XXIII HUMAN SELECTION 511 



the two types required. In this way it is perfectly certain 

 that in a comparatively short period — thirty or forty 

 years perhaps — he would be able to produce two very 

 distinct forms, the one a very fair racshorse, the other an 

 equally good specimen of a cart-horse ; and he could do 

 this without subjecting the two strains to any difference 

 of food or training, since it is by selection alone that our 

 various breeds of domestic animals have in most cases 

 been produced. 



On the other hand, the person who undertook to 

 produce similar results by food and training alone, without 

 allowing selection to have any part in the process, would 

 have to act in a very different manner. He should first 

 divide his horses into two lots as nearly as possible iden- 

 tical in all points, and thereafter subject the one lot to 

 daily exercise in drawing loads at a slow pace, the other 

 lot to equally constant exercise in running, and he might 

 also supply them with different kinds of food if he thought 

 it calculated to aid in producing the required effect. In 

 each successive generation he must make no selection of 

 the swiftest or the strongest, but must either keep the 

 whole progeny of each lot, or carefully choose an average 

 sample of each to be again subjected to the same discipline. 

 It is quite certain that the very different kinds of exercise 

 would have some effect on the individuals so trained, enlarg- 

 ing and strengthening a different set of muscles in each, and 

 if this effect were transmitted to the offspring then there 

 ought to be in this case also a steady advance towards the 

 racer and the cart-horse type. Such an experiment, how- 

 ever, has never been tried, and we cannot therefore say posi- 

 tively what would be the result ; but those who acce2)t the 

 theory of the non-heredity of acquired characters would 

 predict with confidence that after thirty or forty genera- 

 tions of training without selection, the last two lots of colts 

 would have made little or no advance towards the two 

 types required, but would be practically indistinguishable. 

 It is exceedingly difficult to find any actual cases to 

 illustrate this point, since either natural or artificial 

 selection has almost always been present. The apparent 

 effects of disuse in causing the diminution of certain 



