REQUIREMENTS OF A PROOF THAT NATURAL SELECTION HAS ALTERED A RACE IL In its simplest form the theory of natural selection pos- tulates (a) that an individual which is somatically superior has an enhanced probability of prolonged survival as an in- dividual, and therefore of reproduction, because of. this s0- matic superiority; and (6) that, as a purely numerical conse- quence of the enhanced probability of reproduction of such somatically superior individuals, the average condition of the race will tend in time to become altered in the direction of the particular somatic superiority under discussion. Darwin, in The Origin of Species, puts the principle in this way: «(an it then be thought improbable, seeing that varia- tions useful to man have undoubtedly occurred, that other variations useful in some way to each being in the great and complex battle of life, should occur in the course of many successive generations? If such do ocenr, can we doubt (re- membering that many more individuals are born than can possibly survive) that individuals having any advantage, how- ever slight, over others, would have the best chance of sur- viving and procreating their kind? On the other hand, we may feel sure that any variation in the least degree injurious would be rigidly destroyed. This preservation of favorable individual differences and variations, and the destruction of those which are injurious, I have called Natural Selection, or the Survival of the Fittest. Variations neither useful nor injurious would not be affected by natural selection, and would be left either a fluctuating element, as perhaps we see in176 “ SOIENTIA 5, certain polymorphic species, or would ultimately become fixed, owing to the nature of the organism and the nature of the conditions ». IL. In the a priori postulates which underlie the principle of natural selection there are involved a number of logical re- quirements, which I think have never been completely listed and examined. Furthermore, it is an amazing thing in the history of biology since 1859, that so extremely few attempts have ever been made, in fact, to test the principle of natural selection experimentally, or even observationally, by precise and critical specific observations. Considered in all its bearings natural selection is perhaps the most important abstract bio- logical principle ever enunciated. Tons of books and papers have been written about it, Directly and indirectly it has been the guiding light of biological progress since it was first clearly stated. Its effectiveness as a vera causa of organic evolution has been vigorously affirmed and bitterly denied. But singularly little effort has been made either to demons- strate experimentally its operation and the effects of such operation, or to examine into the logical requirements impli- citly demanded in such a demonstration. It is the purpose of the present paper to discuss this latter point, I have been compelled to undertake such an exami- nation of the matter in connection with some studies on na- tural selection in man, in which I have for some time been interested. In actual fact, as will presently be shown by citation of particular cases, the specific investigations of natural selection which have been made, in every case so far as I know, have dealt only with a part of the whole problem. They have mostly shown, for example, that individuals were (or were not) eliminated because of, or at least in possession of, some somatic difference which was either measured or qualitatively observed. But to show that in the struggle for existence the survivors are somatically different from the eliminated, is clearly a long way from proving that in such case the raceREQUIREMENTS OF A PROOF THAT NATURAL SELECTION HAS ALTERED A RACE 177 has been altered as a result of the operation of natural se- lection. But what would a complete proof require? Without further preliminaries let us plunge at once into the discussion of this matter. No attempt will be made to make this discus- sion completely exhaustive. ‘To do so would require a great deal more space than is here available. Furthermore exhaus- tive disenssions of purely logical matters are invariably dull and boring in the highest degree. They become the very quintessence of «the painstaking delineation of the obvious », to use Williams James’ effective characterization of a great deal of the dissertational efforts of earnest but inexperienced graduate students. My more modest aim is merely to call attention to a few points, all of which I think are important, and some of which are certainly commonly overlooked in discussions of this particular aspect of evolution. Il. ‘The basic and minimal logical requisites of an observa- tional or experimental demonstration that natural selection has altered a race in a particular instance, may be subsumed under five general heads, which may perhaps as conveniently as otherwise be labelled as follows: A, Proof of somatie difference between survivors and eliminated. B. Proof of genetic differences between survivors and eliminated. G. Proof of effective time of elimination. D. Proof of somatic alteration of race. E, Proof of genetic alteration of race. Let us consider each of these points in some detail. A. Proof of somatic difference between survivors and eliminated. It is at this point that most of the experimental and observational investigations of natural selection have taken178 “ SOIENTIA ,, their departure. In essence the proof consists in showing that individuals eliminated are different somatically, in some one or more respects, which may be either quantitatively measured or qualitatively described, from individuals which survive. It is quite obviously the first important logical step in the proof of the action of natural selection in a par- ticular ease. The general result of such studies, particularly those of the biometricians, has been to show either (a) that those in- dividuals eliminated by some environmental catastrophe, or by natural enemies, were somatically different from the indi- viduals surviving the same catastrophe or the same natural enemies acting at the same time, or () that the death rate prior to reproduction, or the fertility, of individuals somati- cally defective in their development was different from the death rates, or fertility, of individuals somatically normal. In one or the other of these categories fall the investigations of Bumpus on the English sparrow, de Oesnola on Mantis reli- giosa and Heliz arbustorum, Davenport on poultry (although Pearl with similar material found no significant evidence of selective elimination), Weldon on Carcinus moenas and Clau- silia laminata, and Harris with various plants. But such a result has not been universally obtained. For example, Dewar, as a result of his very thorough obser- vational and bibliographic investigation of the problem of selective elimination among birds in India, comes to the following conclusions: . But the purpose of the present paper is in no sense to teview the results of the study of natural selection. Our sole concern is with the logic of such investigations. This section may then be closed with the statement that plainly item A of the list given above — the proof that eliminated indivi- duals are somatically different from surviving individuals —180 “ SOLENTIA ,, is the first logical step in the demonstration of the action of natural selection in a particular ease, Furthermore, it is the phase of the subject on which most work has been done up to the present time, B. Proof of genetic differences between survivors and eliminated. ‘This vitally important element in any complete demon- stration of the effective action of natural selection in a parti- cular case has been, so far as I am aware, almost completely overlooked or neglected in the discussion and investigation of the subject. Many investigators of natural selection seem generally to have proceeded on the somewhat naive assumption that if a somatic difference is found between the survivors and eliminated in one generation, the succeeding generation will be somatically like the survivors, without ever having tested in any way either the fact as to the inheritance of the parti- cular somatic difference in question, or the precise mechanism of such inheritance if it exists at all. To anyone conversant with modern Mendelian genetics, who thinks about the matter at all, this must indeed appear a naive procedure, For no- toriously it is the case that where the mechanism of inheri- tance is Mendelian it is in many instances wholly impossible to determine the genetic constitution of an individual solely by examination of its soma, Indeed the primary canon of Mendelian experimentation, upon which Bateson so usefully insisted in the early days of Mendelism, is that the experi- mental test in the breeding pen gives the only sure criterion of genetic constitution, and not the somatic condition. The perfect simple example of this principle is, of course, the heterozygote bearing one pair of allelomorphie genes. Somatically such an individual is often indistinguishable from the homozygous dominant individual. This is so, for example, in Mendel’s classic case of yellow and green peas, so far as concerns color. But while heterozygous and homozygous indi- viduals may be, and often are, somatically indistinguishable, and therefore on a complete parity so far as concerns selection‘REQUIREMENTS OF A PROOF THAT NATURAL SELECTION HAS ALTERED ARACE 181 on a somatic basis, they are very different genetically, and in consequence the quantitative effect of the selection in altering the race will be unpredictable and unanalyzable if the only data at hand are somatic (phacnotypic). Since the epoch-making work of Johannsen with beans the principle here under discussion has been fully understood and accepted by geneticists, so far as relates to artificial selection. In the last edition of his Elemente he puts the matter in this way: « Die persinliche, phinotypische Beschaffenheit in bezug auf Vererbung ist irrelevant. Von zwei gleich grossen Bohnen kimnte die eine z. B, einer kleinsamigen Linie angebéren, in dieser also Plusabweicher sein; die andere kinnte gar Minus- abweicher aus einer sehr grossamigen Linie sein. Bohnen mit dem Gewiehte 50 eg. wiirden z. B, in der Linie I Minus- abweicher, in der Linie XIX aber Plusabweicher sein. In bezug auf Grisse persinlich gleich, wiiren sie genotypisch verschieden. Die beiden Bohnen werden — unter gleichen Bedingungen kultiviers — Nachkommen mit wesentlich ver- schiedener mittlerer Samengrisse erhalten ». The futility of selecting somatic differences which do not rest upon corresponding genetic differences is demonstrated by the experiments on artificial selection made by Johannsen. They prove that where there are no genetic differences, there is no alteration of the race by selection of somatic differences, Now, there is no reason whatever to suppose that the same principle does not apply also to the natural selection of so- matic differences. No proof of the effectiveness of natural selection in altering a race can be logically complete until it has been demonstrated that there are genetic differences between survivors and eli- minated, as well as somatic differences. C. Proof of effective time of elimination. The life cycle of most animals and many plants is divided into three periods, as follows: 1. A period preceding reproduction (infancy, adolescence, Prepubertal period, ete.). Vol. XLVIL *182 “ SorENTIA 2. A period during which reproduction takes place, or may take place (maturity, reproductive life, ete.). 3. A post-reproductive or senescent period, during which reproduction does not, and cannot, take place. ‘Thus in man, reproduction does not take place save in very rare instances, statistically speaking, before age 15. Re- production is physiologically impossible before puberty, which rarely occurs before 10 years of age. In 1925, ont of 1,746,909 births in the major portion of the Birth Registration Area of the United States only 1,605, or less than one-tenth of one percent, were born to mothers stated to be under 15 years of age. Only one child of the million and three-quarters was born toa mother stated to be 11 years of age. No child was born to a mother less than 11 years of age. At the other end of life practically no woman ever has children after age 60, and it is physiologically impossible for a woman to reproduce after the involution of the sex organs connected with the menopause has occurred. In 1925 only 61 in 1,746,909 births were to mothers over 50 years of age in the United States. ‘The oldest mother of the year was aged 54. Effective repro- duetive activity of human beings, whether males or females, after age 55 is insignificantly slight, statistically speaking. In 1925, in the United States Birth Registration Area less than one percent of the births were to fathers 55 years of age and over. ‘The bearing of these considerations upon the proof of the action of natural selection in a particular case has been usually overlooked. Obviously if the constitution of a race is to be altered by the action of natural selection, not only must the requirements stated above under A and B hold, but also the following age relations of eliminated and survivors must be fulfilled. No death of an individual ocourring in Age Period 3 (the post-reproductive period) can possibly be selective, in the sense of having any effect upon the race. For all the indi- viduals in this age period have had all the offspring they ever can have. Therefore they have made all the contribution they are ever going to make to the fature of the race. Those dyingAUQUIRENENTS OF A PROOF THAT NATURAL SELECTION HAS ALTERED A RACE 183 in this age period may be both somatically and genetically different from those surviving in this age period, thus fulfill- ing our conditions A and B, but this fact can possibly make no difference to the race. All deaths during Age Period 1 (pre-reproductive) are potentially selective, and are actually so provided our prior conditions A and B (somatic and genetic differences between eliminated and survivors) are fulfilled. The reason is obvious. If an individuai dies (i. ¢., is eliminated) before it reproduces, the germ plasm which that individual bears is permanently substracted from the racial heritage. Whence it follows, if the individuals which are eliminated before they reproduce are somatically and genetically different from tho individuals which survive into the reproductive period, that the mortality is selective. The deaths in Age Period 2 (reproductive period) are in large part potentially selective, and are actually so in the proportion that they occur before the individual has completed reproduction, provided always, of course, that the prior con- ditions A and B are fulfilled, To summarize; it may be said that to establish a logic- ally complete demonstration of the effective action of natural selection it is necessary to have careful regard for the age of eliminated and surviving individuals in relation to their periods of reproduction, in order to be sure that otherwise selective deaths occurred at an age sueh that they could have affected the race. D. Proof of somatic alteration of a race. Even if conditions, A, B and @ have been satisfied, there is still a further requisite for a complete demonstration that in a particular case the race has been altered by natural se- lection. It must be shown by adequate biometrical investig- ation that the race in question is somatically different after the particular event of selection, or after the lapse of a rea- Sonable secular period in the ease of continuing selection, from what it was before. This has rarely been thoroughly184 “ SOIENTIA », done in investigations of natural selection. The investigations of Weldon and Harris seem to me to go the farthest in satis. fying this requirement. E, Proof of genetic alteration of a race. Finally, if conditions A, B, @ and D have been met, it is still necessary to show that the race, following a particular act of selection or after a reasonable secular period in the case of continuing selection, is genetically different from what it was before the selection, if the proof is to be complete. This requirement can only be met, so far as is now known to genetic science, by controlled breeding tests which will show whether or not the constitution of the racial germ plasm has been altered. So far as I am aware this requirement has never been met in any investigation specifically directed to the problem of natural selection. Iv. Because of its philosophical as well as its immediately biological significance the theory of natural selection as a vera causa of organic evolution has been more discussed than any other phase of the subject. Inferential evidence of the operation of natural selection in the organic world has been abundant. Thousands of working biologists have brought forward such inferential and a priori evidence in their published writings, and many others have thought they have observed it, but have not published anything about it. In spite of all this there has been a steadily growing tendency during the past quarterjot a century to doubt whether natural selection has, in real fact, been a factor of primary importance in causing the formation of new species, or in the development of adaptive structures and responses. When such a distinguished fleld naturalist as Dr. Frank M. Chapman, after a monographic investigation (of tropical birds) of extraordinary comprehensiveness and tho- roughness, speaks as he does in the following quotation, itREQUIREMENTS OF A PROOF THAT NATURAL SELECTION HAS ALTERED ARACE 185 must give pause to anyone who upholds the Allmacht of na- tural selection, on the basis of a priori logic solely: «It may be argued that change in environment merely introduces a new set of selective factors which, acting on variations arising wholly independent of the change, perpetuate those that are desirable. «That is, the same variations might have arisen in the old environment and not have been selected, but proving fa- yorable in the new environment they have survived. But I defy anyone to detect the differences between a habitat in the humid Tropical Zone and one in the Subtropical Zone which would give selection the material to work with. ‘The essence of both habitats is luxuriant forest-growth in which color, assured of protection because of its comparative ineonspicuo- usness or because of ever-present opportunity for concealment, runs riot, In general facies the dominant color characters of both tropical and subtropical birds are not unlike. Humming- birds, Trogons, Toueans, Tanagers, and other brilliantly colored birds abound in each zone, not because of the characters which distinguish the zones, but because both zones possess the luxu- riant forest-growth which, as a rule, is essential to the existence of brightly plumaged birds, just as a coral reef is essential to the existence of brightly colored fishes ». But, as has been pointed out earlier, it is not the purpose of this paper to discuss the place of natural selection in the theory of organic evolution. All that I have tried to do is to show, by setting forth in detail a part of the logical requi- rements in the premises, what a difficult and time-consuming task it is bound to be to prove that natural selection has effee- tively altered the course of evolutionary events in a particular case. Fine and useful as has been the biometric and experi- mental work on natural selection, it has, in the best cases, dealt with only one or two of the five aspects of the matter which are here shown to be necessary to a logically eomplete demonstration. And, quite deliberately, this paper deals with only one logical panel, What may be called the environmental panel is not discussed at all. The point here is this. Suppose that186 “sorenTiA.,, in a particular case all of our requirements A to F inclusive had been completely and satisfactorily fulfilled. It would stil] be necessary to show that the changes under D and E had been really caused by the prior events which had oceurred under A, B and , and had not been due to a direct modi fying action of the environment upon soma and germ, going along with, but really distinct from, the selection. It will al- ways be difficult to get such proof, Even under the most favorable of laboratory conditions the demonstration of the precise manner in which an environmental variable produces its results presents a problem of the greatest difficulty, When one is compelled to work with organisms under natural con: ditions the difficulties in the case are bound to be multiplied, Baltimore (U.8.A.), Johns Hopkins University, Institute for Biological Research. Rayrmonp Pzaru