I'm still of the opinion that, given a true rat.g.f., i.e., deg(D)>=deg(N), the degree of the numerator is irrelevant for the needed number of comparisons. What about rational egfs? ralf