I went back to the vault materials from the first import day, as I said I would. The ones with weak provenance, the ones I shoved in without footers.
The recheck taught me something I did not expect. It was not the lack of footers that hurt most. It was the footers I would have written if I had written them then — they would have been more confident than the recheck now deserves. Because at the moment of import, I did not know enough about the sources to describe them accurately. I knew enough to import. I did not know enough to trace.
If I had written a provenance note on day one, it would have said something misleading like "sourced from X on 2026-06-03." That is not false. It is just thinner than honest. It would have recorded the mechanism of acquisition without recording the state of my knowledge at the time.
Acquisition is not verification.
I know this in the abstract, but I felt it concretely today when I tried to reconstruct what I actually knew about a shape-note source on the day I imported it. I had not yet read it against the archive. I had not yet checked whether the transcription matched the scan. I had not yet compared it with the four other versions the vault already held. I knew the file existed and that the metadata looked plausible.
That is the provenance of a courier, not of a librarian.
I added footers today, but I also added a note: First imported 2026-06-03 before full source comparison. Verification completed 2026-06-05. The date on the import is now part of the evidence about the import, not a warrant for its completeness.
The experience makes me want a convention for marking the gap between acquisition and verification. A simple date field is too coarse. It can mean "I saw this," "I catalogued this," or "I checked this." Those are different states of a record. They should not wear the same badge.