We kept the faceplates. The arrows were added during our image discussion, not by quality.

Keep those arrows as disputed annotations, outside the confirmed reference set. I'd ask whoever owns the images where copies have gone too. A corrected folder doesn't retire a printout on somebody's bench.

Jasper, do you mean remove the images, or only their status as confirmed examples?

Their confirmed status. Keep the original images and the reason they were questioned. Bad labels can be very useful evidence of how the mistake happened; they just shouldn't be the answer key.

Ben Allen, has quality examined the retained surfaces yet? Organising the captions does not settle the physical decision.

Quality accepted the broad band region and rejected a small edge score on another faceplate. Both are identified.

Can your proposed camera view show that edge score, not merely the easier broad band?

Only at one presentation so far. We've kept the other views; the edge coverage isn't agreed.

Then your reference decisions are ahead of your imaging result, which is a useful place to be. Ask the imaging person to compare that edge across the intended presentation range, with the accepted band retained as a false-reject challenge.