M code for semiovariogram?

[fork]

Hi all,

Is there some good, sort of simple, and free m-code calculating 
semiovariograms, fitted and experimental? 

I am looking for something that doesn't require compilation.

octave-gpr looks like a good package, but I have trouble compiling packages on 
a MinGW system (windows at work -- yuck!), plus it seems much more complex and 
heavyweight than I need.

It is for calculating a semiovariogram for US counties and a migration index, 
but it would be nice for me to have anyway for many projects.

Since I just need a non-anisotropic semiovariogram, and no kriging, I am 
inclined to write it myself, but I don't mind standing on the shoulders 
whenever possible ...

Thanks!