The observation that there are more males than females at the lower end of the IQ distribution has been confirmed many times and is generally accepted. Because the mean IQ of males and females is the same, there should be a concomitant increase in the number of males at the higher end of the IQ distribution. This, in fact, has been observed and reported many times in the literature over the last 50 years. In 1945, Fraser Roberts reviewed data from two relatively unbiased studies of IQs in British children, both showing greater variability in males but no difference in mean IQ. Using the smaller, more conservative variance (13%), he presented tables showing the relative number of boys/ 100 girls in the extreme percentiles. For both the cleverest and dullest 1%, there were 144 boys per 100 girls. For the cleverest or dullest 0.1%, there were 186 boys per 100 girls.Although conveniently ignored for most of the last 50 years, the published evidence of greater male variability in intelligence and related tests at both ends of the spectrum is very strong, as shown by a recent review [Hedges and Nowell, 1995] of a series of related tests. They performed secondary analyses of six data sets collected over 20 years on large, national samples of U.S. children. Among the 32 tests for reading comprehension, vocabulary, mathematics, perceptual speed, science, social studies, nonverbal reasoning, associative memory, and spatial ability, the variance was greater in males in 30. However, in no test were males more variable in both the high and low end of the distribution, and the direction usually followed a small (10%) but significant difference in means. When there was a higher mean in males, there were more males in the upper percentiles and vice versa. Males were better at mathematics, science, and spatial ability and worse at perceptual speed, associative memory, and reading comprehension, but clearly were more variable. These results are very similar to those reported by Wechsler in 1958 in a chapter on sex differences in intelligence. Among the 11 subtests, males were better in four, females were better in two, and there was little difference in five. Therefore, the similar means in males and females are, in part, a result of the selection of tests in which males do better in some and females do better in others. It could be done differently and, indeed, Wexler, in the same chapter, also sorted the tests to show profiles of maleness and femaleness. However, it never became popular. He did not comment on the small excesses in male variances, but these were largely in the direction of being greater in males. The conclusions remain the same. Males are better at some things, worse at others, but more variable.These results, which indicate that the overall male or female distribution is shifted upwards or downwards for a particular cognitive function, are much more easily explained by testosterone (or estrogen) effects on males in general, rather than the effects of rare genetic variations that would affect only a few. However,...