David Relson wrote: > What you say sounds reasonable. Bogotune needs the variation in ham > scores so that it can adjust the number of false positives. Why? Isn't it done just by counting so and so many errors top down? You might expect that this leads to strange results, but it should work. pi