No, the Sumerian King List Doesn't Encode Ancient Climate Disasters

The eight pre-flood kings of the Sumerian King List reign for absurd, suspiciously tidy spans — mostly integer multiples of 3,600 years, all multiples of 600, summing to 241,200 years. A fringe reading treats those numbers as a garbled memory of real prehistory, with each reign boundary marking an eruption, impact, sea-level swing, or climate shift. The author builds an explicit test of that claim: rescale the list to a fixed 241.2 ka span, anchor the flood boundary to 11.6 ka BP (near the Younger Dryas termination, a choice the text itself does not supply), and check whether the resulting boundaries land on dated Quaternary events.

The scoring uses a Gaussian kernel proximity measure so nearby events count fully and distant ones fade out, avoiding a hard cutoff. Significance comes from an exhaustive permutation test — every possible reign ordering is evaluated rather than Monte Carlo sampled — with q-values correcting for multiple comparisons. Nothing holds up. The primary paleoclimate catalog gives p = 0.35 (q = 0.62); widening to 103 events yields p = 0.148; the single best cherry-picked hit, in a ‘catastrophic’ catalog, reaches p = 0.021 but collapses to q = 0.222 once the search freedom is accounted for. Parallel runs on Biblical patriarch ages, a Manethonian fragment, and early Kish rulers show any old chronology throws off comparable coincidences.

The real value here is methodological rather than archaeological. The author pre-commits to what would count as evidence, flags that the Younger Dryas match is baked into the anchor and contributes nothing to the internal test, and corrects for the freedom to tune anchor, catalog, and bandwidth after seeing the data. An interactive explorer (p-values precomputed in Rust, anchor sweeps recomputed in-browser) lets readers probe the sensitivity themselves. For a technical audience it’s a clean case study in how tunable knobs manufacture apparent patterns — and how disciplined null models and multiple-comparison correction keep you from fooling yourself.