Screen, archival storage and notebook. Illustrative scene.

GUIDE / ARCHIVES

Read a web archive without distorting the timeline

An archive helps recover an older trace. To use it, identify exactly what was observed: a resource, a capture and a passage at a specified time.

AI-generated illustration

01 / GUIDE / ARCHIVES

An observed trace, a bounded conclusion

Define the question before opening a calendar: are you seeking wording, an available document or a page change? Record relevant URL variants. Separate visible information from unknowns; a domain name proves neither its historical author nor continuity of its content.

Guides
Conceptual diagram of three resource states: T0, T1 and T2 observations are distinct. The intervals do not prove continuous presence.
Conceptual diagram of three resource states: T0, T1 and T2 observations are distinct. The intervals do not prove continuous presence.

02 / GUIDE / ARCHIVES

Six checks for a usable archive

These checks document an observation; they do not automatically certify evidence or reconstruct unobserved periods.

01

Retain both addresses

Record the original URL and the capture actually read. Recheck the destination after navigation: an archive may lead to another date or the live web.

Record to retain · Original URL, capture URL and final destination
02

Separate dates

Distinguish the date displayed in the document, capture and consultation. Do not turn the first recovered observation into a certain creation date.

Record to retain · Dates, known precision and retained wording
03

Identify the passage

Retain title, used fragment and context. An isolated sentence may omit a qualification, negation or applicable period.

Record to retain · Passage, title and immediate context
04

Content and resources

Inspect useful text, images and links separately. Replay does not guarantee that components or functions were preserved.

Record to retain · Examined components and missing elements
05

Comparable versions

Compare the same resource and passage across captures. Do not call a redirect, different file or display failure an editorial change.

Record to retain · Versions, observed difference and competing explanation
06

Proportional conclusion

An absent capture alone does not establish absent publication. State the actually observed interval and what cannot be established.

Record to retain · Search scope and explicit qualification

03 / GUIDE / ARCHIVES

A record enabling a second reading

Build a small review record before producing a timeline or historical narrative.

  1. 01

    Define the search

    Record the question, explored URLs, useful period and exclusions. Keep searches without results as part of the method.

  2. 02

    Read the selected capture

    Open the document, check its destination and locate the passage. Record visible information rather than completing unknown dates.

  3. 03

    Compare without filling gaps

    Compare observations, differences and omissions. Treat change as a hypothesis while another explanation remains plausible.

  4. 04

    Write and review

    Cite the capture, retain qualifications and have a reviewer repeat the route. Revisit it if a fuller resource is found.

04 / GUIDE / ARCHIVES

Four reading situations

The record depends on the question the trace must answer.

01

A sentence seems to have disappeared

Compare passages and destinations, then describe both observations. Do not infer an intention to remove content solely from the difference.

Read the dossier
02

A link opens another date

Record each destination and its capture separately; do not attribute all navigation to the starting page’s date.

Read the dossier
03

A capture is incomplete

Describe readable and missing elements. Narrow the conclusion if the missing component is essential to the claim.

Read the dossier
04

A trace must be shared

Prepare a reference with addresses, passage and dates; attach limitations so the recipient knows what still requires checking.

Read the dossier

A useful reference supports another check

Cite the precise page and version read. A portal title, search engine or access date does not replace the document date. Missing information stays missing; retain the uncertainty in your record.

Compose a public-source reference

06 / FAQ

Questions about archived traces

Does the first capture date publication?

It dates a recovered observation. A publication date requires separate evidence identified and interpreted in context.

Does a calendar gap prove absence?

No. Describe available observations and the search without treating the calendar as an exhaustive history.

Is a screenshot enough?

It helps show a passage but does not replace addresses, version and dates in the record.

Does a WARC file remove the need to review?

No. An archiving format does not resolve interpretation: retain relevant resources and check what the document supports.

How should one cite an incomplete capture?

Identify the readable passage and missing resources. After following each link, record the date and address actually displayed; do not present several captures as one state.

07 / Discuss a context

Document a trace before concluding

Describe the question, recovered versions and qualifications that could change the decision.

DOCUMENTED REFERENCE

Read an archive at record level

A page replay helps reading; archive metadata identifies the examined resource. Retain both levels when a conclusion depends on an earlier state.

Read an archive at record level
ReferenceValue or formWhat it identifies
WARC-Target-URIURITarget resource.
WARC-DateUTCStart of capture for this record (UTC).
WARC-Typeresponse / revisit / metadataRecord type.

Implication for the record

Also record HTTP status and payload availability. Do not treat a revisit record as a new full copy without examining its reference to an earlier capture. Archive time does not date the text’s authorship.

Primary documentation

Documentation consulted on .

DOCUMENTED REFERENCE

Memento: separate the requested date from the state returned

Informational RFC 7089, published in December 2013, describes time-based access to Web resources. The two dates below are its header examples; they illustrate protocol vocabulary, not evidence of an inspected capture.

Accept-Datetime2007-05-3120:35:00 GMTMemento-Datetime2007-05-3018:47:52 GMT≠
Memento: separate the requested date from the state returned
ReferenceValue or formWhat it identifies
Accept-DatetimeThu, 31 May 2007 20:35:00 GMTA client’s datetime preference sent to a TimeGate. It is neither a text publication date nor a guarantee that a state exists at that exact instant.
Memento-DatetimeWed, 30 May 2007 18:47:52 GMTThe datetime of the past state reflected in the response. Keep it separate from the requested preference and today’s consultation time.
rel="original"URI-RLinks to the original resource whose state was archived, not necessarily to a currently available file with the same content.
rel="timegate"URI-GLinks to the datetime-negotiation service. Its algorithm selects among known states; the protocol does not impose one universal nearest-state rule.
rel="timemap"URI-TLinks to states known to the server with their archival datetimes. A TimeMap can span several documents; it is not an exhaustive history of all publications.
Varyaccept-datetimeIndicates that selection can depend on datetime preference in the relevant responses. Do not assume it appears at every intermediate step.

Implication for the record

Record original and final URIs, requested date, actual Memento-Datetime and consultation. A returned state can precede or follow the preference depending on selection; a missing state does not prove non-publication. Check embedded resources too.

Is it possible to cite the requested date when the response gives another date?

Not as the date of the state read. Cite the returned state’s date and retain the request as a search parameter. If exact timing matters, explain the gap or seek other observations.

Primary documentation

Documentation consulted on .

DOCUMENTED REFERENCE

BagIt: verify a file collection beyond a ZIP archive

BagIt 1.0, described by Informational RFC 8493, organizes payloads, metadata and manifests. It separates file presence from checksum verification without certifying document truth.

tagmanifest-*.txtmanifest-*.txtdata/…metadata → manifest → payload

The tagmanifest checks the payload manifest, which checks files under data. Arrows show verification coverage, not creation chronology.

BagIt: verify a file collection beyond a ZIP archive
ReferenceValue or formWhat it identifies
bagit.txtBagIt-Version: 1.0bagit.txt contains exactly two ordered lines: version then tag-file encoding. Version 1.0 and the declared encoding identify the applicable reading rules.
data/opaque payload bytesThe data directory holds the payload. Each file is checked as opaque bytes; do not resave PDFs or normalize text before verifying a checksum.
manifest-sha256.txtevery payload file, onceEvery payload manifest must list every payload file exactly once. Paths use / and stay within data; directories themselves are not files to hash.
tagmanifest-sha256.txttag files · no tagmanifestA tagmanifest covers tag files and must include payload manifests. It must list neither other tagmanifests nor payload files: the two checking layers are distinct.
Payload-Oxumoctet count . file countPayload-Oxum combines total payload bytes and payload file count. This fast check can detect missing data but never replaces checksums when declaring validity.
STATUScomplete ≠ validA complete bag has required elements and conforming referenced files. Validity additionally requires successful verification of every checksum in all present payload and tag manifests.

Implication for the record

A valid bag may contain false, substantively incomplete or misattributed documents. Retain origin, collection date and method; a compressed container also has its own digest.

Do total size and file count suffice?

No. Different contents can have equal totals. Payload-Oxum is a preliminary optimization; validity requires prescribed checksum checks.

Primary documentation

Documentation consulted on .

DEEPER READING / PRACTICE

Treat a capture as a dated observation

A current page differs from an archived version. The archive may document a observed status, provided that it distinguishes capture, publication and availability of the related resources.

Updated on .

Priority checks

Keep the original URL, capture URL, timing and visible elements. Check the missing resources and compare several captures when the chronology counts.

Interpretation pitfalls

The first capture found is not necessarily the first publication. A archived page can also mix captured resources at different times.

Working record to retain

The archive report indicates visible state, date of capture, limits and comparison made. The conclusion is limited to the content actually preserved.

PRACTICAL CASE / WORKED EXAMPLE

A capture contains missing resources

The example below is pedagogical. It shows how to move from a document to a qualified proposal, preserving the context, the contradictory elements and the information that the research does not yet allow to establish.

Example context

A archived page retains the text but some images and interactions do not load. The reader seeks to establish the appearance and content of an old state.

Suggested approach

Keep URLs, timetables and visible elements. Compare other captures and check the available resources. Distinguish absence in capture and absence on the site at the time.

Conclusion and next step

The report describes the state actually preserved. The conclusions on the missing elements remain limited by the archive coverage and the consultation conditions.

Questions for deeper practice

Does a capture prove the date of publication?

It documented a state observed at the date of capture. The publication may be previous and some resources may have their own timeliness; these distinctions must be preserved.

DETAILED WORKSHOP / APPLICATION

Qualification of the preserved state and the missing elements

  1. 01 / Prepare the necessary information

    Prepare the original URL, capture, timing and historical question. Identify the resources and interactions necessary to understand the content sought.

  2. 02 / Carry out the approach

    Examine text, images and links that are actually available. Compare other captures if the chronology counts. Keep the differences and distinguish page capture and timing from associated resources.

  3. 03 / Examine an alternative situation

    If an item is missing in the archive, do not conclude that it was missing on the original site. Part conservation limits what observation allows to establish.

  4. 04 / Decide on the next step

    The report is useful when the visible state and its limits are explained. The date of capture does not automatically become a date of first publication.

  5. 05 / Hand over the verification records

    Transmit captures, resources and qualified conclusion. The new occurrences found can specify the history without deleting the limits of the previous pieces.

ANNOTATED WORKING DOCUMENT / REVIEW GRID

State report captured with conservation limits

The report combines the original URL, capture, timing and visible elements. It distinguishes observed status, date of publication and missing or captured resources separately.

On a small screen, scroll the grid horizontally to read all three columns.

Example situations to assess
ObservationSupported interpretationNext step
The content is visible in the capture.The archive documented this status at the date of capture.Keep the reference and specify the scope of the observation.
A resource is not charged.The absence may arise from the conservation and not from the original site.Compare the captures and announce the limit.
The first capture is used as the date of creation.The time of publication is not determined by this single piece.Search for additional information and separate date roles.

Reviewing the working document

The final review describes what was actually preserved. Flag missing functions and resources so that the reconstructed historical account does not exceed what the available archive supports.

REPORTING / EXAMPLE WORDING

Presentation of results: Treat a capture as a dated observation

Point to clarify

The capture indicates a preserved state and does not automatically restore the entire original site.

Before handing over

Do not use the absence in the archive as a general proof of historical absence.

Example of a qualified conclusion

The visible content is described at the date of capture. Lack of resources and separate temporals remain indicated as preservation limits.