418 THE AMERICAN NATURALIST [Vol. LI 



of genes upon what were called unit-characters," is a 

 very indirect one. We now know that new characters 

 may certainly come into being through recombination of 

 genes. Eecombination may result in the origin of new 

 recessive characters, and this process Inay look very 

 much like loss-mutation. And crossing may result in the 

 origin of new dominant characters, color in chickens, in 

 rabbits, extra toes in chickens^, and this process will look 

 very much like positive mutation, the creation of a gene 

 out of nothing. If we except Qilnothera species, dividing 

 the organic world into animals, plants and CEnotheras, 

 for as long as.no solution is found for the baffling delayed 

 and abnormal segregation in CEnothera hybrids, we may 

 sum up as follows : 



Evolution is the result of a combination of all those 

 causes which heighten variability and which limit it. 



The only cause for inheritable variability in multicellu- 

 lar organisms which can be of any account in evolution 

 is mating between individuals of unequal genotype, cross- 

 ing in the widest sense (Amphimixis). 



All those causes which tend to reduce the potential 

 variability of a group of organisms tend to make vari- 

 eties or species of these groups. Such causes are iso- 

 lation, migration, adaptation, selection and espcciallij tlie 

 fact that, either periodically or regularly, the number of 

 individuals of one generation is very much smaller than 

 that of the preceding one. This cause of purification of 

 the typo, which we see in opcn-ation evei-ywhere (think 

 of tlic iiiHul)ci-s of lM)u<(>-fli('s ;i y,>ni- h^ fho last and 

 tlr-t lUTnlioii.). .ru'nlr^ M„ii,. , ru ;n'. 1 1 , of adapta- 

 tion litii.-^. To tlii^ cauM' ^^..l■ki,l^ upon variation 



can iii\<'nl no cartlily use, and for wliicli iioverl lirN^ss 



