A new mode of inheritance is postulated in which a sexual offspring receives a contribution from each parent and selects the better to pass on to its own offspring. This could provide a simple advantage to sex over asex whose magnitude is shown to be of the order of a doubling of fitness in each generation, large enough to cancel the twofold cost of sex. This possible advantage to sex can be realized only if a cell component is in fact inherited in this selectively ambiguous way. No such component is known of, but the eukaryotic centrosome is a possible candidate. The possibility is discussed that the centrosome contains an obligatorily non-digital replicator which has an essential function in the initiation of microtubules. If this theory is true, it has the capacity to apply as widely as sex is found, and it would rescue theories of the long-term maintenance of sex from the necessity to provide a twofold advantage in each generation. If false, the theory will soon be disproved.