I would start by keeping each blank's identity through a loose-length measurement and the setter's seated check, using an agreed reference. We once had a very convincing scatter of protrusion figures that combined different blank lengths with two different measuring positions. Nobody had done anything exotic; they had just filled a heading that didn't say enough. Also note the end contacting the stop and any visible burr or preparation difference. Don't retrofit paired values to the old sheet by matching the shortest-looking blank to the shortest reading. New linked observations are much more useful than tidy guesses.
Keep which end meets the stop in the observations, then. The setter can investigate the contact with those identified samples. Different appearance is a reason to look, not enough by itself to name the cause.