CORRELATION 161 



phenomena an object of study, been applied to a variety of 

 relations which are not of the same nature. The credit of 

 distinguishing them seems to be due to Durken (1922). He 

 recognised three distinct types of association : 



(1) Relation. — The 'unilateral' dependence of a structure 

 for its full expression on an internal factor on which the 

 structure in question itself has no effect (e.g. the dependence 

 in development on the optic capsule of the embryonic lens 

 in the Vertebrata). 



(2) Correlation. — The reciprocal dependence of two asso- 

 ciated parts of such a nature that alteration of the one leads 

 to the alteration of the other (e.g. the reciprocal depend- 

 ence of the extremities and nervous system in vertebrate 

 development) . 



(1) and (2) include all causal associations. 



(3) Combination. — The ' static ' coincidence of variables 

 without any reciprocal or unilateral dependence (e.g. special- 

 isation of several parts for the same function ; dependence 

 of several structures or organs on sex hormones or on an 

 external stimulus (cf. Sumner, 1915) ). 



Graham Kerr (1926) distinguished primary or gametic 

 correlation from secondary or physiological correlation. This 

 is a fundamental distinction of considerable practical value, 

 and forms the basis of our discussion. Robson (1928) discussed 

 the various kinds of correlation in so far as they are contributory 

 to the process of group divergence, pointing out some of the 

 difficulties that are encountered in explaining the origin of 

 groups by the current theories of evolution. In particular 

 he dealt with Pearson's contention (1903, p. 2) that Natural 

 Selection is probably the chief factor in causing correlation. 

 The fact that correlation may be fluctuating or stable according 

 to the degree in which the variables are affected by environ- 

 mental factors, was pointed out by Love and Leighty 



(1914). 



Darwin's views on the importance of correlation in relation 

 to selection and the data which he assembled are discussed in 

 Chapter VII. It is true that in the course of his examination 

 of a large series of cases of correlation he touched on the causes 

 of the phenomena — e.g. he discussed the correlation of variation 

 in homologous parts (1905, vol. ii, p. 389) and the effects of 

 selection (I.e.). He did not, however, take his discussion 



M 



