Yes the customer wants a calibration certificate and yes he wants lines measured on a piece of glass.
I seem to recall some confusion between inspection, calibration, and verification by different groups and the US and Europe; maybe there is some of that going on here.
Calibration of an item does not need to be directly to a "standard" (ultimately the only standards are SI unit items), almost all calibrations are done by a chain of traceability. Even the fancy machine NIST used to measure my master scale, was checked against other scales and end standards.
I brought up 17025 because there are specific reporting requirements required by it, and the customer wants a 17025 accredited calibration certificate (in the future for us, not there yet).
But back to my original question, which is more basic metrology and trying to do things in an industry accepted way:
If we measure the part 3 times and get; 50.000, 50.000, 50.002mm, is it ok to report 50.0007mm (along with the associated uncertainty)?