I see a lot of bubble and boil going on here, so I'll drop another onion into the soup.
Blades that fail in fatigue do so under the loads of combined stress between tensile-bending and cyclic high frequency fatigue (Goodman's diagram).
As such any pertubation in the rotating arc of the blade, e.g. probes in the case add just a tiny bit of stress which while not being causal of the fracture does statistically pick the clock location where it does fracture. Thus for a given fleet problem, statistically most blade fractures will occur at the same clock location