PEOFESSOK PEAKSON's CONTRIBUTIONS TO OSTEOLOGY. 471 



The other conckisions are briefly that, for the race in question, 

 (1) " Women are more brachycephalic and more variable than 



men." 



(2) " The younger generation is more brachycephalic and more 



variable than its parentage.''" 



(3) " Parents of sons are significantly less variable than 



parents of daughters." 



" On the Reconstruction of the Stature of Prehistoric Races " 

 (Karl Pearson). — In this memoir the theory of correlation is 

 applied to the problem of the determination of stature from a 

 knowledge of the long bones. The author begins with the 

 theoretical determination of the most probable value of an 

 organ B for given values of (1) a single organ A, (2) n organs 

 Aj, Ag, A3, .... A^^. The equation giving the most probable 

 value of B in the first case is the regression formula con- 

 necting the two organs. This, it will be remembered, is 



15 -'''^6 = "'''■„. (A -m„), 



which relation may be put into the form 



B = r^ + c, A . 



The probable error of such a determination is -67449 o-^ X 

 ^(1 _ -,.2^^^), If ^ be known for /j individuals, the probable error 

 of the mean value of B is 



•67449 ^, X Vr^^, 



J 2} 



Thus we see that the accuracy of the determination increases 

 with the coefficient of correlation and with the number of the 

 known organs. Consequently, while the reconstruction with 

 extreme accuracy of the unknown organs of individuals is not 

 possible, the exactness with which the mean racial type may be 

 reconstructed depends simply on the number of measurements 

 of the correlated organ obtainable. 



The regression formula for n organs leads to a similar 

 equation to the above, viz. — 



B = Co + Cj A^ + r^ A2+ . . . +f» A„. 



