What is the tolerance range? --I mean 0.0001" OOS on a 0.010" tolerance to me can easily round up/down 0.0005" --But if it's ±0.0002 or less, then nope no way, unless you can affirm it passes some other way.
Maybe I should have been more general with how I worded the moral dilemma and said it measured just barely out of tolerance. But, I did state that the work instructions say to document the nonconformity, which is to imply that it measured enough out of tolerance to be a problem. It was meant to be a moral dilemma, not a technical dilemma.
However, thank you. Your questions did remind me that some places do rounding as a way of avoiding the splitting hairs problem. I haven't worked at a place that does that. For example, the place I currently work at has a policy of displaying one decimal place more than the print is dimensioned to. So a tolerance of +/-.0005 would be reported to 5 decimal places. If the part is reported 10 millionths out of spec it gets rejected. I don't entirely agree with that, but I don't set the policy, I just follow it.
What is the tolerance range? --I mean 0.0001" OOS on a 0.010" tolerance to me can easily round up/down 0.0005" --But if it's ±0.0002 or less, then nope no way, unless you can affirm it passes some other way.
Maybe I should have been more general with how I worded the moral dilemma and said it measured just barely out of tolerance. But, I did state that the work instructions say to document the nonconformity, which is to imply that it measured enough out of tolerance to be a problem. It was meant to be a moral dilemma, not a technical dilemma.
However, thank you. Your questions did remind me that some places do rounding as a way of avoiding the splitting hairs problem. I haven't worked at a place that does that. For example, the place I currently work at has a policy of displaying one decimal place more than the print is dimensioned to. So a tolerance of +/-.0005 would be reported to 5 decimal places. If the part is reported 10 millionths out of spec it gets rejected. I don't entirely agree with that, but I don't set the policy, I just follow it.
that's statistically absurd if you know your measurement device isn't capable to repeating at that tight of a resolution. I would start presenting my case on how silly that is, and start developing a revision to that spec which better aligns with NIST and industry standard, especially when it comes to "state of the art".
start with this white paper as an example (the TUR one, Tolerances Of Uncertainty):
https://www.transcat.com/calibration-resources/white-papers/measurement-uncertainty-basics-for-all-users
