Research guide

How Do We Test an Archaeological Claim?

Start with the exact claim. Identifying material, dating a sample, establishing its context, and interpreting its use are separate steps in an archaeological argument.

By Documentify TV · Sources reviewed 18 November 2026

Start with a sentence you can test

“Something important was found” combines several questions. Write the claim more precisely: a particular sample is wood; its surface was worked; it is ancient; it belongs to a structure. Each statement needs evidence suited to it.

A result can strengthen one statement while leaving the others unresolved. This guide shows how to follow that chain without turning a promising sample into a complete historical identification.

Six questions between a fragment and an identification

These are questions to investigate, not a score or a rule that every find must pass the same tests.
Claim being testedEvidence that could address itConclusion still left open
1. What is the material?Documented examination of the sample's composition or diagnostic structure.Identifying wood does not establish its age or purpose.
2. Was it deliberately worked?Recorded traces assessed against manufacturing, use, damage, and natural alternatives.A worked surface does not identify a particular tool or building.
3. What does the date measure?A suitable dating method, identified sample, laboratory documentation, and stated uncertainty.A sample date does not automatically date its burial or reuse.
4. How did it reach this context?Recovery records, stratigraphy, sample links, and investigation of disturbance or movement.Depth alone cannot establish age or original placement.
5. Does it belong to a structure?Relationships between elements, construction details, and a documented site context.One fragment cannot reconstruct the entire structure.
6. Which structure or historical event?Evidence that distinguishes the proposed identification from plausible alternatives.A resemblance or compatible age alone does not uniquely identify a named vessel or event.

A radar image and a laboratory report do different jobs

Historic England's explanation of geophysical methods describes ground-penetrating radar as recording reflections from the subsurface. Such observations can help select places to investigate. They are not a material identification obtained from a recovered sample.

A useful comparison asks whether the sampled position corresponds to the mapped anomaly, what the surrounding layers contain, and whether another explanation also fits. An image and a fragment should not be treated as one continuous structure until their relationship is demonstrated.

Compare what the evidence actually supports

Different cases illustrate different gaps between observation and interpretation.
CaseThe useful distinctionFollow-up question
Durupınar drillingThe team describes recovered fragments as charred wood. That attribution is separate from a published laboratory identification.Can a report be linked to specific samples and recovery contexts, and what does it establish?
Abric Romaní woodworkingEvidence of deliberate shaping and the proposed weapon function are separate parts of the interpretation.Which observations would distinguish the suggested use from alternatives?
Later settlement at AkéA dated domestic context supports a local occupation history; its scope matters.How many contexts support the proposed sequence, and how far can that sequence be extended?

Read the claimant-issued recovery announcement, IPHES woodworking account, and Aké study for those distinct source records. The linked case files provide the fuller assessments.

Suppose a lab confirms ancient charcoal

Hypothetical example, not a reported Durupınar result: a laboratory identifies charcoal and supplies an ancient date for a sample.

Which parts of the claim would that change?

It would add evidence about the tested material and sampled carbon. We would still need to connect that sample to its recovery record, examine whether it was worked, and establish any relationship to construction.

Historic England's dating guidance explains the old-wood issue: carbon from an inner tree ring can predate the tree's felling. Archaeological association and sample selection therefore matter alongside the numerical date.

A careful description would say what the identified sample establishes and name the remaining questions. It would not treat a date as a vessel identification.

A sample report you can trace

Use these fields as a reading worksheet when a new result appears. A missing field is a question to raise, rather than automatic evidence that a finding is false.

Sample identity and origin
Record the field and laboratory identifiers, material tested, recovery position, layer or borehole interval, and the link between those records.
Method and result
Record who performed the analysis, the method, pretreatment where relevant, and the result with its stated uncertainty.
Association with the proposed event
Ask why this sample dates or identifies the activity of interest, rather than older material, later intrusion, or reuse.
Interpretation and alternatives
Separate the measurements from the author's interpretation. Note which alternatives were examined and which questions remain open.

For drilling comparisons, match positions and layers before treating two accounts as contradictory. Separate boreholes need not encounter identical material.

Does a new publication warrant an update?

A new headline links to the same older paper

It may help people discover the case, but its publication date does not create a new scientific result. Check whether it adds evidence beyond the earlier source.

A new report identifies or dates a documented sample

Review its connection to the case and explain exactly which conclusion changes. Preserve the original article date and add a dated account of the new evidence.

A correction changes a method, sample, or conclusion

Read the correction alongside the original work. Explain its consequences rather than changing a status label without context.

Documentify's evidence tracker separates reviewed developments from routine source checks. A monitoring alert is a prompt to investigate.

Sources and next reading

← All research guides · Evidence update tracker