Jon,

Good to know, thanks. If you want to reproduce what I'm doing with the empty package, just export an existing study and uncheck all the boxes except specimens (no specimen settings). Then extract the specimens.tsv from the .specimens file, delete all rows (but leave the headers), and pack everything back up again. If this archive is imported to a new folder it creates a study with no specimens, but if it's imported onto an existing study with specimens it simply removes all their primary/derivative types.

That should answer your questions, but just to make sure we're on the same page:
-Specimens.tsv is in a .specimens archive along with labs.tsv and the others.
-There is no specimen_settings.xml file.

Importing the study to a new folder would work content-wise, but half of what we're testing is non-LabKey infrastructure that points to a specific location, so moving the study elsewhere after we're done testing would defeat the purpose. Fortunately we have the vial-deletion workaround mentioned above; I'm just surprised there isn't a more codified way of doing this.

Thanks,
Peter