>
>I suppose the method of testing that hypothesis would be rewriting the code in C++?
>
>Does anyone have that kind of expertise?
>
>A C++ version of my code sure would make it easier to talk about outside of the VFP community :-)

This like (might) help:

http://theworld.com/~cme/P1363/ranno.html

One thing I might look at is leveraging something like GUID generation or MD5 hashing or something to generate your random number, then reducing it to the desired range (I think that link talks about doing this - you probably could just take the first X bits of the hash to generate a number which falls into the range you want).