January 23, 2016

Ancient DNA from England suggests strong impact of Germanic invasions

Recently sequenced Roman, Iron Age and Anglo-Saxon DNA sequences from England help to clarify the issue of the impact of Germanic migrations in Great Britain, which seems to have been significant.

Rui Martiniano et al., Genomic signals of migration and continuity in Britain before the Anglo-Saxons. Nature 2016. Open accessLINK [doi:10.1038/ncomms10326]


The purported migrations that have formed the peoples of Britain have been the focus of generations of scholarly controversy. However, this has not benefited from direct analyses of ancient genomes. Here we report nine ancient genomes (~1 ×) of individuals from northern Britain: seven from a Roman era York cemetery, bookended by earlier Iron-Age and later Anglo-Saxon burials. Six of the Roman genomes show affinity with modern British Celtic populations, particularly Welsh, but significantly diverge from populations from Yorkshire and other eastern English samples. They also show similarity with the earlier Iron-Age genome, suggesting population continuity, but differ from the later Anglo-Saxon genome. This pattern concords with profound impact of migrations in the Anglo-Saxon period. Strikingly, one Roman skeleton shows a clear signal of exogenous origin, with affinities pointing towards the Middle East, confirming the cosmopolitan character of the Empire, even at its northernmost fringes.

Notice that I say Germanic rather than Anglo-Saxon because I'm not sure how much can be attributed to these and how much to Vikings, whose genomes were similar. A recent study on British genetics seemed to indicate that the Danish (Viking) origins were clearly more important than the Saxon ones from Low Germany. However... were the original Angles more akin to Saxons or to Danes?

Anyway, the ancient samples are mostly Romano-Briton, from burials at Driffield Terrace, near York (Eboracum), dating to c. 200 BCE and including many decapitated remains. Another sample is from the Iron Age of Melton (East Yorkshire), dated between 200 and 40 CE. Finally a Christian Anglo-Saxon individual from Norton (Teesside, 70 Km north of York), dated sometime between the 7th to 10th centuries. 

Excepted one Roman era outlier (3DRIF-26), who seems an immigrant from the Eastern Mediterranean (autosomal DNA strongly suggests the Levant or Arabia), the rest all fit well with the autosomal genetics of the Iron Age one and modern Welsh. Modern English seem to have, in most cases, at least some Germanic admixture:

Figure 3 - Principal Component Analysis
(a) PCA of the Roman samples from Driffield Terrace (excluding one outlier), one Iron-Age individual and one Anglo-Saxon merged with modern Irish, British and Dutch genotype data. (b) Boxplot of PC1 broken down by subregion. The symbols on the left represent the significance of a Mann–Whitney test performed to compare the Roman population with all other populations in the data set. There were no significant differences between the Roman sample and the present-day Welsh, Northern and North Western English samples included in this analysis; all other regions had significantly different median values for PC1. Population key: Du, Dutch; En, English; Ir, Irish; NS, not significant; Sc, Scottish; Wa, Wales. NS-P>0.05; *0.05>P>0.01; **0.01>P>0.0001; ***P<0.0001.

Using the Dutch average as proxy for continental Germanics and the Welsh average for Romano-Britons, it would seem that modern English are on average, about 1/3 drifted towards Germanics, while the ancient Anglosaxon from Teesside was a bit more than half drifted in that direction. He was still within modern English variance, although rather towards the Germanic extreme of it.

Haploid lineages

The Iron Age sample was a woman with mtDNA haplogroup U2e1e.

The Romano-Brithons (all men), excluding the Eastern Mediterranean outlier, carried all variants of Y-DNA haplogroup R1b1a2a1-M412. It is notable that M405/U106 ("North Sea" subclade) was found in two of them, so it cannot be attributed to Germanic immigration. Another carried a sublineage of the M529 ("Irish") subclade (common also in Great Britain) and two others of the S28/U152 ("Alpine") sublineage (less common in Britain). The remaining two carried upstream L52* (generic "West European") paragroup lineages. See this entry for overall distribution details.

Their matrilineages were all subclades of H1, H2, H6 and J1. Details can be found in table 1

The outlier carried Y-DNA J2-L228 and mtDNA H5. The patrilineage fits well with a West Asian origin (an Italian one also fits) but the matrilineage is much more common in SE Europe, although it also reaches high frequencies in Wales. However the ADMIXTURE analysis strongly negates the possibility that he was European and very clearly supports a West Asian origin instead.

Finally the Anglo-Saxon man carried Y-DNA lineage I1 (most common in Sweden but scattered at low frequencies through Europe) and mtDNA H1.

Other details

The authors estimate that Iron Age and Romano-British samples were typically brunette with brown eyes. There is one exception though, 6DRIF-18, who was probably blond and blue-eyed, as was surely the Anglo-Saxon.

Blood type O was inferred for all Iron Age and Roman era samples, except 6DRIF-22 who was A. The Anglo-Saxon one may have carried type B or A (or AB?)


  1. The results appear to be well within the range of what might have been expected from historical evidence, which reassuringly suggests that both the genetics and the history are probably basically correct.

  2. Blood group O is known as a german to swiss blood marker ( also found in siberia )
    it was introduced into Britain

    1. Are you trying to make a joke or something? Zero (misnamed "O" often) is the most common blood group on Earth. In Europe too. And in Britain as well. It's irregularly mixed with group A everywhere. Group B on the other hand is most rare in the West and increases towards the East, being therefore tentatively associated with Indo-European expansion. It's also much more common in Asia, be it India or China.

      Your link is irrelevant to back your claim and BTW a very bad looking article in all senses: where's the data?, where the conclusions, why do they reproduce old blood maps with such a bad resolution?

    2. LOL, it's a true joke: one of those fake studies they introduced to test the system:

      Ethical issues (Including plagiarism, Informed Consent, misconduct, data fabrication and/or falsification, double publication and/or submission, redundancy, etc) have been completely observed by the authors.


  3. Using the Dutch average as proxy for continental Germanics and the Welsh average for Romano-Britons

    This seems like a big assumption. The same study that assessed that there was a 38% contribution of Anglo Saxons to the modern English population reckons that the contribution to the modern Welsh population is 30%. I'm not particularly convinced by these statistics but I do know that do know that through a whole host of later historical processes causing the movement and dispersal of people that today there's a Germanic (Anglo Saxon and Scandinavian) component in the ancestry of every part of Britain and Ireland. The regional differences lie in the degree of the contribution. Also I'm pretty sure that the Dutch situated as they are at the mouth of the Rhine, an area which has always supported a relatively large population have there own complex and interesting history. I see no reason to assume that the modern Dutch would be that similar to people whose origins stemmed in Lower Saxony and Jutland.

    1. I don't understand where you get that from, Amanda: in the graph above it is very apparent that the modern Welsh average overlaps with the ancient Britons, except the Anglosaxon one of course. Where exactly you get that idea from?

      "I see no reason to assume that the modern Dutch would be that similar to people whose origins stemmed in Lower Saxony and Jutland".

      Dutch are quite homogeneous internally and only show some division in two halves: the Northern one seems identical to Low Germans (Saxons and such), while the southern one may be identical to Middle Germans (former Frankish). Anyhow I do agree that a second reference from Denmark would have been of some help.

    2. Hi Maju,

      Sorry for my poor proof-reading on my original posting. I'll try to do better this time.

      I quite agree with you that the ancient samples from York cluster around the median point for the modern Welsh population. My question is how different is the modern Welsh population to the post Roman Welsh population?

      The figure of a 38% contribution of Anglo Saxon DNA to the modern population of Eastern England comes from the Schiffels paper 'Iron Age and Anglo-Saxon genomes from East England reveal British migration history'. The same set of assumptions lead the authors to conclude that the Anglo Saxon contribution to the modern Welsh population averages at 30%.

      I have to say that both figures seem a bit on the high side to me. Anyway even if the Anglo Saxon component in the modern Welsh population is a bit lower than that, it is still significant and suggests that your assumption that the ancient Welsh are much like the modern Welsh is wrong.

      The genetic continuity between the British and the Dutch (and the Belgians) is apparent in the PCA plots of modern Europe. I am pretty sure that the genetic closeness goes back into ancient times long before the Anglo Saxons and would be related to the migrations of Bell Beaker people and Iron Age Celts amongst others. The Anglo Saxon migrations and later arrival of Danes and other Scandinavians brought exotic new genetic elements from Scandinavia to the isles which had hitherto not been present. Some of these elements must have been already been brought into the Low Countries in some type of earlier migration process which brought German language speakers into that region. I still don't think that makes the modern Dutch into good proxies for the Anglo Saxons.

    3. I must say that I didn't criticize the Schiffels paper back in the day (I was just almost not blogging for much of 2015, personal issues) and now that I face it seriously for the first time, I find it perplexing: the general genome is not compared but instead:

      We determined for each ancient sample the number of rare variants shared with each reference population (Supplementary Note 3).

      That's not a method I'm familiar with and rather looks like the kind of data that medical-oriented research (and not historical genetics) would use. These rare variants are not likely to be informative of the overall distant ancestry. It's very weird.

      Almost all the study insists on using these rare variants. They even develop a method of their own (Rarecoal) to analyze them. Obviously it is an otherwise untested method, whose reliability I can't but suspect.

      Otherwise no IBD/IBS (which should provide approximative genealogies by a well tested method), no ADMIXTURE and only one PC analysis buried in the supp. materials (SF3). In this one the appearance of variation is way too subtle (because they use a lot of European populations that condition the whole graph) but if anything we can discern a tiny "Finnish" tendency in the Saxons versus a "Mediterranean" one in the pre-Saxons. However the only "Celtic" sample in that graph (Scots, Orcadians too) also tends towards Finland relative to English, what makes things more complicated (no Welsh nor Irish samples there, what seems like cherry-picking the data to me and hence suspicious).

      So for me that study is not informative but the data may still be reused by someone in a proper analysis, I guess.

    4. ... "your assumption that the ancient Welsh are much like the modern Welsh is wrong".

      It was not my intention to project such "assumption". I do not assume that in fact. What I merely stated is that the modern Welsh average overlaps too well with the ancient pre-Saxon NE Britons from Yorkshire. If we go through the individual variation of modern Welsh in that graph, some even overlap with the ancient Anglo-Saxon, while others with the Irish average, suggesting a more complex reality that maybe requires further analysis.

      "The genetic continuity between the British and the Dutch (and the Belgians) is apparent in the PCA plots of modern Europe. I am pretty sure that the genetic closeness goes back into ancient times long before the Anglo Saxons and would be related to the migrations of Bell Beaker people and Iron Age Celts amongst others".

      Well, in this study it seems clear that the Anglo-Saxon and Viking invasions pulled Britons closer to continental North Europe, regardless that relationships surely existed earlier (for instance: R1b-U106 is pre-Saxon and is most common in the Netherlands!) It is this extra pull what this paper reveals and not the overall pre-Saxon similitude or differences.

      This is not really new, already Capelli 2003, using only modern Y-DNA, appeared to show that East England has Anglo-Saxon or Danish admixture on the patrilineal side of around 30-40%, with a cline decreasing to the West and SW and reaching a minimum in Wales, which shows almost as little similitude in this aspect with the Germanic mainland as Basques do. The findings of this study are very parallel and comparable to those of Capelli.

      "The Anglo Saxon migrations and later arrival of Danes and other Scandinavians brought exotic new genetic elements from Scandinavia"...

      Not really Scandinavia by NW mainland Europe, Denmark included: only Orcadians seem to have significant Norwegian ancestry, which is clearly distinct from the Frisian-Saxon-Danish one. The real Scandinavians (Norwegians, Swedes) did not partake of those conquests at any significant level, they were busy fishing cod in Iceland or trading pelts in Russia. When we say "Vikings", we basically mean "Danes" and these are much more similar (in genetics, not language) to Low German and Dutch than to Scandinavians proper.

      "Some of these elements must have been already been brought into the Low Countries in some type of earlier migration process which brought German language speakers into that region".

      There was no such process. Germanic was only introduced with the Anglo-Saxons. You may mean Celtic language, you may mean Rhineland genetics rather than Saxon ones. It can get a bit confusing because Rhineland was almost certainly the original homeland of the Celts and British people do show some strong affinity to that area without marked distinctions (see fig.2 in Leslie 2015).

    5. Erratum: "Not really Scandinavia by NW mainland Europe, Denmark included" should read "Not really Scandinavia but NW mainland Europe, Denmark included"

  4. There's also a question that that they touch on in passing but don't examine as to what sort of demographic changes might have occurred in Britain during the Roman period.

    This is something we know hardly anything about. We know that there are several centuries in which Roman Britain is subject to an external continental ruler who make travel possible between Britain and the continent and within Roman Britain in a way that it never was before both because of a new infrastructure of roads and because the Pax Romana dissolved all the problems there would have been before with travelling across multiple tribal boundaries.

    It's quite possible that there was an overall influx of people from Gallia during this period, even it is likely that it ended with a reverse flow of people from Britain to Gallia once the economy started to hit the rocks after the legions left.

    1. We don't see that in any data, do we? Anyhow why would continental Gallo-Romans be attracted to migrate to Britain, which, by most accounts, was rather a backwater place?

      The only evidence we have of migrations is in this dataset, which includes an outlier of quite apparent Near East origins. Some other instances of Britano-Romans may have included people of African origin but, in any case, I doubt these immigrants would be common outside the big cities such as Londinum and Eburacum.

      In this particular case I suspect that the beheaded bodies could correspond to victims of persecution, maybe early Christians, known to have been beheaded in many other cases (much more common than lion-feeding in fact) for refusal to worship the gods of the state. They could be some other kind of rebels of course but the Near Eastern guy fits well with Christians.

      Influences from France in Britain should be at least partly traced to the very first Neolithic and Megalithism. In principle, Britain was settled from NW France around 4000 BCE. So, barring whatever pre-Neolithic remnants and not considering further changes affecting both France and Britain (notably the Celts), the relationship with France is the most basic demographic layer.

      Anyhow, what I was pointing to was a relationship with Rhineland (Germany), which can only be explained by the Celts (or pre-Celtic Bell Beaker influences maybe too but with many doubts until new evidence arises).

  5. I think the study was pretty reasonable. and yes the modern Dutch and Danish could be used as the standard for what the ancient Anglo-Saxons were like genetically. Historically we know that the Anglo-Saxons hailed from Denmark, northwest Germany and Frisia. So those populations are most likely the best choice. However, I believe the contribution of the Anglo-Saxons to the English in future studies would be found to be even higher than 38%, rather closer to half (50%), but that of Wales and Scotland would remain around 30%.It is also a false impression that the Romano-Briton was necessarily dark-eyed than the Anglo-Saxon samples, they were not.

    Blue eyes are the most common eye color throughout the British Isles, at varying frequencies. Wales isn't darker-eyed than England, the differences between Romano-Britons and Saxons were not clear-cut blonde vs brunette.

    ScottishDNA Project and Blue Eye Research Project (2014):

    England per regions:
    Southwest = 35%
    East = 41%
    Southeast = 44%
    Northeast = 47%
    Yorkshire = 49%
    Central = 50%

    Wales = 45%

    Scotland per region:
    Central: 48%
    Northeast: 48%
    North and West: 49%
    Southwest: 49%
    Southeast :57%

    Ulster : 50%
    Munster: 50%
    Leinster : 52%
    Connacht : 53%

    The regions which were firstly invaded by Anglo-Saxons, were the Southeast and East England regions, but the frequency of blue eyes are still lower than Wales.

  6. The Anglo-Saxon genetical contribution to England should be higher than that of Wales/Scotland though 30% is acceptable for the Celtic fringe. However, in England nearly half.

    1. In England "nearly half" ONLY if you use the, already admixed, British Anglo-Saxon as reference. If you use Dutch for reference, it's "only" 30% (what seems a lot to me, honestly, by contrast Goths had at most 5% impact in Spain, probably much less, similar insignificant fraction can be attributed to Muslims, be them Berbers or Arabs). It'd be c. 0% in Wales (although it's probably not that simple).

    2. Probably not 0 in Wales anymore - there was a fair amount of colonization of Wales after it fell to the English.

    3. Well, the issue of how the people of Wales were back in the day remains unsolved by this study, as it only provides ancient references from NE England. We see here that the Welsh average overlaps almost exactly with ancient NE Britons but we also see that the individuals vary quite a bit, most of them between the English and Scottish average ranges. So it's legitimate to wonder if Iron Age Welsh were rather Scottish- or Irish-like on average (relative to the above graph). That may well also apply to parts of England (we see many red dots, like purple ones, well into the Irish strip). It's indeed conceivable to me that what is now SW England was quite different from NE England and instead much more similar to what was then Wales and Cornwall and all them more similar to the Irish. But only more ancient data will clarify that far.

    4. One of the interesting things the POBI project and the Leslie 2015 found was more haplotype structure similarity between Scotland and England than between Wales and England (with I believe the same samples).


      That was also to some extent true in their unlinked ADMIXTURE and PCA analyses as well


      So I wonder if the position of Scots on the PCA which has Dutch-Irish poles, where the Scots are more towards the Irish pole than the Welsh, contra apparent haplotype sharing, is more due to

      a) Scots sharing more haplotypes with England than Welsh do due to recent ancestry, despite ancestry on unlinked SNPs being more divergently Irish-like


      b) Scots having more recent migration from both England / continental Europe and Ireland (Gaels) compared to Wales.

      or something else.

      There's an Irish version of the POBI running at the moment, so that may throw some light on these questions.

    5. Samples seem quite different: Martiniano has many Welsh and probably from all districts, while Leslie has only a few and most are from the westernmost areas. Martiniano has few Scots (not sure where from) while Leslie has lots of Orcadians (clearly absent in Martiniano).

      In Leslie Orcadian peculiarities drive the show: dominate PC1 (vs all) and PC2 (vs North Welsh) and therefore we have basically a Brythonic vs Orcadian polarity, what makes Scots and English appear similarly intermediate but for different reasons: (a) Scots are less Brythonic than Welsh but also less Norwegian than Orcadians, (b) English are more Brythonic and also more Germanic than Scots.

      Here Scots (Orcadian or not) do not matter (too few samples) and the polarity is just Ireland vs Netherlands, Celtic vs Germanic, Neo-/Chalcolithic vs Bronze/Iron Age, pre-IE vs IE even. Sure: we could get a more nuanced approach by getting more Scots in and also a few more continental samples (Danish and French maybe), possibly sampling more English on regional bases too (to compensate for actual numbers and to discern intra-English differences, which should exist in spite of Leslie's suspiciously too homogeneous results). But nobody is doing that analysis. I would keep excluding Orcadians from most analyses anyhow because they clearly distort the playing field.

    6. Ah, I'd have thought they were accessing the same samples, but obviously not.

      Leslie / POBI was very particular about sampling only people whose four grandparents were from the same isolated villages, which is where the balance of relatively high Welsh, high Orcadian, very low mainland Scotland comes from (apparently not many isolated villages in Scotland where people were living and all had four grandparents from the same place, and were possible to get to, was their explanation). I would note that the mainland Scottish isolated groups haplotype clusters break out from Orcadian ones in Leslie's study, as being much more related to English (and more related to English than Welsh).

      The Martiano samples are maybe more weighted towards cosmopolitan samples from Wales and Scotland, which may be why there is so much difference, as well (due to England-Wales migration affecting the cosmopolitan Welsh and English?).

    7. They seem a good example on how different sampling strategies can produce different results, more so when the analysis affects a rather homogeneous macro-population such as are Brits or NW Europeans, where the internal differences are quite subtle.

  7. Maju you stated that the Sardinian culture was less "advanced" than Los Millares.

    I assume that you were referring to the Nuragic civilization, I personally would argue that although the Nuragic culture developed later than Los Millares, architecturally the Nuragics were much more advanced, just look at any major Nuraghe like Santu Antine, Barumini or Arrubiu, with their multiple tholoi and their large corridors, in Santu Antine's case we have corridors superimosed on two floors which is an astonishing feat of engineering for the time.
    Look at the Nuragics sacred wells and their complexity: http://www.academia.edu/2336031/The_nuragic_well_of_Santa_Cristina_Paulilatino_Oristano_Sardinia._A_verification_of_the_astronomical_hypothesis

    And I'm just mentioning a few monuments, since the Nuragics left behind several thousands Nuraghes and hundreds of temples, so they were much more "active" than Bronze age Iberian cultures and than other European cultures at the time in general.

    Even artistically the Nuragics produced a lot, you may not know this but they actually made the first human sized statues in Europe:

    So sorry, but when you just put aside the Nuragic civilization as a "less advanced" culture I can't let that pass.

    1. Not sure when I said that but I can only imagine that I meant that Sardinia, much like Britain, Aquitaine, etc. did not have civilization, as defined by the existence of cities (or towns or large walled settlements, call them what you wish). The point would be that the only known civilizations in Western Europe (or anywhere in Europe, excepted the Balcans) back in those dates were in Southern Iberia.

      As for the origins of the tholos construction technique, I understand that (barring West Asian precedents, separated by some 1000 years of nothing) the Iberian ones are also older.

      Another issue would be if and how the Sardinian nuraghe relate to the very similar Motillas of La Mancha. In that I'm willing to accept that, probably, Sardinian nuraghe are older, what brings many questions to the table. On one side the cultural elements of the Motillas seem to be the same as those of the nearby Bronce de Levante (Valencian Country), on the other side it's clear that some sort of interaction between Iberia and Sardinia was going on, also in terms of bull cult (which seems to replace or compete with an older deer cult but has also relatives in Crete and Anatolia). In very general terms, I'd consider Sardinia to be loosely in the cultural orbit of Iberia and SE France (Italy had not much to offer in that period yet) but the details remain obscure.

      But, of course, I agree that the ancient Sardinians were a vibrant society with, as you correctly say, huge cultural production. This however can also be said of Megalithic Britain, for instance, yet we still do not know of any "city" or equivalent in that area either so early on. Therefore the society has to be considered "rural" and not "civilized" (incipiently urban) yet.

      This probably has implications regarding the economic and political structure but hard to know in detail.


Please, be reasonably respectful when making comments. I do not tolerate in particular sexism, racism nor homophobia. Personal attacks, manipulation and trolling are also very much unwelcome here.The author reserves the right to delete any abusive comment.

Preliminary comment moderation is... OFF (keep it that way, please)