The molecular genetics of European ancestry

Abstract

husband and wife team, Ludwik and Hanna Hirschfeld, published a paper in the Lancet entitled `Serological di¡erences between the blood of di¡erent races ö the results of research on the Macedonian front' (Hirschfeld & Hirschfeld 1919). This was a survey of the frequencies among the Allied soldiers of blood groups A and B, whose Mendelian credentials had by then been established. The Hirschfelds noticed that the blood group frequencies were quite di¡erent among groups of soldiers from di¡erent countries (¢gure 1). For obvious military reasons they did not have access to the ¢gures for Germans and relied on their memories from before the war for these values. It is reassuring to note that even this very ¢rst paper in the ¢eld was not short on speculation, a trend which continues to this day. In their view, humans were divided into two biochemical races, A and B, with di¡erent origins. Although they were uncertain about the origin of race A, the high frequency of blood group B among soldiers from the subcontinent convinced them that `we should look to India for the cradle of one part of humanity. Both to Indo-China in the East and to the West, a broad stream of Indians passed out, ever lessening in its £ow, which ¢nally penetrated Western Europe'. I have drawn a diagram from their data (¢gure 2) using the same techniques as used nowadays. This is a diagram of the genetic di¡erences between the populations. How might these be interpreted? Some relationships look perfectly reasonable. Italians and French are close, with Germans a little further away. But there are some unexpected features. For example, Russia and Madagascar share the same position on the diagram. Is this the evidence of a long-forgotten Russian invasion of Madagascar, or vice versa? Indians are on a separate branch from everyone else and `Negroes', actually from Senegal, are almost as similar to Arabs as English are to Greeks. How are we to explain these bizarre comparisons? One reason for them is that only a single locus is being considered. To counteract this shortcoming, Cavalli-Sforza and Edwards developed a statistical method that was able to compute the accumulated allele frequency data from several loci (Cavalli-Sforza & Edwards 1967). One