Follow-up Comment #1, bug #21954 (project relax): As these are single replicates for each spectra subset, then the variance averaging should have no effect. I.e. both standard deviations (SD) after each spectrum.error_analysis user function call - the replicate SD and averaged SD - must be identical. _______________________________________________________ Reply to this item at: <http://gna.org/bugs/?21954> _______________________________________________ Message sent via/by Gna! http://gna.org/