Stalking the Wild Taboo -Edward M. Miller- Tracing the Genetic History of Modern Man

Tracing the Genetic History of Modern Man

Cavalli-Sforza, L. L., Menozzi, P., & Piazza, A.
The History and Geography of Human Genes (Princeton: Princeton University Press, 1994)

Reviewed by
Edward M. Miller
Professor of Economics and Finance
University of New Orleans
New Orleans, LA 70148

Mankind Quarterly, Vol. 35 (Winter 1994) No. 1-2, 71-108. Posted with permission of Mankind Quarterly, Institute for the Study of Man, 1133 13th St., N. W., Suite C-2, 20005 (Telephone 202-371-2700).


This massive compilation of genetic data on the populations of the world, by documenting the genetic similarities and differences, shows that "races" exist while simultaneously denying the usefulness of the concept. In the course of doing this the book present much useful information about similarities and differences in gene frequencies among the populations of the world. The data is presented in many useful formats including tables, maps, dendograms (descent trees), and principal component diagrams. The interpretation generally presumes neutrality for the various genes, and many interesting conclusions are drawn about the evolutionary history of various populations around the world.

Table of Contents

The Tables of Genetic Distances 3
Trees of Human Descent 5
The African versus all Other Split is Primary 6 Other Splits in the Tree 9
Other Interesting Findings on the World Wide Gene Distributions 17 Conclusions from Principal Components 17 Race 22
Diseases and Gene Frequencies 24
The Regional Chapters 27
Asia 28
Europe 29
Africa 30
The Americas 34
Conclusions 35

Cavalli-Sforza, Menozzi, & Piazza's (1994) new The History and Geography of Human Genes is a very impressive compilation of what is known about the geography and history of human genes. It will be a definitive work of racial analyses (although the authors would not describe it this way). About half of the book (the back half) is an atlas showing of the distribution of a large number of genes for each of the continents, and for the whole world. The extensive atlas section is probably what makes the book so expensive ($175 as advertised), and will unfortunately limit its purchase to libraries, and a few individuals working in the field, who will feel it is an indispensable. It might have been better if the publishers had brought out two books, one the atlas, and the other the text. This would have made the text more compact, and made the book less unwieldy to handle.

There is also an extensive set of tables giving information on allele frequencies for many genes and populations. This data base is compiled from examining 2900 articles from 136 journals (although only 777 involved unduplicated data, listed in the references). It is the mapping and interpretation of this massive amount of data that makes the book so impressive and valuable. Unfortunately, the data compilation is already somewhat obsolete as it only goes to 1986 (p. 25).

The book is organized on a geographical basis, with one introductory chapter, one chapter on the world wide distribution of genes, and then separate chapters on Africa, Asia, Europe, 2 (Australia, New Guinea, and the Pacific Islands), and the Americas. The introductory chapter is a valuable compilation of material about genes, anthropology, archeology, human evolution, and the methods of quantitative genetics. The authors recognize that the book will be used by people who are not experts in all these fields. The introductory material is useful to those lacking training in one or more of these fields. However, the specialist will probably learn little from the sections on his own specialty. A brief section on the "Scientific Failure of the Concept of Human Races" attempts to divert criticism for even studying the subject of how gene frequencies differ across the world (although it probably does reflect the author's true beliefs). Much of the book seems to contradict the anti-race assertions in this section, but it is an effective argument against the most naive ideas of race.

The last half of the first chapter contains useful discussions of such subjects as the problem of identifying populations, and some of the methods that will be used in the rest of the book. Greater detail, and an effort to put the methods into simpler language would have been useful for the general reader.

The second chapter summarizes the data on the world wide distribution of genes. The basic theoretical framework in this book is that gene frequencies are determined by drift. Offspring randomly inherit genes from their parents. The child has a 50% chance of inheriting any given gene from each parent. Many who are not used to thinking in genetic terms think that any trait affected by the genes must be rigidly inherited from the parents. In actuality, because of the random inheritance of genes, genetics provides a theory of human diversity, and helps explain why siblings are usually quite different (Rowe 1994). In small and somewhat isolated populations, such as humans are believed to have lived in during prehistoric times, gene frequencies change appreciably, but randomly, from generation to generation. Over many generations (perhaps 100,000 years) different populations develop different gene frequencies.

As mentioned, one contribution of the book is the extensive table of gene frequencies. The world wide section is based on data for 120 alleles from 42 populations, (although data was not available for all alleles for all populations). A certain amount of averaging of data from different sources was needed to get the relevant gene frequencies, and the data is often for national groups in Europe (such as the English or Danes), and groups of tribes or regions in other parts of the world. Unfortunately, after combining various populations, there is more data than can be easily comprehended. It is helpful that the maps at the back of the book permit the Kreader to get a quick overview of how any allele is distributed.

The Tables of Genetic Distances

A method for simplifying the data is to calculate measures of genetic distance. One of the interesting features of the book is the numerous genetic distance matrices it includes, permitting one to see how closely various populations are related. Before various statistical adjustments, these are the sum of the squares of the differences in frequencies averaged over the various genes. Under a random drift model the distance should be roughly proportional to the time since populations divided, assuming the populations did not differ in size, and the genes were not subject to appreciable selection. Distances are usually calculated as Fst distance, although Modified Nei's distances are given for the worldwide sample of 42 populations. The Fst measure would be 0 if all available gene frequencies were equal, and 1 if they were all completely different (i.e. if every member of one population always had one allele, every member of the other population would lack it.)

These measures of genetic distance are of some interest in their own right. In general one would not expect large differences in frequencies for a particular gene between two populations, if for most genes there is only a small difference in frequencies between the two populations. For instance, anyone arguing that the French and the Germans differed for genetic reasons, would have to contend with the evidence presented here that these two populations are genetically very similar.

Use of genetic distance permits summarizing the worldwide data in a single triangular 42 by 41 matrix. The largest difference in the table appears to be 4573 between the Mbuti Pygmies and the Australians (i.e. the Australian aborigines). In tables the numbers are multiplied by 10,000 for convenience in presentation. Thus, the Pygmies appear to have 45.73% of their polymorphic genes different from the Australians. Thus, it appears that for all pairs of human populations, the majority of the genes will be in common, even if one picks genes whose alleles differ among individuals. Of course, there are a much larger number of genes that appear to have no known variation among humans.

Most of the data is for genes that are either neutral, or close to neutral. In particular, none of the genes discussed are known to affect such genetically controlled characteristics as skin color, hair color, eye color, nose shape, or size. All of these characteristics are known to differ greatly between populations. Likewise, none of the studied genes are known to directly affect such socially important traits as intelligence, criminality, etc (although many such traits are now known to exhibit genetic variability, see Bouchard, Lykken, McGue, Segal, & Tellegen, 1990; Eaves, Eysenck, & Martin, 1989; Herrnstein & Murray, 1994; Miller, 1994a,b; Rowe 1994; Rushton, 1994).

The data for the English, as a population many Americans trace their origins to, can be used to illustrate the nature of the data. In the worldwide sample of 42 populations, the population closest to the English is the Danish (21), and the one most distance the Mbuti Pygmies of Zaire (2373). More important African populations include the Bantu (2288) and the West Africans (1487). For purposes of comparison, the genetic distances between the English and the Japanese is 1244, between them and the South Chinese, 1152, and between them and North American Indians, 947. Estimates of standard errors are provided by the bootstrap method. These estimates indicate that it is unlikely that studying additional neutral genes will changes the conclusions that the English have different gene frequencies than these populations.

The genetic distances between the English and other European populations are small. The two greatest are 404 for the Lapps, and 340 for the Sardinians, two populations that contributed few immigrants to the United States. With major European populations, 22 with the Germans, the distances are 24 with the French , 51 with the Italians, and on up to 204 with the Greeks. In comparison with the much larger genetic distances from the Bantu and West Africans, or the Japanese, South Chinese, or American Indians, the European populations do indeed seem similar to each other. Unless the genes that affect various types of behavior have a frequency difference radically different from the studied genes, genetic differences in behavior between European populations should be small.

Likewise, the various West African populations are similar to one another. The average distance between the various West African tribes is 157, and 211 among the Bantu groups (p. 184). A representative Bantu to West African distance is 188 (p. 175).

Given the large genetic distances between most Europeans, and most Africans, and the similarities within the populations that American slaves and immigrants were drawn from, it does seem reasonable to divide most of the immigrants to America from either Africa or Europe into one of the two conventional groups: Blacks and Caucasians. It is also logical to believe that large genetic differences still exist between the two races. Because the two original grouping differ greatly in skin color, it is to be expected that skin color will convey information about the probability of carrying certain genes, even if population differences in frequency are due only to drift (which is the working assumption for the genes discussed in this book). If the alleles have been subject to climate related selection, as has been argued to be true for intelligence and many aspect of behavior (Lynn, 1991; Miller 1991, 1994; Rushton, 1994), the genetic differences are likely to be larger. Although some would like to argue that knowledge of race is of no use in estimating the probability of someone having a particular trait, a rather simple application of Bayes Theorem shows otherwise . Bayesian statistics show that the posterior estimate should be a weighted average of the information about a particular individual, and the mean for the race he is a member of, with the weights depending on the relative precisions of the information about the individual and the group (Miller 1994c).

Trees of Human Descent

A matrix of genetic distances contains too much data to be readily understood. The data is further presented as dendograms, referred to as trees in the book. Thus, in this section, and in the remainder of the book there is extensive presentation of trees. The populations that are on the same branches are more closely related (as shown by the table of genetic distances). Trees are generally interpreted as having been created by the original human population having divided and subdivided. A rough calibration is attempted from the estimated times of the movement of modern humans out of Africa, and the settlement of Australia and the Americas. The length of the branches leading are portrayed as the relative time since the populations separated. This is not always true, as the authors recognize, since gene frequencies are more affected by drift in small populations, and gene flow between populations makes them more similar even if they had separated many years ago.

Insert about here Fig. 2.3.2.B from p. 79 of book

The African versus all Other Split is Primary A key presentation of the authors results (p. 78) shows trees of 42 populations using frequencies for 120 genes, with genetic distances calculated by two different methods. In both of them, the first split separates African populations from non-African populations. Experiments with bootstrap methods (sampling with replacement from the available data pool to discover how sensitive the conclusions are to changes in the set of genes examined) show that the core African populations (Bantu's, Nilo-Saharans, West Africans, and Mbuti Pygmy) group together 83 and 84 times out of a hundred, showing minor variation in genes studied is unlikely to change the conclusions. When the 42 populations were grouped into nine clusters, Africans versus non-Africans was the first division, and this was true for 98% of the bootstraps (p. 80).

It may be noted that this represents a change from the first results published earlier (Edwards & Cavalli-Sforzaa, 1964, discussed by Cavalli-Sforza et al., 1994, p. 68), which put the first split between a Caucasoid/Negoird grouging and all others. The gene frequency data available then showed the Caucasoids to be more like the Negroids than the Mongoloids. The shift by the Cavalli-Sforza group from grouping the Mongoloids with the Negroids (which would be consistent with modern humans originating in Asia, followed by a branch moving westward, and then subdividing into groups that passed into Africa and into Europe) is explained by the much larger number of loci that are available now, rather than any major methodological difference.

Incidentally, although Cavalli-Sforza et al. here use the terminology of Africans, Europeans, and Asians, it is clear from the populations included in each group that what they really mean is Caucasoids, Negroids, and Mongoloids. They refer to taking thee populations from each continent (p. 68), but the tree (p. 68) shows only two populations from the continent of Europe (the English and the Lapps). The South Turkish were one of the sampled populations, appearing on the same branch as the English. The South Turkish are actually located in Asia (i.e. they are Asian), although they are Caucasoids closely resembling other European populations in gene frequencies. Accuracy and clarity would be improved if the standard scientific racial terms, Caucasoids, Negroids, and Mongoloids, were used, instead of appropriating the well-established terms traditionally applied to those from particular continents, Europeans, Africans, Asians, and giving these terms new, and unusual meanings. Incidentally, later the book uses Caucasian (Fig. 4.10.1 on p. 225) to mean someone from the Caucasus mountains, an accurate usage, but one that could lead to confusion for an unwary reader who only looked at single trees.

There seems to be a general agreement emerging that the first split in the tree of human descent is between Africans and all others. This has been shown by several different methods. As noted, it is what would be expected if modern humans originated in Africa, then moved into the Middle East, and only later divided into other populations.

Nei & Roychoudhury (1993) using 26 populations with the same genes for all populations and a different methodology (neighbor joining) than the Cavalli-Sforza group found that the first split was between Africans and non-Africans, a result that was confirmed in 500 bootstrap replications. The split is again African versus non-African using the same 26 populations but a different tree building method (unweighted pair group method with arithmetic mean). In a test with 15 populations but with more loci (33 loci and 131 alleles), their first split was again between Africans-and non-Africans.

Nei & Livshits (1989) by examining only the three major groups of sub-Saharan Africans (mainly from Nigeria and Cameroon), Europeans (mainly Great Britain), and Asians (mainly Japanese) were able to examine 186 loci, which gave enough data for tests of statistical significance. They found that the distance from the Africans to the Europeans was statistically significantly greater than that from the Europeans to the Asians, even though geography puts Great Britain closer to Nigeria and Cameroon than it is to Japan.

Mountain, Lin, Bowcock, & Cavalli-Sforza (1993) show a tree resulting from using 80 DNA markets on eight populations. The first split is between Africans and non-Africans. The tree of Zhao and Lee (1989) agrees that the largest genetic difference is between Africans and all other populations.

A study of a restriction enzyme haplotype close to the b-globin gene showed "all non-African populations share a limited number of common haplotypes, whereas Africans have predominantly a different haplotype not found in other populations. A genetic distance analysis based on these nuclear DNA polymorphisms indicated a major division of human populations into an African and an Eurasian group" (Wainscoat, Hill, Thein, Flint, Chapman, Weatherall, Clegg & Higgs, 1989, p. 34).

Torroni, Semino, Scozzari, Sirugo, Spedini, Abbas, Fellous, & Santachiara Benerecetti (1990) reported a sharp distinction between Africans and Italians using markers on the Y chromosome. Hammer (1994) has reported a Y chromosome marker (which implies inheritance only from males) which had a frequency of .74 in 611 Africans, but only .07 in 192 Europeans. A tree showed that the first split was again African versus non-African (although the Egyptians grouped with the Africans).

Similar conclusions have been reached by other workers using other genetic markers. Relethford & Harpending (1994) show that a tree constructed using craniometric variation has the first split between Africans and all-others.

Other Splits in the Tree

While the first split in the tree is clear and appears to be well established, the second split is a little surprizing. With the preferred set of distance measurements (Fst), the non-Africans split into Australians, and all others, and then into Southeast Asians, and the remainder. Only then do the Caucasoids separate from the Northeast Asians, Arctic Asians, and American Indians. Using an alternative method for calculating genetic distances, Nei distances, the non-Africans first split into an Australian, and Southeast Asian group, and a Caucasoid, other Asians, and American group. Then the Caucasoids split from the Northeast Asians, Arctic Asians, and Americans. Combining the 42 populations into nine clusters (which increases the number of loci that can be used and reduces the importance of random drift), the non-Africans are then split into a group combining the Australians, Southeast Asians, and Pacific Islanders and into a group including the Caucasoids, Northeast Asians, and Americans.

The results here are surprizing since the Northeast Asians (including Japanese, Koreans, northern Chinese) and American Indians are found to be relatively close to the geographically distant Caucasoids, rather than to the Southeast Asians, who are much closer. This is not what many might have guessed from either the geography or from the similarity of the populations in appearance. Interestingly, detailed inspection of the trees, and the distance matrices show that the Southeastern Chinese (i.e. Hong Kong and vicinity) group with Southeast Asians such as the Filippinos, rather than with the Northern Chinese.

Such an outcome is not impossible. One could imagine the early Middle Eastern population giving birth to a group that moved eastward into Southeast Asia and then on to Australia and New Guinea. Later the Middle Eastern population might have given birth to groups that became the Caucasoids, Northeast Asians, and American Indians.

The authors conduct bootstrap experiments (which in essence repeat the calculations with different sets of genes to see how sensitive the conclusions are to the particular set of genes for which we have data). The conclusions do appear to change depending on the set of genes studied, and the authors suggest that one cannot be confident of the exact order of separation between the branches leading to the Caucasoids, the Northeast Asians, the Southeast Asians, and the Australian and New Guinea populations. They attribute much of the uncertainty to extensive gene flow between Northeast Asia and Southeast Asia, making it hard to produce a tree that fits the data well.

The chief alternative to the extensive calculations undertaken by the Cavalli-Sforza group is another set of calculations done by Nei & Roychoudhury (1993). As already mentioned, these calculations agree that the first split is African versus non-African. However, they place the second split between the Caucasoids and the Greater Asians (Australians, Mongoloids, Americans). The trees they produce correspond very closely to the races as they have been traditionally understood, with their tree grouping populations into groups that are easily recognized as Negroids, Caucasoids, Mongoloids, Amerindian, and Australians. About the only difference from traditional races is that the branch of the tree that leads to the Mongoloids also includes the Australians and New Guinea groups. However, these are on a separate branch. Nei & Roychoudhury discuss why they get a somewhat different tree than Cavalli-Sforza et al. and conclude it is because they use a different method for building trees, neighbor joining, while Cavalli-Sforza et al. use an average linkage method. Nei & Roychoudhury present some cogent reasons for preferring their methods.

Most of the interpretation of the data by Cavalli-Sforza et al. is one of genetic drift, (i.e. the random changes in gene frequencies that occur from one generation to another). The implicit assumption is that population mixing has played little role. However, they do recognize that the theory that observed differences in gene frequencies are due to drift is, at least in principle, testable. For instance, if there is no mixture after separation, all populations that are descended from the same parent population should have approximately the same genetic distances from the various populations descended from another parent population (see table on p. 90). This condition need not be met where there is appreciable gene flow between populations. In general we would expect adjacent populations to exchange genes, and to be more similar than non-adjacent populations.

Certain methods of tree construction produce trees the length of whose branches from the point of origin indicates how much genetic separation has occurred since the populations separated. If the populations are evolving at the same rate, all the branches from a common point should have the same length. Very often this condition is not met.

Perhaps the most striking exception to the predicted pattern is that the branch leading to Europeans is often relatively short. One of the most interesting studies discussed in the book is one that analyzed only a small number of populations (including Chinese, Europeans, two populations of African pygmies, and Melanesians), but collected data on a vary large number of alleles. A tree constructed using this data showed a very short branch leading to the Europeans (p. 91). Several explanations were considered, but the most plausible was mixture. Calculations showed that the European gene frequencies could be explained well by a mixture of the Chinese with a smaller percentage of the pygmies. Obviously, this is not the actual racial history of the Europeans (who are both taller and lighter skinned than either group for instance). The pygmies are fairly close to other Africans in the frequency of their measured genes (the set of measured genes frequencies appears to include no genes that affect height) according to the data in this book.

The evidence that Europeans gene frequencies tend to be intermediate between Africans and Chinese is interesting to those (including the author of this article) interested in behavioral differences between races. Rushton (1994) has presented evidence that on a wide range of characteristics, including intelligence and sexual behavior, the races are ordered Mongoloid, Caucasoid, Negroid. He interprets this as evidence for his differential K theory, while the author of this paper interprets this same pattern as evidence for his paternal investment theory (Miller, 1994a,b). Both have interpreted the fact that so many characteristics had the same pattern as a systematic regularity that called for explanation. It was most easily explained by an evolutionary mechanism, probably taking the form of a common climate related factor producing differences such that the Africans were at the tropical end (or the variable and unpredictable end in Rushton's theory) and Mongoloids at the other end (cold in Miller's account or predictable in Rushton's) with Caucasoids in between. Of course, if Caucasoid's gene frequencies are simply a result of mixtures of two other stocks, the regularity might be explained in other ways (although of course Miller's or Rushton's explanations could still be accurate).

How might European gene frequencies come to be part way between Chinese and African? Part of the explanation is simply geographical. The Caucasoids are located in between the territory of the Negroids and the Mongoloids, and presumably have received genes from both groups. The term Caucasoid is used instead of merely European because it is the Middle Eastern and Indian Caucasoids who are best located to exchange genes with both the Negroids and Mongoloids.

However, a theory of Ammerman & Cavalli-Sforza (1973), discussed in this book (p. 108) provides a mechanism for how the Europeans could come to be intermediate in gene frequencies. They argue on the basis of archeological evidence and gradients of gene frequencies in Europe that agriculture, after emerging in the Middle East, spread into Europe by demic diffusion. By demic diffusion is meant that the early farming populations expanded gradually with each new generation moving further into Europe, with the average rate being about one kilometer per year. The alternative to this account is that the technique of farming diffused without movement of peoples.

Some of the more fascinating work reported in the book is the explanation of gene frequency distributions by the hypothesis of demic diffusion of agriculture. The authors compute first principal components for European gene frequencies. For those not familiar with statistics, the first principal component is a single statistic which condenses as much information as possible about the gene frequencies into a single number. When plotted on a map the component increases systematically with distance from the Middle East. This is explained by the gradual advance of a Middle Eastern farming population into Southeastern Europe. Its gene frequencies were different form that of the original Paleolithic population of Europe. When ever two populations are in contact there is some interbreeding, and genes from the original European populations gradually diffuse into the advancing farming population. It is a fascinating hypothesis, and the use of principal components to support it is ingenious. Many individual genes are distributed as if they had been imported by a population advancing into Europe from the Middle East, with the wave gradually becoming more mixed as the intruding population mixes with the original inhabitants of Europe.

Such an account agrees with what we know about primitive agriculturalists and foragers. Foraging populations are typically very low density, while farming can support much higher densities. Furthermore, a shift to agriculture can plausibly increase the population growth rate, permitting densities to rise rapidly. One of the limitations on population growth in migratory foraging societies is the mother's inability to carry more than one baby at a time. This prevents her from having the next child (or from permitting it to survive) until the first can walk. Thus, births are spaced about four years apart.

In sedentary farming populations births can be more frequent, permitting the population to grow, at least when there is adequate fertile land for expansion. As the population grows, villages every so often become too large and split, with one group leaving to establish a new village. This new village would have been frequently located in a new unsettled area.

Settled farming is a way of life that is quite different from foraging, and one that is in many ways physically harder. Evidence from contemporary foragers shows that they are reluctant to adopt agriculture, and a settled way of life as long as foraging provides an adequate income. Foragers and settled farmers appear to have lived in the vicinity of each other for long periods of time without the foragers taking up agriculture. It also appears that while there is some gene flow between such populations, they basically stay separate.

Thus the Cavalli-Sforza et al. account of demic diffusion of agriculture is plausible. They do illustrative calculations showing that the observed rate of advance (as measured from archaeological sites) is about what would be expected from such a demic expansion (p.108).

Besides its intrinsic interest, what is the importance of whether agriculture in Europe was introduced by demic diffusion or by cultural diffusion? If it was by actual movement of peoples, the current inhabitants of Europe are to a large extent Middle Easterners, rather than descendants of the original Paleolithic inhabitants. Because farming supports a much higher population density , the impact on gene frequencies would be quite large from such an invasions of farmers.

As was noted earlier, gene frequency data suggest that European's gene frequencies appeared to be about what would result from a third African and two-thirds Asian mix. While this mixture could occur by direct diffusion into Europe from Africa or Asia (and undoubtedly there were such gene flows), it is easier to understand if the ancestors of Europeans were originally in the Middle East, possibly even Israel (where there is evidence of a settled culture that stored wild grain, which could have easily shifted to cultivating grains.) Such a population would have been receiving genes from Africa via the Isthmus of Suez (and possibly across the Red Sea) and from Asia.

The evidence of demic diffusion also casts light on the climate in which Europeans evolved. It is a commonplace in evolutionary psychology (also called sociobiology) that the human psychology (and body) was shaped by the extremely long period in which people were foragers, and that we are probably adopted for reproductive success in what is often called the environment of evolutionary adaptation. However, a little thought will show that these environments varied in different parts of the world and ranged from tropical to the cold of Ice Age Eurasia.

The author of this article has argued elsewhere (Miller 1994a, b) that in tropical areas vegetable food was available year around. It has become a common generalization that in hunter-gather societies most of the calories come from gathering and that most of the gathering is done by women, and that the total number of hours expended are low (Lee, 1968). However, examination of the societies used to establish this generalization shows that they were typically tropical societies. In such tropical areas, the females can gather enough food to support themselves and their children. The optimal male strategy is to devote efforts to mating with as many females as possible, and preventing other males from mating with the women he is mated to. Provisioning females and their children is not as strongly selected for (since they will survive in any case).

In Eurasia, the major problem is surviving through the winter when fruits, berries, insects, eggs, and hibernating and migratory animals are unavailable. The common solutions are storage of food (which leads to selection for the ability to defer gratification and for intelligence, see Miller 1991), and the hunting of large animals, such as deer. Unfortunately, women are not effective hunters of large animals (just imagine trying to hunt while carrying a crying baby). Thus, males become the primary supporters of their families during the winters. Females are then selected to look for and attract males who will provision them and their children. Males are selected to form strong pair bonds and to have the personality traits that lead to provisioning.

The ancestors of the Negroids were tropical Africans, and the ancestors of the Mongoloids and Caucasoids were from the cold climate regions of Eurasia. Furthermore, to explain the stronger pair bonds of Mongoloids and other attributes it is necessary to argue they evolved in colder climates. Their stockier build and other features are consistent with this.

That Negroids evolved in tropical Africa, and that Caucasoids and Mongoloids evolved in cold Eurasia is readily accepted. However, some have found it harder to believe that the environment Caucasoids evolved in was appreciably warmer than that for the Mongoloids, especially since there has been extensive publicity given to accounts of Ice Age Europe. It was definitely very cold. Its inhabitants hunted such animals as reindeer and wooly mammoths.

The demic diffusion model makes it likely that the ancestors of modern Europeans were not primarily Ice Age Europeans, but paleolithic Middle Easterners. The gene frequencies of modern Europeans were shaped not only by the cold conditions of Ice Age Europe, but primarily by the conditions in a somewhat warmer Middle East, possibly even in Israel. In turn, the gene frequencies here were influenced by genes diffusing across the Suez Isthmus from Africa.

What happened to the original paleolithic inhabitants of Europe? To a large extent they were absorbed into the populations of the advancing Middle Eastern farmers. However, the evidence presented in this book suggests that the existing population that is closest to the original Europeans is the Basques (p. 276).

If the expansion of Near Eastern farmers affected gene frequencies in Europe, it might have affected gene frequencies into which this farming could have spread (pp. 221-222). Since the book was published, Barbujani, Pilastro, Domenico, & Renfrew (1994) using gene frequency data to argue that not only do European gene frequencies suggest demic diffusion from the Near East, but evidence of such demic diffusion can also be found among the areas once occupied by the speakers of the Altaic languages, and the Asian speakers of the Indo-european and Elamo-Dravidian languages, although only weak evidence was found among the speakers of the Afro-Asian languages.

Other Interesting Findings on the World Wide Gene Distributions

After deriving trees of descent, Cavalli-Sforza et al. compare these with the distribution of language families. They conclude that they are similar. This is not surprising since both languages and genes are argued to spread by the repeated splitting of populations, followed by independent evolution of gene frequencies and languages. Also, people tend not to marry those speaking different languages, and linguistic differences become barriers to gene flows. It should be noted that Nei & Roychoudhury (1993), constructing their trees in a somewhat different manner, found a less close correlation between genetic groups and languages.

Conclusions from Principal Components

Another way the massive amount of data in the table of genetic distances can be condensed is to calculate principal components. In essence, the first principal component is a number which summarizes as much information about gene frequencies as possible. After this is done, a second component can be calculated which summarizes as much as possible of the remaining information and so on. Principal components do not always exist. If the frequency of one allele was completely independent of the frequencies of other alleles, principal components would not exist to be calculated. Principal components are used in other fields. For instance, in psychology the first principal component from a battery of tests is traditionally called g (for general ability), and is usually what the psychometrician means by intelligence.

The first two principal components explained 27% and 16% of the variance respectively (p. 81). Thus there is a high degree of patterning in the distribution of gene frequencies. A graphics technique places the populations on a two dimensional diagram with the first principal component along the base, and the second on the vertical axis. As the authors note, the African populations are in the lower right hand quadrant, and all of the Caucasoid ones are in the upper right hand quadrant. The ones traditionally called Mongoloids are in the left hand side of the diagram, along with the Australian and New Guinean ones. It appears that modern gene frequency data when analyzed with modern sophisticated statistical methods produces something that looks very much like the traditional concept of races. The chief exception is that the Australia and New Guinea populations are in the middle of the left hand side, with populations traditionally considered Mongoloid both above and below. As in their preferred trees (trees and principal component diagrams are merely different ways of simplifying and presenting visually the same information), the Mongoloids seem to fall into a group at the upper left hand corner (including the Japanese, Koreans, Mongols, Ainu, and American Indian groups), and then in the lower left hand quadrant another group including South Chinese, Thai, Indonesians, Malaysians, Filippinos etc.).

As an aside, the reader may note that the second principal component seems to divide populations somewhat by the climatic area in which they are found with the Negroids, South Asians, Australians, New Guineans being at the bottom of the diagram, and the Caucasoids, Northeast Asians, and American Indians groups being at the top. It is possible that the genes that play a large role in determining the second component are ones that are subject to natural selection that is somehow related to climate (possibly through the effects of tropical diseases).

The reader may notice that certain populations that have contributed heavily to the populating of America are close together on the chart. In the upper right hand corner, the Italian, Danish, English, Greeks, (and Iranians) are very close together. Reference to the chapter on Europe shows that most other European populations (such as the Germans, French, Dutch) that helped populate America are very similar to the populations plotted here.

In the lower left hand corner, the Bantu and West African populations come out to be very similar. The European group and the African group are as almost as far apart in the second component as it is possible to be. While the studied gene frequencies do not affect the appearance of individuals, it is plausible that if they did, the difference between the Europeans and Africans would be immediately apparent, and that words would emerge to describe them. Of course, these two population are sharply separated in skin color and other aspects of appearance (due to other genes), and it is not surprizing to find that this difference in appearance has been noticed. People of the two original continents are referred to as white and blacks, Afro-Americans and Euro-Americans, Caucasians and Negroes etc. The second principal component does a good job of separating the two groups of peoples.

There are populations that lie between the two above described clusters. The Berbers are about half way between on the 2nd principal component, and the San and East Africans much closer to other Africans. It is very plausible that these groups located in Africa reflect differing degree of Caucasoid admixture with an African stock. This possibility with regard to the San is discussed in the African chapter. Since these groups contributed relatively little to the populating of America, an impression of a sharper distinction among those of unmixed ancestry than actually exists in the Old World could be created.

One might ask what can principal component maps indicate. The authors argue that when two populations intermingle, all gene frequencies are shifted proportionately in the same direction. As an illustration (not discussed in the book) consider the problem of estimating the percentage of Caucasian intermixture in the African-American population. Consider one gene. The Duffy is a good one, and one that has been classically used. This gene is very frequent in Caucasians but virtually unknown in West African Negroes. The percentage of this gene in an African-American population could be, and has been used, to estimate the percentage of Caucasoid intermixture. If a fifth of the ancestors of the Negroid population were Caucasoids, the frequency of the Duffy gene would be one fifth as high as in Caucasoids. Thus, from one gene, admixture could be estimated. More generally when two parent populations are mixed, the gene frequencies will be w1f1 +(1-w1)f2, where w is the percentage of the daughter population that the first population contributed, and f1 and f2 are the respective gene frequencies. Gene frequencies are always being subjected random changes (drift) and the effects of selection. Thus, one will get slightly different answers depending on the genes studied. The obvious solution is to examine many genes and take an average. Once one had the frequencies of admixture, one could plot them on a map. The first principal component would give a good depiction of the percentage of the invading population in the old. This method would work even if one did not know the populations being mixed. As mentioned, Cavalli-Sforza et al. make good use of a first principal component map of Europe to argue that the observed pattern can be explained by varying mixtures of two populations, an original foraging one, and an expanding Middle Eastern farming one.

Now suppose that one had three populations. One might first compute the gene frequencies to be expected in each of the populations that were mixtures of the first two. The differences between the frequency that could be explained by the mixture of the first two populations, and that observed, might be attributed to mixture from the third. Since the second principal component is constructed to only use information not in the first principal component, its values should indicate the extent of admixture with the third population.

Cavalli-Sforza et al. claim to have conducted simulations which show that the effects of expansions of ancient populations will indeed leave evidence on the principal component maps. Notice that there need be no written evidence of an expansion of the original population, nor does the name of that population, or its gene frequencies have to be known.

In several cases the principal component maps consist of roughly concentric circles, which can be interpreted as indicating mixing with surrounding populations of an original population that underwent a prehistoric expansion. In the discussion, they draw attention to some of the patterns and speculate about what populations might have expanded.

The expansion out of the Middle East with the coming of agriculture is an example that has been discussed. They interpret a pattern of concentric circles around the Sea of Japan as possibly indicating an expansion from that area, possibly of a people similar to that of the prehistoric Jomon culture in Japan (p. 249). Similar maps for Italy are interpreted as possibly providing evidence of the diffusion of the genes of the original Etruscans, who may have come to have a distinctive pattern of gene frequencies through either drift in a small original population, or by immigration to Italy from another area (p. 279).

The book provides principal component maps for the first few principal components both on a worldwide basis, and in each separate continental chapter. Cavalli-Sforza et al. have devised an effective and ingenious mechanism for combining the data provided by the principal components into color maps (used earlier in Menozzi, Piazza, & Cavalli-Sforza, 1978). The human eye can distinguish three primary colors, and by using a separate primary color for each of the first principal components, a map can be prepared which shows the first three components, (which appear to explain about half of the total variance in gene frequencies). The result is some very interesting color maps. One of these is used for the book's dust jacket .


This may be a good place to comment on the views of the authors on race. In the first chapter there is a discussion of "The Scientific Failure of the Concept of Human Races" (p. 19). This opens with the statement that "The classification into races has proven to be futile exercise for reasons that were already clear to Darwin." The reference is presumably to Darwin's knowledge that the races grade into one another, making easy distinctions impossible.

The authors make the point that the measured genetic variability within populations is greater than the variability between populations, which is correct. However, they fail to point out that none of the traditionally studied genes are the ones that relate to such variables as skin color, or nose shape, which are genetic variables that show great variation between populations. It is very likely that many of the genes affecting these traits have gone to fixation in many populations (judging from the absence of dark skin in Swedes, and the absence of non-albino lightly pigmented individuals among the Liberians). At this point, not knowing just which genes influence socially significant traits, we do not know exactly how much of the variation on these traits is between populations, and how much within, although a good guess is that most of the important variation is within populations.

The book states that (p. 19) "However, the major stereotypes, all based on skin color, hair color and form, and facial traits, reflect superficial differences that are not confirmed by deeper analysis with more reliable genetic traits. . ." However, the evidence in the rest of the book serves to disapprove this statement. It has already been pointed out how the trees were calculated, and that the principal component diagrams classify populations in such a way that groups corresponding to races can readily be recognized. The map on the cover makes it easy to recognize the territories of the major races. Australia is a red, sub-Saharan Africa a yellow-green, northwest Europe a green, and China, Japan, Korea a purple, and the New World various shades of purple. Each of the regions corresponds to what are traditionally considered races.

Of course, the map does show intergradations between the major populations. The concept of race as a sub-species implies that such gradations will be found, since if the populations could not interbreed they would be classified as different species, not merely different races. Other maps in the book confirm the existence of races. The map of the first principal component in Africa shows a sharp north to south gradient (p. 191). The contour lines are closer together in the Sahara. A quick glance shows that Africa can be divided into a North African area where live peoples traditionally called Caucasoids, and sub-Saharan Africa where live peoples traditionally called Negroids (the 2 southernmost zones pick up most of sub-Saharan Africa). The map shows a zone in the Sahara where the gene frequencies are intermediate. While such a zone probably does exist on the ground, the actual genetic data for it is weak. Only a few Saharan groups that have been studied (the Tuareg are the most important). The maps are marked with the data points used. Very frequently the data points are for the coast of North Africa, and for points south of the Sahara. In roughly the same way as weather maps are drawn, the computer then fills in the missing lines with zones of smooth transition.

Another very interesting first principal component map is for Asia (p. 250). For this continent the first principal component explains 35.1% of the total variance. The lines run very roughly north south with the extreme values in the Middle East, and in the Far East including Japan, China, and Vietnam. A line running between Burma and India corresponds closely with the traditional Mongoloid/Caucasoid division. It bends to include Tibet in the Mongoloid area, and then proceeds north. As the authors note, the highest values for the Caucasoid pole are not adjacent to Europe, but in the Arabian Peninsula, suggesting a possible gene flow out of that area.

In the far north of Eurasia (where the data is scarce), the Mongoloid line appears to reach almost to the Urals, although there is evidence of considerable mixing in the grasslands of northern Eurasia as populations have moved back and fourth. A widely debated question has been the nature of the Lapps, an Arctic European group speaking a language similar to that spoken in the Urals. The trees show that the Lapps group with other Europeans. Their gene frequencies could be approximated by a mixture of 52.5% Caucasians with the remainder Mongoloid, although another method shows more European mixture (p. 273). The best guess is that this group migrated into Scandinavia from nearer the Urals, bringing a Mongoloid pattern of gene frequencies with them, and then gradually interbred with other Scandinavians, until their gene frequencies had the general European pattern. The Finns, another group that speaks a Uralic language are estimated to be 90% European genes with 10% Uralic, while the Hungarians (also speaking an Uralic language) appear to have a 10% non-European mixture (p. 273).

There are various other small groups around that are difficult to classify into the major racial groups. The Ainu, a traditionally hunter-gathering people of northern Japan, noted for their Caucasoid appearance and hairy bodies prove to have gene frequencies quite close to that of other Japanese, and hence should probably be placed within the Mongoloid major grouping.

Diseases and Gene Frequencies

The worldwide discussion finishes with a section that goes gene by gene, with commentary on the distributions. The details will be mainly of value to those interested in a particular gene. This may be a good point to discuss whether the genes studied are truly as neutral of the theory underlying the book assumes.

The measured distances between populations may be reduced if the genes in question have been subjected to frequency dependent selection. Frequency dependent selection occurs when the less common gene has an advantage. A very important example of frequency dependent selection occurs with parasites and infectious diseases. The body's defenses again foreign organisms depend on identifying them as foreign, which is done by the nature of the proteins on the surface of cells. Genetically, this is controlled by the human leukocyte antigen system, or HLA system. There are several loci, two of which, the A and B are very well studied in different populations. Each of these loci have numerous alleles. The frequency of each loci is treated as a different "gene" in this book. Thus a large part of the data base deals with these loci. "The most important system of markers in our collection, HLA, is represented by 12 alleles and 17 B alleles." (p. 130)

While the population genetics of the HLA system are not very well understood, there is probably a degree of stablizing selection. Otherwise, the observed variability would not have survived (Takahata, 1993). Parasites and disease organisms evolve to have proteins that mimic those in the body. The immune system of an individual who has HLA genes that are relatively rare will find it easier to recognize foreign organisms. If any one HLA allele becomes relatively common, the diseases that attack the carriers of that allele become more common (Jones, 1992, Table on p. 287). The death rate among carriers of that allele increases, reducing the frequency of the allele. Other alleles have an advantage because their body can better recognize the most common pathogens. This mechanism is believed to be what has encouraged the high degree of genetic diversity that characterizes the HLA system. Many alleles are found in both humans, and in species as different as the mouse. (For a readable introduction to the role of parasites in evolution see Ridley, 1994. For a more technical discussion of the human HLA system see Klein, 1990).

There is a brief discussion of known associations with disease, but it is very likely that there are other associations that are not known, including some with diseases which were once important but which are no longer important.

Another very important gene system is the ABO which is vital in typing blood for transfusions. Because of the need for blood typing, it is very well studied, and available for virtually all populations. Certain blood types are known to be more vulnerable to certain diseases, probably because the body can more readily recognize certain invading organisms. For instance O individuals seem relatively resistant to syphilis (p. 126). This may explain why virtually all American Indians (except for Eskimos and some northern Amerind groups) are type O, since syphilis is believed to have been introduced into the Old World by Columbus. Individuals with type A are more vulnerable to smallpox. Tuberculosis (pulmonary) is believed to be more virulent in A individuals than in O or B. Malaria shows a preference for A individuals. Thus, it appears that balancing selection may exist for the ABO blood group.

The frequencies of other genes are believed to be affected by diseases. The Duffy O alle (very high frequency in Africans) confers resistance to a particular malarial parasite, Plasmodium vivax. A number of the G6PD variants produce resistance to malaria. The immunoglobin genes GM and KM, which produce antibodies and play an important role in defense against pathogens, could very well be subject to stabilizing selection. The secretory system FUT2(SE) which brings into "secretions substance responsible for A, B, and the related H substances that are normally found on the red cells of individuals and define their ABO status" (p. 133) is known to affect vulnerability to ulcers, with secretors less vulnerable.

As the brief discussion above shows, many of the widely studied genetic systems that are the subject of this book appear likely to be subject to stabilizing selection (the frequency dependent selection referred to above), such that rare alleles have a reproductive advantage. This would tend to reduce genetic differences between the world's peoples. The effect is probably not enough to make the assumption of neutrality, which underlies much of this book's theory, inapplicable. However, the reader should keep in mind that some gene systems may be subject to stabilizing selection, and others to disruptive selection, and some perhaps to both. A system can be subject to both if allele frequencies tend to a particular equilibrium value under certain conditions, but this equilibrium frequency depends on location. Climatic or cultural differences could make the equilibrium gene frequencies depend on location. Important examples are for malaria where in malarial areas there is a high equilibrium frequency for alleles giving resistance to malaria, and a zero equilibrium frequency in malarial regions . A useful discussion of the distribution of the genes believed to protect against malaria is provided (p. 146-149), although these genes are not used in calculating genetic distances, since their genes reflect selection more than drift.

Incidentally, awareness that many of the easily studied genes appear subject to stabilizing selection is important in evaluating a commonly made argument. It is frequently asserted that only 6.3% of the genetic variation is between races, with the rest being between populations (8.3%) or between individuals within populations 85.4% (Lewontin, 1972). While it is probably true that most of the genetic variability is between individuals, the popular statements do misrepresent the scientific research. A correct statement might be 6.3% of the measurable gene frequencies variation is between races. LewontinTs work (cited on p. 19) dealt with the genes that could be measured at the time he wrote (many fewer than can now be measured). None of these genes affected skin color, nose shape, body build, size, etc. to mention characteristics that differ between races. We can be fairly sure that the genes that were studied (or could be studied given the knowledge then available) were not a random sample of all genes. It appears they overemphasized the genes that were relevant to the body's defenses against disease, and which were subject to stabilizing selection. If this is so, the importance of racial differences is understated.

The Regional Chapters

The remainder of the book is organized in the same way as the Worldwide chapter, except that each chapter focuses on a continental area, and more populations are discussed within each chapter than the few from each continent that were included in the study of 42 populations. Each chapter starts off with a good review of the prehistory of the region, and a history of population movements up to 1500 AD. These are useful to non-specialists, but probably contain little that is not known to the regional specialists.

A distance matrix is then calculated for the selected populations, and used to produce a tree showing the estimated lines of descent. This is then discussed, with emphasis on various interesting issues, such as the origin of particular populations. Principal components are then calculated and discussed. Individual genes are then discussed.

For the chapters on Asia and on Europe, there is a third level. Asia is discussed region by region, (Arctic, Northeast Asia, Southeast Asia, the Indian subcontinent, Central Asia, and West Asia). In Europe, selected regions are given a similar detailed treatment (Italy, France, the Iberian Peninsula, Sardinia), with maps of principal components being presented.


The major surprize to this reader was in the Asian chapter. There is a tendency to think of the third of the human race that is Han Chinese as a homogeneous population. The analysis shows large differences between North China and South China. In a tree with 39 Asian populations (p. 225), the first split puts South Chinese with other Southeast Asian populations, such as the Philippine, Malaysian, Thai, and Indonesian, with the Thai and Viet Muong being the closest. In contrast North China groups with Korea, Japan, and Tibet, as might be expected. However, this group is actually shown as being closer to such groups as the Turkish, Lebanese, and Iranians, traditionally considered as Caucasoid.

A possible explanation is that agriculture emerged twice in China, once in north China for millet, and once in south China for rice, and that these populations then expanded, freezing their gene frequencies. The dividing line is placed between the Yangtse and the Yellow River. Supporting evidence is provided by an analysis of a stratified sample of about 540,000 Chinese surnames from the 1982 Census, which shows a pattern which is argued to be roughly similar to the three Neolithic Cultural areas.

The importance of this finding of a relatively large difference between the North and South Chinese is that much research is done on American or Canadian born Chinese (Vernon, 1982), which are predominantly of South Chinese descent, coming from Hong Kong, Canton, or their vicinity. It may be risky to generalize from this to the whole of Han China.

For those interested in behavior and economic development, the resemblance between South Chinese and the Filipinos, Malays, etc. presents a problem. The South Chinese generally do well on intelligence and academic tests whether tested in the US or in Hong Kong, often better than Caucasoids. Filipinos generally don't do as well. Within Malaysia, the Chinese test much better than the Malays. Within Southeast Asia, the overseas Chinese generally do much better economically than the Malays (Sowell 1994). Thus, it is surprizing to see the small genetic differences between the South Chinese and adjacent populations.


Someone interested in the genetic relationships of various populations will find much of interest in the various chapters on the Continents.

For instance, in inspecting the tree for Europe (p. 268), the Lapps will be found to be the population that is furthest separated from other populations. Next come the Sardinians, which are sufficiently different from other Europeans that their inclusion in the principal components analysis would have required that they be given a component to themselves (p. 291). Their unique gene mix is attributed primely to genetic drift in a small population. The Basques are found to be another distinct group, who are argued to be a remnant of the original Europeans. Iceland is found to be quite distinct from the rest of Europe, which is attributed to genetic drift in a small population. None of these small populations made major contributions to the peopling of America.

A very large cluster puts such Central European peoples as the English, German, Swedish, Italian, Polish, and Russians together (p. 268). Interestingly, the Irish and Scottish are just outside this cluster, even though many think of them as very similar to the English, perhaps because they have been politically united with them. Even though there are historic rivalries between such peoples as the French and the Germans, or the Russians and Poles, the data here shows that any genetic differences are too small to account for much of the differences in national character that some observers claim to see. Needless to say, the similarity in gene frequency among these groups of peoples, which include among themselves such major Europe races, the Nordic, the Alpine, the Mediterranean, and Slavs is strong evidence against any claim for the genetic superiority of the Nordics, or of the Germans, such as the Nazi's reportedly claimed. It is very unlikely that the behaviorally relevant genes could differ much in frequency given the small differences in frequency for the measurable genes. The similarity in gene frequency has been brought about either by these populations being recently derived from a common population or populations (probably a Neolithic farming group spreading from the Middle East followed by later immigrants from the steppes of Asia), or by a high level of intermixture among these various populations. Either of these possibilities would be inconsistent with large differences in the frequency of socially important genes, although it does not make such differences completely impossible.


A somewhat similar situation is found for Africa. Anthropologists traditionally spend much time on small populations that are interesting, but which account for relatively few people. Thus, the African chapter has sections on the Pygmies, the Khoisan, and the peoples of Ethiopia and the Sahara. However, the bulk of the population of sub-Saharan Africa is composed of either Bantu speakers, or West Africans. The populations within both of these large groups are found to differ little genetically from each other. In the case of the Bantu speakers this is believed to be because they spread from a much smaller population originating from near Cameroon. The linguistic, archaeological, and historical evidence for this movement is expounded on. The historical evidence is mainly relevant to South Africa where history shows that the Bantu moved into the area, displacing the Khoisans at about the same time as the Europeans came in. Similarity in languages and archaeological evidence traces the earlier stages of the movement. The genetic similarities between different groups is consistent with the hypothesized movements, and suggests that there were two streams, one moving south first, and the other east into East Africa, and then South (p. 183-185). Because of this relatively recent Bantu expansion, the various Bantu populations do not differ much from each other genetically.

In West Africa, the various population differ from each other a little more, but still resemble each other. The authors hypothesize that this similarity may be caused by an expansion out of a single population that first adopted agriculture. An alternative explanation provides for three such original populations, with only the easternmost (the Bantu speakers) being in a position to expand into southern Africa (p. 185). In any case, the Bantu and the West Africans groups do not differ much genetically.

It was pointed out earlier that the major European populations do not differ much from each other either. Most of the United States is composed of descendants of either the major European populations, or the descendants of slaves from either West Africa or Bantu territory. The two groups are quite distinct in gene frequencies and appearance. On the world principal component diagrams, they are at opposite poles for the second principal component (p. 82). Thus, it is not surprising that in America the difference between descendants of Africans and Europeans has been noticed, and led to people being classified into two races, which have been documented to differ in many traits besides appearance (Herrnstein & Murray, 1994; Miller, 1994a,b,c; Rushton, 1994)

As the book shows, there are numerous populations that are intermediate to these populations in gene frequencies, such as North Africans, East Africans, Nilo-Saharan Ethiopians, inhabitants of the Sahara, and North Africans. There are other groups that have a somewhat different pattern of gene frequencies (Pygmies, Khoisans, Sardinians, Icelanders), but none of these groups contributed much to the United States populations. It can be argued that there are clines in the Old World, with gene frequencies changing gradually from North to South (although relatively rapidly across the Sahara). This doesn't alter the fact that the vast majority of the ancestors of the (non-Mongoloid) United States population can be classified as either Negroid (Bantu or West African), or Caucasoid (European). Of course, subsequent mixing has occurred, and there are many Americans whose ancestry is now mixed.

Outside of the major populations of Africa there are several minor populations that are of interest. The book is filled with fascinating findings about these populations. The Tuareg, who have always been a very mobile people (p. 173) extend over an area stretching from the northern boundary of the dry Sahara (Algeria and Libya) into the Sahel ( p. 171). The authors (p. 173) show that that is a surprizing degree of genetic similarity between the Tuareg and the Beja (whose genetic distance from the Tuareg is only 135), a people in the Eastern Sahara whose territory adjoins the Red Sea.

Since every reviewer must find at least one error, it might be noted that the location of the Beja is different on the map on p. 170 than on the one on p. 171 (which is probably the correct one).

The genetic similarity is surprizing given a relatively large geographic distance. They hypothesize a common origin, perhaps 5000 years ago. This is a long time, but the minimum east-west migration across the Sahara required for the groups to have a common origin is much greater than the width of the Sahara. A few such migrations over tens of thousands of years could greatly reduce or eliminate any genetic differences between North Africa and sub-Saharan Africa. Yet, as the authors document very well, the genetic difference between the Caucasoid inhabitants of North Africa and the Negroid inhabitants directly south in West Africa is quite large (not to mention the obvious differences in skin color and other aspects of appearance). This makes it very likely that the current North African populations did not evolve in place, since if they had they would not be as different from other Africans as they are.

Thus, the large genetic differences north and south of the Sahara present a problem that is not easily solved merely by noting that there is a low density, dry area in between, since large population movements (carrying with them genes) have apparently occurred.

I have developed a theory (Miller, 1994d) that the large genetic difference between the Eurasian populations and the African ones that Cavalli-Sforza et al. document so well is partially due to an early modern movement out of Africa into Eurasia followed by the movement of the Neandertals into the Middle East. This divided the modern population into two segments. Later, a branch, or branches of the European Caucasoid population moved into north Africa.

There is one large area of Africa whose racial affinity has been unclear. This is Ethiopia and adjacent areas. The people tend to have somewhat Caucasoid facial features but dark skins. The gene frequency data suggests that the Amhara (The dominant Ethiopian group) have gene frequencies could be achieved by a mixture of 57% Nilotic African genes with 43% of genes from North Africans (p. 174). Other European populations are similar.Thus, if one must classify these people into one major race, they should be called Negroid. The recorded history of the region and its location makes it very likely that there was an actual admixture of Caucasoid and Negroid peoples here.

Another group that has been the subject of much discussion is the Khoisanid peoples (including the Hottentots, San, !Kung). The San (Bushmen) in southern Africa are a group that physically looks quite different from other Negroids. Baker (1974), and Coon (1965) among others, have argued they are as different from Negroids as Caucasoids are, and should be treated as a separate race from other Negroids. The genetic data reported here shows them to differ more from other sub-Saharan Africans than any of the sub-Saharan groups differ from each other (p. 175).

Interestingly, the San are closer to Near Eastern populations than the adjacent Bantu populations. Their gene frequencies are consistent with their being 56% Near Eastern, with the remainder African. Given that the territory they currently occupy is distant from Caucasoid territory, this is puzzling. However, a possible theory supported by historic remains and linguistic traces, is that they were once were in East Africa, possibly as far north as Egypt. Some mixing with Caucasoids could have occurred then.

The Ethiopian populations, which appear to be a similar Caucasoid, Negroid mix show a considerable genetic distance from the San, suggesting if both are a result of mixture, the mixtures occurred at different times.

An alternative hypothesis, that is supported by mitchorondrial DNA evidence and the San's distinctive morphology, is that they are a relict population of an early race of humans whose territory once covered much of Africa, and are the ancestors of all humans (p. 176). It is interesting to see how modern genetic data supports the earlier idea of Coon that these were a relict of the original populations from which other groups split (1963, 1965). Here the resemblance with the Near Eastern populations is explained by these populations having been derived from the San.

Unfortunately, there is little gene frequency data for Madagascar, and this island is frequently left off of the maps due to lack of information. Madagascar is potentially very interesting because the language of the Malagasy is similar to languages from south-central Borneo. It is generally believed that Madagascar was settled from there by people of Austronesian origin, rather than from nearby Africa (p. 168), whose inhabitants had apparently not yet developed suitable boats.

The Americas

"The genetic evidence for the Americas fully confirm the three waves of migration suggested by dental and linguistic evidence: Amerinds, Na-Dene, and Eskimo" (p. 349). Of course, much interesting detail is supplied. For instance, the high degree of genetic diversity among South American tribes is attributed to drift in numerous small populations.

Australia, New Guinea, and the Pacific Islands are discussed in the final chapter. The genetic evidence is not particularly definitive for Australia and New Guinea.

The book closes with a call for further research, and for collecting data on various small populations of the world before they disappear. Such an effort is underway as part of the Human Genome project.


For the student of race this book makes several points. One is that there is considerable genetic variability between populations. Human populations differ in much more than skin color. This makes it more plausible that they differ in socially and economically important ways including intelligence, personality, disease resistance, sexual behavior etc.

While one can argue about the placement of various small groups, there do appear to be three major groups that include very large number of people, and whose gene frequencies differ. These are the traditional three groups of Negroids, Caucasoids, and Mongoloids. American Indians and Australians constitute other large groupings with distinctive gene frequencies.

Overall, this is a very valuable book that should be in every university library, although its high cost will keep it out of most private libraries.

References Cited

Ammerman, A. J. & Cavalli-Sforza, L. L. (1973). A population model for the diffusion of early farming in Europe. In The Explanation of Culture Change. C. Renfrew ed. pp. 343-357. London: Duckworth.

Austen, R. A. (1979). The trans-Saharan slave trade: A tentative census. In The Uncommon Market: Essays in the Economic History of the Atlantic Slave Trade. Gery, H. A. & Hogendorn, J. S. New York: Academic Press

Baker, J. R. (1974). Race. Oxford: Oxford University Press.

Barbujani, G., Pilastro, A., Domenico, S. D., & Renfrew (1994). Genetic variation in North Africa and Eurasia: Neolithic demic diffusion vs. Paleolithic colonization. American Journal of Physical Anthropology 95, 137-154.

Bouchard, T. J., Lykken, D. T., McGue, M., Segal, N. L., & Tellegen, A. (1990). Sources of human psychological differences: the Minnesota study of twins reared apart. Science, 250, 223-228.

Brues, Alice M. (1990). People and Races. (Prospect Heights, Illinois: Waveland Press).

Cavalli-Sforza, L. L., Menozzi, P., & Piazza, A. (1994). The History and Geography of Human Genes. Princeton: Princeton University Press.

Clark, J. D. (1993). African and Asian perspectives on modern human origins. in Aitken, J., Stringer, C. B., & Mellars, P. A. Eds. The Origin of Modern Humans and the Impact of Chronometric Dating. Princeton: Princeton University Press. pp. 148-178.

Coon, C. S. (1963). The Origin of Races. New York: Alfred A. Knopf.

Coon, C. S. (1965). The Living Races of Man. New York: Alfred A. Knopf.

Coon, C. S. (1982). Racial Adaptations. Chicago: Nelson-Hall.

Eaves, L. J., H. J. Eysenck, & N. G. Martin.1989. Genes, Culture, and Personality. London: Academic Press.

Edwards, A. W. F. & Cavalli-Sforzaa, L. L. (1964). Reconstruction of evolutionary trees. In Phenetic and Phylogenetic Classification. V. E. Heywood and J. McNeill, eds. pp. 67-76. London: The Systematic Association.

Hammer, Michael F. (1994). A recent insertion of an Alu element on the Y chromosome is a useful market for human population studies. Molecular Biology and Evolution, 11, 749-762.

Harpending, H. C., (1994). Signature of ancient population growth in a low-resolution mitochondrial DNA mismatch distribution. Human Biology, 66, 591-600.

Harpending, H. C., Sherry, S. T., Rogers, A. R. & Stoneking, M. (1993). Genetic structure of ancient human populations. Current Anthropology 34, 483-496.

Herrnstein, Richard J. & Murray, Charles (1994). The Bell Curve: Intelligence and Class Structure in American Life. New York: The Free Press.

Jones, S. (1992). Natural selection in humans. In The Cambridge Encyclopedia of Human Evolution. Jones, S., Martin, R., & Pilbeam, Eds. Cambridge: Cambridge University Press.

Kingdon, Jonathan (1993). Self-Made Man. London: Simon & Schuster.

Klein, Jan (1990). Immuniology. Boston: Blackwell Scientific.

Krantz, G. S. (1980). Climatic Races and Descent Groups. North Quincy: Christopher Publishing House.

Laska-Mierzejewska, T. (1982). Dymorfizm Plclowy Czlowieka Odmiany Bialej I Czarnej Na Kubie, Warsaw.

Lee, R. B. (1968). What hunters do for a living. In Lee, R. B. & DeVore, I. Man the Hunter. Chicago: Aldine, 30-48.

Lewontin, R. C. (1972). The apportionment of human diversity. In Evolutionary Biology, Vol. 6, T. H. Dobzhansky, M. K. Hecht, and W. C. Steere, eds. pp. 381-398. New York: Appleton-Century-Crofts.

Loehlin, J. M. (1992). Genes and Environment in Personality Development. Newbury Park: Sage.

Lynn, R., (1991a). Race differences in intelligence: a global perspective, Mankind Quarterly, 32, 254-296.

Lynn, R., (1991b). The evolution of racial differences in intelligence, Mankind Quarterly, 31, 99-121.

Menozzi, P., Piazza, A., & Cavalli-Sforza, L, (1978). Synthetic maps of human gene frequencies in Europeans? Science, 201, 786-792.

Miller, E. M, (1993). Could r selection account for the African personality and life Cycle, Personality and Individual Differences 15, 665-676.

Miller, E. M. (1994a). Paternal provisioning versus mate seeking in human populations, Personality and Individual Differences, 17, 227-255.

Miller, E. M. (1994b). Optimal Adjustment of Mating Effort to Environmental Conditions: A Critique of Chisholm's Application of Life History Theory, with Comments on Race Differences in Male Paternal Investment Strategies. Mankind Quarterly, XXXIV (Summer) No. 4.

Miller, E. M. (1994c).Some Implications of Bayes Theorem for Personnel Selection, Journal of Social, Political, and Economic Studies (in press).

Miller, E. M. (1994d). Did the Neandertals Separate the Africans from the Eurasians? unpublished manuscript.

Nei, M. & Livshits, G. (1989). Genetic relationships of European, Asians, and Africans and the origin of modern Homo sapiens. Human Heriditary, 39, 276-281.

Nei, M. & Roychoudhury, A. K. (1993). Evolutionary relationships of human populations on a global scale. Molecular Biology and Evolution, 10, 927-943.

Pearson, R. (1991). Race, Intelligence and Bias in Academe. Washington: Scott-Townsend.

Relethford, J. H. & Harpending, H. C. (1994). Craniometric variation, genetic theory, and modern human origins. American Journal of Physical Anthropology, 95, 249-270.

Ridley, Matt (1994). The Red Queen. New York: MacMillan.

Rowe, D. C. (1994). The Limits of Family Influence: Genes Experience and Behavior. (New York: Guilford: 1994).

Rushton, J. P., (1994b). Race, Evolution and Behavior: A Life History Perspective. New Brunswick: Transaction Publishers.

Sowell, T. (1994). Race and Culture: A World View. New York: Basic Books.

Takahata, N. (1993). Evolutionary genetics of human paleo-populations. In Mechanisms of Molecular Evolution. Takahata, N.and Clark, A. G. (ed.) Sunderland, Mass.: Sinauer Associates.

Torroni, A., Semino, O., Scozzari, R., Sirugo, G., Spedini, G., Abbas, N., Fellous, M., & Santachiara Benerecetti, A.S. (1990). Y chromosome DNA polymorphisms in human populations: differences between Caucasoids and Africans detected by 49a and 49f probes. Annals of Human Genetics, 54, 287-296.

Vernon, P. E. (1982). The Abilities and Achievements of Orientals in North America. New York: Academic Press.


Miller, Edward M, "Environmental Variability Selects for Large Families only in Special Circumstances: Another Objection to Differential K Theory," Personality and Individual Differences, Vol. 19 (December 1995), No. 6, 903-918.
Miller, Edward M, "Race, Socioeconomic Variables, and Intelligence: A Review and Extension of The Bell Curve," Mankind Quarterly, Vol. XXXV, (Spring 1995), No. 3, 267-291.
Miller, Edward M. and Martin, Nicholas G. "Analysis of the Effects of Hormones on Opposite-Sex Twin Attitudes?," Acta Geneticae Medicae et Gemellologiae: Twin Research, Vol. 44, No. 1, 1995, 41-52.
Miller, Edward M, "Reported Myopia in Opposite Sex Twins: A Hormonal Hypothesis," Optometry and Vision Sciences , Vol. 72, (January 1995) No. 1, 34-36.
Miller, Edward M, "Intelligence and Brain Myelination: A Hypothesis," Personality and Individual Differences, Vol 17, (December 1994) No. 6, 803-833.
Miller, Edward M, "Tracing the Genetic History of Modern Man," Mankind Quarterly, Vol. 35 (Winter 1994) No. 1-2, 71-108.
Miller, Edward M, "The Relevance of Group Membership for Personnel Selection: A Demonstration Using Bayes Theorem," Journal of Social, Political, and Economic Studies Vol. 19 (Fall 1994) No. 3, 323-359.
Miller, Edward M, "Prenatal Sex Hormone Transfer: A Reason to Study Opposite-sex Twins," Personality and Individual Differences, Vol. 17, October 1994, No. 4, 511-529.
Miller, Edward M, "Paternal Provisioning versus Mate Seeking in Human Populations," Personality and Individual Differences, Vol. 17, August 1994, No. 2, 227-255.
Miller, Edward M, "Optimal Adjustment of Mating Effort to Environmental Conditions: A Critique of Chisholm's Application of Life History Theory, with Comments on Race Differences in Male Paternal Investment Strategies." Mankind Quarterly, XXXIV (Summer 1994) No. 4, 297-316.
Miller, Edward M, "The Consistency of Leontief Production Functions with Perfect Substitutability Between Factors," Journal of Financial Management and Analysis., Vol. 7, January-June, 1994, No. 1, 35-43.
Miller, Edward M, "Liquidity: Its Origins and Implications in an Uncertain Multiperiod World with Limited Borrowing." The American Economist, Vol. XXXVIII, Spring 1994, No. 1, 36-46.
Miller, Edward M, "Could r Selection Account for the African Personality and Life Cycle." Personality and Individual Differences, Vol. 15, December 1993, No. 6, 665-676.
Miller, Edward M, "Equivocation in Mathematical Economics Arguments," The American Economist, Vol. XXXVII, Fall 1993, No. 2, 62-66.
Miller, Edward M, "An Analysis of Quality Adjusted Price Indices and Growth Accounting: An Appraisal of the Solow Vintage Model," Journal of Financial Management and Analysis., Vol. 6, July-December 1993, No. 2, 58-71.
Miller, Edward M, "Firm Size Related Implications of the Cost of Accounting Information and Analysis," Review of Financial Economics, Vol. 1, Spring 1992, No. 2, Spring 1992, 68-80.
Miller, Edward M, "On the Correlation of Myopia and Intelligence," Genetic, Social, and General Psychology Monographs, Vol. 118, No. 4, November 1992, 363-383.
Miller, Edward M, "Ricardian Rent, Factor Quality Variations, and the Testable Implications of Production Function Regularity." The Review of Political Economy, Vol. 4, No. 4, October 1992, 467-483.
Miller, Edward M, "Is Aggregation of Capital by its Rent Reasonable? Implications for Growth Accounting," Journal of Financial Management and Analysis, Vol. 5, No. 1, January-June 1992, 33-38.

[The Latest!][Taboos][Stalkers][Thoughts][Library][Web Links]