Ancient DNA | Y-DNA | China | Roman Empire
The Lost Roman Legion of China
In a village of Gansu, on the edge of the Gobi, some people have green eyes and light hair, and a famous theory says they descend from Roman soldiers captured at Carrhae in 53 BC. Paternal lineages of the villagers, ancient genomes from the Han dynasty commandery where Liqian lay, and our own Global25 tests point to a different story: western ancestry is real along the Gansu corridor, but it came from the Silk Road, not from Rome.
Hero image: AI generated illustration of a frontier market in the Hexi Corridor under the Han dynasty: a gate tower with red banners on a rammed earth wall, soldiers in lamellar armour with halberds, an official reading bamboo slips, farmers selling millet, wheat, jujubes and apricots, Central Asian traders in belted kaftans with Bactrian camels and a horse, bolts of silk and bronze mirrors on a stall, and a smoking beacon tower below the snowy Qilian Mountains. An artistic reconstruction, not a historical document.
In 53 BC, at Carrhae in northern Mesopotamia, the Parthians destroyed the army of Marcus Licinius Crassus, one of the richest men in Rome. Crassus was killed, and thousands of his soldiers were marched east as prisoners. What became of them is one of the great mysteries of ancient history, and since the 1950s an answer has been sought more than 5,000 kilometres away, in a village of northwest China.
The village of Zhelaizhai, renamed Liqian, lies in Yongchang County in Gansu, on the old Silk Road between the Qilian Mountains and the Gobi. Some of its inhabitants have green or hazel eyes, light hair or prominent noses, and many of them believe that they descend from a lost Roman legion. The story has drawn journalists, a Chinese and Italian research centre and a Jackie Chan film, Dragon Blade (2015). We review what history, genetics and ancient DNA can say, and add a Global25 test that asks a simple question: if there is western ancestry along the Gansu corridor, does it look Roman?
The legend
The ancient sources agree that many of Crassus's men survived Carrhae as captives. Plutarch puts the prisoners at about 10,000, and Pliny the Elder writes that the Parthian king settled them in Margiana, around the oasis of Merv in today's Turkmenistan, on the eastern frontier of his empire. Nothing more is heard of them.
In the 1940s and 1950s, the Oxford historian Homer Dubs proposed a sequel, which he set out in a 1957 book, A Roman City in Ancient China. In 36 BC, a Han Chinese army under Chen Tang stormed the fortress of the Xiongnu ruler Zhizhi, far to the west near the Talas River. The Han history describes a hundred or so soldiers drawn up in a "fish-scale formation" and a double wooden palisade around the fort. Dubs read the formation as the Roman testudo and the palisade as a Roman technique, and suggested that these men were Crassus's soldiers, who had escaped from Margiana and become mercenaries. Taken prisoner by the Han, they would have been resettled in a new county of Gansu called Liqian, a name that Dubs connected with the Chinese name for the Roman or Hellenistic west.
Historians have been sceptical from the start. No Roman coin, weapon or inscription has been found at Liqian. The fish-scale formation and the palisade can be read in other ways, the identification of the Chinese name with Rome is disputed, and the theory requires prisoners to cross Central Asia twice, first to fight for the Xiongnu and then as captives of the Han. But the legend has proved more durable than the objections, and in the 2000s it turned to genetics.
What the DNA tests found
In 2005, researchers took blood samples from 93 people in and around Liqian. Press reports said that some villagers had DNA of 56% "Caucasian" origin. The results were reported in the press without the details of a scientific publication, and western ancestry, even if real, would not say where it came from.
A formal study followed in 2007. R. Zhou and colleagues typed 12 STR markers and 12 SNPs on the Y chromosomes of 227 men from four Chinese populations, including Liqian. Some 77% of the Liqian Y chromosomes belonged to haplogroups restricted to East Asia. The Liqian clustered with other Chinese populations, especially the Han, and stood apart from Central Asian and western Eurasian groups. Admixture estimates showed a large Han contribution to their gene pool. The authors concluded that, on the paternal side, a Roman mercenary origin could not be accepted, and that the people of Liqian are most likely a subgroup of the Han majority.
Ancient genomes from Liqian's own commandery
Liqian county lay in Zhangye commandery, and Han dynasty burials from Zhangye have now been sequenced. In 2022, Jianxue Xiong, Shaoqing Wen and colleagues analysed the paternal and maternal lineages of 31 individuals from the Heishuiguo cemetery, dated to the Western and Eastern Han, exactly the period of the legend. The men carried mostly O lineages from the Yellow River basin, along with C2, N and one Q, and many of the women northern East Asian maternal lineages such as D4 and D5; no western Eurasian lineage was found on either side. The authors traced the change to a migration of men from the Chinese heartland, recorded in Han documents as the transfer of settlers to the frontier.
In a 2023 study in Science Bulletin, the same team sequenced 30 ancient genomes from the Hexi Corridor. The 17 individuals from Heishuiguo descended from Late Neolithic and Bronze Age Yellow River populations, with no significant western Eurasian contribution. At Foyemiaowan, near Dunhuang at the western end of the corridor, two individuals from the third to ninth centuries stood out, with roughly 50% and 30% western ancestry. Models with Bronze Age steppe-related sources fitted them, although the authors could not pin down the exact source, and the admixture was biased towards women on the western side. The same study found 5 to 20% western ancestry in present-day Gansu groups, with admixture dated to 600 to 1,000 years ago, the era of the Mongol expansion. At Foyemiaowan as a whole, an earlier study of 34 individuals found 8.3% western Eurasian maternal lineages and no western paternal lineage.
Our analysis: how much western ancestry, and what kind?
There are no Global25 data for the Liqian villagers themselves. But there are data for their neighbours: the Hui, Dongxiang, Bonan, Salar, Yugur and Tu of Gansu and Qinghai, Han Chinese from Lanzhou and northern China, the Uyghurs farther west, and the ancient people of Heishuiguo and Foyemiaowan. We modelled each population with an eastern side (Han dynasty Hexi and Tibetan sources) and three possible western sources: Romans (Republican Latium), Iranian speakers of the Silk Road (Kushan-era Uzbekistan) and steppe nomads (the Saka of the Tian Shan). To test the robustness of the result, we ran 96 versions with three Roman references (Republican Rome, Republican Latium, Imperial Rome), four Iranian (Kushan-era Uzbekistan, Parthian Iran, Iron Age Margiana, present-day Tajiks) and four steppe references (Saka, Wusun, Kangju, Iron Age Xinjiang), and two eastern bases.
Western ancestry is clearly present along the corridor. It makes up about 16% of the Dongxiang genome (15 to 22% across versions), 15% of the Salar, 12% of the Hui and Bonan, 13 to 14% of the Yugur, and 5 to 7% of the Tu and Western Yugur. The Han of Lanzhou carry about 3 to 4%, the Han of northern China almost none. The Han dynasty people of Heishuiguo, in Liqian's own commandery, and the main group at Foyemiaowan carry none. The two Foyemiaowan outliers come out at about 56 to 61% and 33%, close to the published estimates, and the Uyghurs at about 60%. In all of these populations, the Roman source receives 0% in the main model.
Which western source fits best?
A simple test is to add one western source at a time and compare how well each model fits. Lower values mean a better fit.
In every population with western ancestry, the Roman sources give the worst fit. For the Dongxiang, the best Iranian source gives a distance of 12.2 and the best steppe source 13.6, against 23.0 for the best Roman one; for the Salar, 10.1 and 11.6 against 19.9; for the Hui, 12.7 and 13.1 against 19.5. The contrast is extreme for the Uyghurs (22.4 for Iranian, 74.8 for Roman) and for the Dunhuang man of the Wei dynasty (24.1 for steppe, 82.8 for Roman). Among the Tu and Western Yugur, who carry little western ancestry, the differences are smaller, but the ranking is the same.
The fingerprint of the western ancestry
A second way to identify the source is to break the western part of each genome into the ancient components that make up all western Eurasians: early farmers from Anatolia, early farmers from Iran, and steppe herders and hunter-gatherers. Romans of the Republic and Empire were about two-thirds Anatolian farmer in this sense. Iranian speakers of Central Asia and the Saka were not.
The western ancestry of the Dongxiang, Salar, Hui and Bonan is only 18 to 22% Anatolian farmer, the rest split between Iranian farmer and steppe ancestry, the same profile as the Kushan-era people of Uzbekistan (27%), present-day Tajiks (23%), the Wusun (15%) and the Saka (13%). Romans, at 63 to 65%, are in a different class. Because the western part of these genomes is small, this breakdown is less precise than the previous tests, but it points the same way.
Would Roman ancestry be detected?
An absence of evidence is only meaningful if the method could have found the evidence. We therefore added known amounts of Roman ancestry to the genome of the Han dynasty people of Heishuiguo, with random noise, fitted the model with different references 100 times for each amount, and recorded what it found. We did the same with Iranian and steppe ancestry, to see whether they could be mistaken for Roman.
Roman ancestry is easy to detect in this setting: 2% added is found as 1.4%, 5% as 4.5%, 10% as 9.5% and 20% as 19.6%. Adding 20% of Iranian or steppe ancestry produces a false Roman signal of only 1 to 2% on average, and at most about 3%. In the Dongxiang, Salar, Hui and Bonan, the Roman share has a median of 0.0 to 0.3% and never exceeds 4% in any of the 96 models. Even on Dubs's own account, a hundred or so Roman soldiers would have been a drop in the population of a frontier county, and two thousand years of intermarriage would have diluted them further; but the genomes of the Gansu corridor give no sign of them at all.
Where do the green eyes come from?
If not from Rome, where does the western ancestry of Gansu come from? The corridor was a crossroads long before the Han. Before the second century BC, it was home to the Yuezhi, a people probably of Indo-European speech, driven west by the Xiongnu, and to other steppe groups. Under the Han and the Tang, Sogdian merchants and other Iranian speakers settled along the Silk Road; non-Han peoples of the Liqian area, known as the Lushui Hu, even ruled the Northern Liang kingdom from 401 to 439. Under the Mongols, large numbers of Central Asian and Persian Muslims moved into northwest China, among the ancestors of the Hui. The genomes record this: the western ancestry of the Hexi Corridor appears in individuals of the third to ninth centuries, it is of Iranian and steppe type, and the admixture seen in today's Gansu peoples is dated to the last thousand years.
A few percent of western ancestry is enough to produce the occasional light-eyed or fair-haired child in a mostly East Asian population. Liqian sits in the middle of a region where such ancestry has circulated for more than two thousand years. The villagers' features are real; the Roman explanation is not needed to account for them.
So, are the villagers of Liqian Roman?
Almost certainly not. The only study of the villagers themselves found that 77% of their paternal lineages are East Asian and that they are closest to the Han Chinese. The Han dynasty people buried in Zhangye, the commandery that included Liqian, carried no western ancestry at all. And the western ancestry that does exist in the neighbouring peoples of Gansu has the fingerprint of Iranian speakers and steppe nomads, not of Romans, with no detectable Roman component even though our tests show that 2% would be found. The legend tells us more about the twentieth century than about the first century BC: about the fascination of East meeting West, and about how a good story can outlive the evidence.
Some doubts remain. The Liqian villagers have never been studied genome-wide, and the 2007 study looked only at the paternal line. A full genome study of the village, with the same tests, would settle the question for good. On present evidence, the people of Liqian are Chinese with a share of Silk Road ancestry, like many of their neighbours, and the lost legion of Crassus is still lost. As with the Kalash and the soldiers of Alexander, which we discussed in our article on the Kalash, a romantic origin story and the genome rarely agree.
Key points
- The theory that the people of Liqian descend from Roman soldiers captured at Carrhae in 53 BC dates from the 1950s and rests on indirect arguments; no Roman object has been found at Liqian.
- A 2007 study of 227 men found that 77% of Liqian Y chromosomes belong to East Asian haplogroups and that the Liqian cluster with the Han Chinese.
- Han dynasty genomes from Heishuiguo, in the commandery that included Liqian, show no western Eurasian ancestry and no western paternal or maternal lineages.
- The Dongxiang, Salar, Hui and Bonan of Gansu carry 12 to 16% western ancestry, but in every model it fits Iranian and steppe sources better than Roman ones, and its fingerprint is that of Central Asia, not of Italy.
- Our tests detect as little as 2% Roman ancestry; the real Gansu populations show a median of 0 to 1% and never more than 4% in 96 models.
- The western ancestry of the corridor comes from the Silk Road: Yuezhi and Saka before the Han, Sogdians and other Iranian speakers later, and Central Asians in the Mongol era.
Methods
All analyses use Global25 coordinates. Models were fitted by non-negative least squares with proportions summing to one; distances are Euclidean distances multiplied by 1,000. The eastern side of each model combined the Han dynasty people of Heishuiguo (11 individuals) or present-day northern Han Chinese with Tibetans. The western sources were Republican Rome and Republican Latium of Italic profile and Imperial Rome (Antonio et al. 2019 and other studies), Kushan-era Uzbekistan, Parthian-period Iran, Iron Age Turkmenistan and present-day Tajiks, and the Saka of the Tian Shan, the Wusun, the Kangju and Iron Age Shirenzigou in Xinjiang. All 96 combinations were run for each population. The fingerprint model replaced the western sources by Neolithic Anatolian farmers (Barcin), Neolithic Iranian farmers (Ganj Dareh), Yamnaya steppe herders and a western hunter-gatherer. Tests added known amounts of a source to the Heishuiguo average, with random noise of length 10 units, and estimated it again 100 times with other references. The Han of Lanzhou are represented by only two individuals, and the Dunhuang outliers by one individual each.
Global25 coordinates used in this article
Gansu and neighbouring populations, the ancient Hexi Corridor genomes and the Roman, Iranian and steppe references used in the models. They can be pasted directly into the Vahaduo tool.
Hui_Gansu,0.030949,-0.386433,0.005423,-0.048958,0.032138,0.016800,0.005674,0.001835,-0.010129,0.001128,-0.061174,-0.008764,0.010066,-0.003788,-0.005377,0.000278,-0.000143,0.000459,-0.004214,-0.003520,0.004148,0.009445,0.008017,0.000683,0.007265
Dongxiang,0.035465,-0.363292,0.004853,-0.045186,0.028191,0.017592,0.004941,0.002459,-0.011534,0.000504,-0.057998,-0.007241,0.008427,-0.004802,-0.003518,0.001727,0.001664,-0.000727,-0.002699,-0.003795,0.004216,0.009642,0.007259,0.000022,0.009441
Bonan,0.030476,-0.385571,0.007703,-0.046431,0.030113,0.015207,0.005376,0.002596,-0.009510,0.002601,-0.064221,-0.007403,0.007017,-0.005869,-0.005286,0.000199,0.001199,-0.000510,-0.004930,-0.003577,0.003846,0.011249,0.008516,0.000148,0.010448
Salar,0.034247,-0.369832,0.004570,-0.045477,0.029399,0.014937,0.005274,0.002919,-0.010918,0.001227,-0.060814,-0.007388,0.007831,-0.005813,-0.004934,0.000897,0.001220,0.001435,-0.003508,-0.004969,0.003813,0.009518,0.008624,-0.000021,0.010147
Yugur,0.034602,-0.374900,0.020264,-0.039524,0.010402,0.004397,0.008546,0.005361,-0.005979,0.004519,-0.060371,-0.005700,0.009108,-0.005679,-0.005112,0.000411,-0.000065,-0.001503,-0.002233,-0.000846,-0.000886,0.010552,0.002067,-0.001233,0.013508
Western_Yugur,0.027531,-0.416367,0.011219,-0.055435,0.037122,0.014781,0.006492,0.001716,-0.007606,0.004305,-0.067381,-0.008168,0.009050,-0.005281,-0.007727,-0.001028,-0.000367,0.001330,-0.002239,-0.004518,0.004921,0.009892,0.008804,-0.000316,0.006002
Tu_Monguor,0.023903,-0.404181,0.009805,-0.046254,0.027943,0.009650,0.005734,0.003600,-0.006013,0.005175,-0.067131,-0.008422,0.009484,-0.003936,-0.007410,0.000849,0.000965,-0.001013,-0.005153,-0.001951,0.002421,0.012860,0.007370,0.001518,0.014298
Han_Lanzhou_Gansu,0.024472,-0.430077,0.009428,-0.060401,0.048932,0.015339,0.005875,0.002999,-0.014828,0.001549,-0.063819,-0.011165,0.010109,-0.005642,-0.008212,-0.001393,0.003325,-0.002217,-0.001257,-0.006315,0.006301,0.007852,0.008134,0.004217,0.008922
Han_northern_China,0.023343,-0.446599,0.008587,-0.064078,0.051058,0.020973,0.005286,0.001832,-0.010629,0.003594,-0.072520,-0.008826,0.011063,-0.005933,-0.007310,-0.001075,0.000594,-0.000138,-0.003349,-0.008862,0.010904,0.007107,0.012232,0.000892,0.001022
Uyghurs,0.065604,-0.165254,0.001371,-0.001938,-0.027152,0.006237,0.006324,0.003273,-0.014131,-0.008971,-0.027724,-0.002609,0.001473,-0.006506,0.002893,0.007009,0.000699,-0.001543,-0.001051,-0.000415,-0.007963,0.000051,-0.000868,0.001912,0.003130
Han_dynasty_Hexi_Heishuiguo,0.017694,-0.449603,0.007508,-0.062163,0.058389,0.018787,0.005491,0.001301,-0.010728,0.006958,-0.073916,-0.009346,0.008866,-0.007131,-0.005096,-0.000398,0.000936,0.001048,-0.005577,-0.011790,0.015405,0.005710,0.013400,-0.003177,0.000131
Dunhuang_Wei_to_Tang_Foyemiaowan,0.020773,-0.447595,0.011973,-0.063954,0.050509,0.021056,0.007403,-0.000750,-0.013294,0.003964,-0.077033,-0.008524,0.009979,-0.008825,-0.004496,-0.003364,-0.000326,0.003753,-0.003787,-0.005800,0.013102,0.005394,0.014127,-0.000075,0.005089
Dunhuang_outlier_Wei_dynasty,0.070570,-0.185842,0.017348,0.002907,-0.005539,0.023427,0.000940,0.005077,-0.027202,-0.014943,-0.042708,-0.004796,-0.006541,-0.017065,0.009636,0.008618,0.001825,-0.000760,-0.000754,-0.008129,-0.004866,0.000371,0.008134,0.006868,0.005029
Dunhuang_outlier_Tang_dynasty,0.047806,-0.282317,-0.000377,-0.029393,0.018773,0.010598,0.000940,0.003923,-0.019430,-0.006378,-0.049528,-0.002248,0.007582,-0.004266,0.000950,-0.000928,0.004824,-0.001267,-0.007919,-0.006753,0.007362,0.007914,0.002465,0.000602,0.004071
Tibetan_avg19,0.023663,-0.410541,0.003612,-0.036312,0.015290,0.006605,0.007928,0.006364,0.004467,0.013294,-0.082075,-0.011911,0.012135,-0.004549,-0.010715,-0.004355,0.003719,-0.003747,-0.006649,0.005279,-0.000164,0.024926,0.007713,0.000450,0.033284
Republican_Rome,0.128783,0.152620,0.038682,-0.013012,0.046514,-0.005179,0.001376,-0.000264,0.022088,0.041185,-0.000673,0.011069,-0.019751,-0.009948,-0.000911,-0.002349,0.004154,0.002317,0.006052,-0.002376,0.000392,0.003180,-0.004384,-0.004252,-0.001847
Republican_Latium_Italic,0.126344,0.155566,0.038372,-0.013223,0.047028,-0.007548,-0.000367,-0.000173,0.023686,0.044967,-0.001411,0.010641,-0.021082,-0.006511,-0.002714,-0.006041,-0.000024,0.002700,0.004022,-0.007824,-0.002035,0.004683,-0.003397,-0.004549,-0.001227
Imperial_Rome,0.108815,0.149950,-0.023263,-0.056876,0.003974,-0.020566,-0.000692,-0.005690,-0.003150,0.021014,0.003633,0.002544,-0.005649,-0.000613,-0.010633,-0.002227,0.004951,0.000981,0.002507,-0.004917,-0.002763,0.002971,-0.001651,0.000685,0.000339
Kushan_era_Uzbekistan,0.094359,0.075555,-0.037561,0.012759,-0.042069,0.010402,0.008178,-0.000831,-0.034565,-0.026734,0.001267,0.000000,0.002111,-0.007501,0.014495,0.013670,-0.009244,0.004016,0.004412,-0.012093,-0.003157,-0.004019,0.000172,0.002193,0.000156
Parthian_Iran,0.092197,0.111962,-0.093526,-0.027778,-0.070398,0.005229,0.005288,-0.000577,-0.063403,-0.030206,-0.005805,0.001836,-0.002825,-0.001032,0.011366,0.021148,-0.004531,-0.003357,0.003237,-0.007379,0.006177,-0.008130,-0.001818,-0.011447,0.012783
Iron_Age_Turkmenistan_Margiana,0.103010,0.093937,-0.017159,0.043282,-0.033699,0.026356,0.001998,-0.004038,-0.044075,-0.041277,-0.011448,-0.001649,0.001264,-0.019817,0.023276,0.016573,-0.012191,-0.001710,0.000188,-0.010630,0.001622,0.000000,0.001048,0.009700,0.001437
Tajiks_today,0.083020,-0.040367,-0.020765,0.014696,-0.039834,0.012271,0.005978,0.003057,-0.020682,-0.012939,-0.011956,-0.001677,0.002574,-0.009874,0.006056,0.011046,0.000196,0.000958,0.001312,-0.002861,-0.007619,-0.001808,0.000185,0.001054,0.003121
Saka_Tian_Shan,0.093172,-0.030031,0.030385,0.051219,-0.040359,0.015697,0.001410,0.000165,-0.025566,-0.031215,-0.008073,0.000343,-0.000170,-0.021567,0.015104,0.009698,0.001285,-0.001593,0.001005,0.002090,-0.017558,0.000954,-0.000088,-0.000000,-0.003130
Wusun,0.102441,0.031482,0.005280,0.052003,-0.047393,0.022172,0.009048,-0.005539,-0.033133,-0.034078,-0.004059,0.001049,0.006987,-0.014863,0.013504,0.002784,-0.014082,-0.001267,0.000817,-0.005190,-0.016845,0.001422,0.000308,-0.002831,-0.002874
Kangju,0.099595,0.049423,0.016091,0.046512,-0.033186,0.018686,0.003173,0.001192,-0.026179,-0.031861,-0.004980,-0.002323,0.000867,-0.021285,0.018096,0.016772,-0.001543,-0.000866,-0.000901,-0.005023,-0.008693,0.000618,0.000514,0.003294,-0.000998
Iron_Age_Xinjiang_Shirenzigou,0.071851,-0.170736,0.035167,0.033713,-0.046509,0.009447,-0.002850,-0.002365,-0.016797,-0.024169,-0.024967,0.000393,0.005779,-0.015052,0.011282,0.001724,-0.004987,-0.003864,0.001791,0.005862,-0.014615,0.007991,0.004175,0.000060,0.004490
References
- Y-DNA Zhou R, An L, Wang X, Shao W, Lin G, Yu W, Yi L, Xu S, Xu J, Xie X (2007). Testing the hypothesis of an ancient Roman soldier origin of the Liqian people in northwest China: a Y-chromosome perspective. Journal of Human Genetics 52: 584 to 591. doi.org/10.1007/s10038-007-0155-0
- Ancient DNA Xiong J, Wang R, Chen G, Yang Y, Du P, Meng H, Ma M, Allen E, Tao L, Wang H, Jin L, Wang CC, Wen S (2024). Inferring the demographic history of Hexi Corridor over the past two millennia from ancient genomes. Science Bulletin, published online December 2023. doi.org/10.1016/j.scib.2023.12.031
- Ancient DNA Xiong J, Du P, Chen G, Tao Y, Zhou B, Yang Y, et al., Li H, Wen S (2022). Sex-biased population admixture mediated subsistence strategy transition of Heishuiguo people in Han dynasty Hexi Corridor. Frontiers in Genetics 13: 827277. doi.org/10.3389/fgene.2022.827277
- Ancient DNA Xiong J, Tao Y, Ben M, Yang Y, Du P, Allen E, et al., Wen S (2022). Uniparental genetic analyses reveal multi-ethnic background of Dunhuang Foyemiaowan population (220 to 907 CE) with typical Han Chinese archaeological culture. Frontiers in Ecology and Evolution 10: 901295. doi.org/10.3389/fevo.2022.901295
- Ancient DNA Antonio ML, Gao Z, Moots HM, Lucci M, Candilio F, et al., Pritchard JK (2019). Ancient Rome: a genetic crossroads of Europe and the Mediterranean. Science 366: 708 to 714. doi.org/10.1126/science.aay6826
- History Dubs HH (1957). A Roman City in Ancient China. London: The China Society.
- History Plutarch, Life of Crassus 31; Pliny the Elder, Natural History 6.47, on the prisoners of Carrhae and their settlement in Margiana.
- Data Global25 coordinates by Davidski (Eurogenes), and the Moriopoulos Collection 2026 of ancient and modern averages. Models, tests and figures: ExploreYourDNA.