210 CERTAIN LAWS OF VARIATION. 



around their mean. In his Presidential Address, 

 already referred to, Mr. Sedgwick has, however, come 

 to a somewhat novel conclusion as to the change of 

 variability with evolution. He points out that through 

 the action of Natural Selection certain variations 

 will be gradually eliminated, and the organisms 

 will become more and more closely adapted to their 

 surroundings. The variability of the species will 

 therefore diminish. Thus breeders have found that 

 " continued selection tends to produce a greater and 

 greater purity of stock, so that if selective breeding is 

 carried too far, variation almost entirely ceases." It 

 follows, therefore, that " variation must have been 

 much greater in past times than it is now. In fact, it 

 must have been progressively greater the farther we go 

 back from the present time." 



This specious and apparently straightforward argu- 

 ment cannot be accepted, however, as it is built up on 

 false premises. It is by no means true that by selective 

 breeding we can reduce variability almost to nil. Pro- 

 fessor Pearson* has calculated that, in the human race, 

 if the parents be selected, then the variability of the 

 offspring will be, on an average, only 9.5 per cent, less 

 than that of the whole race; whilst if the grandparents 

 and any number of more remote ancestors be selected 

 as well, the variability will never become more than 

 10.56 per cent, smaller than that of the race. In fact, 

 the variability of offspring, as compared with that of 

 parents, depends directly upon the correlation between 

 them. If the coefficient of correlation be r, then the 

 variability of the offspring of selected parents will be 



* " Grammar of Science," p. 485. 



