How to create a PDF/A archival scan on Windows

Decide whether PDF/A is the right preservation object

PDF/A is intended for long-term reproducibility, but an archive still needs retention policy, provenance, fixity, storage redundancy, and periodic integrity checks. Keep original image captures when they carry evidentiary value. Choose an archival derivative because it fits the receiving repository and policy—not because a file extension or marketing label sounds permanent.

Prepare stable pages and useful metadata

Correct page order and orientation before archival export. Use conservative cleanup, because an attractive derivative should not erase source information. Add title, author or creating organization, subject, and keywords that will remain intelligible outside the original folder. Unicode metadata matters for names and languages that ASCII cannot represent faithfully.

Create the target and preserve its identity

The frozen build targets PDF/A-3u and can include a local OCR text layer. Save to a new atomic destination so cancellation or failure cannot replace an existing good file with a partial one. Record the application version, profile, date, byte size, SHA-256, source inventory, and whether OCR and cleanup were applied.

1. Preserve and inventory the source pages
2. Verify order, orientation, and conservative cleanup
3. Run OCR and inspect critical values when search is required
4. Choose PDF/A-3u and enter meaningful metadata
5. Save atomically to a new file
6. Open, search, inspect, and run an independent validator

Separate internal checks from independent validation

A PDF header, page count, output intent, XMP packet, embedded fonts, and PDF/A identification metadata are useful checks, but they are not a substitute for a current independent conformance validator. Record the validator name, version, profile, result, and report. If validation fails, preserve both the failed candidate and report, fix the specific rule, regenerate, and validate the new identity.

Failure modes and limits

Repository policies and validator interpretations can differ. Encryption conflicts with normal PDF/A goals, and an internally marked file may still fail an independent conformance rule.

Stop when the source is incomplete, page order is uncertain, cleanup removes meaningful marks, OCR changes a critical value, the requested output contract cannot be verified, or an external validator rejects an archival file. A successful command is not a substitute for inspecting the document.

Recovery and verification

Keep the source, validator report, failed candidate, and profile. Correct the reported rule, regenerate to a new filename, and validate again before accessioning the file.

For reproducible work, retain the application and profile version, source inventory, acquisition settings, language, cleanup operations, output type, final byte size, SHA-256, verification steps, and any independent validation report. Keep the official record separate from temporary previews.