I am surprised to read requirements of agreement to 6mm or even 3mm. Is this intended for a 300m long calibration course or a 50m tape length?
For a 300m cal course, 6mm is 6 parts in 300,000 or 1 in 50,000. That is 50 times smaller than our SCPF of 1 in 1,000. The class 2 steel tapes which we use are only accurate to 1 part in 5,000, so this requirement would be 10 times more accurate than the steel tape markings made by the manufacturer.
Of course I may have misinterpreted Mark's post, perhaps he meant how much difference would be acceptable in one tape length of say 50m. In this case 6mm would be 6 parts in 50,000 or about 1 in 8,300. If that is what Mark meant, then, yes, I would agree that a remeasurement of that 50m interval would be worth doing.
The tapes I have used typically stretch about 6mm to 8mm when you apply 50N force, so basically I am saying that we need to put roughly 10lb force on the tape but it does not have to be precise. However, I do use a spring scale to get it accurate and even then I can sometimes get 6mm difference over the 6 taped lengths of a 300m cal course, but this is much more than adequate for bike calibration.
I once checked a 400m cal course and found the measurer had it short by 100mm. That was due to a mistake he had made and needed correcting. 3mm or 6mm forget it.
EDIT: Mark and Pete posted while I was writing. I agree with what they say.