Fifteen attempts to draw the world, from a Babylonian clay tablet to the triangulated survey of France - laid on one chart, measured by six bearings, and walked by three tours.
Old maps fail badly when judged by one yardstick. A mappa mundi is not a broken road map; a portolan is not a failed cosmology. Each chart here is taken on six independent bearings, scored from 0 to 8 points - one for each principal wind of the compass.
Mathematical fidelity to the shape of the earth: projection, coordinates, scale, the discipline of where things actually are.
How directly the map rests on observation - soundings, itineraries, astronomical sights - rather than inherited authority and rumour.
The weight of symbolic order: theology, ideology, the centre of the world placed where doctrine, not geography, demands it.
How well the map served the task it was actually made for - sailing, taxing, preaching, governing. Accuracy and usefulness are not the same bearing.
The span of world the map dares to hold - from one sea or one kingdom to the whole inhabited earth, the ecumene.
Craft: the quality of drafting, pigment, lettering and engraving - the labour that made the map worth keeping for centuries.
Every map plotted as a position: its age along the horizontal, any chosen bearing on the vertical. Change the bearing to re-sound the whole corpus; select any star to open its dossier.
Note the sawtooth around 1300: the Hereford mappa mundi and the Carte Pisane were drawn within a decade of each other, one near the floor of Witness and one near its ceiling. Accuracy did not simply decline and recover - traditions with different purposes ran in parallel. The ochre dashes trailing to a hollow ring mark transmission: where the surviving artifact is centuries younger than its content, the timeline of what was drawn is not the timeline of what survives.
The selected chart, its wind-rose of scores, a second chart laid over it - and the sources each score must answer to. The History of Cartography volumes cited below are free in full from Chicago.
One chart on this site can be measured without ever seeing it: Ptolemy's, because his coordinates survive as text. For the other nine candidates on the docket the coordinates live on the artifact, not in text - so the bench takes field surveys: pick control points on any digitization, paste them below, and the same least-squares Helmert fit runs, with displacement vectors and a measurable figure.
One control point per line: name, mapX, mapY, trueLon, trueLat - or paste a complete Allmaps georeference annotation (JSON) made at editor.allmaps.org against any digitization linked in the dossiers. Mind each docket caveat below: a projected map wants its own projection inverted before a single global fit means much.
On the Ptolemy dataset - twenty-two positions after the Geographia tradition, rounded, with manuscript families disagreeing by fractions of a degree - the fit recovers the famous diagnosis in one number: Ptolemy's east–west frame runs roughly half again too long, so the whole inhabited world must be shrunk to sit on the ground. Strike Babylon and the Atlantic ports and watch the Mediterranean core tighten - the error is not noise, it is structure, accumulated along his longest itineraries. Longitude was the unsolved problem; latitude, fixed by the sun, behaves far better.
For every other map here, the same measurement needs control points picked on the artifact itself - MapAnalyst and Allmaps exist for exactly that. The literature's verdicts: portolan charts fit astonishingly tightly but sit rotated some 8–10° anticlockwise, the fossil of uncorrected magnetic declination; the mappae mundi resist the fit altogether, which is not their failure but their genre.
Ten charts admit measurement on the artifact itself: seven take a single global fit, three only piecewise. Each carries the caveat that keeps its number honest.