134 RETROGRESSION 



are, apparently, frequently confused, and thoughts are claimed 

 as accurate merely because they are based on, not tested by, 

 statistical data. 



223. In the Law of Ancestral Inheritance we have an instance 

 of an induction expanded, but not tested by a deduction, the latter 

 being regarded as the theory. Here the induction is that, on the 

 average, offspring resemble ancestors (at any rate in certain characters 

 and within three or four generations) in the stated degrees. The 

 deduction the really tremendous expansion of the induction is 

 that ancestors contribute in the stated degrees to the heritages of 

 descendants. Though the universe is a unity, the facts and ' laws ' 

 of which are in harmony with one another, and it is the mission of 

 science not only to interpret the facts in terms of the laws but also 

 to demonstrate the relations between the latter, no appeal is made 

 to any conceived system of reality. The axiom that every induc- 

 tion must be used as a basis for a rigorous deductive inference of 

 consequences is disregarded. In other words the correctness of the 

 thinking, as distinguished from the facts, is left untested. 



224. The law evidently bears on the theory of the Continuity 

 of the Germ-plasm, which supposes that the parents and ancestors 

 of the individual contribute nothing to the child, and on the theory 

 of Recapitulation, which supposes that ancestors are represented 

 during development, not en masse, but serially ; but no attempt is 

 made to show that it accords with these hypotheses, or, if it does 

 not, in what respects they are wrong. It remains as apart from 

 anything else that is known or that has been surmised by science 

 concerning living beings as a miracle. 



225. The Law of Filial Regression furnishes another instance of 

 biometric work. It has been ascertained that when a group of 

 parents differs in any particular from the mean or average (medio- 

 crity) of their race, the offspring on the average differ less, grand- 

 children still less, and so on. In other words, if a group (or 

 individual) in any generation differs from the racial mean, its (or 

 his) predecessors and successors will, as a rule, be found less 

 different. "Thus take fathers of stature 72", the mean height 

 of their sons is /O'S", or we have a regression towards the mean 

 of the general population. . . . The father with a great excess of 

 the character contributes sons with an excess, but a less excess of 

 it ; the father with a great defect of the character contributes sons 

 with a defect, but less defect of it. The general result is a sensible 

 stability of type and variation from generation to generation. . . . 

 Now a man is not only the product of his father, but of all his 



