575. I L4 : :tu . . TY V. Digitized by the Internet Archive in 2011 with funding from University of Illinois Urbana-Champaign http://www.archive.org/details/lecturesonherediOOwash LECTURES ON HEREDITY DELIVERED UNDER THE AUSPICES OF THE WASHINGTON ACADEMY OF SCIENCES WASHINGTON, D. C. 1917 PREFATORY NOTE In 1916 a series of four public lectures on nutrition was given in Washington under the auspices of the Washington Academy of Sciences. These addresses were published later in the Jour- nal of the Academy and reprinted in collective form. This plan having met with approval, a series of lectures on the subject of heredity was arranged for the current year, and Dr. H. S. Jen- nings of the Johns Hopkins University, Dr. Oscar Riddle of the Carnegie Institution, and Dr. W. E. Castle of Harvard Univer- sity were invited to address the Academy upon various phases of this subject. Their addresses, delivered in March and April, have since been published in the Journal of the Academy and are here collectively reprinted in conformity with the lectures of last year. Lyman J. Briggs, Chairman, Committee on Meetings, Washington Acaderdxj of Sciences. Washington, D. C, July 20, 1917. 10 CONTENTS* Page Observed changes in hereditary characters in relation to evolution. Prof. H. S. Jennings, Johns Hopkins University, Baltimore, Maryland ^81 The control of the sex ratio. Dr. Oscar Riddle, Department of Experimen- tal Evolution, Cold Spring Harbor, New York. ! 319 The role of selection in evolution. Prof. W. E. Castle, Harvard University, Cambridge, Massachusetts 369 1 The lectures are reprinted from the Journal of the Washington Academy of Sciences, Vol. 7, 1917. Reprinted from the Journal of th e Washington Academy of Sciences Vol. VII, No. 10, May 19, 1917 GENETICS. — Observed changes in hereditary characters in re- lation to evolution. 1 H. S. Jennings, Johns Hopkins Uni- versity. The problem of the method of evolution is one which the bi- ologist finds it impossible to leave alone, although the longer he works at it, the farther its solution fades into the distance. The central point in the problem is the appearance, nature, and origin of the heritable variations that arise in organisms; the changes that occur in the hereditary constitution. I have for a long time been studying the appearance of heritable variations in certain lower organisms. Having satisfied myself as to the nature of the variations that arise in the creatures that I have studied, I have looked about to see what other workers have found; and to determine whether any unified picture of the matter can be made. Can we bring these facts which experi- mental work has brought out into relation with the method of evolution? Can we say that they exclude any particular theory? 1 A lecture delivered before the Washington Academy of Sciences, March 15, 1917. 282 JENNINGS: CHANGES IN HEREDITARY CHARACTERS Can we say that they leave certain views admissible? Can we go farther and say that they make certain views probable? I shall hardly be so bold as even to ask whether they establish any particular views, though even that has been at times affirmed. These questions have, of course, been raised thousands of times; it is only because knowledge does advance, because experimental work has been enormously multiplied of late, that there is reason to bring them up anew. I am going to try to put before you the present situation as it appears to me. What we may call the first phase of the modern experimental study of variation is that which culminated in the establishment of the fact that most of the heritable differences observed be- tween closely related organisms — between the members of a given species, for. example — are not variations in the sense of alterations; are not active changes in constitution, but are per- manent diversities; they are static, not dynamic. This discovery, like that of Mendelian heredity, was, as you know, made long ago by the Frenchman Jordan; but, as in the case of Mendelism, science ignored it and pursued cheerfully its false path till the facts were rediscovered in recent years. All thorough work has led directly to this result : that any species or land of organ- ism is made up of a very great number of diverse stocks, differing from each other in minute particulars, but the diversities in- herited from generation to generation. This result has in recent years dominated all work on the occurrence of variations; on the effects of selection; on the method of evolution. The con- dition is particularly striking in organisms reproducing from a single parent, so that there is no mixing of stocks; I found it in a high degree in organisms of this sort which I studied. Thus the hi usorian Paramecium I found to consist of a large number of such heritably diverse stocks, each stock showing within itself many variations that are not heritable. 2 Difflugia corona, which I have recently been studying, shows the same condition in a marked degree. 3 As you know, a host of workers have found similar conditions in all sorts of organisms. It led to the 2 See Jennings, 1908, 1909, 1910, 1911. (See Bibliography.) 3 Jennings, 1916. (See Bibliography.) JENNINGS: CHANGES IN HEREDITARY CHARACTERS 283 idea of the genotype (Johannsen) , as the permanent germinal constitution of any given individual ; it supported powerfully the conception of Mendelism as merely the working out of recombina- tions of mosaic-like parts of these permanent genotypes. The whole conception is in its essential nature static; alteration does not fit into the scheme. This discovery seemed to explain fully all the observed effects of selection within a species; but gave them a significance quite the reverse of what they had been supposed to have. It seemed to account for practically all the supposed variations that had been observed; they were not variations at all, in the sense of steps in evolution; they were mere instances of the static con- dition of diversity that everywhere prevails. Jordan, the devout original discoverer of this condition of affairs, maintained that it showed that organisms do not really vary; that there is no such process as evolution; and indeed this seems to be the direct logical conclusion to be drawn. In these days of plots and spies, the evolutionists might almost feel that the enemy had crept into their citadel and was blowing it up from within. Now, this multiplicity of diverse stocks really represents the actual condition of affairs, so far as it goes. Persons who are interested in maintaining that evolution is occurring, that selection is effective, and the like, make a very great mistake in denying the existence of the condition of diversity portrayed by the genotypists. What they must do is to accept that con- dition as a foundation, then show that it is not final; that it does not proceed to the end; that the diverse existing stocks, while heritably different as the genotypists maintain, may also change and differentiate, in ways not yet detected by their discoverers. But of course most of the adherents of the "orthodox genotype theory" do not maintain, with their first representative Jordan, that no changes occur; that all is genetically static in organisms. Typically, they admit that mutations occur; that the genotype may at rare intervals transform, as a given chemical compound may transform into another and diverse compound. We all know the typical instances: the transforming mutations of Oenothera; the bud variations that show in a sudden change of 284 JENNINGS: CHANGES IN HEREDITARY CHARACTERS color or form in plants; the dropping out of definite Mendelian units in Drosophila and elsewhere; the transformation of particu- lar Mendelian units into some other condition. So much then may serve as an outline of a prevailing theory; organisms forming a multitude of diverse strains with diverse genotypes; the genotype a mosaic of parts that are recombined in Mendelian inheritance; selection a' mere process of isolating and recombining what already exists; large changes occurring at rare intervals, through the dropping out of bits of the mosaic, or through their complete chemical transformation; evolution by saltations. Certain serious difficulties appear in this view of the matter; I shall mention merely two of them, for their practical results. One is the very existence of the minutely differing strains, which forms one of the main foundations for the genotype theory. How have these arisen? Not by large steps, not by saltations, for the differences between the strains go down to the very limits of detectibility. On the saltation theory, Jordan's view that these things were created separate at the beginning seems the only solution. Secondly, to many minds there appears to be an equally great difficulty in the origin by saltation of complex adaptive structures, such as the eye. I shall not analyze this difficulty, but merely point to it and to the first one mentioned, as having had the practical effect of keeping many investigators persistently at work looking for something besides saltations as a basis for ■evolution; looking for hereditary changes that would permit a continuity in transformation. Some have been searching in the complex phenomena of biparental inheritance; here Castle is to be first named, and in a later lecture you will hear of the views to which he has been led. Others, like Prof. H. F. Osborn, have been searching from this point of view the paleontological records. Others of us have taken up the problem in uniparental repro- duction; it is here that my own work falls, and of this I will for a moment speak. Where reproduction is from a single pai'ent we meet the problem of inheritance and variation in its simplest form; for JENNINGS: CHANGES IN HEREDITARY CHARACTERS 285 there is nothing which complicates genetic problems so enor- mously as does the continual mixing of diverse stocks in biparental inheritance. In uniparental reproduction we have but one genotype to deal with; we can be certain that no hereditary characters are introduced from outside that genotype. To hope for results on the problem in which we are interested, we must resolve to carry on a sort of second degree research, as it were. That is, we must accept as a foundation the facts before discovered, as to the make-up of the species out of a great number of diverse stocks ; as to the usual effects of selection being nothing save the isolation of such preexisting stocks. What we must do is to take a single such stock — choosing an organism that is most favorable for such work — then proceed to a most extensive and intensive study of heredity, of variation, and of the effects of selection for long periods within such a stock. Such an organism, most favorable from all points of view, I found in the rhizopod Difflugia corona. It has numerous dis- tinctive characters, all congenital; all inherited in a high degree;, yet varying from parent to offspring also; none of these char- acters changed by growth or environmental action during the life of the individual. Long continued work showed that a single strain of this animal, a'l derived by fission from a single parent, does differentiate gradual' y, with the passage of generations, into many hereditar- ily diverse strains. The important facts about the hereditary variations and their appearance are the following: 1. Hereditary variations arose in some few cases by rather large steps or "saltations." 2. But the immense majority of the hereditary variations were minute gradations. Variation is as continuous as can be detected. 3. Hereditary variation occurred in many different ways, in many diverse characters. There was no single line of variation followed exclusively, nor in the overwhelming majority of cases. 4. It gave rise to many diverse combinations of characters: large animals with long spines; small animals with long spines; large animals with short spines ; small animals with short spines ; 286 JENNINGS: CHANGES IN HEREDITARY CHARACTERS and so on, for other sorts of combinations of other characters. Any set of characters might vary independently of the rest. 5. The hereditary variations which arose were of just such a nature as to produce from a single strain the hereditarily different strains that are found in nature. 4 I judge that if the intermediate strains were killed, the two most diverse strains found in nature might well be classed as different species, although the question of what a species is must be left to the judgment or fancy of the individual. Such then were the results of my own studies as to the nature of hereditary variations and how they appear. How do these results compare with those found by other men? If we take a general survey, we find the following main classes of cases: 1 . First, we have the mutations of Oenothera and its relatives : large transformations occurring suddenly. Here is evidently one of the most interesting fields of genetics, but I cannot feel, in view of many extraordinary phenomena in this group, that the bearing on the main problems of genetics is yet clear. 2. Second, we have a large miscellaneous collection of muta- tions observed in various classes of organisms: "bud variations," dropping out of unit factors, and the like — all definite saltations, but not genetically fully analyzed. 3. In Drosophila as studied by Morgan and his associates, we have the largest and most fully analyzed body of facts which we possess with respect to changes in hereditary character in any organism. The changes here are pictured as typical salta- tions ; but of these I shall speak farther. 4. In paleontology, as the results are presented in recent papers by Osborn, 6 the evidence is for evolution by minute, continuous variations which follow a single definite trend. 5. Finally we have the work in biparental inheritance from Castle and his associates: 6 this, as interpreted by Castle, gives evidence for continuous variation, not following a single neces- sary trend, but guided by external selection. 4 The full account of this work is given in Jennings, 1916. (See Bibliography.) 5 See Osbokne, 1912, 1915, 1916. (See Bibliography.) 6 See Castle, 1915 a, 1916, 1916 a, 1916 6, 1917; Castle and Phillips, 1914, etc. (See Bibliography.) JENNINGS: CHANGES IN HEREDITARY CHARACTERS 287 Furthermore, we discover in our survey that there are at least two wel' -marked controversies in flame at the present time : First, we have the general controversy between, on the one hand, those who are mutationists and adherents of the strict genotype view; on the other hand those who, like Castle, be- lieve that we observe continuous hereditary variations in the progress of biparental reproduction. The mutationists attempt to show that the apparent gradual modification of characters observed in breeding is in reality a mere working out of Mende- lian recombinations. Here we have contributions by Morgan (1916), Pearl (1916, 1917), MacDowell (1916), Hagedoorn (1914), and others on the one hand; while the full brunt of the attack- is borne on the other side by Castle. Second, we have a somewhat less lively controversy be- tween the genotypic mutationists and the paleontol^gical up- holders of evolution by continuous variation. Echoes of this we find in recent publicat : ons by Osborn and by Morgan. Now let us look briefly into the points at issue in the contro- versy between the "genotypic mutationsts" and the upholders of gradual change during biparental inheritance. Castle finds that in rats he can, by selection, gradually in- crease or decrease the amount of color in the coat passing by continuous stages from one extreme to the other. As to this, he holds two main points: 1 . The change is an actual change in the hereditary character- istics of the stock; not a mere result of the recombination of Mendelian factors. This is the general and fundamental point at ssue. 2 More specificaly, he holds it to be an actual change in a s'ngle unit factor; this s : ngle factor changes its grade in a con- tinuous and quantitative manner. On the other side, the critics of these views maintain that the changes shown are not actual alterations in the hereditary con- stitution at all, but are mere results of the recombinations of Mendelian factors. And specifically, they find a complete explanation of such results as those of Castle in the hypothesis of multiple modifying factors. 288 JENNINGS: CHANGES IN HEREDITARY CHARACTERS The method in which these modifying factors are conceived to operate is doubtless familiar to you : their application to Castle's work with selection in rats will serve as an example. There is conceived to be a single "main factor" which determines whether the "hooded pattern" shall or shall not be present. In addition to this there are a considerable number of "modifying factors" which, when the "hooded pattern" is present, increase or de- crease the extent of pigmentation. When many of the positive factors of this sort are present, the rat's coat has much pigment; when fewer are present the extent of pigment is less, and so on. The process of changing the extent of pigmentation by selection consists, according to this view, merely in making diverse com- binations of these factors, by proper crosses. This same explanation is applied to a great variety of cases. Castle had carried the war into the enemy's country by predict- ing (or at least suggesting) that the so-called unit characters in Drosophila would be found to be modifiable through selection. 7 Later research by MacDowell (1915), Zeleny and Mattoon (1915), Reeves (1916), Morgan (1917), and Sturtevant (1917) actually verified this prediction; it has indeed been found that the Dro- sophila mutations can be modified by selection. Again the mutationists counter the blow with then- explanation of multiple modifying factors, which are segregated in the process of selec- tion; and they give some real evidence that such is actually the case. Now, into the merits of that particular question, as to whether the apparent effects of selection are really due to modifying factors in the manner set forth, I do not propose to enter. Castle maintains that they are not, and I doubt not that he will show you reason for that point of view. At this point my own dis- cussion will diverge from what I judge that he will be likely to give. What I am going to do is to abandon the ground that Castle would defend, proceed directly into the territory of the enemy, accept the conditions met there, then see where we come out in relation to the nature of variation, the effects of selection, and the method of evolution. ' See Castle, 1915, p. 39. (See Bibliography.) JENNINGS: CHANGES IN HEREDITARY CHARACTERS 289 In no other organism have heritable variations been studied so thoroughly as in Drosophila, and no other body of men have been more thoroughgoing upholders of mutationism and of the multiple factor explanation of the effects of selection, than the students of Drosophila — Morgan, Sturtevant, Bridges, Dexter, Muller, MacDowell, and the others. We may therefore turn to the evidence from Drosophila with confidence that it will be presented with fairness to the mutationist point of view. We shall first ask (1) what we learn from the work on Drosophila as to the possibility of finding finely graded variations in a single unit character. Next we shall inquire (2) as to the re- lation of the assumed modifying factors to changes in hereditary constitution; to the nature of the effects of selection. 1. First, then, what are the facts as to numerous finely graded variations in a single unit factor? Here we have certain remark- able data as to the eye-color of Drosophila; data that are of great interest with relation to the nature of evolutionary change. This fruit fly has normally a red eye. Some years ago a variation occurred by which the eye lost its color, becoming white, a typical mutation. Somewhat later, another variation came, by which the eye color became eosin. By those wonderfully ingenious methods which the advanced state of knowledge of the genetics of Drosophila have made possible, it was determined that the mutat : ons white and eosin are due to changes in a particular part of a particular chromosome, namely, of the so-called X-chromosome, or chromosome I. And further, it was discovered that the two colors are due to different conditions of the same locus of the chromosome; in other words, they represent two different variations of the same unit. Moreover, the normal red color represents a third condition of that same unit. Somewhat later a fourth condition of this same unit was found, giving a .color which lies nearer the red, between the red and eosin; this new color was called cherry. So we have four grades or conditions of this single unit character. And now, with the minute attention paid to the distinction of these grades of eye color, new grades begin to come fast. In the November number of Genetics, Hyde (1916), adds two new grades, 290 JENNINGS: CHANGES IN HEREDITARY CHARACTERS one called "blood," near the extreme red end of the series, the other, called "tinged," near the extreme white end; in fact, from the descriptions it requires careful examination to distinguish these two from red and white, respectively. Thus we have now six grades of this unit. And in the same number of the same journal, Safir (1916) adds another intermediate grade, lying between "tinged" and esoin; this he calls "buff." All these seven grades are diverse conditions of the single unit factor, having ts locus in a certain definite spot in the X-chromosome. Such diverse conditions of a single actor are known as multiple allelomorphs. So, up to date we know from the mutationists' own studies of Drosophila that a single unit factor presents seven gradations of color between white and red, each gradation heritable in the usual Mendelian manner. These grades are the following: (1) Red; (2) blood; (3) cherry; (4) eosin; (5) buff; (6) tinged; (7) white. Three of these grades have been discovered in the ast five months. It would not require a bold prophet to predict that as the years pass we shall come to know more of these gradations, till all detectible differences o 1 ' shade have been distinguished, and each shown to be inherited as a Mendelian unit. Considering that the work on Drosophila has been going on only about seven or eight years, this is remarkable progress toward a demon- stration that a single unit factor can present as many grades as can be distinguished; that the grades may give a pragmatically continuous series. The extreme selectionist asks only a little more than this. Besides showing that a unit factor may thus exist in numerous minutely differing grades, this case shows that a heritable varia- tion may occur so small as to be bare'y detectible. Although the variations do not usually occur in this way, the case presents the conditions which would allow of a gradual transition from one extreme to the other, by means of numerous intermediate con- ditions. In a population in which were occurring such minute changes as are here shown to be possible, we could get by selec- tion such a continuous series of gradations as Castle describes in JENNINGS: CHANGES IN HEREDITARY CHARACTERS 291 his rats. The difference in the two cases is, that in Drosophila variations which are large steps occur as well as do the small ones; and that, according to Castle's conception of the matter, such minute heritable variations occur more frequently in the rat than in Drosophila. But on the showing of the students of Drosophila, there is scarcely any other difference in principle between what happens in Drosophila and what Castle believes to happen in the rat. 2. But as we have seen, the mutationists reject the view that the changes in the coat color of the rat are due to alterations in a single unit factor; they explain this and other cases of the effectiveness of selection on a single character by multiple modify- ing factors. Accepting again their contention, the question is shifted to the nature of such factors. What sort of things are these modifying factors? What is their relation to actual changes in the heritable constitution of organisms? Our direct experimental knowledge of these "modifying fac- tors" is scanty. What we have comes again mainly from the studies of Drosophila, so that we need not suspect it of being colored in such a way as to favor the selectionist point of view. We find data as to certain known modifying factors by one of the workers on Drosophila, Bridges (1916), in his recent important paper on non-disjunction of the chromosomes. And here we are taken back again to the series of eye colors, and indeed to one particular member of the series, the middle member, called eosin. s Bridges tells us that he found a factor whose only effect was to lighten the eosin color in a fly with eosin eyes; this factor indeed nearly or quite turns the eosin eye white. This factor Bridges calls "whiting." Another factor has the effect of lightening the eosin color a little less, giving a sort of cream color; this is called "cream b." A third factor dilutes the eosin color not so much; it is called "cream a." In addition to these, Bridges tells us that he has discovered three other diluters of the eosin color; we will call them the fourth, fifth, and sixth diluters. And finally Bridges tells us of another factor whose only effect s Bridges, 1916, p. 148 (See Bibliography.) 292 JENNINGS: CHANGES IN HEREDITAEY CHARACTERS is to modify eosin in the direction of a darker color: this factor he calls "dark." None of these factors has any effect save on eosin-eyed flies. As you see, these things add tremendously to our gradations in eye color. We had already been furnished seven grades, from white to red; now we have seven secondary grades within a single one of these seven primary grades. Our list of grada- tions of eye color in Drosophila therefore takes now the following form: Heritable grades of eye color, Variations that give modifica- due to diverse variations of a tions of the intensity of eosin, single unit located in Chromo- but are located in other chro- some I. mosomes. 1. White 2. Tinged 3. Buff 4. Eosin . . 5. Cherrv 6. Blood 7. Red 1. Whiting 2. Cream b 3. Cream a 4. Fourth diluter 5. Fifth diluter 6. Sixth diluter 7. Dark Let us hasten to add that these seven new grades are not lo- cated in the same unit factor as are the seven primary ones; their loci are in other chromosomes (or possibly in other parts of the same chromosome). Here again then we have minutely differing conditions of a single shade of color, brought about by seven modifying factors. Bridges makes the following remark concerning them: A remarkably close imitation of such a multiple factor case as that of Castle's hooded rats could be concocted with the chief gene eos'n for reduced color, and these six diluters which by themselves produce no effect, but which carry the color of eosin through every dilution stage from the dark yellowish- pink of the eosin female to a pure white. 9 Now this is an extremely interesting statement, one that must arouse the keen interest of the student of the method of evolu- tion. In Drosophila we could get the same sort of graded results that Castle does with his rats, only in Drosophila this is by 9 Bridges, 1916, p. 149. (See Bibliography.) JENNINGS: CHANGES IN HEREDITARY CHARACTERS 293 means of multiple modifying factors, whereas Castle believes that in the rat it is by actual alterations of the hereditary con- stitution! But what are these modifying factors? And here we come to the astonishing point. These modifying factors are themselves alterations in the hereditary constitution. Bridges leaves no doubt upon this point. He lists and describes them specifically as muta- tions; as actual changes in the hereditary material. Where then is the difference in principle between the condi- tion in Drosophila and that in the rat? In Drosophila there occur minute changes in the germinal material, such as to give, so far as our present imperfect knowledge goes, seven diverse grades of a color which is itself only one grade of another series of seven known grades. By means of these graded changes one could obtain, by the mutationist's own statement, the con- tinuously graded results which selection actually gives. What more can the selectionist ask? There are indeed certain differences in detail, in the notions entertained by the different investigators as to exactly where the changes occur. Castle believes that hi the rat the changes occur all in one unit — in one chromosomal locus — giving a series like the primary series for eye color in Drosoph la. The sup- porters of multiple modifying factors believe, on the other hand — if we are to accept Bridges' account of such factors as typical (and it is the only account we have) — they believe, I say, that these minute changes have occurred in some other part of the germinal material. But this difference :s one of mere detail; it does not touch the fundamental question. This fundamental question is as to the occurrence of these minute changes in the hereditary constitution, and as to the possibility of getting therefrom by selection various grades of a given external characteristic. In this, so far as I can see, there is complete agreement. Now, doubtless, there is a further diversity in the mental proc- esses of the two sets of men, in that the mutationist thinks of all these numerous grades as after all essentially discontinuous, as a series of steps so minute that the difference between one 294 JENNINGS: CHANGES IN HEREDITARY CHARACTERS and the next one is not detectible. His opponent, on the other hand, perhaps thinks of the series as actually continuous. But the difference is not a pragmatical one; when steps become so minute as to be beyond detection, the question whether they exist becomes metaphysical. To put the case in brief, if the mutationists are to show that the existence of multiple modifying factors has any bearing on the general question of the effectiveness of selection, they must show that such factors are not themselves minute changes in the hereditary constitution. Not only have they made no attempt to do this, but in the only well-examined cases they state squarely that such factors are indeed alterations in the hereditary constitution. For the inheritance of such factors as Mendelian units, of course absolutely nothing is required save that the location of the change is in a chromosome. No particular degree of magnitude ; no unity of any other kind is required. But there remains one point brought out by the mutationists which ; s of great importance to the student of the method of evolution. While they must admit, by their own account, that all these grades occur, so that a practically if not actually con- tinuous series can be formed, they of course point out that the changes do not occur in a continuous series. In the eye of Droso- phila variation may occur from red to white directly, without any transit onal stages; or from any grade to any other; the con- tinuous scale is obtained only by arranging the steps in order. Therefore, it is maintained, evolution may have occurred by such large steps, not by continuous gradations. 10 This is of course a matter deserving of serious consideration. But cer- tain other points must be considered also. First, the very facts known for Drosophila show that there is nothing to prevent a passage from one extreme to the other by minute changes, just as is held to occur by the paleontologists and selectionists, although change by large steps occurs also. Secondly, in such cases as the eye color of Drosophila we are dealing with char- 10 See particularly the discussion of this point in Morgan, 1916, p. 7-27. (See Bibliography.) JENNINGS: CHANGES IN HEREDITARY CHARACTERS 295 acters that are already highly developed. We know for example, that this particular character is formed by the cooperation of many separate parts of diverse chromosomes; it is a highly complex product of evolution. Now, we find that one or another of these parts may suddenly cease to perform its function, so that the red color is not completely formed ; there is a sudden change in it; or it may disappear entirely. But is this after all strong evidence that in the original production of this complex character with its numerous underlying functional parts, there was the same change by sudden large steps? Indeed, is it not rather true that such destructive changes in a fully formed character could not be expected to throw light on how that character was built up? I am not unmindful of the fact that there are a few — but only a very few — cases in which there is indication of a positive addi- tion by a definite step, as when the eosin color is produced in white-eyed stock. But here again the underlying apparatus has before had the power to produce eosin and other colors. The white color was due to the temporary suspension of function in parts of the chromosomal apparatus, and it may be doubted whether the restoration of this function throws light on the way the apparatus was first developed. To sum up, it appears to me that the work on Drosophila is supplying a complete foundation, for evolution through selection of minute gradations. The so-called "multiple allelomorphs" show that a single unit factor may thus exist in a great number of grades; the "multiple modifying factors" show that a visible character may be modified in the finest gradations by alterations in diverse parts of the germinal apparatus. The objections raised by the mutationists to gradual change through ^election are breaking down as a result of the thoroughness of the muta- tjonists' own studies. We have already gotten completely rid of the notion that the germinal changes consist only in the drop- ping out of complete units, or that they are bound to occur in large steps. If the recent rate of progress is maintained, when such an organism as Drosophila has been studied for fifty years, instead of eight or nine, there will be no conceivable gradation 296 JENNINGS: CHANGES IN HEREDITARY CHARACTERS of any character that will not have been detected. The only outstanding difficulty is the fact that large changes occur as well as small ones ; this seems perhaps due to the fact that we are witnessing the disintegration of highly developed apparatus in place of its building up. In all this, except the last point, the work on Drosophila is in agreement with my own observation of gradual variation in Difflugia; with Castle's similar results on the rat; and with the conclusions of paleontologists as to the gradual development of the characteristics of organisms in past ages. But there is one point in the paleontological conclusions, as set forth in the recent papers by Osborn, which is not in agree- ment with the experimental and observational results on exist- ing organisms; this I wish to notice briefly. Osborn sets forth that in following given stocks from earlier to later ages, characters arise from minutest beginnings, and pass by continuous gradations to the highly developed condition. This seems in agreement with experimental results, as I have tried here to set them forth. Further, according to Osborn, these developing characters do not show random variations in all directions, but follow a definite course, which might seem to have been in some way predeter- mined. And this is emphasized by the fact that the same sorts of characters (horns,' for example) may arise independently, at different ages, in diverse branches of the same stock, and each follow in later ages the same definite course of development. It would appear therefore from this that there must be some directing tendency, some inner necessity which drives a develop- ing organ to follow a definite course. Evolution is characterized by Orthogenesis, as this phenomenon has sometimes been called. Now it appears to me that we do not observe this in the present day experimental work; by selection we can move in more than one direction. I do not mean that the possible variations are not limited by the constitution of the varying organism; they certainly are. But there is no indication, so far as I can see, that the variations push in one determinate direction only. Now, examining the paleontological summaries further as regards this (I refer to Osborn's papers), we find certain points JENNINGS: CHANGES IN HEREDITARY CHARACTERS 297 that appear to modify seriously, if they do not quite nullify, this conclusion that variations follow a determinate course. First, we do find that diverse courses are followed by given characters, in diverse branches of a given group; this is partic- ularly true of the characters of shape and proportion, which Osborn calls allometrons. I take it from the descriptions that this is likewise true at times for structural and numerical characters. A second point which Osborn sets forth is deserving of partic- ular attention. He states, in agreement with Waagen, that in any given geologic stratum, we do find, in addition to char- acteristics that are in the line of determinate descent, other variations from this line, which are of the sort that constitute what we call at the present time varieties; things that are like the diverse races of DifHugia in my own work. But, say Os- born and Waagen, there is a great difference in principle between these and the others, for those which are in the determinate line of progress persist into the next geologic stratum, while the mere varieties do not. The persistent changes were called by Waagen, mutations (in a sense somewhat diverse from that in which the word is used by de Vries). Osborn expresses the opinion that these "varieties" may be merely non-heritable modifications." But in our present geo- logic period we find just such diverging forms, in great number, and we find that their peculiarities are heritable; this I empha- sized in the introductory part of the present discussion. There is then no reason for supposing that these variations were not heritable in earlier geologic periods; there must have been many races heritably diverse, just as there are now; and these are what Waagen called varieties. Now since this is so, the only difference between Waagen's mutations and his varieties, is that, on looking backward at them, we find that the former persisted and the latter did not. But this tells us nothing whatever about why the latter did not. It is perfectly possible, so far as these facts go, that it was a 11 Osborn, 1915, p. 225. (See Bibliography.) 298 JENNINGS: CHANGES IN HEREDITARY CHARACTERS matter of selection by external conditions; many diverse stocks were present, on an equal footing; some were destroyed, others were not. What ground then is there for saying that the development of given characters followed a definite course, as if predetermined? The conditions described are exactly what we should require to find if in past ages there were many varied stocks, some of which were preserved by the action of natural selection. Looking back over the series from a later age, we are bound of course to find it a continuous development. If the same characteristics were favorable in successive ages — and there is no reason why they should not be so — then the same sorts of variations would be preserved in those successive ages ; a line of development once begun would be continued. And if the same sort of characters are favorable ones in different branches of a family, then similar characters may well arise and follow a similar course of develop- ment, in the diverse branches, as Osborn states they do. But at the same time many other heritable variations arise, that are not in the line of progress, and hence are not preserved through selection; these are precisely the "varieties" described by the paleontologists; the diverse races that I have described in Difflugia and Paramecium, and that are found to exist in all organisms. The conditions described by the paleontologists support strongly the theory of evolution by gradual change, but I cannot see that they tend to establish the view that variations show a tendency to follow a definite course, as if predetermined. The paleontologists appear rather to report precisely the con- ditions which we are bound to find if evolution occurs through the guidance of natural selection operating on a great number of diverse variations, the typical Darwinian scheme. There is one other point which I wish I had time to take up, but I have not. I will merely attempt to state in a few words my impression of it. This is the point made by Bateson (1914) in his Presidential Address before the British Association, and farther developed by Davenport (1916) in a recent paper: the proposition, namely, that since practically all observed variations are cases of loss and disintegration, we are driven to suppose JENNINGS! CHANGES IN HEREDITARY CHARACTERS 299 that evolution has occurred by loss and disintegration. Daven- port combines this idea with the theory that these disintegrating variations follow a definite course, predetermined in large measure by the constitution of the disintegrating material. There are two points worth consideration in dealing with this theory. The first is one of fact; although it is true that many of the so-called mutations appear to be cases of loss and disinte- gration, yet there is no indication that this is the case in such effects of selection as have been described by Castle and myself; variations are not limited to any particular direction. Secondly, it appears to me that this conclusion — that because the variations we see are cases of loss and disintegration, therefore evolution must have occurred by loss and disintegration — it appears to me, I say, that this conclusion involves an error in logic, which makes it unworthy of serious consideration. The syllogism which it involves seems something as follows : 1. Major premise. Evolution has occurred by progress from the visibly less differentiated in structure to the visibly more differentiated in structure. 2. Minor premise. By observation we detect only the visibly less differentiated arising from the visibly more differentiated; we see only a process of decreasing the visible differentiation. 3. Conclusion. The visibly more differentiated must have arisen from the visibly less differentiated, by decrease in the visible differentiation of the latter. The conclusion is absurd; it cannot be drawn save for the fact that while in the two premises we are talking of visible differentiation and disintegration, in the conclusion the ground is shifted to mean something entirely different — a sort of inner, invisible, purely theoretical kind of differentiation and simplicity and disintegration. By putting in the word visible all the way through, the absurdity is brought to light. All that we can legitimately conclude from the two premises is that we have not seen the process of evolution occurring. If we have seen nothing but loss and disintegration, this is indeed the conclusion that we must draw. But I believe that we cannot assert that this is all that we have seen. 300 JENNINGS: CHANGES IN HEREDITARY CHARACTERS To summarize then what I have obtained from experimental work combined with a survey of the work of others, the impression left is as follows: 1. Experimental and observational study reveals that organ- isms are composed of great numbers of diverse stocks differing heritably by minute degrees. 2. Sufficiently thorough study shows that minute heritable variations — so minute as to represent practically continuous gradations — occur in many organisms; some reproducing from a sing'.e parent others by biparental reproduction. 3. The same thing is reported from paleontological studies. 4. On careful examination we find even that the same thing is revealed by such mutationist work as that on Drosophila; single characters exist in so many grades due to minute altera-' tions in the hereditary constitution as to form a practically continuous series. 5. It is not established that heritable changes must be sudden large steps; while these may occur, minute heritable changes are more frequent. 6. It is not established that heritable variations follow a definite course as if predetermined; they occur in many directions. 7. It is not established that all heritable changes are by disintegration; although many such do occur, they cannot be considered steps in progressive evolution from the visibly less complex to the visibly more complex. Evolution according to the typical Darwinian scheme, through the occurrence of many small variations and their guidance by natural selection, is perfectly consistent with what experimental and paleontological studies show us; to me it appears more consistent with the data than does any other theory. BIBLIOGRAPHY Bateson, W., 1914. Address of the President of the British Association for the Advancement of Science. Science, 40: 319-333. Bridges, C. B., 1916. Non- disjunction as proof of the chromosome theory of heredity. Genetics, 1: 1-52; 107-163. Castle, W. E., 1915. Mr. Midler on the constancy of Mendelian char- acters. Amer. Nat., 49: 37-42. Castle, W. E., 1915 a. Some experiments in mass selection. Amer. Nat., 49: 713-726. Castle, W. E., 1916. Can selection cause genetic change? Amer. Nat., 60: 248-256. Castle, W. E., 1916 a. Fur- ther studies of piebald rats and selection, with observations on gametic coupling. JENNINGS: CHANGES IN HEREDITARY CHARACTERS 301 Carnegie Inst. Wash., Pub. 241,Part HI: 161-192. Castle, W. E., 1916 6. Genetics and Eugenics. Cambridge. Pp.353. Castle, \V. E., 1917. Piebald rats and multi- ple factors. Amer. Nat., 51: 102-114. Castle, W. E., and Phillips, J. C, 1914. Piebald rats and selection. An experimental lest of the effectiveness of selection and of the theory of gametic purity in Mendelian crosses. Carnegie Inst. Wash., Publ. 195. Pp.56. Davenport, C. B., 1916. The form of evolutionary theory that modern genetical research seems to favor. Amer. Nat., 60: 449-465. Hagedoorn, A. L. and Mrs. A. C, 1914. Studies on variation and selection. Zeitschr. f. ind. Abst. u. Vererb., 11: 145-183. Hagedoorn, A. L., and Mrs. A. C, 1917. New light on blending and Mendelian inheritance. Amer. Nat., 61: 189-192. Hyde, R. R., 1916. Two new members of a sex-linked multiple {sextuple) allelomorph system. Genetics, 1: 535-580. Jennings, H. S., 1908. Heredity, variation and evolution in Protozoa. II. Heredity and variation of size and form in Paramecium, with studies of growth, environmental action and selection. Proc. Amer. Philos. Soc, 47: 393-546. Jennings, H. S., 1909. Heredity and variation in the simplest organisms. Amer. Nat., 43: 321-337. Jennings, H. S., 1910. Experimental evidence on the effectiveness of selection. Amer. Nat., 44: 136-145. Jennings, H. S., 1911. Pure lines in the study of genetics in lower organisms . Amer. Nat., 46: 79-89. Jennings, H. S., 1916. Heredity, variation and the results of selection in the uniparental reproduction of Difftugia corona. Genetics, 1: 407-534. Mac- Dowell, E. C 1915. Bristle inheritance in Drosophila. Journ. Exper. Zool., 19: 61-97. 1916. MacDowell, E. C, Piebald rats and multiple factors. Amer. Nat., 60: 719-742. Morgan, T. H., 1916. A critique of the theory of evolution. Pp. 197. Princeton Univ. Press. Morgan, T. H., 1917. An examination of the so-called process of contamination of genes. Anat. Record, 11: 503-504. Osborn, H. F., 1912. The continuous origin of certain unit characters as observed by a palaeontologist. Amer. Nat., 46: 185-206, 249-278. Osborne, H. F., 1915. Origin of single characters as observed in fossil and living animals and plants. Amer. Nat.. 49: 193-239. Osborn, H. F., 1916. Origin and evolution of life upon the earth. Scientific Monthly, 3: 5-22; 170-190; 289-307; 313-334; 502-513; 601-614. Pearl, Raymond, 1915. Seventeen years selection of a character showing six- linked Mendelian inheritance. Amer. Nat., 49: 595-608. Pearl, Raymond, 1916. Fecundity in the domestic fowl and the selection problem. Amer. Nat., 50: 89-105. Pearl, Raymond, 1917. The selection problem. Amer. Nat., 51:65-91. Reeves, Edna M., 1916. The inheritance of extra bristles in Drosophila melanogaster Meig. Univ. Calif. Pub. Zool., 13: 49.5-515. Safir, S. R., 1916. Buff, a new allelomorph of white eye color in Drosophila. Genetics, 1: 584-590. Sturtevant, A. H., 1917. An analysis of the effect of selection on bristle number in a mutant race of Droso- phila. Anat. Record, 11: 504. Zeleny, C. and Mattoon, E. W., 1915. The effect of selection upon the "bar-eyes" mutant of Drosophila. Journ. Exper. Zool., 19: 515^-529. Reprinted from the Journal of the Washington Academy of Sciences Vol. VII, No. 11, June 4, 1917 GENETICS. — The control of the sex ratio. 1 Oscar Riddle, Department of Experimental Evolution, Cold Spring Har- bor, New York. No better way to introduce a discussion of the present sub- ject has occurred to me than to avail myself of the words — written less than three years ago — with which Professor Don- caster begins his very excellent book on The Determination of Sex: The question "Is it a boy or a girl?" is perhaps the first which is generally asked about the majority of mankind during the earliest hours of their independent existence; and the query "Will it be a boy or a girl?" must equally often be in the mind, even if it is less frequently expressed in words. This second question raises one of the most widely discussed problems of biology, that of the causes which determine whether any individual shall be male or female, and it suggests the still deeper question, "Why should there be male and female at all?" The problem of the nature and cause of Sex ranks in interest with that of the nature and origin of Life, and it may be that neither can be completely solved apart from the other. Not- withstanding the immense amount of brilliant speculation and re- search which has been devoted to the fundamental problem of Life, it must be admitted that hitherto no satisfactory solution has been found, and in some respects the question of Sex is equally obscure. Hardly any other problem has aroused so much speculation, and about few has there been such great divergence of opinion. In one direction, however, the last few years have seen a considerable advance, and we now know at least something of the causes which lead to the pro- duction of one or the other sex, although of the manner in which these causes act our ignorance is still profound. It is but a short step from the question "Is it a boy or a girl?" to the further question "Why is it a girl instead of a boy?" and yet until recently the answer to this latter question seemed hopelessly beyond our grasp, and even now, although some indications of an answer can be given, they do not touch the deeper problems of the real nature of sex. It is a remarkable thing that apart from the fundamental attributes of living matter — irritability, assimilation, growth, and so, forth — no single character is so widely distributed as sex; it occurs in some form in every large group of animals and plants, from the highest to the lowest, and yet of its true nature and meaning we have hardly a suspicion. Other widely distributed characters have obvious func- tions; of the real function of sex we know nothing, and in the rare cases where it seems to have disappeared, the organism thrives to all 1 A lecture delivered before the Washington Academy of Sciences, March 29, 1917, 320 riddle: control of sex ratio appearance just as well without it. . . . . Sex, therefore, al- though it is almost universally found, cannot be said with certainty to be a necessary attribute of living things, and its real nature remains an apparently impenetrable mystery. From the statements of this rather long quotation we shall have occasion, during the present hour, to dissent only from the impression that the problem of the nature of sex offers difficul- ties of a magnitude comparable with those touching the origin of life, and that its mysteries are apparently impenetrable. Studies on the heredity of sex have indeed made great prog- ress during the last fifteen years. These studies have been con- cerned with sex-linkage and with the so-called sex-chromosomes. Since Doncaster, whom I have quoted, is one of the foremost workers in both of these fields of study, it is particularly signifi- cant that, in his opinion, all of the results thus far obtained from breeding and cytology have thrown so little light on the real nature and meaning of sex, and that from further advances in these fields he apparently hopes for so little. For this reason — • and another, namely that the relation of chromosomes to hered- ity and sex, have probably been extensively treated in other lectures of this series — you will, I trust, not now require a lengthy survey of the relations which the chromosomes are known to bear to sex, but will grant most of the present hour for an examina- tion of the results obtained from studies of a quite different sort ; namely, of experimental attempts to control the sex ratio, to learn the nature of sex, and to control the development of sex. It is quite necessary, however, that all experimental studies on the control of the sex ratio should be carried out with the fullest recognition of the known normal association of the chro- mosome numbers — particularly of the sex-chromosomes — with sex, if the results of such studies are to be used toward a decision of the question whether sex itself has been reversed or controlled. The all-important question concerning abnormal or unusual sex ratios is, of course, the question of their meaning — Has a par- ticular germ cell which had initial tendencies to produce one sex been experimentally forced to the production of the opposite sex? Or — a quite different thing — have the conditions of the riddle: control of SE^fcUWft) ^ experiment decreased or suppressed the production, OPftftid&f^l the union, or modified the chromosomal eonste&Wtiofi, df.'dirt J #f the types of germ cell and left the other type normal and >fttW8- tional? These possibilities for accounting for abnormal Sex- ratios certainly exist and they must be squarely met by decisive experiment. The facts from this sphere which we shall most need to bear in mind while examining our series of ratios are as follows : 1. Sperm and ovum are two kinds of sex cells in respect of their origin from the two contrasted sexes; but, beyond this, it is now clear that in some animals one and the same male pro- duces spermatozoa of two kinds, and that these two kinds are not equal in their prospective sex value. Still other animal forms are known in which the female produces two kinds of eggs, having opposite prospective sex value. Most groups of in- sects and several mammals are known to produce two kinds of sperm, while in moths (Lepidoptera) and birds the dimorphism of the germs exists in the eggs. 2. In forms which reproduce in part parthenogenetically — such as the bee, gall fly, plant-louse, etc. — the sex is known to bear certain relations to chromosome number, or to maturation phenomena in the egg. 2 3. In wide crosses among Echinoderms, Baltzer ('09) and Tennent ('12) have shown that when the cross is made in one of the two possible directions, some of the chromosomes proceeding from the sperms are eliminated and do not take part in embryo- formation. This type of chromosome behavior has been found, however, only in crosses of very widely separated forms. Pure wild species of doves and pigeons have proved to be almost ideal material for obtaining highly abnormal sex-ratios, and for the analysis of the meaning or significance of the modi- 2 Even if one fully concedes the "lethal factor" explanation which Morgan (Science, N. S. 36: 718. 1912.) gives for a particular ratio (1 c? : 2 9 ) in Droso- phila, a similar basis could not apply to most of the pigeon series, since herr every egg formed in the ovary can be accounted for, and in numerous series every egg hatched. These same facts, together with the fact that it is thtf.femiik pigeon that is heterogametic, exclude the action of "assortative mating," as u cause of the sex-ratios obtained in the pigeon. 322 riddle: control of sex ratio fied ratios. And since it has now been shown (we shall here attempt to demonstrate the point) that a real reversal or con- trol of sex has been effected in these forms, the large and dimor- phic ova found generally in doves and pigeons have permitted new lines of investigation on the nature of sex itself. As a final result of these studies — a result that we may very briefly indi- cate in advance of the presentation of the data — we believe that it is now reasonably clear that the two sexes are in fact the TABLE 1 On the Relation Between "Width op Cross" and the Sex Ratio MATINGS WIDTH OF CROSS NO. o" NO. 9 SEX-RATIO Columba X orientalis Families 15 K?2) 15.00: 1 (or 7.50: 1) Alba X orientalis (spring' — before July 1) Alba X orientalis (average) Genera Genera Genera Genera 28 58 36 10 43 33 2.80 1.35 1.09 1.22 1 1 1 Average reciprocal crosses 1 Turtur X orientalis, average recipro- Species 14 18 0.78: 1 Same species 36 34 1.06: 1 Same species (-) 18 20 0.90: 1 expression of the rate of protoplasmic activity — of metabolism — pitched at two different heights or levels. For the pigeon world the data seem quite conclusive, and when we shall have reviewed a part of these data we will undertake to place before you the experimentally modified sex ratios obtained elsewhere among animals, in an attempt to show that this considerable body of evidence supplies further confirmation, and only confirmation, for the modifiability of sex, and for our conclusion that the male sex is an expression of metabolism at a higher level, the female sex of metabolism at a lower or more conservative level. riddle: control of sex ratio 323 The work which first showed the remarkable suitability of the wild pigeons for the analysis of the sex-problem was done by the late Professor Whitman who devoted many years to the study of these forms. Whitman obtained indisputably a pro- found modification of the sex-ratio, and identified in a general way some factors associated with the modified ratios. Whether TABLE 2 Sex Ratios and "Width of Cross" as Reported by Various Authors AUTHOR CROSS = DIFFERENT 2 4-29-07 hatch. cfBl 6-23-15 hatch. cfEl 6- 2-07 hatch. o"B2 6-25-15 hatch. c?E2 6- 4-07 hatch. CI 7- 1-15 infert. ' 9F1 7-14-07 hatch. C2 7- 3-15 infert. 9F2 7-16-07 hatch. Dl 7-28-15 infert. cfGl 8-25-07 hatch. D2 7-30-15 8-13-15 infert. infert. 9G2 8-27-07 hatch. El rMl 2-12-08 hatch. E2 8-15-15 infert. 9A2 2-14-0S hatch. Fl 9-10-15 infert. cfBl 3-18-08 hatch. cTF2 9-12-15 hatch. c?B2 3-20-08 hatch. Gl 9-26-15 infert. 9 CI 4-17-08 hatch. G2 9-28-15 infert. 9C2 4-19-08 hatch. HI 10-10-15 infert. 9D1 5-23-08 hatch. H2 10-12-15 infert. 9D2 5-25-08 hatch. 11 10-21-15 infert. 9 El 6-26-08 hatch. ■c?I2 10-23-15 hatch. 9E2 6-28-08 hatch. Jl 11-15-15 infert. 9F1 S- 9-08 hatch. J2 11-17-15 infert. 9F2 8-11-08 hatch. Kl 12-13-15 infert. 9G1 9-20-08 hatch. K2 12-15-15 12-28-15 infert. infert. 9G2 9-22-08 hatch. LI c?Al 3- 2-09 hatch. L2 12-30-15 infert." 9A2 3- 4-09 hatch. " All of the succeeding 64 eggs produced by this pair — under continued "over- work" — have been tested for fertility. Of these 62 were wholly infertile; the other two hatched (both are males). riddle: control of sex ratio 325 eons and that the wider the cross the higher is the proportion of males. Family crosses produce, in nearly all matings, only male offspring. Generic crosses produce from their "stronger" germs — those of spring and early summer — nearly all males. If, however, the birds of such a generic cross be made to "overwork at reproduction," that is if their eggs are taken from them as soon as laid and given to other birds for incubation, then the same parents which in the spring threw all or nearly all male offspring may be made to produce all, or nearly all, female off- spring in late summer and autumn. At the extreme end of the season eggs capable of little, then of no development, are often found in such series. As the parent birds grow older the time of appearance of females, and of eggs incapable of full develop- ment, is reached earlier and earlier in the summer or spring. The relation of "width of cross" to the sex ratio in one of the many species (Turtur orientalis) with which he worked is sum- marized 3 in Table 1 . Practically every gradation from the wid- est possible (family) cross to inbreeding shows a sex ratio in accordance with its position in the series. 4 The "family cross" shown in Table 3 has also produced only males. In Table 2 I have grouped according to width of cross a num- ber of sex ratios reported by various observers. Here again it is found that family crosses yield only male offspring (20 c? : 9 ) ; generic crosses a ratio of 4.9 c? : 1 9 ; specific crosses 4.3 : 1 ; racial crosses 1.9 : 1. The normal sex ratio, i.e., the ratio for any of these species mated to its own kind, is probably nearly 1 : 1 or at most not higher than 1.3 c? : 1 9 . The method of collecting most of these data renders then objectionable as evi- dence on some important questions, and the numbers are small, but they certainly support the generalization that as the "width of the cross" is increased a relatively higher proportion 3 The matings included in this table were continued by the present writer; both earlier and later work (to 1914) are included in the summary. 4 The specific cross — T. turlur and T. orientalis — whose ratio (0.7S:1) is a seeming exception is in reality not an exception. One of the females used in this cross had been previously "overworked" and threw nearly all females as a consequence. For complete data see C. O. Whitman, Posthumous Works, Vol. II, chap. 4. The Carnegie Institution of Washington, (In press.) 326 riddle: control of sex ratio TABLE 4 Breeding Records — 3914 (St. alb i b* X) 9 St. risoria 641 (old); 1913 = 42 eggs. Series 1 9A1 1- 1 White 140 9U1 7-28 White 144 9A2 1- 3 White, dead 2-3 9U2 7-30 White 151 1st (4) = 2.066 g. 2d (4) = 2.243 g. 9 VI 8-14 White 155 HI 4- 4 Inf. yolk = 1.995 g. cTV2 8-16 Dark 169 H2 4- 6 Inf. yolk = 2.105 g. 9WI 8-22 White 152 911 4-12 White, killed 4-29 W2 8-24 Soft at pole 912 4-14 White 158 (?2) 9X1 8-30 White 161 c?Jl 4-21 Dark, killed 2-25 9X2 9- 1 White 145 9J2 4-23 White 158 c?Yl 9- 9 Dark 161 9K1 4-29 White 147 9Y2 9-11 White, killed bef. 10-12 9K2 5- 1 White 151 9Z1 9-18 White LI 5- 9 Broken 9Z2 9-20 White, dead 10-26 ?o"L2 5-11 Dark (disap.?) 9AA1 9-26 White 141 cfMl 5-18 Dark 161 (?1) 9AA2 9-28 White 146 9M2 5-20 White 163 9BB1 10- 7 White 150 9N1 5-30 White 150 9BB2 10- 9 White 144 9N2 6- 1 White, killed with ext. ?9CC1 10-17 Dark, dead 11-8 cfOl 6- 7 Dark 150 9CC2 10-19 White, dead 11-10 9 02 6- 9 White 150 9DD1 10-26 White 130 (?1) oTl 8-18 Dark 149 9DD2 10-28 White 162 (?2) P2 8-20 Broken cfEEl 11- 6 Dark 152 9Q1 6-28 White 143 9EE2 11- 8 White 143 9Q2 6-30 White 137 9FF1 11-16 White 166 9R1 7- 4 White 154 FF2 11-18 Broken cfR2 7- 6 Dark 162 9GG 11-26 White dead 150 . ?c?Sl 7-12 Dark, dead 7-29 9HH1 12- 7 White 9S2 7-14 White, dead 7-31 cfHH2 12- 9 Dark, dead 1-9-15 9T1 7-20 White 140 9 III 12-20 Dark, dead 11-6-15 o*T2 7-22 Dark 164 9 112 12-22 White, 9 da. embr. 1st 17 = 5 cf : 12 9 ; 2nd 17 = 4 □ " : 13 9 ; last 17 = 2 cT : 15 9 riddle: control of sex ratio 327 of males is produced. It may be noted in passing that this generalization touches the question of the nature of sexual differ- ence; for, studies among the most diverse animals and plants have afforded evidences of the "increased vigor of hybrids," of what Darwin called the "good effects of crossing," and of what has been observed in Mendelian breeding as the "greater vigor of the heterozygote." The means of "increasing the vigor" of the offspring are, therefore, the very same means by which higher and higher proportions of males are obtained; and males, we have concluded from other studies, are characterized by a more active metabolism than that found in females'. A glance at Table 3 will assist in making clear some of the advantages which the pigeons afford in the analysis of sex ratios. First, examining the details of the "family cross" — it is an excep- tionally bad history with almost complete infertility — we note that only males are produced, but that a very great number of eggs failed completely to develop. It might be contended that in such a series only the male-producing eggs are fertilized, and for this reason only males are produced. We may fully grant the point; though attention should be directed to the fact that if this were the whole of the story it is rather remarkable that only 4 eggs of the 18 here shown (6 of 88 in the entire series) were fertilized, since it can be proved in any similar series that at least half of the 18 eggs (also half of the 88) were male-producing eggs. And a further point of interest is that while 4 of the first 18 eggs were fertile only 2 of the last 70 eggs — produced under overwork, or crowded reproduction — were fertile. But to recur to the original point — the pigeon in any event affords an oppor- tunity to study the total production of the animal's ovary; and this particular animal's ovary contains all of the sexually differ- entiated germs. In the second section of Table 3 are given the details of a generic cross, a cross of less widely departed forms than in the preceding case. In these crosses practically every egg can be hatched and the sex of the resulting offspring learned. This was done in 23 of the 24 eggs here recorded. This particular record is one of the many made by Professor Whitman from which he 328 riddle: control of sex ratio TABLE 5 Breeding Records- ■1914 (St. alba d* X) 9 St. risoria 647 (young); 1913 = 18 eggs Series $ Al 1- 9 Wt. yolk = 1.515 g. 9 PI 7- 1 White 150 A2 1-11 Wt. yolk = 1.595 g. 9P2 7- 3 White 15-da. embr. Bl 1-28 Wt. yolk = 1.590 g. 9Q1 7- 9 White 148 B2 1-30 Wt. yolk = 1.685 g. c?Q2 7-11 Dark 164 CI C2 2- 8 2-10 Inf. yolk = 1.445 g. Broken 9R1 c?R2 7-22 7-24 White 152 Dark 172 cfDl 9D2 3- 5 3- 7 Dark 8-da. embr. White ?9S1 S2 8- 3 8- 5 White 13-da. embr. Broken 3-da. embr. cfEl 3-19 Dark 167 JT1 8-12 Dark 174 cfE2 3-21 Dark 180 9T2 8-14 White 164 9F1 3-29 White 154 U 8-20 Yolk = 1.490 g. c?F2 3-31 Dark 190 VI 9- 6 "Blood circle" o*Gl 4- S Dark, killed 5-6 c?V2 9- 8 Dark 170 9G2 4-10 White, killed 5-3 ?cfWl 9-19 Dark, dead 10-16 9 HI 4-16 White 153 9W2 9-21 White, dead 10-14 9H2 4-18 White 153 tfXl 9-30 Dark, dead 10-19 cTIl 4-25 Dark 169 9X2 10- 2 White 145 912 4-27 White 154 Yl 10-29 Inf. yolk = 1.845 g .11 5- 5 3-da. embr. killed 9Y2 10-31 White 15-da. embr. .12 5- 7 3-da. embr. killed Zl 12-27 No dev. yolk = 1.870 g. c?Kl 5-14 Dark 169 Z2 12-29 No dev. yolk = 1.925 g. 9K2 5-16 White 158 cfLl 9L2 5-25 5-27 Dark 179 White 164 9 641 d"s = (170 g.) (cf 170 g.) (5) from 1st = 155 g. 9 's (13) = 149 g. cfMl 9M2 6- 3 0- 5 Dark 169 White 11-da. embr. d"s (3) from 2nd = 165 g. 9 's (11) = 150 g. tfNl 9N2 6-13 0-15 Dark 165 White 150 9 647 rf"s = (166 g.) (c?165 g.) (7) from 1st = 170 g. 9 's (5) = 151 g. cfOl 6-22 Dark, killed 7-13 rf"s (5) from 2nd = 175 g. 9 's 02 6-24 Wt. yolk = 1.968 (6) = 158 gr. 1st 18 = 9 c?:9; 9 2nd 18 = 8 d". 10? ; (1915 = 11 o":21 9) riddle: control of sex ratio 329 learned the following facts: (1) Generic crosses, when hot per- mitted to lay many eggs, produce mostly or only males. (2) Such pairs, when made to lay many eggs (crowded reproduction) produce males predominantly from their earlier, stronger eggs, and predominantly or only females from the later eggs laid under stress of overwork. (3) From the eggs of pure wild spe- cies the first egg of the pair or clutch more often hatches a male ; the second egg of the pair more often produces a female. These generic crosses, then, show practically full fertility and exclude the possibility of accounting for the abnormal sex ratio of either spring or autumn by any "assortative mating" of germs, since the sperms by hypothesis are all alike, 5 and all of the ova are fertilized and the resulting sex of all is known. From series of eggs produced by generic crosses, under "over- work" it is therefore practicable to select a certain number of eggs from near the first and from near the last of the season, and have fair assurance that (in this type of mating) most if not all of the earlier lot are prospectively male-producing, and most or all of the later lot are female-producing eggs. It was this possi- bility that enlisted my own first efforts in the study of sex. And, since a single individual ovum or yolk of the pigeon is large enough to permit a chemical analysis — our first study was to determine whether possible chemical differences between the male and female-producing ova exist and are discoverable. The first analyses of the pigeon's ova were made in April, 1911, and the work has been carried on continuously since that time. Nearly 900 individual yolks have now been analyzed. Among these are represented the eggs of several pure species, and of many kinds of hybrids. The records for the chemical composi- tion of the egg-yolks of a considerable number of individual fe- males is now complete for five consecutive years. Altogether, these studies, and the supplementary ones which developed out of them or along with them, have brought to light a number of facts which I can here only briefly sketch. Before considering the results of the analyses it may be well to make clear the nature of a difference which appeared as soon 5 It is certain that the ova are sexually dimorphic. 330 riddle: control of sex ratio as my first lots of yolk samples were placed on the balances for the preliminary weighings. The balances alone and at once showed that the mass of the yolk of the first egg of nearly all pairs of eggs (from pure species) was less by from (usually) 9 per cent to 15 per cent than the mass of the yolk of the second TABLE 6 Weight of Entire Eggs, and op Yolks, 1913-1915 of 9 641 and 9 647 TOTAL EGGS PRO- DUCED EGGS. YOLKS sex 6* : 9 ord. no. wt. + hr.° no. wt. + hr. 1913 1914 1915 (42) (67) (16) 9 641 1st (19)= 8.532 + 25i 2nd (19)= 9.221 + 16 1st (30)= 8.627 + 6i 2nd (31)= 9.275 + l| 1st (7)= 8.584 + 6 2nd ( 7)= 9.290 + 1 9 641 dead 4-17-15 (older) (5) 1.903 + 15i (6) 2.153 + 20i (4) 2.032 + 139 (5) 2.219 + 106 9 &: 8 9 : 1 cf 9 11 o" : 40 9 6 d 1 : 8 9 {all Total 26 & : 56 9 1913 1914 1915 (18) (51) (45) 9 647 1st ( 6)= 7.246 + 6 2nd ( 5)= 8.062 + 2 1st (25)= 7.478 + 9 2nd (24)= 8.403 + 4 1st (22)= 7.624 + 11 2nd (22)= 8.481 + 3 9 647 dead 2-16-16 (younger) (2) 1.482 + 11 (2) 1.535 + 3i (5) 1.653 + 70 (4) 1.793 + 59 (1) 1.715 + 166 (1) 1.970 + 171 Total 1^:6 9 (all late) 17 &: 19 9 11 &: 21 9 1 (?) 29 c? : 46 9 " The entire egg loses weight on standing; the yolk gains weight on standing. egg of the pair. There were occasional reversals of this relation and also occasional pairs with quite nearly equivalent weight. In the eggs produced by hybrids this relation did not obtain at all. Illustrations of these differences in weight between the egg- yolks of first and second egg of the clutch may be seen in any of riddle: control of sex ratio 331 &ZX COA/77POA s4A/£> /TA/OPVA/ (?0/?/?£Zs47yO/VS- /A/ /O/CaTO/US . ■SA'/PWG' s /0 // /2 A3 /«£ I jj • • •••••••••••• V % •. •• % •• •• •• •• •• •• •• •• % •# A/9 2 /vp 3 A/? 4* A/PS A/p6" A/9 '7 A/98 A/? 9 A/P/O /4A/& C?s 8-13 No. dev. 8-15 Dark o" 8-23 Inc. Inc. No dev. 8-25 White 9 9-15 Inc. White 9 9-17 Inc. 259 2.700 White 9 11-29 73.17 21.40 "25.23 55.52 9323 12- 1 260 2.715 73.02 21.63 ■25.38 55.39 9383 " Calculated. riddle: control of sex ratio 333 the appended tables in which yolk weights are given (Tables 4, 5, 6, 7, 8). Other facts concerning the yolk weights which soon came to light were, that the yolks from an individual bird become larger in the autumn, particularly if the bird is made to lay numerous eggs (i.e., overworked) during the season. A schematic repre- sentation of the dimorphism of the ova, and of their increase in size from spring to autumn is shown (under 1) in Chart 1. A further fact of kindred nature was learned when the study was extended over a period of years, namely that the egg-yolks of an individual bird tend to become larger as the bird grows older ; the yolks of the spring, however, are usually smaller than those of the previous autumn, though larger than those of the previous spring (Table 6). These facts are now established by accurate weighings of more than 12,000 yolks, freed and separated from their surrounding shell and albumen. The details of the chemical analyses of one series of eggs ob- tained in 1912 are given in Table 7. These details we need not here consider, but it will be observed that we find larger amounts of the various chemical fractions (excepting water) in the fe- male-producing egg than in the male-producing egg. This holds true alike for the female-producing egg of the clutch, and for the late eggs, which under these conditions are predomi- nantly female-producing, as compared with the group of earlier eggs which under the conditions of the generic cross are rela- tively male-producing. 6 Not only does the size of the egg in- crease with its later position in the series, i.e., with lateness of season, as shown by a mere comparison of the yolk weights of such a series of eggs, but the percentage of energy-yielding or stored materials increases as much, or probably more, than is indicated by the size, or net weight of the yolk. The per cent- age of water, we shall later see, is greater in the male-producing eggs. For our present purpose the importance of the results of these and other analyses is that they conclusively show: (1) that the 6 In this particular series, 8 of the first 9 eggs incubated produced males; the egg of this group that hatched a female was "a very large egg." The last three hatches were females, 334 kiddle: control of sex ratio male-producing egg of the spring is an egg that stores less ma- terial than does the female-producing egg of the autumn. (2) TABLE 8 Stored Energy or Eggs (1914) of Streptopelia risoria ( 9 558) as Determined by the Bomb Calorimeter NO. DATE WT. OF YOLK ENERGY PER CENT DIFF. 665 Al 6- 6 "1.010 "3,358 666 A2 6- 8 0.970 3,175 b - 5.8 674 Bl 6-19 0.855 2,807 675 B2 6-21 1.000 3,245 + 15.6 699 ■ CI 7-14 1 . 145 3,815 ? 700 C2 7-16 1.463 5,008 +31.3 ? 728 D 8-30 1.395 4,812 E 9- 9 or 10 soft shell, bro ken. Fl 10-17 soft shell, bro ken F2 10-19 soft shell, bro ken 770 Gl 11- 6 1.440 4,837 (?) . • 771 G2 11- 8 1.720 5,797 + 19.8 ? 774 HI 11-20 1.590 + si. loss 4,906 + 775 H2 11-22 1.780 6,015 +22.6 ? 776 11 12- 1 1.640 5,614 777 12 12- 3 1.820 6,255 + 11.4 781 Jl 12-12 1.535 5,302 782 J2 12-14 1.690 5,601 + 5.6 791 Kl 12-23 1.485 5,266 (?) 792 K2 12-25 1.718 5,880 + 11.7 ? ° This egg was not only the first laid during season, but first during life of this bird. » b The percentage differences are based upon a value of 100 per cent for the smaller egg of the pair. That the male-producing egg of the clutch stores less material than does its female-producing mate. (3) That the eggs of old riddle: control of sex ratio 335 females store more materials, and — as has been noted — yield a higher percentage of females, than do birds not old. 7 There- fore, it is evident that the egg of female-producing tendency is one whose storage metabolism is high, as compared with eggs of male-producing tendency. Moreover, the analyses show that during the season successive clutches present higher and higher storage, i.e., the earlier clutches store less — are more male-like; the later ones all store more — are more female-like — and as al- ready noted the eggs of the low storage period give rise (in the generic cross) to males, and those of the high storage period produce females. We here obtain a close view of that upon which sex difference rests. And the facts are now quite beyond question. Un- mistakably, less storage and high storage pertain respectively to the male- and female-producing germs. Unmistakably, our pro- cedures, connected with generic cross, season, and overwork, delivers males from the smaller storages in the earlier eggs. Un- mistakably, the procedures raise the storage in all of the later eggs, and unfailingly we then find that these eggs yield only, or predominantly females. And if we eliminate the factor of wide (generic) cross and mate the female with one of her own or a very closely related species (Table 5), then we see that the pro- duction of males and females coincides from the first with two storage values — with two sizes of eggs (yolks) in the clutch — males from the smaller first, females from the larger second. Only after overwork and season have raised the storage value of the eggs is this situation seriously disturbed. And the dis- turbance — associated with an increase in the storage metabolism of all the eggs — delivers as before, an excess of female offspring (Tables 4, 5, 6). The progressive increase in storage capacity of the eggs during the season — under overwork — is to be interpreted as a decrease in the oxidizing capacity of these same eggs. Living cells in general dispose of ingested food material by storing it or by burning it. If oxidized the products of the oxidation are re- movable and do not serve to increase the bulk of the cell. The 7 See Tables 4, 5, 6. 336 RrDDLE: CONTROL OF SEX RATIO low-storage capacity of the male-producing eggs as compared with the high storage capacity of female producing eggs is there- fore an index of higher oxidizing capacity or as more usually stated, a higher metabolism of the male-producing eggs as com- pared with the female-producing eggs. We may next examine the percentages of water in the eggs of spring and autumn, and in the two eggs of the clutch. These figures for one series of analyses are given along with other analytical results in Table 7. They show a higher water con- tent for the eggs of the spring (male-producers). Indeed, each pair of eggs from the first of the season onward has a slightly higher moisture value than the pair that follows it. The analyses further show a higher percentage of water in the first egg of the clutch, i.e., in the male-producer, than in the second or female-producer in all cases. If the results of my nearly 900 analyses all ran as smoothly as do the 8 of this series there would be no doubt of a perfect correlation of high moisture values with small eggs, i.e., with male-producing eggs — both small eggs of season, and small eggs of individual clutches. The results throughout, however, are not so uniform and smooth as here; there are some series which seem seriously to depart from the order noted above. These cannot be adequately discussed here. We can, however, record our own belief that the situation represented in the table is, in the main, indicated by the moisture determinations obtained in the analyses of eggs produced by pure species. Two ad- ditional methods of determining the amount of water in the yolks, give a satisfactory confirmation of the conclusion that the male-producing ovum contains a higher percentage of water than does the female-producing ovum. It may be remarked at once that the two facts — a higher metabolism, and a higher water value in the same egg (the male-producing one) — are not to be regarded as a mere coinci- dence. They are related facts, essentially correlated in that the more hydrated state of these colloids, which contain only 54 to 59 per cent water, is certainly a more favorable state for a higher rate of (oxidizing) metabolism than is the less hydrated riddle: control of sex ratio 337 state which better corresponds to a condition favorable to in- creased storage.* The results of these analyses (as well as the calorimetric deter- minations to be mentioned later) have an important relation to the question of a modified or differential maturation, by which the changed ratios might be explained. Bearing on this point we may here make the following observations: It has been seen that the sex actually realized corresponds in fact to levels or grades of metabolism; and we now note that the (stor- age) metabolism which was measured was complete before the beginning of maturation, so that if such a differential maturation should occur it must be looked upon not as a cause but rather as a result of the establishment of that grade of metabolism which does here, and under all of the several known conditions, in the clearest way accompany and correlate with each particular sex. But, any assumption of a differential maturation, even 'as a result of or response to these impressed levels of metabolism, brings with it more difficulites than it clears up. Among these it brings the paradox of a rigid selection in favor of the male- producing chromosome-complex in the maturations of the spring, and an equally rigid selection against this same complex in the autumn. Again, it is easily shown by simple breeding tests that such differential maturation does not occur in the spring at least when the female is mated to her own or a closely related species; so that a further assumption would have to be made to the effect that it is the prospective fertilization by a sperm from a wider cross that determines the course of matura- tion! Furthermore, our data on the sex-behavior of series of females from such a wide (generic) cross show that if the male- producing complex was indeed eliminated from the eggs that gave rise to one-half of these females (produced under overwork) these same chromosomes cannot be the real or sole cause of 8 For example, Overton found that withdrawal of water from the cells of Spirogyra was followed by an increased storage or accumulation of starch, etc. Embryonic tissues generally have high water content and show most rapid di- vision, differentiation and growth (not storage), etc. 338 riddle: control of sex ratio masculinity, for as we shall see later a part of these females are strongly masculine, and indeed they show various grades of masculinity. The evidence against a differential maturation as a basis for an interpretation of the controlled sex ratios of pigeons is so strong as to cause its rejection, even if the essential constructive facts on the nature and basis of sex had not yet been learned. The storage metabolism of many male- and female-produc- ing ova, both in reference to egg of clutch and to position in the season, has been determined by means of the bomb calori- meter. The method is very accurate and the results are entirely convincing. The stored energy, or heat of combustion, of nearly 400 egg-yolks has been determined. One such series of determinations, (made in 1914) in which all available eggs of a particular female were burned is shown on Table 8. It will there be seen that the first clutch or pair of the season bore a higher caloric value than the second pair, but is otherwise the smallest of the year. Beginning with the second clutch laid in June the succeeding clutches to December 1 bear higher and higher heat values. In all clutches too, except the very first, 9 the second eggs show a higher storage of heat units than do the first of the clutch. Here we find the conclusions reached from studies on the wieghts of yolks, and on yolk analyses, fully confirmed by a method in which the error involved in the de- termination is wholly negligible. The most accurate method for the study of the storage metabolism of male and female producing ova give too the results most consistent with the breeding data. In other words, we could say, if we wished to make merry with our colleagues, the cytologists, that we here get closest to the facts of sex when we burn our chromosomes! The energy values obtained from the burned yolks, permit an indirect comparison of the water values of the male- and female-producing eggs of the clutch. Such a comparison in- dicates, as in the chemical analyses, a higher percentage of 9 Professor Whitman has observed that the very first egg in life or of the season is more likely to throw a. female than is the first of the clutch of the imme- diately succeeding clutches. kiddle: control of sex ratio 339 water in the male-producing ovum. In addition to these two methods of studying the water values of the two kinds of eggs the value has been obtained direct, by desiccation, on a con- siderable number of samples. The three methods confirm each other. A little later we shall make a further application of the observed facts of higher water values, and of a higher metabo- lism in the male-producing ova. Let us now very briefly consider the other kinds of obser- vations that have been made on the series of eggs, from spring to autumn, produced under crowded reproduction by generic crosses, as these are schematically represented in Chart 1. Curves 2 and 3 on that chart represent facts which were first observed by Professor Whitman on these series of eggs. The curve entitled "developmental energy" (No. 2) represents the observed fact that more of the eggs of spring show the capacity to develop than do those of autumn; and by the use of a con- tinuous (not broken) line or curve is indicated the further fact that the first eggs of the clutch bear throughout the season a similar relation (of higher fertility) to the second eggs of the clutch. The curve marked 3 and designated "length of life" tells again of an advantage possessed by the earlier hatched birds, and of a more limited life-term affixed to the hatches from the later "overworked" eggs. It is probable, moreover, that within the group of clutches giving rise to females only, a longer life-term falls to those birds arising from the first egg of the clutch than from those arising from the second of the clutch. Here, then, as in the preceding curve (2), the smaller eggs of both clutch and season are the eggs which give in their develop- ment the tests of "strength and vigor," while the larger eggs of clutch and season more often display "weakness." The data winch justify curves 4 and 5 as represented on the chart have already been considered. Of the observations upon which curve 6 is based we shall here say only that in general the weight of the parent bird is greatest at the season when the weights of the yolks being produced are smallest, and that when the largest yolks of autumn are being produced the weight of the parent bird is the smallest of the year. Tables 4 and 5 340 riddle: control of sex ratio were prepared originally to make clear certain observations on size of off-spring in relation to their origin from eggs produced under overwork, after continued overwork, and in relation to the order of the eggs in the clutch. The tables themselves tell much of their story and we here forego a further considera- tion of them (see Riddle, '16, p. 406). The seventh curve of Chart 1 refers to a long and rather large series of tests of the sex-behavior of series of birds such as those whose origin is indicated in Table 3 (series of 1908). We have here an opportunity to study and compare sex phe- nomena of particular birds whose sex we have reason to believe had been reversed from its initial sex-tendency; that is to say, where successive pairs of females have originated from succes- ive pairs of eggs in the autumn, under overwork, we have the reasons already given for believing that some or most of such females arising from first eggs of the clutch have had their metabolism depressed to a point sufficient to make them fe- males; but the second eggs of the same clutches should by the same means have been carried to a still more "feminine" level; and though both are females, it seemed possible to differentiate the one sort from the other, and this has been successfully done in a series of tests which now extend through a period of nearly five years. Each female has been given about nine tests, each of six months duration, with (for the most part) another female. In this study, then, female is mated with female and male with male. Such pairs, from a very few selected pairs of par- ents, are kept mated for a period of six months. Most of the birds used, for lack of success with the incessantly fighting males, have been females, and most of the nine or ten successive tests with each bird have been made with her own sisters. The members of the pair are kept apart except when under obser- vation; when put together, as is done twice daily, the records are taken of those females of the pair which behave as males in copulating with their mates. Three facts are definitely es- tablished by the data obtained: (1) The females of the orient- alis X alba cross (they are dark in color) are more male-like in their sex behavior than the females of the reciprocal cross (these riddle: control of sex ratio 341 are white in color). (2) Females hatched from eggs laid earlier in the season are more masculine in their sex behavior than are their own full sisters hatched later in the season. And, several grades of females can be thus seriated according to season of hatch- ing. (3) The female hatched from the first egg of the clutch is more masculine than her sister hatched from the second of the clutch in a great majority of the cases. And in nearly all these latter matmgs the more masculine bird is so predominantly mascul ne that she takes the part of the male a full 100 per cent of the time in copulating with her very feminine clutch-mate sister. (See Riddle, '14a). I may remark in passing that the effect of testicular and ovar- ian extracts (suspensions) have been studied in connection with the work on sex-behavior. The results have clearly shown that the sex behavior of a pair of females is modified by the intra-peritoneal injection of testis (pigeon) extract into the one and ovarian (pigeon) extract into the other. In one case, for example, the more "feminine" female of a pair was given testis extract and her more "masculine" mate received ovarian ex- tract. After the injections the bird formerly more feminine, 16 copulations as a male to 23 by her consort, became very much the more masculine, 27 copulations as a male to onlj' 2 by her consort. To one other kind of fact concern ng the effects of reproduc- t've over-work n changing the developmental and sex phen- omena of the germs of the later part of the season, we ask a moment's consideration. It has been found that some females dead at relatively ad- vanced ages show persistent right ovaries. The right ovary in pigeons normally begins degeneration at or before hatching and is usually wholly absent from the week-old squab. In our study it soon became evident that the persistent right ovar- ies were found almost exclusively in birds hatched from eggs of overworked series. Further study has shown in addition that they arise almost wholly from the eggs of autumn, and predominantly then from the second eggs of the clutch — that is from eggs otherwise known to have greatest or strongest 342 • RIDDLE : CONTROL OF SEX RATIO . - female-producing tendency. These ovaries have sometimes weighed half or more than half as much as the adult left ovary with which they were associated, and have been found in such birds dead at all periods from a few days to twenty-four months. We here attempt no adequate description of this situation, but one can not have observed the frequency of the persistence of this ovary in the birds hatched from the eggs otherwise known to be the most feminine from these overworked series, without conviction that the same pressure which carries the eggs of spring from male-producing to female-producing levels, also carries the earlier female-producing level to another yet more feminine. The several kinds of facts just reviewed in connection with Chart 1 afford clear evidence that sex and characteristics other than sex such as fertility and developmental energy not only bear initial relations to the order of the egg in the clutch, but that sex and these other characteristics are progressively modified under stress of reproductive overwork, until at the extreme end of the season certain aspects of femininity are abnormally or un- usually accentuated. In the light of these facts sex reveals itself as a quantitative modifiable character. And an associa- tion of modifiable metabolic levels with the flux and change of sex, or of sex ratios, has been found and described in precisely this same connection. Let us now take these facts with us in a rapid survey of some experimentally induced and puzzling sex-ratios, and also into a brie? consideration of some important facts of sex that have been learned from embryonic and post-natal stages of organisms. The evidence that higher water values and higher metabolism are associated with male-producing eggs, lower water values with female-producing eggs, is of first importance in connection with our own generalization as to the germinal basis of sex- difference; and is further of much interest as being the means of demonstrating that in the — as I believe — several valid cases of sex-control now known, one thing in common has really been effected; this, though the work has been carried out on a con- siderable variety of animals and though the procedures have riddle: control of sex ratio 343 themselves been most various. The tiling that seems to have been effected in all cases has been the raising or lowering of the general metabolism of the treated germs. In probably none of the cases in which these experimentally induced abnormal sex- ratios were obtained — in other animals than the pigeon — has the observer been able definitely to eliminate all the possibilities of the continued determination of sex by the sex-chromosome ; but several observers have been able to eliminate one or more of these possibilities for their material. And all of those experi- TABLE 9 Time of Fertilization and the Sex Ratio in Cattle Thury and Cornaz.. Diising" Diising" time cf: 9 Early 0: 7 Late 22:0 Early 8: 10 Late 4: 1 Early 3: 10 Late 1 : 1 Russell . Pearl and Parshley. Early Late 31:51 42:34 Early 123: 125 Middle 67: 58 Late 65: 42 [Early 134: 178, ratio = 75.3 d" 100 9 Total 6 Middle 67: 58, ratio = 115.5 d" 100 9 [Late 77: 44, ratio = 175.0 d" . 100 9 ° Work cited by Diising. b Omitting the data submitted by Cornaz in the first announcement of the theory. ments which strongly suggest a real sex reversal or control can now be shown to be in alignment with one or more of the basic facts of sex control now known in the doves and pigeons. When the conditions of these experiments have been such as to lead us to expect an increase of the metabolism, males have been pro- duced in excess, and when the conditions imposed have been obviously capable of depressing the metabolism of the treated germs, these have yielded an excess of females. These facts, therefore, afford much reason for the opinion that sex has been controlled or reversed in a number of very different animals. 344 riddle: control of sex ratio The observed relation of the time of fertilization to modified sex-ratios in cattle is summarized in Table 9. Thury reported in 1862 that from fertilizations made in the early period of heat in cattle an excess of females were produced; and that later (delayed) fertilizations give rise to an excess (all according to Thury) of males. Similar experiments have been several times repeated and these repetitions have all shown an excess of one or the other sex in accordance with such early or late fertiliza- tion. 10 The facts as reported by the several observers, and the totals, are given in the table. We postpone for a moment a discussion of the situation presented by these data except to TABLE 10 Time of Fertilization and Sex Ratio in Sheep Bell". Matings in October, 1S99 cT 10:9 26 = 72.0 per cent 9 Matings time unknown, 1899 c? 179: 9 166 = 48.0 per cent 9 Matings after Novem- ber 15, 1899 c? 23: 9 3 = 11.5 per cent 9 ° Records of a neighboring flock supplied to Dr. Bell by Mr. Macrae. draw attention to the probability that in late (delayed) fertiliza- tion the ovum takes up water before fertilization and gives an excess of males. Connected with these facts obtained from cattle are some par- tially similar data for sheep. From records obtained by Dr. Alexander Graham Bell ('14), made primarily with the object of learning whether certain conditions have an influence on "twin- ning" in sheep, the materials for Table 10 have been taken. Here, again, as in cattle there is probably some evidence for an increased male production from delayed fertilizations. Experiments on the frog and the toad have afforded evidence for the control of sex. Richard Hertwig ('06, '12), and later Kuschekewitch ('10), allowed frog's eggs before fertilization to "overripen," a process during which the eggs take up water — 10 The use of the terms early and late fertilizations assume that some ovula- tion occurs either immediately before, or shortly after, the beginning of heat. riddle: control of sex ratio 345 and obtained (the latter author) in some cases a total of 100 per cent males (Table 11). Dr. King ('12) did the converse of this experiment with toad's eggs — withdrawing water from them be- fore fertilization — and obtained nearly or quite 80 per cent of females in cases where the mortality was less than 7 per cent. The evidence afforded by these experiments on the frog and the toad is thought by many to be inconclusive as evidence for real sex control. Though selective fertilization has been elimi- nated as a possibility by Kuschekewitch, we do not know which is the heterogametic sex in amphibia and there also remains the TABLE 11 Experimentally Modified Sex Ratios in Frogs and Toads AUTHOR TREATMENT HESCXT tf C? 9 Hertwig, R 60 o - — Delayed fertil. (+ H 2 0) 271 per eent 100.00 88.88 Kuschekewitch Delayed fertil. (+ H 2 0) 299 100.00 -a 03 o No delay + H 2 62 41 60.20 King, H. D No delay - H 2 106 85 275 289 27.66 22.73 possibility of parthenogenetic development to account for the excessive male-production in the experiments with the frog. But this appeal makes it impossible to explain the great excess of females obtained by Dr. King on the eggs of the toad, where a selective mortality is definitely excluded, and leaves such doubters to lean upon the rather discredited staff of selective fertilization — a proposition wholly disproved for the related frog and for the pigeon. It may be noted, however, that on the basis of our present knowledge of the "sex-differentials" (to be con- sidered later) in the pigeon's eggs both of these experiments 346 riddle: control of sex ratio might have been predicted to result as these three investigators have reported. The modified sex ratios obtained from the four types of ani- mals just mentioned were all obtained through action upon the eggs, or the egg-stage of the organisms. Some important ex- perimental work, and other very significant physiological and chemical study, has been done on sex in the embryonic and adult CHART 2 Bonellia; Free-martin; Inachus; Frog; Pigeon; Duck; Fowl; Pheasant; Sheep; Human; Stag. Cow Egg ■ [ cfhigh per cent H 2 (?) 9 low per cent (H 2 (?) Pigeon \ ■ Frog Toad Hydatina Daphnids Moths flow fat and P. cf \ High Metabolism [high per cent H2O (high fat and P. 9 \ Low Metabolism [low per cent H2O cf high per cent H2O 9 low per cent H2O -+Adult • Human • Fowl Crab [ (blood) low per cent fat High Metabolism f (blood) high per cent 9 I fat [Low Metabolism c? (blood) low fat and P. 9 (blood) high fat and P. cf (blqod) low per cent fat 9 (blood) high per cent fat cf's from change of food and increased oxygen supply. 9 's from unchanged food and lesser oxygen supply. sex-intermediates — sexual or asexual reproduction influenced by conditions. sex-intermediates — quantitative germinal basis. stages of the organism. Something can here be gained by grouping and treating these several results in a single diagram. Now that the basic problem of sex has been shown to be essen- tially a question of metabolism, a department of physiology and biochemistry, we shall be able to note in connection with Chart 2 (where the principal known facts concerning the relation of metabolism to sex are diagrammatically arranged) that a num- riddle: control of sex ratio 347 ber of data bear'ng on adult sexual differance of the sort we most require are already at hand. Turning now to the diagram we note that egg and adult stages are first distinguished. In the egg of the pigeon we have iden- tified maleness and femaleness by three differentials. Female- ness in the egg-stage being accompanied by low metabolism, lower percentage of water, and higher total fat and phosphorus, or of phosphatides. Maleness is here accompanied by high metabolism, higher percentage of water, and lower total fat and phosphatides. Now there are valid reasons for treating these three differentials not as separate and disconnected facts, but rather as aspects or corollaries of the same fact. For example, a TABLE 12 Sexual Differences of Fat and Phosphorus in the Blood of Adult Fowls and of Man Males (roosters) Non-laying females Laying females Males (man) Females (women) . . AVERAGE TOTAL FAT 15.45 17.87 27.80 AVERAGE TOTAL PHOSPHORUS 6.43 7.42 13.15 RELATIVE AMOUNTS OF PHOSPHORUS 100 115 205 141.4 226.0 high metabolism in a cell is consonant with less storage of fat and phosphatides, and with a more highly hydrated state of the cell- colloids. It follows that where data for either of these three differentials are at hand, for either the germ or adult of any ani- mal, we have in such data evidence of the kind we r are looking for, i.e., evidence for the association of a given type of metab- olism with the germ or adult of a given sex. For what forms then are such data available? t And, what is now known of the persistence of this definite type of differentia- tion of the two kinds of sex germs into adidt stages of the two sexes? Recently, in my laboratory in cooperation with Mr. Lawrence ('16), it has been shown that one of these differentials 348 riddle: control op sex ratio — or one aspect of the differential which our own work has dem- onstrated in the egg — is clearly continued in the blood of the adult male and female. Fowls were substituted for doves in this case in order to increase the size of the samples and thus increase the accuracy of the analytical results. The blood of the male contains less fat and less phosphorus — just as the male-producing egg contains less of these same elements. The data further show that the sexually active (or actively functional) females depart most widely from the male, while sexually inactive females occupy an intermediate position in respect of the amounts of these constituents found in the blood (see Table 12). The results afford fairly clear evidence that in birds the meta- bolic differences of male and female germs persist in the male and female adults. In mammals too these aspects of sexual differences of the adults have been fully demonstrated. Almost simultaneously with the above determinations on birds, data were published by Goettler and Baker ('16) which (as we have pointed out, '16) show that the blood of the human male contains less fat, that of the female more. Further, the basal metabolism of the human male and female has recently been accurately determined by Benedict and Emmes ('15) ; they find that the metabolism of man is 5 to 6 per cent higher than that of woman. Have we any measure of either of our differentials in any mammalian egg? I think that the experiments on sex-de- termination in cattle, together with an observation by van der Stricht, afford some evidence that the water content of the male- producing egg is high, and that of the female-producing egg is low. No one definitely knows whether the ovum of the cow absorbs water in the Fallopian tubes in this interval between ovulation and fertilization, but we do know that every amphib- ian, reptilian, and avian egg that has been investigated does absorb very appreciable amounts of water while being passed from the ovary to the exterior. And van der Stricht has de- scribed phenomena of growth or swelling of the yolk-granules of one mammal — the bat — which, I am sure from my own studies on yolk, indicate the taking up of water by the egg of this mam- riddle: control of sex ratio 349 mal. It is highly probable, therefore, that precisely that time relation which leads to an excess of males in cattle is preceded or accompanied by an increased hydration of the ovum. In mammals therefore there is some evidence that a shift of the metabolic level — as indicated by one partly known differential — is associated with the observed changes in the sex-ratio of the germs which are thus modified. Further, in the adult of one mammal — man — two of the three sex-differentials have been definitely demonstrated. These results for both the egg and adult stages of the mammal are at every point in complete agree- ment with our data for both the egg and adult stages of the bird. How now do the controlled sex ratios obtained in the frogs and toads appear in the light of the sex differentials of our diagram? Clearly the data given in Table 11 arrange themselves in per- fect agreement with the metabolic differentials which obtain in birds and mammals. The data of that table eliminate "de- layed fertilization" as such as being a factor and show that the altered sex ratios correspond with increase or decrease of water as the sole known differential. We next give a moment's consideration to an adult stage in which a change in metabolism was observed in connection with sexual changes. In the spider-crabs Geoffrey Smith ('11) showed that both the blood and the liver of the adult male crabs contain less fat than do the blood and liver of the females. Here once more the facts concerning one of the sex-differentials is known and is in complete accord with all the preceding cases. In these spider-crabs, known to be sometimes castrated by para- sites, Smith and Robson were able to show, moreover, that the parasitized male crabs, which under these conditions gradually assume several female morphological characteristics, are also found to have assumed the type of fat metabolism which characterizes the normal female crab. How much these facts contribute to, and how completely they adjust themselves to, our own general theory, will be realized only after a moment's reflection. Re- cently Kornhauser ('16) has found some of these conditions also in Thelia. 350 riddle: control of sex ratio A glance at the diagram indicates three other groups of ani- mals which experimental work has thrown into the general ques- tion of the control of sex. The information at hand for these forms does not so expressly concern the egg as does that from the preceding cases, but all of these latter groups are concerned with early stages — some of them with the generation preceding the egg whose sex seems influenced by conditions. The results of studies of the first of these groups — Hydatina — are of such a kind as to show that they are in general accord with the meta- bolic differentials of all of the previously mentioned cases of sex- control. One can scarcely doubt that change of food and in- creased oxygen supply are consonant with increased metabolism, just as the studies of Whitney ('14 and later) particularly, and later of Shull ('16), have shown that these changes lead to the production of male-producing daughters. The second of these groups — the Daphnids — have been stud- ied by three independent investigators who agree upon two points that are of importance in the question of the control of sex, and to the general theory of sex as stated here, though the results throw little light on precisely what is causally involved. Issakowitch ('05), Woltereck ('11), and Banta ('15) all find numerous sex-intergrades in a material in which all agree that the type of reproduction — sexual or asexual — is influenced by environmental conditions. All further agree that "unfavorable conditions" (or is it a change from favorable conditions?) tends toward sexual reproduction, while "favorable conditions" favor asexual reproduction. In the third of these groups — the moths — the studies of Gold- schmidt ('12, '14), Goldschmidt and Poppelbaum ('14), Harri- son and Doncaster ('14), and the work of Machida, have dem- strated again sex-intermediates of various grades. Moreover, it has been shown that from among the various geographical races of moths certain matings can be arranged which produce rather definite types of male- or female-intermediates — or sex- intergrades as Goldschmidt elects to call them. And further, from pairs involving still other species still other levels or grades of sex-intermediates may be -freely obtained. A more or less fac- riddle: control of sex ratio 351 torial basis of the phenomena has hitherto been used in the discussion of these results; but recently Goldschmidt ('16) has stated that "very important new facts will be published later which will probably enable us to replace the symbolistic Men- delian language, used here, by more definite physico-chemical conceptions." Such newer descriptions — we would say — is wholly in line with the requirements of present data on sex. In Whitman's and our own material it has been clear from the first that the results far overstep the possibility of treating them in Mendelian terms, for it has been apparent from the beginning that we have had to do not with three or four points merely, but with a flowing graduated line. In the work with the moths, however, sex is clearly described in quantitative terms, and it seems fairly certain that when the functional basis of sex shall have been identified it will be found that sex accords with metabolic grades there, as it does elsewhere. It is clear then that all of the animal-forms for which there is reasonable evidence of sex-control show important correspon- dences with the situation fully elucidated in the pigeons. And that where the sex-differentials known to exist in the pigeon's ova have been traced in adults of the two sexes, the parallel rigorously holds there also. A general classification of male and female adult animals on the basis of a higher metabolism for the one and a lower for the other, was indeed made by Geddes and Thomson ('90) many years ago. It now seems beyond ques- tion that this conclusion of these authors is a correct and impor- tant one. It remains to point out that another very old and much worked line of investigation supplies further confirmatory evi- dence for our present point of view. Studies on the effects of castration, gonad-transplantation, and gonad-extract injection, constitute a large body of observations which deal with sexual phenomena associated with the internal secretions of the sex- glands. These internal secretions, let it be remembered, are themselves metabolites, which have the capacity to influence the metabolism of some, many, or of all the tissues with which they OHWWSW Of 352 riddle: control of sex ratio come in contact or which they may reach indirectly." A par- tial list of the animal forms that have been most studied in this respect is written serially on the top of our diagram — in a position intermediate to egg and adult. The number of these animal forms might be much increased, and the names of the investigators of this aspect of the modification of sex are quite too numerous 12 to be mentioned here. But the present point of interest is that these results, as a whole, demonstrate that the extent of sexual modification in the experimental animal is, in general, in proportion to the immaturity of the treated animal. That is to say, the earlier the internal secretion of the gonad is supplied or withdrawn — the earlier the metabolic change is effected — the more profound is the sexual modification of the in- dividual. All this is of course clearly in conformity with the Law of Genetic Restriction — a principle of embryology that is true alike for all of the known characteristics of the organism. Of the several animals of the list we may here particularize concerning only two or three. The stag is a form that has long been known to show thus a considerable and beautiful series of greater modification of antlers and other so-called secondary sexual characters, in correspondence with castration at earlier and earlier periods in the life of the animal. The free-martin — another Ungulate — is now known to exemplify a much ear- lier point at which the foreign internal secretion begins to act; 11 That changes following the removal of gonads, etc., have for many years been recognized as connected with a changed metabolism may be illustrated from the following quotation from Marshall ('10). "The effects of castration indicate that an alteration in the metabolism, even in comparatively late life, may initiate changes in the direction of the opposite sex" (p. 658). 12 The following partial references are suggested by the particular animals listed in the diagram: Stag, Darwin (1868); Caton (1S81); Fowler (1894); Rorig (1900). Human, Hegar (1893); Selheim (1898); Hikmet and Renault (1906); C. Wallace (1907); Tandler and Gross (1909). Sheep, Shattock and Seligman (1904); Seligman (1906); Marshall and Hammond (1914). Guinea- pig, Bouin and Ancel (1903-9); Steinach (1910-13). Pheasant, Gurney (1888). Fowl and Duck, Darwin (1868); Gurney (1888); Foges (1903); Shattock and Seligman (1906-7); Goodale (1910-16). Pigeon, Riddle (1914 a). Frog, Nuss- baum (1907); Pfluger (1907); Steinach (1910); G. Smith (1912). Inachus and Carcinus, Potts (1909); G. Smith (1910-12). Free-martin, Lillie (1916). Bo- mllia, Baltzer (1914). riddle: control of sex ratio 353 and here, true to the rule that has been established elsewhere in all this general line of work/ the resulting modification is cor- respondingly strong and striking. When, by whatever means, we effect a change in the metabolism (which is the essential thing) at a still earlier stage — in the egg-stage in our own and in some other experimental reversals of sex — then we obtain in- dividuals whose sexual nature is quite thoroughly reversed; 13 in many cases completely so, and in still other cases with varying degrees of completeness. Baltzer's ('14) beautiful experiments with the worm Bonellia best illustrate this fact and show the several stages of modifica- tion not only in one and the same animal form but in the in- dividuals hatching from a single brood. Baltzer finds that when the larvae of this animal are hatched they are capable of becom- ing either males or females. If they happen to become attached to the proboscis of an adult female they become males; if they do not succeed in so attaching themselves they soon settle from the water into the sand or mud of the sea-bottom and there undergo, quite slowly, further development into females (almost exclusively). The plastic, reversible, quantitative nature of sex in this form was shown by this investigator in the following way: Some of the free-swimming "indifferent" larvae were artificially helped to a connection with the proboscis of an adult female. Some of these were permitted to maintain this attach- ment for a very short period; others were removed at progres- sively longer periods, with the very significant result that prac- tically all stages of hermaphroditism were -produced. Those first removed becoming almost perfect females, others with longer and longer periods of attachment, becoming more and more perfect males. Now the conditions under which the two sexes are here de- veloped afford, in our own opinion, good reasons for believing that the larva is stimulated — through its contact with the living 13 The observations of Steche ('12) on the relation of precipitin reactions to sex, as seen in the blood of insects are of much interest. This author thus finds that male and female of the same species present differences as great as do the males of two related species, or as do the females of related species. 354 riddle: control of sex ratio tissue — to a higher metabolism; supporting this point of view is the observed fact that "differentiation" is much hastened in this male individual as compared with the otherwise wholly similar larva that is destined to become a female. What it has been our privilege and opportunity to present is in itself but an outline or summary of result obtained in the modification and control of sex, and of the conclusions that seem to follow from these results. In a closing statement, there- fore, we wish only to direct attention to some consequences of the new knowledge of sex. As a foreword to this statement, however, we would note that not only do the widely different kinds of fact to which we have made reference directly support the view of the basis of sex here presented, but that nothing known of the sex-chromosomes is necessarily opposed to this view although an abundance of the data here presented sharply op- pose the conception that the sex-chromosomes are a cause of sex, or that they are even a necessary associated phenomenon. We may conceive that sexually differentiated organisms, from the first, have had the problem of producing germs pitched at two different metabolic levels; and if two sharply opposed sexes are to result from these two kinds of germs then the two metabolic levels must be measurably distinct. This task of producing and maintaining two kinds of cells pitched at two different levels ultimately falls upon cells, and these have, sometimes at least, produced two different chromosome complexes in connection with or in accomodation to the establishment of these two metabolic levels. Eut, as we have seen, the requisite metabolic level of the germ may be established in the absence of the appropri- ate chromosome complex, and the sex of the offspring made to correspond with the acquired grade or level of metabolism. With these facts concerning the functional basis of sex in mind, and reverting to our first quotation from Doncaster, how little wonder that sex (despite its seeming "lack of func- tion" is "nearly universally distributed," almost coequal with "the fundamental attributes of living matter, irritability, as- similation and growth?" Since some grade of metabolism is of necessity universally present in living matter the basis for two riddle: control of sex ratio 355 sexes is of equally wide distribution in that sexual differentiation results from metabolic differentiation, through the establish- ment of two relatively distinct and relatively stable levels of metabolism. In the same way is accounted for the hitherto puzzling fact that the two sexes must have originated many times, scores, hundreds, or thousands of times, within species previously unisexual, during the long period involved in the evolutional history of organisms. Most important of all, perhaps, is the demonstration that one hereditary character is modifiable, is of a fluid, quantitative, re- versible nature. Seemingly this can only mean that other heredi- tary characters are also modifiable. The methods and results of most studies in modern genetics have asked us to accept a quite different view, namely that no such thing as control of heredity may be hoped for, but that we can only look to a sorting and elimination of germs, or of so-called hereditary factors — and to fortuitous origins or recombinations of the latter — to give us better or more desirable organisms. Surely there is a lot of fatalistic philosophy in that conception. All other aspects of function in biology recognize — and some have already attained — the control of life-processes as their aim and goal. Only in this field of heredity — involving the overwhelmingly important processes of continuance and of becoming — has this aim been accepted by a great and growing body of workers as impossible. If sex has been in fact controlled, if it has a modifiable metabolic basis — as now seems assured — then the life processes involved in heredity like other life-processes, invite the investigator to his full and complete task; territory hitherto labelled ''impossible" is open to investigation. BIBLIOGRAPHY Baltzer,.F. Arch. f. Zellforsch., vol. 2, 1909. Baltzer, F. Mitteil. Zool. Stat. Neapel., vol. 22, 1914. Banta, A. M. Year Book, Carnegie Inst. Wash., 1915. (Also Proc. Nat. Acad. Sci., vol. 2, 1916.) Bell, A. G. Quoted from Popenoe. Jour. Hered., vol. 5, p. 47, 1914. Benedict, F. G., and Emmes, L. E. Jour. Biol. Chem., vol. 20, 1915. These authors give full references to the earlier literature. Reprinted from the Journal of the Washington Academy of Sciences Vol. VII, No. 12, June 19, 1917 GENETICS. — The role of selection in evolution^ W. E. Castle, Bussey Institution. Up to the year 1900 those who believed in organic evolution almost without exception believed in selection as its efficient cause. Then came a period of doubt, inaugurated by DeVries' Mutation Theory and strongly supported by Johannsen's Pure Line Theory. In the minds of many biologists at the present time selection is an obsolete agency in evolution and an adequate explanation of evolution is to be found only in muta- tion and pure lines. I believe this to be a mistaken view, not because mutation and pure Lines are false, but because their applicability is very linuted as compared with the broad field of organic evolution. To universalize them is to hide the world by holding a small object close to the eye. For even if we con- cede the strongest possible claim for mutation as an agency in evolution, viz, that it produces all new and heritable variations, it is still unable to produce evolution without the aid of selec- tion. The production of new variations produces no racial change unless those variations persist, but their persistence depends- wholly upon selection. This is admitted by DeVries, the author of the mutation theory, but overlooked by many of those who have adopted the term mutation, as a scientific shibboleth. But it is idle to enter upon a discussion of either selection or mutation without carefully defining these terms, since both are often used quite ambiguously, the latter in particular being used in several different senses, and so being a cause of misunderstanding where no genuine difference of view exists. Ever since DeVries' original attack in 1900, it has become increasingly common among biologists to refer with disrespect to "Darwinian selection." But Darwin understood by selec- tion any agency which would cause one organism to survive rather than another, and it is not clear that any theory of evo- lution can dispense with such an agency. Since more organisms are born than can survive, some must perish. In a state of 1 A lecture delivered before the Washington Academy of Sciences, April 13, 1917. 370 castle: role of selection in evolution nature, that is, in a state of affairs not actively controlled by- man, those creatures survive which are 'best adapted to their surroundings. This is what Darwin meant by "natural selec- tion." Among organisms under the immediate control of man, as the cultivated plants and domesticated animals, where the determination of what individuals shall become parents rests with man, Darwin recognized the occurrence of "artificial selection." Any legitimate attack on Darwin's views of selection must deal either with natural selection or with artificial selection. But when "Darwinian selection" is mentioned as a term of reproach, the attack is really directed neither against natural selection nor against artificial selection, nor against any other conceivable form of selection, but against one of Darwin's views as to the nature of variability. Darwin recognized two sorts of heritable variations, (1) those which are purely quantitative, plus or minus, as compared with the prevailing racial condition, and (2) those which are wholly different from the prevailing condition. The former we may call "fluctuations," adopting the convenient term of DeVries. The latter Darwin often called "sports." Bateson has called them discontinuous varia- tions, and DeVries calls them mutations. Darwin believed that evolution might result either from the systematic and repeated selection of fluctuations or from the propagation of sports. DeVries doubts whether the systematic selection of fluctuations amounts to much in an evolutionary way, and Johannsen has denied to it any evolutionary effect whatever, on the ground that fluctuations are not inherited. Darwin assigned to the selec- tion of fluctuations a major part in evolution, DeVries assigned to it a minor part, and Johannsen allows it no part in evolution. As regards sports, Darwin assigned to their selection a minor part in evolution (chiefly among cultivated plants and domestic animals); DeVries ascribed to a particular kind of sports (his "mutations") a major part in evolution; and Johannsen ascribes an exclusive part in evolution to a type of variation which would include both Darwin's sports and DeVries' mutations and then some. Johannsen has indeed made a new classification castle: kole of selection in evolution 371 of variations which is both logical and sound, but which has resulted in some confusion owing to efforts to combine it with earlier classifications. He classifies variations into those which are inherited (genotypic) and those which are not inherited (phenotypic) . No objection can be made to this classification except that it raises new difficulties and solves none. For how is one to distinguish a phenotypic from a genotypic variation? Only by trying them out. A variation which is inherited is genotypic; one which is not inherited is phenotypic. Since there is no other way then actual experiment by which to dis- tinguish genotypic from phenotypic variations, we acquire only a new set of synonyms for inherited and non-inherited, a thing for which there was no urgent need. Attempts to combine the classifications of variations made re- spectively by Darwin, by DeVries, and by Johannsen have re- sulted in serious confusion which is largely responsible for the apparently contradictory views held at present concerning selec- tion. There really is no diversity of view concerning selection but only concerning the nature of the material that it acts upon (viz., variations). To complicate the situation still farther we have the discovery of Mendelian unit-characters which introduces a new un- certainty. Are these unit-characters fluctuations or sports? Do they arise solely by mutation or also by the cumulation of fluctuations? These are vital but perplexing questions. As matters stand concerning terminology, we have the term "sport," introduced by Darwin but now largely discarded, meaning any discontinuous, striking, suddenly appearing varia- tion, known to be strongly inherited. Some of the examples cited by Darwin, such as the Ancon sheep, obviously involve Mendelian unit-characters. The term mutant as used by DeVries signifies much the same as Darwin's term sport but involves a particular conception of the circumstances and manner of its origin which is not involved in Darwin's term. Some of DeVries' mutants of the evening primrose involve Mendelian unit-characters, as for example his dwarf mutant (nanella), while others such as gigas do not. 372 castle: bole of selection in evolution The latter involves a double representation of every chromo- some in the cell nucleus; the lata mutant involves the presence of a single extra chromosome. What chromosome changes, if any, are involved in other of DeVries' mutants which do not Mendelize is unknown. Morgan has shown that in Drosophila a unit-character change almost certainly involves a change in a definitely localized part of a single chromosome. But he ap- plies the term mutation to each unit-character variation of Drosophila, of which he has observed over a hundred. Some of these are not at all striking, involving only a slight change in the shape, size, venation, or carriage of the wing, which might easily be overlooked by the ordinary observer. Many of them also fluctuate. Hence it is obvious that Morgan's use of the term mutation is very different from that of DeVries, its origina- tor. To Morgan, mutation as illustrated in Drosophila is simply change by a unit-character. With this conception of mutation, Morgan attempts to combine the genotype concep- tion of Johannsen. He regards unit-character variations as the only kind of genotypic variations and these as fluctuating (if at all) only through the interaction of other unit-characters, each one by itself being incapable of fluctuation. It will be observed that as regards the term mutation, we have a very confused state of terminology which results in much discussion at cross-purposes, because persons using the same term have different things in mind. But in this discussion, however confused its terminology, there are really involved two contrasted sets of general ideas, two alternative lines of explanation of evolutionary change, one favored by Darwin, the other offered as a substitute by DeVries and accepted by Johannsen and Morgan. We may briefly outline them as follows: Darwin DeVries 1. New types are for the most 1. New types are created only part created gradually. abruptly. 2. New types are for the most 2. New types are fully stable. part plastic. 3. One evolutionary change fol- 3. One evolutionary change has lows upon and is made pos- no necessary relation to sible by another. another. castle: role of selection in evolution . 373 4. Natural selection determines 4. Natural selection determines what classes of variations only what classes of varia- shall survive and, in conse- tions shall survive, and exer- quence, what shall be the cises no influence on the variable material subjected to subsequent variability of the selection in the next genera- race, tion. 5. The further evolution of our 5. Evolution is beyond our con- domestic animals and culti- trol except as we discover and vated plants (and of man isolate variations, himself) is to some extent controllable because we can by selection influence the va- riability of later generations. These two sets of contrasted views remind us somewhat of the theological ideas of free-will and predestination respectively, which resemblance will account for the preferences of some biol- ogists but will not prove which is right and which is wrong. This is wholly a matter for evidence. But what conclusion one reaches will depend much upon what sort of evidence he studies. Paleontology, geographical distribution, classification, and experi- mental breeding, all present evidence which must be weighed before a safe verdict can be framed. Paleontology, the study of the actual historical records of evolution found in the rocks, indicates in the case of the most complete series of fossils, as for example of the horse, the camel, and the rhinoceros, that the evolution of these types was a gradual process, though of course their appearance in particular continents may have been abrupt, owing to migration. It indicates further that these and other types, when they first appeared, were plastic, and generalized and varied in many different ways, most of the variations later disappearing and leaving only a favored few lines of specialized survivors. It shows too that one variation paved the way to another. The five-toed horse first becomes four-toed, then three-toed, then one-toed. There is no mutation from five-toed to one-toed, nor from the size of a fox to that of a draft horse. As to natural selection, paleontology is silent, because the causes of extinction are unknown. But on the whole the weighty evidence of pale- 374 castle: role of selection in evolution ontology supports the view that evolution as an age long process has been gradual and progressive, not abrupt and unguided. Geographical distribution and classification favor the same idea. Related species are most often found in contiguous terri- tory. Species not closely related are commonly far separated in space or have been long separated in time. Nothing indicates that of two related species one has sprung suddenly from the other. They are not distinguished from each other, as a sport from its parent form, by some single Mendelian unit-character, but they differ morphologically by a large number of quantita- tive differences, and physiologically they differ to such an extent that frequently they will not interbreed when brought together even though their morphological differences are small, or they will produce sterile hybrids, or those of a blended, intermediate character. In all these particulars they show that they have not diverged by mutation, either in the sense of De Vries or in that of Morgan, but by a gradual progressive process. Finally we come to the evidence from experimental breeding. Some say that this is the only legitimate evidence as regards the method of evolution because it alone is experimental. I should be the last to deny its importance because I have devoted much time to its pursuit in the firm conviction that it could yield valuable evidence, but frankness compels one to admit that this method of study, like all the others, has hmitations of its own. The experimental breeder can study a few successive generations with an intensiveness that is possible by no other method, but his glimpses of evolution at work are momentary as compared with the studies of the paleontologist. He can witness the pro- duction of new sorts but it is doubtful whether any man has witnessed the contemporary production of a new species, in the sense of the paleontologist and the student of geographical distribution. Evolution is undoubtedly at work all the time, but the breeder is not always in a position to say just what is happening. It takes a succession of views in a motion picture to show what objects are stationary and what are moving, and the breeder's view of the evolutionary process often fails to reveal which is which. castle: r6le of selection in evolution 375 On the other hand, the experimental breeder, though he lacks perspective, is dealing with the actual material concerned in organic evolution. He can see and handle it and observe it change under his hands, as no other student of evolution can. But the changes which he observes taking place must be cor- rectly interpreted if valid conclusions are to be reached concern- ing the general process of evolution. At present experimental breeders are divided in their views. The very same facts are interpreted by some as indicating an orderly progress toward definite end results, and by others as nothing but haphazard unrelated chance occurrences. Just now the latter method of interpretation, embodied in the mutation theory, is very popular among experimental breeders, although it has few adherents among students of paleontology, classification, or geographical distribution. The principal tools of the experimental breeder are hybridiza- tion and selection. All are agreed that hybridization (using the term in its broadest sense) is, in the hands of the breeder, a very potent agency in producing variability, upon which selec- tion may then be brought to bear for the production of new or modified types. Lotsy even goes so far as to suggest that all genetic variability is the result of hybridization, but this is flatly disproved by observations of Johannsen who reports the occurrence of mutations in genotypically pure lines of beans, as also by the remarkable series of variations observed by Mor- gan in an inbred race of Drosophila. As regards the action of selection, the most widely divergent views are held by experimental breeders. The mutationists hold that it can do nothing but isolate variations which may sporadically put in an appearance or which may by hybridiza- tion be brought together into new combinations. Those who differ from them, and whom they call selectionists, maintain that selection can accomplish more than the mere isolation of variations because it can, by a series of selections, influence further variability. I confess myself an adherent of this at present somewhat unpopular view. I hold it, not because Darwin, held it, nor merely because paleontologists, systematists, 376 castle: k6le of selection in evolution and students of geographical distribution in general favor it, nor because DeVries and Johannsen have attacked it, but be- cause the facts of experimental breeding, as I understand them, prove it. For DeVries may be claimed the merit of having first system- atically set about testing the effects of selection by actual experi- ment. His methodical selections for many years in succession of maize, buttercups, striped flowers, and four-leaved clover will long be remembered, but they fall far short of conclusiveness because they were not continued long enough to show whether selection had attained all that was attainable under existing variability or whether further variation in the direction of selec- tion would occur, and because DeVries' cultures were not suffi- ciently guarded from hybridization which might conceivably influence the result. These necessary precautions were fully met by Johannsen, who in the case of beans, which are self fertilizing but show fluctuating variation in the size of the seed, proved that selection generation after generation in a particular direction may be without result, so far as any change in average seed size is concerned. Cases of this sort involve "pure lines," those which are devoid of genetic variation to any appreciable extent in the character studied, size of seed. But in other cases, as where Johannsen made his size selections from a field crop harvested from many different plants, he found that average size was influenced by selection, which he reasonably explains on the ground that the material from which selection was made con- sisted of a mixture of pure lines genetically distinct. The correctness of Johannsen's conclusion has been repeatedly veri- fied in the case of other self-fertilizing plants such as wheat and oats. Attempts were at once made to generalize Johannsen's brilliant demonstration of the principle of pure lines in the following ways: 1. Since a line of beans long self -fertilized is devoid of genetic variation in seed size, self-fertilization, if long enough continued, will produce lines genetically pure as regards all characters. Selection can not bring about modification of such pure lines. In respect to this generalization it may be said that it remains to castle: role of selection in evolution 377 be shown that beans are as devoid of genetic variation in other particulars as they are in seed size, which the argument assumes to be true. Further, if various pure lines of beans have come into existence by an evolutionary process (descent from a common ancestor, with modification) it is evident that differences must have arisen which did not originally exist. Suppose we grant Johannsen's (unproved) contention that such differences arise by mutation only. If they arise in this way (or in any other way whatsoever), selection can isolate them, and if they are at all frequent in occurrence, selection can be continuously effective in producing racial changes. It would all come down then to a question of how frequent mutations are in a particular case. Johannsen concedes their occurrence even in beans. It may well be that in some organisms they are commoner than in others and that in beans they happen to be particularly infrequent. 2. Johannsen's case has been further generalized to include all self-fertilizing organisms, which are supposed to fall auto- matically into pure lines (i.e., those devoid of genetic variation) as regards all characters. This too requires proof, but has been found to be a safe working hypothesis in the case of cereals, tobacco, peas, and other economic crops, in the attempted im- provement of which selection of fluctuations, unless preceded by hybridization may be regarded as a waste of time, for the reason that genetic variation is so rare under continuous self- fertilization that the breeder will obtain variation much more quickly by resorting to hybridization. 3. Further, it has been argued that if cross fertilization alone interferes with the automatic production of pure lines, then any organism which dispenses with fertilization altogether, reproduc- ing asexually, must ipso facto constitute a pure line. Jennings sought to test out this conclusion by experiment. He selected size variations in Paramecium which reproduces by fission, with success in the case of mass cultures of unknown origin, but without success in the case of cultures made from single individuals. This was regarded as strong confirmation of the pure line principle until Calkins and Gregory, repeating the experiment on ex-conjugants, were unable to support it. Then 378 castle: role of selection in evolution Jennings, selecting a new species of Protozoa, more favorable for precise quantitative observation, also obtained a different result. He now found that among the observed fluctuations in size, those of a genetic character were included, so that by repeated selection races could be produced which were progres- sively larger or smaller, rougher or smoother. This is fully in harmony with the observations of Stout who found that varia- tions in Coleus arising in asexual propagation were capable of further propagation. It also harmonizes with the observation of Shamel as regards the occurrence in citrous fruits of bud variations which are important enough to warrant propagation in economic work ; and further, with Winkler's clear demonstration of the occurrence in the tomato and the night-shade of gigas like mutations, arising first in single somatic cells, which asexually propagated produce entire plants of a new type which then are self-perpetuating by seed. We also have the observations of East that in the asexual propagation of the potato occasional bud variations may occur which are similar in nature to unit- character variations in reproduction by seed. It is accordingly clear that the pure-line principle does not apply without excep- tion to asexually reproducing organisms any more than it does to self-fertilizing ones. It is true, however, that genetic varia- tions are much less common among such organisms than among those produced by cross-fertilization. Herein lies the justi- fication of present agricultural practice in the breeding of self- fertilized cereals, and of horticultural practice in the propagation by grafts, runners, layers, etc., of superior individual plants. 4. Attempts to extend the pure line principle to organisms which are not self -fertilizing (and this includes all the domestic animals and many cultivated plants) have met with small success. Morgan indeed assumes that it applies to his races of Drosophila up to a certain point, the point at which mutation begins, but the mutations which he recognizes are so numerous, so minute in many cases, and so fluctuating in others, that it becomes a question whether his "mutations" are not just ordi- nary heritable variations. Morgan would undoubtedly admit this since he claims that all heritable variations arise as mutations, castle: r6le of selection in evolution 379 but this is simply juggling with names, giving a new meaning to the word mutation in order to justify a sweeping generaliza- tion otherwise untenable. The test of a pure line is its freedom from any genetic varia- tion, so that selection cannot modify the racial mean as regards any character. As soon as any race of animals or plants changes in response to selection, it must be forthwith excluded from the category of pure lines. The consequence is that no case of a pure line among animals has yet been demonstrated. Never- theless the "principle of the pure line" is in some way or other supposed by the followers of Johannsen to confer on even the higher animals a limited liability to modification in consequence of selection. Thus Pearl having been entrusted in 1908 with a selection experiment for increase of egg production in Plymouth Rock fowls, an experiment which had already been in progress for nine years, decided after a study of the records kept by his predecessor that no improvement whatever had up to that time been made and further that none probably could be made since individual wild birds probably lay, under favorable conditions, as many eggs as their best tame relatives. This reasoning was strictly in accordance with the "pure line principle" and was in fact based on it. Later by changing somewhat the basis of selection, so as to rank, his animals on the basis of their progenies' performance as well as their own, Pearl found that he could considerably in- crease the flock average. Yet he still maintains that he has only more good birds not better ones, than when the experiment began, and in loyalty to the pure line principle he has no expecta- tion of .obtaining better ones in the future, since he already has and has had all along the ne plus ultra sort. One less devoted than Pearl to a generalization of the pure line doctrine would continue hopefully the effort to produce a better fowl as well as to produce more good ones. For the function of egg-production admittedly depends upon many physiological factors (as well as several external ones). These physiological factors must many of them be independently variable and to some extent independ- 380 castle: r6le of selection in evolution ently heritable. Variation in one or more of these factors (by mutation or otherwise) would undoubtedly influence the total productiveness, and the probability of the occurrence of a muta- tion would increase with the number of factors involved. So that even one formally committed to the pure line doctrine, but admitting as Johannsen does that mutations do occasionally occur in puredines, might hopefully continue to look for improve- ment in the standard of egg-production. No other method of detecting and utilizing a favorable variation, when it does occur, can be suggested than the very method of methodical and per- sistent selection against which the pure line advocates direct such vigorous attacks. Morgan is a formal adherent of the pure line doctrine, but pragmatically a selectionist for he admits the great progress made in the improvement of domestic animals and plants by selection, and even that his own mutants of Drosophila fluctuate and yield modified forms in response to methodical selection, as for example the bar-eyed mutant, subjected with success to plus and minus selection by Zeleny. But he attempts to explain these results in harmony with the pure line principle by assum- ing that, whenever a modification is observed in any character, this is due to a mutation, and if a graded series of modifications is obtained, as in the plus and minus selected bar-eyed Droso- phila, this is due to a multiplicity of mutating factors whose action on the chief factor concerned is purely incidental. On this view, however, the attainment of a completely homozygous condition on the part of all factors (if all are indeed Mendelian) would put an end to genetic variability, and selection would then cease to produce effects. Such a completely stable con- dition has, however, rarely been demonstrated. One case is reported by MacDowell, that of a race of Drosophila with an extra number of thoracic bristles. The average number of bristles was increased by selection for six generations but then showed no further increase and could not subsequently be changed either upward or downward by further selection. The race had apparently become a "pure fine" for bristle number. castle: e6le of selection in evolution 381 In the case of certain characters in guinea-pigs I have re- peatedly attempted modification of a racial character by selec- tion within an inbred race, without success. Thus a very dark form of Himalayan albino, after a certain amount of improve- ment by selection, could not be further darkened to any appreci- able extent. A race selected simultaneously for large size and for small size showed so little change that the experiment was abandoned after a few generations. No indication was forth- coming that we could thus ever approach in size either the small wild Cavia Cutleri of Peru, or the large races of guinea-pig kept in captivity by the natives of the same region. Yet evolution had in some way evidently produced these divergent conditions from a single original source. The changes were probably too slow to be observable in the life time of one observer. On the other hand, certain characters of guinea-pigs, rabbits, and rats have been found to respond readily to selection in a particular direction. This is notably true of color patterns which involve white spotting. A selection experiment with hooded rats selected simultaneously in plus and minus directions has produced one race which is black all over except a white patch of variable size underneath, and another race which is white all over except for the top of the head and the back of the neck, which are black. The races do not overlap at all and have not done so for many generations, though they still continue to diverge from each other as a result of continued selection. In similar experiments with Dutch marked rabbits it has been found possible by selection to increase or decrease the amount of white at will. In a series of such rabbits ranging from nearly all black to nearly all white, stages far enough apart to be cer- tainly identifiable behave as Mendelian allelomorphs in crosses, but regularly emerge from such crosses in a slightly modified form, the whiter stages having been darkened and conversely the darker stages whitened. The principle of the pure line manifestly does not apply to these cases. White spotting is apparently a character which from its nature fluctuates con- stantly, such fluctuations having, to some extent at least, a genetic basis, since continuous selection invariably produces a 382 castle: e6le of selection in evolution modified race. Even in wild species, such as the skunks, white- spotting is manifestly a variable character, which no doubt will respond to the selective efforts of our skunk farmers, who de- sire an all-black race. Why white-spotting should be a less stable character genetically than some others, it is impossible to say, but the fact is beyond question. Morgan has suggested that in general the genetic basis of a Mendelian character may be a single molecule, and gives this as a reason for believing in its constancy. But white spotting can hardly fall in with this conception. It seems to me more probably due to a quantitative deficiency in the germ of some substance which normally finds its way into all epidermal cells of the body and which is responsible for the development in them of melanin pigment. Greater and greater deficiencies of this substance cause more and more extensive white areas. Complete or total albinism behaves very differently. It results from a complete change in some color factor which may well be a simple molecule since it appears to be incapable either of contamination in crosses or of modification under selection. Nevertheless the color factor (molecule or whatever it may be) evidently is not so simple but that it can assume at least four mutually allelomorphic forms, as shown for the guinea-pig by Wright, a like number of allelomorphs, though not their exact equivalents, being known also in the rabbit. As regards the agouti factor in mice, rabbits, and guinea-pigs, this too may assume several different allelomorphic conditions, though it is not certain that any one of these fluctuates or can be modified other than by associating with it unrelated genetic factors. The divergent conclusions which students of genetics have reached concerning the stability of Mendelian genes and the consequent effects of selection for their modification are probably due in part to the particular choices which they have made of test cases. A study of albinism alone would lead one to believe in the fixity and constancy of Mendelian genes and the impossi- bility of modifying them by selection. A study of white spotting leaves one with the unshakable conviction that this form of castle: r6le of selection in evolution 383 gene is plastic and yields readily to selection. Where only genes of the former sort are involved, the principle of the pure line is applicable ; where genes of the latter sort are involved, it is not applicable. The divergent results obtained by Jennings when dealing with Paramecium and when dealing with Dimugia indicate that among asexually reproducing organisms, also, genes are involved, some of which are stable, some of which are not. Accordingly, what conclusion we reach as to the applicability of the pure line theory in the breeding of animals and plants will depend upon how common we find stable and plastic genes respectively to be, and in what sorts of variations they are involved. My own opinion, based upon a study through many years of a variety of inherited characters in the smaller mammals, inclines to the view that in such animals very few characters can safely be referred to the agency of perfectly stable genes. Even in color characters, probably the simplest as well as the most studied of inherited characters, there is much fluctuation which yields substantial results to selection by the discriminating breeder. The yellows are not all of one shade, nor the blacks of equal depth. The golden yellow of the Guernsey cow is very different from the fawn of the Jersey or the dark red of the Devon. Yet all are yellows, allelomorphs of black, but each is selected for a different standard to which the breeder must adhere very carefully in his selections, if he wishes to win prizes or sell breed- ing stock. When it comes to size and shape and that consistent inter- relation of parts which the breeder calls "conformation," stable genes cannot be detected. Crosses produce blends as regards size and shape, and conformation is completely dissipated by a cross. That is why the breeder is so reluctant to resort to an outcross unless he is engaged merely in meat or wool production and is not attempting to breed to a type. Aside from color there are very few valued economic characters in our domestic animals which are not inherited after the manner of blends. Weight of carcass, quality of wool, milk production in cattle, egg production in fowls — all these are blending characters which 384 castle: role of selection in evolution in later generations show either no segregation or imperfect segregation (fowls, Pearl 2 ). I do not say that in these cases no Mendelian inheritance is involved, but merely that no stable genes are in evidence, nothing that would preclude the probable effective use of selection in maintaining or raising breed standards. If we turn from the breeding of animals, in which manifestly the pure line principle has little applicability, to the breeding of plants other than those which are self-fertilized, we again find that this principle has a very limited applicability. Prob- ably the most valuable open pollinated field crop in cultivation is maize. But a pure line of maize is not known to exist. An experiment which should have lead to the production of pure fines, if such a thing were attainable in maize, has been in prog- ress at the University of Illinois for the past twenty years. Selection has been made for increased and for decreased protein content of the grain, and also for both increased and decreased oil-content, with the result that steady progress in the direction of selection has in every case been made. The high protein strain now contains twice as much protein as the low protein strain; and the high oil strain contains four times as much oil as the low oil strain. The divergence of the selected lines from each other is not now as rapid as at first but it still continues steadily, with no indication that it is soon to cease, as must be the case if only stable genes were involved. Those characters in maize which directly affect the yield, such as size of plant, or of the grain which it bears, are blending in inheritance and show imperfect segregation subsequently. They are probably all of them quite as amenable to selection as the oil content and protein content of the seed, experimented upon in Illinois. 2 It is true that Pearl (1912) has described fecundity in fowls as "typically Mendelian" in heredity but his figures show that in crosses between Barred Rocks and Cornish Indian Games, the average fecundity of the Fi birds is in both the reciprocal crosses intermediate between that of the respective parent races though nearer the racial average of the sire, which supports his contention that a sex-linked gene is involved, but shows also that this is not the only factor involved. Back-crosses of Fi of both sexes with the pure races give evidence of further blending (or imperfect segregation) on the part of the non-sex-linked factor or factors. castle: bole of selection in evolution 385 Finally, as evidence that even in self-fertilized plants the pure line principle may be inapplicable because of the existence of genes which are plastic, let me cite a very extensive and care- fully executed piece of work on garden peas done by Hoshino. He studied the behavior of flowering time, and showed that its inheritance involves a Mendelian gene coupled with flower color (white or red). The inheritance of flowering time is inter- mediate, but Fi is closer to the late than to the early parent in this character. Segregation is imperfect in F 2 with a range practically all the way from the early to the late parent, but not transgressing this range. F 3 and F 4 families from self-fertilized parents are in many cases quite variable but others are no more variable than the pure parental varieties and so may be treated as practically "constant." A study of the average flowering time of each of the 230 "constant" F 4 families shows that these fall into three main groups, some falling into a modified early group, not quite so early as the early parent, others falling into a modified late group, not quite so late as the original late parent, but most of all falling into an intermediate group occupying the region midway between the parent varieties in flowering time. Considered all together, the F 4 families "constant" for flowering time form an almost uninterrupted series of conditions con- necting the respective parental conditions seen in the early flowering and in the late flowering race. These observations show the existence of a gene for flowering time in peas which is decidedly plastic. That a gene actually exists is shown by its coupling with flower color. That it is plastic is shown by the fact that it emerges from the cross nearly always in a modified form. When the possibility of modifica- tion has been continued as long as the F 4 generation, the majority of the "constant" families are found in the intermediate or middle group. The plasticity is here shown in a tendency of the con- trasted genes to blend into one of intermediate character. It is also shown in data given by Hoshino as to flowering time in parent individuals and their offspring in the late flowering variety. Although this variety is treated by Hoshino as a "pure line," it is evident that within this line itself the later flowering in- 386 castle: kole of selection in evolution dividuals have later flowering offspring and vice versa. In other words selection within this supposed "pure line" is evidently effective. Accordingly either the gene here involved is plastic or the supposed pure line is not pure. From the various lines of evidence which have been cited (and I might have cited many more) it is clear that the pure line prin- ciple, valid as a working hypothesis for seed size in beans and for certain morphological characters in self-fertilized cereals, does not fit in with the observed facts as regards the effects of selection in the majority of the domesticated animals and cultivated plants, nor even with the behavior of certain characters in self -fertilized plants and asexually propagated animals. In the case of such characters as white spotting in mammals, it is evident that a change in the mean of the character in a partic- ular direction in consequence of selection actually displaces in the direction of selection the center of gravity of variation, so that in a very true sense selection makes possible further varia- tion in that same direction. The same is probably true as re- gards protein content and oil content in the Illinois corn experi- ment. It is doubtful whether, outside of that particular experi- ment, maize with as high a protein content as 15 per cent has ever been observed, or maize with as high an oil content as 8.5 per cent. It is not then a misuse of terms to say that the selec- tion has in this case been the cause of further variation in the direction of selection and so an agency in the progressive evolu- tion of a new type. If this is true concerning a single character under experimental study for a period of twenty generations, may it not also be true of entire organisms and groups of organisms subjected to keen competition with all other organisms in a struggle for existence which has continued for millions of generations? If there are characters which are plastic under artificial selection, why need we be skeptical about the plasticity of organisms sub- jected to natural selection? If artificial selection can, in the brief span of a man's life time, mould a character steadily in a particular direction, why may not natural selection in unlimited time also cause progressive evolution in directions useful to the castle: role of selection in evolution 387 organism? I am not ready to say that natural selection is proved as the method par excellence of evolution, but I am not ready to abandon it as the most reasonable explanation of evolution until something better supported than the mutation theory is offered as a substitute for it. At the same tune the fact should be emphasized that biology has benefited greatly from the investigation and the discussion initiated by the mutation theory. Even though the mutation theory cannot be accepted as a general theory of evolution it has done us great good in dispelling or clarifying the hazy notions which formerly existed as to what natural selection could accomplish. Selection, whether natural or artificial, is, as the mutation theory rightly holds, primarily an agency for the elimination of variations, not for their production. It can only act on variations actually exist- ing, and while it can, I believe, continue and extend variation already initiated by shifting in the direction of selection the center of gravity of variation, it cannot initiate new lines of variation. It cannot change a vertebrate into something else nor something else into a vertebrate. It is limited to the modi- fication of existing types of organisms, and to their modification in directions in which they show a tendency spontaneously to vary. BIBLIOGRAPHY Castle, W. E., 1916. Genetics and Eugenics. Harvard Univ. Press. Hoshino, Y., 1915. On the inheritance of the flowering time in peas and rice. Journ. Col. Agr. Tohoku Imp. Univ., vol. 6. Johannsen, W., 1909. Elemente der exakten Erblichkeitslehre. Jena. MacDowell, E. C, 1915. Bristle inheritance in Drosophila. Journ. Exp. Zool., Vol., 15. Morgan, T. H., 1916. A critique of the theory of evolution. Princeton Univ. Press. Pearl, R., 1912. The mode of inheritance of fecundity in the domestic fowl. Journ. Exp. Zool., vol. 13. Pearl R., 1916. Fecundity in the domestic fowl and the selection problem. Amer. Ass. Nat.. Vol. 50. Scott. W. B., 1917. The theory of evolution. New York. Smith, L. H., 1912. Altering the composition of Indian corn by seed selection. Journ. Indust. Eng. Chem., vol. 4. DeVries, H., 1900-1903. Die Mutationstheorie. Leipzig. Wright, S., 1916. Studies of inheritance in guinea-pigs and rats. Carnegie Inst. Wash., Pub. 241. UNIVERSITY OF ILLINOIS-URBANA 575 1L493 C002 LECTURES ON HEREDITY OEUVERED UNDER THE liftSKJHffi! mml, 3 01 019133492