MEASUREMENTS OF RELATIONSHIPS 23 



case measures neither a uniform tendency nor a central tendency 

 of the series of individual relationships. 



The reader will obtain concrete information about the meaning 

 of the different measures of relationship and of their merits in actual 

 practise if he will calculate them for a score of representative rela- 

 tionships and examine them in the light of the entire correlation 

 tables. I have done this for the cosine irU and Median Ratio (or rather, 



X Age o( Musi/and 



FIG. 6. The dotted line is from averages ; the continuous line from medians. 

 The dash line is the regression as calculated from the Pearson Coefficient. 



in order to have the resulting figure comparable directly with the 



At y'QT* 'P 



cosine irU and the Pearson r, for the median of all the ratios : 



and x var ' y \ in the case of nine relationships representing organic 

 y var. x) 



and hereditary and conjugal relations, relations in animals and in 

 plants, relations of definite structural features and complex prop- 

 erties. The results are given in Table XI. They show that the 



