Dear SPM-ers, I am puzzled by a focal spot of unbelievably high significance, which must be due to an error of some sort; but what's the error, that is my question. I performed a correlation analysis on a set of smoothed grey matter maps (so a VBM analysis), and observed a very focal and almost out of the brain significant correlation, with a T=8.69, and p(FWE)=0.024. It looked fishy, so I had a look at the con-image, and resMS image. Now, although the voxel falls within the brain mask (mask.img) and has a value for the con and spmT image (which one would expect) the value for the ResMS image (and also the RPV image) is NaN. I don't understand how the T-value can be calculated of a voxel that does not have an estimated residual variance. Could this be a bug somewhere, or did I overlook something? Many thanks for your answer. Yours, Floris