Soham - I don't think you will be able to resolve the small (.0044) distance difference between one S and 3 others. That corresponds to a sigma squared of only 3.6*10^(-6). If you wanted to resolve them by going to higher k you would have to go out to k=pi/(.0088 A) ~ 357 inverse angstroms, which might be a bit of a challenge. More fundamentally the delr parameter that you float would be nearly 100% correlated with the sigma squared, if you are also floating that. The numerical instability associated with parameter correlation can be expected to give you crazy values for both delr and ss (and correspondingly large error bars), which I think you are getting. So the 4 S atoms are effectively at the same distance. It's to just treat them as a single shell of 4 S atoms and describe the mean and SS of that distribution. It's really all the data can tell you without bringing some other information into the fitting process. hope that helps Grant Bunker