No I meant on the certificates because that would provide a good source of information on this question of what is an acceptable difference between two measurements. I think I'll try to dig into some records to find out what my own experience has been. (I tend to mainly remember the times when 2 measurements agree to a remarkable degree, and block out the other times).
Personally if I get 2 measurements that are off by 0.01% I will remeasure (I think this happened to me once), but I am just balking a little about the effect of this rule on a new measurer.
I agree that a cal course measured incorrectly can significantly affect course distance. Also we shouldn't forget that there are other sources of error with this process: accurate and meaningful measurement of temperature is one; and our tapes themselves are not as accurate as we might like (see the "The Great Tape Stretch", Pete Riegel and Mike Wickiser's 2001 study of 33 steel tapes).
Now what I'm wondering is whether it would be feasible to use EDM technology to check our steel tapes or our calibration courses. Maybe someone can answer this.