Include ordinary brackets that don't need correction too. Last time we followed only the rejects, the resulting allowance made every job look worse than it was.