1 68 THE VARIATION OF ANIMALS IN NATURE 



consisting of proportional differences of parts might arise, if 

 the characters in question were related to the absolute size of 

 the animal and if the latter were of selective value. 



Another kind of correlation of this type is seen in the 

 dependence of a structure on the specific activity of a gland — 

 particularly those of internal secretion. 



(b) As regards the reciprocal dependence of the parts of 

 the living organism little can be said. There is some evidence 

 that in the course of development various parts are dependent 

 for their expression on each other. This fact was indeed made 

 a prominent feature of Driesch's theory of development. 

 According to Jenkinson {I.e. pp. 75—7), the dependence 

 diminishes with age ; correlation is only high during periods 

 of rapid growth, and there is an increasing power of self- 

 differentiation. That certain relations of this kind persist 

 into later life is seen in the dependence of the extremities on 

 the nervous system in the Vertebrata. 



Correlation has often been invoked to supplement the 

 theory of Natural Selection. The modification of apparently 

 non-serviceable structures has thus been attributed to their 

 being correlated with characters influenced by selection. 

 The nature of the correlation has not seriously been studied. 

 It was probably some kind of causal association such as we 

 have been discussing that was in (e.g.) Darwin's mind when 

 he stressed its evolutionary importance. Not only, however, 

 does this kind of correlation require much more study and 

 exploration, but also the efficacy of selection itself (Chapter VII) 

 is open to question. Possibly some differences of size and 

 proportion between species have been produced by selection 

 acting on characters correlatively associated in this way. 

 Whether differences of colour, ornamentation and the arrange- 

 ment of parts are influenced by it is far more problematical. 



(2) We have already seen (p. 162) that we have to deal 

 here with two types of correlation, viz. (a) one due to the 

 coincident effects of various external causes, and (b) another 

 due to the mechanism of heredity. 



(a) There is a variety of ways in which such correlations 

 can arise. Thus Hubbs (1926) shows that low temperatures 

 tend to make fish large, small-headed and small-eyed. High 

 temperatures make them small, large-headed and large-eyed. 

 Schmidt (1930, p. 28) finds that in the Atlantic Cod (as in 



