Hi I am not 100% sure of what the issue is here but there definitely seems to a scenario where there is a difference between assigning peaks to resonances manually and using "initialise roots", I sometime seem to end up with HSQC spectra in projects where some peaks have a chain ID and some dont and some have H and N in their name and some dont. When you sort these by peak names you get essentially two sets of peaks, those assigned by initialise roots and those assigned manually (although I am not sure if it is 100% of them or whether there is another variable). I have not looked into this extensively so the differences in peak names could be arising from another cause but I wonder if this naming difference is what the problem is with the Mars export? Likewise without looking into it I am not sure how to reproduce this but I have seen it more than once. So I think Brians advice to re-use "initialise roots" is a sound one. Cheers Ben