James R. Flynn is an accidental IQ specialist. In the early 1980s, the American-born political scientist thought he might spend a few pages in a planned book on "how to defend humane ideals" grappling with the argument that the gap in IQ scores between blacks and whites was genetically rooted. It was not his first foray into that subject, but this time, as he dug into international intelligence-test data, he came across a puzzling trend: Although the tests were calibrated so that 100 was always the average, the underlying performance was moving steadily upward. Were people getting smarter? Mr. Flynn has calculated that, had they taken a circa-2000 IQ test, Americans in 1900 would have scored a 67, suggesting mental enfeeblement. But one's grandparents (or great-grandparents) probably did not come across as mental midgets in their day. What was going on? More than three decades later, Mr. Flynn is still trying to nail down what it means to be smart, and his legacy is almost surely going to be the "Flynn effect," the term Charles Murray and Richard Herrnstein applied to the rising-IQ phenomenon in "The Bell Curve." Few question the data underpinning the Flynn effect, yet its causes and implications remain very much up for grabs. On a more intellectual level, breaking down IQ tests into their constituent parts has allowed Mr. Flynn to uncover fresh mysteries. The incongruity between rising adult vocabulary scores and flatlining scores for kids and teenagers? Mr. Flynn suggests that the cause is the rise of a self-contained adolescent culture, birthed in the 1950s, that is "impervious to parents' vocabulary." Mr. Flynn also thinks he's found evidence of what he dubs a "bright tax" in the case of analytic ability. The smarter someone is in that subskill, the more that person's scores appear to drop with age. People with average analytic scores appear to maintain those scores from ages 17 to 72, but those with very high scores see them drop as they age. There is no such falloff in verbal skills on the high end. Mr. Flynn's hypothesis is that, when the typical elderly retire, they confront far few analytic problems in their day-to-day lives, even as they continue to use language as much as ever. And while many people think the Flynn effect shows that IQ is largely environmentally rooted, and therefore less politically toxic, a resolute minority remains unpersuaded. Flynn effect or no, the smart money says the IQ wars will continue to rage, in journals if not undergraduate classrooms.