284 Heredity. 



kind. If we suppose that any single variation occurs on 

 the average once in m times, the probability of that 



variation occurring in any individual will be ; and sup- 



HI 



pose that x variations must concur in order to make an 

 improvement, then the probability of the necessary vari- 

 ations all occurring together will be -. Now suppose, 

 what I think a moderate proposition, that the value of 



JL 1 



m* ~ 1000 1 



m is 1000, and the value of x is 10, then - = -7-7. 10 



30 = --ft. A number about ten thousand times as 



great as the number of waves of light that have fallen on 

 the earth since historical time began. And it is to be 

 further observed that no improvement will give its pos- 

 sessor a certainty of surviving and leaving offspring, but 

 only an extra chance, the value of which it is quite im- 

 possible to estimate." 



No one can be more firmly convinced of the great po- 

 tency of natural selection than I am, but I am sure every 

 one will feel that the problem of the origin of species 

 would be greatly simplified if it could be- shown that 

 variations are not fortuitous in Darwin's sense of the 

 word, but that natural selection is in some way provided 

 with variation in those parts where change is needed. 



Mivart has discussed this subject at considerable 

 length. He points out that the modification of domes- 

 ticated animals by the continued selection of slight va- 

 riations, is a very slow process, and after quoting Dar- 

 win's statement that wild species probably change much 

 more slowly than domesticated forms, he continues as 

 follows: "Let us take for an example the proboscis 

 monkey of Borneo. According to Mr. Darwin's own 

 opinion, this form might have been sensibly changed in 



