Does varying twist rate contribute to that uncertainty? I imagine that a cable with a high twist rate is slower because the twist means the individual conductors are longer (maybe just by a couple percent) than one with a lower twist rate. And if the test equipment isn't configured for the pinout scheme being used (A vs B) then it might not even be pulsing a single pair at a time (each pair has a different twist rate) although perhaps that issue is automatically avoided.
> Does varying twist rate contribute to that uncertainty?
Yes, but beyond the effect that you're saying. The distance between conductors affects speed down the cable. Tighter twisting also changes the distance and the amount of dielectric around, which also varies speed of propagation.
These effects can be massive; a typical cat 7 cable has a velocity factor of 80% of the speed of light, varying by a couple of percent; a cat 5e cable more is often more like 65%. If you launch a pulse into a mess of mixed wiring to find out where something is unplugged or cut, you could be off by 20% or more.
