522 THE AMERICAN NATURALIST [Vol. LI 



eral genotypes present. In other words, it was because 

 most material is itself not uniform that an exaggerated 

 idea arose concerning the nature of the variability of the 

 character. 



In the second place, Johannsen's experiments with 

 Princess beans have shown that when the material is 

 homogeneous in successive generations the \'^riability 

 of the character is due' to the environment and is not due 

 to changes in the genotype. 



In the third place, any pure stock (and especially one 

 that has been made homozygous by inbreeding), so long 

 as it does not vary, is an argument for the stability of the 

 factorial basis. When changes occur in it as they are 

 pretty certain to do, the fact does not in itself prove 

 that the gene under observation has changed, for other 

 genes that affect the character may have mutated. Jen- 

 nings has recently said^ that we maintain the constancy 

 of a given gene by assuming that other genes, rather than 

 the original gene itself, have changed. This would be of 

 course on our part a straight evasion of the issue. The 

 criticism would li()](l if the question involved were a purely 

 philoxipliic.-il one. as .Icniiinus iniuiit unintentionally lead 

 the n-ath'T to ln'licvc. l-'ortiiiiatcl y it is becoming more 

 and more jiossible to denionsti'ate that changes of this 

 latter kind do take place; for it is possible with suitable 

 material to show in such cases the exact nature of the 

 change. Wherever it has been possible to do this it has 

 been found that a definite mutation in some gene has 

 taken place, or has been introduced into the culture 

 through crossing. 



In the fourth place it has been found that more than 

 one inntant u^mic may be the mate (allelomorph) of the 

 -am.' iiui ma! uviir. Since no more than two of them may 

 ('xi>t at the -ami' time in a given individual, and since 

 liiika.izc ('\|)('i'iifi»'iits liave sliown in T) rosnpji ihi that these 

 i.mltiph' allrloiMorpli- have the >;\uw llnka-.- relations to 

 all otli.T uviio ( /■. r.. we i 1 it (- i pivt Ww rc-ult. each such 



