Dave Angel wrote: > But I still think there must be code for the gamma function that would > be quicker. But I haven't chased that lead. Log of the gamma function is also an algorithm in its own right (e.g. published in _Numerical Recipes_.) Mel.