External and Internal Factors in Evolution 307 



a moss-rose on a Provence rose is a return to a former 

 state, . . . nor can the appearance of nectarines on peach 

 trees be accounted for on the principle of reversion." It is 

 said that bud-variations are also much more frequent on 

 cultivated than on wild plants. 



Darwin adds : " These general considerations alone render 

 it probable that variability of every kind is directly or indi- 

 rectly caused by changed conditions of life. Or to put the 

 case under another point of view, if it were possible to expose 

 all the individuals of a species during many generations to 

 absolutely uniform conditions of life, there would be no 

 variability." 



In some cases it has been observed that, in passing from 

 one part of a continent to another, many or all of the forms 

 of the same group and even of different groups change in 

 the same way. Allen's account of the variations in North 

 American birds and mammals furnishes a number of strik- 

 ing examples of this kind of change. He finds that, as 

 a rule, the birds and mammals of North America increase 

 in size from the south northward. This is true, not only 

 for the individuals of the same species, but generally the 

 largest species of each genus are in the north. There 

 are some exceptions, however, in which the increase in size 

 is in the opposite direction. The explanation of this is 

 that the largest individuals are almost invariably found in 

 the region where the group to which the species belongs 

 receives its greatest numerical development. This Allen 

 interprets as the hypothetical " centre of distribution of the 

 species," which is in most cases doubtless also its original 

 centre of dispersal. If the species has arisen in the north, 

 then the northern forms are the largest; but if it arose in 

 the south, the reverse is the case. Thus, most of the species 

 of North America that live north of Mexico are supposed 

 to have had a northern origin, as shown by the circumpolar 

 distribution of some of them and by the relationship of 



