Ancient DNA | The first Americans | The Solutrean hypothesis

The Solutrean myth

Did Ice Age hunters from France and Spain cross the frozen Atlantic and found the Clovis culture of North America? The idea fills documentaries and forums. The only Clovis genome ever sequenced, and an actual Solutrean genome from Spain, make it possible to test it directly. We did, in Global25.

About 1%extra Solutrean-related ancestry in the Clovis boy compared with other early Americans, in our test: indistinguishable from noise
3 timesmore of the boy's ancestry is explained by a 24,000-year-old Siberian than by the Solutrean of Spain
4,000 yearsseparate the last Solutrean points in Europe from the first Clovis points in America

Hero image: AI generated illustration of Clovis hunters in the foothills of the Rocky Mountains about 13,000 years ago, one shaping a stone spear point with an antler tool while mammoths graze in the distance. Clothing, tools and the scene are approximations, and it is not a reconstruction of any individual whose genome is discussed here.

It is one of the most seductive ideas in American archaeology. Some 20,000 years ago, at the height of the last Ice Age, hunters of the Solutrean culture of France and Spain, makers of exquisite leaf-shaped spear points, paddled along the edge of the sea ice that stretched across the North Atlantic, living off seals and seabirds, until they reached the east coast of North America. Thousands of years later, their descendants would invent the Clovis spear point and spread across the continent. The first Americans, in this story, were Europeans.

The idea has been championed since the late 1990s by two respected archaeologists, Dennis Stanford of the Smithsonian Institution and Bruce Bradley of the University of Exeter, who set it out at length in their 2012 book Across Atlantic Ice. It has been the subject of television documentaries, and it has found an eager audience online, not always for scientific reasons. It also makes a clear prediction that can be tested: if Clovis people descended from Solutreans, their genomes should show it. Since 2014 we have a Clovis genome. Since 2023 we also have a Solutrean one. We put them side by side.

The first Americans and the Solutrean claimKey dates, from Ice Age Siberia and Europe to the Clovis genomeabout 24,000 years agoThe boy of Mal'taA child buried near Lake Baikal, Siberia, whose peoplelater gave part of their DNA to the first Americans.23,000 to 21,000 years agoFootprints at White SandsHuman tracks in New Mexico, dated in 2021 and confirmedin 2025: people were in America long before Clovis.about 22,000 to 17,000 years agoThe Solutrean cultureIce Age hunters of France and Iberia make fineleaf-shaped points. Clovis comes 4,000 years later.13,050 to 12,750 years agoClovisFluted spear points spread across North Americain about three centuries.about 12,600 years agoThe Anzick boyA child buried in Montana with more than 100 Clovistools, under red ochre. The only Clovis burial known.2014The Clovis genomeRasmussen and colleagues: the Anzick boy belonged toa population directly ancestral to many Native Americans.2015Haplogroup X2a and Kennewick ManKennewick Man carries the oldest X2a lineage known,and his genome is Native American, not European.Ages in calendar years before present. The Solutrean dates are approximate.
Key dates in the debate over the first Americans. Ages are in calendar years before present and approximate; the Solutrean range follows Oppenheimer, Bradley and Stanford (2014), the Clovis range Waters and colleagues (2020).

An elegant idea

The Solutrean hypothesis rests on a resemblance. Solutrean knappers of southwestern France and Iberia, between about 22,000 and 17,000 years ago, made thin bifacial points with a technique called overshot flaking, in which a flake is struck clean across the face of the piece. Clovis knappers in North America, around 13,000 years ago, used the same technique. Stanford and Bradley also pointed to stone tools from the east coast of the United States, including a laurel-leaf blade dredged from the seabed off Virginia, that they believe to be older than Clovis and closer to the Solutrean in style. In their view, the Atlantic ice margin was not a barrier but a highway, rich in marine food.

Critics have answered point by point. The two cultures are separated by some 4,000 years and 5,000 kilometres of ocean, with no trace of the intermediate people in between. Clovis points are fluted, with a channel struck from the base, a feature unknown in the Solutrean. Overshot flaking was invented independently in several places, and the age and context of the Virginia blade have been questioned. Most archaeologists see the similarities as convergence: good stone, skilled hands and similar needs lead to similar tools. For two decades, the debate was mostly about stones.

The Clovis boy speaks

In 1968, construction workers near the town of Wilsall, in Montana, uncovered more than 100 stone and antler tools, many of them coated in red ochre, together with the bones of a small child. The Anzick site, named after the family who owned the land, is still the only known Clovis burial. The child, a boy between one and two years old, lived around 12,600 years ago, during the brief window of the Clovis culture, now dated to between about 13,050 and 12,750 years ago.

In 2014, Morten Rasmussen, Eske Willerslev and colleagues published his genome. The result was unambiguous. The boy belonged to a population directly ancestral to many living Native Americans, and he was more closely related to present-day peoples of Central and South America than to any population of Europe or Asia. His mitochondrial lineage, D4h3a, and his Y chromosome, a branch of haplogroup Q, are both typical Native American lineages of Asian origin. The authors stated that their data were incompatible with the Solutrean hypothesis. In the same year, after consultation with Native American tribes, the boy was reburied near the place where he was found.

Where the European-looking signal comes from

Supporters of the Solutrean idea had one genetic argument that seemed strong: Native Americans share some genetic variants with Europeans that East Asians do not have. The explanation came from Siberia. In the 1920s, Soviet archaeologists excavated the burial of a boy at Mal'ta, near Lake Baikal. In 2014, Maanasa Raghavan and colleagues sequenced his genome. The Mal'ta boy, who lived about 24,000 years ago, belonged to a population that geneticists now call Ancient North Eurasians. Part of their ancestry also went west, into later Europeans, and part of it went east, into the ancestors of Native Americans, who received an estimated 14 to 38% of their ancestry from people related to him.

That shared Siberian ancestry is why Native Americans and Europeans have a genetic link. It does not require anyone to cross the Atlantic: the link runs through Siberia, and it was formed at least 20,000 years ago, on land.

Testing the Clovis boy against a real Solutrean

Until recently, the test the Solutrean hypothesis needed was impossible, because no Solutrean individual had been sequenced. That changed in 2023, when Cosimo Posth and colleagues published the genome of a Solutrean individual from the cave of Malalmuerzo, near Granada in southern Spain, about 23,000 years old. It is a low-coverage genome, but it is in Global25. So is the Anzick boy, along with the Mal'ta boy, East Asian hunter-gatherers, early Americans from Alaska, Nevada, Washington and Brazil, and the hunter-gatherers of Ice Age Europe.

First, a simple question: who is the Clovis boy closest to?

Who is the Clovis boy closest to?Global25 distance from the Anzick genome to ancient and present-day populations (our analysis)0100200300400500600Zapotec, Mexico (today)26Maya, Mexico (today)28Lapa do Santo, Brazil, 9,600 years ago30Pima, Mexico (today)31Spirit Cave, Nevada, 11,000 years ago39Mixe, Mexico (today)42Kennewick Man, 8,500 years ago73Karitiana, Amazonia (today)95Mal'ta boy, Siberia, 24,000 years ago465Devil's Gate, East Asia, 7,700 years ago489Gravettian, Fournol, France545Solutrean, Spain, 23,000 years ago574Magdalenian, El Miron, Spain621Distances x 1000. Green: the Americas. Blue: Siberia and East Asia. Red: Ice Age Europe.
Global25 distance from the Anzick genome to ancient and present-day populations (our analysis). Native American genomes are strongly drifted, which makes them distant from every ancient source in Global25; what matters is the comparison between sources, not the absolute values.

The answer is clear. The closest populations to the Anzick boy are living peoples of Mexico, the Zapotec, Maya and Pima, and the early Americans of Brazil and Nevada, all within 26 to 42 units. Every Ice Age European, the Solutrean included, sits more than 540 units away. The Mal'ta boy and the early East Asians of Devil's Gate cave, in the Russian Far East, are also far, at 465 and 489 units, which says something about Native American genomes rather than about Siberia: Native American populations went through long periods of isolation and small numbers, which makes them distant in Global25 from every ancient source. That is why the next test does not ask "who is closest" but "which source helps most".

Siberia explains the signal, not EuropeHow much better each early American genome fits when a second source is added to an East Asian one (our analysis)0255075100125Anzick, Montana (Clovis)9332Upward Sun River, Alaska11647Spirit Cave, Nevada7723Kennewick, Washington8630Lapa do Santo, Brazil8830Adding the Mal'ta boy (Siberia)Adding the Solutrean (Spain)Reduction of the fit distance (x 1000) against an East Asian source alone (Devil's Gate). Bigger is better.
Improvement of the fit distance when the Mal'ta boy (Siberia, about 24,000 years ago) or the Solutrean of Malalmuerzo (Spain, about 23,000 years ago) is added to an East Asian source (Devil's Gate cave, Russian Far East), for five early American genomes (our analysis). Bigger is better.

We modelled each early American genome with an East Asian source alone, then added either the Mal'ta boy or the Solutrean of Spain as a second source, and measured how much the fit improved. For the Anzick boy, adding Mal'ta improves the fit by 93 units, adding the Solutrean by 32. The same pattern holds for every early American we tried, from the 11,500-year-old infant of Upward Sun River in Alaska to the 9,600-year-old people of Lapa do Santo in Brazil: Mal'ta always explains two and a half to three times more. The Solutrean helps a little only because it shares part of its deep ancestry with Mal'ta; the real signal is Siberian.

Is there any Solutrean left over?

The decisive test is a different one. If the Clovis people had received extra ancestry from Solutreans, it would show up as something the Anzick boy has and other early Americans lack. We therefore modelled the boy as another Native American population, plus Mal'ta, plus the Solutrean, and asked how much Solutrean the model needs.

Did the Clovis people carry extra Solutrean ancestry?Solutrean share in the Anzick genome when compared with other Native Americans, Siberian MA-1 also offered (our analysis)0.0%2.0%4.0%6.0%8.0%10.0%vs Lapa do Santo, Brazil1.1%vs Spirit Cave, Nevada1.1%vs Mixe, Mexico0.5%vs Upward Sun River, Alaska0.0%vs Karitiana, Amazonia7.4%Spiking 5% of the Solutrean genome into Anzick reads as 5.5 to 6%: a real signal would show.
Solutrean share needed by the Anzick genome when it is modelled as another Native American population plus the Mal'ta boy plus the Solutrean of Spain (our analysis). Mixing 5% of the Solutrean genome into Anzick reads back as 5.5 to 6%. The Karitiana value reflects their strong drift and poor fit (distance 46).

Almost none. Against the ancient Brazilians of Lapa do Santo or the 10,700-year-old man of Spirit Cave in Nevada, the Solutrean share is 1.1%; against the Mixe of Mexico, 0.5%; against the Upward Sun River infant of Alaska, zero. Only the Karitiana of Amazonia, a small and very isolated population whose genome fits poorly in this model, give a higher value, 7.4%, which reflects their own drift rather than anything in the Clovis boy.

Could the test simply be blind? We checked. When we mix 5% of the Solutrean genome into the Anzick genome ourselves, the same model reads it back as 5.5 to 6%. A real Solutrean contribution of that size would have been seen. We also report a trap we found along the way: a simpler model, East Asian plus Mal'ta plus Solutrean, gives zero Solutrean to every early American, but it also gives zero when we add 20% Solutrean ourselves, because the Mal'ta source absorbs the signal. We do not use that model, and anyone running these tests at home should be wary of it.

Two honest limits apply. Global25 is a summary of the genome and handles Native Americans poorly, because of their strong drift; formal statistics on full genomes, like those of Rasmussen and colleagues, remain the reference. And no method can exclude a contribution of a fraction of a percent. What the data rule out is the Solutrean hypothesis as it was proposed: a Clovis people descended from European migrants.

Haplogroup X2a and Kennewick Man

The second pillar of the genetic case for the Solutrean idea was a maternal lineage, haplogroup X2a, found in some Native American peoples of the Great Lakes and northern plains. Haplogroup X also exists in Europe and the Near East, so X2a was presented as a European signature. In 2015, Jennifer Raff and Deborah Bolnick reviewed the evidence: X2a has never been found in Europe or anywhere else in the Old World, the X lineages of Europe belong to other branches, and the most basal X2a lineage known comes from Kennewick Man.

Kennewick Man is himself a chapter of this story. When his 8,500-year-old skeleton was found on the banks of the Columbia River in Washington State in 1996, the first anthropologist to examine it described its skull as "Caucasoid", and the find was soon cited as evidence of early Europeans in America. In 2015, Rasmussen and colleagues sequenced his genome: he was Native American, most closely related to living tribes of the region, including the Colville. In our Global25 analysis he sits 73 units from the Anzick boy, and his fit improves three times more with Mal'ta than with the Solutrean. Skull shape, as so often, had misled.

When did the first Americans arrive?

The fall of the Solutrean hypothesis does not mean that Clovis people were the first Americans. At White Sands, in New Mexico, human footprints in ancient lake mud were dated in 2021 to between 23,000 and 21,000 years ago, and independent dating published in 2023 and 2025 supported that age. If it holds, people were in North America during the last glacial maximum, at the same time as the Solutrean culture flourished in Europe. Some fans of the Solutrean idea see a gap to fill there. But nothing at White Sands points to Europe, and every ancient American genome sequenced so far, from Alaska to Patagonia, traces its ancestry to northeast Asia.

The other famous pre-Clovis site is in turmoil. Monte Verde, in southern Chile, dated to about 14,500 years ago, was for 25 years the strongest proof that people reached South America before Clovis. In March 2026, Todd Surovell and colleagues argued in Science that a layer of volcanic ash about 11,000 years old lies beneath the archaeological level, and that the occupation dates to between about 8,200 and 4,200 years ago, most likely 8,000 to 6,000. Tom Dillehay, who has led research at Monte Verde since the 1970s, replied that no such ash was ever observed at the site itself and that the new team had studied deposits outside it; in August 2026 his team, and an independent group of geologists, published rebuttals in the journal PaleoAmerica. The question is open, and the "Clovis first" debate has been revived. Whatever the answer, it concerns the date at which people came from Asia, not whether they came from Europe.

So, did Ice Age Europeans discover America?

The genomes say no. The only Clovis individual ever sequenced is fully Native American. His affinities with Europeans are explained by a Siberian population that contributed to both, and when he is compared with an actual Solutrean from Spain, the share of Solutrean ancestry he needs is indistinguishable from zero, while a test with a known 5% spike shows that such a contribution would be visible. The European link that so many found in Native American genomes was real, but it was formed in Siberia, not on the Atlantic ice.

The first Europeans whose presence in America is proven are the Norse, who were felling trees at L'Anse aux Meadows, in Newfoundland, in AD 1021. They came some 12,000 years after the Clovis boy, to a continent that had been home to the descendants of Asian hunters for longer than any written history.

The Solutrean hypothesis remains a fine example of science at work. It made a clear prediction, and when the data became available, the prediction failed. One Clovis genome cannot describe all the people of the Clovis world, and more ancient genomes from the east coast of North America would be welcome. So far, every one of them points west, across Beringia, to Asia.

Key points

  • The Solutrean hypothesis proposes that Ice Age Europeans crossed the North Atlantic and founded the Clovis culture; it rests mainly on similarities between stone tools separated by 4,000 years.
  • The only Clovis genome, the Anzick boy of Montana, about 12,600 years old, belongs to a population ancestral to many living Native Americans and is closest to peoples of Mexico and to early Americans.
  • Native Americans share some ancestry with Europeans through Ancient North Eurasians of Siberia, related to the 24,000-year-old Mal'ta boy, who contributed to both.
  • In our Global25 tests, adding the Mal'ta boy improves the fit of early American genomes about three times more than adding a Solutrean genome from Spain.
  • Compared with other early Americans, the Anzick boy needs about 1% or less of Solutrean ancestry, while a known 5% spike is read back as 5.5 to 6%.
  • Haplogroup X2a has never been found in Europe, and its most basal lineage belongs to Kennewick Man, whose genome is Native American.
  • The dates of the first arrivals are debated again, from White Sands to Monte Verde, but every ancient American genome traces its ancestry to Asia.

Global25 coordinates used in this article

The Clovis boy, the Solutrean of Spain, the Mal'ta boy, other ancient sources and a few present-day Native American populations. They can be pasted directly into the Vahaduo tool.

Anzick_Clovis_Montana_12600BP,0.051220,-0.298566,0.116530,0.097546,-0.101557,-0.012550,-0.263682,-0.319371,-0.011044,-0.025331,0.005359,-0.003597,0.000743,0.011010,-0.010043,0.006629,0.007432,0.000887,-0.004902,0.003627,-0.004991,0.002102,-0.008504,0.000361,0.002515
Solutrean_Malalmuerzo_Spain_23000BP,0.055773,0.010155,0.039975,0.085918,0.066782,0.019522,-0.003525,0.001154,0.080787,0.036629,0.017863,-0.004646,0.004014,-0.027937,0.019001,0.020419,0.001304,0.010135,-0.013198,0.042020,0.040304,0.002473,-0.022061,-0.095435,0.017483
MA1_Malta_Siberia_24000BP,0.076261,-0.024373,0.036581,0.187018,-0.069859,0.045738,-0.050057,-0.059305,-0.019021,-0.068339,0.030691,-0.012589,0.016353,-0.044039,0.018051,0.021214,-0.008084,0.002280,-0.003771,0.002751,-0.028075,0.005070,0.013188,-0.007471,0.000000
DevilsGate_RussianFarEast_N,0.020868,-0.448187,0.068762,-0.050173,-0.040931,-0.036534,0.004622,0.008000,0.003613,0.021686,-0.051802,-0.005745,0.000545,0.007982,-0.009093,-0.009105,-0.004477,0.007475,0.017723,0.011631,0.008277,-0.026462,-0.020459,-0.001968,-0.013332
UpwardSunRiver_USR1_Alaska,0.037562,-0.305674,0.088623,0.087533,-0.086478,-0.018407,-0.129256,-0.158763,-0.012271,-0.025149,0.003085,-0.006744,0.008176,-0.007982,-0.009093,0.001326,0.001173,0.008742,0.006285,0.013632,-0.009358,0.003957,0.002465,0.003133,0.002275
SpiritCave_Nevada,0.046667,-0.320907,0.119170,0.095608,-0.110790,-0.016733,-0.279193,-0.327909,-0.012476,-0.017495,-0.000812,-0.000300,-0.000149,0.018579,-0.007057,0.006895,0.003520,-0.000127,0.002640,0.002501,0.006364,0.000742,0.003328,0.003374,-0.003353
Kennewick_Washington,0.053497,-0.312783,0.110496,0.083011,-0.099403,-0.014502,-0.226550,-0.265604,-0.012271,-0.023326,0.008119,-0.002398,0.000297,0.011560,-0.004072,0.004773,0.005998,0.004181,0.012193,0.001251,0.003993,-0.011252,-0.005669,0.001084,0.000599
LapaDoSanto_Brazil_9600BP_avg,0.047521,-0.302120,0.118039,0.101907,-0.110098,-0.019871,-0.275257,-0.335121,-0.008334,-0.018497,0.004709,0.001574,0.000111,0.016343,-0.005225,0.006033,0.006030,0.001362,0.002011,0.001157,-0.000624,0.007666,-0.000925,-0.004759,-0.002754
ElMiron_Magdalenian_Spain,0.073985,0.044683,0.101068,0.130493,0.110482,0.032909,0.000470,0.009230,0.085900,0.025149,-0.005684,-0.007343,0.011150,-0.010459,0.034880,0.039644,0.003390,0.004814,-0.016215,0.046772,0.066133,0.007914,-0.035865,-0.130741,0.011017
Zapotec_today,0.053952,-0.298363,0.112533,0.091926,-0.103835,-0.015953,-0.267254,-0.324263,-0.006627,-0.014251,-0.005002,-0.005755,0.003003,0.020038,-0.005836,0.000159,0.002816,-0.000785,0.002615,0.001251,-0.002471,0.005243,0.000099,0.000386,-0.001269
Maya_today,0.056749,-0.294939,0.110766,0.090256,-0.104855,-0.015698,-0.275835,-0.328635,-0.014083,-0.016037,-0.001717,-0.001006,0.001083,0.019995,-0.008783,0.003750,0.007916,-0.000977,-0.001347,-0.001947,-0.002603,0.005617,0.001004,0.000034,0.000359
Pima_today,0.051676,-0.312783,0.107706,0.090570,-0.107343,-0.015953,-0.269933,-0.327309,-0.012517,-0.017713,-0.003248,-0.002218,0.005382,0.019487,-0.009120,0.010952,0.006258,0.006436,0.004626,-0.003152,-0.002496,0.004451,-0.001972,0.003181,0.003329
Mixe_today,0.057822,-0.320704,0.123017,0.094381,-0.107343,-0.015562,-0.278206,-0.334048,-0.012026,-0.014688,-0.007113,-0.006564,0.004460,0.020203,-0.008795,0.004004,0.006102,0.000456,-0.000453,-0.003602,0.001173,0.005861,0.003402,-0.001711,-0.001844
Karitiana_today,0.058809,-0.320061,0.118353,0.108367,-0.113662,-0.013619,-0.323061,-0.382484,-0.015373,-0.017343,0.000947,-0.002698,-0.001635,0.027318,-0.004569,0.004420,0.011017,0.002597,0.000105,-0.000354,-0.005532,0.010902,-0.005546,-0.002530,-0.006786

References

  1. Ancient DNA Rasmussen M, Anzick SL, Waters MR, Skoglund P, DeGiorgio M, Stafford TW, et al., Willerslev E (2014). The genome of a Late Pleistocene human from a Clovis burial site in western Montana. Nature 506: 225 to 229. doi.org/10.1038/nature13025
  2. Ancient DNA Raghavan M, Skoglund P, Graf KE, Metspalu M, Albrechtsen A, et al., Willerslev E (2014). Upper Palaeolithic Siberian genome reveals dual ancestry of Native Americans. Nature 505: 87 to 91. doi.org/10.1038/nature12736
  3. Ancient DNA Rasmussen M, Sikora M, Albrechtsen A, Korneliussen TS, et al., Willerslev E (2015). The ancestry and affiliations of Kennewick Man. Nature 523: 455 to 458. doi.org/10.1038/nature14625
  4. Ancient DNA Posth C, Yu H, Ghalichi A, Rougier H, Crevecoeur I, et al., Krause J (2023). Palaeogenomics of Upper Palaeolithic to Neolithic European hunter-gatherers. Nature 615: 117 to 126. doi.org/10.1038/s41586-023-05726-0
  5. Genetics Raff JA, Bolnick DA (2015). Does mitochondrial haplogroup X indicate ancient trans-Atlantic migration to the Americas? A critical re-evaluation. PaleoAmerica 1(4): 297 to 304.
  6. Hypothesis Oppenheimer S, Bradley B, Stanford D (2014). Solutrean hypothesis: genetics, the mammoth in the room. World Archaeology 46(5): 752 to 774. doi.org/10.1080/00438243.2014.966273
  7. Hypothesis Stanford D, Bradley B (2012). Across Atlantic Ice: The Origin of America's Clovis Culture. Berkeley: University of California Press.
  8. Archaeology Waters MR, Stafford TW, Carlson DL (2020). The age of Clovis: 13,050 to 12,750 cal yr B.P. Science Advances 6(43): eaaz0455. doi.org/10.1126/sciadv.aaz0455
  9. Archaeology Bennett MR, Bustos D, Pigati JS, Springer KB, Urban TM, Holliday VT, et al. (2021). Evidence of humans in North America during the Last Glacial Maximum. Science 373(6562): 1528 to 1531. doi.org/10.1126/science.abg7586
  10. Archaeology Surovell TA, Mendez C, Garcia JL, Luthgens C, Thompson JM, Latorre C (2026). A mid-Holocene age for Monte Verde challenges the timeline of human colonization of South America. Science. doi.org/10.1126/science.adw9217
  11. Archaeology Kuitems M, Wallace BL, Lindsay C, Scifo A, Doeve P, et al., Dee MW (2022). Evidence for European presence in the Americas in AD 1021. Nature 601: 388 to 391. doi.org/10.1038/s41586-021-03972-8
  12. Data Global25 coordinates by Davidski (Eurogenes), and the Moriopoulos Collection 2026 of averages. Admixture models and figures: ExploreYourDNA.