The Ground Beneath the Silence: Persistent Homology and the Test of the Complement Set on the Isthmus of Panama

The complement set, the formally defined space of what was knowable at a given historical moment but was not recorded, has been proposed as a topological object whose shape is measurable via persistent homology. That claim has not been tested on the cartographic corpus it was built to analyze. This article tests it. Two maps of the Isthmus of Panama, Joan Blaeu’s Terra Firma et Novum Regnum Granatense et Popayan (1662) and Herman Moll’s A Map of the Isthmus of Darien & Bay of Panama (1697), are georeferenced using ArcGIS control points, and sixty-eight feature points (settlements, fortifications, hydrographic annotations, coastline markers, legal inscriptions) are extracted by manual annotation. Persistent homology is computed on the resulting point cloud using a Vietoris-Rips filtration projected to UTM coordinates. Five H1 cycles are detected in the combined corpus, the most persistent bounding a region consistent with the Darien interior. A null model comparison against 1,000 bootstrap iterations of density-matched and uniform random point processes shows that the two-map corpus does not reject the null: 45 feature points distributed across approximately 400 × 250 km of the Isthmus produce interior holes of comparable scale under any spatial configuration. The paper reports this as a methodological finding: empirical scaling from the null simulation estimates that approximately 150 to 200 feature points within the Isthmus bounding box, achievable from 20 to 35 maps at current extraction rates, would be required to bring the 95th percentile of the null distribution below the observed persistence, giving the test discriminating power. This motivates the full-corpus analysis. Independently, the Blaeu and Moll persistence diagrams are compared directly: the Blaeu shows a single dominant H1 cycle (persistence 31.7 km), the Moll shows three cycles (largest 17.3 km), and the Wasserstein-1 distance between the two diagrams is 50.2 km, a descriptive difference in topological structure across sixty years and two cartographic traditions, not tested against a null model. The article specifies protocols for LiDAR ground-truth validation and for a knowledge-graph extension integrating non-state sources, and argues that decolonizing the archive is not only a political commitment but a topological operation whose effect on the persistence diagram is formally quantifiable.