Skip to content

SEAD Environmental Archaeology Context

This handbook explains how to read SEAD as both spatial and temporal evidence without asking the data to say more than it does. The central skill is to separate five things that are easy to collapse: a site, the records linked beneath it, a grouped chronology feature, a derived site-level time envelope, and a map interaction.

Begin With The Observation Unit

One normalized SEAD feature represents one upstream site. A site can contain many sample groups, physical samples, analysis entities, dates, datasets, and references. Therefore:

A site point is an index into an evidence dossier, not an archaeological event and not a specimen.

This matters most when time enters the analysis. A 7000–10000 BP site envelope says that captured relations beneath the site establish that overall coverage. It does not say that every observation at the site has a 3,000-year duration.

Read A Site In Four Passes

1. Establish identity and geography

Read the SEAD site identifier, name, coordinate, country assignment, and source page. The identifier is the durable join back to the captured archive; the display name alone is not.

2. Inspect the evidence depth

Use popup rows and the raw archive to see whether the site has sample groups, physical samples, analysis entities, datasets, dating ranges, relative periods, or references. A high analysis-entity count means a richer captured relational neighbourhood, not automatically a stronger chronology.

3. Choose the temporal observation unit

Product Reader interpretation Numeric slider behavior
temporal-evidence feature one interval shared by identified linked chronology rows included when that interval overlaps
dated site summary the widest captured chronology envelope beneath one site useful for overview, too coarse for event-level reading
unresolved site inventory spatial identity without captured numeric chronology withheld from narrowed windows

4. Return to the intended claim

Ask whether the conclusion is spatial, temporal, or record-level. A nearby site can support a spatial-context statement even when its time is unresolved. A contemporaneity statement requires numeric interval evidence on both sides. A claim about a particular sample or proxy requires the linked record lineage, not only the site envelope.

flowchart TD
    Question["proposed statement"] --> Kind{"what kind of claim?"}
    Kind -->|spatial| Space["site identity + coordinate + distance rule"]
    Kind -->|temporal| Time["numeric bounds + overlap rule + uncertainty"]
    Kind -->|record-level| Record["sample/entity/dataset relation lineage"]
    Space --> Claim["qualified result"]
    Time --> Claim
    Record --> Claim

Work Through Three Examples

Agerod V: numeric interval and context

Agerod V (4237) has a normalized site envelope of 7000–10000 BP. The temporal-evidence layer also publishes the linked chronology as narrower features with source-record identifiers. If an atlas window is 8000–9000 BP, only chronology features whose own intervals overlap remain visible. Use the site envelope to understand total coverage and the temporal layer to ask which captured dates support the window.

Borgholm: source label without an interval

Borgholm (3776) retains Quaternary. The label is useful for interpreting the source and deciding what to inspect next. It is not converted into a BP range by this repository, because such a conversion would import a boundary that SEAD did not govern for this row. Borgholm is available in the optional site-inventory layer and absent from the temporal-evidence layer.

An unresolved site

An unresolved site still supplies a source identity and spatial context. Its null temporal fields are not zeroes. When a narrow time window hides the point, the correct reading is “temporal eligibility unknown,” not “site absent in this period.”

Understand The Current Denominators

Population Count
sites in bounding-box review 2,195
assigned four-country sites 2,069
Swedish sites with comparable chronology 370
Swedish sites without comparable chronology 1,555
Swedish numeric chronology features 8,172
chronology claims 25,109
comparable chronology claims 14,264
context-only chronology claims 10,144
temporal-contract-refused chronology claims 60
unresolved chronology claims 641
country-review rows 103
unassigned rows 23

The 25,109 chronology claims comprise 7,377 dating ranges, 10,057 relative period rows, 641 analysis-entity ages, 87 geochronology rows, and 6,947 dendrochronology rows. Exactly 14,264 claims are comparable; 10,144 remain context-only, 60 are explicitly refused by the nonnegative-BP contract, and 641 are unresolved. The atlas renders the comparable Swedish relations as 8,172 interval features without promoting the other claims.

Use The Timeline Correctly

The temporal-evidence layer is the comparison mode and is enabled by default. All 8,172 of its Swedish features are eligible for interval filtering. The site-inventory layer is optional: turn it on to inspect the full 2,069-site four-country spatial population. Claims without comparable intervals remain context-only, explicitly refused, or unresolved and cannot enter a narrowed time window.

stateDiagram-v2
    [*] --> FullExtent
    FullExtent: temporal evidence visible; site inventory optional
    FullExtent --> NarrowWindow: reader narrows BP range
    NarrowWindow: record-level interval overlaps visible
    NarrowWindow: unresolved site inventory withheld
    NarrowWindow --> FullExtent: reader restores full span

This behavior prevents two opposite errors: discarding valuable undated spatial context from the overview, and pretending that undated sites are present in every period.

Use The Governed Sweden Discovery Surface

For Sweden, the Nordic Atlas combines those two reading needs without erasing their distinction. Every one of the 1,925 assigned Swedish SEAD sites appears through either its linked numeric chronology features or one explicitly unresolved feature. The layer therefore supports both full-population spatial discovery and honest interval filtering.

Its rank is an evidence-readiness order. Chronology and bibliography coverage come first, followed by chronology breadth and represented record depth. The order does not measure archaeological importance or current activity. RAÄ density is displayed only as coarse context and cannot supply a site identity, date, or rank contribution.

See Sweden archaeology site discovery for the complete 1,925-site contract and its JSON, CSV, GeoJSON, and Markdown companions.

Compare With Other Evidence Families

For a SEAD–LANDCLIM–AADR comparison, perform three independent checks:

  1. Spatial eligibility: each record is admitted under its own geography and the distance rule is stated.
  2. Temporal eligibility: each compared member has numeric bounds and the overlap calculation is explicit.
  3. Semantic eligibility: the observation units are not treated as interchangeable.

An overlapping SEAD chronology feature, pollen sequence, and human-sample interval can motivate investigation. It does not prove that the people represented by the aDNA sample used the SEAD site or caused the pollen change.

Reuse Contract

A reusable SEAD-derived statement carries:

  • the SEAD site identifier and source page;
  • the coordinate and governed country-membership result;
  • the evidence role, normally environmental-archaeology context;
  • numeric bounds when present, including BP basis and uncertainty notes;
  • original and normalized period labels when present;
  • the temporal comparison posture;
  • the observation unit, explicitly “linked chronology interval” or “site envelope”; and
  • the product rule that admitted, withheld, or excluded the feature.

Do not replace this lineage with a point name, distance, or broad period label.

Access And Recovery Boundaries

The repository mirrors a large relational projection, not the complete SEAD database or browsing experience. The governed artifacts distinguish what is captured from what remains upstream:

  • data/sead/review/access_model.json records site-page and reference visibility;
  • data/sead/review/temporal_review.json records numeric and unresolved site postures;
  • data/sead/review/evidence_legibility_review.json records whether the captured representation is interpretable; and
  • data/sead/review/recovery_requirements.json names remaining gaps and the evidence that would satisfy them.

Recovery remains claim-led. A sample-level chronology needs sample-to-date lineage. A proxy claim needs dataset and analysis lineage. A literature claim needs the site-reference-to-bibliography chain. More site points cannot substitute for any of those relations.