Independent Monte Carlo Verification of Paleopole Orientation Clustering in the Buildreps Database
Year: 2026
Type: report
Abstract: Independent Monte Carlo Verification of Paleopole Orientation Clustering in the Buildreps Database — Version 3.1 This is a pre-registered, independent statistical verification of orientation-clustering claims associated with the Buildreps database of ~1,159 ancient structures (994 in-range). It tests one specific claim: whether the structures' great-circle orientations, intersected with the 47°W meridian, cluster at five proposed former-pole latitudes more than chance. Version 2 (1 June 2026) reported apparent clustering at Pole II (76°N) and Pole III (72.2°N) surviving all four pre-registered and exploratory null models, and characterised it as robust. Post-publication review identified two untested assumptions behind that result: that the five target latitudes were fixed before the data were seen (they were partly identified from the orientation data), and that the 994 structures are independent observations (many belong to dense single-tradition spatial clusters). With both controls applied, the clustering at Poles II and III is not distinguishable from chance — under the assumption-free conditional null with a latitude look-elsewhere correction, the corrected p-values are 0.999 and 0.90; under a spatial-cluster null collapsing nearby structures to one unit each, both return p = 1.0 across linkage thresholds from 25 to 100 km. The 119 site-level contributors to the Pole III window reduce to 25 independent clusters, the largest a single 29-site Mesoamerican cluster in the northern Yucatán. The pre-registered aggregate test statistic, formally significant at p = 0.0001 under the unconditional null, was shown in version 2 to be a hemisphere-selection artifact; that finding is unchanged. Version 3.1 adds three further analyses in response to the data owner's 5 and 8 June 2026 follow-up letters (preserved verbatim as Appendices D and E). A direct simulation of the data owner's own stated peak-finding rule (script 11) reproduces his poles on the observed data but finds no more than chance under realistic nulls; his published high-confidence figures derive from a uniform baseline the great-circle geometry does not support (his per-degree binomial collapses from ~10⁻⁹ to ~0.03 once the geometric baseline is used). A hemisphere-preserving permutation null with a circularity diagnostic (scripts 12–13) finds that the East/West orientation asymmetry he cites is ~96% reproducible from the latitude clustering plus geometry alone, so preserving it cannot serve as independent evidence for a pole. The analytic conclusion is unchanged. Corrected conclusion: No proposed pole shows orientation clustering distinguishable from chance once look-elsewhere multiplicity and spatial non-independence are controlled. One question is disclosed as unresolvable from available evidence: whether the dominant Yucatán cluster represents independent structures of different periods or one tradition, which would require model-independent (radiocarbon, ceramic, epigraphic) dates that do not exist at the needed precision. The conclusion stands under the stated assumption that spatial proximity proxies dependence for that cluster. Scope: This addresses only the statistical clustering claim. It does not test the broader geophysical model (former positions of Earth's rotational axis), which an orientation-clustering analysis cannot evaluate either way; it removes the clustering signal that had been offered as evidence for it. The AI-assisted post-publication review (Claude Opus 4.8, Anthropic) is disclosed in the acknowledgments; the author ran all analyses, verified outputs, and is responsible for the conclusions. Links:- Pre-registration: [10.5281/zenodo.20258204]- Version 2: [10.5281/zenodo.20474028] - Version 3: [10.5281/zenodo.20546301] (superseded)- Version 3.1: [10.5281/zenodo.20593690] (this version)- Repository: GitHub Repo File: `results_v1.0.pdf`SHA-256: `82fea07d4ed8141ecdbe46fbd655e100ad4b4b4bc63e10c7dda72287c2f5ef6e`
Author: Gherbi, Salah-Eddin
Language: English
Subfield: Archeology
Field: Arts and Humanities
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7161490566
Cites:
Cited by:
Related to: