Yeah, I found it with google. I also have Montgomery's "Speeding" paper, and it was instrumental in implementing the progression of f(n). In fact, I just looked at it again, and realized that I didn't read far enough! Section 9 in there may have answers to my question, so I'll try to understand that first. After that I'll probably need to take you up on your offer to answer questions...

And by the way, I thought your thesis was very readable - thanks for putting it where google could get to it.

As to the observation of my buggy version, I guess that incorrectly implementing the progression of f(x) <-- f(x+h) may have accidently set up a condition where the highest factor of p-1, s, divided g(vw) +- g(u) (using montgomery's notation). But this seems to "accidently" happen more often that I thought. I'll shut up now because I'm talking about stuff I don't understand very well yet...

- ben