Most anthropologists recognize that race is a social concept, not a biological one. That is, it stigmatizes some individuals as different and reinforces the privileges of others. There is no evidence that there are large groups of biologically distinct human beings (i.e. subspecies) that correspond to what people refer to when they talk about "race." Furthermore, to base any kind of biological category on a single physical characteristic, such as skin color (which, in itself is incredibly varied and determined by multiple genes), is clearly nonsense.
At the beginning of the 20th century, anthropologists questioned, and subsequently abandoned, the claim that biologically distinct races are isomorphic with (related to) distinct linguistic, cultural, and social groups. Then, the rise of population genetics led some mainstream evolutionary scientists in anthropology and biology to question the very validity of race as scientific concept describing an objectively real phenomenon. Those who came to reject the validity of the concept, race, did so for four reasons: empirical, definitional, the availability of alternative concepts, and ethical (Lieberman and Byrne 1993).
The validity of human races is a subject of much debate. The American Anthropological Association, drawing on biological research, states that "The concept of race is a social and cultural construction. . . . Race simply cannot be tested or proven scientifically," and that, "It is clear that human populations are not unambiguous, clearly demarcated, biologically distinct groups. The concept of 'race' has no validity . . . in the human species."
The first to challenge the concept of race on empirical grounds were anthropologists Franz Boas, who demonstrated phenotypic plasticity due to environmental factors (Boas 1912) and Ashley Montagu (1941, 1942), who relied on evidence from genetics. Zoologists Edward O. Wilson and W. Brown then challenged the concept from the perspective of general systematics, and further rejected the claim that "races" were equivalent to "subspecies" (Wilson and Brown 1953). Claude Lévi-Strauss's Race and History (UNESCO, 1952) enforced this cultural relativist thesis, by the famous metaphor of cultures as trains crossing each other in different directions, thus each one seeing the others as immobile while they themselves are progressing.
One of the crucial innovations in reconceptualizing genotypic and phenotypic variation was anthropologist C. Loring Brace's observation that such variations, insofar as they are affected by natural selection, migration, or genetic drift, are distributed along geographic gradations called "clines" . This point called attention to a problem common to phenotypic-based descriptions of races (for example, those based on hair texture and skin color): they ignore a host of other similarities and difference (for example, blood type) that do not correlate highly with the markers for race. Thus, anthropologist Frank Livingstone's conclusion that, since clines cross racial boundaries, "there are no races, only clines" (Livingstone 1962: 279). In 1964, biologists Paul Ehrlich and Holm pointed out cases where two or more clines are distributed discordantly—for example, melanin is distributed in a decreasing pattern from the equator north and south; frequencies for the haplotype for beta-S hemoglobin, on the other hand, radiate out of specific geographical points in Africa (Ehrlich and Holm 1964). As anthropologists Leonard Lieberman and Fatimah Linda Jackson observe, "Discordant patterns of heterogeneity falsify any description of a population as if it were genotypically or even phenotypically homogeneous" (Lieverman and Jackson 1995).
Finally, geneticist Richard Lewontin, observing that 85 percent of human variation occurs within populations, and not between populations, argued that neither "race" nor "subspecies" was an appropriate or useful way to describe populations (Lewontin 1973). This view is described by its opponents as Lewontin's Fallacy. Some researchers report the variation between racial groups (measured by Sewall Wright's population structure statistic FST) accounts for as little as 5% of human genetic variation2. However, because of technical limitations of FST, many geneticists now believe that low FST values do not invalidate the suggestion that there might be different human races (Edwards, 2003). Meanwhile, neo-Marxists such as David Harvey (1982, 1984, 1992) believe that race is a social construct that in reality does not exist, used instead to extenuate class differences.
These empirical challenges to the concept of race forced evolutionary sciences to reconsider their definition of race. Mid-century, anthropologist William Boyd defined race as:
"A population which differs significantly from other populations in regard to the frequency of one or more of the genes it possesses. It is an arbitrary matter which, and how many, gene loci we choose to consider as a significant "constellation" (Boyd 1950).
In the face of these issues, some evolutionary scientists have simply abandoned the concept of race in favor of "population." What distinguishes population from previous groupings of humans by race is that it refers to a breeding population (essential to genetic calculations) and not to a biological taxon. Other evolutionary scientists have abandoned the concept of race in favor of cline? (meaning, how the frequency of a trait changes along a geographic gradient). The concepts of population and cline are not, however, mutually exclusive and both are used by many evolutionary scientists.
Human Genetic Diversity: Lewontin's Fallacy is an academic paper published in 2003 by A.W.F. Edwards that attacks the conclusion of Richard Lewontin that race is an invalid taxonomic construct.
Lewontin, in a 1972 paper, 'The apportionment of human diversity' and again in a 1974 book, The Genetic Basis of Evolutionary Change, argues that because the probability of racial misclassification of an individual based on variation in a single genetic locus is approximately 30% that race is an invalid taxonomic construct.
Edwards argues that when one takes into account more loci, the probability of racial misclassification rapidly approaches 0%. Edwards argues that the information which distinguishes races is "hidden in the correlation structure of the data."
Edwards argues that both ordination and cluster analyses can reveal the correlation structure of multilocus data.
A caricature of Lewontin's argument is that because humans share 50% of their DNA with carrots, we must be 50% the same