David Reich and his associates have published a paper containing a lot of new genetic data from prehistoric Europe. These data utterly refute the Kurgan hypothesis, and yet Reich and his associates are so stupid that they actually think the data support the hypothesis.
Marija Gimbutas’s Kurgan hypothesis claimed that the people of the Yamna culture spoke a language ancestral to all Indo-European languages around 3500 BC, and that the people of this culture then spread their language west through Europe and east into Asia over the next 3,000 years. The following maps from the Wikipedia article on Indo-European migrations illustrate how this was supposed to have happened:
But we now know that this is totally wrong. The new data show that of seven Yamna males, one belonged to Y haplogroup R1b1a-P297, one belonged to R1b1a2a-L23, and five belonged to R1b1a2a2-Z2103. None of them belonged to R1a, and it was R1a people who spoke a language ancestral to the satem Indo-European languages, which include the Balto-Slavic and Indo-Iranian languages. None of them belonged to R1b1a2a1-L51, and it was R1b-L51 people who spoke a language ancestral to the Germanic and Italo-Celtic languages. So we know that four huge branches of the Indo-European language family were not derived from the people of the Yamna culture.
The idea that the R1b1a2a1-L51 speakers of Germanic and Italo-Celtic languages could have been derived from the R1b1a2a-L23 minority among the Yamna people doesn’t work, because of the timing of the splits in the R1b tree. R1b1a2a2-Z2103 split from its brother clade R1b1a2a1-L51 8,000 years ago, which is around 3,000 years before the time of the Yamna samples. While the TMRCA (time to most recent common ancestor) for R1b-Z2103 is 7,400–5,600 years, the TMRCA for R1b-L51 is 7,600–5,900 years, so R1b-L51 is older than R1b-Z2103. R1b-L51 existed at the same time that the eastern R1b-Z2103 existed, in some other part of Europe. And of course that other part of Europe was Western Europe. There’s no doubt that R1b-L51 originated in Western Europe, because the highest frequencies of the R1b-L51* paragroup are found in France and Ireland.
I’ll now explain what really happened. First, the map below gives a more-or-less accurate representation of the state of things 8,000 years ago, which is right at the time of the split between R1b-L51 and R1b-Z2103. In the map, blue corresponds to the distribution of R1b, purple corresponds to the distribution of I, and pink corresponds to the distribution of R1a.
R1b1a2a-L23* would have been concentrated in Western Europe at this time, but it also would have been spreading east through the rest of Europe. We know that such an outward spread occurred, because R1b-L23* is found all over Europe. The R1b-L23 that stayed in the west would have given rise to R1b-L51, while the R1b-L23 that spread east would have given rise to R1b-Z2103. The R1b-L51 people in the west spoke a language ancestral to the Germanic and Italo-Celtic languages. R1b-Z2103 would have spread into Southeastern Europe, the Pontic-Caspian steppe, and the Middle East. In Southeastern Europe, the R1b-Z2103 people would become the speakers of the Thracian and Greek languages, while in the Middle East they would become the speakers of the Anatolian, Armenian, and Phrygian languages. The split between R1b-L51 and R1b-Z2103 at 8,000 years ago coincides with the estimated divergence of the Anatolian languages from the other Indo-European languages also at 8,000 years ago.
The split between the R1b1a2a1a1-U106 Germanic speakers and the R1b1a2a1a2-P312 Italo-Celtic speakers is estimated to have happened 6,100 years ago. This is well before the start of the Beaker culture 4,800 years ago, which suggests that R1b-U106 arose in Scandinavia from the earlier Megalithic culture. The R1b1a2a1a-L11 that stayed further south would have given rise to R1b-P312, which would then be spread around by the Beaker culture. We know from the new data that the people of the Beaker culture were R1b-P312, so they spoke a language ancestral to the Italo-Celtic languages.
The new data also confirm the association of R1a with the Corded Ware culture. The one Corded Ware male was R1a1a1-M417*, so he would have spoken a satem language that had not yet differentiated into either Balto-Slavic or Indo-Iranian.