Last updated: Version 7.3 — Appendix C (June 08, 2026)

Scientific analysis based on the primary source: Mildner, S. (2026). Geodynamic Reinterpretation Model for Ptolemy’s Germania Magna: General Model Description, Cartometric Foundations, (v7.3). EarthArXiv (Preprint). https://doi.org/10.31223/X5KB51 (📥 Download NEW-v9.0-PDF) 

Disclaimer

This article presents a formal quantitative validation of the geodynamic reconstruction model introduced in the companion article. It does not constitute peer-reviewed research. All results are based on the publicly available Ptolemaic gazetteer of v7.1. The Saale-Unstrut Fragment Impact and related impact hypotheses remain working hypotheses not confirmed in the peer-reviewed impact-cratering literature.

Supplement to: Formal Out-of-Sample Blind Test and Extended Model Validation for Model v8

Last updated: v8[-Preview] (June 10, 2026)

Scientific analysis based on the primary source: Mildner, S. (2026). Geodynamic Reinterpretation Model for Ptolemy’s Germania Magna: General Model Description, Cartometric Foundations, (v8.0Preview). EarthArXiv (Preprint). https://doi.org/10.31223/X5KB51
(📥 Download NEW-v9.0-PDF) 

Builds upon: Mildner, S. (2025/2026). A new interpretation of Ptolemy's Germania Magna: Employing computer-assisted image distortion of a medieval map by Donnus Nicolaus Germanus to examine post-glacial geodynamics in Europe. EarthArXiv. https://doi.org/10.31223/X5313T
(📥 Download v5.0-PDF) (descriptive main publication)

Code and data archive: Zenodo, 10.5281/zenodo.10968193

1. Background and Version History

The companion article established that the kinematic block-deformation model (Model B) outperforms the affine baseline (Model A) by 29–49% in a genuine 70/30 out-of-sample blind test, with the G6 rotation prediction ($r_B=7.5\text{ km}$, estimated from a single training point) representing the strongest individual result. The parsimony argument in §14 of the primary preprint invokes the Akaike Information Criterion conceptually to motivate the significance of the blind-test result. This article presents the Extended Validation Suite underpinning the v8 Model Description.

Changes in v8[-Preview] relative to v7.3:

- The identification of S-C (Carrodunum) has been refined from the Kamenz–Spreetal/Nochten area to Bernsdorf (14.05°E, 51.38°N; Elster Cluster member). The cartometric residual $\mathrm{\Delta }\lambda =-70.0\text{ km}$ is unchanged. The Bernsdorf identification resolves a residual geographic ambiguity and is used as the primary reference for all v8 EC cluster statistics.
- As a consequence of the Bernsdorf calibration, the Model B residuals for the two blind-test EC points S6 and S7 shift marginally: $r_B(\text{S6})=31.8\text{ km}$ (was 28.5 km) and $r_B(\text{S7})=35.2\text{ km}$ (was 33.2 km). Both Wilcoxon results are identical to v7.3.
- The one-sample $t$-statistic for the EC cluster is updated to $t=-15.0$ under the v8 Bernsdorf identification; it remains far beyond any practical significance threshold.
- New: Falsification Test T39 — Moran's I spatial autocorrelation of the 22-point residual field (Section 10).
- New: Computational reproducibility appendix; full data and code archived at Zenodo (Section 13).