Complete recapitulation does not occur in the development of the digits of the wing. Only a rudiment of a fourth digit has been found in the embryonic wing, not, as might be expected, rudiments of five digits of which two disappear. The metacarpals are free, not united as in the adult, and there are separate distal carpals, which in the adult are united with the metacarpals. In other respects the modifications of wings and sternum are so obviously adaptive that it is difficult to believe that the reduction of digits was not due to disuse. This is another of those cases in which the function to which structure is adapted is constant from the beginning of independent life to the end, and there is some ground for believing that in course of time in such cases embryonic recapitulation may be much diminished or disappear. The period of time since birds were first evolved is in all probability immensely greater than that which has elapsed since the blind fish, Amblyoysis, was modified by cave-life, so that we can understand why the eye is developed to a certain stage in the embryo of the blind fish, although it lives in darkness all its life, while embryonic recapitulation in the wing of the bird is very incomplete.
In another class of adaptations the embryonic or larval stage is adapted to new conditions, while the adult condition is either less changed or not changed at all. One of the most obvious examples of this is the allantois in the Amniota. The embryos of Reptiles, Birds, and Mammals all develop two embryonic or foetal membranes, the amnion and the allantois. Of the function or origin of the amnion little is known: to state that it is protective affords little explanation. It seems possible that it is merely the mechanical result of the weight of the embryo and the development of the allantois. The latter is a precocious hypertrophy of the cloacal bladder found in Amphibia, with the function of embryonic respiration. In the water the amphibian larva respires by means of gills and gill slits. In adaptation to terrestrial life it is necessary, if the free aquatic larval stage is to be eliminated, that the embryo should be able to breathe air before hatching. Various Amphibia show how this requirement was met in various ways. In the South American tree-frogs of the genus Nototrema the eggs are developed in a dorsal pouch of the skin of the female, and within this pouch the respiration of the embryo is carried on by a membranous expansion of the second and third external gills on each side. In the Reptilia the bladder is expanded for the same function, and absorbs oxygen and gives off carbon dioxide through the pores of the shell. It is impossible to reconcile the conception of mutation with the adaptive relation between this allantois and the expulsion of the egg enclosed in a shell on land. The transition probably came about gradually from the deposition of the eggs in moist places but not in water. In the midwife toad (Alytes obstetricans) the male carries the eggs about attached to his legs, respiration is effected by enlarged external gills, and the larvae are hatched in water. In the ancestral reptiles external gills may have helped at first, until by the enlargement of the bladder they were rendered unnecessary. In all such cases the absorption of oxygen must be regarded as the stimulus which caused the enlargement of the respiratory membrane. As the allantois could not be absorbed or retracted again into the abdomen, the umbilicus was evolved—that is to say, the scar formed by the union of the folded edge between the body wall and amnion surrounding the stalk of the allantois. It would he difficult for a mutationist to explain how a mutation should affect the development of the cloacal bladder to such an enormous degree, just when it was required for embryonic respiration, and cause the sides of the body to unite ventrally at the time of hatching, cutting off the allantois and the amnion.
T. H. Morgan [Footnote: A Critique of the Theory of Evolution, p.18.] states that a mutation of gametic origin may affect any stage in the development of the individual. This may be true when there are already distinct stages in the life history. The more important question is whether distinct stages can be caused by mutation. It is true that in heterozygous individuals characters may develop more fully in the adult stage than in the young. But when we find different stages evidently adapted to different modes of life, it is impossible to explain them by mutations affecting different stages of life. In such cases as the larval stages of Insects we find the larvae have become adapted to new habits while the adults have remained unchanged, or have evolved quite independent adaptations. For example, the adults in the chief orders of Insects have the typical three pairs of legs, while the maggots or grubs of the Diptera or Hymenoptera have no legs at all, the caterpillars of Lepidoptera have evolved pseudo-legs on the abdomen, and the larvae of Coleoptera have the ordinary legs and no more. This is the reverse of recapitulation: in the case of legless maggots, and caterpillars with pro-legs, the adult is more similar to the ancestor than the larva. But the same principle holds, that where functions and habits are different, there organs are different. No mutationist has yet produced by breeding experiments a caterpillar without the three pairs of thoracic legs and yet developing into a moth that had normal three pairs. Morgan, with all his mutations of the adult Drosophila, says nothing of mutants possessing legs. The only rational conclusion is that legless larvae have lost the disuse, since those larvae which are destitute of legs do not go in search of food but either live in the midst of it or are fed by others, and that the pro-legs of the caterpillar have been developed by the muscular action of the insect in clinging to leaves. Here again the hormone theory, although we cannot pretend to understand the matter completely, helps us to form a conception of the process of heredity and evolution. The disuse of legs in the larva affects the determinants, so that they remain inactive in the presence of the hormones produced in the body generally in this stage. In the adult stage activity of the legs produces hormones which influence the same determinants in the gametes to develop legs, but again in the presence of the different hormones which are present in the body generally in the adult stage. As the habits of larva and adult became more specialised and contrasted, the change became less and less gradual, and the intermediate stage, not being adapted to any transitional mode of life, became an inactive pupa in which the adult organs develop.
In conclusion I will briefly consider the attempts which have been made to prove the influence of somatic modifications or characters on the gametes by direct experiment. The method of Kammerer of inducing changes of habit or structure by conditions, and then showing that the change is in some degree inherited, has already been mentioned. One obvious criticism of this evidence is that it seems to prove too much, for it is difficult to believe that a change produced in individuals would show so much hereditary effect in their immediate offspring. Two other methods are conceivable by which the influence of somatic hormones might be evident. One of these is to graft ovaries or testes from one animal into another which possesses a certain somatic character, and then to see if the offspring produced from these gonads shows any trace of the character of the foreign soma in which it was nourished. C. C. Guthrie [Footnote: Journ. Exper. Zool. (1908), v.] claimed to have done this in his experiments on hens. He grafted the ovaries of two Black Leghorn pullets into two White pullets of the same breed, and vice versa. The black and the white birds bred true when mated to cocks of their own colour. The black hen with white ovary mated with black cock produced four black chicks and two black chicks with white legs, the white hen with black ovary mated with white cock produced some white chicks, some black and some white with black spots. This is held to prove that the transplanted ovaries were functional, because they produced evidence of the character originally belonging to them. On the other hand, the black hen with white ovary mated with white cock produced nine white chicks, and eleven chicks which were white spotted with black, and the white hen with black ovary mated with black cock produced not black chicks but white chicks spotted with black. This was held to prove that the somatic characters of the "foster mothers" were transmitted.
Davenport repeated Guthrie's experiments on different fowls, grafting the ovary from a cinnamon-coloured hen into a white hen, and mating her with a cinnamon-coloured cock. The chicks were exactly similar to those obtained from crossing such a cock with a normal white hen, and Davenport concludes that the engrafted ovary was not functional but had degenerated. It is known to be almost if not quite impossible to remove the ovary completely from a hen, owing to its close attachment over the great post-caval vein. At the same time it is difficult to see how Guthrie could have obtained black and spotted chicks from a white hen mated with, a white cock if the grafted ovary from a black hen had not been functional. One point which Guthrie does not mention, and of which apparently he was not aware, is that the white of the White Leghorn is dominant to colour, the heterozygotes not being pure white but white with spots. Thus when he mated a black cock with a white hen with grafted ovary and obtained spotted chicks, this would have been the result if the original white ovary was functional. None of his results prove conclusively the influence of the soma of the hen into which ovaries were grafted, but would all be explained if some eggs were derived from the part of the original ovary not removed in the operation, and others from the grafted ovary.
The grafting of ovaries in Mammals has often been tried, but very rarely with success. The introduced ovary usually dies and is absorbed. C. Foa [Footnote: Arch. Ital. de Bid. (1901), Tome xxxv.] states that he made bilateral grafts of ovaries from newborn rabbits into adult rabbits, and two months after the operation one of the operated females was fecundated and produced five normal young. In other cases he placed ovaries from new-born young in positions far from the normal position, such as the space between the uterus and bladder, and in one case the female so treated became pregnant, and when killed had a single embryo in one uterus and no trace of the original ovaries in the normal position. But Foa was not investigating the influence of somatic characters on ova in the grafted ovaries, and does not even mention the characters or breed of the rabbits he used or of the young which were produced from the grafted ovaries. Castle [Footnote: W. E, Castle and J. C. Phillips, On Germinal Transplantation in Vertebrates, Pub. Carnegie Institution in Washington (1911), No. 144.] carried out seventy-four transplantations of ovaries principally in guinea-pigs. Out of all these only one grafted female produced young. In this case the ovaries of two different black guinea-pigs about one month old were grafted into an albino female about five months old. After recovery the grafted female was kept with an albino male. She produced six young in three pregnancies, first two, then one, and lastly died with three foetus in the uteri. All these were black, with some red hairs among the black. One of the first two young had a white forefoot. In this case black is dominant, and therefore there is nothing extraordinary in the offspring from a black grafted ovary being black. The presence of red hairs and a white foot is no evidence of the influence of the foster soma, but is due to imperfect dominance. When the same male was mated with a normal black female the offspring were black with red hairs interspersed.
All these experiments are open to the following criticism. It has been the main argument of this volume that there are two distinct kinds of characters in all organisms—namely, those of somatogenic origin and those of gametogenic origin. Theory supposes that somatic modifications by means of hormones affect the determinants in the gametes. But it is obvious that the black and white of Leghorn fowls and of guinea-pigs are gametogenic characters, and are strongly established in the gametes of their respective varieties. It is not even certain that the black or white hair or feathers are giving off special hormones which would or could influence the gametes. The hormone theory only postulates such influence from hormones issuing from tissues modified by external stimuli. It is quite certain that the black colour in Leghorns or guinea-pigs is not due to any external stimulus or influence. The experiments therefore are entirely irrelevant to what has been called the inheritance of acquired characters. All that they can be said to prove is that an albino soma does not convert ingrafted ova of black race into ova carrying the albino character.
It is probably impossible to prove experimentally the influence of a modified soma in one generation. I have endeavoured to find a case which would not be open to the above criticism—that is, to find a character which could be considered somatogenic and which was absent in a closely allied variety. Most of the characters in domesticated varieties are obviously gametogenic mutations, but the lop-ear in rabbits may be, partly at least, somatogenic. Since many breeds have upright ears, we cannot say that disuse of the external ear has produced lop-ears in domesticated rabbits generally, but in lop-eared breeds the ears are much enlarged; and though this may be gametogenic, the increased weight may have been the cause of the loss of the power to erect the ears. I therefore tried grafting ovaries from straight-eared females into lop-eared individuals. The operation was perfectly successful in seven specimens—that is to say, they recovered completely and lived for many months, up to a year or more afterwards, but none of them became pregnant. When killed no trace of ovary was in any of them; in every case it had been completely absorbed, and the uteri and vagina were diminished in size and anaemic. For grafting I used ovaries from young rabbits of various ages from seven days to six weeks or more, but all were equally unsuccessful. Satisfactory evidence by direct experiment of the inheritance of somatogenic modifications due to external stimuli cannot be said to have been yet produced, and, as I have shown, such evidence from the nature of the case must be very difficult to obtain. The indirect evidence, however, which has been considered in this volume is too strong to be ignored—namely, the case of Japanese long-tailed fowls, that of colour on the lower sides of Flat-fishes, and the similarity of the congenital development of the antlers in stags, to the generally admitted effects of mechanical stimulation and injury on the skin and superficial bones of Mammals.
The general conclusions which are logically to be drawn from our present knowledge with regard to the problems of heredity and evolution in animals are in my opinion as follows:—
1. All attempts to explain adaptation by gametogenic mutations, or changes in gametic factors or 'genes,' have completely failed, as Bateson himself has admitted.