Having heard the papers given at this discussion meeting, one must agree that, whether or not one accepts all details of every statement made, there has been in the last 20 years a vast advancein evolutionary comprehension, associated particularly with advances in molecular biology, the theory of group selection, and the application of games theory to much of animal behaviour. On the one hand, our understanding of the nature of the inception of genetic variation has increased out of all recognition; on the other, numbers of phenomena which formerly were the subject of oleographed colour-plates showing ‘bizarre habits’ or ‘outlandish structures’ in popular natural histories are now brought into general theories which give us the means of asking answerable questions about them, and even making predictions. All of this greatly extends the scope of evolutionary theory. There is, however, one major field of enquiry not mentioned by Maynard Smith - rightly, since it has had as yet very little influence on evolutionary theory - and that is natural selection as it actually goes on in the wild. Much of the work we have heard about at this meeting implies natural selection. If it can be shown that of different behavioural strategies directed to the same goal one is more efficient than the others, in expenditure of energy, use of a limited resource, or in resistance to the introduction of other strategies, then the others are less well adapted in these respects and may be expected to be selected out. In general, any demonstration of function implies the possibility of malfunction and the probability of selection acting to keep that function efficient. In some behavioural situations, the implied selection may be intense. A cock robin that is not sufficiently aggressive will not hold a territory and will not breed; but equally, if it is a little too aggressive and drives off both cocks and hens, it will hold a territory but will still not breed. Similarly, in Tinbergen’s remarkable study of eggshell removal in gulls, taking too long away from the nest in removing the shell may leave the chick a prey to neighbour gulls. But staying to protect them and leaving the broken shell in the nest may mean that it is visible to predators, who then kill the chick. In such examples as these, selection against inappropriate behaviour may be from 50 to 100%, and no population geneticist could doubt the efficacy of such a strength.