CORRELATED VARIATIONS. 83 



tionary. It is therefore impossible to say that civilised 

 woman is nearer to the primitive type than civilised 

 man, for while civilised man differs more from the 

 primitive type than civilised woman, so far as absolute 

 size is concerned, he has made only about half her 

 progress in variation, and hardly any progress in corre- 

 lation. The absolute amount of correlation is very 

 high, as the following figures show: 



HALES. FEMALES. 



ORGANS. AINO. FRENCH. AINO. FRENCH. 



Femur and tibia, .83 .81 .85 .89 



humerus, .86 .84 .87 .87 



radius, .79 .74 .70 .78 



Clavicle and humerus, .44 .63 



Humerus and radius, .78 .85 .74 .85 



Tibia and fibula, .89 .96 .97 .98 



Humerus and ulna, .77 .77 .75 .86 



Radius and ulna, .98 .88 .98 .92 



Here we see that the Galtonian constant, r, was in most 

 instances above .8. Between the tibia and fibula it 

 averaged .95, and between radius and ulna .94, so that 

 in these cases the correlation was almost absolute. 



In the Naquada race investigated by Warren * the 

 correlation between the lengths of the long bones in 

 males (as measured in about 60 skeletons) was distinctly 

 higher than for Aino males, but in females (measured 

 in about 90 skeletons) it was either the same or was 

 lower than in Aino females. 



Probably correlation is to some extent affected by sex 

 in most animals. Thus Duncker f determined the cor- 

 relation coefficients of 40 pairs of measurements in male 



* Vide Phil. Trans. 1898, B. p. 178. 



f Wissenschaf tliche Meeresuntersuchungen aus der biologischen 

 Anstalt auf Helgoland, Bd. iii. p. 351. 



