Snake Cult · Research notes · 19 September 2026

What the numbers establish

Evidence packet for Contra d’Huy on the African Root of World Mythology. The root comparisons reuse the project’s independently audited September 2026 analysis. The motif counts use a pinned public catalogue; the archaeological count below audits the Bullroarer Atlas’s European Ice Age corpus.

Three comparisons on the same myth matrix

D’Huy’s 2017 Appendix A prints 13 regional profiles and 28 binary columns. All 364 cells were checked against the source. We preserve the printed block despite a conflict between the text’s stated removal of F45A and the block’s declared 28 columns; no uncertain trait label is silently repaired.

A two-state continuous-time gain–loss model estimates one equilibrium presence frequency and a separate ancestral presence frequency. The latter supplies direction: at equilibrium, rerooting is unidentifiable. See Klopfstein et al. (2015). This is a model of motif change, not a clock or a model of ancient migration.

The unrooted connections match the independently transcribed Figure 1. Each of 23 canonical root edges received 122 starts, combining equal 61-start banks from two representations of that topology. All branch lengths and frequency parameters were refitted. L-BFGS-B used analytic gradients; scalar likelihood and gradient checks were performed in the original analysis. Branch lengths were bounded between exp(−9) and 8; frequency logits between −8 and 8. Several optima are boundary-adjacent.

Each difference compares roots within the same row. Raw likelihoods across rows do not compare model quality because the observation treatments differ. East Africa ranks 23rd in each scan. A South American edge wins the two selection-corrected scans; the exact leading edges are close. No exhaustive joint search over every topology was performed. No matched power calibration specific to this matrix establishes recovery of a true historical root.

These fits show that the chosen African root is not uniquely required by the supplied observations. They do not establish a Balkan homeland, an Ice Age date, or a probability that any geographical origin is correct. Modern regional tips are not ancient populations.

Download all 69 root scores, named comparisons, fit policy, and source checksums · Download the audited printed matrix. The score export preserves the accepted profile hashes and relative source paths. It is an audit summary, not a standalone rerun of the original optimizer.

Counting recorded motifs

The source is Berezkin’s query-engine catalogue at commit 5d055aa (16 May 2023): 926 traditions × 2,138 motif columns. Each count is the sum of a row’s binary motif cells. We do not use the supplied sum metadata, which disagrees with the matrix in 71 rows.

The comparison group includes North Africa and Madagascar. It is not a Europe-only comparator. The exported JSON includes every source area, so readers can inspect the wider distribution rather than only the two selected summaries.

These are unweighted descriptive counts of recorded recurrent motifs. A zero is an absence from this catalogue, not proof of absence from a tradition. Uneven documentation, variable-sized tradition units, selection of recurrent motifs, and differences in cultural history all affect the counts. Row richness is not a measurement of deep divergence between cultural lineages. No significance test, cultural ranking, or homeland inference is attached to these descriptive statistics.

Reproduction script · All counts and source checksum. Download the pinned source, then run python3 count-motifs.py berezkin_new.csv. The script uses Python’s standard library. The input is tab-delimited despite its .csv extension.

Counting the European Ice Age corpus

The Atlas snapshot of 19 September 2026 contains 24 normally visible European Ice Age archaeological records representing at least 45 specimens. A record can describe one object or several objects from a site; counting records alone understates the material corpus. The 13 records corresponding to Barandiarán’s inventory account for 34 specimens. Eleven further records add at least one specimen each.

This count measures the archaeological comparison corpus, including pieces with alternative functional identifications; it does not count successful acoustic tests of 45 original objects. Barandiarán’s six-site concluding category contributes 19 of these specimens; it is the subset shown in the article’s map. The Atlas-wide total does not add those 19 again. Multiple reports of the same Lalinde object are merged; candidate-only records, the rejected Pin Hole attribution, and later archaeological periods are excluded.

Twenty of the 24 records, representing at least 41 specimens, have date ranges at or after approximately 19,000 BP (17,000 BC), the end-of-LGM threshold used here. The dates below retain their source ranges; they are archaeological contexts or cultural assignments rather than 45 direct radiocarbon determinations.

Grouped specimen counts follow Barandiarán’s 2012 itemized descriptions, pp. 314–321, checked against his 2015 synthesis, pp. 153–154. The linked Atlas records supply the other objects and source trails. Download the count, record identifiers and source checksums.

Population history

The article reviews published genomic reconstructions; it does not claim a new genomic inference. Schiffels and Durbin (2014) report effective-size expansion approximately 30–10 ka in CEU, TSI, GIH, and CHB, following a much older non-African bottleneck. Those samples are not every Eurasian population, and effective size is not local density.

Article source quotations supplied by the author’s request were checked against the 2017 paper. Interpretive claims about Europe as a source of the ritual complex are the essay’s argument, not outputs of these numerical comparisons.