Ernest Warren 
327 
way from some other character, and according to de Vries it arose by a sudden 
mutation. It is clear, however, that the somatic nature of an individual tends to 
i:eflect its gametic constitution, and obversely any change in the somatic nature 
of an individual tends to imply a change in the determinants of the germ- cells ; 
and therefore the factor or determinant, say, of the general colouration of the 
corolla in a pale purple heterozygous dominant foxglove differs to some extent 
from that in a darker purple heterozygous plant: and the offspring of these two 
plants, if crossed with the same pollen, and bred under the same environment 
(in its widest sense) would not be the same. I have deduced some evidence in this 
investigation to show that the relationship of parents and offspring with respect 
to the fluctuations in the intensity of expression of a Mendelian character is similar 
to that of a non-Mendelian character provided the parents and offspring are of 
the same class, i.e. all heterozygous or all homozygous. It follows from this that 
a slight variation in the nature of the factor in the parent is accompanied by 
a slight somatic variation in the offspring and will be subjected to natural 
selection. 
It appears to be premature to attempt to visualize these complicated pheno- 
mena in the nuclear behaviour of the sexual cells, although Morgan and others* 
have made ingenious attempts at so doing. In any such attempt it seems necessary 
to assume the existence of factors, determinants, plasms or something, no matter 
what it is called, which bears the heredity influence. Speaking in these mechanical 
terms it may be suggested that characters which exhibit no discontinuity in their 
inheritance, i.e. show no Mendelian relationships, are transmitted by a determinant 
or a system of determinants uniformly distributed through the chromosoines, 
and perhaps also through the cytoplasm of the germinal cells; while those 
characters which do exhibit Mendelian relationships depend on determinants 
which are localized in the chromosomes only, and possibly in definite places in 
the chromosomes. Morgan has recently referred to this last conception as the 
linear arrangement of factors in the chromosome. 
In both cases, whether localized or not, the determinants controlling a particular 
character are variable in nature and are therefore subject to gradual modification 
by the natural selection of the individuals which bear the genital cells. 
Mendelian inheritance in its original simple form, as it sometimes appears in 
the crossing of varieties, would seem to be, so to speak, an incidental phenomenon 
arising through the nuclear changes and the localization of the determinants, and 
it is open to question whether it has had any marked influence in the evolution 
of species. It is conceivable that evolution has taken place in spite of, rather 
than by the aid of, such Mendelian inheritance. The hereditary influence of the 
parent varies slightly through changes from unknown causes in the nature of the 
determinants, and the small resulting somatic variations in the offspring would 
appear to constitute the chief means by which evolution may slowly proceed 
and adaptation to the environment can be effected. 
* Morgan, T. H. aiid otliers, Tkc MechunisiiL uf ilciulcluiii Ilcrcdiiij. New York, 1915. 
