Mel, I completely agree. I don't know how Larry was able to repeat readings like that, but he did it week after week. I'd also like to point out that Larry explained to us that even though the numbers might look great, one cannot make wavefront quality claims above about 1/20th wave based on Foucault data alone. He strongly believed in independant testing with the star test as the final authority.
My tolerance analysis of the Foucault test does indeed show that small errors in judging little r will result in huge wavefront errors. I was only trying to point out that it is possible for someone to use the test to a very high degree of precision after hearing the outright disbelief expressed by some of our members. I should point out that Larry is a professional optics manufacturer and has been making telescopes for many years. His knife edge readings during that beginner class were later proved by autocollimation tests on several of the scopes, including mine.
I know from personal experience that the Foucault test will never be that good for me and I'll always have the proper amount of uncertainty about it's results.
Also, readings to .001" don't necessarily mean accuracy of 1/100 wave, edge readings on very fast mirrors can require that kind of accuracy just to achieve 1/4 wave wavefront, but your point is well taken. Most of us should not put that much faith in Foucault testing.
-- Steve Strickland Lensnut@tpoint.net