Neanderthals in the Nefud, argued entirely from the way a flake was struck

A study in Quaternary Science Reviews argues that the prepared stone flakes at Jebel Katefeh-1, beside an ancient lake in Saudi Arabia's Nefud Desert, were made by Neanderthals between 60,000 and 50,000 years ago ^[4]. No Neanderthal bone has ever been recovered anywhere in Arabia, so the whole population attribution rests on how the stone was struck ^[3].

Two problems had to be cleared in order. The first was the clock. JKF-1 had previously been dated across a span from 90,000 to 50,000 years ago ^[5], a window wide enough to contain both Homo sapiens and Neanderthals, who each moved through Southwest Asia between roughly 100,000 and 50,000 years ago ^[2]. Earlier work put a Neanderthal expansion south into the Levant and the Zagros at roughly 75,000 to 50,000 years ago, with sapiens also present across that region and interval ^[6]. The authors write that because the existing chronological constraint spanned both presences, "examining the source of this heterogeneity has been prohibited" ^[7]. New luminescence dating on the site's sediments, which estimates when grains were last exposed to sunlight, returned about 60,000 to 50,000 years ^[14]. That is a 10,000-year window in place of a 40,000-year one ^[18].

The second problem was attribution. Fossils are scarce in Arabia, which is why lithics are asked to identify populations at all ^[20]. The team built a dataset of 704 complete prepared flakes from parts of Africa, Arabia and the Levant dated between 100,000 and 50,000 years ago ^[9], then ran tests designed to separate the candidate causes of variation: assemblage age, landscape setting, environmental conditions, assemblage size, composition, raw material use, site type and population history ^[8]. Within the reference dataset, samples associated with sapiens and with Neanderthals were told apart with more than 93 percent agreement ^[10]. Population history, not climate or toolstone, best explained the shape of prepared flakes in the model ^[11]. JKF-1 came out on the Neanderthal side, at a median Neanderthal likelihood of 78 percent for the key shape pattern ^[12], with only four artifacts more probably sapiens than Neanderthal ^[13].

Worth being clear about what those two percentages are. The 93 percent figure describes separation inside the labelled reference set ^[10]; the site being classified has no fossil against which the classifier can be checked, which is exactly the condition that makes the method attractive and also makes it hard to falsify ^[3]. And a median of 78 percent is a lean, leaving about 22 percent on the alternative ^[19]. It is a probability statement about flake shape, not a body.

If it holds, the consequence is geographic. The result pushes the known Neanderthal southern range past the Levant into northern Arabia and cuts against the picture of a cold-adapted European population ^[15]. The authors argue Arabia may have been both a route out of Africa and a meeting ground ^[16], and that admixture between the two species "could have occurred over a broad, and currently poorly constrained, geographical area which includes Arabia," given how poorly fossil material and DNA survive at low latitudes ^[17].

Watch whether the classifier is tested against assemblages that do have diagnostic fossils, whether the luminescence ages are reproduced by an independent lab, and whether the other heterogeneous Arabian assemblages get the same treatment or resist it.