1 6 THE VARIATION OF ANIMALS IN NATURE 



evolution should be tested by the results of taxonomy (dealing 

 with both living and fossil forms) and population-analysis. 

 These two studies, more than any others, bring the theories 

 of evolution into contact with the gross facts of nature. We 

 realise their specific limitations and in particular the need to 

 supplement them by observations on habits and behaviour, 

 but we feel that they constitute an acid test of evolutionary 

 theories based on other studies. This test has been insufficiently 

 applied in the past. It is well worth while to try to describe 

 the facts of nature as they actually are and to see what are the 

 simplest deductions suggested. There has been a tendency 

 to ignore or distort certain observations because they fail to 

 fit in with the theories, e.g. some of them seem to suggest a 

 neo-Lamarckian explanation of evolution, but this idea has 

 been nearly always ruled out on a priori grounds. The occur- 

 rence of non-adaptive specific characters, and certain palaeon- 

 tological and other evidence suggest that variants can spread 

 without any adaptive qualifications. But recently mathe- 

 matical theories have been invoked to prove that this is im- 

 possible. We believe it is advisable to make new contacts 

 between theories so obviously developed by deductive methods 

 and the large body of recorded observations from which they 

 have been so long divorced. 



It appears that on the whole modern writers on evolution 

 fall into three classes. The first are impressed by the obvious 

 facts of adaptation. They take variations for granted and 

 tend to describe the assumed effects of selection. The second 

 argue from a relatively few animals which have been studied 

 under laboratory conditions. They tend to assume that, 

 when once a mutation has occurred, it can look after itself 

 and that, as long as it is not harmful, it can spread through 

 a population. The third class, recognising that the spread 

 of variants needs explanation, have given exact mathematical 

 expressions for the efficiency in this respect of Natural Selection 

 without, however, first showing that that process is actually 

 operative in nature. 



In our attempt to evaluate the evidence put forward on 

 behalf of the various theories of evolution we discuss the 

 logical conditions for an exact proof of certain theories and 

 in particular (p. 186) Woodger's account of the stages by 

 which a theory attains the status of an accepted truth. It is 



