STOUT & Boas: STATISTICAL STUDIES IN CICHORIUM 451 
The experiments of Castle and Phillips (714) show that selec- 
tion for variations in color patterns of rats led to a gradual increase 
or decrease in the amount of pigmentation, ultimately giving quite 
diverse patterns. These results have been continuously and con- 
sistently interpreted by Castle to indicate that "genetic factors 
are themselves variable" and that such factors may be altered 
gradually but permanently by repeated selection and "that one 
must reject the conception of modifying factors and conclude that 
the character has a high degree of genetic stability yet is subject 
to continuous genetic fluctuation" (Castle and Phillips, '14; 
Castle, '16a, '16b, '17, and other papers). 
Much the same sort of variability has been demonstrated, 
however, in progenies propagated asexually by Stout (715) in 
Coleus and Jennings (16) in Difflugia. Неге the fundamental 
hereditary characters are shown to be subject to slight and heredi- 
tary alterations. 
Jennings (717) points out that modifying factors, postulated 
by the opponents of the doctrine of fluctuating change in heredi- 
tary constitution to account for a series of graded variations in 
sexually reproduced and cross-bred organisms, are themselves 
alterations. 
Johannsen ('o3) claims that in respect to such a variable 
character as weight and size of seeds an ordinary population of 
beans consists of a number of genotypes (races) which can be dis- 
covered by growing and comparing the progeny of single seeds. 
Each progeny constituting what he calls a pure line was con- 
sidered as breeding true except for occasional mutations. Johann- 
sen thus sought to apply the doctrines of discontinuous mutation 
then so recently announced by de Vries and to establish the geno- 
type theory to account for the isolation of races differing in heredi- 
tary constitution (genotype). The variations appearing within a 
line of progeny were considered as purely due to environment 
(phenotypic). 
There are at least two methods of attacking the problem as to 
the inheritance of variations, such as those of weight of seeds 
within a line of progeny: 
I. By comparing the seed weights of progeny of small and large 
beans irrespective of immediate parentage within the line and 
irrespective of a consideration of individual or partial variability. 
