258 ESSENTIALS OF BIOLOGY 



Therefore, the natural-selection hypothesis is logically sound 

 and it will be actually sound if the condition — that a certain 

 proportion of the individuals of the race we experiment upon 

 display inheritable variations — is realized in wild nature. We 

 must now appeal to the facts of observation. What is *' a certain 

 proportion " ? This means the analysis of the nature of the 

 variations — whether Mendelian, mutational, fluctuating variations, 

 or acquirements. Is the proportion of inheritable variations 

 great enough to enable such a process of transformism as we have 

 suggested above to go on at such a rate as to explain the evolu- 

 tionary career .? Further (as Bergson urged in an argument that 

 still retains its force), the transformist process, or the selection 

 hypothesis, will usually include combinations of favourable 

 variations. What are the probabilities of such combinations 

 occurring at random. In most cases these probabilities will be 

 small ones, so that the rate at which transformism proceeds will 

 be further lessened. Clearly the natural-selection hypothesis is 

 logically strong ; clearly it involves only randomness in respect 

 of the occurrence and combining of variations, and is so far 

 satisfactory (to many minds) in that it dispenses with " purpose " 

 in wild nature. But it is still necessary that the hypothesis should 

 be verified by statistical observations made upon naturally occur- 

 ring populations and such observations — upon an adequate scale 

 — do not exist. 



Two problems more press for treatment, (i) The origin of 

 mutations (which we have already dealt with and (2) the meaning 

 of the '' selection " idea. What is it that " selects " in natural 

 conditions. Plainly it is the urge to live and reproduce. It is 

 rather " auto-selection " that we mean instead of selection " by 

 nature." If the organism is fit it will live and reproduce and 

 so its " breed " will persist, or accumulate. If the organism is 

 not fit it will be exposed to greater risks. The idea of " auto- 

 elimination " is perhaps more clear than that of auto-selection. 

 In Mendelian variants we see, most plainly, the organisms exposed 

 to risks. The combinations of characters studied in Mendelism 

 are mostly such as are disabilities (thus the more obvious ones 

 in man, haemophilia, night-blindness, colour-blindness, the 

 mutilations of extra digits, stumpy-fingers, etc.). If such com- 

 binations of characters occur at random, in the cases of natural 

 populations they must afford the " materials for elimination." 



