176 THE VARIATION OF ANIMALS IN NATURE 



It is instructive to compare the correlations between specific 

 with those obtaining between generic and family characters. 

 Some diagnostic characters are ' good ' and hold for every 

 member of the genus. Others are variable or only present in 

 some members. The permutations of characters amongst 

 related genera or families are also common. In fact, it would 

 seem at first sight that at all stages in divergence the correla- 

 tion between characters was of the same nature and depended 

 on the extent to which the different unit characters had suc- 

 ceeded in permeating populations of different sizes. Highly 

 correlated characters seem to be those for which large numbers 

 of individuals are homozygous. The position of a character 

 in the hierarchy would seem to depend on the extent to 

 which it had spread, and this, in turn, approximately on the 

 time that has elapsed since it first appeared. 



The study of lineages by palaeontologists appears to bear 

 out such a view of evolution. The material studied (Bryozoa, 

 Mollusca, Brachiopods, Echinoids, Mammals) is restricted by 

 certain preliminary requirements. The organism must possess 

 sufficient characters (in the fossil state) to admit of establishing 

 correlations between groups of characters. Indeed, it may be 

 suspected in some phyla (e.g. certain Mollusca) that the number 

 of characters involved is actually too small for the results to be 

 very significant. Secondly, abundant material must be avail- 

 able of approximately the same age. Lastly, the forms studied 

 must occur in an uninterrupted succession of strata, so that 

 the fate of the character combinations may be revealed. 

 We do not wish to deal fully with the palaeontologists' 

 data in the present chapter, but only to note certain 

 general conclusions, of which the most important are the 

 following. 



Each character evolves as a separate unit. In different 

 lineages the same character may evolve at very different rates 

 (cf. Trueman, 1930), so that in one case it is associated with 

 one set of characters and in another with quite a different level 

 of divergence. Correlations between groups of characters 

 are often only maintained at one horizon. As we traverse the 

 strata the associated characters alter. These conclusions, derived 

 from the study of actual fossils, are exactly what one would 

 have expected from a study of living species. In the latter, the 

 variation in correlation and the permutations of characters 



