March 05, 2011

Celto-Germans vs. Balto-Slavs

Here are the first two dimensions of a multidimensional scaling plot of the following samples:
  • Dodecad Ancestry Project: 6 Poles, 12 Russians (Russian_D), 11 Germans, 19 Scandinavians, 6 Mixed Slavs (various West and East Slav combinations), 17 Britons, 17 Irish
  • HGDP: 25 Russians from Vologda
  • Behar et al. (2010): 10 Lithuanians, 9 Belorussians
Applying MCLUST over these first two dimensions and with K=2, the following breakup of individuals ensues:

It's fascinating that Cluster #1 (which corresponds to the assortment of individuals on the left of the MDS plot) includes only Balto-Slavic individuals (65 in total), while Cluster #2 (on the right, includes all 64 Celto-Germanic individuals plus 2 Poles and a mixed Slav.

This surprising concordance is even more striking once we consider that one of the "mixed Slavs" in my sample may be of Prussian origin within present-day Poland. I will be happy to tell the Poles in my sample which cluster they belong to if they write to me at the Dodecad Project e-mail address.

A lot has transpired since the ancient ethnographers divided the little-known peoples of the far north into Keltoi and Skythai, or since the Franco-Russian anthropologist Deniker divided the light-pigmented Northern Europeans into a race nordique and a race orientale. So, it is a bit surprising to see that a basic division of northern Europeans into East and West has stood the test of time. (*)

(*) Minus the Finnic peoples of northeastern Europe who, as has become clear, owe their genetic distinctiveness to a Siberian element in their ancestry, tying them to their linguistic cousins in the east.


Diogenes said...

MCLUST didn't pick it up since it was forced to assign individuals to one of only 2 clusters, but MDS is very suggestive of similar admixture level of both populations in 4 of the 9 Germans, and a smaller Northeastern European component in another German. Similarly for 2 of the Scandinavians. Same apparently applies in reverse to 2 of the 6 Poles, who show possible Northwestern admixture in 2 individuals.

I think this be a fossil of a major ancient admixture event.

In this case Brits-Scandinavians-Germans might represent decreasing population 1 contribution and increasing population 2 contribution. Balts-Byelorussians-Poles the opposite? Maybe with a third Finnic/Siberian element added in for the Russians?

Also, I think the symmetry of the MDS if we exclude Russians seems to posit opposite population movements from 2 core/origin areas (contemporary or not), wherever they are, since if one population had stayed put it would show unequal admixture levels of the other, which isn't the case.

The admixed populations would then represent the area where the two waves clashed.

Average Joe said...

Well done! A minor issue is that in the graph the color dots for the British and the Irish are very similar as are those for the Scandinavians and the mixed Slavs. It makes it a little difficult to distinguish between these ethnic groups.

eurologist said...


As to the Prussian, it depends of course if this person's ancestry is Old Prussian (i.e., native Baltic) or from the occupying Germans.

Would definitely be worthwhile to see at least the first 4 (if not 5) dimensions. Or perhaps you can even do a rotating animated .gif of the first three dimensions?

Fanty said...

That clean split into 2 clusters is amazing to behold.

Unknown said...

What an insolence using Vologda samples as the "Russians" in every Western DNA-research?

The same way you could use a "Russian" sample from Yakutsk or Kazan. The substratum admixture appears too significant in such regions, so this way you will get a good propaganda stuff but will never see who Russians really are. The more appropriate are central regions of European part of Russia like Smolensk, Tver, Voronezh. If the ones are not available that is better to note the used samples are not representative for real Slavic Russians. Only if you make the science, of course.

Andrew Lancaster said...

Compared to geography, I guess the most surprising thing is the distance between Scandinavians and their closest neighbours to the south and east. They are further from them than German are, and that is what seems to cause the big genetic gradient showing in the Poland-Germany area? It would be interest to see what would happen with France and/or Finland added.

Dienekes said...

What an insolence using Vologda samples as the "Russians" in every Western DNA-research?

The blame should be placed on Russia and its policy of not allowing export of DNA. Until then, and until Russian researchers publish their own data, the HGDP sample, and the data submitted to the Project by various Russians who manage to evade the ban (e.g., by expatriates) are the only data we've got.

Njord said...

is there any way to get the ID's of these participants labelled on the plot?

Dienekes said...

As per the project's privacy rules I either post DOD-unlabeled/ethnically-identified results or ethnically-unlabeled/DOD-labeled ones. This is an example of the former.

Fanty said...

That gradient region between Poles and Germans is of course easily explained by the existance of multi-etnian empires like Prussia or the German Empire.

Map, that shwows Polish speaking areas in the 2. GErman Empire:

Or Prussia, annected a large part of Poland and called those provences "South Prussia" and "New East Prussia"

Or just recall General Feldmarschal "Erich von Manstein". Whos true name was "Erich von Lewinski" and who was partly of Kashubian (A slavic speaking minority in Prussia/German Empire) anchestry.
He was adopted by the Manstein family. Thats why his Polish surname was exchanged into a German one.

The Lewinski family served in the Prussian and later German armies for 16 generations.,_Erich_v._Manstein.jpg

Peronaly, I wonder if those 4 Germans are actually from these old Prussian terretories and since 3 of my 4 grandparents are from these parts of the German Empire (East Prussia and Silesia), I wonder if I am of those 4 spots or not.

Diogenes said...

I think you make some sense. German colonization of the East could have led to admixture events explaining this. However 4 out of 9 individuals is a lot and I'm not sure this suffices. Also this doesn't explain why there's such a huge gap between both groups. Poles are also admixed to a large degree apparently and they never colonized the west (post WW2 events can't count). There's no desert or sea between them. There is between Ireland and GB, and between GB and the Continent, as well as between Scandinavia and the Continent. Germans and Poles should fade into each other if admixture within neighbouring populations was the explanation.
I think this points to a more ancient event. The admixed individuals are I believe not exceptions but the rule for border populations (Eastern Germans and Western Poles).
It's not just Germans, Scandinavians (R1a can't be explained by Finnish admixture) and Poles either. Romanians, Hungarians seem to have such a spread among "the gap" in an European MDS plot too. These populations are EXACTLY where you'd expect them in such a theory.

The gap must point to significantly different origins of both groups, and since only food producing technology could explain such a phenomenon by assimilation of sparse (but genetically gradually merging geographically) preexisting populations, I think it's likely that "Celto-Germanic" is actually the large Neolithic wave coming from the Mediterranean and the Balkans, rivers of Central Europe from Western Anatolia, while the Eastern wave may be another food producing revolution coming from the rivers of the Ukraine and ultimately also from some other population at or close to the Near-East. Naturally gradual assimilation of some hunter-gatherers occured along the way, explaining the spread among each population.

I think these waves, particularly the Western one, may thus be totally unrelated to Indo-European linguistic diffusion. Pre-existing cultural similarities may explain the development of different IE branches instead.

Average Joe said...

It is strange that the Scandinavians seem to be more closely related to the British and the Irish than the Germans. I wonder if this is evidence of Viking ancestry in modern British and Irish populations.

eurologist said...

Dienekes: "I think this be a fossil of a major ancient admixture event."

Or ongoing admixture and separation since the neolithic. However, parts of Northern Germany, the Baltic, and Poland were so sparsely populated (due to poor soils and climate) that the core populations could retain their separate genetic identity through the past ~ 9,000 years.

These are the ancient populations I see in the area, going back to before agriculture, but solidified during the first ~2,000 years of agriculture:

1. Central European (From NE France through Hungary, in the NE to the Vistula river; with more Mediterranean and Anatolian/Caucasus admixture in the very south)
2. Ancient Scandinavian
3. Baltic (around Vistula river and East thereof)
4. East European (with Eastern Ukraine as core region)

Then, relatively late, Finnish migration from the Urals, Balto-Slavic later perhaps from the East Belarus/NE Ukraine area, but with impact mostly on the language, only, and then 1,500 ya Slavic expansion into Czech, Slovakian, and Polish areas including the Western Baltic and parts of what is now Eastern Germany.

So, I think there is an ancient East-West gradient in Europe (and a second one to the Baltic) that was maintained by a relatively sparse mixed population in the "center" (NE Germany through Poland/Belarus).

The relatively recent Slavic expansion has actually steepened this gradient by introducing more Eastern Europeans toward this "center."

Other historic events have likewise increased this separation: German re-settlement and pushing out of most Slavs from west of the Oder river and beyond (with the Wend and Sorb minorities exceptions), as well as the mass exodus of ethnic Germans (including "German Prussians") from east of the Oder river after WWII.

Dienekes said...

Dienekes: "I think this be a fossil of a major ancient admixture event."

You got your quotations mixed up.

eurologist said...

I guess Diogenes is just too similar!


eurologist said...

Average Joe: "It is strange that the Scandinavians seem to be more closely related to the British and the Irish than the Germans."

I don't see that in the plot. In pretty much all European PC analyses you can see two additional Scandinavian components (an ancient Scandinavian and a Sami/Finnish one). If you ignore those, then there is a simple gradient between southern Swedes and southern Germans, with northern Germans in between. Neolithic settlers to the British Isles surely were very close to people in the Dutch/Northern German region, which in turn are close to modern southern and western Scandinavians. And all those populations form the core of historic settlers. Even Normans were a mixture of Vikings, Franks, and older northern French/ Belgian populations that surely were not all that different in their autosomal make-up, either.

Of course then you have other components from pre-neolithic inhabitants, prehistoric and ancient seafarers, trades people, and soldiers/occupiers from the west Atlantic and Mediterranean.

Anonymous said...

British Islands were sstrongly influenced by Scandinavians at least for four centuries, this is historically documented and can not surprising.

Andrew Oh-Willeke said...


It isn't inconsistent to think that there are two reinforcing separate population waves on each side. Hence, on one side you have Atlantic European followed by Celto-Germanic and on the other side you have early Eastern Neolithic (e.g. LBK) followed by Slavic.

While intermediate individuals could represent recent admixture, they could also represent cases where the early Neolithic layer and the Indo-European layer don't reinforce each other - for example, a place with Eastern affinities in the early Neolithic that developed Celto-Germanic affinities in the Indo-European wave.

eurologist said...

Andrew Oh-Willeke: "...the Indo-European layer..."

The thing is, though, that there is very little evidence of such.

You can easily explain all current genetic findings without ever knowing about or invoking IE. Which may indicate that either: (i) IE expansion had no significant impact on most European populations, because they were archetypal resistant agricultural settlers, or (ii) that, and in addition, those populations already spoke a language so similar that absorption of a a more global IE was easily done, without any measurable population exchange, at all.

pconroy said...

@Average Joe,

My father to the best of his knowledge has no English or Norman admixture, does however has Ancestry Finder matches in Norway and Sweden, and consistently clusters close to Swedes - in one of Davidski's runs he was in the Swedish cluster.

He also happens to suffer from Dupuytren's Contracture (aka 'Viking Hand'), which I believe Ronald Reagan and Maggie Thatcher also had.

Diogenes said...

Yes. I believe current genetic patterns in Europe were mostly set in the Neolithic with only relatively minor changes since. Pots AND peoples maybe not much later, but at least for the Neolithic Revolution!

Its definitely interesting that despite mostly unscientific and completely wrong conclusions from most 19th and early 20th century anthropologists, their measurements seem to have been correct, as they correlate with genetics. And the Neolithic archeological panorama as well...

Over at Eurogenes, there are some strong indications of this, since the correlation of archeological ancient Neolithic cultures ("pots") with current European genetic elements ("people") is astonishingly close (
However I think pre-set ideas may be blinding some of the analysis over there, which is a pity, since I believe this new ADMIXTURE fixture is what was needed to make sense of all of this. Maybe we can even bypass the need for additional South Russian populations with it.

Its a pity we can't have an Ossetian sample, since seems to me what "Dagestani" (Lezgins) was supposed to be might be them all along. However by finding similarities at the rim maybe we can find the source of the Eastern Neolithic wave. Like Chuvash, Ukranians, Northwest South Asians, Central Asians? If this could be demonstrated maybe we could finally start changing this paradigm for good...

I'm still hoping Dienekes will look into this with his large sample again...

Unknown said...

I think we are seeing the basic split between the R1a population (cluster 1)and R1b (cluster 2). Other splits (eg Atlantic coastal vs other inland migrations) known to be present in Europe are not shown clearly in this spread.

The complete overlap of the Scandinavians with the "celtic" peoples is intriguing, suggesting maybe that the root of both was identical. Brits should be composed of at least 2 major components, with the Swedes more Atlantic coastal based. I am worried we may be seeing a blurring due to the Scandinavians being pooled together.

Diogenes said...

@Annie Mouse
This is a local variety of the wider Neolithic Atlantic, Megalithic Culture, from Iberia to Norway.

Fanty said...

"The complete overlap of the Scandinavians with the "celtic" peoples is intriguing, suggesting maybe that the root of both was identical."

That correlates with the ancient idea (already mentioned by Dienekes), that the people of Northern Europe are only 2: "Celts" in the West and "Skythians" in the East.

The "Germanic" tribes had been estaminated to be "Celtic" by the Romans by the first contacts and it took them another 100 years to realise, that these are a third people in the North. One that lived between "Celts" and "Skythians".

A similiar thing are the "Picts".

Most common people of today usualy believe them to be Celts.

Scientist claim, that culturaly, "Picts" are neither real Celts nor are they "Germanics" but a third cultural group.

While Tacitus (Roman historian from the 1th Century AD)claims them as "Germanic", estaminated by their physical apearance.

Fanty said...

I almost forgot one more "intermixture 2000 years ago" thing:

Tacitus describes a "Germanic" tribe in what is now "Sklovakia", that he believes, must have intermixed with the "Skythians", since their faces show traces of these looks.

Also, worth to mention:
tacitus does not believe the Goth (Who lived in Northern Poland during the time of Tacitus) to be Germanic (but he does not explain what he believes they are) while at th same time he mentiones the "Suiones", from Southeast Sweden as the northern most Germanic people known so far.

And as kind of a proof of how well Tacitus must have known the people of the North, he mentiones that the Finn are neither Germans nor Skythians but a third people.

pconroy said...

These findings, together with Eurogenes, corroborate - or at least are in line with - what I suggested 3 years ago, that Germanic languages are a Creole, with a Baltic substratum and a Celtic superstratum.

It would seem that Baltic speakers stretched all the way to Denmark, then gradually were conquered from the South by Celtic speakers, and what developed was a Creole we know as Germanic.

Average Joe said...

It would seem that Baltic speakers stretched all the way to Denmark, then gradually were conquered from the South by Celtic speakers, and what developed was a Creole we know as Germanic.

Are you saying that the original Celtic speakers were Iberian? If so, how did Celtic reach Iberia? Was it a language indigenous to Iberia or was it brought there from somewhere else?

Average Joe said...

I think we are seeing the basic split between the R1a population (cluster 1)and R1b (cluster 2).

But doesn't Scandinavia have high levels of R1a? Shouldn't that place the region in the eastern cluster rather than the western one?

Anonymous said...

The division of the Northern Europeans into East and West may have stood the test of time but your results don't indicate the length of time those East and West Northern Europeans have resided in Europe or where they came from outside Europe.

Your results just show that they are differentiated and have been since the Ancients, not that long ago in time in reality, listed the Barbaroi as Keltoi and Skythai.

pconroy said...


Well if Tartessian is the oldest form of Celtic attested - from South Western Iberia - then the Celtic language could have spread from there.
Most models assume that Celto-Italic arose in Central Europe - read Southern Germany, Switzerland, Austria - and spread East to France, then South to Iberia and North to Britain and Ireland.

But if Celtic - not necessarily Celtic peoples, just the language - arrived in Western Europe, from a likely source in the Usatova Culture of Romania/Moldova, then that might explain why Celtiberian is Q-Celtic, like Ireland - i.e. the more ancient form - and France/Gaul and Britain were P-Celtic, the newer form.
Last year I linked to a paper that proposed that the insertion of Etruscan into North Central Italy, and it's subsequent expansion towards the North East as the Villanovan Culture forced a sound shift in Celtic, that gave us the P-Celts

Diogenes said...

OK. I may be overextending myself, but here is what I think is the most likely model.

The Baltic region was likely colonized by a first agricultural wave of people coming up the rivers of the Ukraine, I think. The Western North European plain likely received the Atlantic Neolithics, with some Danubian Culture-people contribution.
Maybe then as the Eastern Neolithic people filled the rivers and more fertile lands of the steppe, they adapted to a pastoralist lifestyle, likely advantageous for food production in the grass expanses.
They domesticated the horse, developed lactose tolerance, and improved social organization and animal breeds. Then they exploded demographically, as the Kurgan Expansion.

The Northern European plain had high densities of agricultural people in more fertile land, in the river valleys and some other areas. However as someone pointed out there are many sandy soil areas which are less good for agriculture but ideal for pastoralism. These "Baltic" Indo-European peoples may very well have occupied this niche. Since they had horses and chariots, a tremendous military advantage, and a warlike culture, they became dominant over the agriculturalists and a fusion "melting pot" culture drawing heavily on the Funnelbeaker Culture (with western Atlantic affinities), came into being- the Corded Ware/ Battle Axe Culture.

This was almost exclusively East Neolithic+ related Kurgan in the Baltic Sea area (Baltic branch); mostly "Baltic" with considerable Danubian/"Central European" and "Atlantic" in the east-middle (Slavic branch) and mostly "Atlantic" with some "Baltic" and CE in the North Sea/Scandinavia region (Germanic branch).

Slavic branch culture suffered much more Western Atlantic/Danubian influence than Baltic branch culture. They would adapt western agriculture techiques/seeds/organization to their eastern ones, and would thus be able to bypass the Balts into expansion into Forest zone Russia, then inhabited by a much more primitive neolithic culture- Finns (Combed Ceramic Pottery Culture).

From their Northern European plain base with their military advantages (=horses) smaller bands conquered the Western Atlantic and Southern Danubian and Mediterranean areas, establishing themselves as an elite and spreading Indo-European tongues.

Into Italy proto-Italic (similar to proto-Celtic actually); into France and Iberia and Britain pre-proto-Celtic-like tongues. Like Lusitanian. They were likely fragmented, coming in all directions from Corded Ware area. Linguistic unity could only come later.

"Atlantic" cultures relied a lot on seaborn small distance trade among themselves.
These Atlantic "proto-Celts" in Southwestern Iberia however sat on one of the richest copper and silver mines in Europe/Med.
Cornwall on a motherlode of Tin. This attracted attention from the developed Bronze age cultures of the Middle East. Phoenicians established colonies like Gades (Cadiz) as early as 1100BC.

I think it's likely the Celtic proper, despite its ultimate distant Eastern origins, developed as a lingua franca for this trade, either before or most likely under the strong stimulus given by Phoenician luxury items traded in return for the metals. Q-Celtic linguistic and geographical heterogeneity seems to point to it being an older group with P-Celtic likely an innovation associated with later Gaulish expansion into both Q-Celtic (old-Celtic?) areas and into non-Celtic speaking areas.

Randy McDonald said...

In the case of the German-Polish border regions, the new west of Poland was populated disproportionately by people from the lost east--the former Breslau, now Wroclaw, was resettled substantially by migrants from the former L'wow, now L'viv.

eurologist said...

"It would seem that Baltic speakers stretched all the way to Denmark, then gradually were conquered from the South by Celtic speakers, and what developed was a Creole we know as Germanic."

Archaeology shows continued cultural independence of the region north of the Danube watershed all the way to Denmark and east beyond the Elbe river (and at times farther, to the Vistula), for millennia. The region (except for the Saale/southern Elbe portion) was always poorer than the south, but traded with it even before a Celtic culture was established, there, but more so after - likely based on extreme proximity and the fact that they spoke about the same language (likely not the local "Celtic" - whatever form deviated from proto-Germanic that was - of the aristocrats and business people who started establishing a few new oppidae along the rivers in the south).

There is zero evidence of large-scale military advances northward (the "Celts" in southern Germany were crafty trades people, and lucrative trade is what they were interested in, also established by the formation of new trade towns along the rivers), as there is likewise zero evidence of any advance of something "Germanic" from Scandinavia.

The Celtic culture between much of Switzerland, southern Germany, Austria, and the Czech region was largely carried by an upper class that made large gains through connected, organized trade, but of course protected by an organized military. Celtic military expansion as such is a very late development - post-dating proto-Germanic by many centuries, and with targets solely in the very south. The majority of the people in the hinterland (including much of the south) undoubtedly were little affected by all of this and continued to speak their language - i,e, proto-Germanic.

Diogenes: "Since they had horses and chariots, a tremendous military advantage, and a warlike culture, they became dominant over the agriculturalists"

Horses during that time (from Bronze age until just before Roman times) were useless in the North for many reasons:

- there are no unforested plains you can cross, easily: you had thick underbrush, thick forests, and swamp land - with productive agricultural islands in between. The ox provided transportation through the narrow walk ways and heavy ground, with trails often straight - right through hills (rather than the gently curved trails advantageous for horses and horse carriages).

- the ground can not be passed easily for half a year (October through March or April) by horse over large distances without proper shelter and fodder stations.

- horses were so expensive to take care of in the north, that they were extremely scarce (well-documented by archeological finds) well into the iron age, almost until Roman presence. Hey require expensive (to grow and harvest) oats that you would not grow in large quantities, otherwise. Again, oxen were the work animals on the heavy fields, not horses.

There was no Kurgan expansion into Poland and Germany in terms of a "business model" - a better way of making a life, there. People would not have made it a single season.

Unknown said...

"But doesn't Scandinavia have high levels of R1a? Shouldn't that place the region in the eastern cluster rather than the western one?"

Both Norway and Sweden (36%,42%)are mostly I1 with about even R1a and R1b. But this does not directly reflect the autosomal position because of the unknown relative impact of the maternal genetics (overwhelmingly H, 42%, 46%)

Diogenes said...

Horses are never useless when you have them and the enemy doesn't. They could be few or many but still a major advantage. Horses and Chariots made it to Iberia very early (documented Chariot representations in Tartessian art), certainly not because they were useless.

The impact of the Kurgan people is well documented by Marija Gimbutas in the Northern European plain. Her theory has been gaining increasing acceptance, and could be considered majority view today...

Zones without many trees can always be found, and horses are likely a major military advantage even in less dense forests.
If the Northern European plain was congenial to pastoralism, likely history would be different. As it is, it's still the most congenial zone in Europe after the Ukranian steppe and Hungary. Thats why when much later Mongols/Tatars tried to conquer Europe, they came in through Hungary and Poland (where they were stopped by military defeat only).

arvydas said...

Fanty said: "That clean split into 2 clusters is amazing to behold."
I don't see that "clean split". And if you add the Finns (most probably in the left-upper corner, if we extrapolate Baltic trend) then it becomes a nice tripod.
It is clear that the far right is Ibero-Britannic R1b, but I wonder- what is in the lower-left corner, towards which the Russians are gravitating?