The first line of evidence for a genetic basis of homosexuality came out August 30, 1991, when Science, the most prestigious scientific journal in America, published a short paper by neurobiologist Simon LeVay entitled "A Difference in Hypothalamic Structure between Homosexual and Heterosexual Men." Comparing the brains of dead gay male AIDS victims to those of dead straight men and dead straight women, LeVay found that a tiny region of the hypothalamus (INAH ₃) was larger in men than in women and larger in straight men than in gay men. Although LeVay discussed homosexuality as "biological," he did not at that point suggest that his work indicated that it was genetic or innate. In fact, he said explicitly that "the results do not allow one to decide if the size of INAH ₃ in an individual is the cause or consequence of the individual's sexual orientation." But to the New York Times, he made it clear that he thought it was the cause, suggesting "that the hypothalamic segment could be responsible for inspiring males to seek females."

A Newsweek cover story, "Is This Child Gay?", also featured LeVay's work and characterized him as a "champion for the genetic side."

But where's the genetics?

LeVay studied brains, and the key assumption promoted in the public arena is that brain structure is a direct result of genetic instructions. But it isn't. The brain grows and develops interactively with the experiences of the person; brains are not reliable surrogrates for studying genes.

Unfortunately, LeVay never followed up on this study, and other brain analyses have found his results to be equivocal.

The most interesting aspect of the study is that, aside from the technology, it is conceptually very unmodern. The logic is that the brain is a surrogate for the genes, and that the consistent difference in brain structure implies an innate basis for the thoughts associated with the brain. That's entirely wrong; virtually out of a different century, and affords an excellent illustration of a simple rule of modern molecular anthropology: Genetic conclusions require genetic data.

The second line of evidence for the genetic basis of homosexuality was a study of twins, showing a high concordance for sexual orientation between identical twins. Twin studies have a long and rather sordid history as surrogates for genetic studies. The most notorious were those of Sir Cyril Burt, who found identical twins raised apart to be incredibly similar in IQ scores - so similar, in fact, that the statistics describing their similarities didn't change even when his sample of twins tripled. His power in the English psychology community was so great that nobody dared to challenge him, but upon his death in the 1970s, it was found that he had invented collaborators, ghostwritten book reviews under their names, and generally transgressed virtually every boundary separating the credible scientist from the wacko mad scientist.

Twins are, of course, very powerful cultural figures, the subjects of old mythologies and many hokey novels. In theory, they should be good genetic guinea pigs for nature-nurture experiments, but they aren't in practice. For example, the homosexuality twin study found 52% of identical twins concordant for homosexuality, as opposed to a control group of adopted siblings, only 11% of whom were both gay.

Of course, that leaves 48% of identical twins with different sexual orientations. And the 11% is a much higher estimate of homosexuality among unrelated people than is generally taken as a reliable estimate of homosexuality "out there."

But more significantly, this is a very crude comparison. The same study showed that fraternal twins were concordant 22% of the time, which seems to fit the idea that it is genetic (the value is less than for identical twins and more than for unrelated people); but ordinary nontwin brothers were concordant for homosexuality only 9% of the time. However, genetically, fraternal twins are simply siblings born at the same time. How could the concordance rate for fraternal twins be more than twice as high as the value of for brothers generally, unless the study is actually revealing a strong effect of twinship, rather than of genetic identity?

Twins tend to be treated similarly, and tend to regard themselves as more similar to each other than ordinary siblings - identical twins even more so. This study has not managed to isolate the genetic similarity of twins and bracket it apart from the "overall" similarity of twins. The fact that they are very concordant for a particular trait consequently tells us nothing about genetics.

A second simple rule of molecular anthropology: Similarity among relatives is not necessarily, and often isn't, genetic. See Rule #1.

The third study purporting to show that homosexuality is genetic is the only one, oddly enough, actually to be based on genetic data. Dean Hamer, a glib and charismatic researcher, led a team that found an association between a tiny variant segment of the X chromosome (called X_q28) and male homosexuality, published with considerable fanfare in Science on July 16, 1993.

At least here we have some genetic data.

But what is the nature of these data and how convincing are they for establishing a "genetic basis" for homosexuality?

As already noted, modern genetics relies on establishing a cause-effect relationship between a functional bit of DNA and an observable trait. But this study didn't. It found that brothers who were gay tended to match at this genetic region.

But what is a match? Crucially, the scientists isolated no gene there, and no physiological product affecting sexual orientation is known to be made there. The claim made by Hamer and his colleagues is simply that they found this region to be grossly more similar (83% matching of small genetic marker regions) in a specific sample of gay brothers than at random (50% matching). The result hinged not on a mechanistic analysis but on the statistical difference between the frequency of gay brothers with similar genetic markers in a specific chromosomal region and the random expectation.

The cultural meaning they imparted to the result they reported was that "chromosomal region" implied "functioning gene," which inturn implied "control of trait."

Assuming, of course, that the association was real to begin with. The tricky design of the study makes it very sensistive to a few families matching or not, because the key scientific question is not "Do we have a gene for homosexuality?" but a surrogate question: "Is our 83% result sufficiently different from 50% to be meaningful?" A follow-up study by a different group found no such difference at all. Another follow-up by the original researchers found the difference now to be 67%, rather than 83%, quite a bit closer to the 50% expected at random.

Nevertheless, best-sellers were written; careers and fortunes were made. "Born Gay?" Time asked on July 26, 1993. And the "gay gene" - which has never subsequently been found - entered the popular mind as a fact of science.

For me, however, the most interesting aspect of the study was the scope of the actual claim. How much homosexuality did these researchers believe they had actually explained with their study? From the publiscity, you might expect the figure to be 90%. Or perhaps a more conservative 70% - perhaps they had explained over two-thirds of the homosexuality in our species, which would certainly merit headlines.

In fact, however, when I posed that very question at a conference in 1996, the answer was very different. It came in two parts. First, the result, according to the researchers, was ostensibly only about male homosexuality and had no relevance on female homosexuality; and, second, they believed they had explained about 5% of male homosexuality.

Five percent.

If we make a simplifying assumption that male and female homosexuality exist in the universe in equal proportions, then at best - assuming that homosexuality is a property of a person, not of an act, and assuming all the statistical issues raised are invalid, and assuming there is actually a gene there - they would have accounted for 2.5% of homosexuality in our species.

The third rule of molecular anthropology: There is no science other than behavioral genetics in which you can leave 97.5% of a phenonmenon unexplained and get headlines.

That is the most obvious indicator of the cultural power and meaning of this work, and why it needs to be considered very carefully and regarded very skeptically. Virtually any claim, no matter how ridiculously small, can grab headlines. The question is not, "Do you believe homosexuality is genetic?" After all, the Constitution of the United States guarantees you the right to believe anything you want. The question is, "What have we actually shown scientifically about it?"

And the answer is, almost nothing.