438 THE GRAMMAR OF SCIENCE 



types of mating may exist and serve as a means of 

 differentiation we have made (p. 422) the assumption 

 that all members of a local race are in se and inter se 

 equally fertile. But is such an assumption correct ? Is 

 it a priori in the least likely to be correct ? Why should 

 not fertility be a function of the size of special organs or 

 the intensity of certain characters ? Is it not highly 

 probable that it must be so ? Shortly shall we not find 

 fertility correlated with the other characters of an organ- 

 ism ? There is no difficulty about answering this ques- 

 tion. We have only to form a correlation table such as 

 we have now so frequently illustrated, in which one 

 character will be fertility, and a second any physical 

 character which appears likely to influence fertility. 

 Thus we might take a number of diverse measurements 

 on male and female moths, pair them and note the 

 number of fertile eggs the female lays ; or we might 

 consider the characters of a plant and measure the 

 amount of seed obtained from its seed vessels. Or, again, 

 we might take a number of physical measurements on 

 man and woman, stature, chest girth, pelvic measure- 

 ments, and test whether in the case of marriages lasting 

 during the fecund period ^ of their lives there is any rela- 

 tion between the size of family and these characters. So 

 far as I know, no material ad hoc has so far been col- 

 lected, and yet the question of whether fertility is cor- 

 related with other characters is of immense importance 

 for the theory of evolution. 



Let us see exactly what will happen if fertility be 

 correlated with any other character. Let us to simplify 

 matters suppose autogamic or pangamic mating so that 

 such a frequency polygon as that on p. 422 represents 

 the distribution of individuals, or mid-parents (see p. 470) 



1 We must take the fecund period if we wish to disentangle " causes." A 

 death-rate, non-selective with regard to the characters in question, may obscure 

 otherwise the intensity of the correlation. Like ^ the physicist, the biologist 

 must first deal with the simplest possible fields, i.e. isolate his sources of 

 change, and then having studied these individually, proceed to a synthesis in 

 which the resultant effect no longer exhibits any individual source of change 

 in the full intensity of isolated action. 



