Xo. 509] 



The high values of these correlation coefficients for the rat 

 as compared with man suggest an interesting question : Are 

 we to conclude on the basis of these results (and similar ones 

 obtained by Kellicott from his study of the toad) that there is 

 a general tendency for the various parts of the body to be more 

 closely correlated in lower organisms than in man? In the 

 writer's opinion such a conclusion is at present very doubtful 

 for two reasons. In the first place the human data on which 

 correlation studies have been made are meager and, from their 

 method of collection, not altogether trustworthy. In tin; second 

 place the coefficients of correlation published by Donaldson (the 

 same considerations hold with reference to Kellicott 's toad 

 data, though not to so great a degree relatively) probably have 

 spuriously high values. This arises from the fact that they 

 are deduced from material which is very heterogeneous in 

 respect to age. The biometric constants for all characters which 

 change with age by growth will have their values affected in 

 such material. It is a well-known fact, of which the mathe- 

 matical proof was first given by Pearson, that heterogeneity of 

 material operates to increase apparent correlation. To such 

 an extent may this occur that several sets of data, each of which 

 taken alone shows no correlation whatever between two charac- 



apparent correlation of the age (— growth J 

 the material. It seems desirable if brain-i 

 correlations are to have full significance for 

 poses, that they be based either on adult ma 

 growth changes have reached a minimum, or s 

 which is homogeneous in respect to some 

 period of the life cycle. 



A further point of less practical signifies 

 parently overlooked by Donaldson in his dis 

 tion results is that in at least all of his cases ii 

 sion lines are plotted in the paper the corre 

 skew. In such cases, of course, the correlal 

 not be taken as the true measure of the 

 Instead, resort must he had to the <-orrelatioii 



