It's true, on the one hand, that the major dimensions of personality measured by psychologists--extroversion, neuroticism, etc.--have "heritability" of around 40 percent or 50 percent. Technically, this means that around half of the variation in these traits among individuals is statistically accounted for by genetic differences among them, and that the rest is attributable to environmental differences. In real, everyday life, heritability estimates mean less than meets the eye, but we won't get into that now. (Though I should note that these estimates never say anything at all about genetic differences between different ethnic groups, a methodologically intractable issue.) Let's just accept these "heritability" numbers at face value and ask what they might mean for the compatibility of adoptive siblings. Will these pseudosiblings be too "different" to get along?

If so, then natural siblings are in trouble, too. Because it turns out that in terms of personality, two natural siblings are, on average, almost as different as two randomly selected members of the population at large.

But how can this be? How can siblings be so different if heritability statistics say that genes are important for personality, and given that siblings have significantly more genetic commonality than other people do? (At least, siblings have significantly more commonality within that relatively small subset of genes that differ widely among the population at large; and those are the genes that matter when we're talking about heritability.) The answer has to do with a fairly recent conceptual innovation in psychology: the idea of "nonshared environment."

The influence of environment on personality--the 60 percent or 50 percent of variation in personality not accounted for by genetic differences--isn't confined to things that siblings share, like their neighborhood, their parents' income level, and so on. It includes things siblings don't share. Like: Which child got Cruella De Vil as a kindergarten teacher? It also includes things siblings can't share. Like: Which child is the firstborn, and thus has all kinds of experiences unique to firstborns, such as the undivided attention of parents during infancy? (Firstborns have higher IQs, on average, than later-borns.)

An important implication of the idea of nonshared environment is this: While you are more like your sibling than your neighbor genetically, and while you are more like your sibling than your neighbor in terms of some aspects of early environment, you are less like your sibling than your neighbor in terms of other environmental aspects. And these latter aspects can be very important. In some cases, they more than compensate for the genetic commonality with your sibling. That's why, in terms of personality, siblings are almost as different as strangers on a train.

This point is central to Frank Sulloway's justly celebrated book Born to Rebel. The fact that siblings can't share the same birth order, he believes, leads to all kinds of subtle and consequential differences among them. Firstborns, for example, can get away with pushing their dinky siblings around, and thus they tend to become, through positive reinforcement, more aggressive than the latter. And indeed, Sulloway shows, in terms of aggressiveness, the average firstborn is more like a randomly selected firstborn from the population at large than like his or her younger siblings. So too with other selected traits, like deference to authority and conscientiousness.

Sulloway explains this striking fact partly by reference to a "family ecology" model. The idea is that people are designed by natural selection to compete with their siblings for parental favor, and that in doing so they become systematically different from their siblings. If your older sibling has occupied the "conscientious establishmentarian" niche, you may occupy the "carefree, artsy, creative" niche.

If you want to learn more about birth order and personality, read Sulloway's book. But for our purposes, the moral of his story is this: So much for the idea that mixing unrelated children through adoption will lead to an unnatural and combustible amount of diversity in personality. If Sulloway is right (and he's got the simplest explanation of the data I've seen), then not only are we not designed to be reared amid people like ourselves, but we are designed to become in some respects systematically unlike the people we're reared with! For better or worse, families have been mixing oil and water for millennia, regardless of whether any of their children are adopted.