What, if Anything, is a Race?

Is the human population subdivided into races? The answer depends on one’s understanding of this term and people speaking of race should therefore first specify what exactly the word means to them. The original division of the human species into races was based on visible, conspicuous differences among individuals. European travelers to Africa noted that people on this continent tended to have dark brown or black skin, kinky black hair, brown eyes, a broad nose, and thick lips. Similarly, explorers returning from Central and East Asia reported that the inhabitants of this part of the world had yellowish skin, coarse, straight hair, and a fold of skin extending from the eyelids across the inner corner of the eye (a feature called the epicanthic fold). These characteristics contrasted with those of the Europeans who tended to have fair skin, light-colored hair, either straight or wavy, blue eyes, a narrow nose, and fairly thin lips. These three types came to be called the Negroid, Mongoloid, and Caucasoid races, but we shall refer to them as Africans, Asians, and Europeans, respectively. Later, the list of races was extended to include peoples of other continents, subcontinents, and island groups, specifically American Indians, Australians, Indians, and Oceanians (Melanesians, Micronesians, Polynesians). Closer study, however, revealed a considerable variation within the major races in the characters used initially for their delineation. Thus, for example, 36 shades of skin pigmentation have been described, ranging from jet black to almost white. Using these, as well as other characters, additional races - up to several hundred - were described and anthropologists were neither able to agree on the most suitable classification system, nor on how detailed the subdivision should be.

As the number of races grew, so did the skepticism. The large number of races, the existence of intermediates, and the hierarchical grouping of clusters of individuals suggested to some anthropologists that the characters on which the classification was based varied continually and hence that any partitioning into discrete groups was arbitrary and without biological significance. It appeared that any two groups delineated by one or more physical characters could be linked up by a series of intermediates providing a smooth, uninterrupted transition from one group to the other. The continuous line seemed to stretch from family units all the way to the global population. Moreover, the classification based on one visible character often did not match that based on other characters. From all these observations, some anthropologists concluded that it was an exercise in futility to classify humans into races and that it would be best to abandon the whole concept of human race. This conclusion also appeared substantiated by the misuse and abuse of the concept in some countries for the purpose of discrimination, oppression, and even genocide.

The proposal to scrap the concept of race altogether is currently only one extreme in a range of views. It is certainly not shared by all anthropologists and is by no means the majority opinion of the public at large. It appears to be a conclusion reached more on the basis of political and philosophical creeds than on scientific arguments. Correspondingly, anthropologists who do hold this opinion often attempt to shout down their opponents rather than convince them by presentation of facts. Their favored method of argumentation is to label anybody who disagrees with them as racist. The public, however, seems unimpressed by their rhetoric. It refuses to believe that the differences they see are a mere figment of their imagination. A lay person can tell with a high degree of accuracy where individuals come from just by glimpsing their features…

Except for some anthropologists, everybody else seems to be able to distinguish people from different parts of the world at a glance by their outward appearance. This, apparently, is also the view of some governmental administrators in countries with programs designed to fight racial discrimination. Obviously, there is a credibility gap on this issue between some anthropologists on one side and the public, as well as the governments of some countries, on the other.

One way to settle the arguments among anthropologists and to reconcile anthropologists with the public might be to move away from physical characters and focus on the genes. If races are real, they should have a genetic basis separable from environmental and cultural influences. Genetics might help to resolve the issue in two ways. First, the physical characters used in the classification of races are controlled by genes which could therefore be used to determine whether there is discreteness at the genotypic level where there appears to be continuity at the phenotypic level. The differences should then reduce to the presence or absence of genes responsible for the particular character and so provide a quantitative measure of racial differentiation. Unfortunately, none of the genes controlling skin color, hair color and texture, or lip and nose shape have been identified. These characters are determined by multiple, interacting genes, so their identification is not easy. But in the near future, the genes will undoubtedly become known and it will then be possible to establish whether there is a correspondence between their distribution and any of the classification schemes that anthropologists have designed for the human species…

The second way in which genetics may contribute to the resolution of the question is by the analysis of genes that are not involved in the control of the physical differences between races. In the preceding chapters we described several examples of genetic variation within the human species. In addition to the systems mentioned there, numerous other variable systems are known, probably none of which has anything to do with human morphological diversity. Nevertheless, all these systems can be used to test the hypothesis of genetic racial differentiation. Provided that the races separated a long time ago, random genetic drift should have diversified their genetic composition even in the absence of selection. It can be expected that the longer ago the races diverged, the greater the differences between them will be. Even if there has not been enough time to “fix” different alleles in distinct races, at least differences in gene frequencies should have been generated. If, on the other hand, races are classification artifacts or totally arbitrary categories, this should be reflected in the pattern of genetic variation in the human species. An ideal way of testing these assumptions would be to determine the genetic variation of the entire human species and check whether it sorts out into groups that correspond to the morphologically or geographically defined races. Since, however, the testing involves gene frequencies and the latter can only be obtained by comparing individuals from predefined groups, this approach is not feasible. Instead, the starting point of the genetic test must be the groups defined by anthropologists as races. If the comparison of genetic variation within each race with that between races reveals no significant differences, the whole concept of the human species being genetically subdivided into groups would be seriously undermined.

As far as we can tell, the first geneticist to measure the genetic differences between human races was L.L. Cavalli-Sforza. In a study published in 1966, he summarized genetic distance data for 12 loci (most of them coding for blood group antigens) in seven major human populations and reached the conclusion that of the total genetic variation observed in the human species, less than 15 percent accounts for differences between races. Cavalli-Sforza apparently found this observation unremarkable, or at least unworthy of a comment. The geneticist who did make a big story out of it was Richard C. Lewontin in 1972.  He divided the human species into seven races (Caucasians, Black Africans, Mongoloids, South Asian Aborigines, Amerinds, Oceanians, and Australian Aborigines), and each race into a series of populations (Caucasians into Arabs, Armenians, Basques, Bulgarians, Czechs, etc; Black Africans into Abyssinians, Bantu, Burundi, Batutsi, etc; Mongoloids into Ainu, Bhutanese, Bruneians, Buriats, Chinese, etc; etc) delineated by morphological, linguistic, historical, and cultural information. He then compiled gene frequency data for nine blood group systems, four serum proteins, and four soluble enzymes obtained by typing of the individual populations by immunological (antibody-based) and protein electrophoresis methods…

The assessment consists of comparing genes sampled separately from individual populations (races) with those sampled randomly from the whole species. The main consequence of the subdivision is that individuals no longer mate randomly with each other within the species; they now tend to mate within each population. As a result, individuals within each population become increasingly more similar to one another genetically but more and more different from individuals of other populations, and the proportion of heterozygotes in the populations decreases through extinction of alleles. The probability that two genes sampled at random will be different is lower if the genes are taken from the same population than if they are taken at random from the entire population comprising the species. Working against these expectations is gene flow, the exchange of genes between populations. Gene flow increases heterozygosity within populations and decreases it between populations by spreading alleles through the entire species and so homogenizing the species genetically.

Lewontin’s study revealed that 85.4 percent of the genetic variation that exists within the human species is contained within the individual populations and that the remaining 14.6 percent is accounted for by differences between human groups. Of the 14.6 percent, 8.3 percent accounts for differences between populations, as defined by Lewontin, and 6.3 percent for differences between races. Of course, since Lewontin’s distinction between populations and races is arbitrary, some anthropologists might want to call some or all of his “populations” races.

Lewontin’s analysis was later repeated by several investigators using ever larger compilations of gene frequencies and different methods of assessing genetic variation within and between groups. Most of the measures of variation were related to the Fst statistic which was introduced by Sewall Wright to determine the effect of fragmentation of a population on its genetic make-up. Wright called the measure fixation index (hence the letter F) because it gauges the increase in fixation of alleles within a subpopulation as a result of the fragmentation… Earlier we mentioned that genetic variation is influenced by two opposing processes: random genetic drift within a subpopulation and gene flow between subpopulations. When an equilibrium is reached between the processes, Fst measures the number of migrants between the subpopulations.

Regardless of the method used in the analyses, all investigators reached estimates very close to that obtained by Lewontin: the differences observed between the subdivisions (populations, groups of populations, races) represented 10 to 15 percent of the total genetic variation found in the human species. Formally, these findings demonstrate, first, that the species is indeed subdivided into genetically definable groups of individuals and, second, that at least some of these groups correspond to those defined by anthropologists as races on the basis of physical characters. They do not, however, settle the arguments regarding the merits of the racial classification. Unfortunately, Lewontin did not specify before initiating his analysis how large the difference has to be in order to call the groups “races”. Consequently, the results of the studies have led population geneticists to two diametrically opposite conclusions. Lewontin called the observed differences trivial and proclaimed that “racial classification is now seen to be of no genetic or taxonomic significance” so that “no justification can be offered for its continuance.” This view is echoed by most authors of similar studies, who seem to be surprised that genetic variation within populations is greater than that between them. By contrast, Sewall Wright, who can hardly be taken for a dilettante in questions of population genetics, has stated emphatically that if differences of this magnitude were observed in any other species, the groups they distinguish would be called subspecies.

One can extend Wright’s argument even further. The more than 200 species of haplochromine fishes in Lake Victoria differ from each other much less than the human races in their neutral genes, although they are presumably distinguished by genes that control differences in their external appearances. The same can be said about at least some of the currently recognized species of Darwin’s finches and about other examples of recent adaptive radiations. In all these cases, reproductively isolated groups are impossible to tell apart by the methods used to measure differences between human races. Obviously, human races are not reproductively isolated (interracial marriages are common and the progenies of such marriages are fully fertile) but the external differences between them are comparable to those between the cichlid fishes and Darwin’s finches. Under these circumstances, to claim that the genetic differences between the human races are trivial is more a political statement than a scientific argument. Trivial by what criterion? How much difference would Lewontin and those who side with him consider nontrivial?

By mixing science with politics, geneticists and anthropologists are committing the same infraction of which they are accusing other scientists, whom they themselves label as racist. Even worse, by dismissing the genetic differences as insignificant, they play into the hands of genuine racists who can easily demolish this claim and so further their own agenda. It is intellectually more honest to acknowledge the differences and then point out that they by no means imply supremacy of one race over others. This can be done by demonstrating that the differences are in genes that cannot be linked to any features that would be required for the preeminence of a particular race.

As for the notion of continuity in genetic variation across the entire human species, it must not be confused with the hierarchical structure of the variation. There are indeed different levels of grouping within H. sapiens, which are the result of population history, genealogical patterns, geography, cultural differentiation, and other factors. Smaller groups can be clustered into larger ones and these into larger ones still, as is apparent from any phylogenetic tree drawn for human populations. The differentiation and clustering is more pronounced for populations inhabiting Siberia, the Amazon basin, or other such areas in which the traditional organization of human societies has not yet disappeared completely. It is far less apparent in areas of large population expansion and mixing, in which the frequencies of some genes can be shown to form clines - relatively smooth gradients decreasing from a center of distribution to the periphery. But even in Europe, with its long tradition of intermarriages and easy opportunities for mixing, Guido Barbujani and Robert R. Sokal could uncover a patchy distribution of allete frequencies and zones of sharp changes in genetic variation, and attribute them to physical and cultural barriers to gene flow. The hierarchical nature of the groupings, of course, begs for an answer to the question: which of the groups should be called races? It could be: all, none, or any, according to one’s preference. The name is not important. What is important is to acknowledge the existence of differentiation and its significance for the reconstruction of human history.

A related question is how old the differentiation may be. Here the answer depends on whether one is a multiregionalist or a uniregionalist and if the latter, to what version of the uniregional hypothesis one subscribes. Multiregionalists have no problem in explaining the 10 to 15 percent difference between the human groups. Since they assume that the differentiation began up to 2 my ago, when H. erectus established founding populations in the different regions, there has been sufficient time to accumulate the differences. Uniregionalists, who assume that the differentiation into groups began after the exodus of H. sapiens from Africa, are at a disadvantage, because calculations indicate that only under highly unrealistic assumptions (e.g., no gene flow between populations) would the time interval suffice for the origin of the observed differences. Henry C. Harpending and his colleagues therefore proposed in 1993 that the ancestral human population was already subdivided before its expansion from Africa so that the race differences are older than the date of the exodus. An alternative considered by Harpending and Alan R. Rogers is the possible colonization of new environments after the exodus took place by small bands which then expanded locally. The small sizes of the founding populations could account for the local losses of neutral genetic variation and thus for the relatively large differences between the populations.

- Jan Klein and Naoyuki Takahata, 2002.  Where Do We Come From?  The Molecular Evidence for Human Descent.  Springer, New York.  pp. 382-91 (excerpts).   You should buy this book.