But why eliminate mutations in this way, rather than correcting more of them by better proofreading? Kondrashov has an ingenious explanation of why this makes sense: It may be cheaper to allow some mistakes through and remove them later. The cost of perfecting proofreading mechanisms escalates as you near perfection.

According to Kondrashov's calculations, the rate of deleterious mutations must exceed one per individual per generation if sex is to earn its keep eliminating them; if less than one, then his idea is in trouble. The evidence so far is that the deleterious mutation rate teeters on the edge: it is about one per individual per generation in most creatures. But even if the rate is high enough, all that proves is that sex can perhaps play a role in purging mutations. It does not explain why sex persists.