Research Guides
Verify the Research File, Not Just Its Filename
Two files called “final-report.pdf” can contain different evidence. Before handing a research download to an editor or using it to reproduce a calculation, record which exact file you inspected. A filename and a reassuring thumbnail are not enough to settle that question.
This is a manual file-handoff workflow, not a claim that Colquhoun.shop provides preservation storage or checksum verification.
1. Establish a baseline before editing
Keep the original download separate from annotated, converted or cleaned copies. Record its public source, retrieval time, filename and byte size. Use an approved local tool to compute a SHA-256 checksum and record the algorithm alongside the result. Do not upload confidential research to an unfamiliar online checker.
The Library of Congress describes fixity checking as comparing a newly computed checksum with an earlier one to detect file changes. Its example also shows why equal file sizes and similar-looking images are not enough to establish that file contents are unchanged. Library of Congress: File Fixity and Data Integrity
2. Compare the received copy with the recorded baseline
After a transfer, compute the checksum of the received file with the same algorithm. Compare the complete values, not a short fragment. Keep the baseline record separate from your working edits so a later export does not silently replace the reference you intended to check.
In an invented handoff, the researcher sends a public report and its baseline record to an editor. The editor checks the received copy before marking it up. The record now distinguishes the original evidence from the editor's annotated derivative, even if both files look nearly identical.
3. Investigate a mismatch before choosing a copy
A different checksum establishes a difference in the compared file bytes; it does not explain the cause. Check whether someone re-exported the PDF, changed metadata, converted line endings in a text file or intentionally edited the content. Do not immediately call the mismatch corruption or tampering.
Preserve both versions while investigating. If a change is expected, give the derivative its own identity and record its relationship to the original. If the change is unexplained, pause the dependent claim or calculation instead of selecting whichever filename sounds newer.
4. Keep integrity separate from provenance and accuracy
A match is useful evidence that a copy agrees with the baseline under the chosen check. It does not prove who authored the file, whether the baseline was trustworthy or whether the figures inside are true. Those questions require source and content review.
Likewise, a checksum does not preserve a missing file. Maintain an appropriate authorized copy and recovery plan; a string of characters cannot reconstruct the report. A saved dashboard view is a different object again, as explained in the guide to saving a research view without mistaking it for an archive.
5. Use a small file-verification ledger
- Identity: source URL, retrieval time, filename and byte size.
- Baseline: checksum algorithm, complete value and creation time.
- Copy: received or stored file checked and comparison time.
- Result: match, mismatch or check not completed.
- Explanation: known transformation or unresolved difference.
- Use: quotation, calculation or editorial decision that relies on this version.
The goal is a reproducible handoff, not an impressive-looking hash. A reviewer should be able to identify the evidence file, verify the copy and see which transformations happened before a claim reached publication.