Keep the record intact.

What these checks establish

A baseline records each source snapshot, editorial JSON dataset and original PDF access copy by relative filename, byte length and SHA-256 digest. Verification compares the current bytes with that baseline and reports missing, added or changed files. The report is excluded from its own inventory so checks can be appended without changing the content baseline.

The recorded recovery test built a compressed preservation package outside the source repository, extracted it into a new temporary directory and compared every restored file with the baseline. The generated local recovery copy is held outside the source repository. Independent offsite-copy evidence, where recorded below, is separately verified and scoped. Preservation packages are not published as downloads.

The original deposits remain accessible through Zenodo, and the website source is held by the site platform. Those separate services provide source provenance; they do not establish that an independently administered backup of the complete archive has been made. The inventory covers source and catalogue content, rather than a complete disk image or every website dependency.

Future checks use python scripts/stewardship.py --verify. After reviewed additions, an explicit baseline refresh records the changed scope. A scheduled workflow can run the same verification; a timetable alone is not evidence of a completed check.