Genetics 3(5): 422-499 (1918)
The Rice Institute, Houston, Texas.
[Received January 14, 1918]


In numerous breeding experiments there is positive evidence that the factors concerned undergo no sensible fluctuation, nor sensible contamination during segregation. But, unfortunately for a clear and simple proof or disproof of the generality of these principles, Mendelian theory demands, and experiment has proved, that not infrequently multiple factors and other complications quite consistent with factor constancy should be met with, that lead to results superficially similar to those which would arise from factor variability or contamination, although not really due to such a cause. Consequently, unless apparently irregular results have been subjected to a rigorous factorial analysis, they are equally reconcilable either to an explanation consistent with, or to one contradictory to, the "genotype conception." But the fact that these particular results cannot be used as arguments for factor constancy does not mean that they are arguments against it, and valid evidence against factor constancy could not be secured except by analyzing such results and proving that factor fluctuation is the only (or the more probable) explanation.

As a matter of fact, whenever a factorial analysis of the questionable cases was possible and has been made, evidence of the presence of the multiple factors or other complicated genetic phenomena to be expected on the strict genotype conception has always been obtained, and in no case has it been found possible to show that the factors underwent fluctuating variation or contamination. Of course, this furnishes "presumptive evidence" for all such cases, and, taken together with the positive evidence for factor constancy previously referred to, tips the scales heavily on this side. Opponents of the latter view have thus found it necessary hitherto to base their argument on the questionable cases in which factorial analysis was impossible or lacking, and they have not admitted that these cases really supply an argument for neither view. Accordingly it seems necessary to find more typical cases of the above sort which can be dissected, and to expose the factor mechanism there involved if we are ever to reach a decision that will be accepted by both sides.

The case of beaded wings in Drosophila appears to be particularly favorable for this purpose, from the point of view of both contending parties, because its inheritance seems most markedly "non-Mendelian," and because it exists in the organism best suited for a rigorous factorial investigation. Beaded was the third variation discovered by MORGAN in these flies, and it and truncate from the start formed the two great unconformable cases that quite refused to fit in with the ideal Drosophila scheme of things. The following are some of the more important peculiarities of this character and of its mode of inheritance.

(1) In the first place, the character is irregular in its manifestations. It consists of a sort of lopping off of the edge of the wing, which may be evinced at any point and to almost any extent (see figure 1). In some flies only a small piece of the margin seems gone; in others the margin may be cut into at intervals—if this happens along the outer edge of the wing it results in the "beaded" appearance, on account of the breaking up of the marginal vein; in still other flies a large part of the border of the wing seems to have been cut away, along either a straight or a wavy line, and at either the inner or outer edge, or both. The exact nature and amount of the deformity is seldom the same in the two wings of the same fly, although a good deal of correlation does exist. These differences between the two wings cannot of course be ascribed to heterozygosis, unless we accept the extremely dubious doctrine of somatic segregation; they are therefore an indication either of factor inconstancy or of "somatic" (non-genetic) fluctuability. Whether or not the differences between the wings of different flies is due to one of the latter causes could not be decided; however, by these observations alone.

FIGURE 1.—Flies from "pure" beaded stock; an "average" beaded on the left, an "extreme" beaded on the right.

(2) (a) In the second place, when MORGAN bred beaded flies together, he obtained, besides the beadeds of various types, some flies with wings entirely normal, and for years he was unable, by selection, to secure a strain consisting entirely of beaded flies. (b) This latter variability was proved to be not merely somatic, for it was found that the normals produced in this stock gave, on the average, a much larger proportion of normal flies among their offspring than did the beaded flies. (c) The tendency towards normal on the whole overbalances that towards beaded, for it was found, when the beaded stock was allowed to run for many generations in mass culture without selection or supervision, that the beadedness tended to disappear from it more and more, i.e., there was reversion towards normal.

If we believe in factor constancy, these genetic variations so continually occurring in the beaded stock must be referred to a different source from the somatic variations between the two wings of the same fly, and in order to account for them we would have to suppose that beaded flies for some special reason are always heterozygous, like Andalusian blue fowl or like yellow mice. If, on the other hand, we believe in factor fluctuation, all of the above results, of both preceding paragraphs, are directly explained on the simple basis of a variability in the factor for beaded itself, and it is then further to be noted that the results would fall into line with the widespread notion that factors vary more in the reverse direction-regressing towards their ancestral conditionthan in the original direction of variation.

1Or else multiple allelomorphs.

(3) (a) Not only the variation from beaded to normal, but also the variation in the intensity of "beading" was shown to be at least partly genetic, because MORGAN was able by selection to obtain stocks differing distinctly and consistently in the degree of their beading; in fact (b) he succeeded, by selection, in bringing the extreme and the average intensities of this character both far beyond their original points. These facts too can be covered by the one assumption of factor inconstancy; denying this interpretation, we must add to our previous apparently unrelated assumptions of somatic variability and persistent heterozygosis the assumption of multiple factors.1

(4) The climax came when, after very many generations of the above selection work, a stock pure for beaded was, in some inexplicable way, finally obtained. The result came almost suddenly, as though a certain critical point in the "strength" of the factor had been passed, owing to the long selection; or, to be more specific, as though the beaded flies were now so far removed from their normal ancestors both in number of generations and in intensity of character that reversion to the latter no longer took place. In this more stable stock the character was on the whole well developed, and although it was still somewhat variable its range of fluctuation was not as wide as before, and it seldom approached very closely to the normal type. This result, like the others, falls naturally in line with the view of factor variability; the alternative view would require still another postulate, apparently disconnected with the previous ones,—namely, that a new factor or factors had appeared in the stock, which somehow did away with, or obscured, the tendency to "reversion."

To sum up, all the different peculiarities evinced by the character beaded point consistently to the explanation of factor inconstancy, and, neglecting the evidence against such a conclusion afforded by the great body of work in genetics, it would seem by far the most probable interpretation here; to avoid it, at least four apparently unrelated assumptions, two of them seemingly unusual to a high degree, are necessary.

An elaborate study of the character, beaded wings, was carried out several years ago by DEXTER (1914). His extensive and exemplary investigations have, besides confirming points mentioned above, brought to light a number of further facts which have an important bearing on the problem. He has found, firstly, that the character is markedly affected by various environmental conditions; drought, for example, tends to inhibit its appearance. These effects, if they are really exerted on the factor itself, must be but temporary, as succeeding generations, if brought up under the original conditions, will exhibit the character in its original form. In other words, there exists in beaded stock a sort of variation that is purely somatic, which is caused by external influences. Secondly, DEXTER found that beaded is linked to pink eyes and to ebony body color, and a factor for beaded must therefore lie somewhere in the third chromosome, as do the factors for both these characters. In some cases, however, he found that an intensifying factor was also present in the beaded stock, and this factor proved by its linkage relations to lie in the second chromosome. Hence there are at least two pairs of factors concerned in the character beaded, one chief factor in chromosome III, and one accessory factor in chromosome II. Two of the four postulates which I have shown to be requisite for an explanation of the beaded case on the theory of factor constancy were thus proved by DEXTER to be true: these are the postulates of "somatic" (non-genetic) variability and that of multiple factors. This would seem to be rather a strange coincidence if the results, or part of them, are really due to factor variability. DEXTER accordingly concluded that the explanation of the peculiar behavior of beaded was to be found not in factor variability, but in complications of a different kind.

There was as yet, however, no experimental basis for the other two suppositions which the adherent of factor constancy must make,— namely, that the stock was at first necessarily heterozygous, and that later a new factor or factors appeared, that somehow obscured or did away with this heterozygosis. Moreover, these latter postulates were, after all, much more unusual than the other two, and so the more important part of the problem still remained with only a vague and entirely hypothetical explanation. It is true that DEXTER had found that the intensifying factor in chromosome II was not present in all the beaded flies which he examined, and that on the basis of this he had suggested, as an explanation of the difficulties in question, that perhaps this intensifier might act as a lethal, like the factor for yellow in mice. For if this were true, it would be impossible to obtain stock pure for the intensifier, and since flies without the latter might sometimes be normal in appearance, an all-beaded stock could never be obtained. But the mere fact that the intensifier was not present in all beaded flies examined affords very little evidence for the assumption that it is lethal, and a serious objection to such an assumption might perhaps be found in the fact that in some strains at least of the pure stock that was finally obtained, all of the flies are beaded to a high degree, as if the intensifier was present in all.

These difficulties and contradictions gave an air of unsoundness to the whole argument for factor constancy in the case of beaded wings. Opponents of the genotype idea, had they come across the case, would undoubtedly have urged with much conviction (for themselves at least) that the discovery of an intensifying factor and of somatic variability did not by any means clear up the chief uncertainties of the case, nor weaken very much the argument which this case, taken by itself, unquestionably provided for factor variability.