A Pre-Registered Configuration Test for the Göbekli Tepe Pillar 43 "Date Stamp"

Abstract

Two published readings assign the animal carvings of Göbekli Tepe's Pillar 43 to constellations and conclude that the pillar encodes a date connected to the Younger Dryas event. This paper audits the published probability and then runs the spatial test both sides have pointed to in print, with the analysis plan (metric, null distributions, decision rules and the author's own prior) locked and hash-stamped before any coordinate was extracted. The audit. The published figure of "less than one in 5 million" reproduces exactly and rests on the authors' own fit judgments, seven of eight of which sit at the floor of their scale, so no possible disagreement can lower it. A mildly neutral judge moves it by three orders of magnitude; acknowledged but unpriced selection factors add another; the ±250-year precision rests on a sub-feature choice where the constellation-level window is ±1,130 years; and the link to the Younger Dryas specifically is priced at 8 % in the paper's own main text. The pre-registered tests. Neither published mapping passes. Sweatman's assignment ranks in the middle of all 40,320 ways of assigning his own constellations to his own carvings (p = 0.55); Burley's ranks near the middle of its null (p = 0.38); every one of 10,000 random near-ecliptic constellation sets fits the pillar better than either. An opportunity audit on Pillar 56, a denser pillar to which no date has been fitted, shows that 0.42 % of honestly-paired random 8-figure subsets clear this paper's own "weak evidence" threshold using no animal identity at all, purely from the freedom to choose which figures to compare. Finally the same locked machinery is pointed at a different target using the same locked digitization of seven of its twelve carvings, reduced to the number of sites by a rule fixed in advance, against the Pre-Pottery Neolithic sites of the surrounding region: it too returns no configuration evidence (d = 0.64, p = 0.12), though it fits better than either published sky mapping, while the best of the 5,040 relabelings of the same seven sites reaches d = 0.081, which is what an unregistered version of that run would have reported. What this establishes. The spatial component of both published readings is indistinguishable from arbitrary, so whatever support they have rests entirely on animal-identity judgments, which the two authors resolve in mutually incompatible ways. Two contradictory instruments have been used to reach the same conclusion, which locates the conclusion upstream of both. The result does not bear on the Younger Dryas impact hypothesis itself, which is a geological question, nor on non-spatial readings of the iconography. A separate finding is reported along the way: the positional record for the region is in worse condition than the method. One public-database coordinate for Sefer Tepe reproduces Göbekli Tepe's own to four decimal places, two further entries fall hundreds of kilometres away, and Karahan Tepe appears in the same journal at two published positions about 19 km apart. All three pre-registrations, the SHA-256 lock, every run script, the digitizations, the gazetteer and the raw outputs are included so that each number regenerates. Version 2 (2 September 2026). No result, table, figure or number changed. The prose is repunctuated: 56 em dashes are unpacked into commas, colons, parentheses and separate sentences, here and in the paper. The code and data archive is unchanged from version 1. Version 3 (18 September 2026). No result, table, figure or number changed. Every statement in the paper that quotes or characterises someone else's work was checked line by line against the source texts, and the corrections are listed in a new section, "Corrections in this version". The substantive ones: three readings attributed to Burley (2013) appear only in his 2017 comment, and the era it leaves open is stated definitely in 2013; Pillar 56 was described as never examined, whereas Sweatman and Tsikritsis read it for Taurid symbolism in 2017 and deferred a fuller reading, so the claim is narrowed to the absence of a date-stamp mapping; one quotation had been assembled from two clauses of a source sentence; the excavators' objection concerns the headless man, not the vulture's identification; Burley suggested a quantitative spatial analysis rather than calling for one; the 8 % figure is in its source's main text, not its appendix; and the Enclosure H excavation report, previously described as unconsulted, has been read and is now cited. Neither test is affected: the configuration test evaluates the 2017 comment's own overlay, and the opportunity audit uses no animal identity at all. The code and data archive is unchanged from version 1.