No. 642] VARIATION IN INDIVIDUAL GENE 



43 



cases stand alone. Aside from them, there are only two 

 instances in which two (or more) new mutant genes have 

 been proved to have been present in the same gamete. 

 Both of these are cases in Drosop]iiIa—iXY)OYi(H\ by 

 Mnller and Altenburg (1921) -in whicli a i;;nnctc con- 

 tained two new sex-linked lethals; two cnscv aiv not a 

 greater number than was to have been I'xpi'ct^'d from a 

 random distribution of mutations, judging l)y tlie fre- 

 quency with which single mutant lethals were found in the 

 same experiments. Ordinarily, then, the event that 

 causes the mutation is specific, affecting just one par- 

 ticular kind of gene of all the thousands present in the 

 cell. That this specificity is due to a spatial limitation 

 rather than a chemical one is shown by the fact that when 

 the single gene changes the other one, of identical com- 

 position, located near by in the homologous chromosome 

 of the same cell, remains unaffected. This has been 

 proved by Emerson in com, by Blakeslee in Portulaca, 

 and I have shown there is strong evidence for it in Dro- 

 sophila. Hence these mutations are not caused by some 

 general pervasive influence, but are due to accidents " 

 occurring on a molecular scale. AVhen the molecular or 

 atomic motions chance to take a ])articu];ir ^ovm. to wliich 

 the gene is vulnerable, then the iiuitatioii oc('iir>. 



It will even be possible to (Ictcniiiiio wliotlifr tlie entire 

 gene changes at once, or wliotlicr the uvnc con-i-t- of 

 several molecules or ])articlf-. one ol" wliidi may rlmiige 

 at a time. This point can he settled in ()r,<;'ani-in> lia\ ing 

 determinate cleavage, by studies of the distrilnition of the 

 mutant character in somatically mosaic mutants. If there 

 is a group of particles in the gene, then when one par- 

 ticle changes it will be distributed irregularly among the 

 descendant cells, owing to the random orientation of the 

 two halves of the chromosome on the mitotic spindles of 

 succeeding divisions,^ but if there is onlv one jtarticle t(» 



