[[p. 196]] Mr. Weldon's letter on this subject, in Nature of May 3, calls, I think, for a few further observations. He first criticises the statement that "the survival-mean must, on cessation of selection, fall to the birth-mean," by showing that there are probably cases in nature in which the survival and birth-means may coincide, owing to the removal by selection of all individuals above and below the mean, they being approximately equal in number. This is, no doubt, the case with certain characters of a species, but probably never with all or even with most characters. Darwin states that in France and Germany white pigeons are killed off by kites, and that on the coast of Ireland black fowls are also killed off by sea-eagles. These and other analogous facts render it probable that in many species of animals colour is kept to the inconspicuous and protective mean tint by the elimination of all individuals which vary much on either side of it, and thus, as regards colour, the birth-mean and the survival-mean may be almost identical. But with many other characters this is not the case. In sheep, cattle, and horses it has been observed that when the larger lowland breeds are taken to bleak mountain regions they gradually dwindle in size, only the smaller and hardier of each generation surviving the severe winter and spring climate and the comparatively innutritious food. Here the elimination is clearly in one main direction; and the absence of this selection due to the transference of the whole body of such reduced individuals to a milder climate and better pastures, would no doubt lead to a slight increase of average size, indicating that the birth-mean had been above the survival-mean. So also in the case of the half- [[p. 197]] wild horses of Circassia, which are greatly exposed to attacks of wolves and to extreme vicissitudes of climate, swiftness, strength, wariness, and a hardy constitution must be kept at a high level of efficiency by the elimination of the less gifted in these qualities; so that here again the birth-mean must be below the survival-mean. In such cases as these there seems no difficulty in the fact that the mean characters do not change for many generations; for this is in accordance with Darwin's principle that natural selection "cannot produce absolute perfection, but only relative perfection." When the average characters of a species have reached a point such that it can permanently maintain itself in a given area, then no further change will occur; but, the less efficient being constantly weeded out, the survival-mean will be necessarily a little above the birth-mean. Both means will, however, be sensibly permanent as long as the environment remains unchanged.

    Mr. Weldon says that it has not been shown that, in some given case, Panmixia does in fact occur; and further, that in the only case which has been experimentally investigated--that of the stature of civilised Englishmen--the consequences said to result from it do not, in fact, occur. To obtain absolute evidence of Panmixia, or of the action of Natural Selection, is extremely difficult, because we cannot first compare and measure minutely a large number of individuals in a state of nature, and then follow those same individuals throughout their lives and see how nature deals with them. We can, however, observe what happens in the case of semi-wild animals, and the examples already cited show that natural selection must, and actually does, act on the character of colour, weeding out those which diverge on both sides towards whiteness or blackness, and in the case of physical and mental activities destroying those which fall below the standard of excellence requisite for the preservation and continuance of life.

    In our domesticated animals, on the other hand, we find what are probably examples of the effects of Panmixia. The wing-bones of our pigeons, fowls, and ducks, as compared with wild individuals, were found by Darwin to be decidedly reduced in size in proportion to the leg-bones; but a part of this may be due to disuse in the individual, and to determine the share of the two causes seems impossible. There are, however, a few characters in which we see Panmixia alone at work in our domesticated animals. Such are, for example, the constant appearance and increase among them of prominent unsymmetrical markings, as in dogs, cats, cattle, and horses. Such markings never occur in wild races, or if they occur in individual cases they never increase; and I have given reasons for thinking that symmetrical colour and marking is kept up in nature for facility of recognition, a factor essential to preservation, and to the formation of new species. In this case, there can be no question of disuse, while as we know that white and unsymmetrical individuals do occasionally occur in wild species, but never increase, the fact of their increase under domestication must be due to the absence of whatever form of natural selection eliminates them in nature; that is, to Panmixia. Another illustration may perhaps be found in the fact of curled tails appearing in domestic pigs and some races of dogs, while no wild animal is known which has a curled tail. We can hardly doubt that the special form of tail in each animal is of use to it, and that any abnormality, like a curled tail, would be eliminated under nature. Its appearance and perpetuation under domestication is therefore a fair example of Panmixia.

    The slow increase of the stature of civilised Englishmen, which Mr. Galton is said to have proved, may, it seems to me, be partly a result of Panmixia, and partly due to more healthy conditions of life acting on the individual. It is, I presume, a fact, as generally stated, that old armour shows that the knights of the middle ages were rather short men. This may have been a result of natural selection, because, as a rule, the strongest and most active men are rather under than over middle height; while tall men would certainly be more exposed to danger, would have to carry a greater weight of armour, and by thus overloading their horses would be under a disadvantage in battle. Tall men would thus be killed off rather faster than short men; and the same might be the case even after the disuse of armour, so long as rapine and civil war prevailed over a large part of the country. But during the last two centuries of comparative peace tall men have been under no such disadvantage, and their survival may have aided in bringing about the slight increase of average stature which has been observed.

    One other point in Mr. Weldon's communication requires notice. He considers that the frequent occurrence of abnormalities and the wide range of variation in many species, show that "natural selection is in most cases an imperfect agent in the adjustment of organisms." This conclusion does not appear to me to be a logical one, since it ignores the admitted fact of the exceedingly intermittent character of selection and its constantly varied locus of action. Each species of animal is subject to a number of quite distinct dangers--hunger, cold, wet, disease, and varied enemies--and all these are separately intermittent in their action. Some affect the species at one time of the year only, some at another; but most of them only reach their maximum of intensity at long intervals--once or twice, perhaps, in a century. Whether cold winters or hot summers, excessive drought or excessive wet, deep snow or phenomenal hail or wind-storms, all are intermittent and occur with extreme severity only at long intervals. These intermittent waves of meteorological phenomena have their corresponding "waves of life," as Mr. Hudson well terms them, such as phenomenal swarms of locusts or of wasps, of caterpillars, mice, or lemmings, and to a less conspicuous degree of almost every living thing. It follows, that during a succession of favourable seasons variation can go on almost unchecked, and even hurtful abnormalities and imperfections may survive for a few years, but soon there comes a check to the increase, and the most abnormal forms die out; while after a greater or less interval either adverse seasons or an increase of living enemies weed out all the extreme disadvantageous variations, leaving only the pick of the typical form to continue the race. This may occur again and again, each special period of stress affecting different organs or faculties--now abnormal colour, now deficient agility, now again incaution or a weak digestion--till in turn every departure from the best adapted mean form is eliminated, to again arise and again be extinguished as favourable or unfavourable conditions prevail. Thus, I am fully in agreement with Mr. Thiselton Dyer when he said: "I feel more and more that natural selection is a very hard taskmaster, and that it is down very sharply on structural details that cannot give an account of themselves." (Nature, vol. xxxix. p. 9.) The appearance of imperfect adjustment is thus only a temporary phenomenon, while that there is an underlying permanent adjustment is indicated by the long-continued identity of specific characters to which Mr. Weldon refers.

    As it is very important to obtain some direct evidence of the action of natural selection, I wish to suggest a mode of doing so which might probably be successful. There is much evidence to show that the migrating birds which visit us in early summer are very largely old birds which have lived through two or more migrations; and, consequently, that of the large number of young birds which migrate in autumn for the first time a very small proportion return to our shores. If this is so, then the extreme severity of the selection during migration would afford us the opportunity of determining some of the physical characters which influence it, combined no doubt with mental characteristics which we have no means of gauging. I would suggest, therefore, that two or three common species of migrants should be chosen, of which the young birds of the year can be distinguished with certainty. Of these birds a number of observers should collect specimens just before their autumnal migration, and should carefully record the characters fixed upon in the case of the young and old birds separately. Probably the weight, the total length, and the length of the wing, would be sufficient, since heavy birds with comparatively short wings would hardly be adapted for a long-continued flight. By laying down the dimensions of some hundreds of specimens in curves of variation, whatever difference existed between the young and old birds would be easily detected; and this difference would presumably be the difference between the birth-mean and the survival-mean, so far as the selective influence of migration is concerned. In the following spring another set of specimens of the same species should be collected and measured; and we should then perhaps be able to determine the characters which had led to the selection of the young birds which had survived the double migration.