search this blog


Saturday, November 26, 2016

Big deal of 2016: the territory of present-day Iran cannot be the Indo-European homeland

It's been a pretty big year. Not as big as many of us had hoped for, but there's still a few more weeks till 2017, so who knows what will happen?

In any case, for me 2016 will be the year in which we finally saw hard data - courtesy of Lazaridis et al. and Broushaki at al. - that crossed off the territory of present-day Iran from the list of potential Proto-Indo-European (PIE) homelands.

Now, it's true, Lazaridis et al. and Broushaki et al. were somewhat vague in what their data meant in this context, and indeed, both sets of authors left open the possibility that what is now Iran might prove to be the PIE homeland. However, their data left no doubt: look elsewhere for the PIE Urheimat.

Why? Here it is, in point form, as simply as I can put it:

- the Mesolithic and Neolithic peoples of the Zagros range and South Caspian region were highly distinct both in terms of genome-wide genetic structure and uniparental markers, and clearly only contributed meaningful gene flow to South Asia, not to Europe or Anatolia

- after the descendants of Neolithic farmers from the Zagros range, or at least their very close relatives, migrated to South Asia, the territory of present-day Iran, as per the data in Lazaridis et al. and Broushaki et al., saw waves of migrations from the west and north that dramatically shifted the population structure of the region, bringing it closer in this respect to the Levant and Europe

- most of the Neolithic samples from Iran in Lazaridis et al. and Broushaki et al. came from near the proto-Elamite homeland in the Kor River basin in central Fars, which strongly suggests that their close relatives who streamed into South Asia were not Indo-Europeans, but rather speakers of languages closely related to Elamite

- even though Lazaridis et al. successfully modeled Early/Middle Bronze Age steppe populations, including Yamnaya, as largely of Iran Chalcolithic origin, this appeared to be a coincidence, because the Chalcolithic samples from Iran showed fairly typical South Caspian uniparental markers, such as Y-chromosome haplogroups J and G1 and mitochondrial (mtDNA) haplogroup U7, which are conspicuous by their absence from an exceptionally wide range of Bronze Age steppe samples.

I do realize that many readers won't accept these arguments for emotional reasons, because the PIE homeland debate is an emotional one for a lot of people. Nevertheless, if you decide to argue with me in the comments, make sure you have a coherent argument.

Actually, the territory of present-day Iran has never been a serious contender in the PIE homeland debate. This is something that many Iranian scholars will freely admit.

Not only was central Fars the proto-Elamite homeland, but much of what is now western Iran was the stomping ground of the Hurrians. The Indo-Europeans, in the form of Indo-Iranians such as the Medes, only got there late in the game, in all likelihood from the Eurasian steppes.

But a lot of newbies to the PIE debate don't know this, or they don't want to know it. I guess to them, Iran makes sense because it's in between Europe and South Asia? Or maybe it's the fact that the word Iran is kind of similar to the word Aryan? No idea, really, but like I say, it's now all over.


mickeydodds1 said...

David, if you think that endless quibbling and complaining, here on this particular blog, will cease, then you are sorely mistaken.
Nothing but nothing will even moderate let alone stop the quibbling. In fact, I confidently predict that the quibbling will continue long after the blog's own eventual demise.

Or EinSof said...

Who proposed Gutians were Indo-Iranians? I've heard that they might be related to Tocharians, but not Indo-Iranians.

Davidski said...

Who proposed Gutians were Indo-Iranians? I've heard that they might be related to Tocharians, but not Indo-Iranians.

Fair enough. I'll change it to the Medes.

FrankN said...

Dave: "most of the Neolithic samples from Iran in Lazaridis et al. and Broushaki et al. came from near the proto-Elamite homeland in the Kor River basin in central Fars"
The samples come from Kermanshah Province, which is in the basin of the Karkeh River, a Tigris tributary.
That area is mostly Kurdish today, neighbours Lor, and is approx. 1,000 km NNW of Fars. Any connection towards a "proto-Elamite homeland" is highly speculative, given the little we know about Elamite language.
"In the central Zagros the best known Bronze Age sequence comes from excavations at Godin (Gowdīn) Tepe in the Kangāvar valley (Young and Levine). In the last quarter of the 4th millennium an enclave of lowland traders (or indigenous admini­strators with strong ties to the [Mesopotamian] lowlands) had been established there (periods VI-V; see Weiss and Young). (..)
The oval enclosure at Godin was appar­ently abandoned in some haste, for numerous pots and other objects were left on the floors of the buildings. Following a brief (?) hiatus the settlement was reoccupied around 2700 B.C. by people with a very different material culture (Godin IV, “Yanik period”; see “Northwestern Iran” below), including dark, burnished pottery with incised and white-filled decoration. These people had apparently migrated to the Kangāvar area (and to the Qazvīn and Malāyer plains) from north­western Iran and ultimately from across the Caucasus (Burney and Lang, p. 59).

Dark burnished ware is the hallmark of the Kura-Araxes culture. Further illustration on the expansion of Kura-Araxes, which continued further towards the SE during the subsequent Aras culture:

One of the earliest attested Indo-Iranian words is baghdadu "god-given" ->Baghdad, in the Code of Hammurabi (early 18 cBC). Baghdad lies some 450 km W of Kermanshah.

Various Sumerian and, to a lesser extent, Elamite words (e.g. kura "to sear" < PIE *gwher "hot, burn"; pa-at "foot" < PIE *ped) can be traced back to PIE, suggesting it was once spoken not too far from Elam and Sumer.
have been traced back to PIE.

Davidski said...

Read more carefully what I said: there were no migrations from Iran and surrounds to the steppe.

You don't see that as a problem?

Azarov Dmitry said...

@Davidski Read more carefully what I said: there were no migrations from Iran and surrounds to the steppe.

Distribution of relic subclades of Ra1 and R1b haplos as well as archeological and linguistic evidence makes the Iranian Plateau the best choice as a source of Ra1 and R1b folks. You overestimate importance of autosomal DNA data while in fact it can be unreliable and even misleading due to limited number of availibal ancient DNA samples.

Davidski said...


We only deal with hard data here, not soft data like you.

FrankN said...

Dave: "there were no migrations from Iran and surrounds to the steppe."
As far as archeology tells us, there were migrations (or at least intensive cultural exchange) between the Urmia lake region (Leyla Tepe) and the NE Pontic (Maykop), and the former lies in Iran.
You might want to be a bit more specific when it comes to geographical statements.

Davidski said...


Hard data. Not soft data.

FrankN said...

Preferably, as we are dealing with a linguistic question here, "hard linguistic data"! That data is in my post above. Mind dealing with that first, before discussing aDNA that it inevitably "soft" when it comes to linguistics?

Davidski said...

There's a consensus among historical linguists that the PIE homeland was on the steppe.

So you just admitted that there's nothing to argue about.

FrankN said...

Dave - if you truely believed there was unanimous consensus about the PIE homeland, you wouldnt have issued this post.

Davidski said...

Sure I would, and I did, because I want to show that ancient DNA backs historical linguistics.

Estess said...

Dave, newbie question here, what are you implying when you say that western Iran was the stomping ground of the Hurrians? Thanks

FrankN said...

You mean, like the historical linguistics discussed here:

Consensus, eh?

Why don't you acknowledge the multitude of open issues with the Steppe theory:

- unable to explain the genesis of Hitthite, Italic and Celtic (Gimbutas'/ Anthony's theory on the latter two - CA spread from the Steppe along the Danube - has clearly been disproven by aDNA);
- unable to explain PIE influence in Sumerian and Elamite;
- unable to explain the similarity between PIE and Proto-Semitic (3/4 consonant roots, semantic differentiation by Ablaut, various shared roots, etc.)

All those are "hard" linguistic facts.

Rob said...

From recollection, most Yamnaya mtDNA is already present in Mesolithic and Neolithic central - eastern Europe, and any movement from central Asia / northern Iran had already occurred by the Neolithic, and some subsequent movements were more circumscribed around the Caucasus.
Yamnaya as a phenomenon represents the culmination of development of preceding, Eneolithic groups from the steppe and adjacent syncretic pastoral-farmer cultures, of which 2 or 3 distinct sub-groups expanded somewhat in succession. One can further attempt to hypothesize where from exactly, but I won't for now.

Davidski said...


Figure of speech. Just means they were there.

Estess said...

Dave, I understood the expression. My question was related to how the Hurrian being there ties to the reference about Fars being the proto-Elamite homeland. Thanks

Davidski said...

It doesn't.

But it's interesting that the non-Indo-European Hurrians were native to northwestern Iran and surrounds, and they were ruled by an Indo-European Indo-Aryan elite of foreign origin.

FrankN said...

Here is an analysis of Hurrian-Sumerian lexical matches - 6 out of 65 analysed word pairs, 2% probability for chance occurence. The most probable explanation brought forward is Hurrian SE-wards expansion with the Kura-Araxes culture.

Now, there is also quite some evidence of PIE ("Euphratic") influence on Sumerian. And, actually, the Hurrian-Sumerian matches are mostly substrate terms (hand, dog, liver, meat, rain, who?), while the IE ones tend to be adstrate (trade terms like wool, copper, cattle, flour, vessels) or even, possibly superstrate (e.g. Sum. dug "to say" ~ Lat. dicere, PIE *deyḱ "to point out", Sum. ner "prince, hero" < PIE *h2nēr "man, hero"). Makes me wonder whether the "Hurrian people, IE elite" pattern started just with the Mittanni, or was in fact already present within (eastern) Kura-Araxes.

"the non-Indo-European Hurrians were native to northwestern Iran and surrounds"
From the link above: “The Hurro-Urartian (in the following: HU) linguistic family consists of two closely related languages: Hurrian (with several dialects) and Urartian. Historical Hurrian was spoken in the southeast of present-day Turkey, in northern Syria and northern Iraq at least from the 2nd half of the 3rd millennium to the end of 2nd millennium BC. Urartian is attested in the 1st millennium BC as a language of the Urartian empire (present-day Armenia and neighboring areas).

Several Russian linguists have connected Hurro-Urartian to NW Caucasian languages. Whatever homeland ("native to") that ultimately yields, NW Iran was at best a periphery to it.

Ten203 said...


Were R1b "Indo-Europeanized" by R1a folks? from current data we have, which haplogroup is most likely to be Proto-Indo-European? It seems like R1a Indo-Europeanized R1b.

Lista04 said...

Hello @FrankN

A little correction: those two people have attempted to connect Hurro-Urartian to "NE" Caucasian (not to "NW" which has no known relation to either).

huijbregts said...

I made a fuzzy clustering of the datasheet in the 'high adventure' thread (12 nov).
The results have much improved, thanks to the use of the Gustafson-Kessel algorithm.
One of the eight clusters has the next membership values

Corded_Ware_Poland 0.958
Unetice_EBA 0.944
Corded_Ware_German 0.936
Alberstedt_LN 0.919
Srubnaya 0.897
Bell_Beaker_German 0.877
English_Cornwall 0.870
Dutch 0.862
Sintashta 0.861
Scottish 0.859

This appears to be a Steppe_LNBA cluster, lower in this column we find most of the Northern European populations. Iranians are nowhere to be seen here.
But some Iranians have a considerable presence in a Russian cluster: Hotu 0.745, Iran_Late_Neolithic 0.745, Iran_Neolithic 0.670.
The membership matrix can be downloaded in GK4_8.csv

Olympus Mons said...

On the linguistic side. What do you think about the fact that if you go to the core words of a PIE homeland, most likely the only people that would probably understand them speaking these days, I really mean today, would be the Portuguese? Those words sound amazingly the same. PIE and Portuguese
A Original PIE guy would talk about his cows (wakkā/Vaca) and his Pigs (porko/Porco) and his plough (Arājō/Arado) and about his dog (kwōn/Cao) and the goat (kaprā/Cabra) and how he hunted deer (kerwos/Cervos).

Even Spanish dos not have some of those words . Heck the only people that use the word for hill (Doenti/Doente) are the Portuguese.

We all know that the Portuguese do identify themselves as Lusitanian and by pretty much AD all others spoke Celtic, Ligurian, etc so much younger them Proto-celtic, proto-italic, but the Lusitanian is still identified at that time as PIE. Heck all others had by them loose the -P and the Lusitanian still retain it.

So the lusitanians, that lived in the land of the origin of Bell beakers spoke PIE far longer than anybody else in western europe?

OH, and does anyone knows if those guys at Max plank still are saying that my shulaveri-Shomu were the origin of PIE? :)

Olympus Mons said...

correction "for ILL (Doenti/Doente) are the Portuguese.

Grey said...


"Makes me wonder whether the "Hurrian people, IE elite" pattern started just with the Mittanni, or was in fact already present within (eastern) Kura-Araxes."

It's interesting reading about the historical and modern Fulani, Toureg, Hausa etc of Africa and seeing how the pattern of mobile pastoralists dominating settled populations seems to constantly recur.

The other thing is how often they displace settled populations.


"- unable to explain PIE influence in Sumerian and Elamite;
- unable to explain the similarity between PIE and Proto-Semitic (3/4 consonant roots, semantic differentiation by Ablaut, various shared roots, etc.)"

Logically doesn't that imply either
1) PIE homeland must have been in contact with Sumer/Elam/Semitic
2) the people of Sumer/Elam/Semitic started in contact with the PIE homeland and then moved.

So if DNA says the PIE homeland was on the steppe then logically the only option is 2.

FrankN said...

@Grey; " if DNA says the PIE homeland was on the steppe then logically the only option is 2. - the people of Sumer/Elam/Semitic started in contact with the PIE homeland and then moved."

Logically - that explains why Bedouins are full of Steppe ancestry.

Shaikorth said...

Aren't various proposed Sumerian - X contacts (IE, Uralic, Austroasiatic, Basque, Turkic...) considered non-mainstream theories in linguistics?

By non-mainstream I mean accepted as widely or less widely as this one:

Gioiello said...

@ Olimpus Mons

"On the linguistic side. What do you think about the fact that if you go to the core words of a PIE homeland, most likely the only people that would probably understand them speaking these days, I really mean today, would be the Portuguese?"

Don't speak about what you don't know. All these Portuguese words come from Latin.
If your Shulaveri theory is like that, it is ridiculous and done from someone who speaks about what he doesn't know.

"correction "for ILL (Doenti/Doente) are the Portuguese".
In fact "doente" is like Italian "dolente" from Latin "doleo", present partiple: dolens, dolentis, dolenti, dolentem, dolens, dolente as mons, montis, monti, montem, mons, monte.
Shut up! You are ridiculous.

MfA said...

Wow that archatlas website is amazing, thanks for sharing. Another great presentation from there.

Rob said...
This comment has been removed by the author.
Olympus Mons said...

Gioello.... Where does Latin como from?
If you dont know that those words ( like doenti) were in lusitaian writings, with archaism older than any proto (proto celtic , proto ligurian, etc) let alone latin or celtic itself, then... Tell me again what is your expertise again?

So... Porcom is latin? Not suinus? Or wakka? ... Gioello... Give me a break.

There is only one language that still pronounces those words like PIE ( eg kwon that is said exactly as cao in portuguese).
That has to explained.

Olympus Mons said...

What is that suppose to mean?

Rob said...

Just a reference for yours, anyone's parousal
It's a book floating around on the web. It suggests a stratigraphic gap just prior SS.

Samuel Andrews said...

"From recollection, most Yamnaya mtDNA is already present in Mesolithic and Neolithic central - eastern Europe, and any movement from central Asia / northern Iran had already occurred by the Neolithic"

U4 and U5a do very rarely pop up in Mesolithic and Neolithic Central-East Europe. T1a pops up a few times in Neolithic Europe. J1b1a1 and I amoung others don't. Those typical Yamnaya haplogroups appear in pre-Steppe affected Central/East Europe because they were distantly related to Yamnaya not because there's no signifcant Steppe ancestry or because EHG/CHG were in Central/East Europe in the Neolithic.

I've looked at LNBA Central European mtDNA in detail. There's a huge Steppe presence. There's no room for debate on this topic. There's a huge amount of Steppe mtDNA in modern Europeans as well and some in SC Asia.

Rami said...

Everyone knows the PIE homeland is in the Pontic Steppe/North Caucasus but denying the peoples of that area have no connection with Iran, is a bit ridiculous, there are connections. Kotias and Neolithic Iranians are closely related to each other. The diversification of R points to Iran as well.

Davidski said...

Kotias and Neolithic Iranians are closely related to each other.

Different enough.

The diversification of R points to Iran as well.

This claim is based on modern DNA, which can be very deceptive.

Rob said...

No, you've misunderstood
I agree that there is steppe mtdna in Central Europe, and most of it was already present in the Carpathian-Pontic region by the Neolithic , with some additional flow from nearby 'east-central european' farmers (eg from Romania / Bulgaria . So this comments were more in regard to the issue of gene flow from Iran

EastPole said...

I got similar results with Gustafson-Kessel algorithm. But I still have doubts whether this method is correct.
After removing outliers and doing PCA to the data we have 4 relevant dimensions on which we do fuzzy clustering.
But these dimensions are not of equal importance.
Their eigenvalues are:

In your Steppe_LNEB cluster you have Sintashta and Irish and Scottish while Eastern Europeans like Poles or Russians are in different cluster. This cannot be correct.
On PCA plot showing all dimensions I marked Sintashta red squere, Irish blue and Polish yellow.
You can see that in 2 most important dimensions i.e. 1 and 2 Polish is closer to Sintashta while in two less important dimensions 3 and 4 Irish is closer to Sintashta:

If all dimensions were of equal importance the clustering would be correct. But they are not of equal importance, first and second dimensions explain 82% of the variance while third and forth 15% of the variance.

When Gustafson-Kessel algorithm is applied to the first three dimensions only I got following clusters:

Sintashta is in Eastern European cluster with Russians and Poles.

Davidski said...

By the way, if anyone's wondering if the Ancient North Eurasians (ANE) originally came from Iran, nah, they probably arrived in Iran well after MA1's time.

That's because the ANE in Iran_Neolithic appears to be more closely related to the more recent AfontovaGora3 than to MA1. And that's despite the fact that AfontovaGora3 is damaged to shit.

Yoruba Iran_Neolithic MA1 AfontovaGora3 D 0.0084 Z 1.076 SNPs 152797

Yoruba Iran_Neolithic Bichon AG3 D 0.0145 Z 2.097 SNPs 207372
Yoruba Iran_Neolithic LaBrana1 AG3 D 0.0219 Z 3.282 SNPs 202184
Yoruba Iran_Neolithic Loschbour AG3 D 0.0208 Z 3.122 SNPs 205080
Yoruba Iran_Neolithic Villabruna AG3 D 0.018 Z 2.715 SNPs 198754

Yoruba Iran_Neolithic Bichon MA1 D 0.0043 Z 0.796 SNPs 530552
Yoruba Iran_Neolithic LaBrana1 MA1 D 0.0115 Z 2.099 SNPs 509319
Yoruba Iran_Neolithic Loschbour MA1 D 0.0106 Z 2.026 SNPs 525613
Yoruba Iran_Neolithic Villabruna MA1 D 0.0052 Z 0.946 SNPs 468082

Yoruba AG3 Caucasus_HG Iran_Neolithic D -0.0145 Z -2.354 SNPs 207474
Yoruba MA1 Caucasus_HG Iran_Neolithic D -0.0192 Z -3.647 SNPs 531080

Ebizur said...


I am skeptical about the value of such unsystematic comparisons of subjectively similar lexemes. Case in point, Korean has 발 (/par/ [paɭ] ~ [pa.ɾV]) "foot" and 그을 (/kɯɯr-/ [kɯ:ɭ] ~ [kɯ:ɾV]) "to be singed, to become sooty, to become tanned," which seem to be matches for *ped and *gwher as specious as pa-at and kura.

I have performed comparisons of various pairs of languages that are very unlikely to be related in any meaningful sense and found that between 10% and 20% of their basic vocabulary will match to a similar degree as the examples you have provided through random chance.

FrankN said...

@Samuel Adams:

- Blätterhöhle (D), ca. 8650 BC (2x)
- Les Closeaux (F), ca. 8000 BC
- Falkensteiner Höhle (D), ca. 7200 BC
- Inzigofen (D), ca, 6700 BC
- Les Vignolles (F), ca. 6370 BC
- Motala (SE, ca. 5700 BC (5x)
- Stora Förvar (SE), ca. 5400 BC
- LBKT Hungary
- Derenburg (LBK), ca. 5000 BC
- Brześć Kujawski (PL, Lengyel), ca. 4350 BC
- Bom Santo (POR), 3780 BC
- Fuente Hoz, Alava (ESP), ca. 3200 BC
- Benzigerode (D, Bernburg). ca. 3000 BC
- Ostorf (D, FB/Forager), 2950 BC

- Stora Förvar (SE), ca. 6900 BC
- Bad Dürrenberg (D), 6850 BC
- Spignas (LIT, Kunda Culture), 6350 BC
- Yuzhnyy Oleni Ostrov (Karelia), ca. 5500 BC (3x)
- Popova (Karelia), ca. 5000 BC (2x)
- Gurgy Les Noisats (F, RRRP), ca. 4700 BC
- Alto de Reinoso (ESP, Megalith.), ca.3650 BC
- Barcelona (ESP), ca. 3250 BC
- Perdigões (POR), ca. 3150 BC
- Pitted Ware Sweden, 3200-2000 BC (12x)
- Lánycsók (HU, Starcevo)

"Very rarely..". Pls. do your homework next time before posting !

mtDNA J1b hasn't so far been identified in Yamnaya. J1 (no further sub-clade determined) has been reported 4x from Treilles (F), twice from Obernai (F), plus once from Gurgy (RRRP). So, J1b may as well have arrived in Germany with Michelsberg from the West, as from the Steppe, or from anywhere else.

mtDNA I has been found once in Yamnaya - I wouldn't call that "typical". The earliest report, AFAIK, is from Paternanbidea (Navarra), ca. 5200 BC. Evidence for Vasco-Caucasian, or what do you think, Sam?

FrankN said...

@Ebizur: Both my PIE-Elamite examples were taken from the linked Starostin paper, where etymology and sound comparisons are discussed in far more detail.

Being a Nostratic-ist, Starostin provides also possible Altaian connections for Elam. kuur "to sear", and those may in fact have reached Korean (IMO rather via the Seima-Turbino phenomenon, we are possibly dealing with metalurgical terminology here).

He rejects a PIE connection for Elam. pu(n) "full" (Lat. plenus) for lack of the "l" in Elamite. For a similar reason, he would probably also have discarded your Korean "foot" [Still, the closeness to PIE *ped is intriguing].

Here is another link to that Starostin paper:

As I have noted elsewhere, Elamite appears to be quite a linguistic bastard, with connections to a wide range of language families:
- KA (cuneiform) "fish" is related to Proto-Austroasiatic *ka, Proto-Munda *qa "fish", and was passed on from Elamite to Sumerian and even Hithitte.
- mak "to eat" can be connected via ProtoNDrav *moq to Indonesian makan, but also to proto-Mongolian *meke "to chew, suck". Which of these two routes connects to ProtKor *mak "to eat, drink" I leave you to judge on.
- There are a couple of Elamite-Austronesian parallels that may, or may have not, bridges via Dravidian and Austroasiatic.
- li-im "fire" is connected by Starostin to Proto-Kartvelian *lemec "fire"
- ki "one" has parallels in Omotic and East Cushitic. The same applies to su-ut-me "night" (Dime suut-u, Ari soyt-i, etc.)

Gaspar said...

All facts

1- PIE began in the steppe

2- 1st split from PIE was ~4000BC in Anatolia

Conclusion has to be that a major migration occurred between Steppe and anatolia to force a split on a language.

who migrated?....what HG markers are we talking about

André de Vasconcelos said...


I really don't get your point here, what does Portuguese have to do with PIE? It's just another language derived from Latin, there's nothing special about it.

According to the word for 'ill' is translated into 'aigros', not 'doenti'. I have no idea how accurate it is, but it's irrelevant because the Portuguese word came from Latin, not some obscure IE language.
If you were referring to the 'doenti' found on the Lusitanian inscription of Lamas de Moledo, it actually happens to be a verb akin to the Latin 'dant' (third-person plural present active indicative of 'dō', meaning 'to give'), so there's absolutely no connection.

And yes, as Gio said, all the words you used are derived from Latin, and they all do have a Spanish equivalent, regardless of how common it is used nowadays.
vaca(pt/es), vacca(lat);
porco(pt), puerco(es), porcus(lat);
arado(pt/es), aratrum(lat);
cão(pt), can(es), canis(lat);
cabra(pt/es), capra/caper(lat);
cervo(pt), ciervo(es), cervus(lat);
doente(pt), doliente(es) meaning 'hurting', doleo(lat).

I also find it odd that you say "We all know that the Portuguese do identify themselves as Lusitanian" when there is very clearly no cultural continuity, we know so very little about who the Lusitanians were. We do know they gave the Romans a bloody bad time, which is why they got a special fame other peoples of the region did not achieve. I'd say most of our pre-roman IA cultural heritage comes from further north, even the name of the country gives that away, we're basically just southern Galicians and even genetics hint at us being so similar there's virtually no difference.

Davidski said...

Years ago there was that mtDNA U5a from the Mycenaean Grave Circle B.

That looks rather ominous now, because U5a is common on the Bronze Age steppe. Can't wait for the Mycenaean Y-HGs and genome-wide data.

Rob said...

And what's your genome -wide guess ?

Davidski said...

With enough samples, I reckon they'll be all over the place; from very steppe-like to Greece LN.

I had a brief look at Kum4, which is from that general location and time frame, and wow, he's not much different from the Swedish Battle-Axe genome. Even if the number and quality of markers skewed the results, that's still an interesting outcome.

Jijnasu said...

would be great to see more ancient dna from mycenean greece. Along with data from S and Central Asia it would pretty much end the debate

Samuel Andrews said...


I'm tired of this. Just accept the facts!! Ive already soent hours explaining to people for two years after working less hours on mtdna. Go do the research yourself because I know an hour out of my day won't change your mind. My blog is mtdnaatas

Gioiello said...

@ OM and André de Vasconcelos

I thank André for the exhaustive response to OM (in these hours we sleep in Tuscany).
To OM I would say that in Italy not only there were Latin "porcus", but also "porca" (see Mallory%Adams, The Oxford Introduction... p. 168), and also in Ligurian river "Polcèvera", perhaps from *porko-bhera", see Irish "orc", demonstrating very likely that Lusitanian and Old Celtic languages of Iberia derived from Italy and expanded along the Atlantic corridor, what has also a genetic witness with the Cardials and Bell Beakers, with R1b1 and subclades. We'll see from the next aDNA of Bell Beakers what came from Italy and what formed in Iberia.

Ebizur said...

FrankN wrote,

"[Starostin] rejects a PIE connection for Elam. pu(n) "full" (Lat. plenus) for lack of the "l" in Elamite. For a similar reason, he would probably also have discarded your Korean "foot" [Still, the closeness to PIE *ped is intriguing]."

For what similar reason?

Please note the correspondence set Middle Chinese 出 */t͡ɕʰiuɪt̚/ > Mandarin chū, Korean /ʨʰur/ ([ʨʰuɭ.C] ~ [ʨʰu.ɾV], Japanese shutsu, Vietnamese xuất "to go out, to come out, to exit, to leave, to emerge, to appear; to send out, to put out, to publish, to produce, to offer, to advance (money), etc." for evidence of a */t/ > /r/_# sound change in the history of Sino-Korean. (The same change has affected the Korean pronunciation of every Chinese character whose Middle Chinese reading has ended in */t/.) If you still find it not speciously similar enough, cf. Manchu bethe [pət.xə] "foot (of a human or of a mountain), lower leg," Manchu fatha [fat.χa] "paw, hoof, claw, foot (of non-human animals)," Manchu fatan "sole (of the foot), bottom," or Korean 바탕 [pa.tʰaŋ] "foundation, basis; background, ground; fundamental nature (of a person); bout (of wrestling); field, range (for archery or shooting)."

"- mak "to eat" can be connected via ProtoNDrav *moq to Indonesian makan, but also to proto-Mongolian *meke "to chew, suck". Which of these two routes connects to ProtKor *mak "to eat, drink" I leave you to judge on."

The basic Korean word for "eat" is 먹 /mʌk-/ ([mʌk.C] ~ [mʌ.gV]). It is also used in many contexts in which "eat" would be an inappropriate translation to English, e.g. /mur-ɯr mʌk-/ "to mʌk water" = "to drink water," /mur-mʌk-/ "to get wet, to get soaked; to fail, to flunk (a test)" -- literally, "(something) mʌks water," /sur-ɯr mʌk-/ "to drink an alcoholic beverage," /jak-ɯr mʌk-/ "to take medicine," /ta:mpɛ-rɯr mʌk-/ "to smoke tobacco," /apʰjʌn-ɯr mʌk-/ "to smoke opium," /tʰa:n-nɛ-rɯr mʌk-/ "to breathe in fumes from burning charcoal," /kʌp-ɯr mʌk-/ "to get cold feet, to get scared, to get intimidated." Possibly deeply related words within Korean include /masi-/ "drink; inhale" and */misk/ > /mik͈i/ "bait; lure, decoy."

As for Malay/Banjar/Indonesian makan, you must first compare it with its clear cognates, such as Javanese mangan "eat," Fijian kana "eat; meal," Māori kai "eat; food," Tagalog kain "eat," etc. The morphosyntactic properties of the languages should not be ignored (e.g. the tendency of Austronesian languages to inflect words with prefixes, infixes, and circumfixes in addition to suffixes; Korean has no such tendency, so it is unlikely that Korean /mʌk-/ can be related to the aforementioned Austronesian words for "eat" unless one assumes that Korean is a drastically altered descendant of Malay, Banjar, or Indonesian).

What is your source for Mongolian */meke/ "to chew, suck"? I am familiar with Khalkh /mere-/ "to chew, to nibble, to gnaw (a bone, etc.)," which resembles Middle Korean /mɯr-/ > Modern Korean /mur-/ "to bite, to snap at, to nip; to hold between one's teeth, to have in one's mouth; to latch onto (a patron); to take (bait); to clench, to grit (one's teeth)." I have never heard of a Mongolian word that might descend from a hypothetical */meke/.

It seems to me that such Nostraticists and other long-rangers and lumpers are fond of casting their nets so broadly that they draw in any vaguely similar word regardless of its recent history and etymology.

Gioiello said...

@ Ebizur

"It seems to me that such Nostraticists and other long-rangers and lumpers are fond of casting their nets so broadly that they draw in any vaguely similar word regardless of its recent history and etymology".

I appreciate very much your argumentation. I used a similar one in demolishing the huge work of Dolgopolskij about the link between Afroasiatic and Indo-European, but I have always considered even the monogenesis theory of the language of Alfredo Trombetti.
In genetics it seems to me that I am winning all my battles. Why? Because many years ago, when we haden't all the tools we have now, I formulated 4 golden principles about the STRs mutation rate:
1) mutations happen around the modal
2) there is a convergence to the modal as time passes
3) sometime a mutation goes for the tangent
4) there is rarely a multistep mutation

Thus I ask you if the same may happen in the languages.

Ryukendo K said...

Just dropping this here:

Rami said...

Different enough?
They plot very close together David , Kotias just has a smidgen of Villabruna, while Iran_N has a bit of SW Asian and a hint of South Asian, that does not make them much different. They are closely related to each other, they form the same continuum of peoples, I do not posit that much in uniparental markers. Chalcolithic Armenians have L1 and so do indigenous Austro Munda tribals in India, but are completely different genetically. Point of admixing between Basal Eurasians and Paleolithic Siberians ( ANE) happened on the Iranian plateau at some point.

Olympus Mons said...

( and to Andre, but to him a separate answer)
Maybe this will help (

I am not a linguistic. But before talking about something I read the best experts. Something obviously either of you did (Proper, Blazek, etc).
I could find ten ways to reply. But don’t really know the one that would get trough to you guys. Could try to explain that to ALL LINGUISTICS around the world, Lusitanian is an uncategorized language, to ALL LINGUISTICS Lusitanian is a Strait PIE language with archaism that predate Proto-all (celtic, ligurian , italic). This means that is older and does not derive from!
Even at language structure close to AD, when all languages in Iberia was categorized as Celtic, italic so forth, Linguistics still define Lusitanian as PIE which is strange because is surrounded by other languages that are.

Could explain that Lusitanian did not have a written code, hence it’s written using other languages such as latin code. And was written in roman time, therefore has naturally Latin words…. But this might do the trick!

The text I am posting at the bottom, is the oldest Lusitanian text (Arroyo de la Luz) and to explain it, linguistics (such as Blazek) to explain its ethnonyms, linguistics rely on Latin, Old Indic, Greek, Lithuanian, Avestan, Armenian, etc. What it means is that there is no direct transduction from Celtic, italic, lingurian (or even Proto, proto, proto). Does it do the trick?


huijbregts said...

I want to explain what I find attractive in fuzzy clustering.
I am South Dutch. nMonte calculations show often some admixture that differs from the general Dutch population.
It depends on the nMonte specification but I often find Croatian, Slovanian, Bosnian or Hungarian.
So I am inclined to think that I do have some Balkan admixture, which is supported by the fact that my closest ancient match on GEDmatch is BR2.
The first impulse was to run nMonte with all of the Balkan populations, but as these populations are closely related, the specification would be overfitted and the result pretty useless.
My next thought was: can I somehow compare my DNA with a Balkan cluster?
And here the concept of fuzzy clustering is very promising.
In the classical clustering each item is classified into one specific cluster.
In fuzzy clustering, each item is a member of all the clusters, but with differing membership percentages, which add to 100%.
In my case a perfect result would be:
NW_European LBK Balkan
Dutch 89% 10% 1%
Huijbregts 70% 20% 10%

Thanks for your extensive comment.
It is pretty specific, so my answers will be even more specific. I will not post them here but in my dropbox
Pretty busy now. I hope to complete it on Tuesday.

Olympus Mons said...

Fair enough. But i did not pick random words!
I picked words that must be at the deepest core of PIE. cow, pig,Plough,dog. Words of things that are found with PIE lithics everywhere. I just stated the fact that those words are pronounced in a way that still resembles the way present day portuguese say it. Like Dog (Kwon) whcih sounds like in portuguese (Cao). Even spanish say Perro) and Arajo (plough) surely is pronouced like arado. Remember the portugusese language is prolific in the "-R" sound of the PIE speaking pronouciation.

Its like the fact that the Origin of the word AND for PIE that the lusitianian use INDI and is even these days only used by Porguguese and Galician (AINDA).

I wasn't making a big deal out of it. But remember. If pie came from the east, sure is strange to have an AD language (Lusitanian) at the far end of Western europe that was so unadulture PIE in a sea of already very transformed pie (actually all around them in thousands of miles) and right at the spot of Bell beakers origin.

Olympus Mons said...

@ Andre de Vasconcelos.
You don’t seem to have an idea of your own. You just look for plausible deniability in others arguments. Should I ever become a communist, form a party and need a propaganda “meme” creator… You are hired!

To the point.
Dude. YOUR country was “born” 5400BP. The borders your country has are not a fluke or a random choice by a adolescent prince or the pope. YOUR country was born when 5400 years ago a group of warriors locked the entrance from the Guadiana into the lowlands (alentejo). Each with no more than 100 man, locked near estremoz and the Guadiana river with lots of arrows and even horses. Juromenha , Sao Pedro made larger paraiso possible (over 10 Ha). As just decades later locked the middle Guadiana (Porto carretas, Mercador, etc) made Porto Torrao possible (over 300ha, over 20,000 people).
2000 years later was still valid. Google a map of Phoenician, or carthage… notice something strange in Iberia? Especially carthage, punic wars, Hannibal, etc maps notice how they never crossed that gateway? – It took the Roman empire to get into those “borders”
YOUR country is not south Galicia! Sure Lusitanian is famous for their fierce fighting against the roman but the people that define the borders of your country were the TURDULI tribes. Yes they were Lusitanian. Has were the Galician for that matter or even Cantabrians.
Turduli locked the south, Turduli Bardili south of Tagus, Turduli Oppidani (Oppida- the fortified ones ) north of Tagus river where the Bell beakers arise , the Turduli veteres (Veteres - the ancient ones) south of douro river. –Oh do you know that those areas still have the Higher percentage of R1b in Portugal? – Apart from Guadiana river.
This is YOUR country. Those borders. Are still today the Portuguese borders. You should respect it a bit more.

huijbregts said...

I have doubts some of your points, but you seem to be right that the model is too noisy.
If I add a noise parameter to the Gustafson-Kessel model, a lot of the population get a lot less than 100% membership.
Generally the East_European populations are relatively unaffected, but especially the Indian populations may loose as much as 80% of their membership%.
An alternative interpretation might be that the model needs a lot more clusters.
Anyhow the model needs some reworking.

Gioiello said...

@ OM

I'd want having time to study this inscription, and read many books about that. At a first sight it seems to me due to an Italic dialect linked with Osco-Umbrian. Vilar writes about "puppid": "Formas parecidas con evoluciones fonéticas similares se encuentran en algunos dialectos itálicos: cf osco púd, pid, pitpit (Religión, lengua y cultura prerromanas de Hispania By Francisco Villar, María Pilar Fernández Alvarez, p. 682).
Of course that could just demonstrate how Lusitanian may descend from the the agriculturalists migrated to Iberia from Italy beginning from 7500 years ago (see Zilhao) and with later waves.
Anyway that doesn’t demonstrate anything about your theories.

André de Vasconcelos said...

Firstly a bit of advice, work on your manners and attitude. It's not that the discussion regarding your theories aren't interesting, they could be, but your obnoxiously defensive attitude is horrible. Mocking people who disagree with you, and accusing them of being rubbish doesn't help. When was anyone ever persuaded by being attacked?
And how dare you say I don't "have an idea of my own"? You don't even know me and we very rarely interacted. I don't often post here, and when I do it's mostly to ask for details on something I'm interested - usually to David or Frank - or to point out obvious flaws on things I might know a bit to modestly contribute to the discussion, as was the case.

I'd feel like returning the favour and provocking you back, but I won't drag this blog discussion down any further, I enjoy reading it too much to tarnish it like you do.

You didn't actually addressed the things that were said about those Portuguese words that are "of things that are found with PIE lithics everywhere" - words derived from Latin, not Lusitanian. Anyway, Blazek had the following to say about the Lusitanian language:
"The etymological and grammatical analysis of the lexical and onomastic corpus
of Lusitanian, which is at our disposal, indicates that this language belongs
into the circle of the Western Indo-European languages, with closest cognates
with the Italic and Celtic languages, but undoubtedly different from them, representing its own group comparable with the statute of Italic, Celtic, Germanic,
etc. On the other hand, the specific isoglosses connecting Lusitanian with Indo-
Aryan/Iranian apparently reflect the peripheral archaisms, which are also typical
for Italic and Celtic." (Vaclav Blazek (2006), Lusitanian Language p.15).
This doesn't add up to the idea that it's a "Strait (straight) PIE language with archaism that predate Proto-all". I remain unconvinced by your argument, it seems to be just a Western IE language, but I'm not a linguist either.

Regarding "my country". If you believe it's intimately connected to some unknown blokes who lived five millenia ago, and decide to ignore the impacts of subsequent other peoples and events in the region, go ahead, no one can stop you from doing so. But don't judge people who think differently from you, and don't call out for "respect" on it, especially when no disrespect was done in the first place.

Lastly, regarding R1b, what subclade are you talking about? And what about the values in Spain, where R1b (total) tops in the East rather than (south)West?

Which are you referring to?

André de Vasconcelos said...

That post was meant for OM, which I forgot to add on the header. Apologies. But anyone is free to butt in anyway.

Olympus Mons said...


Simple. Show me italian 7.500 bc or anywhere agriculturalists that had hill settlements oh 1ha averaging 4.6m round huts with 1.5/2m silos surronded by bended walls to create a central space for work with a commom hearth, fixaded on making beads and antrophomorphic figures...enough.

And that is fine. We can start trailing them. If you dont have that then its also fine. It only means that you are having a lot of hallucinations but definitely not talking about humans. Because humans have package of traits by which they live by, that will only change on exteme subjugation events.

So. Point to those agriculturalists traits.

Gioiello said...

@ OM

I am not so polite as André de Vasconcelos and who knows me knows that I am politically incorrect, thus I say that I think that you are too much stupid for meriting my answers, so my discourse with you is closed. About that first migration of agriculturalists from Italy to Iberia 7500 years ago (only the first wave) spoke just a Portuguese, Zilhao, more than 15 years ago.
So far, as an Arbereshe would say,: Ec mirr ka bytha!

Olympus Mons said...

@ Andre de Vasconcelos.
a. All I said was that core PIE words are still pronounced in Portuguese as they were in PIE (Dog- Kuon/cao, cow-Wákka/Vaca, Plough – Arajo/arado, etc). A curiosity. Even Spanish say Perro for dog or changed pronunciation very much. Portuguese still talk with that “-R” Pie sound. Just a curiosity. Same as saying Doenti/Doente for ill (when all others use “newer” words). Or still using the word “Ainda” along side with galicians with are the only that follow that from old PIE from the Lusitanian word INDI.

b. Game over is: Lusitanian retain “-P” long, long, after everyone else had drop it. So Older. Much Older. So older that to explain it most linguistic always reverts back to Indo-aryan, Iranian or Sanskrit. Mostly Sanskrit and Mittani.

c. If Lusitanian language is the remains of the Bell beaker language, It would solve a lot of conundrums. Like the “celtic from the south”. If Celtic language was in part a derive of Lusitanian (which remained archaic) so no need to blow their mind about components of celtic in tartessian which is so much older. So some Celtic components in Lusitanian language and deities can actually be the other way around. But celts themselves are a mix and evolution of the bell beakers mixing with CWC, mixing with whomever. And lots flowing back into Iberia over time.

d. As per Pereira et al – Largest districts of R1b In Portugal:
Setubal 80% r1B (Turduli Bardili) ; Castelo branco, Viseu 75% (lusitanians); Near the coast over - 65% Lisbon, Leiria, Aveiro, Porto (Turduli Oppidani and veteres)….

Olympus Mons said...

that is just your Sunday lunch Wine talking. And we know how you like it. - So, talk to you when you sober up!

I have a bottle of Pure "Touriga nacional" wine for you when we meet someday.

Olympus Mons said...

Just to finish this issue.

Regarding those turduli that better represent structure of the settlement of chalcolithic portugal...
So, turduli... this is what was said about them:

" their exact ethnic affiliation remains obscure. However, recent linguistic studies of the few funerary inscriptions they left behind seem to DEMONSTRATE THAT THE EARLY TURDULI SPOKE AN INDO-EUROPEAN LANGUAGE OF THE ANATOLIAN BRANCH VERY SIMILAR TO MYSIAN, though they later also included people of Celtic, Illyrian, and even Ligurian origin.[1]"

See... Mysian.
Mysian, that earlier were Troy, that earlier were Kumtepe, that (I Say) early Kumtepe (KUM6) was a Shulaveri womem, that came from Shulaveri dispersion from south caucasus, that Max planck guys say were at 4900bc the origin of the PIE language and of the CHG dispersal to a lot of places. - Just a curiosity! :)

FrankN said...

@Ebizur: Thx for your explanations- qualified and fascinating as always!

Mongolian */meke/ "to chew, suck" comes from that Starostin paper (p. 157), without further sources given. Possibly, he is referring to own reconstructions, however qualified they ultimately have been.

"It seems to me that such Nostraticists and other long-rangers and lumpers are fond of casting their nets so broadly that they draw in any vaguely similar word regardless of its recent history and etymology".

Yes and no:
No: There is evidence of "ultra-conserved words" that may have been around for 70,000 years or more, e.g. "I", "Who?" The more frequently specific words are used, the higher is their chance of being conserved.
External assessments:
[Note that even the latter, while overall critical, states "Linguistic science has known for decades of certain “stable” meanings in which originally cognate words are replaced relatively slowly through time."]

Thus, I deem it quite likely that certain "basic" words have been around since the Paleolithic, and as such are present across various language families. "Foot" may be one of these. I have furthermore noted that various language families use *tV/*dV for the semantic cluster "eat/ taste/ tongue/ teeth" ; IE (English) being one of them. Others include Afro-Asiatic, Nilotic, and Uto-Aztecan. *ik(w) or simil. for "water / drink" is also extremely widespread (a.o. IE, Ainu, Hlai, Mayan, E. Khoe), as is *cu "run, leave".

Yes: I aggree that the net may often be cast too broadly. More importantly, we get more and more knowldege about wide-ranging migrations/ contacts during the last 5-8 millennia, be them Trans-Eurasian (e.g. Sintashta/ Seima-Turbino), Eurasian-African, circum-Indian Ocean / Africa (Banana transfer to Cameroon), Trans-Pacific (Coconut, sweet potato), or Trans-Beringian (Dene-Yenissean). Quite a lot of what is landing in the net may reflect Wanderwörter / borrowings related to these migrations, and needs to be sorted out first before the questions of Nostratic/ Eurasian, and/or "Paleo-words" can be addressed in earnest.

André de Vasconcelos said...


This is going way off-topic, and I'm derrailing it further, but anyway..

a. The pronunciation is not the same, although there are similarities, but that's really not unexpected as Portuguese is relatively conservative for a Romance language, both in its written and spoken form (at least since the middle ages).
'Wakkā' isn't pronounced as 'vaca' . Besides the long 'a' at the end, 'w' is a "short u", not an actual modern "v". In it's Classical Latin form it sounded like /ˈwak.ka/
'Perro' has an unrelated and different origin. The word derived from Latin was 'can', which does sound more like 'kuon' than 'cão' which has a nasal sound that was previously absent. In Galician-Portuguese the word was also 'can'.
The "guttural R" you're referring to (I think?) is a modern sound change, possibly from 1800s French influence, it wasn't pronounced like that in the past. If that is indeed the case, as it seems to be, then there's no connection to whatever old PIE dialect that was spoken in the region previously. If you could link us something that showed/proved guttural R in, for instance, medieval Portuguese it'd be really good and I'd love to read it.
Not going over the doenti/doente thing again, different meanings and origins. But remember that a lot of 'o' in Portuguese are actually pronounced as 'u'. As in 'arado' /ɐ.ˈɾa.ðu/, so just because the words might look similar when written, they can actually sound more different than expected.
Oh a joking note, I do really hope you don't don't pronounce 'doente' as "duenti" like a rural southerner.

b. I showed you the conclusion Vaclav Bladek wrote on the Lusitanian language. It does not agree with what you say, just because it has some archaisms, like the -P sound.

c. Obviously I can't say it doesn't date back to the Beaker period. Maybe it does, maybe it doesn't. I'm just as excited to find that out as you are, but I'm not optimist we ever will.

d. Yes, yes, but I asked about Spain, not Portugal. That, and the subclades are what I want to know.

Olympus Mons said...

The problem I have with you is the obvious "bias" inclination you have. If we don’t know bias to what, it becomes difficult and somewhat irritating. Its just that simple.

*Of course Bladek agrees with me. You even write in his conclusion! - Lusitanian is a PIE language that shares traits with all western languages but has to be considerate a different branch, as he states “representing its own group comparable with the statute..” .
If we accept that it has a “own group” and in just a few inscriptions we find such old archaisms that do not even exist in Proto-celtic, in Proto-Italic in proto-germanic ANYMORE, let alone in the celtic or latin itself….them it has to be said that maybe it was LUSITANEAN having influences in the other PROTO something. Not the other way around. What was the event out of their territory that went all over Europe like a wild fire in a hot summer day? – Bell beakers!

*If you read blazek, or B. Prósper, you notice that to establish those parallelism to other western Europeans languages but not being the same, he, most often than not, has to go back to Old Indic, meaning Indo-Aryan or Sanskrit and accepts that the same archaism derivations are seen in Proto-celtic, proto-germanic, etc but complety rules out Lusitanian being a derivation of those Proto-celtic or proto anything in western Europe.

*”-P” just means that Lusitanians retained a PIE archaism that not even by 1500BC anyone else in Proto still had kept. So “-P” says that Lusitanian was old. Very old form of PIE. – Heck just need to remember what Strabo (estrabao) said about the Turditani, south of Turduli (but sort of same stock) … They told Strabo they spoke “roman” but their original language was spoke by then and with laws, rules and verses that were 6000 years old (their words not mine. That looks really old).

Anyway, officially PIE by linguistics. See the Lusitanian.

►Albanian (4)
►Anatolian (10)
►Armenian (3)
►Balto-Slavic (25)
►Celtic (16)
►Germanic (103)
►Graeco-Phrygian (10)
►Indo-Iranian (318)
►Italic (88)
►Tokharian (2)

Gaspar said...


One must take in consideration that even the romance languages have there isogloss splits from ancient the major split - La Spezia-Rimini line

Even the simple word of Vaca ( as you mentioned )is only used in North-italian languages, while Muca is the rest of Italy

Grey said...


interesting stuff

it's like reading the history of all the pastoralist groups in the Sahel - a constantly repeating pattern of conquest or displacement of the settled populations along the edge - like an interior African version of the steppe.

Grey said...


"Logically - that explains why Bedouins are full of Steppe ancestry."

no doubt there'd be other circles to square but the basic point stands

if you have two propositions

1) PIE homeland is either north or south of the Black/Caspian seas.


2) linguistics says one or more of now southern associated populations were in contact with PIE

then *if* the second one is true (and i have no opinion on that)

then either
1) PIE homeland was south and in contact with the southern associated populations
2) PIE homeland was north and one or more of the southern associated populations were originally in contact and moved south

so people who think the PIE homeland was in the north might want to either
1) disprove the linguistic link
2) look for evidence of a once near-steppe population who were displaced far to the south
who might have influenced the languages of the southern populations

EastPole said...

„Tracing the biogeographical origin of South Asian populations using DNA SatNav”

Kippou said...

Hello! I'm new on GedMatch. I'm from Morocco and I have done DNA testing on FTdna and Ancestry Dna. Well the results are kind of different when it comes to detecting the middle eastern and African dna. I would like to know which admixture calculator is best suited for me. I have north african, iberian, italian, sub-saharian, east african, and middle eastern dna. Thank you so much ^^

FrankN said...

@Grey: I agree. We need a theory for the PIE genesis, homeland, and/or its early evolution some 5-8 kya that accounts for the linguistic closeness to/ shared vocabulary with each of Uralic, Caucasian, Afro-Asiatic and Sumerian. I leave out Elamite here, it is too much of a linguistic "bastard" to be helpful in this respect, but Dravidian may comprise "Nostratic" parallels that go beyond what can be expected from language contact with IndoAryan.

The Steppe theory in its current form works well as concerns Uralic and Caucasian relations, but fails on the Afras and Sumerian aspects. So, as you say, "people who think the PIE homeland was in the north might want to (..) look for evidence of a once near-steppe population who were displaced far to the south who might have influenced the languages of the southern populations." That task doesn't come easy, given today's genetic makeup of "southern" populations, and the aDNA we have so far.
Moreover, the PIE-AfrAs link looks the most "basal", as it relates to commonalities in root building (2-4 consonants, semantic differentiation by Ablaut) rather than "only" shared vocabulary. For Western IE, a way out of the PIE-AfrAs dilemma could be assuming EEF to have spoken some kind of (Para-)Afras, as is e.g. brought forward by Vennemann's "Germania Semitica" (somewhat modified - Vennemann proposes the Megalithic as bearers of [Para-]Semitic). But I don't see yet how that approach can work for IndoAryan.

A "southern" homeland theory first of all faces the obstacle that during the 3rd mBC many candidate regions were (already) occupied by other language families - Sumerian, Akkadian, Hurro-Urartian etc. OTOH, this "Babylonian" language mix provides for most of the language contacts PIE requires, except for Uralic. But - this is more thinking out loudly than coming up with a full-fledged theory - what about the SW Caspian coast/ NW Iran? It
a) isn't a region that during the CA was clearly dominated by a non-IE language family,
b) has connectivity to Uralic (EHG) via the Caspian Sea and the Volga/ Ural rivers, signified a/o by occurrence of Caspian/Caucasian yDNA J much further north, and apparent Iran_Hotu relatedness to AG3.

Davidski said...


Start with the Eurogenes K13 and the four ancestors oracle. Then the K15. They should give you accurate results.

Chris Davies said...

@FrankN - I have observed a link between PIE and pan-AfrAs word roots also. But I think that I have noticed, in addition to that relationship, more in the western IE langs eg. Proto-Celtic, Proto-Italic, Proto-Germanic, quite a number of cognates in Chadic (and not Berber or Semitic) &/or Cushitic/Omotic or in some instances even Benue Congo.

Grey said...


so something like Hyksos or Sea Peoples but substantially earlier.

Olympus Mons said...

@ Chris Davies

Hi Chris. Can you elaborate a bit more?

Qagan said...


There are a lot of ADMIXTURE calculators on gedmatch you can try. Like Davidski have mentioned, probably start with Eurogenes K13 and K15 first. Then after that maybe you can try some Dodecad calculators like K12b, K7b, Eurogenes K36 first. Then after that if you are more interested, tried doing some ADMIXTURE calculators that used ancient genomes like Iran Neolithic K6, Near East Neolithic 13.

Do you have an account on Anthrogenica btw?

FrankN said...

1- As you asked for it above: I have a few times gone on record to qualify Portuguese as "Spanish spoken with a heavy Swabian accent" (Swabian as per ). Please take that as it is, i.e. (a) reflecting my tendency to occasionally make provocative statements, and (b) an intuitive observation by an outsider (North German), who has been exposed to all three languages/ dialects in question without mastering any of them. I haven't undertaken a systematic comparison of Spanish-Portuguese vs. German-Swabian sound shifts. Moreover, while the above statement to a certain extent reflects my knowledge about Suebian settlement in Portugal, the perceived similarities may as well go back to Celtic presence in Swabia.

2. Aside from the name "Portugal" (Gaulish Harbour, referring to Porto), as mentioned by AdV, there is quite some toponymic evidence for strong Celtiberian (Galician) presence. Let's just look at a few well-secured Celtic toponyms:
- briga "hill, hilltop settlement": Present in Conimbriga (Coimbra), Lacobriga (Lagos), possibly also in the Bracari, who lent their name to Braga. Other occurences of that root include the Briganti (British midlands), Bourg-en-Bresse/F (Brigantium), Bregenz/AT (Brigantium), and Sarrebruck/D (Sarabriga).
- bona "castle (capital?)": Lisbon (castle of the Lusi[tani]); elsewhere Narbonne, Bonn, Vindobona (Vienna), maybe Yerewan (Erebuni).

3. Lusitani: "Lisbon" suggests "Lusi" as ethnonym. Thus, -tani should have meant something like "lands". Of course reminiscent of Indo-Aryan, and all those "(s)tan"s between Kurdistan, Rajastan and Kirghisztan. How come Portugal and Central Asia sharing a similar toponymy, with little of it in-between (Wurs-ten [< Wierde "wharf"] in the Elbe-Weser triangle as possible exception), is a quite intriguing question.

4. I don't think that single words such as "porcom" are sufficient to qualify Lusitanian as non-Celtic. Lusitanian has evidence of having undergone sound shifts similar to P-Celtic , e.g. PVMPI "five". The P-Celtic Kw>P shift took place prior to the 5th cBC, implying that from then on P-Celtic had regained the ability to voice "P" (Q-Celtic is different, Old Irish shows re-appearance of the "P"-sound only around 700 AD). Hence, Lusitanian "porcom" may well represent a borrowing from Latin.

Aram said...

If Lusi is a ethnonym then it is very interesting. Many early IE tribes had a word Lu as a root.
Lydia cuneiform Luddu. There is also Luddu in India
Lycia cuneiform Lukka
Hypotheticaly Lullubeans are similar to Luwians living close to Gutians.
Also Slavic Ludi. Some slavic tribes had a Lud root. Meaning free men free people.
Burgundian Leudis .
More speculatively Latium can be from same root but it could be a non IE substratic word as some scholars proposed. No idea here.

Aram said...

Mtdna I was found in Levant PPNC.

I1679 AG037C 31.988 35.976 F Levant_N PPNC This study (S1679.E1.L1) half 1240K.capture 6900-6800 BCE 'Ain Ghazal Jordan 0.04 47421 I

It is absent from Your spreadsheet because it was published later.

Kippou said...

@Davidski, Thank you :)
@Qagan, No,i don't. I didn't know about that forum. I will make an account :)) Thank you for your answer.

Alberto said...


"more thinking out loudly than coming up with a full-fledged theory - what about the SW Caspian coast/ NW Iran?"

With Leilatepe in that area acting as a bridge between North Iran and Maykop, what do you think about what this paper proposes? (In German, abstract in English)

We do have a rather obscure find of R1b in Vucedol from the early 3rd M. supporting that possible connection.

Qagan said...


You welcome. They can tell you a lot about genetics there.

Olympus Mons said...

Considering it Celtic violates the elementary principles of Linguistic reconstruction.

"Porcom" should be enough to considered it non-Celtic.
But if not, then "Trepobala" has to do the trick. And if that doesn't, then development of IE. *d to Lusit. r, that appears in so many lusitanian texts, are non existent at all in celtic.

Or … the use of the conjunction indi ‘and’, that is not attested in Celtic at all but amazingly still one of the most used by today Portuguese language (ainda);
Or the development of a present inflection of the root *do- ‘to give’, doenti3, unparalleled in Celtic.
Or … the basic word order is of the type ODS (= direct object + indirect object + subject).

Actually Trebopala and Tencopala, Goddeses, are unique in the context since to explain it, like so many other words linguists have to revert to Indo-aryan in these two cases without hesitation to VEDIC godess Vis-pala (like those two above, literaly "protector of the dwelling").

And always remember how Karl Horst Schmidt has define it : “tentatively” defines Lusitanian
as «a language that belongs to the southern group of western IE languages
and borders on ….. EASTERN IE”

Amazing right. FrankN ... give it up. :=)

Again --- Lusitanian has so many archaism that is considered a very old PIE.
So, isn't it intriguing that in the EXACT PLACE WHERE BELL BEAKERS ARISE, people living there even at AD times spoke a PIE predating all those Protos (proto-celtic, proto-italic, proto-Ligurian, etc)?

Olympus Mons said...

Have you read my suppl II

isnt it what this dna paper is saying? -

I thought on that one I had gone clearly off the reservation but apparently not that much, right?

batman said...


"Considering it Celtic violates the elementary principles of Linguistic reconstruction."

Elementary jokes are also jokes.

""Porcom" should be enough to considered it non-Celtic."

In the Germano-Celtic and Uralic Europe you'd find a Porca/Porka/Purka/Porsa in every pig-farm - from I-beria to Si-beria.

FrankN said...

@Alberto: Thx for the link!

It affirms other papers demonstrating technological spread out of W. Iran over the Caucasus into Maykop during the 4th mBC, and then, after a longer pause, by the early 3rd mBC also into the Steppe (Yamnaya) and the S. Balkans (Ezero/ Schneckenberg/ Vucedol). A few takeaways that aren't captured in the English abstract:

1. Two of the copper tools analysed, namely (a) spouted shaft-hole axes, and (b) flange-hilted daggers, can technologically be traced back to W. Central Asia. For (a), Ivanova mentions a late 5th/ early 4th mBC find from Susa, late Ubaid clay models from S. Mesopotamia, and 4th mBC finds in the Kura Valley and Mudigak (S. Afghanistan). She supects an origin in W. Iran. Flange-hilted daggers (b) appear to have originated further NE, earliest finds, also from the late 5th/ early 4th mBC, are reported from S. Turkmenistan, NE Iran (Tapeh Hesar) and Sialk (W. Iran).

2. The third tool type, spouted hollow chisels ("Tüllenhohlmeißel") seems to have originated within Maykop (earliest find in the Maykop Kurgan, mid-4th mBC). Early 3rd mBC finds of such chisels in E. Anatolia and N. Iran indicate that trans-Caucasian technology exchange worked both directions.

3. A central theme of the article, reflected in the title "Stop and Go", is a description of the spatio-temporal dynamic. It took more than half a millennium before these copper tools spread across the Lower Don, i.e. out of Maykop, into the Steppe and the Balkans. However, when they finally crossed that border during the early 3rd mBC, the spread was fast and geographically wide-spread [Note that the enigmatic burial of Egeln, Saxony Anhalt, ca. 2750 BC, which Rob a few weeks ago asked me to look into, also provided a flange-hilted dagger with closest parallels in Bulgaria (Ezero Culture). The relevance of flange-hilted daggers within the BB package is well known, I guess.]

4. To explain the “stop” she points to low population density, cultural fracturing and poorly developed (exchange) networks in the Steppe during the 4th mBC. The appearance of Yamnaya by the early 3rd mBC meant cultural and technological integration over a wide area, and coincided with increasing population density. Without going into detail, she concludes that it required Yamnaya to establish the “socio-technical system” within which the new tools gained demand (as weapons, but also for wood processing -> wagons), as well as the resource base to allow them being regularly produced.

5. She doesn’t delve into the genesis of the Yamnaya phenomenon. Obviously, Maykop and ultimately W. Iranian traditions played an important role w/r to metallurgy. However, I wonder whether increasing population density and cultural integration in the Steppe wasn’t also promoted from the West (CT) and North (GAC), with the former possibly strengthening agriculture and the subsistence base, while the latter being demonstrated builders of long-distance exchange networks. To put it differently: While Maykop provided the metal tools, CT had the ard and milk cows, GAC the wheeled wagon and pig-breeding experience. In any case – the Steppe apparently was a sink, not a source, but as such sink became an important interface for cultural fusion.

What we now need is aDNA from Leyla-Tepe, Maykop, W. Yamnaya, GAC, CT, Ezero, Schneckenberg, Vucedol. A little more Kura-Araxes would also be nice, and OM (but not only him) would of course love to see some Ubaid and Shulaveri-Shomu aDNA. A very humble wishlist for Christmas…

Olympus Mons said...

Thsoe are NOT my words.
Are statments and seem to represent a consensus of Linguistics that are experts in Lusitanian (germans, Polish, Spanish, portuguese, British, Russians). - So, take it to them. :)

Azarov Dmitry said...


Damn it! You pretty much accurately described my model of R1a expansion from the Iranian Plateau.

Rob said...

@ FrankN/ Alberto

Yes a great paper indeed. Overall, I tend to agree with Dave, in that there is no evidence for migration from somewhere between Armenia & Iran between to the steppe, at least on the present evidentiary basis. We have aDNA from Armenia Chalcolithic (c. 4000 BC) & Iran (4000 BC), but the 'CHG' present in Yamnaya is not from there, each time choosing archaic Satsurblia in runs. Moreover, its 'ANF' comes via something linked to LBK.

So if there was a migration from NW Iran to Majkop, it seems to have ended there, and not spread to the steppe. Any technological spread beyond could arguably have been by 'adaptation'/'acculturation'/ borrowing.
However, this isn't quite a satisfactory answer. Ivanova points out, as others have, the learning of complex new metallurgical techniques would have been guarded and passed on personally, not easily 'adapted'.
It is also interesting to note that deposition of such axe classes show concentration in 2 major regions : Majkop and the Carpathian basin, and in much of the steppe & CWC the deposition of these axes is rare. Yet genetically, BA Hungary looks like a hypothetical fusion between Dnieper-Donets (80%) and Anatolia Chalcolithic (20%)

As for the origins of Yamnaya, it seems most of it goes back to 2 broad Eneolithic networks which existed between the Danube & Don.

Alberto said...


Thanks for the summary, it's always much better to read your summaries than to try to find our way with automated translations.

How this transmission of technology relates to the movement of people is still difficult to asses. The suggestion is that the more mobile Yamnaya people were indeed responsible for it, rather than suggesting a migration of the Maykop people. But we're lacking ancient DNA from all those places to know the details of how it might have happened.

We'll need to understand the exact genesis of Yamnaya first, and then we might also be dealing with slightly different movements from the Black Sea to the CWC area and to the Balkans, the latter being more metal workers and not just mobile pastoralists, so maybe also with a more noticeable "Maykop" admixture (whatever that turns out to be).

I subscribe that wishlist. If not for Christmas, may it be for the first half of 2017 at least.

Qagan said...


Hope to see you there in Anthrogenica. I am a member there as well.

Rob said...

@ Alberto & Frank

I have noted that Battle Axe Sweden (RISE 94) looks like an outlier for northern Europe, being different to the LN/ EBA samples, although plotting normally.

"Armenia_Chalcolithic" 54.65
"Hungary_HG" 16.45
"Eastern_HG" 16.05
"Satsurblia" 10.55
"Loschbour" 2.3
"Villabruna" 0

"Yamnaya_Samara" 36
"Armenia_EBA" 34.3
"Hungary_HG" 12.3
"Esperstedt_MN" 5.85
"AG3-MA1" 4.15
"Israel_Natufian" 3.25
"Villabruna" 3.1
"Loschbour" 1.05
"Satsurblia" 0

IIRC, this individual was R1a1, possibly Z645+, whilst all of German LN/ CWC was Z645-.
Can't say much from one sample, probably an outlier at best. But to me it looks like he came from Majkop.
Frank- was there anything different about his burial ?

Nirjhar007 said...

Are you sure that you didn't make any mistakes?. It looks quite outstanding ! .

Alberto said...

Yes, interesting, that's a lot of Armenian admxiture over Yamnaya.

I hope the long awaited Maykop samples will come soon. As time passes without new samples we only get new questions with a difficult answer.

FrankN said...

@Rob, Alberto e.a.:
Hope you are still reading this and haven't moved on to later entries..

Good background on the Ivanova paper is provided by this conference reader (Tbilisi 2014):

I especially recommend reading:

1)Stöllner/ Gambashidze (p. 102-124): Excellent overview on Caucasian metallurgy (unsurprisingly):
- Leyla Tepe reminiscent of NW Iran (Seh Girdan, Tappeh Ghabrestan, Arisman),
- Little innovation and metallurgical finds in early KA, strong increase in the early 3rd mBC (as Balkans, Yamnaya),
- Trans-Caucasian export of "prestige items"(gold, spiral-rings, antimony as silver surrogate) from Georgia into Maykop/ Steppe from the late 4th mBC onwards, by the early 3rd mBC mining possibly turning into a "business", i.e. full-time activity.

2) Inanishvili (p. 233-2444): Archeo-metallurgical analyses, dealing in particular with Copper-Nickel alloys in Maykop and Mesopotamia, and tracing the ores back to the Talmes and Meskan deposits (West Iranian Plateau, Anarak distr., ca. 500 km SE Teheran).

3) Mirtskhulava/ Chikovani (p. 32-41) give a good overview on Chalkolithic cultures. I just wish they had refrained from ethno-linguistic interpretation towards the end - never a good idea for the CA, and even less so in the Caucasus. Anyway, here is their take on Leila-tepe: "Nakhichevan, Mughan and Karabakh late Eneolith sites keep following the traditions of Shulaveri-Shomu Tepe culture while the sites of Leila-Tepe type strictly differ from them and they are attributed to the north Ubaid-Uruk cultural circle. There is nothing either pre-existent or subsequent of Leila-Tepe pottery in the Caucasus and scholars suppose that the culture moved to the north Caucasus and took part in the formation of Maykop culture. (..) The excavators have quite pronouncedly stated that there is no sign of any contacts between Leila-Tepe group and Kura-Araxes culture."
This implies that Armenia_CA (early KA) aDNA can't be used as a proxy for Leila-Tepe/ Maykop.

4) Shanshashvili/ Narimanishvili (p 245-264) provide insight into the enormous diversity of dwelling constructions and settlement structures. This seems to indicate that the Caucasus during the CA/BA has no less been multicultural than today, implying a warning against uncritically extrapolating singular aDNA results.

5) Cultraro (p. 125-138) draws attention to N. Caucasian (Maykop, KA) influence in the N Aegean (esp. Poliochni / Lemnos, but also Troy I/II). Parallels are typological (shaft-hole axes) and technological (lost-wax moulding), and, as elsewhere, they only gain impetus in the N. Aegean from the early 3rd mBC onwards. Cultraro also glosses over typological parallels between the N. Aegean and the Balkans.
Surprisingly, N. Aegean tin bronzes (from early 3rd mBC onwards, dominating by ca. 2600 BC) share their isotopic fingerprint with those from Velikent/Daghestan, KA Armenia (Shiraz), and the Persian Gulf (Tell Abraq, VAR), suggesting a common ore source. Armenian copper has been ruled out in this respect. Cultraro interpretes these patterns as additional evidence (on top of the spread of agricultural and ceramic innovations) for Ivanova’s suggested trans-Anatolian “royal road”.
In this context, I find it noteworthy that Ivanova 2014, while demonstrating metallurgical similarities between the Steppe and the Balkans, has refrained from any explicit suggestions on the paths and directions of technology transfer. The pattern may indeed be Circum-Pontic (North and South). Btw, there is also some archeological evidence, e.g. copper pins, for eneolithic maritime contact between the Sinope-Samsun area and Thrace.

FrankN said...

“Btw, there is also some archeological evidence, e.g. copper pins, for eneolithic maritime contact between the Sinope-Samsun area and Thrace.”
This was from memory, which can be deceiving, so I looked up the source again (old, but well-made, though the chronologies may require review and updating):

P. 217ff: “The Büyük Güllücek group in Central Anatolia, the Ikiztepe II group in the Black Sea Littoral and the Orman Fidanligi group in North-Westem Anatolia are more or less contemporary to each other. All groups share some similar features in pottery, such as homed handles and white painted decoration.
(..) Homed handles of Büyük Güllücek, Ikiztepe II and Orman Fidanligi groups show remarkable similarities to homed handles from the Karanovo III and III-IV cultures. However, characteristic forms and decoration of Karanovo III and III-IV are absent in these groups. Copper flat axes and a dagger from Büyük Güllücek have parallels in Bulgaria. However, the Bulgarian samples may be dated later than Büyük Güllücek samples. Comparable material for the Büyük Güllücek, Ikiztepe II and Orman Fidanligi pottery also comes from the Aegean islands and Westem Anatolia. It seems that all these groups are related to the Besiktepe-Kumtepe la culture in Westem Anatolia.
(..)According to pottery evidence, Karanovo III-IV, Büyük Güllücek, Ikiztepe II, Orman Fidanligi and Besiktepe-Kumtepe la should be more or less contemporary. However, dates from Anatolian groups are later than the Karanovo III-IV culture. The Besiktepe-Kumtepe la culture dates ca. 4900 / 4800-4500 / 4400 cal. BC and Karanovo III-IV dates 5200 - 4900 cal. BC. Dates from Ikiztepe E show ca. 4400 / 4300 cal. BC. However, most of Ikiztepe II dates are determined by the large errors. On the other hand, dates from Besiktepe are respectful.

(..)The Gelveri Group in Central Anatolia still poses a problem in Anatolian archaeology. The Gelveri pottery shows similarities to the Pre-Cucuteni, Maritsa, Sava and Boian cultures of the Balkans.

(..)During the Late Chalcolithic period there must have been some contacts between the Balkans and the Black Sea Littoral of Anatolia. Similarities are especially apparent in the metal finds and figurines, also to some extent in the repertoire of shapes. The metal finds - flat axes and wedges - from Dündartepe have good parallels in the Gumelnita-Karanovo VI context. Figurines found in Ikiztepe I sounding C and F, and the figurines from Dündartepe show good similarities to Vinca, Cucuteni and Karanovo VI figurines. It seems likely that the pottery from the Black Sea Littoral of Anatolia was the product of local development closely related to the Balkans. During the Late Chalcolithic period. Central Anatolia has a distinct pottery repertoire. The only similarities between the Balkans and Central Anatolia are graphite painted pottery and some figurines from Alisar. Central Anatolian Late Chalcolithic sites can be dated ca. 3750-3250 cal. BC. However, these dates match the Transitional Period in the Balkans.

p. 237: “Similarities in material culture, especially metal objects and figurines (Fig.V.7:11-14) during the Late Chalcolithic period, indicate that there must have been some contacts between the Gumelnita culture in the Balkans and the Black Sea Littoral region of Anatolia. The lack of similarities between the Black Sea Littoral of Anatolia and Turkish Thrace indicates that this contact probably happened by sea..

Angantyr said...


Since it is likely that information on the burial context of RISE 94 has only been published in Swedish, I have been trying to find something, but I haven't succeeded.

Supplementary table 1 in Allentoft 2015 states that RISE 94 comes from "grave 26:1", and there is an object in the Swedish National Heritage Board's register with this number at the right location (see However, the register has no information about any archaeological investigation of this site, and neither has Google.

Swedish science journalist Karin Bojs has written briefly about RISE 94, both in a newspaper article and in a book, and stated that his grave is "clearly" Battle Axe. She also seems to say that an isotope analysis has been made, and that it showed that he appears to have lived his whole life in the same area. Unfortunately I don't have the book, which might include a reference to the source of this information.

By the way, what does RISE 61 from Denmark look like autosomally? He was roughly contemporary and had a similar Y-DNA.

batman said...

RISE 94 and 81 were probably part of the same cattle-farmers that beamed out of Jutland, producing a 'danish' dynasty. Untill IA Scania was an island and part of the 'lowland' (continental, kelto-german) cattlefarmers with Rb1.

Simulataniously mainland Sweden (Sormland/Uppland) hosted a hub of the 'highlanders' (oriental, suevian) cattlefarmers based on a dynasty of R1a. Thus we find some of the oldest R1a-derivates between Middle Norway and Carelia.

The graveform in Lilla Beddinge in southern Scania was a 'normal' stone-pit-grave, similar to Motala, with roundstones creating a chamber, covered with slabs to create a simple grave-chamber. Preservation was obviously an idea behind this practice.

A closer description of Lilla Beddinge can be found in Elin Fornanders thesis from 2011:

"Lilla Bedinge comprises the largest known cemetery associated with the Swedish Battle Axe Culture. The site, extending over an area of about 240×30 m, is located only about 1 km from the present day coast line. The majority of the at least 14 identified and excavated flat earth inhumations graves are located on a NE–SW oriented moraine embankment, whereas four of the graves are found on the flatter grounds to the SE. The site also includes a number of Late Bronze Age cremation graves, and two other find spots for BAC inhumation graves are known in the nearby region."

"Grave 49 was excavated by Hansen 1934. It constitutes a N–S oriented subsurface oval stone construction with pointed edges, measuring about 4.5×2 m, where flat stone slabs form a roof over a chamber with an original height estimated to about 0.6–0.7 m. Fragments of wood indicate the presence of planks in the chamber. On the stone paved floor of the chamber three adult individuals had been placed in a line in sitting crouched positions facing southwest. Between the northern and middle skeleton fragmented remains of three children (initially only two were identified), representing two infants and a juvenile, were recovered. Further, some very brittle diaphyses of a fourth adult have been identified. The only recovered find is a bone needle deposited next to the northern skeleton (Hansen 1934; Malmer 1962:162p ; During unpublished notes). According to Malmer (2002:141) the grave can be dated to Period 4, and an unpublished radiocarbon date from the northern skeleton falls within the interval 2580–1980 cal. BC (2σ, 3850±105 BP, Ua-2758, During unpublished notes)."

Davidski said...

RISE94 is a low coverage, non-UDG treated sample with pretty heavy deamination signals that make it look more basal.

You'd need a whole group of Swedish Battle-Axe samples, preferably UDG-treated, to say anything meaningful about their origins.

But I reckon RISE94 is a fairly unadmixed example of someone from the western edge of the Bronze Age steppe. Can't see it having any Maykop links. Z645 is really unlikely to be found in Maykop.

FrankN said...

Could RISE94 be the burial described on p52 in batman's link?

"Håslöv, Gustav Adolf parish, Raä 13:1
The burial site at Håslöv, northeast Scania, was discovered and investigated at a gravel quarry in connection to road construction work in 1956 (Malmer ATA dnr 4.7.1957). The site is found just over 2 km southeast of the burial site at Åraslöv, and the present day proximity to the coastline is about 10 km. Three flat earth inhumation graves, all of which are included in the present analysis, have been identified at the site, located on a southern slope. Grave I (Grave 26 in Malmer 1962) comprised a funnel shaped stone construction under which a more or less intact skeleton was found in crouching position with the head to the north, facing east. Remains of a wooden construction were identified around the body. The grave goods consisted of a bone awl, a flint blade and a fragmented amber pendant (Malmer 1962:917,ATA dnr 4.7.1957). Due to the amber find, Malmer (1962:221) interprets the feature as a woman‟s grave, but an osteological analysis identifies the skeletal remains as a man (Jantsch & Ranåker 2001:32). The grave has been typologically dated to Period 5 (Malmer 1962:198)"

Parish & location seem to fit, Grave 26 w. Malmer, Grave 1 (of 3) on the site.

Rob said...

@ Angantyr & FrankN

Thank you for your contextual comments. ! So it seems there was nothing unusual about the individual.

@ Dave

It's on the lower end of coverage, but if coverage was the issue, wouldn't pseudo -African mixture arise ? It did not, and I don't think poor coverage would give pseudo-Arm Chalcolithic appearance, esp given that it's present still in many modern (south) Europeans. Of course, we should all like this confirmed further, and my suggestion that this chap might have 'come from Majkop' shouldn't be taken literally, as for starters Swedish battle Axe is hundreds of years after even the youngest Majkop phase, nor was I suggesting that Z645 needs to come from there either (although I do think that early CWC R1a xZ645 was largely replaced subsequently). But there's no reason to doubt *some* direct movement from the Caucasus to Europe, even northern Europe, could have occurred, just as there was movement from central Europe to the steppe (perhaps limited albeit).
As I also demonstrated with BA Hungary, there were movements from several regions aside from the common masses from the Don steppe, as the early Bronze Age eastern Euro admixture in Hungary doesn't look steppic, but from some yet unsampled group like C-T/ Dnieper-Donets, Usatavo, etc.

Rob said...

@ Angantyr

The other BAs appear to be regular LNBA Nordics , Corded Ware - like

FrankN said...

"Could RISE94 be the burial described on p52 in batman's link?"

Apparently not, because that burial is Raä 13:1, RISE94 is Raä 26:1 and lies some 500m further SE. In any case, the area seems to have had quite a burial density around 4500 BC. Regional centre?

Angantyr said...


Rob is right, RISE94 must be the man buried at the Håslöv site, alias Raä Gustav Adolf 13:1, and Allentoft 2015 inconsistently uses Malmer 1962's grave number 26 with a ":1" that does not correspond to Malmer 1962 (which names grave II at the same site grave 27). Everything fits perfectly, and Allentoft 2015 doean't seem to use Raä numbers for any of the Swedish samples. If Raä 26:1 had been excavated, this would have been clearly documented.

This area has massive amounts of neolithic archeology, from Funnelbeaker times and onwards.

Alberto said...


Thanks for the paper with so many interesting materials. I just had time for a quick glance, but the summaries you made are already very informative.

Yes, I expect a lot of circum-Pontic movements in every direction. It's a very central area between Europe and Asia that will require high resolution by extensive sampling all around it before we can have a full picture.

Seinundzeit said...

Here are some models.

All of these were done using my standard setup, which only uses ancient West Eurasian references (with the addition of Ust-Ishim and Mota), and countless contemporary references for African, Siberian/East Asian, indigenous South Asian, and Oceanian/Australasian ancestry.

There are many more reference populations, but obviously, I'm only showing the ones for which samples/populations scored percentages.

Most importantly, my aim is to test everyone under identical conditions. I really want these results to be totally comparable, in terms of the variation on which these models are built. Only in the case of South Central Asians was an additional population excluded (Vietnamese_South). I did this because Vietnamese_South really eats into the South Central Asian Jarawa/Onge/Agta percentages. For some Pashtuns, the Jarawa/Onge/Agta just disappear from the models when the Vietnamese are included.

The Vietnamese do not have this effect in South Asia, so they were included in all other models, as I had Austroasiatic and Northeastern Indian populations in mind.

Hopefully, with aDNA from outside West Eurasia, we'll arrive at even more solid models. Right now, this setup works best with people who have around 10% or more West Eurasian admixture.

Regardless, let me explain some of my choices here.

I have included Samara_Eneolithic, rather than Yamnaya, or later steppe populations like Sintashta.

Mainly, I did this because I really wanted to zero-in on that extra ANE affinity many South Central Asians seem to display, especially in comparison to the currently sampled Neolithic Iranians.

Also, Lazaridis et al. (2016) did model South Asia with Samara_Eneolithic. In fact, it was one of the steppe/Eastern European populations which allowed for the successful modelling of all 20 South Asian populations. As a result, I really wanted to see how things would look using Samara_Eneolithic.

In addition, I wanted to see distinctions, with regard to the ANE affinity seen in South Asia. Or, in other words, I wanted to see if there really are any differences in the origins of the ANE affinities often seen in South Asia. To that end, I also included two purely ANE references, MA1 and AG3.

Anyway, what I've found has been very pleasing. Peninsular South Asians prefer AG3, while northwestern South Asians + South Central Asians prefer Samara_Eneolithic.

To me, these results are reflective of differences in genetic input from ancient steppe populations (ones which had origins in Eastern Europe/the Pontic Caspian steppe). The AG3 percentages seem to reflect an ANE substratum that was local to Central Asia. Although, it seems that things get muddled for South Indian Brahmins (and some North Indians). But that was expected, as we can get close to perfection, but we can’t fully attain it.

Be that as it may, the results themselves are obvious; there is a lot of extra ANE/EHG-related admixture in South Central Asia, approximately equal to the amounts seen in Northern Europe (and with Pamiri peoples + Chitrali Dardic people, more ANE/EHG-related admixture than what is seen in Northern Europe).

As a reminder, Samara_Eneolithic is far more EHG-shifted compared to Yamnaya (and Yamnaya itself is around 50% EHG-related).

Seinundzeit said...

Continuing from where we left off...

To represent true ASI (and not some hybrid component, one which is usually mixed between Andaman-related and Neolithic Iranian-related ancestries), I have used the Jarawa (a population from the Andaman Islands, they should be genetically identical to the Onge) and the Agta (a "Negrito" population from the Philippines).

I've found that populations from Punjab and beyond prefer the Jarawa, while people west of the Indus river (and from Central Asia) prefer the Agta.

This isn't surprising, as the Agta are 30%-40% Southeast Asian (and 40% Andaman-related, with an additional 10% of Australian/Papuan-related ancestry), and South Central Asians (Kalash, Pashtuns, and even Sindhis) prefer East Asians to Onge when it comes to d-stat comparisons (a clear bias towards Dai versus Onge, in some d-stats David ran for us), so they'll naturally prefer the Agta as an ASI reference.

Please, see below.

Seinundzeit said...

Austroasiatic populations of India:


70.30% Jarawa
25.00% Iran_Neolithic
4.65% Vietnamese_South


68.35% Jarawa
23.65% Iran_Neolithic
7.25% Vietnamese_South


67.45% Jarawa
25.95% Iran_Neolithic
6.60% Vietnamese_South


65.10% Jarawa
34.35% Iran_Neolithic
0.55% AG3


65.1% Jarawa
21.6% Iran_Neolithic
11.0% Vietnamese_South
2.3% Agta


65.20% Jarawa
33.35% Iran_Neolithic
1.45% AG3


64.30% Jarawa
19.65% Iran_Neolithic
16.05% Vietnamese_South


62.9% Jarawa
20.3% Iran_Neolithic
16.8% Vietnamese_South

Khasi (very different from other Austroasiatic populations in India, which isn't surprising)

47.95% Vietnamese_South
22.45% Jarawa
18.60% Iran_Neolithic
7.05% AG3
3.95% Mongola

Seinundzeit said...

South Indian populations:


59.45% Jarawa
39.10% Iran_Neolithic
1.45% AG3


50.65% Jarawa
46.10% Iran_Neolithic
3.25% AG3


55.2% Iran_Neolithic
38.7% Jarawa
6.1% AG3


54.10% Iran_Neolithic
38.85% Jarawa
7.05% AG3


56.35% Iran_Neolithic
37.80% Jarawa
5.85% AG3


60.10% Iran_Neolithic
33.65% Jarawa
6.25% AG3


59.95% Iran_Neolithic
32.25% Jarawa
7.80% AG3

Tamil Nadu Brahmin

58.0% Iran_Neolithic
25.6% Jarawa
16.4% AG3 + MA1

Seinundzeit said...

North Indian/eastern Pakistani populations:


55.70% Iran_Neolithic
35.25% Jarawa
9.05% AG3


57.55% Iran_Neolithic
33.80% Jarawa
8.65% AG3


54.15% Iran_Neolithic
33.45% Jarawa
12.35% AG3


52.40% Iran_Neolithic
24.25% Jarawa
23.35% AG3

UP Brahmin

50.25% Iran_Neolithic
23.80% Jarawa
19.25% MA1
6.75% Samara_Eneolithic


61.25% Iran_Neolithic
26.55% Jarawa
12.20% AG3


59.45% Iran_Neolithic
25.90% Jarawa
14.65% AG3


50.9% Iran_Neolithic
24.9% Jarawa
12.7% MA1
11.5% Samara_Eneolithic


52.10% Iran_Neolithic
23.55% Samara_Eneolithic
20.00% Jarawa + Agta
4.35% MA1

Punjabi (Lahore)

55.75% Iran_Neolithic
28.90% Samara_Eneolithic
15.35% AG3 + MA1


61.25% Iran_Neolithic
21.10% Samara_Eneolithic
17.65% Jarawa + Agta

Seinundzeit said...

South Central Asian populations (northern/northwestern Pakistan, Afghanistan, Tajikistan, etc)


51.95% Iran_Neolithic
16.10% Samara_Eneolithic
15.05% Agta
12.35% MA1
4.55% Mongola

Pakistani Pashtun, Khyber Pakhtunkhwa (of the Yusufzai tribe, and that too a real Yusufzai)

59.15% Iran_Neolithic
26.80% Samara_Eneolithic
12.85% Agta
1.20% Villabruna

“Pathans” (Pakistani Pashtuns, the ones first collected by the HGDP. This group is curated, only includes the 13 samples that cluster together in ChromoPainter/fineSTRUCTURE analyses. These 13 samples are quite distinct from Sindhis/Punjabis, while the other ones overlap with Sindhis/Punjabi Jatts. Collected in the tribal areas, of the Karlani confederacy)

55.85% Iran_Neolithic
31.10% Samara_Eneolithic
11.10% Agta
1.95% Villabruna

Pakistani Pashtun, tribal areas (of the Mohmand tribe)

39.10% Iran_Neolithic
34.15% Samara_Eneolithic
15.45% Iran_Chalcolithic
11.00% Agta
0.20% Mongola
0.05% Samara_HG
0.05% Iran_Late_Neolithic


54.10% Iran_Neolithic
35.25% Samara_Eneolithic
10.55% Agta
0.10% Villabruna

Myself (Pashtun with roots in both Afghanistan and Pakistan)

54.70% Iran_Neolithic
28.65% Samara_Eneolithic
10.15% Agta
3.60% AG3
2.20% Mongola
0.65% Iran_Late_Neolithic
0.05% Iran_Chalcolithic

Afghan Pashtun, Ghazni (central Afghanistan, of the Ghilzai confederacy)

38.50% Iran_Neolithic
31.50% Samara_Eneolithic
18.80% Iran_Chalcolithic
9.70% Agta
1.45% Villabruna
0.05% Samara_HG

Afghan Pashtuns (samples collected in the provinces of northern Afghanistan, probably hodgepodge of Ghilzai and Durrani)

32.65% Samara_Eneolithic
30.75% Iran_Chalcolithic
23.95% Iran_Neolithic
9.05% Agta
2.15% Mongola
1.35% Iberia_Chalcolithic
0.05% Villabruna
0.05% Iran_Late_Neolithic

Afghan Pashtun, Khost (central eastern Afghanistan, of the Karlani confederacy)

33.35% Iran_Neolithic
32.65% Samara_Eneolithic
17.30% Iran_Chalcolithic
7.75% Agta
4.15% LBK_EN
3.10% Iberia_Chalcolithic
1.55% Mongola
0.15% Iran_Late_Neolithic
(Out of all Pashtuns, it is this individual, and a Waziri, who probably have steppe ancestry from Scythian-related groups)

Afghan Pashtun, Kandahar (southeastern Afghanistan, of the Durrani confederacy)

36.05% Iran_Chalcolithic
27.70% Samara_Eneolithic
26.05% Iran_Neolithic
5.45% Agta
4.35% Mongola
0.20% Karelia_HG
0.10% Villabruna


42.70% Samara_Eneolithic
30.15% Iran_Chalcolithic
15.90% Iran_Neolithic
7.10% Agta
4.05% Mongola
0.10% Samara_HG


42.75% Samara_Eneolithic
32.55% Iran_Chalcolithic
10.35% Iran_Neolithic
4.35% Iberia_Chalcolithic
4.20% Agta
4.10% Mongola
1.10% Villabruna
0.60% Samara_HG


45.45% Iran_Chalcolithic
39.25% Samara_Eneolithic
7.70% Mongola
2.65% Iberia_Chalcolithic
2.40% LBK_EN
1.85% Karelia_HG
0.85% Agta
0.05% Barcin_Neolithic

Seinundzeit said...

Pakistani Balochistan/West Asian populations:


42.00% Iran_Neolithic
31.70% Iran_Chalcolithic
18.00% Samara_Eneolithic
7.25% Agta
1.00% Villabruna
0.05% Iberia_Chalcolithic


65.65% Iran_Neolithic
16.95% Samara_Eneolithic
6.75% Iran_Chalcolithic
5.40% Agta
5.25% Villabruna


45.00% Iran_Neolithic
19.95% Iran_Late_Neolithic
11.35% Iran_Chalcolithic
4.40% Villabruna
4.30% Agta
1.45% Iberia_Chalcolithic


53.45% Iran_Chalcolithic
18.35% LBK_EN + Barcin_Neolithic
13.10% Iran_Neolithic
9.50% Samara_Eneolithic
3.30% Mongola
2.30% Karelia_HG


61.75% Iran_Chalcolithic
25.85% LBK_EN + Barcin_Neolithic (mostly Barcin_Neolithic)
11.45% Jordan_EBA
0.60% Samara_Eneolithic
0.15% Levant_Neolithic
0.10% Iran_Late_Neolithic
0.10% Iran_Neolithic

Seinundzeit said...

Northern European populations:


51.65% Iberia_Chalcolithic
35.65% Samara_Eneolithic
12.10% Villabruna
0.45% Karelia_HG
0.15% Barcin_Neolithic


47.85% Iberia_Chalcolithic
36.85% Samara_Eneolithic
15.30% Villabruna


36.55% Iberia_Chalcolithic
36.25% Samara_Eneolithic
26.15% Villabruna
1.05% Karelia_HG

General Observations:

If one knows a thing or two about South Central Asian anthropology, linguistic relations, and if one has some knowledge of aDNA research, the results seen with South Central Asians are of tremendous value.

Not much I can add, in terms of general observations.

Seinundzeit said...


Punjabi (Lahore) are 28.90% Jarawa, mistakenly have them as 28.90% Samara_Eneolithic.

Davidski said...


Onge are now in the spreadsheet.

Seinundzeit said...



Eventually, I'll add them to my setup, and we'll see how things look with them in the mix (although, I'd imagine things shouldn't be too different from what we see with the Jarawa, who have the same biogeographic origins).

Jaydeep said...


Thanks for all the hard work.

I find one thing strange. If Punjabi from Lahori does not show any Samara_Eneolithic proportions, why do the Sindhi, Baloch or Gujarati show it when they live further down South ?

It is also interesting that all the steppe affinity of South Indian populations goes back to the Palaeolithic. If we can put faith in this model, it clearly implies that there is no recent steppe ancestry in Peninsular India that can be traced to any Bronze Age migration.

Seinundzeit said...


No problem.

I truly enjoy doing this, since I have a great interest in figuring out the genetic history of Central and South Asia.

Also, I think the Punjabi_Lahore samples are, for the most, Lahori Christians (Lahori Christians are mostly of Punjabi Chamar origins).

With that in mind, only GujaratiA and GujaratiB (higher caste Gujaratis) show Samara_Eneolithic.

Although, I think Brahmins (and other higher caste Indians) throughout Peninsular India do have steppe ancestry via Bronze Age migrations. But, it gets confusing in their case, since they already have that ANE affinity native to Central/South Asia, and this method is amazingly good but not completely perfect (as no method can be).

Nirjhar007 said...


So you are Pakth ! ;) .

You are doing a nice job . I do have a query , what differences do you see in case of N Indian Brahmans and S Indian Brahmans groups regarding Samara and Andronovo type ancestry in modeling?.

Seinundzeit said...


My family is from the region described as belonging to the ancient Paktha, so if those ancient accounts are real, I am a Pakth (lol).

And thank you.

Comparing South Indian Brahmins with those from northern India, it seems Tamil Brahmins have around the same amount of ASI as Brahmins from Uttar Pradesh (just slightly/moderately more than Uttar Pradesh Brahmins), but less Samara_Eneolithic/Yamnaya/Andronovo affinity, and more Neolithic Iranian affinity.

Although, this is based on comparing Brahmins from Uttar Pradesh and Tamil Nadu. It might be more worthwhile to compare South Indian Brahmins to Brahmins from the areas of North India in which they have historical origins.

I'm not sure if Uttar Pradesh fits the bill? (does it? as I honestly don't know the history involved)

Nirjhar007 said...

Make no mistake , it is not an exaggeration when one suggests as I am doing now , that you do belong to one of the oldest known groups of IE family :) .

Yes Uttar Pradesh do fit the bill. But of course Kashmiri Pandits will be Ideal . Interesting observations but expected too , thx !.

Jaydeep said...


That explains the issue with the Punjabi from Lahori, I suppose.

Regarding South Indian ANE affinity, it is indeed confusing. South Asians have affinity to both MA-1/ AG3 as well as to Iran_Neolithic. That South Asian Neolithic and Iran Neolithic populations stem from a common source is without doubt. When we try to impute steppe ancestry in South Asia , the best sources are steppe_eneolithic and Yamnaya. Yet both of them have ancestry from a southern source closely related to Iran_Neolithic. The other source of their ancestry is EHG which is a combination of WHG and ANE. Since ANE and Iran_Neolithic ancestry in South Asia appears to be pretty old (present atleast since the start of the Neolithic, it is a complicated matter and a conventional explanation of a Bronze Age steppe migration is not the answer.

According to the most popular archaeo-linguistic theory Indo-Iranians were the last people to move out of the steppe PIE homeland and Sintashta represented the archaeological culture of these people. But clearly Yamnaya and steppe_eneolithic are better fits for steppe ancestry into South Asia. This puts the conventional steppe theory into tatters.

Seinundzeit said...

Since I can't directly respond at Anthrogenica, I'l respond to some observations posted there, here.

To Pegasus:

* The Baloch have more Iran_Chl, probably because they have ancestry from further west on the Iranian plateau. They speak a language related to the Kurdish group of languages, so this is a very solid result. As expected, they also have more Samara_Eneolithic.

* By contrast, the Brahui are a Dravidian (Elamite-related?) isolate, so one would expect them to have much lower Iran_Chl, and lower Samara_Eneolithic. The Baloch are basically Brahui + minor/moderate amounts of something from the western end of the Iranian plateau.

*With Iran_Chl in the mix, there is almost no Barcin_Neolithic ancestry in South Central Asia.

The very minor percentages seen in some populations are now, without a doubt, reflective of Sintashta/Andronovo/Scythian-related admixture. This is evident in the case of the Khostwal Afghan Pashtun, and a Waziri Pakistani Pashtun.

In addition, the LBK_EN percentages are closely tied to Iberia_Chalcolithic percentages, so that is the only evident conclusion.

*There is no reason to suggest that the Paniya have low Iran_Neolithic-related ancestry. It wouldn't make any sense, if their West Eurasian ancestry wasn't almost entirely ancient Iranian (in terms of genetic affinity).

In addition, MA1/AG3 are very different from Ust-Ishim.

MA1/AG3 are distant relatives of Kostenki/Vestonice/El Miron-related groups, and the Villabruna cluster seems to be admixed with MA1/AG3-related populations, but predominately derived from the Kostenki/Vestonice/El Miron-like stream of ancient populations.

By contrast, Ust-Ishim is a "basal" Eurasian, but not of the sort that played a huge role in the genetic history of the ancient Near East (and in the genetic history of modern West Eurasia).

In fact, MA1/AG3 and Villabruna/Bichon are more closely related to Australian Aborigines than they are to Ust-Ishim, who is very divergent (although, extra archaic Homo Sapiens admixture in Australian Aborigines would muddle such a comparison).

All things considered, the Paniya have nothing even remotely like a straight shot of 30%-50% ANE admixture, or even 30%-50% Ust-Ishim admixture.

They have around 40% admixture from ancient people related to Neolithic Iranians from the Zagros + the Mesolithic Iranians from the Hotu/Belt caves. No real debate there.

*With regard to why I used Samara_Eneolithic, rather than Yamnaya or Sintashta, please re-read my "Preliminary Remarks".

Also, David's remarks are right on the mark, they explain a lot, with regard to why the Baloch and Brahui differ.

To Dmxx

*I have included Iran_Hotu in my standard setup, no one scores him. Instead, Iran_Neolithic + AG3 is preferred for South Indians.

*Also, it would be incorrect to say that I have used MA1.

Most South Indians score AG3, who is much more recent than MA1, and is purely ANE.

EHG is, as per the academic literature, 30%-40% Villabruna-admixed.

I would encourage people to read the models, and my remarks, much more carefully.

Seinundzeit said...

Another note, for DMXX:

The West Asian models are actually reflective of "overfitting", and thus not necessarily what we want.

The Pashtun models aren't reasonable, they are perfect, in the sense that we really want fits that are as tight as that, but not as tight as what we see with West Asians.

West Asians are overfitted, because so many of my ancient Near Eastern references are mixtures of each other.

Harry Parihar said...


Hey Sein this may sound like a tall order but can you run the South Asian samples from ANE K6 Run Plot by Khana as shown here

if you can pull it off it would much appreciated and I think you may have to contact Khana in order to get the samples.

Much Thanks and Keep up the excellent work for you and David both.

Seinundzeit said...

Hey Harry,

If I can get access to the Global_10 PCA coordinates for these samples, I will definitely do that.

Also, my standard setup is meant to be global, any population can be modeled with it.

Bur I've now tailored a modelling setup with West Eurasians (defined as anyone who is 70% or more genetically derived from ANE-related, WHG-related, and Basal Eurasian-related ancestries) in mind, but which also works with tribal South Indians.

It's somewhat different, not exactly like my standard setup.

FrankN said...

@Sein: Great job, very informative!

Some musings:
1.) I wonder whether Barcin picks up a Hellenic signal from Alexander the Great's time,

2.) Iberia_CA is a problematic source population for the apparent Levantine/ SCA admix in the El Portalon samples. See for more details (last paragraph)

3.) The fits for Austronesian/ S. Indian pops aren't particular well, indicating that some source populations are still missing. In this respect I am in particular thinking of

a) The Tibetan highland, archeologically evidenced as LGM refugium. Could the Sherpa work in this respect, or is their specific drift setting them too far apart? [Some investigation into the relation of Sherpa to MA1/AG3 might in any case be informative]

b) Research on Anopheles DNA indicates that the termination of the Monsoon during/ after the LGM created an isolated refugium in the Eastern Himalaya foothills (E.Bihar/ W.Assam). In the absence of aDNA, such a hypothetic "Eastern Himalayan" component is of course difficult to identify, but may still be worthwhile to look out for.

c) The Circum-Indian Ocean tradition (Malay settlement of Madagascar, spice trade etc.) is invariably messing up things when it comes to ASI. I wonder whether a population from around the Sulawesi Sea, where in all likelyhood the settlers of Madagascar originated, can filter out a bit of this. [I also wonder whether in fact S.Vietnamese act as a proxy for the combination of S. Chinese (Dai) and Eastern Indonesian influences].

Seinundzeit said...


You make good observations, and raise interesting questions.

Something you might find of interest, a model for the Onge:

73.55% Jarawa
11.70% Iran_Neolithic
10.00% Ami
3.00% Bougainville
1.75% Afontova Gora3


I was surprised at how good this turned out.

The Onge seem to be a much more cosmopolitan spin on the Jarawa.

They have minor West Eurasian-related, minor East Asian-related, and minor Australian/Oceanian-related admixture, all of which is missing in the Jarawa.

I read some anthropological research on these peoples, and it seems this is a very reasonable result.

The Onge have been exposed to far more externally oriented cultural/trade relationship/contacts, over a far longer period of time, compared to the Jarawa.

One only has to examine the history of these islands, to see the dynamics at play.

Regardless, based on everything I've read, the Jarawa are a far more isolated people (they've maintained active hostilities with outsiders, till very recently).

Anyway, these are some models, using a new setup tailored for West Eurasians:

Me (again, I'm a Pashtun with roots in both Afghanistan and Pakistan)

39.45% Iran_Neolithic + Iran_Hotu + Iran_Late_Neolithic
27.45% Samara_Eneolithic
14.80% Iran_Chalcolithic
14.10% Austroasiatic_Savara
4.20% Mongola


Afghan Pashtun, Ghazni (Ghilzai)

39.15% Iran_Chalcolithic
30.75% Samara_Eneolithic
15.90% Austroasiatic_Savara
14.20% Iran_Neolithic


Usually me and my Afghan Pashtun friend have around the same amount of Jarawa/Onge/Agta, but it seems they have slightly more of the Austroasiatic-related admixture.

To check how the Savara actually stack up, in terms of deep ancestry, I tried the same model with them that I used for the Onge. Here is the output:


64.75% Jarawa
27.30% Iran_Neolithic
7.90% Ami
0.05% AG3


Not a great fit, while the Onge model was very tight.

I guess this means that Jarawa-related groups are far from being perfect references for South Asian-specific ENA. But they are perfect for the Onge, since they're from the same islands! No surprises there.

Regardless, I wanted to see if other "Negrito" and Oceanian peoples score any West Eurasian admixture, in the context of this new ENA setup. This would constitute a good "sanity check":


52% Ami
32% Jarawa
16% Bougainville



59.00% Ami
28.25% Jarawa
12.75% Bougainville



100% Bougainville


Yeah, with the same setup used on the Onge, no other "Negrito" peoples are displaying any West Eurasian affinity (and they all have great fits, like the Onge).

And a Papuan population turns out to be 100% Australian, no East Asian or West Eurasian admixture (although considering the intense differentiation between Papuans and Australians, this fit was expected to be really bad).

So, it is 100% airtight, the Andamanese Onge have West Eurasian, East Asian, and perhaps even very minor/slight Oceanian admixtures, even though they are still around 75% Jarawa-related.

This actually explains a lot.

Anyway, for fun, I ran my data through the ENA/South Eurasian setup (again, only for fun, wanted to see what would happen):

56.75% Iran_Neolithic
33.65% MA1/AG3
7.95% Ami
1.65% Bougainville


Another surprisingly good fit, but not meant to be taken too seriously.

Qagan said...


Hey there again! How are your results? Are you already registered in Anthrogenica?

Seinundzeit said...

Okay, so I was able to produce an exceedingly good ASI ghost, using some of the Austroasiatic populations as a base.

Essentially, it is the Austroasiatic_Savara, but without 20%-30% West Eurasian admixture, and it really improves the South Indian models.

I'll post some stuff next week (after today, this week will be very busy for me).

khana said...

^^ Excited to see the ASI ghost.
Working much too hard Sein :)

Seinundzeit said...


I think you'll really like this experiment.

After I find some time to do additional tests with this experimental population, in the context of multiple setups, I am definitely going to send you the coordinates.

This "ghost" does amazing things for South Indians, North Indians, and even eastern Pakistanis. It's clearly very close to the population that we were missing for South Asia; it just works way better, compared to the Jarawa/Agta/Onge.

Hopefully, once we get an early Mesolithic genome from Peninsular South Asia, and David runs it through his PCAs, I will be looking to see how well it matches my ghost (lol). Personally, I am willing to bet the correspondence will be tight.

Even people from West Asia, Central Asia, and Afghanistan/northwestern Pakistan see noticeable improvement. The patterns finally make sense for everyone, rather than just for some, or for most.

Regardless, it is intelligent/brilliant/interested people like you who make this stuff worth sharing. :)

So yeah bro, I'll make sure to eventually send the coordinates your way.

Qagan said...


Wow very interesting! I wonder what the indigenous South Asian component from Mesolithic period will be like!

I know that the Jarawa/Andamanese seems to completely lack West Eurasian but are still pretty different from the actual "ASI"

Would Bonda, Gadaba have the least West Eurasian admixture among in mainland South Asia at around 15-20%?

I will appreciate your generous reply to my question.

Thank you.

Seinundzeit said...


I have to hit the sack, but I don't want to leave you hanging, so I'll try to quickly chime in.

The Jarawa are definitely free of West Eurasian admixture, although the Onge seem to be 10%-15% West Eurasian.

Since this 10%-15% West Eurasian ancestry is lacking in the Jarawa, and since it is lacking in other "Negrito" populations throughout Asia and Oceania, it can't be a fluke.

Rather, it's a product of historical interactions between these people and populations from Southeast Asia + South Asia (as the Onge also seem to be 10% East Asian, and have an extra affinity towards Oceanians when compared to the Jarawa).

After reading up on the anthropology and history of the islands, it all makes sense.

Regardless, the Bonda are probably the least West Eurasian-admixed of all mainland Indians. They seem to be only 20% West Eurasian (and that too Iranian Neolithic-related admixture, nothing else).

But they are also 15%-20% Southeast Asian.

The Savara are around 25%-30% West Eurasian, but their Southeast Asian ancestry is very negligible.

So, I used the Savara to base/ground my artificial population. It just lacks their 25%-30% West Eurasian admixture.

So far, the results have been really good, everything finally clicks together.

Once I get time later this week, or early next week, I'll post the output here.

Qagan said...


Well that's weird how the Onge have 10-15% West Eurasian admix while the Jarawa don't have it at all. I find it strange as they are very close geographically, being neighbors.

Can you explain why and how Onge acquire this 10-15% West Eurasian admix while the Jarawa didn't?

Hmm ok so Bonda have the least in mainland at 20% West Eurasian. Do Khasi, Garo have similar amounts of West Eurasian as Bonda or more?

Oh ok so that's the reason why Savara is used as ASI proxy.

Yes I look forward to seeing the results.

Davidski said...

I don't think Onge have any West Eurasian admixture. The inferred West Eurasian admixture in this test is probably a quirk of modeling ancestry along PC clines, because Onge are basal to other East Eurasians. From memory, something similar shows up with D-stats.

Seinundzeit said...


I was reading some very interesting material on this topic.

Although, I should mention that the Andamanese_Onge samples might be mislabeled.

There is a good chance that these samples actually come from Great Andamanese people, who are well known to have varying amounts of Indian and Burmese ancestry.

I think this is what's really happening, as none of the other "Negrito" and Oceanian people I've tested show any West Eurasian admixture, and the neighboring Jarawa certainly don't seem to have any ancestry like that.

Anyway, I created another ASI ghost, based on South Indian tribal populations, and cross-checked with the least East Asian-shifted Austroasiatic Indians. The results have been very interesting.

I'll post the output here, next week.

Seinundzeit said...


Could you test this with some quick d-stat comparisons, using these particular Andamanese_Onge samples, and the Jarawa?

The Jarawa seem even more basal, and yet they don't show any West Eurasian admixture in any modelling I've tried, so I think these Andamanese_Onge are mislabeled, they might actually be Great Andamanese (indigenous Andaman people who have a well documented history of some Indian and Burmese ancestry).

I guess you could compare the Andamanese_Onge you used in the PCAs with the Jarawa you also used in the PCAs, in the context of Iran_Neolithic, MA1, and Southeast Asian affinity.

Thanks in advance.

Davidski said...


I don't think they're mislabeled and they're unlikely to have West Eurasian ancestry. Here are some D-stats. Looks like my memory was a bit hazy.

Mbuti Loschbour Dai Onge -0.0112 -2.607
Mbuti Loschbour Han Onge -0.0116 -2.681
Mbuti Loschbour Jarawa Onge -0.0088 -1.746

Mbuti Iran_Neolithic Dai Onge -0.0159 -3.545
Mbuti Iran_Neolithic Han Onge -0.0145 -3.264
Mbuti Iran_Neolithic Jarawa Onge 0.0001 0.024

There might be something a little bit off about them. I always use a couple of individuals from this batch that appear the most sound, but to err on the side of caution, I wouldn't delve into the population history of the Andamanese Islands with them or the PCA method.

Seinundzeit said...


In the PCA-based modelling, they differ rather significantly from the Andamanese Jarawa.

So this is pretty interesting.

If d-stats can't detect any shift towards West Eurasians versus the Jarawa, I'm guessing they behave oddly due to random drift? If it isn't due to admixture, that has to be it.

Davidski said...

Not sure what's happening. Might take them out of the datasheet just to be sure.

Seinundzeit said...


There are some very interesting patterns related to all of this.

For example, if one creates an ASI ghost using an Indian Austroasiatic population as the primary base, one creates an artificial population that behaves just like the Onge, but which provides much better fits. It's amazing how the percentages correspond very closely, yet the models look much better with the Austroasiatic-based ghost.

But if one creates an ASI ghost using the Pulliyar and Paniya, the percentages closely follow what we see using the Jarawa. But again, the fits are far, far better than what we see using the actual Jarawa.

I'll eventually post models using the Onge, models using the Jarawa, models using a hypothetical ASI ghost based on Indian Austroasiatic populations, and models using an ASI ghost based on South Indians. That'll showcase the phenomenon I have in mind.

All things considered, I wonder about what these differences demonstrate? Perhaps, genetic substructure in Mesolithic South Asia?

Shaikorth said...

Sein, have you tried modeling Jarawa like you did Onge, with Onge as one of the sources? If the results look the same it probably means there's no real difference between the two.

Seinundzeit said...


When I switch places between the Jarawa and Onge, this is how the Jarawa turn out:

92.7% Onge
3.9% Ust-Ishim
2.5% Papuan


Compared to when the Onge are tested in the same setup:

72.90% Jarawa
11.90% Iran_Neolithic
11.00% Ami
2.45% Papuan
1.75% AG3


I have no clue about what we should make of these differences.

Shaikorth said...

Interesting, can't outrule Onge being more admixed if Ust-Ishim+Papuan (sans Denisovan) approximates original Andamanese stock but that's an if. How about something similar to what van Dorp et al. did with Ari using Chromopainter output: both Jarawa and Onge tested without Jarawa and Onge donors.

Seinundzeit said...

That's an interesting idea, modelling them as Ust-Ishim + Australian/Oceanian + East Asian + ANE:


84.25% Ust-Ishim
7.10% Ami
5.35% Bougainville
3.30% AG3



84.05% Ust-Ishim
10.10% Ami
3.25% AG3
2.60% Bougainville


Differences persist, but the differences aren't as intense/strong as in their other models.

Shaikorth said...

Yeah, based on that result I'd say they're quite similar and that there doesn't seem to be Iran_N in Onge. Basu et al. had many issues but along the first 4 dimensions in their South+East Asian PCA Jarawa and Onge were indistinguishable.

Regarding your ASI construct, if it gives percentages that are similar to ones you get with Jarawa but better fits, wouldn't that back the idea that ASI and Andamanese indeed derive from a single root, but isolation since the Paleolithic has added much population-specific drift to the islanders, worsening fits? Population-specific drift is the kind of thing that shows on PCA dimensions.

Seinundzeit said...


For what it's worth, I excluded Iran_N from the fit, as they both get substantial percentages with those samples included, with the Onge a little more. In that context, the Onge also have less Australian/Oceanian and more East Asian, than the Jarawa.

Since the d-stats can't detect any West Eurasian shift, I suppose it really does boil down to genetic drift making these populations differ in random ways?

Although, I think you've hit the nail right on the head, with regard to the Andamanese-ASI relationship.

For example, when I try to model myself (and other people from South Central Asia) using the Onge, the Jarawa, and my simulated ASI populations (one based on tribal Austroasiatic Indians, and another based on tribal South Indians), me and all other South Central Asians choose the simulated ASI populations, rather than the actual Andamanese populations, even though the proportions are quite close (often identical) to what we get using the Jarawa and Onge.

The fits are (so far) the best I've ever obtained, extremely good. Especially compared to what I was previously finding (when all I had were the Andamanese populations and "Negrito" people from the Philippines).

Also, these simulated populations allow for much cleaner models, with a very precise delineation of what is East Asian (of both the northern and southern kind, in the case of Central Asia/South Central Asia for the former, and in the case of northeastern India/Bangladesh for the latter) and of what is "indigenous" to South Asia:


51.40% Iran_Neolithic
20.85% Samara_Eneolithic
17.35% ASI (South Indian-based)
8.50% Iran_Chalcolithic
1.30% Iberia_Chalcolithic
0.60% ASI (Austroasiatic-based)

(0% from both the Jarawa and the Onge)



36.10% Iran_Neolithic
33.90% Samara_Eneolithic
16.70% Iran_Chalcolithic
13.30% ASI (7.65% South Indian-based + 5.65% Austroasiatic-based)

(0% from both the Jarawa and the Onge)


Afghan Pashtun, Ghazni (Ghilzai)

36.95% Iran_Chalcolithic
30.25% Samara_Eneolithic
18.35% Iran_Neolithic
13.75% ASI (Austroasiatic-based)
0.70% Mongola

(0% from the Jarawa, the Onge, and the South Indian-based ghost)


Afghan Pashtun, Khost (Karlani)

31.80% Samara_Eneolithic
26.80% Iran_Chalcolithic
22.15% Iran_Neolithic
8.65% ASI (6.00% Austroasiatic-based + 2.65% South Indian-based)
5.30% Barcin_Neolithic
3.10% Mongola
2.20% Iberia_Chalcolithic

(0% from both the Jarawa and the Onge)



39.15% Iran_Neolithic
30.00% Samara_Eneolithic
16.65% Iran_Chalcolithic
8.90% ASI (South Indian-based)
5.30% Mongola

(0% from the Jarawa, the Onge, and the Austroasiatic-based ghost)


Again, this is definitely the best fit I've ever found for myself, and the most sensible one yet.

All things considered, I just think it's very interesting that simulated populations (based on tribal South Indians and Indian Austroasiatics) allow for a huge improvement in the fits, and take the place of Andamanese populations.

The improvement in modelling also applies to South Indians, who finally have tight fits.

It all comes back to what you've said concerning ASI and the Andamanese deriving from a single root, but with the Andamanese being very distant to ASI due to isolation/drift.

Qagan said...


I see so probably mislabeling problems probably swapping Onge for Great Andamanese.

It seems that Indian ancestry is responsible for West Eurasian admixture in Great Andamanese then. But you also mention Burmese ancestry. Do Burmese also have some West Eurasian admix?

Look forward to seeing the other ASI ghost.

Thank you again.

khana said...

@Sein, Did you model the 2 ghost ASI populations based on something like a midpoint between Jawara-Onge (Austroasiatic ASI) and Onge-Paniya (South Indian ASI)?

I'm curious to see if something like that cleans up the fits for my first run, overcoming the PC8/9 dimension issues.

Have you tried modeling with the new ASI ghosts in one-on-one runs of Yamnaya-Samara, Andronovo, Shintashta or the Corded-Ware pops, against Samara_Enolithic? I wonder if with the ASI ghosts, populations will consistently choose Samara Enolithic over the other steppe groups consistently.

I have a few more coordinates for you to try. I know you are busy this week so I'll e-mail you them a bit later in the week.

Seinundzeit said...

Hey Khana,

What I did was rather easy. Which is why I was so surprised/pleased, I didn't expect the models to improve (especially to such a substantial extent).

At the end of the day, the South Indian-based ghost is just the non-West Eurasian half of the Pulliyar population. Using the Jarawa/Onge, the Pulliyar are consistently around 50% West Eurasian (Iran_Neolithic-related) and around 50% Andamanese-related.

So, it's pretty easy to use the Pulliyar PCA coordinates right in conjunction with the PCA coordinates for Iran_Neolithic. With the Pulliyar and Iran_Neolithic populations side-by-side, one can then arrive at the non-West Eurasian half of their ancestry.

For what it's worth, their non-West Eurasian ancestry actually constitutes a clade with the Jarawa. For example, one can model their non-West Eurasian ancestry as 100% Jarawa, just like how the Kosipe/Papuan can be modeled as 100% Bougainville. And just like how the Kosipe/Papuan fit very poorly as 100% Bougainville, this South Indian-based ghost fits really bad as 100% Jarawa. But, the Jarawa are the closest population to them, out of all contemporary populations.

The Austroasiatic ghost was arrived at via the same simple method, on the assumption that the Savara are 25%-30% West Eurasian (which is what I've seen in my modelling). This ghost is far more similar to the Onge. Although, the same applies to this simulation, it fits very poorly as 100% Onge, just like how the South Indian-based simulation fits very poorly as 100% Jarawa.

Both simulated populations allow for a massive improvement in how the models fit, for everyone from Tajikistan (and eastern Iran) to deep inside South India. But both simulations differ as much from each other (in terms of the proportions they provide) as the Jarawa and Onge.

I find that the South Indian-based ghost provides better models (in terms of how tight the fits are) compared to the Austroasiatic-based ghost.

But to repeat, even the Austroasiatic-based ghost allows for way better fits, and way cleaner models, compared to anything I've found using the Jarawa/Onge/Agta.

Also, I am definitely going to try some models with Yamnaya-Samara, Andronovo, Sintashta, Srubnaya, etc, in conjunction with these ASI ghosts.

Honestly, my choice of Samara_Eneolithic was driven by a desire to see if the ANE affinity in South Central Asia could be modeled as having origins in some local forager substratum (to that end, MA1, Afontova Gora3, and Iran_Hotu were included) or in admixture from the ancient steppe/ancient Eastern Europe (or perhaps from ancestry with roots in both of these supposedly ANE-rich streams of ancestry).

In that sense, it was/is quite successful (Tajiks, Pashtuns, Kalash, etc, all prefer Samara_Eneolithic, while South Indians and scheduled caste North Indians all prefer AG3/MA1).

Now that it's obvious that ANE affinity in South Central Asia is steppe/ancient Eastern European in origins, not anything local to South Central Asia, I'll do some models with later steppe populations.

Regardless, absolutely, I'd love to try some more samples, so please do send those coordinates, whenever you want. And I'll make sure to send you the coordinates for my simulated populations.