378 READINGS IN EVOLUTION, GENETICS, AND EUGENICS’ 
measured and curves of variability made for each. It was found that a 
different curve and mean resulted in each set. If the largest and the 
smallest individual in any pure line is isolated and allowed to produce 
a set of progeny, the mean and curve of variability will be the same, 
because both the large and the small individual belong to the same 
genotype, though varying phenotypically. 
In conclusion therefore we may say that, according to Johanssen, 
organisms that appear to be alike, or are alike somatically, are identi- 
cal phenotypically; but organisms, whether alike somatically or not, 
that have the same determiners are genotypically identical or belong to 
the same genotype. 
ARE DETERMINERS (GENES) CONSTANT OR VARIABLE ? 
In our study of the causes of mutations we were forced to admit 
that we are almost wholly ignorant of the causes of mutations. We 
infer that in the majority of cases the change occurs within the germ 
cell and in the gene itself. In this pure-line work where the genes are 
unmixed by intercrossing we should have a splendid opportunity of 
testing the possibility of genes varying or becoming modified. In 
none of these experiments in pure lines was there any indication of 
genes being modified, but some further work by Jennings seems to 
imply that he has changed his position with reference to the modifi- 
ability of genes. Using another protozoan, Difflugia, he found that 
he did succeed in markedly shifting the mean by selection, and thus 
seemed to prove that genes were modifiable. This work is open to two 
comments. First, protozoa are not suitable materia] for testing the 
distinction between germinal and somatic changes, because the whole 
individual is but a single cell; therefore any change that is passed on 
may be merely a somatic change. Second, Jennings, in making his 
selections, did so on the basis not so much of an individual itself as 
upon the characters of its ancestors for some generations back. He 
was, therefore, working more with genotypic than with phenotypic 
considerations. If, as he claims, there was a progressive modification 
of characters, such as numbers of spines, beyond the limit of vari- 
ability in the original stock, the results would seem to warrant the 
conclusion that genes are variable and that selection might be effective 
in establishing new types within pure lines. 
Castle, as the result of a long and elaborate experiment with 
“hooded rats,” at first thought that the gene for the hooded pattern 
was variable and could be enhanced by selection; later, however, he 
