Basically with enough variables, one can divide almost any sample into subsamples. An example of this is that with a few skull measurements, one can do a pretty good job of separating skulls of 18th-century white Americans from 19th-century white Americans. You can't do it with a single variable, of course, but with a combination of variables, statistically, with more and more variables you'll do better and better and better in dividing individuals into the groups in which they're purported to belong. But that doesn't mean that those groups have any sort of underlying biological integrity, or any sort of underlying real integrity. It's just a matter of statistics. And it doesn't necessarily produce a sorting that we can all agree upon. Variation is always, to some degree, random.

The question, though, is really about race as a scientific or analytical category. It doesn't work as such, for a number of reasons. For one, definitions of race are always based on social definitions. They are socially defined, and thus entirely fluid and unstable, and they vary from time to time and place to place. Secondly, on the biological side, we've all come to realize the incredible amount of variation within any so-called race. So the greater the amount of variation within, the greater the number of variables that you're going to need to define a race. But why even begin to go down that road when there really is no underlying analytical or biological reality in the idea of race in the first place. On a grand scale, I really can't find a reason to think that races would have any sort of reality to them, in terms of selection and evolution.

The concept of race involves not only differences between different races, but similarities within any one race. Although we can use characteristics, genetic or otherwise, to make statistical distinctions between groups of people, such distinctions can be misleading because they do not capture what people generally mean when they talk about "race." This is because within any group defined by those statistics, there is more genetic difference than similarity; we cannot use race defined statistically as a guide to genetic similarity or relatedness.

Traits are inherited independently unless the genes that code for those traits are very close together on the DNA. Most observable physical characteristics that are influenced by genes - such as skin color, hair texture, nose shape and height - are inherited independently of each other. Among the dark-skinned people of Africa we find populations with the tallest and shortest average heights of any people in the world. Another example: there are very dark-skinned people in Africa, India and Southeast Asia, yet even though people from geographically distant places might have the same skin color they often differ with respect to hair texture, nose shape and other physical characteristics. Furthermore, observable physical traits such as skin color do not correlate with particular internal traits; we can't say that if somebody has genes that would cause darker skin, that that correlates with some other things that she might also have inherited.

Also, humans are genetically very similar. We're still a very young species, and a lot of the traits that may be inheritable and may have something to do with behaviors or cognitive processes are probably very old in our species. There's no reason to think that whatever genes exist that might contribute to those kinds of traits would be distributed non-randomly; there is no reason to think that gene variants for particular behaviors or personality traits would be enriched in one group of humans, particularly when the groups include millions of people. Those kinds of traits exist in every population.

One of the fascinating things that's come out of genetics in the last ten or fifteen years, is the discovery that human beings don't have much genetic variation. As Pilar was saying, we are apparently a young species. And if you compare the genetic diversity in a group of chimpanzees with the genetic diversity among humans all over the world, what you find is that chimpanzees are very much more diverse from one another than humans are, in spite of the fact that chimpanzees all look alike to us. And that's because they're a much older species. They've had a lot more time to differentiate.

Also, in terms of the way this question is framed, we're eliding different uses of the term race. On the one hand we're talking about this classic, essentialized race where if you have one drop of non-white blood, you're in that non-white category. And now suddenly in this question, we're talking about average differences between mega-populations. And these are actually quite different conceptions.