Ancient genomes link Celtic and Germanic spread to migrations, but language is inferred

Two Nature papers using 4,587 imputed ancient genomes link Celtic to a Late Bronze Age movement out of eastern central Europe and Germanic to an elusive group that reached Sweden about 4,000 years ago, while stressing that DNA cannot show which language anyone spoke.

Peer reviewed · Ancient-DNA analysis · Language link inferred

Researchers at the University of Copenhagen and collaborators report that migrations visible in ancient genomes line up with the spread of early Celtic and Germanic languages. For Celtic, the best fit is ancestry moving west from eastern central Europe between 3,200 and 2,800 years ago. For Germanic, they propose a group that reached Sweden from the Baltic region by about 4,000 years ago. The catch: the language link is an inference, since genomes record ancestry, not speech.

Evidence at a glance

Is this a big deal?

A solid step for the population-history half of the story, and an open question for the language half. The datasets are large and the Celtic result favours one of four tested models. But the authors themselves say the findings do not definitively establish the Celtic homeland.

What scientists found

The Celtic paper tested four proposals. Celtic took shape with the Bell Beaker expansion (roughly 4,750 to 3,800 years ago), or in Atlantic Bronze Age networks, or in inland France and possibly northwestern Italy, or in eastern central Europe alongside the Urnfield culture. It found three processes in the genomes. Steppe-related ancestry arrived with Bell Beaker-associated groups from about 4,400 years ago. People moved into Britain from the south between 3,800 and 3,200 years ago. Then, from 3,200 to 2,800 years ago, ancestry linked to the Knovíz–Urnfield-associated group spread west from eastern central Europe into France, Britain, northeastern Iberia and Italy.

Only that third movement overlaps most linguistic estimates for when Proto-Celtic formed and fragmented, mostly 3,600 to 2,800 years ago. The authors say this makes it the most parsimonious explanation, a statement about the simplest fit, not a demonstration.

The Germanic paper sequenced 712 ancient genomes and analysed them with 4,009 published ones. It identifies an archaeologically elusive population that entered Sweden, probably from the Baltic region, by about 4,000 years ago and spread across Scandinavia by about 3,500 years ago. The authors propose it as a possible vector for early Germanic. Later movements include a southward migration by about 1,650 years ago into presumed Celtic-speaking areas. During the Migration Period (1,575 to 1,375 years ago, roughly 375 to 575 CE by SciQuest’s conversion) the same northern European ancestry appears in Anglo-Saxons in Britain and in Langobards in southern Europe. A northward back-migration into Denmark and southern Sweden coincided with the documented emergence of Old Norse.

How they found it

Ancient DNA is old, fragmentary and often scarce, so the team first sequenced hundreds of skeletons and filled in missing positions statistically (imputation) for individuals with enough data. Then they asked who shared long stretches of identical DNA, which is a sign of recent common ancestry, and modelled each individual as a mixture of earlier groups. Spatiotemporal kriging, a mapping method that interpolates between sampled sites, turned the results into movements across space and time.

The key move was to compare those genetic timelines with linguistic dating of Proto-Celtic and with archaeology. Because the four Celtic models make different predictions about where and when ancestry should shift, the genomes can favour some and not others. The Celtic data are open at the European Nucleotide Archive (accession PRJEB87274).

Why it matters to you

This will not change daily life or health. It does speak to a long-running argument about where two language families began. Nature‘s news report puts Germanic speakers at over one billion and Celtic speakers at roughly 2 to 3 million. English is a Germanic language, and the Anglo-Saxon result is part of that background, though the papers do not claim to explain how English itself arose.

What the evidence does not show

It does not show which language any individual spoke. The authors of a piece in The Conversation write that “Ancient DNA cannot reveal the language or languages a particular individual spoke.” Kroonen, a historical linguist and co-author, is quoted in Nature‘s news report: “You can easily adopt a new language without changing your genes.”

The Celtic homeland is not settled either. Partial support remains for western models, and the paper calls a subtler westward signal in the La Tène period one that “needs further confirmation.” Sampling is thin in places: cremation during the Urnfield period leaves few Late Bronze Age genomes, and coverage between the Netherlands, western Germany and northern France is limited. “Knovíz-related” ancestry stands for a broader group, not one culture, and its origin and formation are unclear. The Germanic authors add that they cannot rule out more complex scenarios.

Linguistic date estimates are also wide and come from different methods. Proto-Celtic formation, for instance, is placed mostly between 3,600 and 2,800 years ago, with the greatest overlap at 3,200 to 3,000 years ago.

How this fits what we already knew

Most historical linguists accept that Indo-European languages spread from the Eastern European steppe, and earlier DNA work showed large steppe migrations about 5,000 years ago. The Conversation authors say earlier thinking tied Germanic’s arrival in Scandinavia to steppe migrants about 4,800 years ago. The new result suggests a second steppe-ancestry group about 800 years later (4,800 minus 4,000), which they propose, hedged, may have introduced the precursor of Germanic.

For Celtic, the new data give mixed support: an earlier Atlantic movement into Britain gives partial support to western origins, while the eastern movement fits best overall. See also how ancient proteins are reshaping human origins for another case where molecular evidence meets older assumptions, and 13,800-year-old evidence from Alaska for how archaeology and genetics cross-check each other.

Independent expert view

SciQuest found no on-record comment from a researcher outside the author team. The quoted voices in Nature‘s news report are co-authors, and ScienceAlert and The Conversation likewise rely on or come from the team. Nature‘s report notes that linguists disagree about where Celtic originated, without naming them. An independent view is the main gap in this story, and SciQuest will add one if it appears.

How other headlines framed it

• Nature news (Ewen Callaway) treated it as thousands of genomes pointing to migrations that “may have spread” early forms of the languages, with the language-versus-genes caveat voiced by a linguist co-author.
• The Conversation, written by three of the researchers, framed it as tracking origins and spread, with explicit hedged wording.
• ScienceAlert chose the stronger “rewrites the story” framing, though its text repeats that DNA alone cannot show spoken language.

The strongest framing was the “rewrites” headline. The papers themselves describe a proposal and a best fit, not a rewrite.

What happens next

More genomes from thinly sampled regions would help most, especially Late Bronze Age Europe between the Netherlands and northern France, where cremation limits the record. The authors also say the La Tène signal needs confirmation, and independent teams can now test the models against the open data. The linguistic dates may shift too, which would change which movement fits best.

In one sentence

Ancient genomes from 4,587 people show migrations that fit the timing of Celtic and Germanic language spread, but fitting the timing is not proof that the migrants carried the languages.

Read next

• Ancient proteins mark Xujiayao fossils as Denisovan: another case of molecular evidence qualified by its limits.
• Alaska base camp from 13,800 years ago: how archaeologists trace migrations from the ground.
• The most ancient civilization in the world: the deeper background to early societies.

Sources and disclosure

1. McColl, H. et al. “Tracing the spread of Celtic languages using ancient genomics.” Nature, 4 November 2026.
2. McColl, H. et al. “Tracing the spread of Germanic languages using ancient genomics.” Nature (2026), page dated 28 October 2026.
3. Callaway, E. “Rise of Germanic and Celtic languages is written in ancient DNA.” Nature news, 4 November 2026.
4. Kroonen, G., Willerslev, E. and McColl, H. “We used ancient DNA to track the origins and spread of the Germanic and Celtic languages.” The Conversation, 4 November 2026.
5. Cockerill, J. “Ancient DNA Rewrites The Story of How Germanic And Celtic Languages Spread Across Europe.” ScienceAlert, 4 November 2026.

Disclosure: This article was drafted with AI assistance and checked against the sources above, including the open-access text of the Celtic paper and the first part of the Germanic paper. Calculations by SciQuest: 4,800 minus 4,000 = 800 years; 1,950 minus 1,575 and 1,375 gives roughly 375 and 575 CE; 578 new plus 4,009 published = 4,587. The graphic is a SciQuest original built from the papers’ reported values. Spotted an error? See our corrections policy.