American Philosophical Society collection
The early 1970s saw two papers move the goalposts on race, first in academia and then throughout society. One was by Walter Bodmer and L.L. Cavalli-Sforza. The other was by a third geneticist, Richard Lewontin.
Bodmer and Cavalli-Sforza (1970) conceded that human races exist while denying that they differ statistically in intellectual capacity, at least on the basis of current evidence. Lewontin (1972) took a different tact: human races don’t exist. Period. No races, no race differences.
He came to this conclusion after analyzing the way various genes vary among human populations, specifically genes whose variants (‘alleles’) produce different blood groups, serum proteins, or red blood cell enzymes. Surprisingly, only 6.3% of this variation was accounted for by large continental races (i.e., ‘Caucasoids’, ‘Mongoloids’, and ‘Negroids’). Another 8.3% was accounted for by sub-racial populations. The rest—over 85% of human genetic variation—existed only among individuals of the same population.
This pattern had been known for some time with respect to blood groups. But researchers had assumed that some kind of balanced polymorphism was inflating within-population variation, perhaps one that hinders the spread of contagious diseases (1). By the early 1970s, however, the same pattern was appearing with other ‘structural’ proteins. The building blocks of flesh and blood were turning out to be remarkably the same in all humans. As Lewontin concluded:
It is clear that our perception of relatively large differences between human races and subgroups, as compared to the variation within these groups, is indeed a biased perception and that, based on randomly chosen genetic differences, human races and populations are remarkably similar to teach other, with the largest part by far of human variation being accounted for by the differences between individuals.
Human racial classification is of no social value and is positively destructive of social and human relations. Since such racial classification is now seen to be of virtually no genetic or taxonomic significance either, no justification can be offered for its
(Lewontin 1972, p. 397)
How did Cavalli-Sforza react? According to a recent interview, he saw this paper as a turning point in human genetics:
The between-population genetic variation observed with 650,000 SNPs on the 52 populations of the HGDP is 11% (Li et al. 2008) with a very small standard error. It becomes 16% for the X chromosome, as is expected if nearly all the genetic variation is due to drift—that is, the role of natural selection is very limited. The ca. 30-year-old estimate by Lewontin (1972) of this quantity (15%) was based on other markers and populations and was a reason to encourage banning the use of the word race in humans. In any case the new value is even more supportive of dropping the word race.
Yet back in the 1970s, and even long after, his reaction was …. silence. From 1972 to 1989, Google Scholar lists 426 publications for which Cavalli-Sforza was an author or co-author (2). None of them cited Lewontin’s 1972 paper. He first commented on it in 1993, some twenty years after the fact (3).
Why the two-decade silence? If Lewontin’s paper had been so important in human genetics, why did Cavalli-Sforza take so long to acknowledge this importance?
It is hard to enter a silent person’s mind. A better approach would be to ask whether any thinking person had reasons for rejecting Lewontin’s finding, or rather the way he spun it. Such reasons fall under three headings:
1. A small genetic difference can make a big cultural difference
Even if human populations differ only slightly in certain genetic predispositions, these slight differences can have big effects.
For instance, the historical economist Gregory Clark has argued that the slow but steady demographic expansion of the English middle class from the 12th century onward gradually raised the population mean for predispositions to non-violence, deferment of pleasure, and other future-oriented behavior. Although the embryonic middle class was initially a small minority in medieval England, its descendants grew in number and gradually replaced the lower class through downward mobility. By the 1800s, its lineages accounted for most of the English population (Clark 2007, pp. 124-129, 182-183; Clark 2009).
The 1800s also saw the triumph of Victorian morality in England. This triumph was due not to a massive change in the gene pool, but rather to a slow incremental change that had finally reached a critical mass. The English middle class could now impose its behavioral norms on the whole population, thereby abandoning the ‘two-tier morality’ of other class-stratified societies.
2. Lewontin’s finding is true only if you look at one gene at a time
Genes vary much more within than between human populations only if we take one gene at a time. This pattern reverses if we aggregate variation at several gene loci. The more we aggregate, the more this genetic variation will exist between populations and not within them.
This fact was known to Cavalli-Sforza back in 1966 when he was constructing his first phylogenies of human populations: “it is desired that the number of genes considered be as high as possible in order to increase the reliability of the conclusions.” (Cavalli-Sforza 1966). When he and another colleague later aggregated data from 75 gene loci of 144 individuals belonging to 12 human groups in Africa, Asia, Europe, and Oceania, he found very little genetic overlap among the groups. Most individuals clustered with other members of their regional group (Mountain & Cavalli-Sforza 1997). This point has also been made by Mitton (1977, 1978), Edwards (2003), and Sesardic (2010).
Clearly, two groups are easier to tell apart with several criteria than with one. With enough criteria, any overlap will shrink to zero and all individuals can be unambiguously assigned to either group. This is basic logic. But all this proves is that human populations are identifiable. It doesn’t prove that the differences between them are greater than the differences within them.
3. The way a gene varies within and between populations will itself vary as a function of the gene’s selective value
When genes vary between populations, they do so usually because the population boundary separates different environments with different sets of selection pressures. Genes that differ across this boundary are necessarily genes that make a difference, i.e., that have high selective value.
In contrast, selective value is necessarily low for genes that differ within a population despite similar selection pressures (unless the different variants form a balanced polymorphism).
The two kinds of genetic variation are therefore not comparable.
And this leads to another problem. Yes, we have a lot of data on the way genes differ between populations, but that data comes largely from genes with little or no selective value—the ones that are most likely to differ within populations! When population geneticists look for a gene worth studying, they tend to choose one that responds weakly to natural selection. This choice is deliberate. They want the gene to be as close to selective neutrality as possible so that it changes at a predictable rate (i.e., only through random mutations). It thus provides a reliable ‘clock’ of population history.
Population geneticists also prefer to study genes whose protein products are easy to find and measure in body tissues. Such ‘structural proteins’ are similar in different species or even different genera. Humans and chimps, for instance, look very much alike if we compare the protein building blocks of their body tissues. These two species have diverged from each other largely through evolutionary changes at a higher level, particularly regulatory genes that control the pace and timing of development.
This point was grasped by Stephen J. Gould (1977, 406). He explained how such genes provide a misleading picture of genetic variation:
The most important event in evolutionary biology during the past decade has been the development of electrophoretic techniques for the routine measurement of genetic variation in natural populations. Yet this imposing edifice of new data and interpretation rests upon the shaky foundation of its concentration on structural genes alone (faute de mieux, to be sure; it is notoriously difficult to measure differences in genes that vary only in the timing and amount of their products in ontogeny, while genes that code for stable proteins are easily assessed).
I remember telling a geneticist that Lewontin’s finding applied only to genes with low selective value. He laconically replied that there was no evidence that things would be any different for genes with higher selective value.
Actually, there is real-world evidence. The same genetic overlap that Lewontin found between human populations also occurs between many species that are nonetheless distinct in anatomy, physiology, and behavior (see previous post).
And Cavalli-Sforza in all this?
He was certainly aware that culture can amplify slight genetic differences. This was, in fact, part of his dual transmission theory—now known as gene-culture co-evolution (Stone & Lurquin 2005, p. 104-108).
He had also been aware since the mid-1960s that the genetic overlap among human populations is a function of the number of genes under consideration. In addition to his 1997 article with Joanna Mountain, this principle has been implicit in most of his work on human populations.
When questioned directly on this subject, with reference to Edwards’ reply to Lewontin (Edwards 2003; Khan 2006a), he diplomatically answered that both were right:
Edwards and Lewontin are both right. Lewontin said that the between populations fraction of variance is very small in humans, and this is true, as it should be on the basis of present knowledge from archeology and genetics alike, that the human species is very young. It has in fact been shown later that it is one of the smallest among mammals. Lewontin probably hoped, for political reasons, that it is TRIVIALLY small, and he has never shown to my knowledge any interest for evolutionary trees, at least of humans, so he did not care about their reconstruction. In essence, Edwards has objected that it is NOT trivially small, because it is enough for reconstructing the tree of human evolution, as we did, and he is obviously right.
What about the third counter-argument? Was Cavalli-Sforza aware that genetic variation within populations is not comparable to genetic variation between populations? We see some awareness in his 1971 textbook, where he argues that most polymorphic genes have little selective advantage. Only in two cases are they subject to strong selection pressures. One case involves balanced polymorphisms. The other involves “transient polymorphisms”—genes quickly moving to fixation in those populations where they are advantageous (Cavalli-Sforza & Bodmer 1971, pp. 732-735). Such genes are thus more likely to vary between than within populations.
So perhaps he was aware. Or perhaps not. Even less clear is what he was thinking during his long silence on Lewontin’s 1972 paper. This paper was, after all, in Cavalli-Sforza’s own field of study. It was also widely commented on by other human geneticists. So why the silence?
One reason was his tenuous professorship at Stanford. It was this position that had propelled him to academic stardom, and he may have decided that discretion is the better part of valor. His pragmatism is recounted by a former colleague, Anthony Edwards:
When in the 1960s I started working on the problem of reconstructing the course of human evolution from data on the frequencies of blood-group genes my colleague Luca Cavalli-Sforza and I sometimes unconsciously used the word ‘race’ interchangeably with ‘population’ in our publications. In one popular account, I wrote naturally of ‘the present races of man’. Quite recently I quoted the passage in an Italian publication, so it needed translating. Sensitive to the modern misgivings over the use of the word ‘race’, Cavalli-Sforza suggested I change it to ‘population’. At first I was reluctant to do so on the grounds that quotations should be accurate and not altered to meet contemporary sensibilities. But he pointed out that, as the original author, I was the only person who could possibly object.
And the others in all this?
Lewontin’s paper met with either enthusiasm or silence. Two attempts at rebuttal were published in 1977 and 1978 by Jeffrey Mitton, a zoologist at a second-tier university. Another one was made much later, in 2003, by Anthony Edwards, a geneticist who no longer held an academic position. All three papers used the second counter-argument, i.e., within-population diversity exceeds between-population diversity only if you consider one gene at a time. Although I know several zoologists who are aware of the third counter-argument, none of them has ever written it up for publication.
Why did Lewontin’s paper meet with so little opposition? First, there was the wave of attacks on “racist” professors during the early 1970s, and the chill that subsequently spread through academic life. Many felt it best to be prudent. Second, there was the tenure-track system, which compelled untenured professors to ingratiate themselves with key members of academia. This system had always existed but was now being manipulated to advance an ideological agenda.
Thus began the soft totalitarianism of the late 20th century, not with a bang but with a whimper—or rather a silent acquiescence.
1. If your surface proteins differ from your neighbors’, you are less likely to be infected by contagious pathogens. There is thus selection for variability in surface proteins within each population.
2. In some cases, several references actually refer to a single paper (because of errors or variations in transcription). This overcount would not lead to an undercount of references to Lewontin’s 1972 paper.
3. He first cited Lewontin’s 1972 paper in 1990. His first substantive comments came three years later, when he cited it to show that genetic variation occurs mainly within human populations whereas cultural variation occurs mainly between them (Cavalli-Sforza 1993). This seems to be the only one of his publications that discusses the implications of Lewontin’s 1972 paper. All in all, he cited it three times in the 1990s and once in the 2000s.
Bodmer, W.F. and L.L. Cavalli-Sforza. (1970). Intelligence and race, Scientific American, 223(4), 19-29.
Cavalli-Sforza, L.L. (1966). Population Structure and Human Evolution, Proceedings of the Royal Society of London. Series B, Biological Sciences, 164, 362-379.
Cavalli-Sforza, L.L. (1993). “How are values transmitted?” in M. Hechter, L. Nadel, and R.E. Michod (eds), The Origin of Values, New York: Aldine de Gruyter, pp. 305-317.
Cavalli-Sforza, L.L. and W.F. Bodmer. (1971). The Genetics of Human Populations, San Francisco: W.H. Freeman and Co.
Cavalli-Sforza, L.L. and F. Cavalli-Sforza (2008). La génétique des populations : histoire d'une découverte, Paris: Odile Jacob. (translation of Perché la scienza : L’aventura di un ricercatore).
Clark, G. (2007). A Farewell to Alms. A Brief Economic History of the World, Princeton University Press, Princeton and Oxford.
Clark, G. (n.d.). The indicted and the wealthy: surnames, reproductive success, genetic selection and social class in pre-industrial England,
Edwards, A.W.F. (2003). Human genetic diversity: Lewontin’s fallacy. BioEssays, 25, 798-801.
Gould, S.J. (1977). Ontogeny and Phylogeny. Belknap Press: Cambridge (Mass.)
Khan, R. (2006a). 10 questions for A.W.F. Edwards, Gene Expression, August 28, 2006.
Khan, R. (2006b). 10 questions for Luigi Luca Cavalli-Sforza, Gene Expression, August 24, 2006.
Lewontin, R. (1972). The apportionment of human diversity, Evolutionary Biology, 6, 381-398.
Manni, F. (2010). Interview with Luigi Luca Cavalli-Sforza: Past research and directions for future investigations in human population genetics, Human Biology, 82, 245–266.
Mitton, J.B. (1977). Genetic differentiation of races of man as judged by single-locus and multilocus analyses, American Naturalist, 111, 203-212.
Mitton, J.B. (1978). Measurement of differentiation: reply to Lewontin, Powell, and Taylor, American Naturalist, 112, 1142-1144.
Mountain, J.L. and L.L. Cavalli-Sforza (1997). Multilocus genotypes, a tree of individuals, and human evolutionary history, American Journal of Human Genetics, 61, 705-718.
Sesardic, N. (2010). Race: a social destruction of a biological concept, Biology and Philosophy, 25, 143-162.
Stone, L. and P.F. Lurquin. (2005). A Genetic and Cultural Odyssey. The Life and Work of L. Luca Cavalli-Sforza. New York: Columbia University Press.