Garrett,

But any internal epsilon is not appropriate for the epsilon required by the analysis being performed.

>True, but VFP includes that epsilon in the = function. I bugged the fact that it misses it in EMPTY() comparisons a while back: I'm not sure of the current status.