Ancient DNA | Avar Khaganate | 7th century Hungary
The Avar woman who looked east
In grave 7 of an elite cemetery on the Hungarian plain lies a woman whose genome is almost entirely Northeast Asian. With five men buried in the same period, she lets us measure how far, and how fast, the Avars came.
Hero image: AI generated illustration of an Early Avar woman with gilded pendants on the Hungarian plain in the 7th century. It is not a reconstruction of the woman buried in Kunpeszér grave 7.
Sometime around 640 AD, a woman was buried in grave 7 of a small cemetery near Kunpeszér, between the Danube and the Tisza. She wore gilded bronze pendants of the type archaeologists call Mezöszilas, a string of beads, and carried an iron awl and a knife. The men around her were buried with silver belt mounts, bows and quivers. Her grave was not the richest in the cemetery. Her DNA is the most remarkable.
Her genome, published by Maróti and colleagues in 2022 under the code KFP-7, belongs to a person whose ancestors had lived, only a generation or two earlier, on the steppes of Mongolia. In our reanalysis she is the most eastern of the six Early Avar elite individuals in the Global25 dataset, more eastern than any of the warriors, and much more eastern than the man buried a few metres away in grave 6.
We are not claiming she was a queen. Nothing in her grave says so, and no Avar queen has ever been identified by DNA. What she offers is something rarer: a clean genetic snapshot of the first Avar generation in Europe, before mixing with local people had started to dilute it.
The people who came out of nowhere
In 557 or 558 AD, envoys of a people called the Avars arrived at the court of Justinian in Constantinople, introduced by the Alans of the North Caucasus. They asked for land and gold. Byzantine writers had never heard of them. Ten years later, in 567 and 568, under their khagan Bayan, they crossed into the Carpathian Basin, the Lombards departed for Italy, and the Avars built a khaganate that would dominate central Europe for more than two centuries, until Charlemagne's armies broke it in the 790s.
Where did they come from? The timing pointed east. In the 550s the Türks destroyed the Rouran Khaganate, the empire that had ruled the Mongolian steppe since the 5th century. Many historians proposed that the European Avars were Rouran refugees. Others followed the Byzantine historian Theophylact Simocatta, who claimed that the Avars of Europe were impostors, "Pseudo-Avars" from two Central Asian tribes, the Uar and the Chunni, who had borrowed a feared name. The archaeological record could not settle it: the Avars brought new belts, new weapons, and they are often credited with bringing the stirrup to Europe, but objects travel more easily than people.
What the genomes already told us
Three studies changed the debate. In 2020, Csáky and colleagues typed the Y chromosomes of 26 individuals from 7th century elite burials. Most high status men in the Danube to Tisza region carried haplogroup N, in a few closely related haplotypes, including the man buried at Kunbábony with more than 2 kg of gold, a candidate for a khagan. The authors concluded that the elite arrived as a group of related families and stayed largely endogamous for generations.
In 2022, Gnecchi-Ruscone and colleagues sequenced 66 genomes in a paper in Cell. The earliest elite of the Danube to Tisza region could be modelled with 88 to 98% of their ancestry from an Ancient Northeast Asian gene pool, best represented by Xianbei and Rouran period individuals from Mongolia and Inner Mongolia. The authors linked this to the fall of the Rouran and estimated that the Avars covered more than 5,000 km from Mongolia to the Caucasus in a few years, and settled in Hungary about ten years later: the fastest long distance migration that ancient DNA has yet reconstructed.
The same year, Maróti and colleagues published genomes from all three nomadic periods of the Carpathian Basin, Huns, Avars and conquering Hungarians, including a set of Early Avar elite burials from the Danube to Tisza interfluve. Those are the six genomes we reanalyse here.
Six graves
The six individuals labelled Early Avar elite in the dataset come from five sites in central Hungary. Archaeological descriptions and dates below are from the supplementary information of Maróti et al. 2022. The eastern shares are from our G25 models, described in the methods section.
| Sample | Site and grave | Sex | Date (archaeological) | Grave goods | Eastern share (Rouran / Xianbei pole) |
|---|---|---|---|---|---|
| KFP-7 | Kunpeszér, Felsöpeszéri út, grave 7 | Female | 630 to 660 | Gilded bronze Mezöszilas type pendants, beads, iron awl, knife | 90% / 100% |
| KV-3369 | Kiskörös, Vágóhídi dülö, grave I | Male | 660 to 700 | Headgear with gold mounts, belt with glass inlaid mounts | 87% / 100% |
| CS-465 | Budapest, Csepel Kavicsbánya, grave 465 | Male, 45 to 50 | 630 to 660 | Sword with gold plated suspension loop, bronze buckle, arrowheads | 87% / 100% |
| FGD-4 | Fajsz, Garadomb, grave 4 | Male | c. 630 | Silver earrings, pressed silver belt mounts, bow, quiver | 86% / 99% |
| KFP-6 | Kunpeszér, Felsöpeszéri út, grave 6 | Male | 630 to 660 | Silver and gilded bronze belt mounts, bow, quiver | 44% / 51% |
| DK-701 | Dunavecse, Kovacsos dülö, grave 701 | Male | 630 to 660 | Sword, bow, quiver, stirrups and bit, silver belt mounts, Byzantine jug | 11% / 12% |
Two things stand out at once. Four of the six, including the only woman, are almost entirely eastern. And the two other men, buried with the same kind of prestige objects, are not: one is half eastern, the other is overwhelmingly European.
The woman in grave 7
KFP-7 sits at the eastern end of every model we ran. With the Rouran genome as the eastern pole she is 90% eastern, with the Xianbei genome 100%. Her nearest ancient populations in the full Global25 reference set are all in Mongolia and the eastern steppe: Asian individuals from the Golden Horde period in Kazakhstan (distance 0.024), Late Xiongnu from Buryatia (0.029) and Mongol period individuals from Mongolia (0.029). Nothing in Europe comes close.
That matters because women are the best test of a migration. Warrior elites can be recruited, adopted, or joined by ambitious locals. A woman who is genetically indistinguishable from the eastern steppe, buried in a Hungarian cemetery within a few decades of the conquest, is evidence that the Avars did not arrive as a male war band who took local wives. They came as families. This fits what Csáky and colleagues inferred from the Y chromosome: the elite arrived as a group of related families, women included.
Her neighbour in grave 6, a man buried with silver belt mounts, a bow and a quiver, is a different story. With 44 to 51% eastern ancestry, he is roughly the child of one steppe parent and one European parent, or the grandchild of such a union. The genetic distance between the two individuals of the same cemetery, 0.288, is about half the distance between Mongolia and central Europe in this space. Kunpeszér was not a closed community. Mixing had begun within the very first generations.
The man with the Byzantine jug
DK-701 from Dunavecse had one of the most complete prestige assemblages of the group: a sword, a bow and quiver, stirrups and a bit, silver belt mounts, and a Byzantine jug. Genetically he is 11 to 12% eastern, and his closest matches in the reference set are medieval individuals from western Russia with a Uralic profile, not Avars at all.
This is the second lesson of the six graves. Avar elite status was a set of objects and practices: the belt, the horse gear, the weapons. It could be held by people of very different origins. The ruling families were eastern. Their followers and allies did not have to be.
Elite and everyone else
The lower half of Figure 1 puts the six elite genomes in context. Across all Avar periods, most individuals carry a modest eastern share: the median is 35% for other Early Avar period individuals, 28% for the Middle Avar period and 29% for the Late Avar period (Rouran pole). Only 8 of the 59 other Early Avar period genomes reach 80%, the level of the four eastern elite individuals.
The contrast with the Huns is sharp. The 21 Hun period individuals from Hungary have a median of zero eastern ancestry in the same model, with only a few individuals carrying any at all. The Huns left almost no genetic trace of the east in their subjects. The Avars brought a large eastern population and kept a substantial eastern share for more than two centuries, which is also why the 10th century Hungarian conquerors, at a median of about 27%, look so similar to Late Avars in this single measure, despite a partly different eastern source.
Rouran or Xianbei?
Our two eastern poles give different numbers for the same people, and that is worth explaining rather than hiding. The Rouran pole is a single genome, and in Global25 it sits further east than the Avar elite. To reach the Avar position, NNLS has to add 10 to 14% of a western source, even for KFP-7. The Xianbei pole sits slightly closer to the Avars, so the same individuals fit at 99 to 100% eastern, with no western source needed.
The truth for these four individuals is somewhere in that range, and the published qpAdm figure of 88 to 98% sits exactly inside it. We use the qpAdm range as the headline, and the G25 numbers as directional confirmation. The fit is also slightly better with the Xianbei pole (distance 0.045 to 0.072) than with the Rouran pole (0.063 to 0.083). This does not show that the Avars were not Rouran: one low resolution Rouran genome cannot represent a whole empire, and Gnecchi-Ruscone and colleagues found that the same Rouran individual fits several Avar elite genomes as a single source in qpAdm. It does show that G25 cannot separate Xianbei and Rouran period Mongolians cleanly, as both belong to the same Ancient Northeast Asian gene pool.
Is a zero really a zero?
In the Xianbei model, KFP-7 receives 0% western ancestry. Before reporting that, we tested how sensitive the model is. We mixed known fractions of a Lombard period Hungarian average into her genome and re-ran the model. A 2% spike is not detected at all, a 5% spike returns 1.6%, an 8% spike returns 4.7%, and a 10% spike returns 6.8%. For CS-465, KV-3369 and FGD-4 the model recovers spikes almost exactly from 2% upwards.
So the honest reading for KFP-7 is not "zero European ancestry" but "less than about 4 to 5%". The Xianbei individual used as a pole already carries a small western share of its own, and the model cannot see below that level. Our models also cannot tell whether the small western share in the Rouran model comes from the North Caucasus (Alans) or from local Carpathian people: the two western poles are only 0.141 apart, far too close to split reliably, so we only report the total.
What the six graves tell us
The first Avars of the Hungarian plain were a population from the Mongolian steppe, almost unmixed, and they arrived with their women. Grave 7 at Kunpeszér holds one of them, and she is as eastern as any warrior in the sample. Within a generation, men of mixed and even mostly European ancestry were buried with the same belts, swords and horse gear. The genetic origin of the Avars was in Mongolia; their power in Europe was built by recruiting everyone else.
Theophylact's Pseudo-Avars may still have existed somewhere in the confederation. But the core that crossed the Carpathians in 568 was not borrowing a name. It had travelled from the far side of Asia, faster than any population we can follow in ancient DNA.
Methods
We used Global25 scaled coordinates by Davidski for the six Early Avar elite genomes and for reference populations, and the Moriopoulos 2026 collection for the Rouran period genome from Mongolia. Each target was modelled by non negative least squares (NNLS) with a sum to one constraint as a mixture of one eastern pole (the Rouran period genome or an Iron Age Xianbei genome from Mongolia), an Alan average from the North Caucasus, and a Lombard period average from Hungary. Condition numbers of the source matrices were 4.7 (Xianbei) and 5.4 (Rouran); the eastern poles are 0.52 to 0.63 from the western poles, while the two western poles are only 0.141 apart, which is why we report the western share as a total. Group distributions in the lower panel of Figure 1 include all individuals in the dated ancient file with at least 30% coverage. Spike in tests add a known fraction of the Lombard period average to the target and re-fit to establish the detection floor. G25 results are a proxy; we report the published qpAdm range as the headline and our numbers as directional confirmation. Grave goods, sex and archaeological dates are from Maróti et al. 2022; the dates encoded in the sample labels of the Global25 file differ for some individuals, and we follow the archaeological publication.
Global25 coordinates used in this article
The six Early Avar elite genomes, the Rouran period genome and the three model poles. They can be pasted directly into the Vahaduo tool.
Hungary_EarlyAvar_Elite:KFP-7.SG,0.034147,-0.399103,0.095034,-0.028747,-0.081554,-0.054384,0.022326,0.025845,0.013908,0.019135,-0.025008,0,-0.002379,0.015551,0.009093,-0.003978,0.000913,-0.005321,0.00088,0.018384,-0.022585,-0.008903,-0.018487,0.000482,-0.002036
Hungary_EarlyAvar_Elite:KFP-6.SG,0.085367,-0.132019,0.077687,0.017119,-0.020619,-0.013108,0.013866,0.022153,0.008181,0.007472,-0.013316,0.003747,-0.009068,0.002477,0.001086,-0.001326,-0.009909,0.002154,0.004148,0.006003,-0.010606,-0.000742,-0.014666,-0.001566,0.000718
Hungary_EarlyAvar_Elite:KV-3369.SG,0.043253,-0.380824,0.098051,-0.028424,-0.092325,-0.051037,0.025146,0.024691,0.006136,0.016219,0.002598,0.003897,-0.005946,0.002615,0.003664,0.004243,0.001434,0.004814,0.005154,0.011255,-0.018343,-0.008408,-0.010476,-0.005543,0.005868
Hungary_EarlyAvar_Elite:CS-465.SG,0.036423,-0.388948,0.084852,-0.020995,-0.073244,-0.039602,0.016686,0.026537,0.007976,0.014032,-0.016888,0.000899,0.001338,0.006193,0.000679,-0.006099,-0.001825,0.000633,0.005154,0.014882,-0.020838,0.003462,-0.017624,-0.001325,0.001916
Hungary_EarlyAvar_Elite:FGD-4.SG,0.039838,-0.380824,0.079572,-0.01292,-0.078168,-0.041834,0.014336,0.01823,0.001227,0.007107,-0.02241,-0.001049,0.002527,0,0.008822,0.000398,-0.004433,-0.008742,0.002388,0.014632,-0.013102,-0.003957,-0.019103,-0.003976,-0.001078
Hungary_EarlyAvar_Elite:DK-701.SG,0.118376,0.066009,0.069767,0.054264,0.02185,0.013387,0.012456,0.011769,-0.008181,-0.025149,0.000325,0.000749,0.012636,0.017891,-0.008415,0.009414,0.02034,0.003421,0.004399,0.007879,-0.010107,0.000247,-0.007025,0.000964,0.005269
Mongolia_Early_Medieval_Rouran_(n=1),0.043253,-0.459019,0.064865,-0.036822,-0.077245,-0.058288,0.026086,0.004846,-0.014521,0.007654,-0.047742,0.0003,-0.005798,0.020919,-0.000814,0.005171,0.013821,-0.003674,0.019232,0.002751,-0.035562,-0.035117,-0.033154,-0.00735,-0.014609
Mongolia_IA_Xianbei,0.037562,-0.38387,0.075801,-0.023902,-0.058472,-0.039324,0.013631,0.00923,0.000205,0.013668,-0.046281,0.002248,-0.003568,-0.001239,0.00665,0.010077,0.001565,0.003801,0.006034,0.023011,-0.023084,-0.018795,-0.026252,-0.005784,0.002515
Russia_Alan.SG,0.109498,0.104193,-0.024437,-0.008592,-0.028251,0.003904,0.007191,-0.006046,-0.044954,-0.025076,0.003443,0.010551,-0.016234,-0.006496,0.005782,-0.002678,0.005189,-0.005752,-0.005757,0.003752,0.003594,0.004402,0.00175,0.000048,-0.002515
Hungary_Langobard,0.12881,0.13185,0.060276,0.047966,0.037879,0.01506,0.002898,0.006384,0.001551,-0.001336,-0.007578,0.003809,-0.007631,-0.004484,0.017712,0.004386,-0.005357,0.000496,0.006484,0.004398,0.00652,0.005296,-0.00114,0.011889,-0.00001
References
- Ancient DNA Gnecchi-Ruscone GA, Szécsényi-Nagy A, Koncz I, et al. (2022). Ancient genomes reveal origin and rapid trans-Eurasian migration of 7th century Avar elites. Cell 185(8): 1402 to 1413. doi.org/10.1016/j.cell.2022.03.007
- Ancient DNA Maróti Z, Neparáczki E, Schütz O, et al. (2022). The genetic origin of Huns, Avars, and conquering Hungarians. Current Biology 32(13): 2858 to 2870. doi.org/10.1016/j.cub.2022.04.093
- Y-DNA Csáky V, Gerber D, Koncz I, et al. (2020). Genetic insights into the social organisation of the Avar period elite in the 7th century AD Carpathian Basin. Scientific Reports 10: 948. doi.org/10.1038/s41598-019-57378-8
- Kinship Gnecchi-Ruscone GA, Rácz Z, Samu L, et al. (2024). Network of large pedigrees reveals social practices of Avar communities. Nature 629: 376 to 383. doi.org/10.1038/s41586-024-07312-4
- Press Medievalists.net (2022). Genetic study connects the Avars to the fall of the Rouran Khaganate. medievalists.net
- Data Global25 PCA coordinates by Davidski, and the Moriopoulos 2026 collection of Global25 coordinates.