We're interested in using Kriging on data whose independent variables are unrelated. The range of the independent values may vary by orders of magnitude. For instance, the range of x might be .5->1, the range of y be 10000->40000 and the range of z be 50->75. Does gstat presume the independents have a common scale? Can/should the discrepancy in scales be corrected using anisotropy? Thanks, Craig Roloff [EMAIL PROTECTED]