Ancient DNA | Old Kingdom Egypt | Mystery file
The first complete ancient Egyptian genome
A potter buried in a clay pot at the dawn of the pyramid age. Four fifths North African, one fifth from the far side of the Fertile Crescent, and a result that both camps of the internet's longest-running argument about ancient Egypt have been misreading since the day it came out.
Hero image: AI generated illustration of a rock-cut tomb with a burial jar above the Nile valley. It is not the actual Nuwayrat tomb or pot.
In 1902 the Liverpool archaeologist John Garstang opened a rock-cut tomb in the cliffs at Nuwayrat, a village about 265 kilometres south of Cairo, near Beni Hasan. Inside was a large ceramic pot, and inside the pot, folded, was a man. He was not mummified. He had simply been sealed in, and the pot and the dry tomb did what embalmers could not: they kept his teeth well enough for his DNA to survive. The skeleton went to Liverpool, sat in a museum store through the bombing of the Second World War, and in July 2025 became the first ancient Egyptian ever to have his whole genome sequenced.
The paper, published in Nature by Adeline Morez Jacobs, Linus Girdland-Flink and colleagues, has a quiet title (Whole-genome ancestry of an Old Kingdom Egyptian) and an explosive subject. Few questions on the internet generate as much heat as "what were the ancient Egyptians?", and here, for the first time, was a complete genome from the age of the first pyramids. Within hours it was being quoted as proof that the Egyptians were Black, as proof that they were not, and as proof that modern Egyptians are not their descendants. In this article we go through what the study actually found, then rerun the key models in Global25 with every caveat on the table.
Why it took forty years
Egypt should be the promised land of ancient DNA: thousands of mummies, precise chronologies, a written history. In practice it has been a graveyard for genetic data. Heat destroys DNA, and embalming chemistry does not help. The first claim of mummy DNA, in 1985, is now thought to have been largely modern contamination. In 2017 Verena Schuenemann and colleagues sampled 151 mummies from Abusir el-Meleq in Middle Egypt and obtained mitochondrial genomes from 90, but genome-wide nuclear data from only three, all from the first millennium BC or later.
So until 2025 there was no nuclear genome at all from the Early Dynastic period or the Old Kingdom, the age of Djoser, Khufu and the pyramids of Giza. Pontus Skoglund, senior author of the new study, put it simply: forty years after the first pioneering attempts, nobody had managed it. The Nuwayrat man was dated to 2855 to 2570 BC, straddling the end of the Early Dynastic period and the start of the Old Kingdom. His DNA was extracted from the cementum of a tooth and shotgun sequenced to an average depth of 2.02x.
A potter in a rich man's grave
Joel Irish and his team read the skeleton as closely as the genome. The man was 44 to 64 years old at death, stood about 157 to 160 centimetres tall, and had heavily worn teeth and age-related osteoarthritis. His bones record a working life: enlarged areas on the pelvis consistent with long hours seated on a hard surface, marked arm musculature, severe arthritis in the right foot and strain in the neck. The pattern matches the posture of potters in Egyptian tomb scenes, sitting low, turning a wheel, looking down at the clay.
That is a paradox, because the burial is not a labourer's burial. A pot inside a rock-cut cliff tomb was a form of treatment normally reserved for people of higher social class. Irish's own reading is that the man may have risen through exceptional skill. Isotopes in his teeth are consistent with a childhood in the hot, dry Nile Valley: he grew up in Egypt. The Face Lab at Liverpool John Moores University later produced a facial reconstruction from the skull.
The genome belongs to one man, from one place in Middle Egypt, who may not have been typical of the population. The authors say so in their own paper, and every conclusion below inherits that limit.
What the paper found
The core result is a two-source ancestry model fitted with qpAdm. No single ancient population fitted the Nuwayrat genome on its own. The best fit (P = 0.12) was a mix of 77.6 plus or minus 3.8 per cent ancestry related to the Middle Neolithic people of Skhirat-Rouazi in Morocco (about 4780 to 4230 BC) and 22.4 plus or minus 3.8 per cent related to Neolithic Mesopotamia (9000 to 8000 BC), a cluster that includes early farmers from Nemrik 9 in northern Iraq. An f4 test pointed the same way: the extra West Asian affinity was maximised with Neolithic Mesopotamia (Z = 3.2).
Three things about that result matter for everything that follows.
- The labels are proxies, not origins. Nobody is claiming the man had Moroccan ancestors. There are no Neolithic genomes from Egypt itself, so the closest available North African farmers are used as a stand-in for the local population of the Nile Valley.
- The Levant was tested and it lost. A model that used a Bronze Age Levantine population as the West Asian source was rejected (P = 0.013), and the two-source model failed again when Bronze Age Levantine groups were added to the reference set (P = 0.021). Three-source models allowed only a small Levantine share of 1.1 to 4.7 per cent. The West Asian ancestry in this man leans east, toward Mesopotamia and the eastern Fertile Crescent.
- No substantial sub-Saharan component. Both qpAdm and ADMIXTURE excluded any substantial ancestry related to the 4,500-year-old Mota genome from Ethiopia or to populations from central, eastern or southern Africa.
The uniparental markers fit the same picture. His Y chromosome falls in haplogroup E1b1b1b2b, part of the large E-M35 family that is common today in North Africa, the Horn of Africa and the Levant. His mitochondrial lineage is reported as I/N1a1b2, within the West Eurasian macrohaplogroup N. The team also compared him with the three Abusir el-Meleq genomes dated between 787 and 23 BC and found those later Egyptians carried much more ancestry of the Bronze Age Levant type, about 64.5 plus or minus 5.6 per cent in the best model.
Finally, the paper ran a pigmentation prediction. It gave brown eyes, brown to black hair, and skin "ranging from dark to black", with a lower probability of intermediate skin colour. That single sentence is where most of the online fighting started.
Our analysis: first, measure the noise
The Nuwayrat genome is in the Global25 datasets in two versions produced from the same sequencing data, labelled "merged" (87.7 per cent of SNPs covered) and "ss" (70.7 per cent). The Moriopoulos 2026 collection carries the merged version under the spelling "Neurat". The distance between the two versions of the same man is 0.021. That is our noise yardstick: any difference smaller than about 0.02 is within what the data can resolve for this sample and should not be interpreted.
With that in hand we refitted the paper's model with NNLS, using the Skhirat-Rouazi Middle Neolithic average and the Nemrik 9 Neolithic sample. The two sources are well separated (distance 0.220) and the model is well conditioned (condition number 1.8).
Then we asked the same question as the authors: which West Asian population does the job best? We kept Morocco MN fixed as the African side and swapped in the main Neolithic populations of West Asia one at a time.
The trap: why Global25 cannot settle Mesopotamia against the Levant
Figure 2 contains an uncomfortable result that we are not going to hide. When the West Asian source is a Bronze Age Levantine population, for example Early Bronze Age 'Ain Ghazal in Jordan (fit 0.043) or Middle Bronze Age Sidon (0.047), Global25 fits the Nuwayrat man about as well as, or slightly better than, the Mesopotamian model (0.050). The difference is smaller than our noise yardstick. The paper's qpAdm, which works on hundreds of thousands of SNPs and tests the fit formally against a set of outgroups, rejected the Bronze Age Levant models. Global25 cannot reject them.
The reason is collinearity. Bronze Age Levantines were not a pure population: they formed from Neolithic Levantine farmers plus a large inflow from the east. In our own decomposition, Middle Bronze Age Sidon comes out as about 48 per cent Nemrik 9-like, 39 per cent Levant PPNB, and small Anatolian and Iranian shares. A Bronze Age Levantine source therefore already contains Mesopotamian-type ancestry, and a 25-dimensional summary of the genome cannot tell "20 per cent Mesopotamian" apart from "35 to 40 per cent of a population that is half Mesopotamian". When we open a seven-source model with everything at once, the condition number jumps to 24.5 and the weights scatter meaninglessly across the Levant, Morocco and Iran. That is the textbook signature of a model that has run out of resolution.
Our verdict: Global25 independently confirms the size of the West Asian share (about a quarter) and that it fits Mesopotamian Neolithic sources better than Levantine, Iranian or Anatolian Neolithic ones. Whether it came through a Bronze Age Levantine intermediary is a question only the formal tests in the paper can answer, and they said no.
The sub-Saharan question, with a floor
The most shared claim about this genome is that it has "zero sub-Saharan ancestry". The most shared counterclaim is that it predicts dark to black skin, so the man was Black. Before weighing either, we need to know how small a sub-Saharan share our method could actually detect. We added known amounts of the Mota genome (Ethiopia, 4,500 years ago) to Nuwayrat, added random noise of the same size as the difference between the two versions of the genome, and asked the model to find it again, 500 times at each level.
So in the paper's own framework, the answer is at or below 3 per cent, consistent with the authors' statement. Two nuances keep this honest in both directions.
First, the number is model dependent. When the West Asian side is a Bronze Age Levantine proxy rather than Mesopotamia, a Dinka-related share of 3 to 4.5 per cent appears, right at our floor. We read that as the model using a small African component to compensate for a North African proxy that sits 3,000 kilometres to the west. The true local population of the Neolithic Nile Valley, which has never been sequenced, may simply have carried a little more northeast African ancestry than the farmers of Morocco. That is a statement about Egypt's own deep population, not about migration from further south.
Second, "no sub-Saharan component" does not mean "no deep African ancestry". The North African source itself is a mixture. In our decomposition, Skhirat-Rouazi is about 72 per cent Levantine Neolithic-like and 28 per cent Iberomaurusian (the Taforalt foragers of Ice Age Morocco), and van de Loosdrecht and colleagues showed in 2018 that the Taforalt people themselves carried roughly a third of their ancestry from a sub-Saharan African-related source. North Africa's genetic history is its own thing, with ancient African roots, and it does not fit neatly into either box of the modern argument.
Where he falls on the map
On the first component, which separates African-related from West Asian ancestry, the Nuwayrat man is slightly more African-shifted than present-day Copts and close to present-day Egyptian Muslims. On the second, which runs from the Neolithic farmers at the bottom toward Iranian and Caucasian ancestry at the top, he sits low, near the Natufians and the Neolithic populations of the Levant and North Africa. Modern Egyptians sit higher on that axis: they carry more of the later, eastern-shifted West Asian ancestry that arrived after his time.
Four snapshots of Egypt
To follow the population through time, we fitted the same three sources to four targets: the Nuwayrat man, the two Late Period individuals from Abusir el-Meleq, present-day Copts and present-day Egyptian Muslims. The sources are Morocco MN for the Neolithic North African side, Middle Bronze Age Sidon as a proxy for West Asian ancestry of the Bronze Age type, and the Dinka and Yoruba for sub-Saharan ancestry.
The picture is clear even allowing for the collinearity. The Abusir individuals of around 650 BC give 67 to 69 per cent Levantine Bronze Age-like ancestry, matching the paper's independent 64.5 per cent. Between the Old Kingdom and the first millennium BC the population of at least Middle Egypt shifted substantially toward West Asia. The paper cannot date that shift, and neither can we, but the second millennium BC, with the Hyksos period and centuries of New Kingdom contact with the Levant, is the obvious window to test when more genomes arrive.
The present day adds a second layer. Copts fit as the Abusir population plus about 5 per cent sub-Saharan ancestry. Egyptian Muslims carry more, about 12 to 13 per cent, with a worse fit to the ancient source. That matches Schuenemann and colleagues, who found in 2017 that modern Egyptians share about 8 per cent more ancestry with sub-Saharan Africans than the ancient Abusir mummies did, most of it added in the last 1,500 years through Nile trade, mobility and the trans-Saharan slave trade.
His closest living relatives
The nearest living populations are the Copts, both in Egypt and in the Sudanese Coptic community, at about 0.053 to 0.055. Egyptian Muslims are at 0.081. Nubians, Ethiopians, Sardinians and the English are all three to five times further away, and the Dinka and Yoruba more than ten times. None of these numbers means "descendant" or "not descendant". A distance of 0.05 is more than twice the noise yardstick: no modern population is a close match, which is exactly what you expect after 4,700 years of history layered on top.
What each side gets wrong
| Claim you will see online | What the genome actually shows |
|---|---|
| "The ancient Egyptians were Black Africans." | This man's ancestry is overwhelmingly North African and West Asian, with no substantial sub-Saharan component above a floor of about 3 per cent. The skin prediction is a separate question, covered in the next row. |
| "The skin prediction proves he was Black." | It predicts skin colour, not a population. Pigmentation is shaped by a handful of genes that do not track overall ancestry, the prediction from a 2x genome carries real uncertainty, and a dark-skinned North African with no recent sub-Saharan ancestry is not a contradiction. |
| "The ancient Egyptians were white Europeans." | Neither source population is European, and present-day Europeans are among the most distant populations in Figure 6. |
| "The Mesopotamian quarter proves foreign invaders founded Egypt." | This is the old "dynastic race" theory in new clothes. The admixture cannot be dated, and archaeology suggests the shared ancestry goes back to the Neolithic. The man himself grew up in the Nile Valley. |
| "Modern Egyptians are Arabs, not the descendants of the pharaohs." | Modern Egyptians carry later West Asian and sub-Saharan layers, but the ancient Egyptian population is still a major part of their ancestry, and Copts remain closest to the Abusir mummies of the first millennium BC. |
| "Copts are the pure descendants of the ancient Egyptians." | Copts are the closest living group, but they also carry the West Asian shift that happened after the Old Kingdom, plus about 5 per cent more recent sub-Saharan ancestry. |
What this genome cannot tell us
One genome is a spotlight, not a floodlight. The Nuwayrat man lived in Middle Egypt. Upper Egypt, closer to Nubia, has almost certainly always been more connected to populations further south, and the Nile Delta more connected to the Levant. He was one man of relatively high status, possibly a craftsman who had risen, and his family may not have been typical. The North African side of every model uses farmers from Morocco, because the Neolithic Nile Valley has not been sequenced. And the pigmentation prediction comes from a low-coverage genome, with the uncertainty that implies.
What the genome does do is anchor the start of the story. At the dawn of the pyramid age, at least one Egyptian was a mix of the Neolithic peoples of North Africa and the eastern Fertile Crescent, with no meaningful ancestry from south of the Sahara. Over the following two millennia the population of Middle Egypt shifted toward the Levant, and in the last 1,500 years it gained a modest sub-Saharan layer. Neither camp of the internet argument predicted that story, and that is usually a sign that the data are finally speaking for themselves.
Global25 coordinates used in this article
From the Moriopoulos 2026 collection and Davidski's Global25. The Nuwayrat man is spelled "Neurat" in the Moriopoulos file. Paste directly into the Vahaduo tool.
Egypt_Early_Dynastic-Old_Kingdom_Neurat_(n=1),0.011382,0.129988,-0.043746,-0.122095,0.008925,-0.052710,-0.029141,-0.006000,0.078128,-0.005832,0.008607,-0.017984,0.044152,0.000000,0.006650,0.009546,-0.003260,-0.007348,-0.010559,0.023761,0.002496,-0.002844,0.009860,0.012532,-0.003832
Egypt_Late_Period_Abusir-el_Meleq_(n=2),0.050082,0.146744,-0.042426,-0.118380,-0.002001,-0.049782,-0.015628,-0.004731,0.047449,0.006196,0.009012,-0.011465,0.029732,-0.005849,-0.000815,-0.002056,-0.007431,-0.001963,-0.003520,0.012006,0.005615,0.001669,0.003635,0.003916,-0.000598
Morocco_MN_Skhirat-Rouazi_(n=3),-0.005312,0.145559,-0.022879,-0.111866,0.038981,-0.055592,-0.026869,-0.001846,0.106898,0.008747,0.010177,-0.021681,0.048265,-0.014496,0.015020,0.001502,0.002738,-0.022213,-0.036662,0.028597,-0.006322,-0.026998,0.014009,-0.013616,-0.001357
Iraq_N_PPN_Nemrik_9_(n=1),0.064879,0.134050,-0.080327,-0.071060,-0.048009,-0.015618,0.004230,-0.010384,-0.032110,-0.017312,0.007470,-0.007793,0.012042,0.003716,-0.007329,0.008353,0.006389,0.008995,0.000628,0.000375,0.016097,0.010263,-0.010106,-0.019641,0.004910
Jordan_N_PPNB_(n=6),0.068863,0.159269,-0.031489,-0.133239,0.026774,-0.064377,-0.013552,-0.007346,0.079253,0.033774,0.018025,-0.018059,0.034613,0.001399,-0.017372,0.001017,0.010735,-0.001520,-0.006222,0.015174,-0.001477,0.005626,0.003389,-0.004478,-0.001497
Jordan_EBA_'Ain_Ghazal_(Levant_EBA_South)_(n=3),0.075882,0.147929,-0.059962,-0.114020,-0.005642,-0.047690,-0.010889,-0.013307,0.033678,0.007532,0.015752,-0.019682,0.041526,-0.000872,-0.003122,0.022894,0.001825,0.001014,-0.003058,0.017092,0.005074,0.011706,-0.003862,0.004258,-0.007105
Lebanon_MBA_Sidon_Canaanite_(n=5),0.081270,0.146845,-0.061320,-0.097934,-0.010525,-0.038933,-0.004794,-0.006600,0.011453,0.010169,0.010490,-0.009771,0.021585,0.005505,-0.007573,0.006523,-0.000156,-0.001191,0.001509,0.004252,0.004267,0.007543,-0.001257,0.001591,-0.000503
Iran_N_Ganj_Dareh_(n=7),0.044066,0.066155,-0.156344,0.006875,-0.124067,0.022630,0.015679,-0.000395,-0.082102,-0.054827,-0.001021,-0.001713,0.004948,-0.008120,0.033833,0.055801,-0.006482,0.009284,0.009679,-0.035035,0.007540,-0.029800,-0.011973,-0.037320,0.022188
Ethiopia_LSA_Mota_4500BP_(n=1),-0.511066,0.043668,0.000754,0.000969,-0.002770,-0.011435,0.050997,-0.045229,0.089172,-0.087838,-0.012991,-0.002997,-0.031219,0.000688,0.021580,-0.029965,0.027772,0.039273,0.001760,-0.009004,0.000374,0.006183,-0.003451,-0.002410,-0.000838
Egyptian_Copt_(n=27),0.017729,0.142455,-0.049604,-0.104384,-0.000371,-0.043313,-0.014851,-0.002682,0.046742,-0.000362,0.007449,-0.012015,0.028901,0.002011,0.000624,0.001569,-0.003961,-0.001525,-0.003607,0.004888,0.001147,0.000706,0.000571,0.002179,-0.000501
Egyptian_Arab_Muslim_(n=74),-0.020893,0.134029,-0.042424,-0.084544,-0.004164,-0.032185,-0.013353,-0.000132,0.036472,-0.000461,0.006568,-0.009880,0.021547,0.000424,0.000264,0.001747,-0.001368,-0.002936,-0.003867,0.003126,-0.000623,-0.000660,0.001811,0.002086,-0.000196
Dinka_(n=33),-0.577601,0.053054,0.001211,-0.008026,-0.003022,-0.002392,-0.013324,0.015908,0.079653,-0.098794,-0.020397,0.023579,-0.041837,-0.000113,0.008851,-0.017184,0.019648,-0.011905,0.021068,-0.020498,0.000760,0.003432,-0.004037,-0.001807,0.005784
Yoruba_Nigeria_(n=38),-0.630071,0.063791,0.021674,0.016422,0.000462,0.013159,-0.044732,0.047597,-0.048806,0.033464,0.005316,0.001850,0.024443,0.001351,0.012854,-0.010502,0.007229,0.000570,0.006086,-0.003097,0.001080,0.002372,-0.001314,-0.000701,0.000076
Methods in brief
Global25 scaled coordinates were merged with first occurrence priority from Davidski's population averages, modern averages, the Moriopoulos 2026 collection (all averages, no sims) and the dated individual ancients file; scale compatibility was checked on shared populations. Distances are Euclidean over 25 dimensions. NNLS models use a sum to one constraint enforced by a heavily weighted extra equation, with results normalised; condition numbers and pole separations are reported. The noise yardstick is the distance between the two versions of the Nuwayrat genome (0.021). Noise replicates add the difference vector between those versions, with dimensions shuffled and signs randomised, to the target. Spike-in tests add a known fraction of a source to the target and re-fit 500 times under noise to set the detection floor. Principal components were computed on 211 present-day population averages with ancient samples projected. G25 results are a proxy; where the published qpAdm results exist, they take precedence.
References
- Genomics Morez Jacobs A, Irish JD, et al., Girdland-Flink L, Skoglund P (2025). Whole-genome ancestry of an Old Kingdom Egyptian. Nature, published 2 July 2025. doi.org/10.1038/s41586-025-09195-5
- Genomics Schuenemann VJ, Peltzer A, Welte B, et al. (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
- Genomics Simões LG, Günther T, Martínez-Sánchez RM, et al. (2023). Northwest African Neolithic initiated by migrants from Iberia and Levant. Nature 618: 550 to 556. doi.org/10.1038/s41586-023-06166-6
- Genomics van de Loosdrecht M, Bouzouggar A, Humphrey L, et al. (2018). Pleistocene North African genomes link Near Eastern and sub-Saharan African human populations. Science 360: 548 to 552. doi.org/10.1126/science.aar8380
- Genomics Gallego Llorente M, Jones ER, Eriksson A, et al. (2015). Ancient Ethiopian genome reveals extensive Eurasian admixture in Eastern Africa (title as corrected by the 2016 erratum). Science 350: 820 to 822. doi.org/10.1126/science.aad2879
- News Liverpool John Moores University (2025). Researchers sequence first genome from ancient Egypt. ljmu.ac.uk
- News Max Planck Institute (2017). The first genome data from ancient Egyptian mummies. shh.mpg.de
- Data Global25 PCA coordinates (Davidski) and the Moriopoulos 2026 collection of Global25 averages.