Who Were the Headless Gladiators of Roman York? DNA Finds Six Local Britons and One Man From the Near East

Hero image: AI generated illustration of two beast fighters facing a lion in a timber arena outside Roman York around AD 200, with the walls of the legionary fortress in the distance. York's amphitheatre has never been located, so the arena is a reconstruction based on smaller provincial amphitheatres, and clothing and weapons are approximations. The two men stand for the local and foreign men buried at Driffield Terrace. They do not portray any real individual, and nothing is known about the occupation of the man from the Near East.

In 2004, archaeologists from York Archaeological Trust began digging in the gardens of Driffield Terrace, a residential street on the south-west side of York. Under the lawns lay a Roman cemetery unlike any other in Britain. Nearly everyone buried there was a man between 18 and 45. Many were taller and more robust than the average Roman Briton, many carried healed injuries, and about seven in ten had been decapitated.

Since then, the "headless Romans" of York have been called gladiators, soldiers and victims of an imperial purge. Ancient DNA cannot tell us what these men did for a living. It can tell us where they came from. The answer is less exotic than the legends, with one striking exception.

A cemetery of young, violent men

The cemetery lay beside the Roman road that left Eboracum, Roman York, towards Calcaria, modern Tadcaster. It was used from the late 1st to the 4th century AD, and the men discussed here lived between the 2nd and 4th centuries. When the first results were announced in 2010, researchers had examined 82 skeletons of well-built young men. Many showed old fractures and other healed wounds, and some showed signs consistent with hard combat training.

Decapitation is the defining feature. In the burial studied in detail in 2025, the head had been removed with a single blow from behind, between the second and third neck vertebrae, and then placed back in its normal position in the grave. Across the cemetery, the authors of that study estimate that about 70% of the burials were decapitations.

Gladiators, soldiers or victims of an emperor?

Three explanations have been proposed. The first is a gladiator cemetery: the age and sex profile, the robust skeletons and the many healed injuries resemble those of a suspected gladiator burial ground at Ephesus, in modern Turkey. The second is a military cemetery, since York was a legionary base and Roman army recruitment also selected young, tall men. The third is executions.

That third idea was set out in 2011 by Janet Montgomery, Christopher Knusel and Katie Tucker. They argued that the prominent position of the burials beside a main road, and the controlled, repeated way the heads were removed, point to men of some status who were executed, rather than common criminals or battle casualties. They linked the cemetery to the Severan period, when York briefly became the centre of the Empire. Septimius Severus, born at Leptis Magna in North Africa, campaigned in Scotland and died at York in 211. His wife Julia Domna came from Emesa, modern Homs, in Syria. After his death, his son Caracalla had his brother Geta murdered and purged his supporters.

Isotopes then added a geographical twist. In 2011, Gundula Muldner, Carolyn Chenery and Hella Eckardt analysed 68 bone and dentine samples and 18 tooth enamel samples from the cemetery. The childhood origins of the "headless Romans" turned out to be significantly more diverse than those of people buried in other Roman cemeteries of York. Two men had eaten a significant amount of protein from C4 plants, probably millet, a crop not grown in Britain at the time. These were the first such values reported from any period of British history. In their chapter, Montgomery and colleagues judged that four of the six men they sampled had probably or possibly grown up outside Britain, with suggested origins ranging from the Alps or central Europe to warm regions around the Mediterranean.

The lion bite

In April 2025, a team led by Tim Thompson of Maynooth University published the analysis of one more skeleton from the site in PLOS ONE. The man, numbered 6DT19, was aged 26 to 35 and was probably buried in the 3rd century AD. Both sides of his pelvis carry pits, depressions and crushed areas. The team compared them with bite marks made by zoo animals and concluded that they were made by a large cat, most likely a lion.

The marks did not heal, so they were made around the time of death. Their position on the pelvis suggests a big cat feeding on a man who was dying or already dead, rather than the first strike of an attack. Afterwards he was decapitated, like most of his neighbours, and buried with butchered and gnawed horse bone in the grave. The authors describe it as the first physical evidence for human and animal gladiatorial combat from the Roman period anywhere in Europe.

The find makes a strong case that at least one man in the cemetery died in an arena, probably as a bestiarius, a beast fighter. It does not prove that all of them were gladiators. York's amphitheatre has still not been located, and the lion-bitten man is not one of the men whose genomes were sequenced.

Seven genomes: six Britons and a stranger

In 2016, Rui Martiniano, Daniel Bradley and colleagues from Trinity College Dublin, the University of York, Durham, Reading and other institutions published the genomes of seven men from the cemetery in Nature Communications. DNA was extracted from the dense petrous bone of the inner ear and sequenced at about 1x coverage. For comparison, the team also sequenced an Iron Age woman from Melton, in East Yorkshire, dated between 210 BC and AD 40, and an Anglo-Saxon man from Norton-on-Tees.

Six of the seven men were genetically similar to the Iron Age woman from Melton. Among present-day populations, they were most similar to the Welsh, followed by the Irish and the Scottish, with the modern English ranked only ninth. They also differed significantly from people living in Yorkshire today, while the Anglo-Saxon man sat closest to East Anglians, between the English and the Dutch. The authors read this as a trace of the later migrations that reshaped eastern England after the Roman period.

All six local men carried Y-chromosome lineages of haplogroup R1b, the most common lineage in western Europe: two belonged to R1b-U106, one to R1b-U152, one to R1b-DF63, a branch of R1b-L21, and two could only be placed under R1b-L52. Their predicted appearance was mostly brown eyes and dark hair. One of them, 6DRIF-18, probably had blue eyes and blond hair.

The seventh man, 3DRIF-26, was different. In a principal component analysis his nearest neighbours were present-day Palestinians, Jordanians and Syrians, and the modern genomes with which he shared the most alleles were Saudi Arabian. His Y chromosome belongs to haplogroup J2, his mitochondrial DNA to H5, and he probably could not digest milk as an adult. His teeth tell the same story: an unusually low strontium ratio and an extreme oxygen value, which the authors attribute to a childhood in an arid environment. Yet his burial and his skeleton were indistinguishable from those of the men around him.

Our test: what Global25 sees in the seven men

All seven genomes are in the public Global25 datasets, with coverage between 48 and 82% in the dated file. We checked that the scales are compatible: the average of the six local men and the published Global25 average for the same group are only 4 units apart (distances multiplied by 1,000). To judge what counts as close, we measured how far single genomes of similar quality lie from the average of their own population, using 223 Bronze and Iron Age genomes from England. The median is about 26 units. That is the noise built into any single ancient genome.

We then modelled each man with four sources: Iron Age East Yorkshire, an average of genomes from the region before the Roman conquest; Roman-era Lebanon; Egypt in the early first millennium BC, from Abusir el-Meleq; and present-day Yemenis from Amran, since no ancient genome from Arabia is available. The sources are well separated (50 to 243 units apart) and the condition number is 11.0.

The six local men come out at 98 to 100% Iron Age East Yorkshire. The Near Eastern share never exceeds 2%, below the detection floor: when we built synthetic local genomes with no Near Eastern ancestry and added real single-genome noise, the model returned up to 6% in 95% of 500 runs. These men descend from the people who lived in the region before Rome arrived.

3DRIF-26 comes out at 1.4% British and 98.6% Near Eastern, with a fit of 20 units, as good as any single genome can be expected to reach. The Near Eastern part is split 25% Levant, 52% Egypt and 22% Arabia, but that split is not reliable, as explained in Trap 2 below.

The man from the south

Distances to present-day populations show where he fits best. The closest averages are Egyptian Copts (40 units), Yemenis from Amran (43), Negev Bedouin (45), Palestinians (47), Saudis from Mecca (49) and Jordanians (50). Lebanese Christians (62) and Syrians (71) are further away, Iraqi Arabs and Cypriots further still (86), and modern Italians from Lazio, the region of Rome, are at 131. The English average is 225 units away.

Ancient genomes give the same direction. The closest single ancient genome in the dataset, at 27 units, is a man buried at Marsala in western Sicily around the 1st century BC, who carried Near Eastern ancestry: another migrant within the Roman world. Late Period Egyptians from Abusir el-Meleq follow at about 32 units. Roman-era genomes from Lebanon sit at 54 to 57 units, and their average fits him poorly on its own (57 units, against 20 with three Near Eastern sources).

This agrees with the 2016 study on Palestine, Jordan and Arabia, and is less favourable to Syria. In Roman terms, our test points to the southern half of the Roman Near East: Judaea, renamed Syria Palaestina in the 2nd century, the province of Arabia Petraea created in 106 in what is now Jordan, the Negev and north-west Arabia, or Egypt. Global25 cannot choose between them.

A first-generation migrant

Could he have been born in Britain to a Near Eastern father and a local mother? We built synthetic genomes from the Near Eastern sources with known amounts of East Yorkshire Iron Age ancestry, added real noise, and ran the same model 500 times for each scenario.

A true 0% British share came back at no more than 6%. One British great-grandparent (12.5%) returned 6 to 19%, one British grandparent (25%) returned 19 to 31%, and one British parent (50%) returned 44 to 56%. His 1.4% rules out a British parent and makes a British grandparent very unlikely. Combined with the isotopes, which record where he lived as a child, the simplest reading is that he was born and raised in the southern Near East and came to Britain later in life.

One stranger among many locals: Roman Britain in 31 genomes

Was he typical? The published evidence says no. In 2024, Christiana Scheib, Toomas Kivisild and colleagues reported genomes from 52 people buried at eight rural sites in Cambridgeshire during the Roman occupation. They found little genetic differentiation between sites, no evidence of long-distance migrants from elsewhere in the Empire, and only one possible case of temporary mobility within a family. Their conclusion was that the genetic patterns of present-day Britain formed after the Roman period.

The other known exception was published the same year by Marina Silva, Thomas Booth, Pontus Skoglund and colleagues. A man buried alone in a ditch at Offord Cluny, Cambridgeshire, between AD 126 and 228, carried ancestry related to the Caucasus and to Sarmatian groups of the steppe. His isotopes show that he spent his early childhood in an arid part of eastern continental Europe, eating millet, before moving west. In 175, Marcus Aurelius sent 5,500 Sarmatian cavalrymen to Britain, but nothing in his grave links him to the army.

We gathered all Roman-era genomes from Britain in the public Global25 files dated between AD 43 and 410 with coverage of at least 40%: 31 in total, from York, Cambridgeshire, Dorset, Gloucestershire and other English sites. Each was modelled with a British Iron Age source and five distant ones: Imperial Rome, Roman-era Lebanon, Egypt, Roman-era Algeria and the Sarmatians (condition number 30.5).

Twenty-nine genomes fall inside the detection floor, between 0 and about 15% distant ancestry. Two stand alone at 100%: 3DRIF-26 and the Offord Cluny man. In Global25, the Offord Cluny man's closest ancient matches are Bronze and Iron Age genomes from Armenia, 24 to 28 units away, which fits the Caucasus-related ancestry reported by Silva and colleagues.

This does not mean that Roman Britain had no migrants. The sample is small and mostly rural, and the floor hides anyone with less than about 15% distant ancestry: the grandchildren of a migrant would be invisible here. Isotope studies of Roman York, including the Driffield Terrace cemetery itself, point to more people who grew up elsewhere. But it shows how rare a genome like that of 3DRIF-26 still is.

Three traps in the data

Trap 1: Welsh, Scottish or English? Global25 cannot tell

The 2016 study found the six local men closest to the modern Welsh. In Global25, the average of the six is 11.6 units from Orcadians, 12.7 from the Scottish, 13.7 from the English, 14.9 from the Irish and 15.2 from the Welsh, but also 16.2 from the Dutch and 17.7 from the Danes.

For an average of six genomes of this quality, the expected noise is about 11 units, so all of these values are a statistical tie. Global25 confirms that the men belong to north-western Europe and to Britain's Iron Age population. It cannot reproduce the ranking of modern British populations, which the 2016 study obtained from the full sequence data rather than from a 25-dimension summary. Here the published result has to be the headline.

Trap 2: the three Near Eastern sources overlap

Levant, Egypt and Arabia are only 50 to 76 units apart in Global25, much less than the distance that separates any of them from Britain. Dropping one source at a time moves the split a lot. Without the Lebanese source, 3DRIF-26 becomes 61% Egypt and 33% Arabia. Without the Egyptian source, he becomes 36% Levant and 63% Arabia, and the fit worsens from 20 to 32 units. Without the Arabian source, he becomes 30% Levant and 69% Egypt.

Two things do not move. His British share stays below 6% in every version, and the southern sources always outweigh the Roman-era Lebanese one. The "Egypt" label should not be taken literally either: the Abusir el-Meleq genomes are themselves close to Levantine populations, and here they act as a marker for the southern side of the Near Eastern cline. Nor is the southern shift a simple matter of African admixture: adding an East African source to the Lebanese one gives fits of 40 to 49 units, twice as poor as the three-source model.

Trap 3: modern populations carry later history

Egyptian Copts come first in the distance ranking, but that does not make him an Egyptian. Every modern Near Eastern population has received gene flow in the last 1,800 years, from the Arab expansion to the trade routes of the Red Sea and the Sahara, and no Roman-era genome from Palestine, Jordan or Arabia is available in the dataset for a direct comparison. The best ancient match is a man buried in Sicily, himself a migrant. The region is clear. The town is not.

What this means

Most of the men buried at Driffield Terrace were not exotic fighters shipped in from the edges of the Empire. Six of the seven sequenced so far descend from the Iron Age people of the region, the same population that lived in East Yorkshire before the Roman conquest. Whatever brought them to a violent death, they were local.

The seventh man is the exception that makes the cemetery famous. He grew up in the southern Near East, nearly 4,000 km away, had no British parent, and was buried in exactly the same way as the local men around him. The lion bite on another skeleton shows that animal fights were staged at York, and that at least one man in this cemetery was caught up in them. What the men did for a living remains open: DNA records origins, not occupations. York Archaeology has said that further genome work on the cemetery is under way, and more genomes may settle how many of the men grew up far from Britain.

Global25 coordinates used in this article

The model sources, the seven men from Driffield Terrace and the Offord Cluny man. They can be pasted directly into the Vahaduo tool. Labels for the seven men follow the 2016 study; in the original files they appear as England_IA_Roman.SG and England_IA_Roman_oMiddleEast.SG.

Source_England_EastYorkshire_MIA_LIA,0.129921,0.132671,0.060461,0.049281,0.038579,0.015967,0.002367,0.004038,0.005157,0.005578,-0.004419,0.006765,-0.014940,-0.016593,0.022234,0.008675,-0.003763,0.001507,0.001118,0.002528,0.006595,0.004301,-0.002386,0.006490,0.001116
Source_Lebanon_ERoman,0.083850,0.146575,-0.055060,-0.092055,-0.016516,-0.030492,-0.002350,-0.007384,0.006817,0.009355,0.010555,-0.008942,0.015560,-0.001514,-0.011446,0.015204,0.010909,0.003125,-0.000461,0.001959,0.001165,0.004616,-0.006450,0.006025,-0.005069
Source_Egypt_ThirdIntermediatePeriod,0.050651,0.146744,-0.041672,-0.118704,-0.001846,-0.049085,-0.012573,-0.004154,0.048063,0.006196,0.008932,-0.014387,0.028989,-0.003578,-0.002375,-0.004973,-0.007888,0.000760,-0.006347,0.014257,0.005678,0.003462,0.004437,0.001326,0.000420
Source_Yemenite_Amran,0.045855,0.135428,-0.061525,-0.101445,-0.014860,-0.042152,-0.008729,-0.007763,0.045010,-0.004256,0.011750,-0.022962,0.046510,0.005259,0.003994,0.017256,-0.014137,0.004253,0.001832,0.015704,0.006925,0.009521,-0.006330,0.004174,-0.004781
Source_England_MIA_LIA,0.129779,0.136413,0.061080,0.046330,0.040824,0.015618,0.002547,0.004296,0.007408,0.007442,-0.004195,0.006657,-0.014601,-0.013960,0.021424,0.006352,-0.008610,0.002043,0.001879,0.003700,0.005347,0.003303,-0.002178,0.005328,-0.001293
Source_Italy_Imperial,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
Source_Algeria_NumidoRoman_Berber,0.091628,0.150298,-0.024324,-0.052165,0.008155,-0.021335,0.003407,-0.000347,0.006954,0.021048,0.002273,0.003822,-0.003940,0.000138,-0.002375,-0.002851,0.005411,-0.002154,-0.000063,-0.005565,-0.005116,-0.005626,0.002280,0.001145,0.004431
Source_Russia_Sarmatian,0.109649,0.048915,0.034004,0.065838,-0.028056,0.026727,0.000392,0.000230,-0.030542,-0.036568,-0.004249,-0.003372,-0.002626,-0.024956,0.019544,0.013237,-0.002108,-0.001921,-0.003980,0.000333,-0.011064,0.004183,0.000657,0.007290,0.003772
England_York_Roman_3DRIF-26,0.054635,0.153345,-0.051288,-0.100453,-0.005539,-0.045459,-0.013866,-0.009692,0.034360,0.002916,0.004547,-0.019183,0.026610,0.000275,-0.002850,0.004110,-0.004955,-0.002534,-0.006034,0.011130,0.004617,-0.001484,-0.001232,0.005302,-0.008742
England_York_Roman_3DRIF-16,0.127482,0.142174,0.063356,0.054910,0.039700,0.020638,0.001175,0.000923,0.006749,0.009294,0.000974,0.002847,-0.011150,-0.007707,0.013165,0.001458,-0.005867,0.003421,0.003017,-0.004877,0.004367,0.003091,-0.002711,0.010001,-0.001437
England_York_Roman_6DRIF-3,0.129758,0.136081,0.061471,0.051357,0.036007,0.025658,0.005875,0.002769,-0.001227,-0.005103,-0.006496,0.008692,-0.019772,-0.001927,0.030130,-0.011933,-0.023339,0.002407,0.000126,0.005753,0.006364,0.001607,-0.002958,0.012170,0.001796
England_York_Roman_6DRIF-18,0.126344,0.142174,0.059585,0.055233,0.046162,0.020359,0.004935,0.011999,-0.003477,0.005103,0.000812,0.011989,-0.017691,-0.018717,0.022394,0.004110,-0.006389,-0.003421,0.010684,0.000875,0.009358,0.001360,-0.003821,0.012050,0.000958
England_York_Roman_6DRIF-21,0.127482,0.127957,0.060716,0.047481,0.025543,0.014781,0.004700,0.010846,-0.001841,-0.000547,-0.017051,0.014387,-0.009514,-0.008808,0.020629,0.003447,-0.008605,-0.003041,0.003771,0.005628,0.000998,0.000989,0.002095,0.004097,-0.001317
England_York_Roman_6DRIF-22,0.124067,0.139128,0.061471,0.047481,0.043393,0.016733,0.003995,0.003000,0.003272,0.001640,-0.004384,0.004346,-0.010704,-0.018579,0.015336,0.001724,0.000522,0.002280,0.005028,0.002001,0.006239,0.006801,-0.013188,0.002530,0.003113
England_York_Roman_6DRIF-23,0.127482,0.140143,0.059208,0.046512,0.037545,0.011992,0.008225,0.007384,0.001227,0.001093,-0.009419,0.009442,-0.015609,-0.006881,0.032844,-0.001591,-0.018645,0.005828,0.000000,0.008379,0.002745,-0.004946,-0.011462,0.006989,0.002395
England_OffordCluny_Roman_Sarmatian_C10271,0.094473,0.129988,-0.046386,-0.023579,-0.029852,-0.008367,0.006345,-0.006692,-0.051131,-0.022962,-0.000650,0.009292,-0.013825,0.000413,0.009636,-0.005967,0.001565,0.002027,0.001257,0.013256,0.004118,-0.004451,0.002835,-0.006386,-0.003233

References

1. Ancient DNA Martiniano R, Caffell A, Holst M, Hunter-Mann K, Montgomery J, Muldner G, McLaughlin RL, Teasdale MD, et al., Collins MJ, Bradley DG (2016). Genomic signals of migration and continuity in Britain before the Anglo-Saxons. Nature Communications 7: 10326. doi.org/10.1038/ncomms10326
2. Osteology Thompson TJU, Errickson D, McDonnell C, Holst M, Caffell A, Pearce J, Gowland RL (2025). Unique osteological evidence for human-animal gladiatorial combat in Roman Britain. PLOS ONE 20(4): e0319847. doi.org/10.1371/journal.pone.0319847
3. Isotopes Muldner G, Chenery C, Eckardt H (2011). The 'Headless Romans': multi-isotope investigations of an unusual burial ground from Roman Britain. Journal of Archaeological Science 38(2): 280 to 290. doi.org/10.1016/j.jas.2010.09.003
4. Isotopes Montgomery J, Knusel CJ, Tucker K (2011). Identifying the origins of decapitated male skeletons from 3 Driffield Terrace, York, through isotope analysis: reflections of the cosmopolitan nature of Roman York in the time of Caracalla. In Bonogofsky M (ed.), The Bioarchaeology of the Human Head. University Press of Florida, 141 to 178. doi.org/10.5744/florida/9780813035567.003.0006
5. Ancient DNA Silva M, Booth T, Moore J, Anastasiadou K, Walker D, Gilardet A, et al., Montgomery J, Skoglund P (2024). An individual with Sarmatian-related ancestry in Roman Britain. Current Biology 34: 202 to 212. doi.org/10.1016/j.cub.2023.11.049
6. Ancient DNA Scheib CL, Hui R, Rose AK, D'Atanasio E, Inskip SA, et al., Kivisild T (2024). Low genetic impact of the Roman occupation of Britain in rural communities. Molecular Biology and Evolution 41(9): msae168. doi.org/10.1093/molbev/msae168
7. Ancient DNA Patterson N, Isakov M, Booth T, Buster L, Fischer CE, et al., Reich D (2022). Large-scale migration into Britain during the Middle to Late Bronze Age. Nature 601: 588 to 594. doi.org/10.1038/s41586-021-04287-4
8. Ancient DNA Cassidy LM, Russell M, Smith M, et al., Bradley DG (2025). Continental influx and pervasive matrilocality in Iron Age Britain. Nature 637. doi.org/10.1038/s41586-024-08409-6
9. Ancient DNA Schuenemann VJ, Peltzer A, Welte B, et al., Krause J (2017). Ancient Egyptian mummy genomes suggest an increase of Sub-Saharan African ancestry in post-Roman periods. Nature Communications 8: 15694. doi.org/10.1038/ncomms15694
10. Ancient DNA Haber M, Nassar J, Almarri MA, Saupe T, Saag L, Griffith SJ, et al., Xue Y, Scheib CL, Tyler-Smith C (2020). A genetic history of the Near East from an aDNA time course sampling eight points in the past 4,000 years. American Journal of Human Genetics 107(1): 149 to 157. doi.org/10.1016/j.ajhg.2020.05.008
11. Press release University of York (2025). Study reveals skeletal evidence of Roman gladiator bitten by lion in combat. york.ac.uk
12. Press release Trinity College Dublin (2016). Scientists sequence ancient British 'gladiator' genomes from Roman York. tcd.ie
13. Data Global25 PCA coordinates by Davidski, and the Moriopoulos 2026 collection of Global25 coordinates.