Approach | mapping al-shawwām

Methodological Foundations

Every location, boundary, and spatial relationship in the platform derives from documentary evidence: travel accounts, administrative records, geographic dictionaries, or scholarly secondary sources. We do not manufacture coordinates. Where the platform derives anything beyond what a source states, the derivation is a rule anyone can read, and the result is marked as derived.

This produces an intentionally incomplete map that reflects the uneven documentation of historical geography. Some regions (like major trade routes) are densely described; others appear sparse not because they were uninhabited but because they were less frequently documented by the sources we currently access.

From Text to Map: The Process

1. Source Selection

Primary sources — travel accounts, administrative records, geographic dictionaries — and authoritative scholarship are weighed for reliability and geographic specificity before anything is extracted from them.

2. Transcription

Print and manuscript pages are transcribed by recognition models checked against pages a human transcribed by hand. A page's condition — damage, marginalia, column layout — travels with its text.

3. Extraction, Cross-Checked

Each entry is read several ways — strictly, generously, in context — and a synthesis keeps only what survives the comparison. No single reading is trusted on its own.

4. Verbatim Capture

What a source asserts is recorded as the source wrote it: names unstandardized, dates as printed, affiliations only where the text states them. Where the source is silent, the record stays empty.

5. Anchoring & Validation

Places are matched to a curated gazetteer and administrative hierarchies, and every record must trace back to its source citation. Failed matches are flagged rather than silently repaired.

6. Interpretation, In One Place

Reading a historical claim into modern terms — including which modern polity a historical place falls within — happens in a single documented, revisable step, kept apart from capture so each can be audited on its own.

Followed end to end, the process looks like this: a page is read, captured verbatim, matched to the gazetteer, and rendered as a marker with its citation attached. A scanned map runs the same discipline in parallel — georeferenced, its boundary traced and dated — and the two arrive in the same database, where a marker, a boundary, and a citation can be read together. The Data Pipeline page walks both chains, and shows where they meet.

Handling Uncertainty

Historical geographic data is inherently uncertain. Place names change, boundaries shift, sources contradict each other, and coordinates are often absent. Rather than hide this uncertainty, we make it explicit:

Explicit derivations — Where the platform infers, as when tracing a historical district to its modern successor, the inference is late, documented, and reversible
Multiple attestations — When sources disagree, we present all versions with citations
Temporal precision — We distinguish between "18th century" and "September 1742" level temporal specificity
Spatial precision — Point locations vs. regional descriptions vs. route segments have different geometric representations

Arabic Text Challenges

Working with Arabic historical sources introduces specific methodological challenges:

Orthographic variation — Standardizing spellings across manuscripts and printed texts
Transliteration — Following consistent systems (typically IJMES) while respecting local pronunciation
Name variants — Tracking how places are referenced differently in Arabic vs. Ottoman Turkish vs. European sources
OCR limitations — Historical Arabic typography creates unique challenges for text extraction

Arabic text is treated as a first-class problem, not an afterthought: normalization and name matching are built into the pipeline, and the original forms always survive them.

Ethical Considerations

Historical geography is never neutral. The sources we access, the places we prioritize, and the boundaries we draw all reflect power structures, both historical and contemporary. We acknowledge:

Archive bias toward state records and elite perspectives
Geographic bias toward trade routes and administrative centers
Linguistic bias toward Arabic and European-language sources
Contemporary political implications of historical boundary visualization

While we strive for transparency about these limitations, we recognize that simply documenting bias does not eliminate it. This remains an ongoing methodological challenge.