FLUCTUATING VARIATIONS 81 



definitely in regard to small quantitative variations. A tall 

 father does not necessarily have tall children. Where the 

 characters in which the two parents differ are such 35 

 readily blend, regression towards the mean of the stock will 

 occur. 



(2) Even with very thorough isolation segregation of like 

 individuals and very consistent selection, it is doubtful whether 

 a new race could be evolved from the cumulative increase of 

 small quantitative variations, e.g. in stature or colour of hair. 

 It is doubtful whether any domestic races have so arisen. It is 

 not in this way that dwarf-races and giant-races have been 

 formed. They arise from sudden discontinuous variations or 

 mutations, which are often peculiarly heritable, which are any- 

 thing but liable to be swamped by inter-crossing, and which 

 sometimes exhibit Mendelian inheritance. 



(3) The result of the gradual accumulation of small 

 quantitative variations may be very important in a long time, 

 just as a small sum may become large from interest accumulated 

 for centuries ; but it is difficult .to believe that minute fluctuations 

 in quantity would always have sufficient selective value to ensure 

 their persistence. 



There are several reasons why selectionists have restricted 

 themselves so much to continuous variations as the raw material 

 of evolution, (i) Until lately we have known comparatively 

 little in regard to discontinuous variations or mutations. (2) It 

 was hastily concluded that these changes were not likely to 

 be transmitted a generalisation in part due to preoccupation 

 with teratological non-viable freaks. (3) In many cases related 

 species can be arranged in a gradual series with intermediate 

 forms linking the extremes. 



Now, there is no need to hamper the Evolution Theory by 

 restricting selection to minute variations. We know that sports, 

 mutations, or discontinuous variations are frequent, and that 

 they are remarkably stable in their hereditary transmission. 



6 



