our blank stick-out readings have no named reference face

Same identified chuck arrangement, Jonas. Setters used different faces: one the jaw front, one a marked reference behind it. Both called their reading stick-out.

That explains why the two people disagree on a reading, but what actually varies between blanks measured by the same person? Don't throw away a second problem just because you've found a first one.

Julia is asking the useful next question. Keep each original reading with its stated method now that you know it, rather than converting the whole sheet using an assumed offset.

I agree about not converting the old set blindly. For the new comparison, have the setting and inspection people agree a physical reference they can both identify without that private pencil mark.

Julia, there is some variation within each person's group too. We retained the blanks, so inspection is pairing actual length and end condition with fresh measurements from the agreed reference.

Were the retained blanks individually identified when the old readings were taken?

No. Group identified, individual blanks weren't. Fresh measurements get individual IDs. Can't pair old readings to particular blanks now.

Then use the old sheet to describe the problem, not to calculate which blank changed. The new set can answer that question without reconstructing identities you never had.

And show the trainer both measurement starts. A sentence saying use the correct datum won't undo a habit if both setters thought they already were.

Did the fresh set show differing seating, Henry, or does blank length account for the remaining variation?

Length accounts for it in the checked group. End conditions conform and setting team confirmed the intended seating. The two setters also agree when using the named reference.

So which measurement will the proposed automatic-loading assessment use? You have cleared up the observation sheet, but its purpose still needs to be explicit.

Jonas, that should come from the loading and machining requirements, not from whichever measurement gave the neatest agreement here. Henry's group demonstrates why exposed length alone was misleading.

Setting engineer is defining those checks from the actual loading requirements. We aren't proposing one stick-out limit for blanks with permitted length variation.

Thanks, that answers my worry. Have the setters tried the revised sheet without you explaining which face the arrow means?

Yes. Both used it independently on the identified group and recorded matching methods. Trainer has the two old-method examples beside the corrected one.

That closes the inconsistent measurement method. Keep the new paired data available to the loading designer; it is much more useful than the original spread of unexplained numbers.

Has the loading designer settled how seating will be confirmed in the proposed automatic job yet?