BASIS OF THE PEOBLEM 77 



either that organism does not possess some character which other 

 organisms possess and which have enabled them to survive, or 

 it possesses some character which other organisms do not possess, 

 and which at some crucial moment has told against it. This 

 subject could be considered at great length, but it must suffice to 

 say here that the more experienced naturalists are, and the 

 greater the knowledge they have of the conditions of organic life, 

 the more it is borne in upon them that the death-rate is upon 

 the whole selective, and that the best adapted types have a better 

 chance of survival than other types. 



From the nature of the case it must be difficult to obtain 

 statistical evidence of natural selection. There are, however, 

 certain cases known which are of great interest. Bumpus, for 

 instance, after a storm in America collected 136 sparrows, all of 

 which had been injured. Of these 72 recovered, while the re- 

 mainder died. He weighed and measured all the specimens and 

 compared the figures for the survivors with the figures for those 

 which had perished. It was found that the average type of the 

 latter was larger and heavier than the average type of the former. 

 It was also observed that there was a less wide range of variability 

 among the survivors than among the dead, showing that the 

 favoured type approximated more closely than the others to the 

 average type of the species. This favouring of the average type 

 has been shown to occur by other observations, and it may be 

 supposed that normally it is the average type of any species 

 which is best fitted to the particular niche in nature occupied 

 by the species, and that therefore adaptation will be measured 

 by the nearness of approach to this type. When, on the other 

 hand, circumstances are changing, some deviation from the 

 average type will be favoured. 



9. This leads us to ask in more detail what it is that happens in 

 the process of selection. Selection has been likened in its action 

 to a sieve separating the fit from the unfit. But fitness is measured 

 by the characters exhibited, and these characters may be either 

 of the nature of mutations or of modifications. What is exhibited 

 is selected quite apart from the underlying nature of the character ; 

 but it is only when variations of the nature of mutations are 

 selected that there results any change in the composition of the 

 germinal constitution of the species. So far as selection is merely 

 that of modifications within a strain, there will be no change in 



