In 1993, a National Cancer Institute researcher named Dean Hamer made what seemed to be an astonishing discovery about the genetics of human behavior. He had located a link to male homosexuality on the X chromosome, Hamer reported in Science. The story was splashed across front pages around the country. At last, overly doting mothers and early cross-dressing games were off the hook and the predilections of everyone from Walt Whitman to Liberace could be explained by a few errant proteins. Hamer's article, based on an examination of the DNA of 40 gay brothers, led to mass-market book deals and minor celebrity. There is only one small, underreported glitch: Hamer's results have never been replicated. Two subsequent studies showed much weaker evidence of a gay gene; a third, published April 23 but overshadowed by the massacre at Littleton the day before, found no evidence at all. "There is no hint or trend in the direction of the initial observation," George Ebers, a Canadian investigator involved in the study, said in Science.

A quick trawl through the headlines of the 1990s finds a similar fate for other front-page genetic breakthroughs. Despite much-publicized discoveries of genes for schizophrenia, manic-depression, alcoholism and bipolar disorder, the precise genetic components of these illnesses continue to elude science. The same goes for personality traits. The much-trumpeted discovery of a "novelty-seeking gene" in 1996 hasn't been replicated -- nor have various "depression genes." This is not to say that progress isn't being made in parsing the biological components of behavior. But in any given person, the interplay of genes and the environment is a horrendously complex story. Individual genes produce quite subtle effects, and the more we learn about DNA, the clearer it is that any particular gene's potential can be shut down or enhanced by complex biochemical pathways contingent upon things like sleep, nutrition and stress.

A decade ago the field of behavior genetics was aflutter with the hope that molecular biology would home in on what makes each of us tick. But "the fog is lifting very slowly," says Kenneth Kendler, a professor of psychiatric genetics at Virginia Commonwealth University in Richmond. "We've learned that in psychiatric disorders, there are no single genes of really large effect. If there were, we'd have found them already." Genes for certain conditions, such as Huntington's or sickle cell disease, are known as simple Mendelian traits because they follow the straightforward model of inheritance noted by the Austrian monk Gregor Mendel last century. "Mendelian traits are like a trumpet call. The genetic signal blasts right through," says Kendler. "But the genetic effects in [most behavioral disorders and traits] are like whispers in a busy train station. It's hard to distinguish them from the background noise."

That behavior is a tricky business to predict was elegantly demonstrated by another study published in Science this month. A group of researchers at three universities -- in Oregon, upstate New York and Edmonton, Alberta -- ran a set of identical experiments on eight different mouse strains, each bred to show distinct behavioral attributes. The idea was to see whether they would act according to type. They didn't. In some tests, genetically identical mice acted differently depending on the lab that tested them. A strain of mice lacking a receptor for the neurotransmitter serotonin -- a substance whose imbalance has been implicated in various addictions and mood disorders -- was expected to drink more alcohol and show more anxiety than the other mice strains. But all three teams found that the serotonin-mutants didn't booze it up any more than the others. And all strains of mice tested in Alberta, it turned out, were mellower than the New York and Oregon mice. Must be the weather, eh?

Genetic programming, in other words, isn't nearly as efficient as scientists might hope. Peter W. Nathanielsz, a Cornell University obstetrician whose research focuses on the fetal environments of sheep, writes in a new book that some of the most significant programming of human health -- and behavior, potentially -- occurs in the womb. This isn't exactly a new observation, but advances in neuroscience highlight the fact that the migration of neurons to the precise area of the brain where they belong during pregnancy is a dicey business that can be easily disrupted. Even in a healthy pregnancy, chance has a large impact on the prenatal formation of the brain. Even twins with identical sets of genes have brains that look different.

What happens in the womb is one of the new frontiers of the nature-nurture debate. Epidemiology has already shown how important the womb can be. Over the past several decades, David Barker of Southampton University and other researchers have been studying the outcome of the thousands of Dutch babies conceived or carried during the hunger winter of 1944-45, when the Nazis kept food from reaching people in Amsterdam and the surrounding area to punish the Dutch for the Allied "Bridge Too Far'' invasion. Immediately after the war, this population's nutrition returned more or less to normal, but 50 years after their births, hunger winter babies have much higher rates of diseases like diabetes than controls. Scientists attribute the higher diabetes rates to the "thrifty fetus" phenomenon -- the fetus' pancreases were programmed in the womb to process much lower levels of glucose than became available after birth. "Babies who prepare in the womb for a thrifty existence after birth pay the price if they live a life of over-consumption in a situation in which food is plentiful," writes Nathanielsz, who has observed the same phenomenon in animal experiments.

Rats whose mothers eat low-protein diets suffer from various behavioral and learning problems. If the rats are female, they often pass these traits on to their offspring, even if the second-generation mother's diet is normal, Nathanielsz writes in his book "Life in the Womb: the Origin of Health and Disease." Malnourished rats take many generations of healthy eating to return to normal, in a kind of Lamarckian pattern of inheritance, he writes. "The environment of the womb is of extreme importance in building the body and the brain," Nathanielsz says. "If things go wrong there can be a permanent price to pay."

Twin studies, which compare the sharing of traits in fraternal and identical twins, have provided pretty convincing evidence of a genetic component to most aspects of behavior. They show in study after study that genetically identical twins, even when separated for much of their lives, are far more similar than fraternal twins in everything from IQ to heart disease to shyness. Twin studies produce statistical estimates of the genetic contribution traits -- for example, that IQ is 50 to 80 percent genetic. But even identical twins can have remarkably different womb environments. About a third of all identical twins are nourished with separate placentas, and in as many as 10 percent of the twins who share a placenta, one twin essentially steals womb nutrients from the other, a process that has been called "runting." A few studies that compared the congruence of the behavior of identical twins nurtured in different types of womb environments found "significant personality and IQ differences," says Kendler. Pondering how these studies of womb environment could skew behavior genetics studies, Kendler admits, "is enough to make someone like me nervous."