Large-scale Migration into Britain.

No, this isn’t about the causes of Brexit, I’m abbreviating the title of Large-scale migration into Britain during the Middle to Late Bronze Age by Nick Patterson, Michael Isakov, and a long list of coauthors ending with David Reich, published in Nature last month. At that link you can only get the abstract unless you’re a subscriber, so a more useful one is Juan Siliezar’s Harvard Gazette story about it:

New research reveals a major migration to the island of Great Britain 3,000 years ago and offers fresh insights into the languages spoken at the time, the ancestry of present-day England and Wales, and even ancient habits of dairy consumption. The findings are described in Nature by a team of more than 200 international researchers led by Harvard geneticists David Reich and Nick Patterson. Michael Isakov, a Harvard undergraduate who discovered the existence of the migration, is one of the co-first authors.

The analysis is one of two Reich-led studies of DNA data from ancient Britain that Nature published on December 22. Both highlight technological advances in large-scale genomics and open new windows into the lives of ancient people. […] The researchers analyzed the DNA of 793 newly reported individuals in the largest genome-wide study involving ancient humans. Their findings reveal a large-scale migration likely from somewhere in France to the southern part of Great Britain, or modern-day England and Wales, that eventually replaced about 50 percent of the ancestry of the island during the Late Bronze Age (1200 to 800 B.C.).

The study supports a recent theory that early Celtic languages came to Great Britain from France during the Late Bronze Age. It challenges two prominent theories: that the languages arrived hundreds of years later, in the Iron Age, or 1,500 years earlier at the dawn of the Bronze Age. Previous research has shown that large-scale movement often accompanied language changes in pre-state societies. The Reich team argues that this untold migration event makes more sense for the spread of early Celtic languages into Britain.

“By using genetic data to document times when there were large-scale movements of people into a region, we can identify plausible times for a language shift,” Reich said. “Known Celtic languages are too similar in their vocabularies to plausibly descend from a common ancestor 4,500 years ago, which is the time of the earlier pulse of large-scale migration, and very little migration occurred in the Iron Age. If you’re a serious scholar, the genetic data should make you adjust your beliefs: downweighting the scenario of early Celtic language coming in the Iron Age [and early Bronze Age] and upweighting the Late Bronze Age.”

There’s other interesting stuff (“the researchers found that the ability to digest cow’s milk dramatically increased in Britain from 1200 to 200 B.C., which is about a millennium earlier than it did in central Europe”), and I’m sure there are plenty of Hatters who will want to dig into it. Thanks, Bonnie!


  1. John Emerson says

    While these results may be scientiically and historically interesting, we have to ask ourselves whether they are really worth it, given the way that the Celtic fantasists are sure to make use of these facts.

    At least this doesn’t bring the Celts much closer to Stonehenge.

  2. The Celtic fantasists were probably rooting for “1,500 years earlier at the dawn of the Bronze Age.”

  3. David Eddyshaw says

    Bah! My Celtic forebears were too busy levitating the stones for the Pyramids to bother with cheap knockoffs like Stonehenge.

  4. Knowing as I do nothing on the subject of genetics but a thing or two on the subject of historical linguistics, I will say that, assuming that this 1200-800 BC population movement did indeed trigger language shift in (at least parts of) the British Isles (which seems likely, but is not certain), then this new language is unlikely to be the direct ancestor of any of the Modern Celtic languages (Both those of the Goidelic and those of the Brythonic branch). Because of this we have no way of knowing whether it indeed was Celtic or not (but see below).

    The reason for my saying this is that it has proven well-nigh impossible to present even a single clear difference between reconstructed insular Celtic (spoken sometime between 100 BC and 100 AD) and its contemporary Celtic sister, Gaulish, attested (roughly) during the same time period. For this reason it is unclear, in fact, whether there are any solid grounds for treating Proto-insular Celtic and Gaulish as separate languages/dialects.

    This is very unlike the situation when comparing Gaulish and its Southern Celtic sister language, Celtiberian: despite both being attested by very reduced/fragmentary corpora, there is a set of features setting Gaulish sharply apart from Celtiberian (The latter makes use of fully declined forms of the Indo-European relative pronoun *yo-, which in Gaulish + Insular Celtic is an undeclined verbal suffix (?clitic), for example).

    If Insular Celtic had indeed descended from the language brought to the British Isles by these invaders, no later than 800 BC, I would expect Insular Celtic to be as sharply differentiated from Gaulish as Celtiberian is. Since it is not, I can only assume that Insular Celtic must have arrived in the British Isles at a much later date.

    If my feet were held to the fire and I was forced to guess, I would say that the 1200-800 BC invaders probably spoke an Indo-european language and that this invasion was the first Indo-Europeanization of the British Isles (In whole or perhaps only in part): if this language was Celtic (which I think is extremely likely) it would have been far more archaic than Gaulish or Insular Celtic, indeed it might well have been quite Celtiberian-like (Actually, I would consider calling it Para-Celtiberian: PC). For these Early British Celtic/PC speakers Proto-Insular Celtic would probably have been an easily-learnt L2, so that we might think of the British Isles as having been Celticized in two waves: an Early British Celtic/PC wave associated with the 1200-800 BC invaders, and a much later wave (whether it left genetic traces or not will have to be determined by future research: mark you, since Proto-Insular Celtic would have been so easily learnable by Early British Celtic/PC speakers, a scenario whereby Insular Celtic spread via mechanisms of language spread other than demographic expansion is certainly possible), with the latter *and not the former* being directly ancestral to all Brythonic and Goidelic languages.

    Cue the outrage from the Celtic fantasists in three, two, one…

  5. Those sound like very reasonable ideas to me.

  6. Trond Engen says

    The immediate lesson is that Celtic probably came with the very front of Hallstatt-affiliated cultures rather than with later movements within Hallstatt. That will have implications also on the continent, but it may be too early to tell which.

    I’ve not been able to find a draft version online, so it’s not much more to say except what’s said by Reich and Isakov in the Harvard Gazette.


    The studies are not only important for Great Britain, where we now have far more ancient DNA data than in any other region, but also because of what they show about the promise of similar studies elsewhere in the world.


    “It’s sort of incredible that we have geneticists, we have statisticians, we have archaeologists, linguists, and even chemical analysis coming together. I think that the fact that we’re able to like merge all these fields and have an actual insight that’s culturally important is a great example of interdisciplinary science.”

  7. Trond Engen says

    … and now recalibrating for what Etienne says.


    It’s reasonable that Celtiberian spread with the front of Hallstatt, and that Core Celtic developed somewhere else in Hallstattia. Does this have to be a land-based region, or could it be centered around the Atlantic trade network and the Channel? And/or could there have been a back-migration into the European continent?

  8. David Marjanović says

    I’ll just caution against underestimating the diversity within “Gaulish”, i.e. Celtic within more-or-less-historical Gaul, which covers a lot of space and a noticeable amount of time.

    It also seems clear that robust linguistic contacts between Celtic languages across the Channel continued well into Roman times.

  9. Yes, that’s what I’m starting to wrap my head around. It’s so easy to assume that the insular Celts went off to the insulae and the continental Celts stayed on the continent and they never spoke to each other again. Brexit avant le mot!

  10. I haven’t read this latest evidence, but if indeed it restricts the Beaker Culture replacement of inhabitants of Britain to 50% rather than the 90% which, if I recall correctly, was the figure given by Reich himself (in “Who We Are and How We Got Here”) my prejudices suggest it is less unbelievable.

  11. Trond Engen says

    I don’t think this means a downward adjustment of the 90% replacement. This is a later and more gradual migration that brought peoples with a higher proportion of European Farmer ancestry.

  12. A 50% replacement after the 90% replacement? Blimey! DNA reconstructions of prehistory are fascinating and are gradually revealing some fascinating stuff. I will hang on to a bit of my scepticism, though. I think we are very much at the early stages of this research (the picture has changed dramatically since Bryan Sykes’ “The Blood of the Isles” just over 15 years ago, when we were told the bulk of our DNA was present here in the Neolithic at the latest) and it would not be surprising if there were vastly new pictures to come).

  13. Exactly. It’s quite a thrill to be able to watch a science being so thoroughly revolutionized in real time!

  14. Trond Engen says

    Not so much being revolutionized as being built up from scratch, which means that the dots are few and får between, and there are still major discoveries to be made in the gaps.

  15. Dmitry Pruss says

    It’s tough call to me. Yes, their sampled remains from Iron Age Britain have an unusual excess of old European Farmer ancestry (EEF) which has been reduced to lower levels in the British Isles during the earlier Bronze Age migrations. And the same kind of ancestry is also enriched in Urnfield remains they were able to sample.

    But these interesting British samples came from a small area on the Channel’s shore and may or may not be possible to generalize to a wider swath of Britain (one could argue that different peoples may have traveled in the Channel without necessarily impacting the inland areas). And Y-chromosomes in England, as they existed in the subsequent era, are distinct from anything known in the early continental Celts so far. And generally Europe was rather thoroughly homogenized in Bronze Age, making predictions based on broad ancestral components somewhat less exact. One might need to seek confirmation in more narrowly specific DNA phenomena, like Y-chromosomes and especially sharing of autosomal haplotypes.

  16. Substantial discussion also ongoing at Eurogenes, with similar reservations as Etienne how this is not at all necessarily the Insular Celtic spread event.

  17. Trond Engen says

    “793 newly reported individuals” sounds like a lot. You mean to say that only a few (late and local?) genomes actually show a shift towards European Farmer ancestry? I didn’t know that. But even so, the population level ⁷shift was already remarked upon in the previous study, so I understamd this to be more about being able to connect it to a timed wave of immigration rather than internal development. Not having read the paper, I may very well be wrong.

  18. Savalonôs says

    Etienne’s comments make a lot of sense. I’m still vaguely weirded out by the idea that Proto-Goidelic was the same language as gaulish / Brittonic as recently as ca. AD 1, just because Irish seems so distinctive. But, I gather that the Primitive Irish evidence indicates that Goidelic was undergoing very rapid changes in the first few centuries AD, so I guess that’s when it became distinctive. Perhaps the speakers of the gaulish dialect that became proto-goidelic had entered Ireland very recently prior to the primitive irish period. Perhaps their language underwent rapid changes at that point because of intensive contact with the earlier local language, which we just missed having direct evidence for. Who knows, perhaps that earlier local language was some kind of astounding Celtic-Afroasiatic creole that had developed after the late bronze age migrations of early Celtic speakers!

    P.S. naturally, I’m mostly goofing when I say Afro-Asiatic specifically—no Vennemannian, I—it would be whatever language EEFs spoke. Could be AA, or Hurro-Urartian, or Hattic, or Sumeroid, or maybe even Elamo-Dravidian. Most likely if, by some near miracle, we recover written evidence of the last pre-IE language spoken in Ireland, it would not be discernibly any of the above (Vasconic would not be out of the question).

  19. Savalonôs says

    P. S. Patterson et al’s findings extend only to England and Wales, correct? So, I suppose any inferences drawn about the genetic or linguistic history of Ireland or Scotland on this basis are limited to the speculative.

  20. But, I gather that the Primitive Irish evidence indicates that Goidelic was undergoing very rapid changes in the first few centuries AD, so I guess that’s when it became distinctive.

    Yes. As I said here, “Primitive Irish looks reassuringly like Latin.”

  21. David Marjanović says

    Perhaps the speakers of the gaulish dialect that became proto-goidelic had entered Ireland very recently prior to the primitive irish period.

    Check out the first half of this comment and the replies to it.

  22. >check out the first half

    So just the first 15,000 words?

  23. David Eddyshaw says

    As I’ve remarked before, St Patrick (that famous Welshman) would probably have found Irish fairly easy to acquire. The changes that make Irish so very Irish largely postdate his time (and indeed, the sheer glorious irregularity of Old Irish itself shows that those changes hadn’t been around long enough for their effects to get smoothed out by analogy.)

  24. Savalonôs says

    Great comment, easily one of my top 500 favorite comments from that thread. Seriously, though, I’m fairly certain I read the first half of that comment before, half forgot about it, and had it in the back of my head. So I took this as an opportunity to read the next 25% or so. Fascinating stuff.

  25. David Marjanović says

    So just the first 15,000 words?

    Yes 🙂

  26. Nick Patterson says

    A clarification. 90% replacement and 50% Steppe are not in conflict.
    Northern European BB (and British BB) have about 50% Steppe, 50%
    European farmer (pre-Steppe or middle Neolithic) ancestry. The point is that there is little or no dilution of the Steppe ancestry in British BB compared with mainland European BB which means the replacement must be nearly complete.

  27. Adrian Martyn says

    Delighted to find this site – more like it are sorely needed.

    This is a wonderful genetic insight, but cannot prove what they propose. The earlier dates were not made by linguists but geneticists, right? More than a bit cheeky for geneticists to try the same approach to re-adjust the goalposts.

    If they are serious scholars, the linguistic data should make them adjust their beliefs and upweight the Iron Age. Fair’s fair …

Speak Your Mind