Modification and Variation. 311 



interesting, but one well-nigh impossible to test by obser- 

 vation. If accepted as a factor, it would serve to account 

 for the inordinate growth of certain structures, such as 

 the exuberance of some secondary sexual characters, and 

 for the existence of determinate variations, that is to say, 

 variations along special or particular lines of adaptation. 



Such determinate variations are, however, explicable on 

 the theory of natural selection. Writing in 1892, I put 

 the matter thus.* " Take the case of an organism which 

 has in some way reached harmony with its environment. 

 Slight variations occur in many directions, but these are 

 bred out by intercrossing. It is as if a hundred pendulums 

 were swinging just a little in many directions, but were at 

 once damped down. Now, place such an organism in 

 changed conditions. The swing of one or two of the pen- 

 dulums is found advantageous ; the organisms in which 

 these two pendulums are swinging are selected : they mate 

 together, and in their offspring while these pendulums 

 are by congenital inheritance kept a-swinging, the other 

 ninety-eight pendulums are rapidly damped down as 

 before. 



" Let us suppose, then, that the variation in tooth 

 structure, in a certain mechanically advantageous direc- 

 tion, be such a selected pendulum swing. That particular 

 pendulum, swinging in that particular direction, will be 

 the subject of selection. The other pendulums will still be 

 damped down as before, and in that particular pendulum 

 variations from the particular direction will be similarly 

 damped down. It will wobble a little, but its wobbling will 

 be as nothing compared with the swing that is fostered by 

 selection. In this case, then, selection will choose between 

 the little more complexity that is advantageous, and, the 

 little less complexity that is disadvantageous. The little 



* Natural Science, vol. i. April, 1892, pp. 100, 101. 



