Sir

We are students aged 16–18 in a Texas high school. Our biology teacher Vidya Rajan asked us to comment on the paper by A. J. Tatem and colleagues (Nature 431, 525; 200410.1038/431525a); we believe the projection on which it is based is riddled with flaws.

The idea of women running faster than men — although not novel (see B. J. Whipp and S. A. Ward Nature 355, 25, 199210.1038/355025a0; and Correspondence Nature 356, 21, 199210.1038/356021b0) — is interesting, but one cannot draw these conclusions based on generalization by extrapolation. Tatem et al. used a domain of 104 years to extrapolate to a domain of 252 years. It is not logical to say that the first 104 years will have data with exactly the same regression as the next 148 years. Using similar reasoning in 1992, Whipp and Ward suggested that women would run the marathon faster than men by 1998. This has still not happened.

In Tatem and colleagues' study, men were measured for 32 more years than women. This ignores the possibility that women might be reaching a plateau: had women's times been unexpectedly high before 1934, one could trace a decreasing rate of change for post-1934 Olympians.

Improvements due to the increase in numbers of women running are likely to level off as the rate of increase in participation slows down (see http://www.olympics.org.uk/olympicmovement/olympicissueswoman.asp).

Finally, both men and women may reach a physiological limit beyond which they cannot progress.

With these factors taken into consideration, the predictions made from the extrapolation seem less than sound.