BY THE SELECTION OF SOMATIC VARIATIONS. 11 



that the hereditary bearers of characters appear to be split up and 

 modified by crossing, giving in some cases quite new characters. 



Thus in the period of less than 20 years the literature passes from the 

 confident treatment of characters as units with a simple shorthand 

 representation to a discussion of "factors" that under some conditions 

 work together as a multiple unit and under other conditions separate, 

 producing equal or different expressions. The assignment of different 

 values to the assumed factors as diluters, intensifiers, inhibitors, and 

 the conception of multiple factors that can separate out, giving aber- 

 rant ratios or new expressions and even almost endless intermediate 

 gradations, both of so-called qualitative and quantitative characters all 

 reduced to descriptive terms, add nothing to the fundamental concep- 

 tions of Darwin regarding variability in hybridization. The extreme 

 application of the multiple-factor hypothesis simply means that small 

 variations are inherited equally as well as are large variations. 



The greater number of Mendelians, mutationalists, and adherents 

 of the doctrines of pure lines seem to hold that the unit factors are 

 changeless. Others still accept Darwin's general views of the modi- 

 fiability of the fundamental units. The latter view is especially well 

 developed by Castle (1912), who states (p. 356) : 



"In my experience every unit-character is subject to quantitative variation, 

 that is, its expression in the body varies, and it is clear that these variations 

 have a germinal basis because they are inherited." 



Morgan (1913) considers that factors are labile aggregates subject 

 to rearrangement, that processes of mutation and reversion are reversi- 

 ble, and that in eversporting varieties mutation and reversion are regu- 

 lar processes. 



Bateson (1902, p. 201), in a defense of early Mendelian views, makes 

 the following admission: 



"We have to consider the question whether the purity of the gametes in 

 respect to one or another antagonistic character is or is likely to be in the case 

 of any given character a universal truth. The answer is unquestionably No, 

 but for reasons in which ancestry plays no part." 



More recently (1914, p. 322) he has expressed the view that the 

 conception of multiple factors is in his mind an admission that there 

 are imperfect segregations. To quote further: 



"Segregation is somehow effected by the rhythms of cell-division, if such an 

 expression be permitted. In some cases the whole factor is so easily separated 

 that it is swept out at once; in others it is so intermingled that gametes of all 

 degrees of purity may result." 



Thus it seems that the short-hand system of representing hypo- 

 thetical germ-cell units is often not only cumbersome but inaccurate. 

 Some phases of it may be useful as descriptive terms, but the method 



