I just read with interest
Participation Explains Gender Differences in the Proportion of Chess Grandmasters on the blog Pure Pedantry. It's a discussion of the research article
Sex Differences in Intellectual Performance: Analysis of a Large Cohort of Competitive Chess Players by Chabris and Glickman.
It's a nice discussion of some recent research looking at boys' and girls' chess ability. They were looking at chess because of all the interest in math & science ability, of course, but chess is nice because it's a reasonably objective system of ranking based on tournament results. They looked at a large sample -- 250,000 tournament players over 13 years -- and consider a number of explanations for the observation that boys and men dominate elite chess.
It seems that the hypothesis that explains their data best is a quite simple one: lower participation rates of girls in chess mean a lower chance that the girls who would be really good at it even try it. When participation rates are equal (which they are, in certain zip codes), abilities are equal. Other popular hypotheses, like the idea that males have higher variance (so more occur in the top tail) are not supported by the data.