Preservation
The packing slip
A manifest is a file that accompanies a set of archived files and describes their contents: which files, how many, what size, with which checksum, under which identifier.
It’s the equivalent of a delivery note. Without one, a folder of 12,000 files is a mass: you can neither prove it’s complete, nor detect that three are missing, nor verify that none has changed.
What it concretely enables
- Verifying completeness. The declared count and the actual count must match.
- Verifying integrity. Each checksum can be recomputed and compared.
- Unambiguous identification. A dated deposit identifier you can refer to.
- Accessioning. An institution receiving a deposit checks the manifest before accepting the batch.
- Establishing partial loss. Knowing exactly which files are missing beats knowing that some are.
What a good manifest contains
- A relative path per file, never an absolute one — an absolute path doesn’t survive the first move.
- A checksum per file, with the algorithm named explicitly (
sha256, not “checksum”). - Size in bytes, which catches truncation before you even compute a hash.
- Total file count and total volume.
- Creation date and the tool that produced it, with its version.
- The deposit identifier, stable and citable.
The rule that decides whether it’s useful
A manifest must be readable without the software that wrote it. A plain text format — CSV, JSON, or the archival world’s bagit layout — can be read with anything in twenty years. A proprietary binary format turns your integrity evidence into a second preservation problem.
Corollary: the manifest travels inside the package, not beside it in a database somewhere else.
In practice
- Produce a manifest for every deposit, even a small one.
- Use relative paths and a text format.
- Name the checksum algorithm explicitly.
- Store the manifest with the files it describes.
- Check it on receipt, not only on dispatch.
Obscura Flow builds documented packages with checksums and a readable inventory, verifiable after the fact from remote storage — the manifest is there to be checked, not just to be filed.