I^ECUNDAL SELECTION IN HUMAN RACES. 93 



of these results which I had previously reached by other methods. 

 The points in which our results most fully correspond are that fertility 

 tends to increase till checked by some other form of selection ; that 

 transformation of race may be produced by this principle ; and that, 

 in a species well adjusted to the environment, the typical form of the 

 species is not only the best adapted, but it is the most fertile. Con- 

 cerning "a tendency to increasing fertility in man," he says: "We 

 can not doubt that reproductive selection would steadily tend to alter 

 the mean fertility in man, unless it were somehow held stringently in 

 check. It is a point which seems to me of the utmost significance 

 that (as revealed by the statistics of 4,390 Anglo-Saxon families) 

 allowing for the proportion of the unmarried in the population, about 

 one-fifth to one-sixth only of the adults produce quite one-half of the 

 next generation, and any correlation between inheritable (physical 

 or social) characteristics and fertility must thus sensibly influence 

 that next generation. " (The Chances of Death, pp. 82, 83.) In regard 

 to the transforming influence of this principle he says : " I think there 

 is quantitative evidence that types of life may change without the 

 action of organic or inorganic environment, i. e., solely owing to some- 

 thing inherent in their constitution. One such factor of evolution, 

 genetic selection, I shall refer to later." (The Grammar of Science, 

 p. 376.) Of the relation of fertility to type he says: "For stable 

 races there is a strong tendency for the character of maximum fer- 

 tility to become one with the character which is the type." (The 

 Grammar of Science, p. 444.) 



The importance of applying statistical methods of investigation to 

 the problems of human evolution can not be overstated ; but, for the 

 full success of these methods, it is necessary that the nature of the 

 factors in their fundamental relations to each other should be clearly 

 apprehended and clearly stated in the definitions of the terms by 

 which the different influences are designated. This necessity seems 

 to require that we should have some knowledge of the probable factors 

 before we can even collect the statistics that will be of avail in giving 

 a quantitative measurement of the effects of any one factor. We 

 must have a clear conception both of the scope and of the limits of a 

 given factor or we may ascribe to it effects that are produced by other 

 factors, or ascribe to other factors effects produced by it. By way of 

 illustration, we may ask what is the scope and what are the limits of 

 the terms fertility and reproductive selection as used by Karl Pearson ? 

 When he says (The Chances of Death, p. 81), "Hence it would seem 

 that any characteristic or organ — such, for instance, as stature or size 

 of pelvis in the mother — correlated with fertility would be progres- 



