The extent to which characters may be changed through selection varies greatly in different forms. In some cases selection apparently can do nothing at all. In the experiments on corn which have been carried out at the University of Illinois, it has been found that the protein and oil content has been increased in some strains selected for high content and decreased in those selected for low content during the twenty years in which the investigation has been carried on. In Drosophila it has been possible to demonstrate the existence of modifying factors, to determine their location by their linkage relations, and to show their effect on the character under investigation. Sometimes these modified factors may arise as new mutations in the course of selective breeding. They may be located here, there, or anywhere in the chromosome complex. Occasionally successive mutations occur in the same locus of a given chromosome. This has occurred several times in the factor for eosin eye color and a few times in the factor for bar eye.

Selection seems to have not the least effect upon either the location or the nature of the new germinal variations that make their appearance. A scrutiny of the numerous mutations that have arisen in Drosophila shows them to be of the most varied character and quite without definite relation to utility. Most of them are recessive; several are dominant; others partially dominant. They are just the fortuitous variations which Darwin postulated as forming the raw material out of which selection might build up perfected structures. Some of these variations show much greater viability than others. Several of them are lethal, that is, when present in the duplex state they cause the death of the organism. From the standpoint of utility the majority of these mutations must be treated as failures. Occasionally, however, if not in Drosophila at least in many other forms, the new variations that arise are apparently better adapted than their parent stock and tend to supplant it under conditions of free competition.

Recent work in genetics has furnished us with a much clearer picture than we formerly had of the nature and origin of the variations which constitute the material for the transformation of species. Particular heritable variations are the results of changes in localized parts of chromosomes; they behave in general as Mendelian unit factors; they occur sporadically in various parts of the different chromosomes; and while they may affect predominantly one or at least a few features of the body, there is reason to believe that they change to a certain extent the entire constitution of the organism. A variation in a particular locus of a chromosome may produce as its most obvious result a change in eye color, but it may also influence the development of wing characters or general viability. Particular characters are thus the products of a number of germinal factors, and in selection experiments characters may be modified, not through any change in the factor or factors to which the character in question is mainly due, but to changes in other modifying factors whose chief influence is upon quite different bodily structures.

The development of any character is dependent upon a number of elements in the germ plasm. We can no longer speak of a localized determinant of a character in Weismann's sense, as a particle of the germ plasm which produces the character much as a seed gives rise to a plant. The term, unit character, so frequently employed a few years ago, is being replaced by the expression, unit factor. And there is reason to believe that not only is every character the result of many unit factors, but that every unit factor may influence to a greater or less extent a multitude of different characters.

What is now known of variation affords little support to the doctrine of orthogenesis according to which variations are prone to keep on accumulating quite independently of selection along definitely directed lines. Variation in any one direction apparently has not the least

effect upon the direction of the next variation that makes its appearance. Variations have been accumulated along particular lines in the past, as has been abundantly demonstrated by series of fossil forms, and in some cases the changes in structure have been so gradual as to preclude the occurrence of mutations of any considerable extent. But it is quite possible to account for such development by the theory of natural selection. The fact of evolution along specific lines tells us very little of the causes by which such changes have been brought about. Orthogenetic development in the etymological sense of the term may be a fact, but that does not prove that living beings possess any inherent tendency to evolve along straight lines.

I have already alluded to the divergent views regarding the potency of natural selection. While one may argue indefinitely as to whether natural selection can or cannot account for this or that structure, the status of the theory has become, I believe, more firmly established than it was in the time of Darwin. Whatever one may think of the adequacy of natural selection to account for the evolution of organic life, there can no longer be any doubt that organisms are preserved or eliminated on the basis of differences of hereditary constitution. The studies of Weldon on Carcinus and Clausilia, of De Cesnola on mantids, of Bumpus on the English sparrow, of Davenport on birds, of Tower on potato beetles, and of the Morgan school on Drosophila, have shown the actual operation of selective elimination, and the studies of Pearson, Ploetz, MacDonald, and various other investigators of the selective death rate in man indicate that, notwithtsanding our advances in medicine and hygiene, natural selection continues to operate with considerable vigor. The reproach of Lord Salisbury in regard to natural selection that "No man has ever seen it at work" has now been definitely shown to be without foundation.

It must be borne in mind, however, that natural selection is at best only a proximate category of explanation. It is a general term for a multitude of processes which bring about a differential death rate. The old objection so often advanced with an air of novelty that natural selection does not account for variation may be admitted without reserve. Darwin, who was under no illusions in regard to this point, never attempted to explain variation by natural selection. The occurrence of variation is made one of the presuppositions of his theory, and, for aught that Darwin could see, and for aught that we can see now, variations are pretty much haphazard occurrences. They arise no one knows why, and no one can foretell when. When they appear they are subjected to the action of many forces within and without the organism-forces whose action is generally indirect and in many cases obscure-and the fittest survive, and we call the procedure natural selection. According to the theory of natural selection then, the causes of evolution are the agencies, whatever they may be, which cause the hereditary qualities of organisms to vary, and the agencies which favor the elimination of certain variations and the preservation of others. When we have explained evolution in terms of natural selection, therefore, we have made only a first step, although from certain points of view a very important first step, toward a final explanation.

Considering the fundamental importance of the problem, it is surprising that very little investigation has thus far been devoted to the discovery of the causes of variability. We may count on the fingers of one hand the researches of any importance bearing on this topic. Tower has claimed to have obtained variations in the offspring of potato beetles by subjecting the parents to different conditions of heat and moisture during the maturation of the germ cells. Stockard has induced what appeared to be true hereditary variations by subjecting

guinea pigs to the fumes of alcohol. A few other suggestive results have been obtained with higher forms, but what has been accomplished scarcely represents even a good beginning upon the problem. More success has been attained in the induction of variability in unicellular organisms, especially the bacteria, but the relation of the variation produced in these forms to the congenital variability of higher types is a matter of dispute. Much of it, apparently, is akin fundamentally to the somatic variability of multicellular organisms, and its transmission may be due to quite other causes than those by which the inheritance of higher forms is determined.

Discussion of the nature and causes of variability naturally recalls the address delivered by Professor Bateson on the occasion of his presidency of the British Association in 1914. Professor Bateson, the leading English exponent of Mendelism, took as his theme a general survey of recent work in genetics. He pointed out the fact, now amply demonstrated, that much of what had heretofore passed as variability is merely the product of crossing and the subsequent segregation of characters in various new combinations. The appearance of novelty in these variations is, as Professor Bateson assures us, quite illusory. Most apparently new forms are just the kaleidescopic combinations of old elements in new patterns, adding nothing that is really new in substance. Professor Bateson held, however, that not all variation was of this kind. This extreme view had been developed by the Dutch botanist, Lotsy, who, being impressed with the multitudinous variation obtained by crossing distinct types, and by the ways in which variations really due to crossing many generations back appear to arise de novo, attempted to account for all variation, and hence the whole process of evolution, as the result of varied combinations of germ plasm produced by hybridization. But Bateson, alive to the weak points in Lotsy's position, and perceiving that there is