BIOMETRY 133 



variable characters in very variable species. He himself terms the 

 chosen characters ' moderately exceptional.' l Bassett hounds 

 were established very recently, and display the variability of new 

 types. In many characters human beings display the variability 

 which follows cessation of selection. The reproduction of coat- 

 colour in the lower animals, and of eye-colour in men, tends to be 

 1 alternative.' 2 Probably human stature has never been very rigidly 

 fixed by Natural Selection, and variability has been influenced, in 

 England at any rate, by a great intermixture of races. Artistic 

 faculty and health are not only very variable, but develop under 

 the influence of an environment which, as regards these characters, 

 is very varied. 3 



221. Not until ancient, stable, non-exceptional characters, for 

 example heart, lungs, and the like, have been investigated shall we 

 be in a position to formulate or perceive the impossibility of 

 formulating a numerical Law of Ancestral Heredity (i.e. to indicate 

 the average rate of reversion). When that is done, I think it will 

 be admitted that variability, including the tendency to reversion, 

 is an adaptation regulated by Natural Selection, and, therefore, 

 that the degree of variability differs with the stringency of Natural 

 Selection in every species and character. I think, also, that it will 

 then be admitted that the true Law of Ancestral Heredity is 

 formulated when we declare that, In any character the tendency 

 to retrogression (i.e. reversion to ancestors} is proportionate^ on the 

 average^ to the speed of the antecedent progression. 



222. The Law of Ancestral Inheritance is an example of the 

 biometric and statistical method of inquiry. As a method biometry 

 is, of course, like experiment, excellent. It enables the patient 

 worker to gather many obscured truths which, but for it, would 

 lie beyond our reach. But it is extraordinarily slow and laborious, 

 and, like experiment, has its limitations as a means of discovery 

 and as a test of thinking. So many biological facts are patent, and 

 so much biological thinking can be tested by reference to this patent 

 evidence that, apparently, it has been difficult to find subjects 

 suitable for biometric inquiry, except when the patent evidence 

 is ignored. Occasionally, therefore, such inquiry has been directed 

 to the discovery of matters already known, or which can be made 

 known at a thousandth part of the labour involved and with an 

 even greater degree of certainty. Inferences founded on materials 

 furnished by biometry require to be tested as carefully as any other ; 

 but, as in experiment, the two functions of discovery and testing 



* Natural Inheritance, p. i. * See 239, 277, 278. 3 See chapter xxii. 



