Archer
··Omega Qualified WatchmakerSo, assuming is is demagnetized.
Did you check that the watch keeps reasonable time? Because if the mainspring is only slightly too strong, rebanking can happen only dial up or dial down, at full wind. As soon as friction increases and the spring unwinds, it can disappear. Impact on timekeeping may be slight.
Of course if the mainspring is really too strong, it happens on all positions and not only at full wind. Big impact on timekeeping if it is regularly rebanking.
If the hairspring is sticky, deformed, out of the regulator, or if a small hair is tangled in it, you can have all sort of weird stuff happening, but usually on all positions, full wind or not. Two coils touching, in particular, may induce weird rate variations if they come in contact and stick at high amplitude. Usually, also a visible impact on timekeeping.
So, you may check what happens at full wind, half wind, dial up and on the side. But without opening it you may not be able to know much more... had to diagnose anything remotely.
Not sure if you regularly service 321's, but with a factory mainspring I've had them rebank for hours on end. As in still rebanking the following day...
The thing that can fool you sometimes is that immediately after winding, it might not rebank. It tends to start after the watch has been sitting for a while. My guess is that this happens because when the torque is very high, something is tipping slightly causing a bit of extra drag. As the torque drops a bit, this effect goes away and the amplitude actually goes up. I've seen this quite a few times, and this movement is, in my experience, very sensitive to rebanking with the amplitude at the high end.