1 66 THE VARIATION OF ANIMALS IN NATURE 



we examine a series of geographic races. Throughout con- 

 siderable tracts a positive correlation may hold : in other 

 territories the correlation may be entirely dissolved. Intra- 

 racial correlations in pigmental characters, on the other hand, 

 are even more pronounced than are interracial ones. Darker 

 races, like darker individuals, tend to have more extended 

 coloured areas in their pelages, deeper pigmentation in the 

 skin of their feet, broader (and longer) tail stripes, etc' 



Other papers of Sumner's (e.g. 1918, 1920, 1923) make it 

 quite evident that the character complexes which distinguish 

 subspecies are by no means highly correlated, and certainly 

 his evidence concerning the behaviour of these complexes on 

 crossing shows (1923) that they fail to behave as units. The 

 systematic analysis of species and geographical races has 

 yielded similar results, and there is a good deal of evidence 

 that the characters distinguishing such groups vary inde- 

 pendently (cf. Swarth, in Linsdale, 1928, p. 257 ; Mertens, 



193^ P- 205). 



The discussion as to the kinds of correlation (p. 161) shows 

 that they may be reduced to two fundamental types : (1) one 

 in which the characters stand in relation to each other as 

 cause and effect, and (2) one in which their association is 

 coincidental (' combination '). 



(1) This includes (a) the dependence of one part on another, 

 and (b) the reciprocal dependence of two parts on each other. 



(a) A structure may depend, as we have seen, on another 

 structure on which it has no effect itself. Certain of the 

 phenomena of development have been interpreted as due to 

 various kinds of stimuli (chemotaxis, thigmotaxis) exerted 

 by one part on another. The classical example is the failure 

 of the lens of the vertebrate eye to develop if the optic 

 capsule fails to make contact with it. Other examples are 

 discussed by Jenkinson (1909, p. 273 and foil.). 



The dependence noted here affects the main architecture 

 of the parts rather than the characters which distinguish 

 species. But certain characters of proportion are obviously 

 influenced by growth principles, and Huxley (1932, passim) 

 in particular has applied the principle of heterogonic growth 

 to explain certain differences between species. (See especially 

 the case of the Lucanid beetle, Cyclommatus tarandus.) It may 

 therefore come about that correlated specific differences 



