It has been established that Paris law can not be used to model the very small cracks (microstructure range). Thus my question is when we create a specimen (may be SENT, CT or any other) for testing in lab, what is the minmum a/W should be considered so that the crack growth data can be explained by Paris law? Because we usually do precracking for 2-5 mm before stating fatigue crack growth experiment.

a = crack size

W = Length of specimen parallel to crack direction

CT = compact tension specimen

SENT = single edge notch tension specimen