search this blog

Wednesday, September 12, 2018

Avars and Longobards

Most of the "barbarians" from today's Amorim et al. paper have made it into the Global25 datasheets. Look for the samples with Collegno and Szolad in their labels. Same links as always...

Global 25 datasheet (scaled)

Global 25 datasheet

Global 25 pop averages (scaled)

Global 25 pop averages

Here's my usual Principal Component Analysis (PCA) of West Eurasian variation with the same individuals. As seen in the paper, the two females from Avar burials are very European indeed, with no hints of any recent Asian ancestry. The relevant datasheet is available here.

And this is my Global25-derived North European PCA featuring a subset of these samples that plotted firmly with present-day populations from north of the Alps, Balkans and Pyrenees. The aforementioned Avars (red dots) are sitting within the Polish cluster. The relevant datasheet is available here.

See also...

First real foray into Migration Period Europe: the Gepid, Roman, Ostrogoth and others


zardos said...

Only a thin layer at the top of the Avar society showed East Asian features. The majority of the population was local and came under Avar rule, especially Slavs.

EastPole said...

“As seen in the paper, the two females from Avar burials are very European indeed, with no hints of any recent Asian ancestry. …The aforementioned Avars (red dots) are sitting within the Polish cluster”

So at least one thing seems to be clear now. Avars the Hyperboreans, which influenced Greeks were not Mongols.
Recently a book have been published: ‘A Story Waiting to Pierce You: Mongolia, Tibet and the Destiny of the Western World.” By Peter Kingsley.
Some review:
“Kingsley’s story starts with Abaris Skywalker, Pythagoras’ “teacher,” an Avar-ambassador from the people of Hyperborea, from Mongolia, a shaman carried and being carried by the golden arrow indicative of his mission, and thus able to cover incredible distances. One ancient account says Abaris was carrying his arrow and the other says the arrow was carrying Abaris. Both accounts together corroborate the basic shamanic principle: “those who carry an object of power in their hand are carried by it” (p. 13).
All the strands of our evidence point to a Mongol shaman, at the dawn of Western culture, who came to purify the land of the Greeks, constantly moving in a circle around the land. Here Abaris met Pythagoras, recognized him as a divine incarnation (a procedure described by Iamblichus,8 but also still a definite procedure among Tibetans and Mongols for identifying a great being who from compassion decides to become embodied), and gave Pythagoras his arrow. Instead of the sublime rationality we have come to associate with the supremely Greek god Apollo, there is a Mongol who walks as the servant of a Central Asian Apollo by virtue of an unbroken trance “holding his god within him.”9
What this means, according to Kingsley, is that the mutual understanding between Abaris and Pythagoras is, in reality, the more intimate and comprehensive relation of a god (Apollo) trusting and understanding a god (Apollo), “two bodies but only one being, like recognizing like”

Greek mystery religions and so called Orphico-Pythagorean tradition, which so much influenced Greek philosophy, came from Avars, but Avars-Hyperboreans were not Mongols. Kingsley is completely wrong.

Vedic religion, as described in Rigveda, shows many similarities to the Greek mystery religions and to Slavic religions and this is because they have the same origin. They all came from Eastern European Hyperboreans, not from Mongolia.

Look at solar cross on diadem from Drotowo, Poland, Trzcieniec culture 1400 BC:

And look at Avar solar cross from Panonia:

Matt said...

Dropped individual samples onto a neighbour joining tree with population averages:

huijbregts said...

I could not find the Hungary_Medieval and Italy_Medieval samples in the Global 25 data. My mistake?

Davidski said...

They're definitely in the sheets linked to above.



huijbregts said...

Thats odd. When I clicked your link I got a view of an outdated version in my downloaded map. Anyway thanks.

André de Vasconcelos said...

To be honest I find the study of the southern folks more interesting than the northern ones, who we already know who/what they are. The southern ones kind of vary from the very southernmost europeans to around as "north" as Iberians

James Goblin said...

females are unimportant, as the barbarians married locals.

Samuel Andrews said...

Some of the Italian Medieval samples are overwhelmingly East Mediterranean, similar to Anatolia BA. The purely 'southern' samples in Hungary for the most part cluster with Balkan_IA/BA & Bavaria Medieval outlier. Some though have a Anatolia_BA pull.

Samuel Andrews said...

Also has anyone noticed Anatolia_IA:MA2197, Hellan-era Anatolian, clusters close to Balkan_IA. He has no Anatolian ancestry. Now, we see Anatolians in Medieval northern Italy. What's going on?

Hopefully, David & Matt & others can pinpoint ethnic affiliation in this diverse group from Lombard burials. I bet, the samples that are actually ethnic Lombards will show affinity to their modern Germanic relatives.

Samuel Andrews said...

Wonder if Kindred group CL2 in Collegno are not of mixed southern/northern ancestry like Figure 3 in the paper suggests.

I wonder if they're a northern Italian family who are roughly intermediate between northern & southern Europe.

Here's one member of that group.



Matt said...

Couple of runs of re-processing the Global25 data through Past3 PCA, with averages for other populations and the samples from this post:

Narrower set:

Wider set (more South and Central Asians with ENA admix and North African):

Not sure if these are informative (so many pop labels and hard to read) but throwing them out there anyway.

Mikkel Nørtoft said...

One of the males (SZ15) from Szolar, Hungary shows Y-haplogroup R1a-Z284 (on the European Z645 branch that didn't go Indo-Iranian Z93), and which is quite common in Swedish and Norwegian populations today. A later node of Z284 mostly found in populations of Scotland from what I can gather at Yfull:

Here is my own R1 tree with ancient individuals plotted:

Haven't looked at the other samples yet :)

Samuel Andrews said...

I have a decent West Eurasian PCA working using past3. It includes the SC Asian Neolithic samples.

All the 'southern' individuals, except for one, have no relatives in the burials, so probably people straight from foreign non-Lombard populations.

Uruk Osmerius said...

@Samuel Andrews

I doubt they were natives to Italy, the oldest member of the kindred group is a likely migrant:
"In Kindred CL2 the man in grave CL49, of the first known generation, also had a likely non-local 87 Sr/86 Sr value, making him a migrant to Collegno. His son, daughter and nephew however (CL53, CL47 and CL57) had local strontium isotope values, suggesting that they spent their entire lives close to Collegno. This means that the woman in grave CL102 and the man in grave CL49 were first-generation migrants to Collegno, while all individuals who were in the third generation in either kin group were likely born locally."

The supplementary info states that CL47 was buried with transalpine objects:
"Some female tombs (CL48, CL47, CL147 and perhaps others) contained non-Italian but transalpine objects (Supplementary Figure 20) which suggested the presence of some individuals of a different origin that differs from both Italy and Pannonia, especially women who may have moved from neighboring regions (possibly from the Frankish-Burgundian area in south-eastern Gaul, starting from the nearby Val di Susa) due to exogamy, perhaps accompanied by some subordinate individuals."

I think its more likely these people are migrants from Gaul or Noricum.

Garvan said...

Mikkel Nørtoft said...

Thanks for your R1 tree, I will keep this for a reference.

A small error .. Rathlin Island (or any part of Ireland) is not in Great Britain, so the "Great Britain only" tag looks strange.

Bob Floy said...


Youcould just change "Great Britain" to "United Kingdom" and you'll be correct.

Open Genomes said...


What happened to the most interesting one, CL31, who looks like he's part-East Asian?
He has a Caucasus / Kirghiz G2a2a1 Y-DNA, the same as DA117, and H6a1b mtDNA, like Khvalynsk Eneolithic and the pre-Hittite MA2208.

Davidski said...


CL31 has a very high contamination estimate, so it's difficult to say how relevant his results are to Medieval Italy.

Samuel Andrews said...


Do you have RISE61 on your tree.

Corded Ware, Denmark, RISE61 R1a1a1b1a3b1(a) Z282+, Z284+, Z287+, CTS8401+, Z281+

Samuel Andrews said...

I was wrong saying some of the Italian samples are Anatolian. It is more likely they're southern Italians. This includes three samples: CL38, CL30, CL25.

They're more Mediterranean than modern southern Italians. We could make all kinds of theories of what these extremely Mediterranean Medieval samples mean about Italian origins.

IMO, they explain why modern Italians go away from the Beaker-EEF cline towards Bronze age Anatolia. IMO, these Medieval Italians mostly descend from an ancient Anatolian people who settled in southern Italy & made a big impact on the whole peninsula.










huijbregts said...

I collected the samples of Germany_Medieval, Italy_Medieval_Collegno and Hungary_Medieval_Szolad.
A cluster analysis with mclust shows two very distinct clusters.
The majority cluster contains all of the G10 Germany_Medieval samples, 12/17 Collegno samples and 21/28 of the Szolad samples.
The minority cluster contains 5/17 Collegno samples and 7/28 of the Szolad samples.
The minority samples are: CL23, CL36, CL57, CL94, CL121
SZ27, SZ28, SZ32, SZ36, SZ37, SZ40, SZ43
Italy_Medieval_Collegno:CL53 seems to be an outlier.


Very surprised that there has not been more discussion of the fact that the Collegno individuals with Southern ancestry are primarily local or from a geographic range broadly within Northern Italy... Their extremely southern ancestry coupled with the geographic location seems to indicate significant migration-era northern input across the Italian peninsula. Or is my reading wrong?

Aniasi said...

I'm a bit confused here. Then they mean Avar burial, do they mean a culturally "Avar" form of internment, or a burial from that time period?

Open Genomes said...


You could try to do Global25 CL31 and I can see what happens. Maybe with some restriction of the smaller components it makes sense. What I've seen from from the study is about 20% East Asian, which doesn't seem likely unless someone East Asian was handling the sample.
Amorim (2018) Figure 2c PCA showing ~20% East Asian in CL31

As I said, this is what we could expect from the Y. As you can see, the idea of ast Asian "Avars" (or in the case of Lombard Italy, "Bulgars" who are documented as having settled there) in some quarters is "controversial". If there are "unexpected" East Asian components (e.g. Japanese), then it wouldn't be all that useful. Let's try it and see what comes out.

VesteinofVestfirðir said...

Samuel Andrews has a really good point about the EEF/CHG affinity of the southernmost Collegno samples. In fact, perhaps we can return to the Aeneid of Virgil and the Ab Urbe Condita of Livy, both of which speak of a supposedly-legendary "Trojan" origin of the Latins?

Legends are legends - sure, but I recall that you wrote about a major Caucasus shift in Minoans in an earlier post this year:

Steppe admixture in Mycenaeans, lots of Caucasus admixture already in Minoans (Lazaridis et al. 2017)

Davidski said...

I haven't had a chance to take a really close look at these samples yet, but I will soon, and I'll probably do a blog post about that Sicilian-like or Jewish-like trio from Collegno.

These people might show isotopic ratios typical of the Collegno region, but I'd say that they're likely to be the descendants of migrants from another part of the Roman Empire, probably the Aegean region.

It's worth keeping in mind that at around this time Italy wasn't just populated from the north by Germanics, who then mixed with the supposedly Sicilian-like locals, but from almost all directions, and the Aegean region was likely to have been an important source of migrants for obvious reasons.

Until someone sequences remains from cemeteries associated with local cultures we won't know what the general population in Northern Italy was like at the time. For all we know, they may have still resembled the North Italian Bell Beakers.

Philippe said...

@ Eastpole,

Hyperborea is Bronze Age Britain.

Mikkel Nørtoft said...

Now the R haplogroup tree has been updated with Longobards and Niederstotzingen samples.
Find your favorite Longobard :)
(might want to refresh/delete cookies or whatever your browser needs)

ǵenh said...

CL38, CL30, CL25. It's very unlikely that these samples represent the average population of northern Italy. At the same time, it's unlikely that they all came from the same source.

CL30 is R1b1a1a2a1a2 S116, he can't be a migrant from the Aegean region, he is probably mixed on his maternal side (mtDNA I1b) and is roughly similar to modern far southern Italians. Instead CL38 is E1b1b1a1b1a3 and mtDNA is X2. CL25 is a female and mtDNA is H.

Open Genomes said...

There are two "Germanic" women, one STR_500, and one SZ19, who cluster very closely with each other, but are outliers to the typical "Germanic" or even part-East Asian members of the group.

Running Global25 clustering nMonte3 with scaled data shows something fascinating:

STR_300 nMonte3 Modern populations only

SZ19 nMonte3 Modern populations only

On the Ward's distance-squared cluster analysis with scaled data, both of these cluster very closely with each other and I5769 from the Balkans Iron Age.

STR_300 appears to be about 1/3 Mainland Greek, about 1/3 "Sardinian-like", with the remainder divided between Turkish Trabzon (in the cluster with Greek Trabzon), Cypriot, and Greek from Crete.

SZ19 is even more "Sardinian", 37.6%, and 1/3 Albanian and the remainder divded between Italian Abruzzo and Greek Trabzon.

I think that the combination of "Sardinian" (a Sardinian-like population) and Greek means that STR_300 is likely a Greek-speaking South Italian from the Roman Empire. SZ19 is substantially more "Balkan", but also has "Italian Abruzzo", which is a region to the east of Rome. She is likely an Italian, more of a Central-South Italian (Roman?) with some Albanian-like (Illyrian?) Balkan admixture.

What's so important about this is that we can likely see the makeup of the Classical Greek colonists of Southern Italy, and the typical Mediterranean Central-South Italian citizens of the Roman Empire. An ancient inhabitant of Magna Graecia would have been about 1/3 native "Sardinian-like" (pre-Indo-European) South Italian, with the remainder being from a diverse Greek population.

A typical Central Italian / Roman would still have some Greek ancestry, but also be more typically "Italic", with substantial ancestry from the pre-Indo-European inhabitants of Italy.

SZ19 is buried in a grave that is not typical of the other graves, with a bronze bracelet, which is not found in Germanic burials, but more typical of Classical Mediterranean burials. STR_300 however was buried with a "Frankish-Alemannic" brooch.

People have been looking for the aDNA of typical "Romans". FN2 was a Roman solider who was buried in Munich around 300 CE, but at least half of his ancestry was from Northeast Spain, and it seems he had a paternal grandfather who was Germanic. His G-L497 subclade is G-Y64315 which has a 2400 ybp tMRCA with a Swede. His grandfather was probably Gothic.

So here we have at least three "Romans", a 3rd century Roman solider of mixed Mediterranean Empire and Gothic ancestry, a woman of mixed Central Italian and Balkan ancestry, and a South Italian Greek-speaking Roman woman from Magna Graecia.

@David, yes, the women are "Ashkenazi-like" (this is why Ashkenazi Jews had to be excluded from the populations!) because Ashkenazi Jews have ancestry from Central and Southern Italy, the result of Judean admixture with (primarily female) Roman converts. It was mostly young men who settled or were forcibly deported to Roman Italy from Judea, so these of necessity (converted) and married local Roman and South Italian women. The proto-Ashkenazi population originated in Central and Southern Italy before about the year 987, when they were resettled by the Ottonians in the Rhineland. We see this heritage most strikingly on the mtDNA.

RobertN said...


thank you for the reference on the Kingsley book regarding the Hyperboreans. This seems to dovetail nicely with the book by Geoffrey Ashe called "Dawn Before the Dawn". Even if the majority of Avars were a Europoid population, it's still entirely possible that the impetus for this culture lay in Mongoloid Altaic shamanic elites. The correspondences are there if you read the books, your Slavic chauvinism notwithstanding.

RobertN said...


The Siberian Altai region also has some things in its favor which could connect it to the Hyperborean myth.

Samuel Andrews said...


Even compared to me, you're jumping the gun too soon. There isn't enough information to come up with such detailed conclusions.

EastPole said...

I think the story of some groups with East Asian component escaping from the steppe and settling among Slavic(Hyperborean) Avars/Abars/Obry in Pannonia shows some similarity to the story of Tatars settling in Poland thousand years later.

“Pseudo-Avars” escaped from the Turks:
“The next author to discuss the Avars, Menander Protector, appeared during the 6th century, and wrote of Göktürk embassies to Constantinople in 565 and 568 AD. The Turks appeared angry at the Byzantines for having made an alliance with the Avars, whom the Turks saw as their subjects and slaves. Turxanthos, a Turk prince, calls the Avars "Varchonites" and "escaped slaves of the Turks", who numbered "about 20 thousand"”

So some ‘pseudo-Avar’ steppe groups escaped from the Turks and settled among Avar/Abar/Obry (Hyperborean) Slavs. Because they were good soldiers they were accepted by Slavic elites.
They married and mixed with Slavs.

Similar stories were observed thousand years later where Tatar groups escaping from the steppe were settling in Poland and adopting Polish language and customs. They were good soldiers and were accepted, later participated in every significant military campaign. The upper nobility of Tatars became Polish nobility. So if you select skulls with Mongoloid morphological traits you will find East Asian DNA among Polish nobles.
But of course Poland was a Slavic country with Slavic population and Slavic elites. So no surprise that following ‘Avar’ period in Pannonia Great Moravia was a Slavic country, with Slavic population and Slavic elites.

MaxT said...

Will you be adding these ancient Avar burial samples to Basal-rich K7 spreadsheet?

Spreadsheet has some Avars but i'm not sure if these are ancient or modern (?), they also lack Asian admixture.

Matt said...

@Davidski, totally off topic, is this python script - popstats by Pontus Skoglund - at all compatible to run with how you have your data stored?

Thinking again recently about how the f3 outgroup stats work and can be decomposed into different combinations of f2 stats (according to this paper, and so was wondering if you could easily use this script to run a set of f2 statistics, to run those simpler measures through some analyses in PAST3.

This script is was what was used to run the f2 statistics in the paper from Rodríguez-Varela on Guanche in late 2017 -

If not, no problem.

Davidski said...


That looks like it might be useful and it's fairly straightforward to implement. I'll check it out this weekend.

Open Genomes said...

@David, about the "trio" of CL38, CL30, and CL25:

CL38 is Y E-V13 and mtDNA X2:
CL38 restricted clustering nMonte3

CL38 is at least 57.6% non-European. To a great degree, his non-European ancestry appears to be something like Anatolian Greek / Cypriot / Lebanese. The European ancestry of CL38 appears to be French, with some East Sicilian.

CL30 is R1b-P312 and mtDNA I1b:
CL30 restricted clustering nMonte3

CL30 is 63.4% non-European, perhaps somewhat less "Anatolian" than CL38, although he's 15% Georgian Laz. CL30 appears to be mixed with Central European, including Germanic, not with South Italians.

CL25 is mtDNA H.
CL25 restricted clustering nMonte3

CL25 is 52.8% non-European, and in that 29.8% Levantine. CL25 appears to be more admixed with both North and South Italians than the others, and includes Greek ancestry.

My take on these is that while they are part Italian or European, their non-European ancestry appears to come from somewhere like Antioch, which would have had large numbers of Greeks, Anatolians, and northern Levantines. I think they are rather too Levantine to be just regular South Italians.

There is a possibility that they are of part-Jewish ancestry.

Removing Ashkenazi Jews from the test, we get the following:
CL38 restricted nMonte3 without Ashkenazi Jews, but including other Jews

CL30 restricted nMonte3 without Ashkenazi Jews, but including other Jews

CL25 restricted nMonte3 without Ashkenazi Jews, but including other Jews

Interestingly, it's CL38 who this way appears to have a substantial Italian Jewish component, 37.8%, although this is a slightly worse fit than without any Jews at all.

This isn't a Jewish cemetery. The first generation is buried with weapons and other grave goods, and even a horse sacrifice.

A more "restricted" look at CL25 with a "distance penalty", which produces a worse fit, shows her to be something like a mix of a Byzantine Anatolian Greek, South Italian, and Sicilian.

CL25 more restricted clustering nMonte3 with a distance penalty

The Levantine component ancestral however needs explaining, since this not typically "Greek" or "South Italian". They don't seem to be "Aegean" either, but more likely of partial East Mediterranean Greek ancestry, mixed with the local Levantine population.

knedokesto said...


Too many unsupported conclusions.

Arza said...

@ Davidski

Is there any chance for adding Alemannic genomes to your spreadsheets? Some of them are intriguingly eastern...

Ancient genome-wide analyses infer kinship structure in an Early Medieval Alemannic graveyard

Philippe said...

@ RobertN

"The Siberian Altai region also has some things in its favor which could connect it to the Hyperborean myth."

Interestingly the Tarim mummies seem to be closely related to celts. They also wore tartan and had red hair. There are apparently similarities between the tartan and textiles found at the Hallstatt salt mine.

Open Genomes said...


Using ancient and Medieval samples, CL25 doesn't look very "Aegean" at all, although she's clearly part Greek, and mixed with Central European:

CL25 restricted clustering nMonte3 pre-Modern samples only

She's 24.4% Levantine. Here overall non-Central European ancestry seems more typical of a Hellenistic Greek from the Eastern Mediterranean. She could be some sort of Greek-Phoenician mix, but it cannot be from North Africa, since any North African component is lacking.

CL30 is 22.2% Levantine Chalcolithic, and 9.2% BMAC. BMAC of course was present in the Bronze Age Levant.

CL30 restricted clustering nMonte3 pre-Modern samples only

CL38 is 40.8% Anatolian Bronze Age, 4.0% Armenia Chalcolithic, 11.3% Minoan Lasithi, and 9.2% Levant Bronze Age South. He's lacking in any Mainland Greek or Balkan.
He seems to be much more Anatolian than the others, but still somewhat Levantine.
I think he's a good candidate to be a kind of Anatolian Byzantine mix with some Levantine, so probably from a region of Anatolia near the LEvant.

CL38 restricted clustering nMonte3 pre-Modern samples only

Of course, all of these have local Central European admixture, even some Germanic admixture, so they seem to be 1/2 to 1/4 "Eastern".

Open Genomes said...

@knedokesto can you give us a better explanation as to why these three individuals would be substantially Levantine and Anatolian, but not in any way North African?
What would you say they are then?

Samuel Andrews said...

GEDmatch results for the northern European samples are very Germanic. Closest to Scandinavians, Dutch, North Germans.

They're ethnic Lombards.

Samuel Andrews said...

Lombards & Alemani results are very similar. Probably basically identical. Both tribes were Elbe Germanic. Do they have any Celtic ancestry? My guess is they're overwhelmingly descended from proto-Germans who lived in Scandinavia.

Chad Rohlfsen said...

Starting up a new post...

Samuel Andrews said...

It looks for sure, among modern people, Lombards are most similar Scandinavians. This is cool because they claimed to have originated in Scandinavia.

Using nMonte for K36 results, all high coverage Lombards score about 100% in Scandinavian countries.

Eurogenes K15
West_Norwegian + West_Norwegian + West_Norwegian + West_Norwegian @ 4.340765

Orcadian + West_Norwegian + West_Norwegian + West_Scottish @ 3.649648

Tabassaran + West_Norwegian + West_Norwegian + West_Scottish @ 6.332824

Swedish + Swedish + West_Norwegian + West_Norwegian @ 8.660437

Davidski said...


I got the Popstats thing running. What sort of stats were you interested in? I'm assuming f2 stats?

Maybe post a few and I'll see if I can produce coherent results.

Matthew said...

CL38 and CL30 look like Greek islanders. CL25 is closer to Sephardi and Itallian Jews.

Matt said...

@Davidski, cheers, thanks for that.

I'm not sure whether you can mass run these stats if I give you a big list and which pops you've got in the data file, so first, to go big, a big set of f2 stats from every combination 113 populations with labels from the G25 list:

(There's a lot of redundancy in this list, but I figure if you've got a script to run a big list and they take seconds, then that's no problem, and this file is not much use if it is).

If that's a no go, then about 46 stats from the main 7 differentiated ancient West Eurasian populations, an African outgroup and a couple of intermediate populations (Unetice and SHG): .If that's not still too long! If it is, I'll have a rethink about if there are any few stats which might be interesting.

For either set I basically want to try and put all these into PAST3 in a matrix (with assumption that f2(X,X) is always 0 and only f2(X,Y) has a positive value) and see how they behave through PCA, PCoA and trees.

Open Genomes said...


What about all that Levantine admixture in CL38, CL30, and CL25?

Excluding Ashkenazi Jews, it's CL38 who looks more like Italian Jews than the others.

I'm thinking that these three are a mix of ancestries from servants / slaves of the Lombards and the Romans before them. It seems that many slaves of the Romans were Hellenistic Greeks of mixed origin from the region of Antioch on the Orontes, and also Black Sea Caucasians from Lazica (northeast Turkey). We also see Gaulish and Germanic ancestry in these three.

Juvenal, Satire III: On the City of Rome (c. 118 CE)

Since at Rome there is no place for honest pursuits, no profit to be got by honest toil---my fortune is less to-day than it was yesterday, and to-morrow must again make that little less---we purpose emigrating to the spot where Daedalus put off his wearied wings, while my grey hairs are still but few, my old age green and erect; while something yet remains for Lachesis to spin, and I can bear myself on my own legs, without a staff to support my right hand. Let us leave our native land. There let Arturius and Catulus live. Let those continue in it who turn black to white; for whom it is an easy matter to get contracts for building temples, clearing rivers, constructing harbors, cleansing the sewers, the furnishing of funerals, and under the mistress-spear set up the slave to sale.

It is that the city is become Greek, Quirites, that I cannot tolerate; and yet how small the proportion even of the dregs of Greece! Syrian Orontes has long since flowed into the Tiber, and brought with it its language, morals, and the crooked harps with the flute-player, and its national tambourines, and girls made to stand for hire at the Circus. Go thither, you who fancy a barbarian harlot with embroidered turban. That rustic of yours, Quirinus, takes his Greek supper-cloak, and wears Greek prizes on his neck besmeared with Ceroma. One forsaking steep Sicyon, another Amydon, a third from Andros, another from Samos, another again from Tralles, or Alabanda, swarm to Esquiliae, and the hill called from its osiers, destined to be the very vitals, and future lords of great houses. These have a quick wit, desperate impudence, a ready speech, more rapidly fluent even than Isaeus. Tell me what you fancy he is? He has brought with him whatever character you wish---grammarian rhetorician, geometer, painter, trainer, soothsayer, ropedancer, physician, wizard---he knows everything. Bid the hungry Greekling go to heaven! He'll go. In short, it was neither Moor, nor Sarmatian, nor Thracian, that took wings, but one born in the heart of Athens.

Open Genomes said...

@David, speaking of Medieval Germanic people, how about some of these in Global25?

Ancient genomes from Iceland reveal the making of a human population 01 Jun 2018

Opportunities to directly study the founding of a human population and its subsequent evolutionary history are rare. Using genome sequence data from 27 ancient Icelanders, we demonstrate that they are a combination of Norse, Gaelic, and admixed individuals. We further show that these ancient Icelanders are markedly more similar to their source populations in Scandinavia and the British-Irish Isles than to contemporary Icelanders, who have been shaped by 1100 years of extensive genetic drift.

Fanty said...


"nterestingly the Tarim mummies seem to be closely related to celts"

R1a Z93 Celts that must have been then, because thats what they had been tested for.

I know that these mummies had been claimed to look "European" or even "Celtic" for decades now and that DNA tests revealed fair hair and fair eyes and everyone was predicting R1b. But turned out to be R1a Z93.

Samuel Andrews said...

Cl38, CL30, Cl25 have little or no Levantie ancestry. European Jews have a lot of Levantie ancestry. So their affinity to European Jews is superficial.

Davidski said...


I may have spoke too soon. I can run D, f3 and f4 stats with this thing, but I can't run f2 stats. Not sure if it's a bug, or I'm missing something. Might have to get in touch with PS.


My gmail seems to be down, so replies to e-mails will have to wait.

Matt said...

@Davidski, understood, if you get running then we can pick up from there.

In meantime, would it be possible to run off this (very long) list of f3 stats?

Basically testing f3 outgroup stats with Ust-Ishim and Papuan against the classic African / Chimp outgroup.

Ust-Ishim and Papuan are not a true outgroup (for modern West Eurasians, in theory because Basal Eurasian is outgroup to them and because of ENA flow with ancient West Eurasians). But wondering if they would be outgroup enough to serve at limiting effects of branch specific genetic drift (e.g. like Kalash vs Pashtun, where you have similar populations on a deeper level effected by different levels of pop specific genetic drift), while poss not having some issues that might affect African / Chimp outgroup.

Philippe said...


So do you think the Tarim mummies were a different group to the Tocharians? Apparently Tocharian is Centum, more closely related to Celtic and italic.

Matthew said...

@ Open Genomes said...
I think CL38, CL30, and CL25 all look like Greek islanders.

Samuel Andrews said...

Btw, CL38 & CL30 & CL25 have way more CHG/IranNeo than the Minoans or Myceneans did. Roughly twice as much. Southern Italians have more of this than Minoans! This, erases the idea Minoans were specially rich in CHG/IranNeo even in a European context.

Minoan-like pops can't explain the CHG/IranNeo-like ancestry in southern Europe. For, Italy it has to be populations more rich in that kind of ancestry such as CL38/30/25. The, Greek islanders obviously need extra layers of this ancestry that the Minoans can't explain.

MaxT said...


Tarim mummies are most likely descendants of Sintashta culture (Steppe_MLBA). They conquered or replaced previously existing Afanasievo/Yamna (Steppe_EMBA) in that region.

Allentoft et al (2015): "Afanasievo culture persisted in central Asia and, perhaps, Mongolia and China until they themselves were replaced by fierce warriors in chariots called the Sintashta (also known as the Andronovo culture)"

Tramin mummies are also very light pigmented like Sintashta (Steppe_MLBA) folks due to higher admixture from CWC, unlike Afanasievo/Yamna (Steppe_EMBA) who were darker.

Philippe said...

Thanks. So who were the Tocharians?

MaxT said...


Tocharian being centum, they were most likely Afanasievo/Yamna (Steppe_EMBA), who were earliest IE migrants to that region.

Sintastah (Steppe_MLBA) came later, conquered and mixed with previous IE group aka Tocharian.

Tocharians would have continued to exist after this mix, preserving old Centum influences (EMBA) and new Indo-Iranian invader influences (MLBA).

Philippe said...


I did a bit more reading. Apparently the mummies tested by Li et al. (2010) were R1a1a but not R1a Z93. This was stated by one of the study's authors:

Also the mummies that they tested were from the Xiaohe cemetary and are all from before 1500 BC (see above).

The celtic-looking mummies, with tartan and red hair apparently date from 1000 BC onwards, they come from a different cemetery and they don't appear to have been tested. The celtic-looking Cherchen Man, for example is dated to 1000 BC and is from the Zaghunluq cemetery:

Please correct me if this is wrong.

This article notes various other similarities between Cherchen man (and the associated mummies) with the Celts:

"Based on the mummy, the museum has reconstructed what Cherchen Man would have looked like and how he lived. The similarities to the traditional Bronze Age Celts are uncanny, and analysis has shown that the weave of the cloth is the same as that of those found on the bodies of salt miners in Austria from 1300BC.

The burial sites of Cherchen Man and his fellow people were marked with stone structures that look like dolmens from Britain, ringed by round-faced, Celtic figures, or standing stones. Among their icons were figures reminiscent of the sheela-na-gigs, wild females who flaunted their bodies and can still be found in mediaeval churches in Britain. A female mummy wears a long, conical hat which has to be a witch or a wizard's hat. Or a druid's, perhaps? The wooden combs they used to fan their tresses are familiar to students of ancient Celtic art."

(The article refers to a dna test, without giving a link, but it was written in 2006 before any y-dna tests were done and I don't know what test it is referring to or whether these particular mummies were actually tested or not).

Davidski said...


Anything European-related in the Tarim Basin and surrounds after 1500 BC is gonna be from Andronovo, and thus R1a-Z93.

Have a look at the map here on page 6...

By the way, the big problem with the Tarim Basin aDNA results is that they're based on PCR, not NGS or capture, which means that despite the best efforts of the researchers to avoid contamination, they might well be the result of contamination. What we need are ancient capture data from the Tarim Basin mummies and actual Tocharian remains.

Matt said...

@Davidski, thanks, I've filtered that list out for labels that match your pops file:

Matthew said...

PCA with all the samples from Lombard period cemetery at Collegno (6th-7th century A.D.), Italy

PCA with all the samples from Lombard period cemetery at Szólád (6th century A.D.), Hungary

PCA with Collegno and Szólád samples

Matt said...

@Davidski, thanks for running that data for me.

Here's a little example of what I was looking to use these stats for, with Levant_N:

First a couple of strongly correlated stats, the stats using Gorilla and Mbuti as outgroup:

They're strongly correlated (and note magnitude is comparable). There are some very slight differences, which may reflect archaic ancestry, or African admixture, or just statistical ascertainment errors.

Another pair B of correlated stats, those using Ust Ishim and Papuan outgroups:

Again strong correlation, with exception of Australian sample, which sharply deviates the line towards Papuan (Onge notably does not, nor the East Asians, so this stat is fairly unsensitive to deep sharing between East Eurasians, and by extension most likely by any even deeper "Crown Eurasian" sharing between WHG+ANE and East Eurasians).

Before we move on, also note that Kalash and Pathan have almost identical positions on both of these stats. This means there's no real sign of inflation of differentiation through strong group specific drift in Kalash, as we see should we look at the comparable Fst score measure!

Now to compare an less correlated pair, C, Mbuti and Papuan stats:

What we see here is that as well, as with the obvious direct recent African ancestry effect, the stats which tend to break linearity are those which are rich in Near Eastern ancestry which is divergent to that in Levant_N! These tend to be both ancient Iranian and to an extent recent Arabian/Natufian like influenced.

Looking at the same pair less correlated pair of equivalent Yamnaya stats shows a similar phenomenon: Populations with divergent Near Eastern ancestry much less close to Yamnaya in the Mbuti conditions, so e.g. CWC_Baltic_Early or CWC_Germany, which are very close to Yamnaya in the Papuan stat, are subtly pushed down the "rankings".

Matt said...

Now this probably all seems very obvious so far, but I think it helps explain why:

1. Sardinian, among moderns, converges so closely with all ancient South European samples; it's very rich in "West Farmer" ancestry in a stat that is very sensitive to even slight differences in "East Farmer" vs "West Farmer" vs "Natufian".

2. Likewise, Lithuanians tend to be pretty close to ancient Steppe influenced North Europeans, because they are relatively poor in "West Farmer", through a combination of high Euro HG, less Neolithic ancestry and a relatively low West:East farmer ratio.

This suggests to me that when we're trying to look at very fine distinctions in the rank order of which populations are closest to which ancients (e.g. Bronze->Iron Age Europeans) , and we want a measure that functions like "Fst without group specific drift", then I suspect that the f3 stat with Ust / Papuan as outgroup may be closer to that. So, we *may* be better off running a f3 outgroup stat with Ust Ishim or Papuan as a complement to what is normally run with an African outgroup or outgroup basal to Homo as well. The shifts are fairly slight in the grand scheme of things, but if we're dealing with such fine scale ancestry, it can matter.

As to why the Mbuti and basal to Homo outgroup have this behaviour, not clear what this reflects, whether it's something real (e.g. African->Basal Eurasian somehow?) or some kind of statistical measure issue problem. Hopefully this is more sorted in future.

Few more graphics just to firm up ideas. Here's some more sets of pairs like pair C: There are some really obvious deviations in "distance" to Mycenaean and Barcin, but also some stuff visible at the very fine grain for Beaker_Britain and Baltic_BA.

Using just the Ust Ishim stats to build PCA and trees;

vs using just the Mbuti stats:

The Ust Ishim stats seem to build trees which are a little more structured among present day Europeans at least, and a little easier time structuring the Near East together.

Ebizur said...

Just in case any readers have missed my post at Anthrogenica regarding this recently published Chinese paper:

"A study of genetic diversity of three isolated populations in Xinjiang using Y-SNP"

LIU Shuhu, NIZAM Yilihamu, RABIYAMU Bake, ABDUKERAM Bupatima, DOLKUN Matyusup

College of the Life Sciences and Technology, Xinjiang University, Urumqi 830046

Abstract: The Keriyan, Lopnur and Dolan peoples are isolated populations with sparse numbers living in the western border desert of our country. By sequencing and typing the complete Y-chromosome of 179 individuals in these three isolated populations, all mutations and SNPs in the Y-chromosome and their corresponding haplotypes were obtained. Types and frequencies of each haplotype were analyzed to investigate genetic diversity and genetic structure in the three isolated populations. The results showed that 12 haplogroups were detected in the Keriyan with high frequencies of the J2a1b1 (25.64%), R1a1a1b2a (20.51%), R2a (17.95%) and R1a1a1b2a2 (15.38%) groups. Sixteen haplogroups were noted in the Lopnur with the following frequencies: J2a1 (43.75%), J2a2 (14.06%), R2 (9.38%) and L1c (7.81%). Forty haplogroups were found in the Dolan, noting the following frequencies: R1b1a1a1 (9.21%), R1a1a1b2a1a (7.89%), R1a1a1b2a2b (6.58%) and C3c1 (6.58%). These data show that these three isolated populations have a closer genetic relationship with the Uygur, Mongolian and Sala peoples. In particular, there are no significant differences in haplotype and frequency between the three isolated populations and Uygur (f=0.833, p=0.367). In addition, the genetic haplotypes and frequencies in the three isolated populations showed marked Eurasian mixing illustrating typical characteristics of Central Asian populations.

Key words: Isolated population; Y-chromosome; SNP; Haplogroup; Genetic diversity

Citation: Liu SH, N, Yilihamu, R Bake, et al. A study of genetic diversity of three isolated populations in Xinjiang using Y-SNP[J].
Acta Anthropologica Sinica, 2018, 37(1): 146-156

1/39 = 2.56% C3c1
1/39 = 2.56% G2a2b2a1a2a
1/39 = 2.56% H
1/39 = 2.56% H1
1/39 = 2.56% J1a2b3
10/39 = 25.64% J2a1b1 (M260, Page14, M92)
1/39 = 2.56% O3a2c1a
1/39 = 2.56% R1b1a1
8/39 = 20.51% R1a1a1b2a (L342.2, S278.2, Z94)
6/39 = 15.38% R1a1a1b2a2 (Z2124)
1/39 = 2.56% R1a1a1b2a2a
7/39 = 17.95% R2a (P249, L266, PF6108, P267, PF6109, M124)

2/64 = 3.13% C3c1
1/64 = 1.56% H1
28/64 = 43.75% J2a1 (L26, PF5110, Page55, S57, L27, PF5111, S396)
9/64 = 14.06% J2a2 (L581, S398)
1/64 = 1.56% J1a2b
5/64 = 7.81% L1c (M357, L1307)
1/64 = 1.56% N1c1a1a
1/64 = 1.56% O2a1a
3/64 = 4.69% O3a2c (F130, F131, P164, CTS4723, F422, F299, F422, F427, CTS11109, CTS12099)
1/64 = 1.56% O3a2c1a
1/64 = 1.56% Q1a1a1
1/64 = 1.56% R1b1a1
1/64 = 1.56% R1a1a1b2
2/64 = 3.13% R1a1a1b2a2
1/64 = 1.56% R1a1a1b2a1a
6/64 = 9.38% R2 (M479, PF6107)

3/76 = 3.95% C3b
1/76 = 1.32% C3f
5/76 = 6.58% C3c1 (M77, M86)
1/76 = 1.32% D1a
2/76 = 2.63% D3a
2/76 = 2.63% G2a
1/76 = 1.32% G2a1a1a
1/76 = 1.32% G2a2b1b
1/76 = 1.32% G2a2b2a
1/76 = 1.32% H2
3/76 = 3.95% J2a1
3/76 = 3.95% J2a2
1/76 = 1.32% J1a2b2
1/76 = 1.32% N1b
1/76 = 1.32% N1c2
1/76 = 1.32% N1c2b
1/76 = 1.32% N1c1a1
2/76 = 2.63% O3a2c
1/76 = 1.32% O3a2c2
1/76 = 1.32% O3a2c1a
2/76 = 2.63% Q1b1
1/76 = 1.32% Q1a2
1/76 = 1.32% Q1a1a1
1/76 = 1.32% Q1a2a1c
1/76 = 1.32% Q1a2a1c1
2/76 = 2.63% R1b1
1/76 = 1.32% R1b1b
1/76 = 1.32% R1b1a1a
7/76 = 9.21% R1b1a1a1 (M478)
1/76 = 1.32% R1a1a1b2
1/76 = 1.32% R1a1a1b2a
1/76 = 1.32% R1a1a1b2a1
4/76 = 5.26% R1a1a1b2a2 (Z2124)
1/76 = 1.32% R1b1a1a2a2
6/76 = 7.89% R1a1a1b2a1a (Y7)
4/76 = 5.26% R1a1a1b2a2a
5/76 = 6.58% R1a1a1b2a2b (Z2122)
1/76 = 1.32% R1a1a1b1a2b3
1/76 = 1.32% R2
1/76 = 1.32% T1a1a

Ebizur said...

These peoples would be considered "Uyghurs" according to the PRC's official classification. However, the Keriyaliks and the Lopliks are linguistically and culturally somewhat distinct from other Turkic-speaking populations of NW China who are now officially grouped together as Uyghurs. The Lopliks are known for their traditional fishing culture and associated words in their strange dialect. The Keriyaliks are famous mostly for their inhabiting a remote oasis in the south-central part of the Taklamakan Desert.

Matt said...

OK, so one more thing on the stats discussed above, some graphs taking the opportunity to run those stats in my above post against Fst to see the correlations:

Correlations are dominated by the general characteristic in which the "distance" as measured by Fst is a) very much higher between isolated ancient West Eurasian populations than it is in either of the f3 based measures, using Mbuti or Ust Ishim as an outgroup, and b) noticeably higher in drifted subpops (e.g. Kalash).

Within that, I also see clear pattern where at the low end of distances, the Mbuti stats have lose fine scale correlation with Fst for increasingly Near Eastern populations (so this is more apparent Iron Gates->Yamnaya->Barcin->Levant N).

Simon_W said...

What we have to keep in mind when discussing the isotopic evidence for local or non-local origins in Collegno: The authors had checked the isotopic levels within a day's march around Collegno - so non-local doesn't necessarily always mean non-local to northern Italy!

Matt's PCA with the Collegno samples looks highly interesting. Some observations:

The female CL47 is an interesting case: Here isotopic levels suggest local origins, but in the PCA she's close to the French Swiss. And she was buried with grave-goods from the Burgundian area (southeastern France or western Switzerland), which matches her DNA. I'd say she could be of western Swiss Gallo-Roman origin, in spite of having grown up in Collegno.

CL23, CL36, CL57 and CL94 look more or less normal North Italian. CL36 is close to modern Emilians. I made a wild nMonte run just to check what he could be composed of:

[1] "distance%=2.3663"



Seems to be predominantly a mix of Iron Age east-central Italians (if they happened to be similar to LBA Dalmatia) and some substantial Celtic admixture. But some later exotic accretions are not lacking, especially Samaritan-like admixture. Very interesting. I myself (1/4 from Romagna) get considerable Samaritan-like admixture in this sort of test.

Then, CL121 looks like having roots somewhere in the northern parts of southern Italy or in the southern parts of central Italy, judging from Matt's PCA. Somewhere between Lazio/Umbria/Marche on the one hand and Abruzzo/Molise/Campania on the other hand, with a slight Sardinian shift. A made the same wild nMOnte run with this individual:

[1] "distance%=2.247"



Again, like in CL36, the strongest type of ancestry is LBA Dalmatian-like. Seems to be typical for east-central Italian Iron Age folks. The exotic admixture is now rather from Anatolia and the northern Levant. And indeed, the Beaker_Sicily reflects the Sardinian-like shift I noted.

In the PCA CL36 and the Emilians look fairly close to the Tuscans. So I had a go at these as well:

[1] "distance%=0.7578"



I don't know how significant this is, but they come out rather different. Strong central European Beaker affinity here, probably reflecting the elevated Protovillanovan ancestry in the area. Strong Peloponnese_outlier and Anatolia EBA as well, possibly related to the Etruscans! LBA Dalmatian-like ancestry on the other hand low, quite unlike in the other samples. Apart from that there is various minor exotic admixture, especially from Greece and Anatolia.