BIOLOGY AND STATISTICS 75 



is wanting, that new species have arisen exclusively 

 through the accumulation by natural selection of 

 variations of a strictly indefinite, fluctuating, or normal 

 kind. We have already seen reasons for believing that 

 this is very far from being the case, and future chapters 

 will be found to add considerably to the force and 

 quantity of the evidence already adduced. 



Normal variations, strictly speaking, are individual 

 differences which can be supposed to depend upon a 

 large number of small factors or causes factors so 

 numerous and so minute that the numerical distribu- 

 tion of the individuals examined, when ranged in order 

 according to the feature chosen for examination, is 

 found to conform closely to that which would be 

 expected on the mathematical theory of chance. 

 Such a distribution will only result when the differences 

 considered can be strictly regarded as lying upon a 

 linear scale, and when they are also evenly distributed 

 along that scale. That is to say, the biometrician 

 deals with continuous variations of a quantitative kind. 

 It is to be hoped that these somewhat obscure sayings 

 will be more easily understood in the light of what 

 follows. 



The facts of variation have not been found readily 

 amenable to precise definition, but we shall endeavour 

 to make plain by the aid of a few examples what kinds 

 of variations do and what kinds do not appear to be 

 legitimate objects for the application of biometrical 

 methods. Thus it may be thought that the biome- 

 trician is outrunning his license when he ranks the 

 colours shown by the iris of the human eye in a con- 



