Methods
Retaining source copies
Preserve the material actually read, subject to rights, and bind its citation with a content digest.
Problem
A URL identifies a place to request material, not the bytes a researcher saw. Pages move, vanish, redirect or change in place, and linked files may be replaced without a new address. A citation to the current URL cannot prove what supported an earlier assertion. A retained copy, recorded observation time and content digest make that earlier reading inspectable. Retention must still respect licence, privacy and access restrictions; a forbidden copy is not a valid shortcut.
Steps
Identify the material
Capture source title, issuing body, original URL or identifier, retrieval time and visible version or publication date. Specify the representation read, such as a PDF revision, HTML page or data release. Redirects and dynamic pages may yield different content at the same address, so record the final fetched location and relevant request context where they affect interpretation.
Preserve a lawful copy
Store the material used in a form that keeps its text, context and pagination or structure readable. Apply licence, privacy duties and reader permissions before retaining or sharing it. If a full copy cannot be kept, record that limitation and preserve permitted metadata or excerpts. Do not imply that an inaccessible retained item can be independently checked by every reader.
Bind citation to content
Calculate a cryptographic digest over retained bytes and record the algorithm and digest beside the citation. A digest detects whether a later file is byte-for-byte identical; it does not prove publisher authorship or factual truth. Verify that the cited passage and locator resolve in the retained representation before relying on it for a claim.
Monitor the live location
Periodically test whether the original URL still resolves to materially equivalent content. A changed page may be a correction, so keep both the retained version and a new observation with its own date. Report link rot, redirects and substantive revisions separately. The earlier citation should continue to point to the version actually read.
Checks
- The retained representation contains the cited passage.
- A digest detects byte changes.
- Original location and observation date are available.
- Access and licence limits accompany the copy.
Pitfalls
- Treating a live URL as an immutable archive.
- Mistaking a digest for publisher authentication.
- Overwriting a copy when the live page changes.
A worked example
A fictional record cites a regulator's guidance page for a deadline. When the analyst read it, the page said that submissions were due by 30 June. The analyst saved the page as received, recorded the final URL after redirects, the time of retrieval, and a SHA-256 digest of the saved file, and noted the page's own "last updated" line.
Six months later the live page says 31 July. A reviewer asks which date applied when the analysis was written. The retained copy answers: 30 June, on a page last updated on a certain date, read on a certain day. The live page is a later observation, and the difference is a revision, which may or may not be a correction. Without the copy the analyst would have had only memory and a link that now says something else.
Variations and edge cases
Not everything can be kept. Licensed material, personal data and restricted documents carry conditions on copying and sharing. The method records what was permitted and what was done: a full copy held in a restricted area, an excerpt, or metadata only. Where only metadata is kept, the record says that independent verification requires access to the source, so no one assumes that every reader can check.
Dynamic pages are awkward. A page whose content depends on a query or a script may not be captured by saving the HTML. Capture what the reader saw, in a form that renders, and note the capture method. Public web archives can supplement a private copy and provide independent timestamps, and a protocol for time-based access to prior versions of web resources exists, but archive coverage is incomplete, so they do not replace the record's own copy.
Digests and what they prove
A cryptographic digest identifies a sequence of bytes. If a later file has the same digest, it is the same file; if it differs, something changed, which may be only a timestamp in the markup. Standards describe how to name content by its hash, and the secure hash algorithms are standardized. A digest does not say who published the document or whether it is true. Where authorship matters, the source's own signature, where it has one, is a separate check. For long-term keeping, preservation guidance such as the OAIS reference model describes what to store alongside the object so that it stays interpretable.
How the method is checked
Test retrieval, not storage. Pick citations at random, open the retained copy, confirm that the digest matches and that the cited passage is in it. Test also that the copy can be read years later: file formats, fonts and compression can age. Record the format and keep a rendering that does not depend on special software. Review the retention and access rules at set intervals, and do not delete copies that decisions depend on.
References
- RFC 6920. Naming Things with Hashes.
- NIST. FIPS 180-4: Secure Hash Standard (SHS).
- RFC 7089. HTTP Framework for Time-Based Access to Resource States: Memento.
- CCSDS 650.0-M-2 (ISO 14721). Reference Model for an Open Archival Information System (OAIS).
- Klein et al. (2014). Scholarly Context Not Found: One in Five Articles Suffers from Reference Rot. PLOS ONE.
Found an error on this page? See corrections for what counts as one, how to report it and the dated log.
Enquiry
Test this method on one entity.
Name an entity where "Retaining source copies" would be checked in practice, with its sources and its readers. We reply with what a first record would cover and what it would need from you.
Or write to [email protected] in your own words.