Search for a Celtic DNA test and you will find products that promise a Celtic percentage. The problem is that Celtic names a family of languages and a pair of material cultures, not a population geneticists can sample. What a test can honestly do is compare your genome with published Iron Age individuals from Britain, Ireland and Gaul, the regions where Celtic languages were spoken, and report resemblance, a modelled share and shared segments. This guide explains where the word stops and the evidence starts.
Key facts
Figures as published, one source per row.
| Britain was remade twice before the Iron Age | Beaker-associated people replaced about 90 percent of Britain's Neolithic gene pool within a few centuries of 2450 BC, bringing steppe ancestry and the R1b lineages that dominate the islands today.Source: The Beaker phenomenon and the genomic transformation of northwest Europe (2018), Olalde et al. |
|---|---|
| The likeliest Celtic-language vector was Bronze Age, not Iron Age | A large migration into southern Britain between about 1300 and 800 BC raised farmer-related ancestry from roughly a third to about half, and the Iron Age that followed saw comparatively little new inflow.Source: Large-scale migration into Britain during the Middle to Late Bronze Age (2022), Patterson et al. |
| The Irish genome was already in place by 2000 BC | Three Early Bronze Age men from Rathlin Island carried steppe ancestry and R1b-M269, and their genomes already resembled present-day Irish, Scottish and Welsh people more than anyone else.Source: Neolithic and Bronze Age migration to Ireland and establishment of the insular Atlantic genome (2016), Cassidy et al. |
| Iron Age Gauls were locals with regional structure | Genomes from Iron Age France show continuity with the Bronze Age population of the region, a north-to-south genetic gradient, and individual mobility across it, rather than a single Celtic people.Source: Origin and mobility of Iron Age Gaulish groups in present-day France revealed through archaeogenomics (2022), Fischer et al. |
| There is no Celtic cluster | In the fine-structure map of 2,039 British people, the Welsh, Cornish, Scottish and Northern Irish clusters were among the most different from one another, not a shared Celtic group set against an English one.Source: The fine-scale genetic structure of the British population (2015), Leslie et al. |
| R-L21 is older than the Celts | The R1b branch marketed as Celtic is carried by most Irish and Welsh men and is common in Brittany and Scotland, but it spread with Beaker-associated people around 2500 BC, more than a thousand years before any Celtic language is attested. |
| What Ancestrify reports | The named ancient individuals your raw DNA shares segments with, each with its site, date and published source, €29.99.See Ancient Matches |
Celtic is a language family and a material culture#
Linguists use Celtic for a branch of Indo-European: Gaulish, Celtiberian and Lepontic on the continent, all extinct; Irish, Scottish Gaelic, Manx, Welsh, Cornish and Breton in the islands and Brittany. Archaeologists use it for two successive central European styles, Hallstatt from about 800 to 450 BC and La Tène from about 450 BC to the Roman conquest, whose swords, torcs and curvilinear art spread across much of Europe. Greek and Roman writers used Keltoi and Galli for peoples of Gaul, Iberia, the Alps and the Danube, and never for anyone in Britain or Ireland. The idea that the Irish, Welsh and Scots are Celts dates from the eighteenth century, when the languages were classified together and the name was extended to their speakers.
None of these three usages describes a population that can be sampled. A language spreads by learning; a style spreads by trade and imitation; neither requires a body to move. So the first thing to say about any Celtic DNA test is that its label is borrowed from philology, and the genome has never heard of it.
What Iron Age genomes actually show#
Britain and Ireland were remade twice before anyone there could have spoken a Celtic language. Olalde and colleagues showed that Beaker-associated people replaced about 90 percent of the Neolithic gene pool within a few centuries of 2450 BC, bringing steppe ancestry and the R1b lineages that still dominate the islands. Cassidy and colleagues had found the same shift in Ireland: three Early Bronze Age men from Rathlin Island, around 2000 BC, carried steppe ancestry and R1b-M269 and already resembled present-day Irish, Scottish and Welsh people more than anyone else. The Irish guide walks through both resets.
Then came a second, quieter change. Patterson and colleagues sequenced roughly a thousand ancient individuals and found a large migration into southern Britain between about 1300 and 800 BC, raising farmer-related ancestry from roughly a third to about half. The Iron Age that followed, the period archaeologists call Celtic, saw comparatively little new inflow. The authors' conclusion was that if a movement of people carried Celtic languages into Britain, the Bronze Age migration is the more plausible vector than anything in the Iron Age. Iron Age Britons, in other words, were mostly the descendants of people already there, with fine regional structure and, as the Iron Age kinship study showed, communities organised around women's lineages.
Across the Channel, Fischer and colleagues sequenced Iron Age individuals from sites across France, the Gauls of Caesar's account. They found continuity with the region's Bronze Age population, a north-to-south genetic gradient, and individuals who had clearly moved across it. Northern Gauls sat close to Iron Age Britons; southern Gauls leaned toward Iberia and Italy. Hallstatt and La Tène objects travelled far more freely than the people who made them.
Put the three together and the picture is consistent. The Celtic-speaking world was genetically a set of regional Bronze Age populations that adopted a language and a style, not a people who spread from a homeland carrying either.
The Celtic nations today#
The six modern Celtic nations, Ireland, Scotland, Wales, Brittany, Cornwall and the Isle of Man, are defined by language survival, and their genomes reflect geography rather than a shared origin. Leslie and colleagues typed 2,039 British people with all four grandparents from one area and found seventeen clusters. The Welsh clusters were the most distinct in Britain and preserved the oldest ancestry; Cornwall separated from Devon along the Tamar; Orkney stood apart with Norwegian input; Northern Ireland grouped with western Scotland. What the map did not show was a Celtic fringe united against an English core. The Welsh, Cornish, Scottish and Northern Irish clusters were among the most different from each other in the whole sample.
That is exactly what the ancient DNA predicts. Each region descends from the same Beaker and Bronze Age base and then diverged on its own, with Wales the least touched by later migration, Scotland layered with Gaels, Britons, Angles and Norse, and Ireland shaped by its own founder history and later Norse and English input. The Scottish and Welsh guides give each its own reading. Bretons, for their part, are western French with an insular tilt from the fifth to seventh century migrations out of Britain, and cluster with their neighbours in Normandy and the Loire rather than with Wales.
The R-L21 problem#
The Y-chromosome branch R1b-L21 is the one most often sold as Celtic. It is carried by most Irish and Welsh men, is common in Scotland, Cornwall and Brittany, and is rare east of the Rhine. The frequency is real; the label is not. L21 arose around 2500 BC and spread with the Beaker-associated people, more than a thousand years before any Celtic language is attested anywhere. It marks the Bronze Age settlement of the Atlantic edge, and it is found in men whose ancestors spoke Gaelic, Brittonic, English, Norse and French. Downstream branches such as M222 in the north of Ireland or DF21 in Britain can say something about medieval clan structure; none can say a word about language. The free Clade Finder will place your paternal line on the tree and show where it is common today, which is all a haplogroup can honestly do.
What a real analysis offers#
Since Celtic cannot be a source, the useful question is which Iron Age populations your genome resembles and by how much. Ancestrify's catalog carries Iron Age reference populations from exactly the regions that spoke Celtic languages: Celtic Britons from southern England, the Arras and Durotriges groups, Brittonic Scotland and Wales, Gaul, Hallstatt Austria and Germany, Celtiberians, Helvetians and Pannonian Celts. Each is a set of named, published individuals.
In the qpAdm analysis, an Iron Age British, Irish or Gaulish source can be offered alongside its competitors, an early medieval continental source for the Anglo-Saxon-era stream, a Viking Age Scandinavian one, a Roman-era Mediterranean one, and the model reports the weight on each with a standard error and a p-value for the fit. For most people from the islands the Iron Age source carries the bulk of the model, because that is where most of their ancestry comes from; the interesting number is what remains once it is accounted for. In Global25 coordinate space the same populations appear as distances, and a modern Welsh or Irish row sits notably close to the Iron Age British and Irish averages, closer than most modern Europeans sit to any Iron Age sample. And Ancient Matches scans your file against every individual in the panel, so a match to a specific woman from a Dorset Iron Age cemetery or a man from a Gaulish necropolis appears as shared stretches on your chromosomes, identity by state, with the evidence stated on the row.
None of that is a Celtic result. It is a measured resemblance to the people who lived in the Celtic-speaking world, which is the most a genome can offer and more than a label can.
Frequently asked questions#
Is there such a thing as Celtic DNA?#
No. Celtic is a language family and a set of archaeological styles. The people who spoke Celtic languages were regionally distinct Bronze Age populations in Britain, Ireland, Gaul, Iberia and central Europe, and no allele or haplogroup separates them from their non-Celtic-speaking neighbours.
Are the Irish, Scots and Welsh genetically Celtic?#
They descend mostly from the Beaker and Bronze Age populations of their islands, which is a real shared base, but they differ from one another as much as from the English, and no genetic marker defines the group. What they share is language history, not a genetic signature.
Which haplogroup is Celtic?#
None. R1b-L21 is the branch most often marketed as Celtic, and it does dominate Ireland, Wales and Brittany, but it arose and spread with Beaker-associated people around 2500 BC, long before any Celtic language existed. It marks the Bronze Age Atlantic settlement, not a language.
Can a DNA test tell me if I have Celtic ancestry?#
It can measure how closely your genome resembles published Iron Age individuals from Britain, Ireland and Gaul, model your ancestry from those sources with error bars, and show the segments you share with specific people from the period. That is a statement about populations and individuals, never about a language or a culture.
Terms used here are defined in the glossary.
Sources and further reading#
- Olalde, I. et al. (2018). The Beaker phenomenon and the genomic transformation of northwest Europe. Nature, 555, 190 to 196.
- Patterson, N. et al. (2022). Large-scale migration into Britain during the Middle to Late Bronze Age. Nature, 601, 588 to 594.
- Cassidy, L. M. et al. (2016). Neolithic and Bronze Age migration to Ireland and establishment of the insular Atlantic genome. PNAS, 113, 368 to 373.
- Fischer, C.-E. et al. (2022). Origin and mobility of Iron Age Gaulish groups in present-day France revealed through archaeogenomics. iScience, 25, 104094.
- Leslie, S. et al. (2015). The fine-scale genetic structure of the British population. Nature, 519, 309 to 314.
