Well, Peter, my opinion is based on my own experiences, but especially based on watching others do Foucault tests. The variability in people's readings was a real eye opener to me years ago. A technique I saw practised more than once was to place the knife edge where it ought to be, and see if the shadows were close. If they were, then the zonal measurement was considered 'on'. People also like to take a couple three zonal measurements and assume that they truly represent the mirror's curve, calculating figures like 1/30.12 wavefront error. Where do they get the .12 from?
A careful amateur who has used the test repeatedly and constantly verifies it against other tests, can probably achieve 1/10 wavefront. Otherwise, it's 1/4 wavefront at best, in my opinion. I really like the Caustic test for accurate measurements, the Ronchi test for smoothness, and the star test to verify and fill in the cracks (extremely sensitive to overall correction, any edge problems, and so forth).
I hope that at some amateur meet someday, we can have a mirror with known curve, that we can test against, to see how consistent amateurs are or could be.
Just my observations...Mel