UNIVERSITY Of JUL 2 7 1915 4. Studies from the Laboratory of Physiology and Phar- macology of Indiana University. THE EFFECTS OF INBREEDING AND SELECTION ON THE FERTILITY, VIGOR, AND SEX-RATIO OF DROSOPHILA AMPELOPHILA By W. J. MOENKHAUS Reprinted from Journal of Morphology Vol. 22, No. 1, March 20, 1011 fJNfVFRs*7> or ff f wo»s f twwarv JUL 2 7 1915 535.77 ma*. to^.3 Reprinted from Journal of Morphology, Vol. 22, No. 1 March, 1911 THE EFFECTS OF INBREEDING AND SELECTION ON THE FERTILITY, VIGOR AND SEX RATIO OF DROSOPHILA AMPELOPHILA W. J. MOENKHAUS Indiana University , Bloomington , Indiana CONTENTS Introductory 124 Material and methods 124 Inbreeding and selection on fertility and vigor 126 1 Introductory 126 2 Sterility 127 a Character of the sterility 127 b Degrees of sterility 128 3 Inbreeding and vigor 131 4 Sterility and selection 134 5 Discussion of results 138 Sex-ratio and selection 141 1 Introductory 141 2 The normal sex-ratio 141 3 Control of sex-ratio by selection 142 a History of strain 206 • 143 b History of strain 207 147 c Discussion 147 4 Influence of male and female in determining the sex-ratio 148 5 Discussion of results on sex-ratio 151 153 Summary Literature cited 123 154 124 W. J. MOENKHAUS INTRODUCTORY The present report includes the results of two series of experi- ments on the fruit fly — Drosophila ampelophila. One set con- cerns itself primarily with the effects of inbreeding and the other with sex-ratios. The experiments on inbreeding grew out of work I had been carrying on on hybridization. In these hybridiza- tion experiments the effects on the developmental processes of hybrids between species too remotely related were especially emphasized. The converse of these experiments was, naturally, to study the effect upon the young between individuals too closely related. Fishes, upon which all my experiments in hybridization were made, do not lend themselves for purposes of inbreeding without elaborate breeding facilities. Mice seemed suitable for this purpose but, both at the outset of these experiments and since, these creatures have proven miserable failures in my hands. Among the insects, I tried the common willow beetle but this proved to throw only one generation annually in this latitude. It was desirable to have an animal with a brief life history, whose food could be easily obtained at all seasons and in which the sexes could be readily distinguished. In these respects the fruit fly is almost ideal. The facts herein considered confine themselves to this species. The experiments on sex-ratio suggested themselves in connec- tion with the inbreeding experiments and so were carried out along with the latter and after they were completed. MATERIAL AND METHODS The strain which is mostly under discussion in my inbreeding experiments came from a well-filled female that was taken from the window of my residence in Bloomington. Other strains *were started at the onset. Some of these came from the banana bunches at the various, groceries and others came from fruit which I had laid out for this purpose. None of these were carried further than two or three generations excepting two, called 6 and 7 in my records. The latter was discontinued after the tenth generation INBREEDING AND SELECTION TN DROSOPHILA AMPELOPHILA 125 since it had been from the beginning apparently less prolific. The strain 6 was carried for over seventy-five generations and is the one on which the experiments in inbreeding of this report are based. For vivaria, tall stender dishes, tumblers, quinine bottles and lamp chimneys were given a trial. They were discarded in favor of 8-dram shell vials. These were compact, so that a large number of matings could be kept in a small space, and they were most con- venient in manipulating the pairs during the frequent changes to new cages that was necessary all along. The open end of the shell vial was closed with a plug of absorbent cotton, not too compact, so as to afford some ventilation. The flies are strongly positive to light, so that the vials could be laid with their bottom toward the light and the cotton plug removed with safety for the introduction of food etc. Small trays holding fifteen of these vials were used and in this way the experiments could be readily and compactly stored in the incubator, or they could be packed into a valise to be taken along wherever I went. The food was exclusively well-ripened bananas. To prevent the larvae irom pupating in the food, narrow strips of blotter or filter paper were introduced in which they seemed to be especially fond of pupating. It is, of course, apparent that the greatest care had to be taken to avoid contamination from flies without. The stock food had to be scrupulously watched and the instruments kept clean to avoid the introduction of eggs laid on them by extraneous females. The bananas, especially, as they come from the stores, are likely to be infected with eggs and larvae if the skin be in any way bruised or split. The brothers and sisters were paired off, always within the first ten or twelve hours of their life as imagos. Up to this time mating has not occurred. In fact I have never found a pair that copulated during the first twenty-four hours or, if so, that pro- duced fertile eggs. 126 W. J. MOENKHAUS INBREEDING AND SELECTION ON PERTILITY AND VIGOR 1. Introductory That continued inbreeding acts deleteriously on the fertility and vitality of a race is a belief so firmly and generally established that it is seldom questioned. This has its origin largely in the common experience of breeders whose observations, unfortun- ately, are too often unreliable. There are not wanting experi- ments such as those of Van Guaita (’98) and Bos (’94) and others, scientifically conducted, which bear out this conclusion. On the other hand, it is refreshing to encounter in the literature such reports as that of Gentry (’05) who believes not only that inbreeding is not necessarily harmful, but also that it maybe beneficial to conserve and intensify the good points in his breed. Gentry’s experiments were made on Berkshires. The most pro- longed tests of close inbreeding that have been recorded were made by Castle (’06) on the same species with which the present paper deals. He inbred (brothers with sisters) for fifty-nine gener- ations. He concludes that such close inbreeding does not neces- sarily result in a loss of productiveness and of vigor; at least that inbreeding cannot be regarded as a causal factor. Some of his results so nearly parallel those of the present writer that further reference to his results will be made in the body of the paper. During the early part of October, 1903, a number of pairs were started breeding. These came from various sources in Bloomington. These different pairs were reared for the most part only a few generations, excepting pair No. 6 which was continued for about four and one-half years. During this time over seventy-five generations were produced. Toward the close of this period no exact count was kept of the generations so that only an approxi- mate figure can be given. Five pairs of brothers and sisters were mated in each generation to insure against accidents that might terminate the strain if but one mating were made. Along at the fifth and sixth generation it became more and more difficult to keep the strain alive with the five pairs of brothers and sisters that were mated each generation. The failure of an INBREEDING AND SELECTION IN DROSOPHILA AMPELOPHILA 127 occasional pair to produce young had hitherto been attributed to accidental conditions of food, etc., but this no longer seemed a satis- factory explanation of all the failures to produce young. This con- dition, was, therefore looked into more thoroughly. This was done by laying out instead of five pairs a much larger number from the offspring of a given productive pair. The greatest care was taken with the food, temperature etc. and it soon developed that a variable per cent of the pairs were sterile. These sterile pairs were to all appearances normal. It was clear now that, while inbreed- ing had not reduced the general vitality of the strain thus far, there had appeared a high degree of sterility. 2. Sterility a. Character of the sterility. Examination of all the matings brought out the fact that in all cases eggs were present in large numbers. This seemed to suggest that the difficulty lay in the larvae either failing to emerge from the egg envelope or, succeeding in this, failing to carry themselves through the feeding stage or the transformation. By a careful search of the food of the sterile pairs, after suffi- cient time for the larvae to mature had been allowed, it became evident that the difficulty lay at a time earlier than the pupal stage for none of the latter could ever be found. The food sup- plied these sterile pairs was the same as that of the fertile ones since it could not be foretold which pairs were going to prove infertile. Furthermore, the infertile pairs were usually kept for from twenty to thirty days, the best of food being supplied them from time to time. The same search showed that no larvae were pres- ent, at least so far as direct inspection of the food under a dis- secting microscope could be depended upon. It was always possible, of course, that the larvae failed to carry their development very far, and, thus, being small when they first emerge from the egg, might have been overlooked. It became necessary, consequently, to take the eggs as they were laid from time to time and keep them under observation to see whether the larvae ever emerged. This was done by placing a piece of banana 128 W. J. MOENKHAUS in the vial with a sterile pair and from time to time removing the eggs one by one with the point of a needle and placing them on a piece of moist filter paper in a separate vial. Usually twenty were placed in each vial and some food added for the larvae, should they emerge. Inspection of the eggs after twenty-four, forty-eight and seventy-two hours would readily reveal the number of eggs that had produced larvae. I have laid out thus at a great expense of time literally thousands of eggs from many infertile pairs, in many cases all the eggs that a given pair produced during the first twenty-five days of its life, but I have never seen a single egg that had hatched. Eggs of fertile pairs thus laid out will readily hatch so that all the larvae will have taken to the food twenty-four hours after the eggs are deposited. Such infertile pairs copulate frequently and it would seem that impregnation should follow. I have never sectioned the eggs to see whether spermatozoa enter the eggs or whether they con- tain partially developed larvae which fail to hatch. I have, however, been able to determine in this strain which of the sexes is at fault. This was done in the following manner. After a pair by sufficient trial had proven itself infertile, the male was mated to a virgin female of a fresh strain that had not been inbred and possessed a high degree of fertility, and the female was simi- larly mated with a male, usually one whose fertility had been estab- lished. Sixty-four such cases were tried and in no case did the females fail to produce young and in no case did the males pro- duce any although repeated copulations took place. It is evident from the foregoing, that, in this strain, the sterility lies exclusively in the male and that the female has lost, apparently, nothing in fertility. Castle (p. 735) reports, on the other hand, that either sex may be sterile. However, Castle took no account of the eggs and larvae but merely the production of pupae, so that his steril- ity cannot be with certainty compared to mine. It would seem, however, that in some strains infertility may be strictly confined to the males and in others to both sexes. That sterility is com- plete for all males, when it occurs, is shown by both our results. b. Degrees of sterility. The foregoing experiments concerned themselves with such pairs as were completely sterile. Other pairs INBREEDING AND SELECTION IN DROSOPHILA AMPELOPHILA 129 of brothers and sisters from the same parents, however, were fer- tile. Judging from the productiveness of these, there was often a wide divergence. It would seem that, as a result of inbreeding, we had a condition of fertility ranging from absolute infertility to comparatively high fertility among the different pairs of brothers and sisters from any given pair of parents. To test this the follow- ing experiment was carried out: About two-hundred eggs from each of fifteen pairs of flies were laid out after the fashion indicated above. Ten of these pairs had been inbred for seventeen genera- tions while five belonged to fresh stock that had not been inbred. Of the ten pairs of the inbred strain, five belonged to a strain which had arrived at a very low degree of fertility, namely only 36 per cent of the forty- two pairs tested were fertile (table 3, seventeenth generation, strain, A) . These five pairs were brothers and sisters to many of the sterile pairs considered in the preceding section. The other five pairs (of the ten inbred) were from a strain which had been held by selection to a high degree of fertility, namely 97 per cent of the thirty-four pairs tested were fertile. Both of these strains were descended from common great grandparents (table 3, seventeenth generation, strain B). We have, thus, for comparison three conditions, namely, (1) eggs from a highly infertile inbred strain; (2) eggs from a highly fertile inbred strain; and (3) eggs from a presumably norma strain that had not been inbred. It should be added that the five pairs were taken at random and were not selected. Approxi- mately the first two-hundred eggs of each pair were laid out in batches of about twenty to twenty-five to the vial. The number of eggs that hatched was noted in each case and also the number that emerged as imagos. Table 1 gives the summary of results. From this table it appears that from the eggs which were taken from the inbred pairs with low fertility practically as large a per cent (97.27) hatched as from the eggs that came from the inbred pairs that showed a high fertility (98.2). The same is true in regard to the number that produced imagoes, 86.8 per cent and 85.1 per cent respectively. The fact clearly brought out here is that when infertility arises in this strain it arises suddenly and JOURNAL OF MORPHOLOGY, VOL. 22, NO. 1. 130 W. J. MOENKHAUS does not present all intergradations. In other words, one does not find that among a large number of brothers and sisters some pairs whose eggs only partially hatch and other pairs that range in this respect, on the one hand, to perfect fertility and, on the other, to complete sterility. The fertility is either completely lost or it is of a high degree. Furthermore, when we compare the inbreds with the normals (not inbred) in regard to the percentage of eggs hatched no essential difference is observable. It would seem, therefore, TABLE 1 Inbred { low fertility ) PAIRS NUMBER OP EGGS PLACED NUMBER OF EGGS HATCHED NUMBER OP IMAGOS EMERGED PER CENT OP EGGS HATCHED PER CENT OF IMAGOS EMERGED A 193 184 160 95.3 82.9 B 200 188 169 94.0 84.5 C 2or 197 182 98.0 90.5 D 198 198 180 100.0 90.9 E 123 123 104 100.0 84.5 Total 915 890 795 97.27 86.8 Inbred {high fertility ) A 201 198 182 98.5 90.5 B 173 172 156 99.4 90.1 C 204 200 161 98.0 78.9 D 197 193 165 97.9 83.7 E 175 169 145 96.5 82.8 Total 950 932 809 98.2 85.1 Normals {not inbred) A 215 211 193 98.1 89.7 B 70 70 48 100.0 68.5 C 153 152 132 99.9 86.2 D 224 218 144 97.3 64.2 E 158 155 144 98.1 91.1 F 146 127 109 87.7 74.6 G 223 222 205 99.9 91.9 Total 1189 1155 975 97.2 82.0 INBREEDING AND SELECTION IN DROSOPHILA AMPELOPHILA 131 that the pairs that had not completely lost their fertility, in so far as hatching their eggs is concerned, had suffered no deteriora- tion whatever as a result of seventeen generations of closest in- breeding. A fact of further importance brought out by table 1 is that of the percentage of eggs that successfully produced imagos. This does not differ essentially in the two groups of inbreds nor do these differ essentially from the normals. Castle used as his measure 1 productiveness/ meaning thereby the number of pupae that were successfully produced. Making allowance for some pupae which do not emerge, the imagos produced in my experi- ments were an approximation to his ‘productiveness.’ Inbreed- ing, consequently, does not affect adversely the productiveness of pairs that show any fertility at all. Castle found that his strains showed an annual fluctuation in productiveness, the period of least productiveness falling in the late autumn and early winter. My own experiments extended over about four and one half years and, although I have been on the lookout for this, I have never observed it. As Castle himself suggests, this fluctuation was probably a function of the tempera- ture of the room. My flies were kept in a room which varied from 60 to 80 degrees and, when this was not possible, they were placed in an incubator kept at about the same range of temperature. It may also be that the productiveness of his strain ran low at this time of the year because they were placed in new hands at the open- ing of the college year. My observation has been that it takes some time for a new man to learn all the conditions that make for a favorable rearing of these creatures so that Castle’s low produc- tive periods may be merely a measure of the training period of the experimentor. 3. Inbreeding and vigor At the outset of the experiments it was the expectation of the writer that such rigorous inbreeding would early and violently show itself in the vigor and fertility of the animals. In this, how- ever, he was largely disappointed. In the strain that is here under consideration no untoward results could be detected during the 132 W. J. MOENKHAUS first five or six generations. As previously stated, up to this time the method consisted in placing pairs of brothers and sisters in each of five vials to insure against mishaps. These mishaps con- sisted of drying up of the food, attacks of fungus and in some cases the escape of the flies themselves during the process of feeding etc. Those pairs that produced young were regarded as having es- caped these various possible mishaps and were taken as indica- tions of the vitality and productiveness of the strain. The expecta- tion at that time was that any deleterious effect of the inbreeding would show itself in the offspring of any of the pairs. Conse- quently, when a given pair would produce offspring that was num- erous, all well formed, vigorous, and in no apparent way differing from normal offspring, to see whether some slight influence might not be present that could not be detected by ordinary observation a definite measure was taken of (1) their rate of reaction to light and gravity, (2) the total number of eggs produced and (3) the percen- tage of eggs which hatched and emerged. An attempt was made to determine their length of life but this proved too prolonged to allow one to carry it out together with all the other incidents of the already too laborious experiments. The reaction of this animal toward light and against gravity is well known. To get a measure of the rate of reaction the ani- mals were made to travel through a glass tube that had been blackened for 16 cm. on the inside. This tube had a light placed at one end and was inclined about twenty-five degrees. From a glass vial the flies were admitted, one at a time, into the tube and the time from the moment of entrance into the blackened portion of the tube to their emergence was recorded. It was found essen- tial that the t\^o batches of flies (inbreds and normals) should be of the same age, be reared under the same conditions and that the temperature of the room be the same for the two batches. The results are as follows: at a temperatureof 27.2° C. 133 normals took 16 seconds, average, to travel the distance, and 140 inbreds took 15.4 seconds. The two sexes in these two groups were about equal in number. In both groups the males travel the distance on an average in three seconds less time. It is clear from this that the normals and inbreeds are equally responsive to these two INBREEDING AND SELECTION IN DROSOPHILA AMPELOPHILA 133 agents and that the latter have not suffered in this regard as a result of inbreeding. In order to determine the total number of eggs produced it was necessary to isolate the pairs and twice each day pick off all the eggs that had been deposited in and around the food provided. This proved to be a most laborious process, for the eggs are too small to be followed safely with the naked eye and had to be removed individually with the point of a needle. Too much value must not be attached to this measure for the reason that the rate and, therefore, probably the number of eggs deposited seems to depend somewhat, at least, on the condition of the food present, and for the TABLE 2 Strain 6 Number of generations inbred 2 3 5 6 8 9 Number of days eggs were counted 27 30 34 34 23 32 Total number of eggs laid 433 617 480 724 455 516 Strain 7 Number of generations inbred 2 3 5 6 9 10 Number of days eggs were counted 26 33 29 23 33 28 Total number of eggs laid 654 662 539 498 907 429 reason that only the eggs deposited during the first twenty-five or thirty days were counted. These creatures live to be very much older. We have kept females alive 153 days, but after the first twenty-five or thirty days the eggs come only in small numbers. Table 2 gives the actual counts of several females of both strains 6 and 7. We see from the above counts that no material reduction has occurred in egg production during nine and ten generations of inbreeding. Such variations as occur may, of course, represent individual differences in the females. The data given in table 1 of the relative hatching and emerging qualities of the young of normals and of pairs inbred for seventeen generations shows that there is no difference in this respect. 134 W. J. MOENKHATJS In so far as the above determination may be taken as a measure of the vitality of this species we are justified in concluding that from six to seventeen generations of inbreeding no appreciable deterioration has resulted. No such exact determinations were made in later generations, and it is possible that eventually the effects of inbreeding would manifest themselves, but my observa- tions during seventy-five or more generations does not lead me to believe this. J/.. Sterility and selection Along at the thirteenth and fourteenth generations the sterility had become very pronounced. Of the offspring of some of the pairs, more than 50 per cent of the males were sterile. On the other hand, while practically all pairs showed at least some degree of sterility this varied very much in the different brothers and sis- ters of the same brood. That this sterility was a direct physiolog- ical result of the inbreeding seemed to me very doubtful. To find the effects of inbreeding showing itself in such a specific way upon the males only, did not, to say the least, meet expectations. Furthermore, sterility was not wholly wanting in forms that had not been inbred. It was highly desirable to continue the experiments on inbreed- ing, and yet to keep the strain alive, it was necessary to find some way to eliminate this high degree of sterility. The process that was most effective was selection. By continuing the strain of those pairs whose offspring showed the highest degree of fertility but at the same time continuing the rigorous inbreeding, it was possible almost completely to eliminate the sterility. This at the same time gave one of the severest tests as to whether inbreeding was the responsible factor, for if the sterility could be eliminated by continuing the very process of inbreeding the latter could not well be held to be the cause of it. This was done as follows: In the fourteenth generation three fertile pairs of brothers and sisters from the same brood were iso- lated and mated. The offspring of each of these were mated in pairs to determine the degree of sterility. By reference to table Generation 6^ (From brood with 84.6% fertility) INBREEDING AND SELECTION IN DROSOPHILA AMPELOPHILA 135 (53%) (47%) (96%) (4%) (77%) (23%) (85%) (15%) (69%) (31%) 136 W. J. MOENKHAUS 3, it will be seen that the pair marked A produced offspring out of which nine of twelve pairs tested were infertile; pair B produced offspring of which four pairs out of fourteen tested were infertile and pair C threw offspring with five pairs out of fifteen infertile. We have here, then, three pairs showing a wide variation in the degree of fertility of their offspring. Pair A showed 75 per cent of the pairs infertile and pairs B and C approximately the reverse ratio. In the further progress of the experiment pair C was dis- TABLE 4 Strain A NUMBER PAIRS TESTED NUMBER PAIRS FERTILE NUMBER PAIRS INFERTILE PER CENT PAIRS FERTILE PER CENT PAIRS INFERTILE 18 (1) 52 27 25 51 49 18 (2) 51 37 14 72 28 18 (3) 52 37 15 71 29 18 (4) 56 45 11 80 20 18 (5) 28 19 9 69 31 Average for 238 pairs 69 per cent. Strain B 18 (1) 15 15 0 100 0 18 (2) 14 14 0 100 0 18 (3) 19 19 0 100 0 18 (4) 22 15 7 68 32 8 (5) 23 23 0 100 0 Average for 93 pairs 92.5 per cent. continued so that only pairs A and B were used. I shall in the further description of the experiment refer to the descendants of A as strain A and of B as strain B. Before entering upon the experiment of selection it was neces- sary to ascertain whether, without selection, the descendants of pairs A and B continued to show a low and high fertility respec- tively. Accordingly, a single one of the fertile pairs of the 15th inbred generation of strain A and B was tested. Reference to the table shows that in strain A 27 pairs or 57 per cent of the forty seven pairs tested were infertile, while in strain B none of the thirty- INBREEDING AND SELECTION IN DROSOPHILA AMPEL.OPHILA 137 seven pairs tested were infertile. The same process was repeated with a pair of the sixteenth generation of the two strains. Strain A showed twenty-seven or sixty-four per cent of the forty-two pairs tested infertile and strain B one or three per cent of the thirty-six pairs tested. Up to this point in the experiment only a single pair in each generation was tested as to the fertility of its offspring. It might well be that by chance in each case a pair of low fertility was taken in strain A and a pair of high fertility in strain B. To eliminate this possible error five pairs were taken in each strain and the fertility of their offspring determined. It was further desirable to obtain an estimate of the variability in the fertility of the pairs in the two strains as well as to get a more correct estimate of the average fertility of both. In the diagram these five pairs are designated as 18 (1), 18 (2), etc. Table 4 shows the number of pairs of offspring tested for each pair and the number and per- centage of pairs fertile and infertile. The fertility thus varied in strain A from 51 per cent in 18 (1) to 80 per cent in 18 (4), with an average fertility of 69 per cent. In strain B the fertility was much less variable in the different pairs, the only exceptions being 18 (4), the average fertil- ity being 92.5 per cent. We now have definitely established two strains, one' of low and another of high fertility. The important part to be empha- sized here is that this was produced by the process of selection from among the variable offspring of generation fourteen of the inbred strain. To make the experiment more complete it was now neces- sary to obtain a highly fertile strain out of the one with low fertility. Accordingly strain B was discontinued at this point and attention restricted to strain A. Five pairs, 19 (1), 19 (2), 19 (3), etc., were taken from among the offspring of 18 (4) because this showed the highest percentage of fertility. These were tested in the same way as in the preceding generation. Table 5 gives the details. By selection it will be seen that the average fertility was raised from 69 per cent in the 18th generation to 75 per cent in the 19th generation. Among the five pairs used one 19 (2) showed an unusually high fertility (96 per cent) . This pair was accordingly JOURNAL OF MORPHOLOGY, VOL. 22, NO. 1 138 W. J. MOENKHAUS TABLE 5 NUMBER PAIRS TESTED NUMBER PAIRS FERTILE NUMBER PAIRS INFERTILE PER CENT PAIRS FERTILE PER CENT PAIRS 1 INFERTILE 19(1) 50 27 23 53 47 19 (2) 51 49 2 96 4 19 (3) 51 39 12 77 23 19 (4) 52 44 8 85 15 19 (5) 35 24 11 69 31 Average fertility of 239 pairs 75 per cent. taken to select from. Five pairs were again taken as before. The results appear in table 6. Thus it will be seen that all five pairs showed a uniformly high degree of fertility. The average fertility of all the pairs was raised to 93. 8 per cent. 5. Discussion From the above series of experiments a number of important facts are brought out. 1. Sterility, as it appeared in the strain under consideration, is strongly transmissible through inheritance. 2. It is readily controlled by selection. 3. Inbreeding is probably not the physiological cause of it. That this sterility is transmissible cannot be doubted. The faithfulness with which this occurs appears in the strains A and B. Both were derived from a common pair that showed a variability with respect to this character in the three pairs of .its offspring TABLE 6 j NUMBER PAIRS TESTED | NUMBER PAIRS FERTILE NUMBER PAIRS INFERTILE PER CENT PAIRS FERTILE PER CENT PAIRS INFERTILE 20(1) 41 37 4 90 10 20 (2) 45 42 3 93 7 20 (3) 45 44 1 97 3 20(4) 36 33 3 91 9 20 (5) 44 42 2 95 5 Average for 211 pairs 93.8 per cent. INBREEDING AND SELECTION IN DROSOPHILA AMPELOPHILA 139 tested. One of these possessed a high degree of sterility, while the two other pairs showed a low degree. The descendants of the latter constituting strain B, retained this low degree of infertility through- out. Similarly the descendants of the former, constituting strain A, retained their high degree of infertility up to the time when selection away from this condition was introduced. In the latter process the transmissibility of the character is again emphatically revealed. In the eighteenth generation, pair 4 showed a lower degree of sterility than any of the remaining four pairs of brothers and sisters. Breeding from this pair at once showed offspring with a decided decrease in sterility, compared with the eighteenth generation, the average of the nineteenth generation being 75 per cent of the pairs fertile as compared to 69 per cent of the latter. Again, in the nineteenth generation, pair 19 (2) showed a much lower degree of infertility than the other pairs. Continuing the strain from this pair, this character is faithfully reproduced in the offspring in that they average fertility of the latter is raised to 93.8 per cent. It is important to note in this connection that Castle, in his experiments upon Drosophila, found that productiveness (which as previously noted is quite a different thing from the sterility here considered) was similarly transmissible and amenable to selection. Furthermore, Castle’s experiments would seem to indi- cate that this character of productiveness behaves, in inheritance, after the Mendelian fashion, low productiveness acting as the recessive character. We have evidently to do here, both in the pro- ductiveness in Castle’s experiments and in the sterility in my own, with characters that are germinal for they behave as such. In the strain upon which my experiments were made we have the further remarkable condition that the infertility is inherited only by the males. It is clear that whatever the causal factor or factors to which the sterility may be attributed, it is relatively insignificant compared to the effect of selection upon it. Furthermore, the modification is a germinal one. That inbreeding may be responsible for its prevalance in the strain seems probable, but that it is responsible 140 W. J. MOENKHAUS for its origin is not believed. We have seen that the general vital- ity of the strain, as measured by its productiveness and its reaction to light and gravity, did not suffer as a result of seventeen gener- ations of closest inbreeding. Failing in this, it is not probable that its effect would show itelf in so specific a way as the sudden and complete sterility in certain males of the strain. The improba- bility is further supported by the fact that the inbreeding may be continued unabated if only care be exercised in the selection of the brothers and sisters to be mated, thereby even eliminating practically what sterility may have existed. It is much more probable that the sterility arose spontaneously in this strain or that it is present to a varying degree in this species. With the character present and highly transmissible and subject to selection it is only necessary to carry on indiscrim- inate breeding to have the character appear in varying intensi- ties depending upon the chance combinations. The rule of inbreeding would be only to intensify the chance combination of the character and to insure the more or less continued presence in the successive generations. That this character of sterility is not unique to this inbred strain is evident from its rather frequent presence in pairs not inbred. In my own experience this sterility nearly always showed itself in the males. In one instance I found among a brood, besides a sterile male, two females that failed to deposit eggs although eggs were evidently present in the oviducts. Similarly Castle found in his strain a considerable amount of sterility, and this in some cases among the females. We see, therefore, that sterility is not altogether rare even in broods that were not inbred. The same facts doubtless hold for the character of productive- ness. Castle has shown this to be transmissible and amenable to selection. Inbreeding does not produce it but is instrumental, with indiscriminate mating, in intensifying it, or, if the strain be not eliminated thereby, of preserving it in the strain. INBREEDING AND SELECTION IN DROSOPHILA AMPELOPHILA 141 SEX-RATIO AND SELECTION 1 . Introductory The once rather generally accepted notion that nutrition was an influential factor in the control of sex, based on the experiments of Yung (’85), Born (’81), and others, has given place to the now as commonly accepted idea that sex is determined prior to or at the time of fertilization and is independent of the food. The experimental work of Cuenot (’99) King (’07) and others, and the splendid cytological researches of Wilson and his students are largely responsible for this change of view and have been so fre- quently reviewed in the various recent discussions of the problem of sex that they need not be further detailed here. The writer tried some starvation experiments on Drosophila in 1904. During the past year more extensive experiments were carried on under his direction by Mr. Claude D. Holmes, on the effects of starvation during successive generations upon the sex- ratio. These are published under a separate title (TO). It will suffice in this connection, to state that the results coincide with those of recent workers, namely that nutrition does not affect the sex-ratio. 2. The normal sex-ratio One fact was very apparent in these earlier tests and in all sub- sequent experiments, that, under the varying conditions in these creatures were reared, there was the same persistance of the pre- dominance of females over males. Below (table 7) is given the TABLE 7 FOOD TOTAL NUMBER REARED NUMBER OF MALES NUMBER OF FEMALES RATIO Bananas 10506 4972 5534 1:1.113 Grapes 2161 995 1166 1:1.171 Tomatoes and grapes 4048 1943 2105 1:1.083 Bananas 10218 4757 5461 1:1.14 Total 26933 12667 14266 1:1.126 142 W. J. MOENKHAUS summary of four determinations on a large scale to obtain the normal sex-ratio. The flies were reared in the following manner. Mason jars containing a large quantity of food were exposed to flies in nature. The jars were left open until the larvae began to pupate when all flies were excluded by tying a guaze over the top. As the imagos emerged from time to time they were preserved and the sex-ratios determined. For 26933 individuals, the ratio was one male to 1.126 females. In regard to these determinations only one question, so far as I can see, can be raised. This is the academic one of the greater mortality of the males during development or, to push the matter back a little further and to make it applicable to recent develop- ments in our idea of sex, the greater mortality of the male deter- mining sex cells. In reference to this it may be pointed out that the developmental conditions were as nearly normal as one can imagine. There was an abundance of food, air, light and mois- ture, and the larvae pupated in the remnants of the food in much the same manner as one finds them doing in nature. In this con- nection the experiments of Miss King (’07) on the influence of food on the sex ratio of Bufo are of importance. In this she finds that the mortality among the males is not greater than among the females. From these facts and from the knowledge that has come to me from the extensive rearing of Drosophilas for six years I am convinced that the sex-ratio in this species is not one of equality. 3. Control of sex-ratio by selection If the sex-ratio of this species, then, is that of 1 male to every 1.126 females, this should be regarded as specific just as any other of the specific characters of the species. It should, therefore, be subject to fluctuations and to control like other specific characters. Starting with this conception of sex-ratio, I wished to see whether it were possible to control this, within limits, of course, by the process of selection. The results of these experiments I propose to detail below. To apply the selective process on the sex-ratio, the following simple method was employed. Two pairs were selected from INBREEDING AND SELECTION IN DROSOPHILA AMPELOPHILA 143 nature, the one showing a high, the other a low female ratio. These were selected as the parents of the two strains to be developed. From among the offspring of each of these two pairs a number of single matings were made. From among these the pair that showed the most favorable ratio in the desired direction was selected to continue the strain. The same process was repeated as often as desired. From a number of pairs taken from a banana bunch in Bloom- ington June 12, 1907, two such pairs were obtained. These two pairs go by the numbers 206 and 207, showing the following ratio : 206— 52 <*■: 135 9 or 1:2.59 207— 84 * : 75 $ or 1 : 0.89 A. Strain 206 ( high female ratio). The 206 strain will, for convenience, be called the female strain and the 207 strain the male strain, although, as will appear, the latter never developed into a predominantly male strain. In tables 8 and 9 are given in diagramatic form the results of selection for five generations in the former and six generations in the latter. At the margin the genera- tions are numbered 1, 2, 3 etc., and the sex-ratios are indicated. The sex-ratio of the eleven pairs of brothers and sisters mated from the first generation of the female strain (206) varied from 1 :93 (76