present votes on the dispersed-fragments-issue seem to indicate to broaden the semantics of <msPart> to allow for the capturing of fragments.
I personally would welcome this change but it will introduce another issue for me: I’m working with 19th century letters and for a ‚standard’ letter I only give information about the country, settlement, repository and idno in msIdentifier. When we have fragments, though, this information would be moved to the msParts, leaving an empty msIdentifier. I wonder if it does make sense to make msIdentifier optional? Maybe as an alternative to msPart+ ?
My argument would be, that the combination of the msParts *is* the msDesc, rather than enforcing a singular msIdentifier as some sort of glue.