I have a methodological concern about the qpAdm design used to model Iron Age France.

If I understand the source construction correctly, the “local” Bronze Age proxy used for France is not a specifically French Bronze Age source, but a combined France/Iberia Bronze Age-related source. More importantly, the French Bronze Age individual included in this proxy appears to have an autosomal profile close to Basque/Iberian variation, and is paired with an Iberian genome. This makes the source a poor proxy for the broader local Bronze Age background of France, especially for northern, eastern or central French Iron Age targets.

This matters because qpAdm results are highly sensitive to source choice. If the local Bronze Age proxy is shifted too far toward a Basque/Iberian-like profile, then Iron Age French individuals will mechanically require an additional Central European / Knovíz / Hallstatt-related source in order to fit. In that case, the inferred Central European contribution may partly reflect the inadequacy of the local proxy rather than a real demographic input from Hallstatt-associated populations.

A stronger test would require modelling the French Iron Age targets against several alternative local Bronze Age sources: regional French Bronze Age proxies where available, eastern/northern/central French Bronze Age individuals separately, France Bronze Age without Iberian individuals, and only then Central European / Knovíz / Hallstatt-related sources in competition. It would also be useful to show whether the chosen France/Iberia Bronze Age source itself is representative of Bronze Age France, or whether it is an outlying southwestern/Basque-like proxy.

As it stands, the model risks building the desired historical conclusion into the source design: a weak or unrepresentative “local” proxy can artificially create the need for a Central European Hallstatt-related component. This should be addressed before drawing strong conclusions about the demographic spread of Celtic languages into France.