Westendorp and Kirkwood reply

We found1 that aristocratic women had low average family sizes, the reasons for which are discussed by Jim Cummins2. In these circumstances, any impairment of fertility is likely to result in an increased likelihood of remaining childless. Excluding childless couples from the analysis is therefore counterintuitive. In our opinion, the data from the British aristocracy do not support Ligtenberg and Brand's conclusion that “once you have children, the number you have makes no difference to your life expectancy”.

A study3 of 822,593 women from the Norwegian census of 1970 found that, among post-menopausal women, those with larger numbers of children (more than four) also had higher mortality rates. This is consistent with our finding of a negative association between longevity and reproductive success; furthermore, it suggests that family size does matter.

In his News and Views article about our paper, Daniel Promislow4 suggested that environmental, rather than genetic, factors might explain the trade-off between longevity and reproductive success, which we showed was similar for women and for men. For example, a large family might increase environmental stress and mortality risk for both parents. If this was the case, spouses’ lifespans should be correlated. We found a statistically significant correlation, but it accounted for only 2% of the variance in age at death. The weakness of this correlation argues strongly against environmental factors playing a major role in the trade-off, and supports the hypothesis that genetic factors are important.