Moreover, Fisher realized that it isn't strictly the numbers of males and females that are held at 50:50 by natural selection, but what he called the “parental expenditure” on sons and daughters. Parental expenditure means all the hard-won food poured into the mouth of a child; and all the time and energy spent looking after it, which could have been spent doing something else, such as looking after another child. Suppose, for instance, that parents in a particular seal species typically spend twice as much time and energy on rearing a son as on rearing a daughter. Bull seals are so massive compared with cows that it is easy to {110} believe (though probably inaccurate in fact) that this might be the case. Think what it would mean. The true choice open to the parent is not “Should I have a son or a daughter?” but “Should I have a son or two daughters?” This is because, with the food and other goods needed to rear one son, you could have reared two daughters. The evolution-arily stable sex ratio, measured in numbers of bodies, would then be two females to every male. But measured in amounts of parental expenditure (as opposed to numbers of individuals), the evolutionarily stable sex ratio is still 50:50. Fisher's theory amounts to a balancing of the expenditures on the two sexes. This often, as it happens, turns out to be the same as balancing the numbers of the two sexes.

Even in seals, as I said, it looks as though the amount of parental expenditure on sons is not noticeably different from the amount spent on daughters. The massive inequality in weight seems to come about after the end of parental expenditure. So the decision facing a parent is still “Should I have a son or a daughter?” Even though the total cost of a son's growth to adulthood may be much more than the total cost of a daughter's growth, if the additional cost is not borne by the decision maker (the parent) that's all that counts in Fisher's theory.

Fisher's rule about balancing the expenditure still holds in those cases where one sex suffers a higher rate of mortality than the other. Suppose, for instance, that male babies are more likely to die than female babies. If the sex ratio at conception is exactly 50:50, the males reaching adulthood will be outnumbered by the females. They are therefore the {111} minority sex, and we'd naively expect natural selection to favor parents that specialize in sons. Fisher would expect this too, but only up to a point – and a precisely limited point, at that. He would not expect parents to conceive such a surplus of sons that the greater infant mortality is exactly compensated, leading to equality in the breeding population. No, the sex ratio at conception should be somewhat male-biased, but only up to the point where the total expenditure on sons is expected to equal the total expenditure on daughters.

Once again, the easiest way to think about it is to put yourself in the position of the decision-making parent and ask the question “Should I have a daughter, who will probably survive, or a son, who may die in infancy?” The decision to make grandchildren via sons entails a probability that you'll have to spend more resources on some extra sons to replace those that are going to die. You can think of each surviving son as carrying the ghosts of his dead brothers on his back. He carries them on his back in the sense that the decision to go the son route to grandchildren lets the parent in for some additional wasted expenditure – expenditure that will be squandered on dead infant males. Fisher's fundamental rule still holds good. The total amount of goods and energy invested in sons (including feeding infant sons up to the point where they died) will equal the total amount invested in daughters.

What if, instead of higher male infant mortality, there is higher male mortality after the end of parental expenditure? In fact this will often be so, because adult males often fight and injure each other. This circumstance, too, will lead to a {112} surplus of females in the breeding population. On the face of it, therefore, it would seem to favor parents who specialize in sons, thereby taking advantage of the rarity of males in the breeding population. Think a little harder, however, and you realize that the reasoning is fallacious. The decision facing a parent is the following: “Should I have a son, who will likely be killed in battle after I've finished rearing him but who, if he survives, will give me extra specially many grandchildren? Or shall I have a daughter, who is fairly certain to give me an average number of grandchildren?” The number of grandchildren you can expect through a son is still the same as the average number you can expect through a daughter. And the cost of making a son is still the cost of feeding and protecting him up to the moment when he leaves the nest. The fact that he is likely to get killed soon after he leaves the nest does not change the calculation.

In all this reasoning, Fisher assumed that the “decision maker” is the parent. The calculation changes if it is somebody else. Suppose, for instance, that an individual could influence its own sex. Once again, I don't mean influence by conscious intention. I am hypothesizing genes that switch an individual's development into the female or the male pathway, conditional upon cues from the environment. Following our usual convention, for brevity I shall use the language of deliberate choice by an individual – in this case, deliberate choice of its own sex. If harem-based animals like elephant seals were granted this power of flexible choice, the effect would be dramatic. Individuals would aspire to be harem-holding males, but if they failed at acquiring a harem they would much prefer to be females {113} than bachelor males. The sex ratio in the population would become strongly female-biased. Elephant seals unfortunately can't reconsider the sex they were given at conception, but some fish can. Males of the blue-headed wrasse are large and bright-colored, and they hold harems of dull-colored females. Some females are larger than others, and they form a dominance hierarchy. If a male dies his place is quickly taken by the largest female, who soon turns into a bright-colored male. These fish get the best of both worlds. Instead of wasting their lives as bachelor males waiting for the death of a dominant, harem-holding male, they spend their waiting time as productive females. The blue-headed wrasse sex-ratio system is a rare one, in which God's Utility Function coincides with something that a social economist might regard as prudent.

So, we've considered both the parent and the self as decision maker. Who else might the decision maker be? In the social insects the investment decisions are made, in large part, by sterile workers, who will normally be elder sisters (and also brothers, in the case of termites) of the young being reared. Among the more familiar social insects are honeybees. Beekeepers among my readers may already have recognized that the sex ratio in the hive doesn't seem, on the face of it, to conform to Fisher's expectations. The first thing to note is that workers should not be counted as females. They are technically females, but they don't reproduce, so the sex ratio being regulated according to Fisher's theory is the ratio of drones (males) to new queens being churned out by the hive. In the case of bees and ants, there are special technical reasons, which I have discussed in The Selfish Gene and won't rehearse here, for expecting the sex ratio to be 3:1 in favor of {114} females. Far from this, as any beekeeper knows, the actual sex ratio is heavily male-biased. A flourishing hive may produce half a dozen new queens in a season but hundreds or even thousands of drones.

What is going on here? As so often in modern evolutionary theory, we owe the answer to W.D. Hamilton, now at Oxford University. It is revealing and epitomizes the whole Fisher-inspired theory of sex ratios. The key to the riddle of bee sex ratios lies in the remarkable phenomenon of swarming. A beehive is, in many ways, like a single individual. It grows to maturity, it reproduces, and eventually it dies. The reproductive product of a beehive is a swarm. At the height of summer, when a hive has been really prospering, it throws off a daughter colony – a swarm. Producing swarms is the equivalent of reproduction, for the hive. If the hive is a factory, swarms are the end product, carrying with them the precious genes of the colony. A swarm comprises one queen and several thousand workers. They all leave the parent hive in a body and gather as a dense cluster, hanging from a bough or a rock. This will be their temporary encampment while they prospect for a new permanent home. Within a few days, they find a cave or a hollow tree (or, more usually nowadays, they are captured by a beekeeper, perhaps the original one, and housed in a new hive).