Hi Chris

Thanks again for your detailed answers. I thought it might have something to do with removal of duplicates. I also switched on merging, so this must be why dcm2niix merged series across two different study dates, since they had the same series number and the same geometry, just different numbers of slices.

I had to switch on merging because some series ended up divided, perhaps because of differences in slice thickness or some other feature of CT. It looks like my correct approach has to be to start a different instance of dcm2niix for each study day to ensure that scans are merged within each numbered series, but not across days.

Many thanks
Paul