Where Did Indo-Europeans Really Come From? The 2025 Answer

Hero image: AI generated illustration of an Eneolithic steppe community of the lower Volga in the fifth millennium BC. It is not a reconstruction of any individual.

English, Hindi, Persian, Russian, Spanish, Greek and Armenian are all descended from a single language spoken somewhere, by someone, more than six thousand years ago. Linguists have reconstructed thousands of its words: its terms for wheels and wool, for a sky father and for horses. What they could never agree on was where its speakers lived.

Three answers have dominated the debate. The steppe hypothesis, associated with Marija Gimbutas and later David Anthony, put the homeland in the grasslands north of the Black Sea and the Caspian. The Anatolian hypothesis of Colin Renfrew tied the spread of the languages to the first farmers leaving Anatolia around 7000 BC. A third family of models, from Gamkrelidze and Ivanov to the language tree study of Heggarty and colleagues in 2023, looked to the region south of the Caucasus, in Armenia and the Iranian plateau.

In 2015, ancient DNA seemed to settle the question for Europe. Haak and colleagues showed that people of the Corded Ware culture in Germany carried about three quarters of their ancestry from the Yamnaya herders of the Pontic and Caspian steppe, who had arrived only a few centuries earlier. Most of the Indo-European languages of Europe, and through the later Sintashta and Andronovo horizons the Indo-Iranian languages of Asia, could be traced back to the steppe.

But one branch refused to fit. And it was the oldest one.

The problem of the Hittites

Hittite, Luwian and Palaic, the Anatolian languages of the second millennium BC, are the earliest Indo-European languages ever written down. Most linguists agree that they were the first to break away from the family tree: they lack features, such as a feminine gender and parts of the wheel vocabulary, shared by all the other branches. Any theory of Indo-European origins has to explain how the ancestors of the Hittites reached central Anatolia.

When the first Bronze Age genomes from Anatolia were published, they showed something awkward for the steppe camp. In 2018, Damgaard and colleagues found no trace of the steppe ancestry that marked the Yamnaya and the Corded Ware people. The signature of Yamnaya ancestry is a large share of genes from the eastern hunter-gatherers (EHG) of Russia, and the Hittite period people of Kaman-Kalehöyük simply did not have it. If the Indo-Europeans came from the Yamnaya, why did the oldest Indo-European speakers carry none of their hunter-gatherer ancestry?

The answer, published on 5 February 2025 by Iosif Lazaridis, Nick Patterson, David Anthony, David Reich and more than a hundred co-authors, is that the Yamnaya were not the beginning of the story. They were the second chapter.

A cline between the Caucasus and the Volga

The team assembled genomes from 435 individuals, 367 of them newly reported, dated between 6400 and 2000 BC. Among the most important were people of the Eneolithic, the copper age of the fifth millennium BC, buried at sites in the steppe between the North Caucasus foothills and the lower Volga: Progress-2 and Vonyuchka in the Stavropol region, Remontnoye near Rostov, and Berezhnovka on the Volga near Volgograd.

These people formed a genetic gradient, which the authors call the Caucasus to lower Volga cline, or CLV. At its southern end stand the Neolithic farmers of Aknashen in Armenia and the Maikop people of the North Caucasus, with ancestry from the Caucasus and the Near East. At its northern end stands Berezhnovka, with roughly equal parts of Caucasus and eastern hunter-gatherer ancestry. Remontnoye falls in between. This was a world of intermarriage between the mountains and the steppe, two thousand years before the Yamnaya.

We can measure the same gradient with Global25. Modelling each group as a mix of Aknashen and the Karelian eastern hunter-gatherers, we obtain a simple position along the cline:

The Volga groups offer a second test. Lazaridis and colleagues describe a Volga cline in which the Khvalynsk people of the middle Volga were a mix of Berezhnovka-like CLV ancestry and local hunter-gatherers, in proportions that vary from grave to grave. Their middle Khvalynsk group carries 57.3 ± 1.7% CLV ancestry, and the older Ekaterinovka cemetery 24.3 ± 1.3%. Our Global25 models, with Berezhnovka and EHG as the two sources, return 58.6% for the Khvalynsk average and 28.6% for Ekaterinovka, with fit distances of 0.013 and 0.021. The CLV people, in other words, were already spreading north up the Volga by the early fifth millennium BC.

How the Yamnaya were made

The second step happened on the Dnipro and the Don. Around 4500 BC, groups carrying CLV ancestry pushed west into what is now Ukraine. There they met the descendants of the Neolithic hunter-gatherers of the Dnipro valley and of the Lower Don, and mixed with them. The result is the Sredni Stog culture, known in Ukrainian as Serednii Stih. Its people form a third gradient, the Dnipro cline, between CLV and the Ukraine Neolithic hunter-gatherers.

The Yamnaya came out of this world. In the companion paper by Alexander Nikitin, David Reich and Ukrainian colleagues, a single man from Mykhailivka on the lower Dnipro, dated 3635 to 3383 BC, is already genetically Yamnaya, three centuries before the Yamnaya burial mounds spread across the steppe. The site shows continuity from the Eneolithic to the Bronze Age, and the authors call it a likely epicentre of Yamnaya formation.

In qpAdm, Lazaridis and colleagues find that about four fifths of Yamnaya ancestry comes from CLV people and one fifth from the Dnipro and Don hunter-gatherers: 17.7 ± 1.3% if the hunter-gatherer source is the Ukraine Neolithic, 22.5 ± 1.8% if it is the Lower Don site of Golubaya Krinitsa. They date the mixture to 4038 ± 48 BC. Then something unusual happened. The genomes show a population bottleneck: the Yamnaya descend from a group of a few thousand people that started growing rapidly between about 3642 and 3374 BC, and within a few centuries spread from Hungary to Kazakhstan.

Global25 reproduces the main result. The Samara, Kalmykia and Ukrainian Yamnaya come out at 84 to 87% CLV, against about 80% in qpAdm. The Yamnaya of Rostov on the Don carry more hunter-gatherer ancestry, 29%, which agrees with the paper: it models the Don Yamnaya as Core Yamnaya plus 20.6 ± 1.1% additional Ukraine Neolithic ancestry. With Golubaya Krinitsa instead of the Ukraine Neolithic as the hunter-gatherer source, the CLV share falls to 65% in Samara, because Golubaya Krinitsa itself already sits partway toward CLV (its separation from Berezhnovka is only 0.112). That is the same direction of change as in the paper, but stronger.

One model does not replicate well, and we report it. The paper's best fit for Core Yamnaya is 73.7% Sredni Stog plus 26.3% Remontnoye. In Global25, the same model gives 92% Sredni Stog and 8% Remontnoye for the Samara Yamnaya. Only Mykhailivka returns the published ratio (74.5% and 25.5%). The two sources are just 0.112 apart, and our Sredni Stog average pools four sites of different ages and very different compositions, from 55% to 95% CLV. A pooled proxy of this kind is not precise enough to separate a 26% contribution from an 8% one, and the qpAdm figure remains the reference.

The other wave: Usatove

The CLV people did not only mix with hunter-gatherers. Nikitin and colleagues describe a parallel movement, around 4500 BC, that went further west and bypassed the foragers entirely. In the Usatove culture near Odesa, migrants with CLV ancestry mixed with Trypillian farmers of the Cucuteni-Trypillia world in roughly equal parts. In Global25, Usatove from Mayaky fits as 58.6% Trypillia and 41.4% Berezhnovka (fit distance 0.022, condition number 1.2), with no room for Ukraine Neolithic hunter-gatherers (0.0%). Close to half and half, as the paper says.

Three successive waves out of the Caucasus to lower Volga region, in the words of the authors, each absorbing whoever they met: farmers in one place, foragers in another. It was this flexibility, they argue, that allowed CLV genes and culture to spread across Eurasia.

The Anatolian branch

Now the Hittites. If the CLV people were the common ancestors, then the Anatolians should carry CLV ancestry without the extra hunter-gatherer ancestry that the Yamnaya picked up on the Dnipro. That is exactly what Lazaridis and colleagues found. The Bronze Age people of central Anatolia, from the Early Bronze Age to the Old Hittite period, are a mix of local ancestry related to Neolithic Mesopotamia (the site of Çayönü in southeastern Turkey) and a CLV component. Depending on which CLV group is used as the source, that component is 10.8 ± 1.7% (Berezhnovka and Progress, the "BPgroup"), 19.0 ± 2.4% (Remontnoye) or 33.5 ± 4.8% (Chalcolithic Armenia). The more southern the source, the larger the share, because a southern source carries more of the Caucasus ancestry that the Anatolians already had. The mixture is dated to 4382 ± 63 BC.

Our Global25 values follow the same order and the same scale: 9.0% with Berezhnovka, 13.3% with Remontnoye, 29.8% with Areni. All three are a little lower than the qpAdm estimates, the Berezhnovka figure within the published error.

The decisive point is what is missing. When the authors added eastern or Ukrainian hunter-gatherers to the model, their contribution came out negative (minus 3.4 ± 2.6% EHG, minus 2.3 ± 2.7% Ukraine Neolithic). The people who brought CLV ancestry to Anatolia had no more hunter-gatherer ancestry than the Berezhnovka group. They were not Yamnaya. They were their cousins, who left before the Yamnaya were formed.

In Global25, adding EHG or the Ukraine Neolithic as a third source to the Hittite model also gives exactly 0.0%. But a zero means little without knowing what the method can detect, so we ran a spike in test, adding known amounts of each hunter-gatherer group to the Hittite profile and fitting again. The result is sobering. An added 2, 3 or 5% of EHG is not detected at all, and a 10% spike reads as only 4.9%. For the Ukraine Neolithic, 5% reads as 2.2% and 10% as 7.4%. For Kaman-Kalehöyük and Ovaören the floor is even higher. Global25 therefore cannot, by itself, exclude a hunter-gatherer contribution of up to about 5 to 10%. The reason is geometric: EHG and Berezhnovka point in nearly the same direction from Cayonu, so the solver can absorb a little EHG into a little more Berezhnovka. On this question, the qpAdm result, which has an error of under 3 points, is the one to rely on, and our zero is consistent with it but not independent proof.

There is a second trap worth showing. If we let the solver choose between Berezhnovka and the Samara Yamnaya as the CLV source for the Hittites, it picks the Yamnaya (9.6%) and gives Berezhnovka 0%. This does not mean that the Hittites descend from the Yamnaya. The two candidate sources are only 0.055 apart in Global25, the condition number jumps to 8.9, and spiking in 5% or 10% more Yamnaya simply moves the Yamnaya share up while Berezhnovka stays at zero: the model cannot tell them apart. Distinguishing the Yamnaya from their CLV ancestors requires the outgroup statistics of qpAdm, and that is precisely where the 2025 paper is strongest.

Proto-Indo-Anatolian

Put together, the genetic picture lines up with what linguists have long said about the shape of the family tree. Lazaridis and colleagues propose that the CLV people of 4400 to 4000 BC were the speakers of proto-Indo-Anatolian, the ancestor of both Hittite and of all other Indo-European languages. One group of their descendants moved south into Anatolia, with no Dnipro hunter-gatherer ancestry, and gave rise to the Anatolian branch. Another moved west onto the Dnipro, mixed with local foragers, and after a bottleneck became the Yamnaya, the speakers of proto-Indo-European in the narrow sense. From the Yamnaya came the Afanasievo people of the Altai, most likely the ancestors of the Tocharian languages, which form the second split of the tree, and then the Corded Ware and Bell Beaker people of Europe and the steppe pastoralists who carried Indo-Iranian into Asia.

The linguistic dates fit. Independent estimates put the Indo-Anatolian split between 4300 and 3500 BC, overlapping the genetic window.

So where is the homeland?

The honest answer is: in the steppe, but at its southern edge, and with deep roots in the Caucasus. The CLV people lived north of the Caucasus, in what is now southern Russia, but a large part of their ancestry came from farmers related to those of Neolithic Armenia. In that sense, each of the old camps was partly right. The steppe hypothesis is right about the people who carried most of the branches. The southern models are right that a Caucasus and Near Eastern component sits at the very root of the family, in the genomes if not necessarily in the language.

Some questions remain open. The route by which CLV ancestry reached central Anatolia, across the Caucasus or around the Black Sea through the Balkans, is not settled; the absence of European farmer ancestry in the Anatolians makes a Balkan route harder to argue. Genes are also not languages. The case that the CLV people spoke proto-Indo-Anatolian rests on the match between genetic and linguistic trees, not on any inscription. And proponents of a southern Caucasus homeland, such as the authors of the 2023 language tree study, still read the same evidence differently. What has changed is that the steppe model now has a candidate population for the common ancestor of all known branches, Anatolian included, and that population is in the data.

Steppe ancestry in Europe today

The descendants of the Yamnaya transformed the genetic map of Europe. With three sources, Western hunter-gatherers (Loschbour), Anatolian Neolithic farmers and the Samara Yamnaya, modern Europeans fall on a clear north to south gradient.

Lithuanians, Poles and Norwegians carry more than half of their ancestry from Yamnaya-related people. English, Hungarians and the French of Paris are in the 40 to 50% range, southern Europeans in the 26 to 40% range, and Sardinians, the most farmer-rich population of Europe, at about 5%. The Basques, who speak the only surviving pre-Indo-European language of western Europe, still carry 26%: language and genes do not always travel together. These are simplified models, and the fit distances for northeastern Europeans (0.085 for Lithuanians) show that a fourth source, related to Siberian and Baltic hunter-gatherers, is missing there.

Every time you speak a language of this family, you are using words whose ancestors were first spoken by people who lived between the Caucasus and the Volga six thousand years ago. Their genomes have finally been found.

Methods

We used Global25 scaled coordinates by Davidski and the Moriopoulos 2026 collection, merged with first occurrence priority (Global25 population averages, then Global25 modern averages, then the Moriopoulos all averages file without simulations, then the individual dated ancient samples). Scale compatibility was checked on populations present in both sets (distances 0.005 to 0.029). Distances are Euclidean in the 25 dimensional space. Admixture models use non negative least squares with a sum to one constraint enforced by a heavily weighted extra row (weight 1000). Sources: EHG = Russia_Karelia_HG; Ukraine Neolithic = Ukraine_N; CHG = Georgia_Kotias.SG; Aknashen = Armenia_Aknashen_N; Remontnoye (n=2); Progress-2 (n=2); Berezhnovka (n=4); Khvalynsk (n=22); Ekaterinovka (n=32); Golubaya Krinitsa (n=2); Sredni Stog = average of Dereivka II (n=3), Igren (n=2), Vynogradne (n=1) and Krivyansky (n=1); Mykhailivka (n=1); Cayonu (n=1, the only Mesopotamian Neolithic genome available); Areni = Armenia_C_Areni; Trypillia = Verteba (n=21); Usatove = Mayaky (n=9). Condition numbers, pole separations and fit distances are given with each figure. Spike in tests added 2, 3, 5 and 10% of EHG or Ukraine_N to each Anatolian target before refitting. The PCA in Figure 1 is computed on the 21 plotted groups only. Dates, sample sizes and qpAdm figures are taken from Lazaridis et al. (2025) and Nikitin et al. (2025) and are not recomputed here.

Global25 coordinates used in this article

The source and target populations of our models. They can be pasted directly into the Vahaduo tool.

EHG_Karelia_HG,0.125775,0.034020,0.131803,0.212535,-0.006616,0.056196,-0.019741,-0.021229,-0.001125,-0.088293,0.016807,-0.020157,0.035158,-0.033099,0.017032,0.028639,-0.020470,0.001837,-0.006411,0.009817,-0.008797,0.012922,0.008257,-0.023859,-0.007185
Ukraine_N_HG,0.128336,0.091398,0.147580,0.183142,0.063576,0.055708,-0.000313,0.011538,0.020282,-0.049644,0.001299,-0.014999,0.024281,-0.015815,0.029169,0.034772,-0.000435,0.002164,-0.006672,0.030869,0.035677,0.013571,-0.012130,-0.064989,-0.000160
CHG_Georgia_Kotias,0.089920,0.103584,-0.088246,-0.004845,-0.089555,0.020359,0.023266,-0.001154,-0.130486,-0.080002,-0.007632,0.024878,-0.052626,0.007707,0.026465,-0.029700,0.027772,-0.010769,-0.023757,0.037518,0.030945,-0.006677,0.005300,-0.020846,0.001557
Armenia_Aknashen_N,0.097888,0.128972,-0.080704,-0.058463,-0.046470,-0.008646,0.012456,-0.005538,-0.050108,-0.011481,0.005846,0.003447,-0.011893,-0.000550,-0.004886,0.015115,0.027381,0.004814,0.009302,-0.008754,0.011355,0.004575,-0.004190,0.001928,0.002754
Russia_Caucasus_Maikop,0.104717,0.121864,-0.048271,-0.017442,-0.047086,-0.005578,0.007520,-0.007384,-0.048472,-0.015672,0.007470,0.009142,-0.011298,-0.004266,0.001764,-0.008088,0.008084,0.003801,-0.006285,-0.005127,-0.003119,-0.000866,-0.018734,-0.019159,0.011137
Russia_Rostov_Eneolithic_Remontnoye,0.108132,0.097490,-0.022627,0.043282,-0.050471,0.018267,0.002938,-0.015807,-0.065243,-0.054033,0.006090,0.006294,-0.008920,-0.016445,0.026126,0.004508,0.001760,-0.002471,-0.004274,0.013444,-0.004180,0.007048,0.001171,0.002169,-0.004790
Stavropol_Eneolithic_Progress,0.114392,0.075657,0.016216,0.115796,-0.054317,0.049642,0.000705,-0.008308,-0.073526,-0.083829,0.007145,0.001649,-0.003568,-0.025047,0.031351,0.002718,-0.012386,-0.003357,-0.002200,0.006753,-0.006613,0.002968,0.009798,0.006145,-0.002874
Russia_Volgograd_Eneolithic_Berezhnovka,0.112685,0.071341,0.021590,0.118542,-0.060858,0.049085,0.000646,-0.015807,-0.073220,-0.083373,0.009743,-0.000712,-0.007247,-0.026183,0.037764,0.008519,-0.002640,-0.001267,-0.004180,0.010318,-0.001778,0.001144,0.013804,0.006025,-0.002155
Russia_Samara_Eneolithic_Khvalynsk,0.119876,0.053084,0.069047,0.157669,-0.036440,0.052736,-0.008930,-0.016426,-0.048240,-0.085808,0.012260,-0.005327,0.012656,-0.036107,0.031796,0.016031,-0.007230,-0.002868,-0.006062,0.010880,-0.005865,0.007369,0.011804,-0.002372,-0.005704
Russia_Samara_Eneolithic_Ekaterinovka,0.123000,0.045477,0.105382,0.187230,-0.018946,0.057948,-0.016561,-0.016528,-0.027668,-0.087764,0.016254,-0.011680,0.023451,-0.041596,0.027309,0.015766,-0.008829,-0.003603,-0.004007,0.012045,-0.005198,0.012033,0.011458,-0.008736,-0.006436
Sredni_Stog_avg_4sites,0.114297,0.083908,0.043164,0.115043,-0.026826,0.046935,-0.000137,-0.006586,-0.043828,-0.063046,0.004006,-0.001099,-0.001134,-0.014857,0.028049,0.012574,-0.005590,-0.004555,-0.007243,0.010667,0.006291,0.004349,0.004473,-0.017196,-0.003862
Ukraine_Eneolithic_Mykhailivka,0.113823,0.069056,0.031678,0.096254,-0.024928,0.048806,-0.001880,-0.000923,-0.057267,-0.060502,0.000162,0.010191,-0.012785,-0.003441,0.037052,0.024264,0.001695,0.010895,-0.012318,0.016758,-0.007487,-0.000618,0.015159,0.029763,0.007784
Russia_Samara_EBA_Yamnaya,0.125838,0.089254,0.042908,0.115456,-0.027868,0.044685,0.004491,-0.002949,-0.054858,-0.072996,0.001858,0.000350,-0.001652,-0.023610,0.037263,0.015734,0.000000,-0.001478,-0.001704,0.012506,-0.003120,0.001374,0.011229,0.018436,-0.004524
Ukraine_EBA_Yamnaya,0.120083,0.091398,0.047140,0.107721,-0.040469,0.040300,0.005640,0.007500,-0.057266,-0.078453,-0.001056,0.007268,0.005575,-0.016445,0.021376,0.008884,-0.008149,-0.000253,0.001320,0.009879,-0.006489,0.010387,0.006039,0.020304,-0.004550
Ukraine_Eneolithic_Trypillian_Verteba,0.124989,0.168578,0.033079,-0.045097,0.063353,-0.024675,-0.000369,0.002363,0.038012,0.059157,0.002397,0.006252,-0.013259,-0.000931,-0.024927,0.000196,0.016348,0.002968,0.008392,-0.003525,0.000802,0.006259,-0.008598,-0.015963,0.000108
Ukraine_Eneolithic_Usatove_Mayaky,0.118882,0.129198,0.012529,0.013315,0.006223,0.004059,0.000601,-0.001795,-0.013976,-0.002491,0.004403,0.003064,-0.008061,-0.011010,-0.000151,0.005804,0.010793,-0.000746,0.002388,0.000333,0.001248,0.002336,0.000890,-0.007444,-0.002488
Turkey_Mesopotamia_N_PPN_Cayonu,0.087644,0.154360,-0.036581,-0.095608,0.004001,-0.036814,-0.003525,-0.009230,0.006954,0.028976,0.013478,-0.002398,-0.006541,0.012248,-0.025787,-0.007823,0.004433,0.000760,0.005908,0.004377,0.002620,0.008779,-0.008258,-0.003615,-0.005269
Armenia_C_Areni,0.110693,0.133796,-0.033187,-0.027213,-0.028313,-0.007600,0.001821,-0.007038,-0.025770,-0.005922,0.007876,0.007456,-0.008660,-0.002236,-0.003834,-0.009812,-0.008638,0.002945,0.000691,-0.006722,-0.001716,0.004420,0.000893,-0.003675,0.002455
Turkey_OldHittitePeriod,0.101872,0.151314,-0.036769,-0.075582,0.000770,-0.024403,-0.005522,-0.005539,-0.006851,0.029249,0.007632,0.007343,-0.014495,0.002201,-0.014183,-0.004442,0.012582,-0.000950,0.009553,-0.011818,-0.004867,0.007543,-0.003882,0.003193,-0.005449
Turkey_Anatolia_MLBA_Kaman-Kalehoyuk,0.105002,0.151568,-0.042332,-0.082365,-0.004078,-0.027470,-0.002409,-0.007788,-0.011402,0.028429,0.009744,0.007006,-0.012079,0.003096,-0.013843,-0.004475,0.011669,-0.002154,0.008799,-0.007410,-0.003182,0.006183,-0.004807,0.003073,-0.001407
Turkey_N,0.117902,0.180087,0.003426,-0.101059,0.051240,-0.047969,-0.003799,-0.006846,0.036167,0.080678,0.008261,0.011309,-0.024164,0.000579,-0.042712,-0.010370,0.022556,0.001388,0.013649,-0.010448,-0.014261,0.005693,-0.004904,-0.003750,-0.004436
Luxembourg_Loschbour,0.130897,0.109677,0.203645,0.198000,0.162492,0.059125,0.015041,0.038075,0.100217,0.016219,-0.015427,-0.017235,0.019921,-0.001239,0.061346,0.070670,0.002608,0.007348,-0.008925,0.065406,0.117543,0.010387,-0.049422,-0.173639,0.019519

References

1. Ancient DNA Lazaridis I, Patterson N, Anthony D, Vyazov L, Fournier R, Ringbauer H, Olalde I, et al., Reich D (2025). The genetic origin of the Indo-Europeans. Nature 639: 132 to 142. doi.org/10.1038/s41586-024-08531-5
2. Ancient DNA Nikitin AG, Videiko M, Patterson N, Ivanova S, Dergachev V, Kotova N, Lillie M, Potekhina I, et al., Reich D (2025). A genomic history of the North Pontic Region from the Neolithic to the Bronze Age. Nature 639: 124 to 131. doi.org/10.1038/s41586-024-08372-2
3. Ancient DNA Haak W, Lazaridis I, Patterson N, Rohland N, Mallick S, Llamas B, et al. (2015). Massive migration from the steppe was a source for Indo-European languages in Europe. Nature 522: 207 to 211. doi.org/10.1038/nature14317
4. Ancient DNA Damgaard PB, Martiniano R, Kamm J, Moreno-Mayar JV, Kroonen G, et al. (2018). The first horse herders and the impact of early Bronze Age steppe expansions into Asia. Science 360: eaar7711. doi.org/10.1126/science.aar7711
5. Ancient DNA Lazaridis I, Alpaslan-Roodenberg S, Acar A, et al. (2022). The genetic history of the Southern Arc: a bridge between West Asia and Europe. Science 377: eabm4247. doi.org/10.1126/science.abm4247
6. Linguistics Heggarty P, Anderson C, Scarborough M, et al. (2023). Language trees with sampled ancestors support a hybrid model for the origin of Indo-European languages. Science 381: eabg0818. doi.org/10.1126/science.abg0818
7. Full text Lazaridis et al. (2025), author manuscript hosted by the Reich Lab, Harvard Medical School. reich.hms.harvard.edu
8. Full text Nikitin et al. (2025), repository copy, Borys Grinchenko Kyiv Metropolitan University. elibrary.kubg.edu.ua
9. Data Global25 PCA coordinates by Davidski, and the Moriopoulos 2026 collection of Global25 coordinates.