Chromosome Breakage by X-Rays and The Production of Eggs From Genetically Male Tissue in Drosophila

Numerous experiments have been performed to test the genetic effect of X-rays, but until the very recent work of Little and Bagg on mice no case of a permanent, i.e. a genetic, change in a chromosome or in one or more of its contained genes had been critically demonstrated to have been produced by X-rays or, indeed, by any other environic agent. But even in Little and Bagg’s work, which seems to prove that heritable abnormalities can be produced by X-rays, it is not certain whether the effects are really due to changes in the chromosomes themselves or to changes in chromosome distribution, occasioned by mitotic irregularities that result in lines with one or more chromosomes extra or missing. Mavor’s work had shown that effects of the latter type were in fact produced by X-rays, in Drosophila, but his experiments were confined to studies of chromosome and genedistribution—non-disjunction and crossing-over—and were not calculated to answer the other question, of whether permanent genetic alterations could be produced. On the other hand, experiments (unpublished) of Altenburg performed in 1918 had shown that “visible mutations” could not readily be induced by X-rays in Drosophila, at least not in sufficient numbers to be detectable by ordinary breeding and inspection methods.

Nevertheless, what we know of the nature of X-ray action would lead us to infer that X-rays must sometimes cause heritable changes in the chromosomes. This conclusion is based not so much on the experimental and cytological findings that X-rays do affect chromosome behaviour (see e.g. Strangeways and Oakley, 1923) as on the a priori considerations, first that they cannot fail to penetrate to and into the chromosomes, and second, that, having penetrated, they must at times give rise there to beta rays. The beta rays produce along their path a concentration of ions so high that it is bound to produce drastic chemical changes in practically any organic substance present there; at the same time little injury may be done elsewhere in the cell, after the ions have had a chance to scatter and become neutralised. Hence at times the X-rays must happen to produce localised chromosome changes before sufficient harm is done to the remainder of the cell to prevent its reproduction. The only question in this connection is the statistical one of how frequently the chromosome would happen to be “struck” before the cell as a whole is done irreparable injury. We are ignorant of so many of the quantitative factors concerned here that this question can scarcely be answered at present by a priori calculation, despite the suggestive work of Crowther (1924); an adequate attack upon it requires genetic experiment.

When the results of such experiments have been obtained, they must then be analysed in the effort to determine what portions of the genetic changes produced (if there are any) may have been due to the process above described, and what portion to a more indirect, diffuse after-effect of the X-rays.

In the experiments herein to be reported it has been found by genetic means that X-rays can cause permanent changes in the chromosomes, of a morphological nature. Whether these are associated with gene mutations, of a sort similar to those which cause character changes, cannot yet be ascertained. The morphological changes, which are of the nature of breakages, altering the length and shape of the chromosomes, have been demonstrated in connection with the peculiar double X-chromo-somes discovered in Drosophila, melanogaster by L. V. Morgan (1920).

The double-X, which arose “spontaneously,” in a fly derived from yellow stock, by the end to end union of two of the familiar rod-shaped single X’s, was found by L. V. Morgan to be V-shaped, like the autosomes, owing to the fact that the spindle fibre is attached in the middle, at the point of union of the two X’s. This chromosome was used in the present study because it is especially suitable for the detection of breakages (as will be explained in the next section), but there seems little reason to doubt that X-rays would be found to cause similar alterations in the other chromosomes also, if appropriate means of observation of the latter could be employed. It is true that this double X-chromosome is “abnormal,” representing an attachment between two pre-existing chromosomes (or chromosome halves); but, after all, what is normal to a species to-day was not normal to its ancestors, and we have reason to believe that this consideration applies to chromosome attachment (and breakage) as well as to other characters, the normal chromosomes of present-day forms containing various loci where in the past fusion (as well as breakage) occurred, and was perpetuated (see McClung, 1917).

As to the way in which the breakage was produced the writers believe—for reasons to be given later—that most of the breakages were brought about indirectly, and that we do not yet have evidence of changes caused by the direct “striking” action mentioned above which our knowledge of X-rays would lead us to expect as an occasional “accident.”

In the choice of genetic material for investigating possible chromosome breakage by X-rays there were three main points to be considered. First, the chromosome studied should present the maximum a priori chance for breakage to occur. Second, it must be of such a nature that individuals inheriting a broken piece would be likely to live. And third, these individuals should be readily distinguishable as such, by means of visible characters differentiating them from their normally produced sibs. The double X-chromosome is favourable material in all three respects ; none of the other chromosomes of Drosophila melanogaster are so suitable.

With reference to the first point, it should be noted that the double-X is longer than any other chromosome found in Drosophila melanogaster, and that there is consequently more chance for a break to occur in it. Furthermore, its V-shape givesit a certain intra-chromosomal morphological differentiation. As compared with a straight chromosome, then, our chances would be increased for finding an effect in one of its regions or another—just as in chromosomes II and III Plough (1917, 1921) and Muller (1925) have found that temperature and X-rays, respectively, affect crossing over in the bent portion at the apex of the V, but not in the distal regions. Lastly, as L. V. Morgan has reported (1922), the double-X does occasionally, though very rarely, break in ordinary (untreated) cultures. For the X-rays to cause breakage, then, they might only have to increase the frequency of a process already occurring.

With regard to the question of viability, it must be noted that, although the two long autosomes (called chromosomes “II” and “III”) resemble the double-X more closely than do the other chromosomes in both size and shape, and therefore in the a priori likelihood of a comparatively high breakage frequency, nevertheless it is likely that most fragments formed by breakage of these chromosomes would be lethal to the individuals containing them. In Bridges’ case of translocation it was found that when a section only 10 units long is absent from the right end of one member of the pair of second chromosomes (“II”), the individual cannot survive; if breakage is more apt to occur near the apex of the V a still larger piece would usually be broken off, and the resultant individuals would be still less likely to be viable. In the offspring of Bridges’ triploids all combinations having chromosomes II and III present in numbers different from each other (i.e. two H’s and three Ill’s, or three H’s and two Ill’s) were completely inviable. Though the ratio change here involved an entire autosome rather than a section of one, nevertheless the amount of ratio change per locus $(13)$ is smaller than in ordinary cases of loss of a section of one homologous chromosome in a diploid $(12)$⁠, and goes to indicate the relatively low viability of zygotes in which the gene proportions of the autosomes are disturbed. Even in individuals in which one member of the pair of tiny chromosomes IV is absent, the viability is appreciably lessened.

The above results are understandable when we consider that the normal development of all characters depends not only on the presence of many genes, but on their presence in normal numerical proportions, determined in the course of a long continued natural selection acting in conjunction with the gene proportions that were in existence in the ancestral line. In this line there were practically always a pair of each of the autosomes and development therefore became adjusted only to the proportions thereby existing. On the other hand, the X-chromosomes, by virtue of the sex-determining mechanism, might be present either in single or double “dose,” and so normal development had to remain compatible with a wider latitude of interproportions between X-chromosomal and autosomal genes. Hence we find that individuals with a fragmentary double-X are often viable.

In this discussion it has been assumed that when a chromosome breaks in two, one section of it is lost. This is to be expected, inasmuch as there is normally but one point of spindle fibre attachment in a chromosome, and this is fixed genetically ; the fragment lacking such an attachment would lack a necessary portion of the mechanism for effecting its transfer to the nuclei of the daughter cells during mitosis. The only way for both fragments to persist in the germ track would be through the acquisition of an additional spindle fibre attachment, located on the fragment from the region which normally did not possess one. In that case, however, segregation of a broken chromosome from its unbroken homologue should occur normally at the maturation divisions, the two fragments together going to the opposite pole from the entire chromosome (as has been found by McClung in the case of two single chromosomes versus the double chromosome formed by their union). The result would be gametes with the full complement of genes, and viable zygotes having normal gene ratios and characters identical with their sibs. Here the third factor previously referred to should enter to obstruct observation, for we would have no means of detection of individuals with the broken chromosomes except cumbersome linkage tests applied to each of them. In the case of the double-X, on the other hand, two such fragments would tend to separate at segregation, as they would be partially homologous to one another, and individuals different in character from their sibs would result.

We may now consider the method by which breakage of the double X-chromosome is detected in the breeding tests. This involves as a preliminary an understanding of the usual method of inheritance in lines containing the double-X As was shown by L. V. Morgan, females having the double-X and the recessive yellow body colour caused by the pair of mutant genes for yellow contained therein will give rise to double-X yellow daughters like the mother (inheriting a Y-chromosome from their father), and sons inheriting the single-X with the sex-linked characters of their father (grey body if the father was grey), and getting the “empty” Y-chromosome from their mother. This is shown in fig. 1. Thus inheritance of sex-linked characters in such lines is direct, or unisexual, from mother to daughter, and from father to son, instead of crisscrossing ; it might also be described as “100 per cent, non-disjunction.”

There may be in addition to the above a few viable paternally coloured offspring of “superfemale” type, having the double-X from the mother and also a single-X from the father. They will be grey if the father was grey. Such flies are always sterile, and their viability is very low, but the number of them that hatch fluctuates considerably from experiment to experiment. Aside from these flies, any viable exception to the unisexual method of transmission of the sex-linked characters which involves an entire chromosome must be caused either by equational non-disjunction of the single X-chromosome halves in the father or by breakage and disjunction of the previously connected arms of the double-X in the mother. The former event would give daughters homozygous for the paternal sex-linked characters, whereas the latter event will (when the individuals containing the disjoined arms are viable) give inheritance of the ordinary “criss-cross” type, as shown in fig. 1. By introducing into the males a character like bar eye, which is partially dominant, or by having the male dominant in respect to one or more genes and recessive in respect to one or more others, of which the female has the opposite allelomorphs, one can easily distinguish the effects of non-disjunction in the father from those of chromosome breakage in the mother. In the crosses here reported, and shown in the figure, the yellow double-X females were crossed by forked bristle (recessive) bareyed males. All yellow, non-forked non-bar sons necessarily had resulted from chromosome breakage. So also had the nonyellow, non-forked semi-bar daughters that were fertile (although the sterile three-X daughters also were non-yellow, non-forked, and slightly bar). Non-disjunction in the father would have given homozygous non-yellow, forked, bar females, but no such individuals were found.

In the previous year (1922-23) an extensive series of counts had been made, of progeny from a cross identical with that chosen for the X-ray experiments, namely, double-X yellow females by forked bar males. It was thought that these might serve as controls for the X-ray work. The total count had consisted of 30,402 males, there being an approximately equal number of females. Exact counts of females were not made, as the purpose of the experiment had been such that the interest lay chiefly in the males, but all females as well as males were examined in order to discover any exceptions to completely unisexual inheritance. Among the males not a single exception was found. There were eight female exceptions, non-yellow, non-forked, incompletely bar, these including both the class of “superfemales” and that resulting from chromosome breakage. There were at least three of the latter type, as this many were fertile; the remainder left no offspring. One of these three contained a lethal factor near the middle of the chromosome derived from breakage of the double-X; the others gave viable yellow nonforked non-bar sons. Even if we were to consider that all eight female exceptions had been derived from cells in which breakage had occurred, and that there had been an equal number of male exceptions formed, all of which had been killed by lethals, we should have a proportion of only one exception due to breakage in 3800. The minimum figure of 3 among the approximately 60,000 total gives only 1 in 20,000.

As the rarity of breaks in this extensive series of counts had been so very great it was thought safe, during the next year, to do a small preliminary X-ray test without further controls. Stock of double-X had to be obtained afresh, however, from Columbia University, as the first stock had been killed by heat during the Texas summer. Twenty-six newly hatched virgin females were exposed to the rays, bred in five mass cultures for a week, and then again exposed to a dose identical with that used before. After this they were transferred to new culture bottles, where they were left another week.

In giving each of these doses of X-rays, a current of 4 milliamperes was used, at 50,000 volts. The flies were exposed for eleven minutes, at a distance of 2.4 cm. from the target. The dosage thus used may be expressed as follows : ⁠, or 18 H (“Holzknecht units”). This mode of expression is applicable because it has been shown that in much of the X-ray work with biological material the latter is affected directly in proportion to amperage, time, and voltage, and inversely as the square of the distance from the target, and Mavor has confirmed this finding specifically for the case of the effect of the rays on crossing over.

The results, in brief, were as follows: From the first cultures, amongst 328 female and 321 male offspring, there were 7 male and 16 female exceptions to unisexual inheritance, and one intersex which was abnormal in many respects, and sterile. All five cultures produced male exceptions, though one culture gave a total of only 11 individuals. From the second cultures 54 females and 65 males were produced, including 3 male and no female exceptions.

The above results, when compared with the previous ones, seemed to prove decisively the effectiveness of X-rays in causing breaks of the chromosome. It was decided, however, to undertake a second experiment, with simultaneous controls, and to study here more in detail the time of action of the rays in the reproductive cycle. In this experiment 61 females (counting only those which afterwards proved fertile) were used as control parents, 29, called T₁, were subjected before mating to a dose of 18 H, and 29, called T₂ to twice this dose, i.e. 36 H. These females were all taken from the same stock bottle and were apportioned at random amongst these three series. Each female, whether treated or control, was placed in a separate vial, with a single male; the cultures were started simultaneously and given the same food, heating and other treatment. Every three days the parents were transferred to a new culture, and if the male had died a new one was supplied. The parents were kept for thirty days in all, being carried through ten series of cultures, and separate records were thus kept of the count of each female for each 3-day period in which she laid viable eggs.

The number of mothers in each series giving no exceptions, and of those giving the various numbers of male and female exceptions found, is shown in Table 1.

Since the male exceptions all undoubtedly arise from chromosome breakage, we may compare first the numbers of mothers producing one or more of these. It will be seen that there were 2 in the series “treated once” (18 H) and 5 in that “treated twice” (36 H), or 7 altogether among the 55 treated mothers. On the other hand, among the 61 control mothers there were but 3 giving any male exceptions. Although the chance of such a difference in frequency of male exceptions between random lots, if the X-rays had not affected the breakage frequency, appears by ordinary methods of calculation to be as much as 1 in 5, this figure does not take into account the much larger number of offspring per pair produced by the controls, and it will be seen later that when the time and grouping of the exceptions are considered the evidence for an effect of the rays becomes very strong. Evidence in the same direction is to be found in the fact that the series with the double dose gave more exceptions than the one with the single dose. On the other hand it is strange to find even three exceptions among the controls, if these had approximately the same frequency of exceptions as the untreated flies of the preceding year.

Turning now to the mixed class of mothers that produced female exceptions—whether “super females” or breakage cases—we find that among the total of treated mothers 26, or 47 per cent., gave such exceptions, whereas among the controls there were 36, or 58 per cent. Here the difference seems in the other direction, but, though a larger difference than before, it is not nearly so large proportionately to the figures concerned, and it turns out to have practically no statistical significance. However, when considered in the light of the fertility tests, which were attempted in the case of most of the exceptional females, the meaning of these figures becomes reversed. As Table II. shows, a large proportion, probably more than half, of the exceptional females produced by treated mothers were fertile, i.e. due to broken chromosomes, and this is true more especially in the case of the T_g derivatives ; on the other hand the great majority of female exceptions produced by the controls were the sterile “superfemale” forms. Allowing for this, we find a decided excess of female exceptions due to breakage in the treated series, particularly in the portion that was doubly treated. This, then, taken in connection with the parallel results concerning the males, increases considerably the probability that the X-rays were effective in causing breakage.

The results of the second experiment leave no doubt, however, that the untreated flies of the second year differed markedly from those of the preceding year, in respect to both their rate of breakage and the viability of their three-X individuals. Thus it is seen to be highly important in this work for controls to be run simultaneously, and to involve individuals taken from the same lots as those treated. What the cause of the difference in the untreated lots may be cannot be guessed at present, beyond the possibility of “invisible” genetic differences influencing the results.

In Tables III. to VI. the results of each 3-day period are separately presented, both the number of offspring and the number of mothers of each category being given. Here, on examination of the male exceptions—the most critical cases—it will at once be seen that the data acquire greater importance in relation to the question of X-ray effectiveness. For of the seven treated cultures in which male exceptions were found, five belong in the second period, in which the eggs were laid from three to six days after the treatment. None of the control cultures yielding male exceptions belong in this period. If we consider only this period the difference between the per cent, of treated and control cultures yielding exceptions (9.3 per cent, and 0.0 per cent.) becomes about three times its “probable error,” even though the total numbers are small. The chance of so great a difference arising in random sampling, even if the numbers were as small as those observed, is about 1 in 20. It can be shown to be similarly probable that the per cent, of male exceptions produced by the treated series themselves is greater in the 3-to 6-day period than later. This period of maximum X-ray effectiveness does not coincide with that found by Mavor for the effect on crossing over ; there a considerable influence was exerted on eggs laid after the first six days, and for many days thereafter, and little influence before that time. As the present series of flies were bred at ordinary room temperature in the winter time (18° to 22° C.) the greater speed of action of the rays in this case can scarcely be ascribed to heat.

Another significant difference between treated and control series is disclosed by a comparison of the mode of distribution of the exceptions amongst the “families” of the treated and control sets. So far as the families (i.e. offspring of a single pair) that contained only female exceptions are concerned no very decisive difference is noticeable ; this is to be expected because, as above explained, a large proportion of these exceptional families were so classified on account of the occurrence in them of three-X individuals, which should have a similar distribution in the two sets. But in the case of the families that produced male exceptions we find in the control series a much greater tendency for plural production of exceptions than in the treated series ; this difference becomes more distinct when we separate the treated families in which the male exceptions occurred in the 3-to 6-day period from the others.

Amongst the former families—having exceptions presumably caused by X-rays—there is no more tendency for plural exceptions than chance would readily allow. The most probable number of families that would have more than i, more than 2, etc., (male) exceptions, if the distribution were purely a random one, is given by the formula etc., where n₀ is the total number of families, n₁ the number having one or more (male) exceptions, n₂ the number having two or more, etc. The actual figures conform very nearly as closely to those so calculated as would be possible for the small number of exceptions involved ; there being 49 families containing none, four containing 1, and one containing 2 male exceptions in this period. On the other hand, of the two families in the treated series which produced male exceptions in some other than the 3-to 6-day period one contained 4 male exceptions and the other 1, and of the three control families that contained male exceptions one contained 2, one 4, and one 5 male exceptions. These latter figures are quite beyond the limits of a random distribution of exceptions and, in spite of the small number of families concerned on both sides of the comparison, there can be no question of the significance of the difference in distributions.

It is further to be observed concerning the distribution of exceptions that in those families which produced plural exceptions—whether control or treated—the exceptions were almost invariably grouped within a definite period, i.e. they tended to be found in a single culture bottle. Thus the distribution really departed much farther from that to be expected on the basis of chance than would be brought out by a mere comparison of the families as a whole.

Inasmuch as the proportion of exceptions produced in the treated series above had been somewhat less than in the preceding uncontrolled experiment, but the proportion in the controls much greater than in the untreated series of the preceding year, it was decided to repeat the experiment again, in such a way that a larger number of flies would be produced during that period which the above experiment had indicated to be the more critical for revealing an X-ray effect. Young virgin females from the same stock as that used in the last experiment were employed, and divided at random into a series of 50 (fertile) control and 155 (fertile) treated mothers. The dose of rays was increased to 55 H, which is the same as the minimum dose used by Mavor in his work on crossing over. Immediately after the raying both series of mothers were mated (each to a single male), and two days thereafter the pairs were placed, separately, into the culture vials from which the counts were to be made. They were left in these vials for eight days and then discarded.

The results are given in Table VII. Inspection shows at once the clear difference between treated and control series, which extends in this case not only to the exceptional males but also to the exceptional females. For 14 of the 155 treated cultures produced exceptional males and 23 (including 2 of the above 14) produced exceptional females, whereas only 1 of the 50 control mothers gave exceptions (males). In the treated series here the per cent, of exceptional offspring produced by breakage is considerably above that in the second experiment, and is equal to that in the preliminary treated series; the per cent, of mothers yielding exceptions due to breakage has risen similarly. The difference between the proportions of families producing exceptions in the two series of the third experiment is about eight times its own probable error, both in the case of male and female exceptions ; this could not possibly have arisen through “chance.” And if we allow for the higher fertility of the controls, and also combine the results from both male and female exceptions, the difference becomes very much more than eight times its probable error.

The reason that the numbers of female exceptions are so different in the two series of this experiment becomes evident when we examine the results of the fertility tests conducted upon them (see Table II., last line). The data show that all the females which lived long enough to receive a fair test were fertile, i.e. they had arisen through breakage of the double X-chromosome and were not three-X forms. The question of why no “superfemales” were hatched in this experiment cannot be answered, except by pointing to the well-known variability in their viability at different times : thus, for example, among the untreated families of the preceding year it will be recalled that 5 or fewer “superfemales” were found among about 60,000 flies, whereas among the controls of the second experiment about 1 per cent, of the females were of three-X type. It is fortunate for the purposes of the third experiment, however, that no “super females” were viable there, as the difference in number of breakage exceptions in treated and control lots was thereby brought out much more distinctly than in the second experiment, where “superfemales” had been so abundant as to obscure the effect of the X-rays on the production of female breakage exceptions.

If we now examine the distribution of the exceptions amongst the cultures we find that, except for three cultures that produced relatively very large numbers of exceptions, the figures are extremely similar to those to be expected on chance grounds (i.e. if there is no correlation between exceptions). Neglecting these three non-conforming cases, the actual figures of 120 mothers producing o, 26 producing 1, 4 producing 2, and 2 producing 3 exceptions are matched by the following figures expected on the basis of random distribution : 120 producing o, 26 producing 1, 5 producing 2, and 1 producing 3 exceptions; both male and female exceptions are here included as they have a similar method of distribution. The result agrees thus far with the findings for that period (three to six days) in the preceding experiment in which the exceptions were most probably caused by treatment

On the other hand, the three treated cultures that gave many exceptions (6, 10, and 21 respectively) clearly fall entirely outside the range of a “chance” distribution, as does the single control culture, which gave exceptions (9). As there were three times as many treated as control mothers, it is very probable that the above three treated cultures represented cases in which the breakage was brought about by the same general cause—independent of X-rays—as in the single control case. This too agrees with the findings of the preceding experiment, in which the controls, when they did produce (male) exceptions, tended to give several at once, as did also the treated cultures at those periods when the exceptions were probably not due to treatment (as judged by the fact that they were not more likely to occur in treated than in control series at that time).

Just as surely as we may conclude that X-rays cause breakage, so also may we regard it as established by these results that, whatever the cause or set of causes may be which bring about breakage in the controls, these factors must vary considerably in their incidence, from one experiment to another. The difference in proportions of both male and female exceptions, between the 60,000 flies of the preliminary untreated series and the combined controls of the later experiments, admit of no other interpretation. The influence or influences at work here are pretty surely neither “environic” nor “developmental,” for the numerous flies of the preliminary series varied considerably in regard to the temperatures under which they were raised and maintained, the kind of food they received as larvae and as adults, the sort of culture vessel used, and the length of time they were kept before the eggs that furnished the counts were laid, yet in no group of these flies were any male exceptions, or any considerable number of female exceptions, found. Any peculiar environic conditions that might be supposed to have caused the production of breakage exceptions by the controls of the second year must almost certainly have existed, in even more extreme degree, in one or more even larger groups of flies in the first year. By elimination, then, we may take it as highly probable that the relative absence of breakage throughout all flies of the first year was due to something in their genetic composition differentiating them from the flies of the second year. The locus of the gene or genes concerned cannot, however, be determined from the present data.

It was shown that the exceptions resulting from the breaks not due to X-rays had a much greater tendency to occur in clusters than those caused by the X-rays. How is this clustering to be explained ? On a priori grounds it too might be accounted for on the supposition that the females in particular cultures happened to have a certain genetic composition (or a peculiar environment?) which, acting similarly on many of their contained germ cells, caused breakage in a number of them. In our cases, however, genetic differences are excluded as the cause of the clustering within a given experiment because clusters were found to be limited not only to given families, but to given cultures, i.e. to batches of eggs of a given period laid by a particular female; the tendency to produce exceptions was not carried over to the eggs laid before and after that period by the same female. The environic conditions in the different cultures of the later experiments were very similar, and in the second experiment, since each mother was removed to a new culture vial every three days, no considerable differences in bacterial floras could have developed in the yeast-sprayed banana-agar while the mother was present. It therefore appears as though the clustering were not due to the action of some common genetic or environic agent acting on the different germ cells of a female separately, and we must rather conclude that the breakage in each case of plural exceptions occurred in one cell that was at a relatively early stage in the germ cycle, and that this cell left as descendants a number of mature eggs which inherited the “same” broken chromosome. On this view, then, the breakages in the controls of the second year (caused by special genes existing in them and not in the first year’s controls) must have occurred at an earlier stage in germinal development than thp breakages caused by X-rays, as the latter exhibited no tendency to clustering.

We find practically conclusive evidence for the above explanation of the clustering of exceptions when we consider the sex ratios in the groups of exceptions. Amongst the seven groups or “clusters” containing five or more exceptions per group, only three groups contain flies of both sexes, these yield a total of 15 exceptional males and 20 exceptional females. All four of the other groups, which together include a total of 30 exceptions, contain exclusively males. Such a distribution unquestionably differs significantly from the normal approximate equality in numbers of the two sexes. Most or all of the exclusively male groups, then, must contain an X-chromosome which is relatively or completely lethal to females.

This female-lethal character was evidently caused by some peculiarity in the manner of breakage of the chromosome, and this peculiarity existed only in these particular cases. For, in the first place, no such lethal effect had been resident in the double-X before breakage, as shown by the fact that the exceptional males of these male clusters had numerous viable non-exceptional sisters, which inherited the unbroken double-X. In the second place, as we have seen, no disproportionate elimination of females occurred amongst the exceptions produced by the three mothers that yielded cluster of exceptions of mixed sex, nor did this occur among the exceptions produced by the action of the X-rays (but instead a distinct tendency towards an excess of females). Hence the broken X’s in’ the offspring of a female which produced a cluster of exceptions, all males, must have been alike in some characteristic traceable to their mode of breakage, and must have been different in this respect from all the broken X’s of a cluster of exceptions of mixed sexes.

This very specific likeness in the breakage of the X’s of a given cluster, and unlikeness in different clusters, could hardly have come about by the separate action of some common agent upon the different cells of a given mother at a given time, in such a way as to break the chromosome in an identical manner in all the cells of four of the mothers, and in another manner in all the cells of the other three mothers. Rather must we conclude that the peculiar chromosome which is lethal to females was formed in a single ancestral oogonium in each of the four cases in question, and this is why it exerted the same peculiar lethal action on all the female offspring that received it.

What could there be about the manner of breakage of the double X-chromosome that would render females receiving it inviable, but not males ? A mutant sex-limited lethal gene, dominant in females over two normal allelomorphs but not dominant in males over even one normal, would formally satisfy requirements, supposing the gene to be on the duplicated portion of a broken double-X possessing one perfect and one incomplete arm. Or a mutant sex-limited lethal that is dominant in the female over one normal allelomorph, but is not lethal at all in the male, would give the results obtained, if the gene were located on an unduplicated portion of a broken double-X. But it would be highly improbable for such a special kind of mutation as either of these to have arisen four times independently in the broken X’s, and never to have been observed formerly in lines having single X-chromo-somes. On the other hand, these are the very effects that would have been produced if a large enough portion of the broken arm of the double-X had remained attached to the unbroken portion (forming a J-shaped chromosome) to cause the death of females receiving the J from the mother and a single-X from the father. Inasmuch as three-X “superfemales” are usually inviable it would be very improbable if all zygotes having just less than three but more than two X-chromosomes possessed normal viability. If, then, the double-X may break at any point along the length of one of its arms we should expect some of the resulting chromosomes—those more nearly like single-X’s—to give viable females when in combination with a single-X from the father, and others—those having a longer attached arm—to give inviable or poorly viable females.

The fact that both kinds of broken double-X’s—those lethal and those not lethal to females—were produced by breakage in the controls accordingly indicates that the locus of the breakage not caused by X-rays is variable, being sometimes nearer and sometimes farther from the point where the two arms are attached to each other. In the cases where exceptional males only are produced it is evident that, even though the imperfect arm of the J-shaped chromosome is relatively long, nevertheless it cannot contain the sex-determiners, since this J, when in combination with the Y-chromosome from the father, yields an apparently typical male. The cause of death of the females here would consequently lie, not in an excess of the sex-genes themselves, but in an excess of other sex-linked genes, in proportion to the rest of the genetic make-up.

The conclusion that the locus of “natural” breakage of the double-X is variable was also arrived at by Morgan, Sturtevant, and Bridges (1923), presumably on the basis of evidence of a different and more direct kind than that given above.

Consideration of the chromosome configuration in the germ cells in which the double-X has broken “spontaneously” (i.e. not because of the X-rays), reveals a curious situation. We have seen that the evidence shows that these breaks commonly occur asymmetrically, so as to produce a J-shaped X and an incomplete fragment smaller than a single-X. Inasmuch as the latter fragment lacks the locus at which spindle fibre attachment may occur, it is obvious that it cannot be transported at the time of cell division and hence, failing to become incorporated in the nuclei of the daughter cells, must finally disappear, as similar fragments are known to do elsewhere. Since the breakage occurs in most of these cases in oögonia (as shown by the clustering of exceptions), other germ cells are thus produced by cell division which contain as their complement of sex chromosomes only the J-shaped X, together with the Y-chromosome—which, it will be remembered, is also present in double-X females. These cells, then, have an XY combination much like that occurring in all diploid male cells, including spermatogonia. They must contain, moreover, the proper proportions of genes to produce maleness, in spite of their X-chromosome being J-shaped, since individuals of the next generation which receive the J-shaped X, and a Y, are found to be males. Nevertheless these XY cells formed in the females eventually produce functional eggs, not spermatozoa. Since genetically these cells are male, their sex must somehow be “reversed,” so that instead of being regulated by their genetic composition they act as true female cells.

There seem at first to be only two possible explanations of this non-expression of the genetic sex. One is that the surrounding female tissue, through influences such as hormones, forces the XY cells to act as female. Such an explanation is scarcely in accord with the prevalent idea that sex hormones do not exist in insects, and with the experimental findings that gynandromorphs in Drosophila sometimes have gonadic as well as somatic tissue of both sexual types. If it be argued that in the present case the supposed influence might be transmitted through direct cell-contact, it should be observed that the XY cells do not occur singly, surrounded everywhere by female-type tissue, but must occur in groups (derived from an ancestral oogonium), and in these groups there must be included not only directly germinal XY tissue but also XY cyst nursecells that were derived from the XY germinal cells not long before the growth period was entered upon. The idea of the XY cells being forced to form eggs by influences originating in XX tissue is therefore highly improbable.

The other explanation of the “sex reversal” is that the original female “tendency” that is present in the cells before breakage occurs may reside in persistent substances or structures that are able to determine egg development even after the genetic cause of these substances has disappeared. If this were the case, we could calculate through approximately how many cell divisions such substances must have been able to persist and retain their effectiveness. One of the mothers in which the asymmetrical breakage occurred produced 10 male exceptions. These must have come from eggs containing the J-shaped chromosome, fertilised by Y-containing sperm. There were probably about an equal number of eggs with the J which were fertilised by X-containing sperm and which consequently died. And to parallel this total of 20 eggs that had received the J-shaped X there were most probably another 20 which received the Y, but which originally had a broken X that entered the polar body. Similarly, the female which produced a cluster of 21 exceptional males and females probably also formed at least 40 eggs, altogether, from cells which before reduction contained a broken double-X. Now to produce 40 cells requires at least six cell divisions. But it has been found by Plough (1917) and others that the last four cell divisions in Drosophila, before maturation, only separate the oöicyte from nurse-cells in the cyst about it. Hence the breakage must have occurred about ten cell divisions before the oöcyte was formed. In each division the hypothetical “female substances” would have had to be diluted by half, and since $(12)10$ is a little less than $11000$ these substances would probably have been diluted over one thousand times before their specific result—oogenesis—began to be produced. This in itself seems practically a reductio ad absurdum.

We are thus confronted with a peculiar dilemma. Apparently cells of male genetic composition metamorphose into ova neither because of any special substance or influence exerted upon them, derived from other cells of the female body, nor because of any “egg-forming substances” that might be present within their own protoplasm—derived from an earlier period in which they still possessed the female gene-complex. How then is the oogenesis of these XY cells to be explained ? There appears to be but one remaining possibility worth consideration. That is, that the process of oö genesis really—chemically—begins in much earlier germ cells than the oöcytes, even though its first conspicuous morphological effects appear in the latter ; and that, after the early period in which oogenesis is initiated, each step in the process in turn produces the next reaction of the series, in chain-like connection, until the definitive ova are laid down, without the further direct intervention of the genetic determiners that decided the nature (whether male or female) of the original reactions.

Thus we find that (in Drosophila) what has been considered the most distinctive and fundamental process in sex-determination—namely, gametogenesis—falls into line, from the standpoint of its developmental physiology, with the amphibian limb, the lens of the eye, and various other structures recently investigated in the embryogeny of diverse forms, in that the later course of development of the cells in all these cases is early determined through certain (as yet) invisible reactions, which subsequently lead through a regular sequence of linearly interdependent steps to the final organ or character, while other cells, of identical genetic composition, and in an identical somatic environment, may meanwhile be travelling an entirely different developmental road, merely because they were given a different start. In the case of gametogenesis we know in addition that the sex-determining gene-complex enters in a differentiating rôle to decide the nature of this start (but only of this start) ; in the case of the somatic characters referred to we do not yet know any of the genetic factors involved in starting the “self-differentiation,” because organisms differing in these factors have not been bred together and studied genetically.

The above conclusions should not be construed as meaning that after the beginning of such “self-differentiation” the genes within the cells in question no longer play a part in development, nor that many somatic influences from other regions of the body are not still essential, but only that those particular genetic differences that were found to determine the initiation of the sequence of reactions no longer exert an influence on these cells (except through the processes they have already started), and, similarly, that certain changes, or differences, in “external” somatic influences, of given types which were studied, no longer have an effect on the differentiation of the cells under consideration, although they may have played an important rôle in earlier stages.

We have seen that the clustering of individuals containing broken chromosomes in the controls indicates that breakage there usually occurs a considerable number of cell-generations previous to maturation. Conversely, the fact that, with a few striking exceptions obviously caused by the same factors as those operating in the controls, the individuals arising from breaks in the treated series of all the experiments conform closely to a random distribution shows that these breaks must have occurred in late stages of the germ cycle. In the second experiment this idea receives corroboration by the direct method of studying the time of appearance of the individuals containing breaks, in relation to the time when the X-rays were administered, and the period of breakage determination is thereby fixed as being either in the very late obgonial or the very early oöcyte stage, for these are, according to the work of Plough and Metz, the stages of development of Drosophila eggs three to six days before laying. No breakage Weis produced in germ cells in earlier stages of the cycle, for if this had been the case groups of exceptions (each group arising from a single affected gonium) would have occurred in the treated series in greater numbers than in the controls—contrary to the actual findings—and periods later than that of three to six days would have shown a larger proportion of exceptions in the treated-than in the control series.

Designating late oögonia or oöcytes three to six days before laying as in the “susceptible” stage, we can calculate from the results the approximate per cent, of susceptible cells in which breakage was caused by the X-rays. In the calculations based on the second experiment only the males can properly be consider, since the female exceptions were not all due to breakage; this makes the total numbers for the 3-to 6-day period—1 exception in a total of 245 in the T₁ series, and 5 in 172 in the T₂ series—very small and necessarily subject to much error. The resulting per cents, of exceptions—0.4 per cent, for T₁ and 3.0 per cent, for T₂—are nevertheless similar to those obtained in the third experiment, where larger numbers were obtained, and where the numbers of both females and males are significant. In the third experiment, excluding the three clusters of exceptions, there were 40 exceptions in the treated series, in a total of 2024 flies. The mothers however, were in the cultures from the second to eighth day after treatment. Assuming that all these exceptions were produced between the third and sixth day, and that about 3/5 of the total flies counted came from eggs laid then (since fertility is still low on the third day, and conditions for viability are worse on the later days), we find that 3.3 per cent, of the susceptible cells were affected here.

Examining further the results of those breaks that were caused by treatment, we find a sex count of 45 females and 27 males amongst all the exceptional individuals thus produced in the third experiment. There would be a chance of about 1 in 20 of obtaining figures deviating as widely as this from a 1 : 1 ratio, if the material really tended to give equality of the sexes ; and a chance of 1 in 2 of figures differing this much from a 1 : 2 ratio, if the latter were the “true” ratio in the material. As between these two possibilities, then, the odds are about 10 to 1 in favour of a 2 : 1 rather than a 1 : 1 ratio, although of course it may well be that not all the females were alike in their sex-producing potentialities. The matter could be decided by individual tests of a large number of the exceptional daughters to determine how many of them carried a sex-linked lethal gene, and hence gave 2 : 1 ratios; but unfortunately the writers did not have the opportunity to make such tests. It will be recalled, however, that one of the three exceptional females tested in the first year did prove to contain a lethal in her chromosome derived from the broken double-X ; this lethal was near the locus of “vermilion,” in the central region of the chromosome.

It is rather to be expected that lethals would be relatively frequent in the double-X, and consequently in its breakage derivatives, because there would be a tendency towards the automatic accumulation of recessive mutant genes (of which lethals are by far the most numerous) in this chromosome. For ordinarily natural selection acts to weed out sex-linked lethals, through killing the males carrying them; in the case of the double-X, however, the individuals carrying the chromosome are all females, in which the eliminating action of the lethal in one arm of the chromosome could not ordinarily be exerted because of the presence of a dominant normal allelomorph in the homologous arm. Crossing over between the arms of the double-X might, however, allow some elimination of lethals by causing both arms, homozygously, to contain the lethal. This could only occur if crossing over took place in the “4-strand stage,” and the lethal-bearing section of one chromosome-half that had been derived from the preliminary equational splitting could change places with the homologous non-lethal-bearing section of the other half.^* Hence, without more knowledge of crossing over conditions in various parts of the double-X, it is not possible to predict the rate of lethal accumulation within it merely on the basis of previous work on the rate of origin of sex-linked lethal genes. It will be seen, nevertheless, that the above considerations afford a distinct basis for expecting a larger number of 2:1 sex ratios among the breakage exceptions, and among the offspring of the exceptional females, than among the general population, and that to explain such an occurrence we do not have to invoke speculative hypotheses like the idea of a lethal being formed by the act of breakage.

The frequency of all-male clusters of exceptions among the controls led us to the conclusion that in breakage not caused by X-rays the position of the break is variable, so that sometimes a relatively large, sometimes a small fragment remains attached to the unbroken arm of the chromosome. In the case of the exceptions caused by X-rays there are no clusters formed to guide us concerning this question, but we might use the total sex ratios shown by the X-ray exceptions to judge whether or not chromosomes with a large enough fragmentary arm to be lethal to females were common among them. The count of 45 ♀ : 27 ♂ would indicate that this is not the case, and that therefore the treated chromosomes are usually broken at or near the point of union of the X’s. Here, however, we enter into the difficulties discussed in the preceding paragraphs, inasmuch as we do not know how often recessive sex-linked lethals occur to kill off the exceptional males, and so we cannot judge from these data alone how much higher the proportion of males would have been in the absence of this oppositely acting influence

Fortunately, this disturbing factor may be eliminated to some extent by a comparison of treated with control sex ratios among exceptions, cis there would be the same chance for preexistent recessive lethals to reduce the proportion of males in control as in treated lots, and so a significant difference between the respective sex ratios would be referable to a difference in the mode of breakage. The exceptions among the controls included a great majority of males, as we have seen, giving a sex ratio utterly unlike that found among the treated exceptions. This result, then, would argue for a real difference in mode of breakage under the two conditions, breaks in the controls having a tendency to leave a larger fragment attached to the unbroken arm of the X than breaks in the treated ; in other words, breaks caused by treatment would be more apt to occur near the attachment point of the V-shaped chromosome. This conclusion cannot be regarded as thoroughly established, however, until a much larger number of cases of “control breaks” have been found, for comparison of the sex ratios here with those in genetically identical treated material, or until the work can be repeated with double-X’s known not to contain recessive lethals.

The first question that may be raised here is whether the action is direct or indirect. By a direct action is meant a localised destruction of chromatin through chemical change, caused by ionisation which happens to have been initiated at the particular point in the cell where this bit of chromatin is lying. For, as indicated in the introduction, it is commonly held that the chemical action of X-rays is first exerted unevenly and discontinuously, through Beta rays produced by them at certain points where, owing to the exact positions and modes of motion of the electrons and the local intensity and mode of incidence of the incoming X-waves, the energy of the latter could be absorbed and translated into atomic and molecular alterations. The Beta rays, which are fast-moving electrons, then produce an ionisation that is at first concentrated along their immediate path. Since it is probable, however, that one portion of the chromatin would be as likely as another to be affected by these “hit-or-miss” but supposedly drastic darts—which, so far as the directly affected part is concerned, might almost be compared to bolts of lightning—we must conclude that the breaks which were found were probably not brought about in such a direct fashion. This is indicated both by the probable localisation of the breaks caused by X-rays near the apex of the V, and by the fact that they are caused to occur only in the cells undergoing maturation.

It is more likely, then, that in causing the breakage the rays act indirectly, by bringing about a general change throughout the cell, or cell nucleus, which in turn reacts upon the chromosomes so as sometimes to induce their breakage. This recalls Mavor’s conclusion regarding the mode of action of X-rays in affecting crossing over, for he has inferred that this action too is indirect, since it can be exerted long after the application of the rays has ceased.

In view of the above correspondence between the breakage and the cross-over effects, we are led to ask the question, whether the peculiar action of X-rays in breaking the double-X in a given region, and only at maturation time, may not be due in some way to an influence of the radiation upon crossing over between the two identical arms of the double-X. Is crossing over simply made so abnormal that breakage of the chromosome sometimes results (at its weakest point, or point of greatest strain) ? This hypothesis may be pretty definitely answered in the negative, for the careful work of Mavor has demonstrated that the effect on crossing over is exerted mainly in ova which were exposed to the rays more than six days before they were laid, and that the effect continues to be found even in those treated as long before laying as fifteen days, whereas there is no doubt that the breakage of the double-X is produced mainly in eggs treated three to six days before laying, and little or not at all in those treated much earlier. The breakage action, then, although probably indirect, cannot be a result of the action upon crossing over, but it is dependent upon something else occurring at or near the time of maturation.

Our conclusion that the breakage caused by X-rays was probably indirect does not mean that X-rays may never break a chromosome by means of a direct “stroke,” but it indicates at least that with practicable doses such an action would be extremely rare, and breakage through some indirect effect much commoner. In so far, now, as this evidence can be taken to indicate that localised drastic effects due to such doses of X-rays as are consistent with fertility do not often occur, it also becomes less likely that single gene mutations or deficiencies will often be brought about through such a direct action.

That mutations could be caused indirectly, through changes in the general conditions in the cell brought about by X-rays, would remain a possibility, though a rather speculative one, considering the failure of all previous attempts to prove that mutations can be caused by means of substances given in the food or by general environic changes. If genetic changes did result from an indirect action of the rays, however, there would seem to be more chance for specific types of mutations to be caused repeatedly by this means—as Little and Bagg seem to have found — than if the changes resulted from a random “striking” action. Moreover, regional deficiencies, since they must depend on a spatially defined event, smaller in extent than a chromosome but larger than a gene, should be still less likely than mutations to follow from an indirect action.

Some experimental evidence with regard to mutations and deficiencies may be gained from the present work. For 19 of the control females in the second experiment were mated with (forked bar) males that had been treated with 18 H of X-rays. If a “visible” mutation had occurred in the X-chromosome of a male, in a cell destined to give rise to a male offspring, it would have been evident in the latter. A mutation, deficiency, or loss in this X-chromosome in a region including either the locus of non-yellow (left end) or that of forked or bar (near the right end) would have been evident in the female offspring. Yet no such cases were found amongst the 2587 offspring of these males that were counted. This is also evidence against the common occurrence of breaks in the single X-chromosome, when X-rays are applied. It may be noted further that not even non-disjunction of the two halves of the single-X in the male was found to occur in the cells of these X-rayed males; as explained in the second section, this would have been detectable through the appearance of non-yellow, forked bar females, but these did not occur here, or in any of the other counts. The dosage used, however, was not large, and the offspring in question are not numerous enough to disprove a slight effect of the rays in any or all of the above respects.

In order to obtain additional evidence concerning the possible production of deficiencies, crosses were made of X-rayed males homozygous for the six recessive genes for roughoid, hairy, scarlet, pink, spineless, and ebony, which are scattered through the third chromosome, to X-rayed females homozygous for the normal allelomorphs of these genes. A dosage of 55 H was used. Any viable deficiencies in the eggs, of mutations, or losses due to breakage, which included one or more of the above loci, would result in exceptional offspring showing the corresponding recessive character. Among the 772 offspring counted, however, all were normal. Owing to the fact that deficiencies in autosomes are less likely to be viable than those in the X-chromosome, these figures too may be little more than “straws in the wind,” yet their agreement with the preceding data and with the ideas above arrived at at least indicates that such effects of X-rays are not common.

It may be that the abnormal intersex found among the offspring of the first flies treated involved a mutation or deficiency, but as this fly died after a few days without leaving any offspring there is no way of testing this possibility. The fly was non-yellow, non-forked, slightly bar, and hence must have contained an X-chromosome from the father and a double-X or a part of a double-X from the mother. Most probably it was, in the main at least, a triploid like the triploid intersexes found by Bridges that have two X’s and three of each of the long autosomes, but it was not a typical triploid intersex. It would be a peculiar coincidence, however, if mutation and triploidy had occurred simultaneously.

If the above intersex represents a case of triploidy caused by X-rays, then X-rays can cause non-disjunction of both long autosomes as well as of the X-chromosome. Mavor (1921) in two of the papers in which he proved the effect of X-rays in producing non-disjunction of the X’s, drew attention to its apparent non-occurrence in the other chromosomes, but it is probable that the lack of data indicating such a phenomenon is rather due to the difficulty of detecting the effect, and the inviability of the zygotes which would thereby result. Only in cases where the non-disjunction involved both sets of long autosomes simultaneously would individuals having a viable balance of genes result, as Bridges’ findings regarding the viability of different classes of offspring of triploids prove; such individuals, if they reeeived three sets of long autosomes, would be “supermales” or intersexes, according to whether they received one or two X-chromosomes. Fertile triploid females would result only from non-disj unction of the X’s, occurring simultaneously with non-disjunction of both long autosomes; since in the double-X stock the X’s are permanently joined there should be a far better chance here of obtaining triploid females than in stock containing ordinary X’s. It had been hoped to obtain double-X triploids by the X-ray treatment, since these would solve many genetic problems, but only the one intersex appeared.

In order to obtain further evidence concerning the effect of X-rays on non-disjunction in the autosomes the writers, at the time of the third X-ray experiment with the double-X, treated with the same 55 H of X-rays some males and females containing different recessive fourth chromosome genes (eyeless and shaven respectively), in order to get some data from this chromosome on the point just raised. The fourth chromosome is a better indicator of non-disjunction than the other autosomes, because otherwise diploid individuals having one or three of these chromosomes are fairly viable, as Bridges has shown. Among a count of 851 offspring two shaven flies, caused by non-disjunction in the male, were found, but none occurred in a much larger count of flies from a similar cross of untreated parents.

We are not in this study primarily concerned with the three-X individuals that occurred, but it is of interest to note that among these no tendency to clustering was found, either in individual cultures or in families. The causes leading to viability of the triple-X’s must consequently have acted at random, there being no differences of either genetic or environic type between the mothers or the culture bottles of a given experiment, of such a sort as appreciably to influence the relative chances of development of the triple-X individuals. Nevertheless, since numerous three-X forms were found in the second experiment but a negligible number in all the other experiments combined, there must have been some agent, capable of acting on an entire series of cultures, which influenced “superfemale” viability. This agent was probably not environ ic solely, as the cultures during the first year had been so varied in their environments ; probably, therefore, genetic factors influenced the viability of the “superfemales.”

It may, finally, not be out of place here to call attention to another result, unconnected with the primary problem, which issued from the experiments with the double-X. As each of the double-X females, both in the first and second year, had been crossed with forked bar males, a total count was obtained of over thirty-five thousand sons that had derived their X-chromo-some from a bar-eyed father. If the reverse “mutation” of bar eye to normal occurs in the male with a frequency at all similar to that in the female (one “mutation” in 1500 to 2000 gametes), a fair number of these sons (20 ±) should have had normal eyes. Every one, however, showed the bar eye. This affords confirmation, on a considerable scale, of Sturtevant and Morgan’s finding that the reverse mutation of bar is dependent on crossing over, for of course there was no opportunity for crossing over in the bar-eyed male parents since crossing over occurs only in the female of Drosophila.