The alphabet that refused to grow

Claim (verbatim)

Genevieve von Petzinger's survey of the geometric (non-figurative) marks in Ice Age Europe - the core of her University of Victoria thesis and the popular distillation The First Signs (2016) - reduced the whole engraved and painted repertoire of the Upper Palaeolithic to roughly 32 recurring sign types recorded across some 370 decorated sites in France and beyond. The striking datum is not the number but its behaviour over 25,000 years: most of the types are already present near the beginning, in the Aurignacian, and the inventory barely enlarges thereafter. If these marks were free decoration, the count of distinct types should keep climbing as more sites and more millennia are added, the way an open-ended ornamental vocabulary does; a rarefaction curve would never flatten. If instead they are a bounded, conventional sign-system - a closed set held in cultural memory - the accumulation curve must saturate, and the most recent (Magdalenian) sites should contribute almost no new types that the earliest sites did not already carry. The missing knowledge is what the signs meant; but their systematicity is measurable without decipherment, from the shape of the discovery curve alone. That curve is the fingerprint that separates a lexicon from a doodle-book, and it is countable from a site-by-sign incidence table.

Prediction clause (verbatim)

Prediction: reconstructing the site x sign-type incidence matrix from von Petzinger's published corpus and ordering sites by archaeological period, a species-accumulation (rarefaction) curve of distinct sign types will saturate - the estimated asymptotic richness (Chao1 or first-order jackknife) will fall below 45 types, and the number of NEW types first appearing in the youngest (Magdalenian) third of sites will be under 15 percent of the number contributed by the oldest (Aurignacian-Gravettian) third (primary clause: the <15 percent late-third novelty fraction, with the saturation ceiling as its corroborating clause; the verdict follows the novelty fraction). Disambiguation: composite/positional variants collapse to their base type per von Petzinger's own typology. Coverage guard: if fewer than 150 sites carry a scored sign inventory, the test voids for insufficient sampling rather than resolving.

Kill-dataset (verbatim)

Kill (not yet built): von Petzinger 2015 (University of Victoria thesis, "Geometric Signs of the European Upper Paleolithic") appendices and The First Signs (2016) tabulate the ~32 types against site occurrences - build the site x sign incidence matrix, run rarefaction/Chao1 on it, and compute the fraction of types first appearing in the youngest period-third versus the oldest. The decisive dataset is the incidence table digitised from those appendices plus the site period assignments.

Novelty / leakage triage

already answered in the literature

Von Petzinger's own thesis work and its popular distillation The First Signs (2016) already report the primary clause. She states as her central, deliberately counter-intuitive finding that roughly three-quarters of the ~32 geometric sign types are already present in the Aurignacian and that the repertoire barely enlarges across the following ~25,000 years - she expected an open-ended vocabulary that kept growing and instead found early establishment and stability. The site-by-sign temporal distribution underlying that claim is tabulated in her corpus, so the predicted '<15% late-third novelty' fraction restates a published result; only the formal rarefaction/Chao1 packaging of the corroborating saturation ceiling remains to be computed.

• G. von Petzinger, The First Signs: Unlocking the Mysteries of the World's Oldest Symbols (New York: Atria/Simon & Schuster, 2016), on ~3/4 of sign types present by the Aurignacian and the stability of the inventory thereafter
• G. von Petzinger, University of Victoria research on the geometric signs of the European Upper Palaeolithic (thesis and associated sign-inventory / site-occurrence tables)
• E. Dutkiewicz, S. Wolf & N.J. Conard, 'SignBase, a collection of geometric signs on mobile objects in the Paleolithic,' Scientific Data 7 (2020) - the digitised sign corpus over which the rarefaction operation would run

Predictions

No prediction registered yet.