Crusades data fields · Terra Chartarum

Why these fields exist

The Crusades flagship makes three arguments, and each rests on a distinction that is easy to lose in a spreadsheet.

Stage, and why it has no position

A stage of Matthew Paris’s itinerary is a cell in a strip diagram. The itinerary is a vertical sequence of towns with day-marks drawn between them, and it has no projection whatsoever: no scale, no orientation, no coordinates.

So a stage record has nowhere to put a position, and the field that looks like one is called modernReference. That is the coordinate of the modern town the stage names, supplied so the diagram can be compared with a map. It is never something the manuscript gives.

depictedDays is the number of day-marks the diagram draws before a stage. It is the manuscript’s claim about the journey, not a measurement of it, and nothing in this project converts it into a travel time.

The six states of a campaign

A single route line from Venice to Constantinople would destroy six different kinds of claim. They are kept as separate records:

`stateKind`	What it is
`intended_destination`	Where the crusade contracted to go
`negotiated_diversion`	Where it agreed instead, and on what terms
`travelled_route`	Where the fleet actually went
`attack`	An assault at a place, on a date
`partition_claim`	Territory a treaty assigned
`durable_control`	Territory somebody actually held

held carries the difference the whole proof turns on: held, claimed_not_held, or not_applicable. The Partitio Romaniae assigned an empire among people who held very little of it, and a record that could not say so would republish the document’s wishful thinking as geography.

Three states carry no geometry at all, and that is a finding rather than a gap. The intended destination was a clause in a contract; the partition’s boundaries are disputed in the scholarship; and what was durably held is contested throughout and already modelled properly by the Venetian possessions layer.

The seven registers of a city

Jerusalem is not one kind of object, and a table that held it as one row would have made the essay’s argument impossible to state. roleKind carries which kind of claim a record is:

`roleKind`	What it asserts	Drawn
`contemporary_narrative`	What a witness says happened, and when	no
`sacred_centre`	The world is arranged around this city	no
`pilgrimage_destination`	What the journey was for	no
`textual_construct`	The land has an order, and this is it	no
`cartographic_construct`	The land ruled into squares	no
`network_node`	Something passed through here	yes
`cartographic_memory`	The centring outlived its subject	no

contemporary_narrative is also the only register with Muslim-authored sources behind it: cru-jer-ibn-al-qalanisi (via Gibb’s translation, for the loss of Jerusalem in 1099) and cru-jer-baha-al-din (via Conder and Wilson’s translation, for its recovery in 1187) - the other six registers run on Latin sources only, which is the gap docs/crusades/islamic-perspective-scoping.md records.

Only the ports are drawn. The middle of a mappa mundi is not at 31.78° N; it is in the middle, and a pin there would convert a claim about what the world is arranged around into a claim about where a city is.

catalogueObjectId joins a record to an object in this site’s own collection. sourceId is null for exactly one register: later cartographic memory cites no source in the audit, because an early-modern map that centres Jerusalem is a witness to the early modern period and nothing in this corpus may use it as a witness to the twelfth century.

Geometry provenance

editorial_generalisation, modern_reference, not_spatial.

There is no value for a documented track, because no source in this corpus gives one. Both route lines are generalisations between named ports, drawn so the shape of a campaign is legible, and dashed so nobody measures them.

Review state

raw, normalized, reviewed, approved, published. Everything here is raw except two records, both normalized: cru-jer-qalanisi-1099 (pp. 47-48 of Gibb’s translation of Ibn al-Qalanisi) and cru-jer-shaddad-1187 (pp. 118-120 of Conder and Wilson’s translation of Baha’ al-Din) - the only folios actually read anywhere in this corpus. No witness is cleared for publication, which is what the source audit records and what the release manifest asserts.