86 PROCEEDINGS OF TJIK A NA K ).M ICA 1, AND A NTH KOl'Ol.OtilrA I. 



object in both cases is to find certain numbers which shall be char- 

 acteristic of the race or races existing in the populations of various 

 countries, and by this means to trace alfinities and differences. 



1. THE EMPIRICAL METHOD. 



Anders Retzius, the Swedish anthropologist, was the founder of 

 the Empirical Method. In 1840 he published his method at Stockholm. 

 He measured the maximum length and breadth of the skull, but he 

 considered the absolute values of these dimensions of very little import- 

 ance compared with their ratio, the <-<'i>/Kt!i<- ///</c,>: Strictly defined, 

 the cephalic index is the breadth multiplied by 100 and divided by 

 the length. 



The cephalic index has been accepted by most anthropologists as 

 one of the best criteria of race from the time of Retzius down till 

 quite recently. There are now, however, a number of anthropologists 

 who have expressed doubts as to its value. 



Retzius divided races into dolichocephalic and brachywphalic, and 

 subdivided these again into orthognatkwts and prognathous. 



The classes of skulls denoted by the terms dolichocephalic and 

 brachycephalic are denned differently by different anthropologists. 

 Retzius used centres to define his groups, the centre of the dolicho- 

 cephalic group being a cephalic index of 75, and of the brachycephalic 

 group, an index of 837. Welcker used limits, and introduced an 

 intermediate group, viz., orthocephalic. Dolichocephalic skulls (ac- 

 cording to Welcker) were those with a cephalic index of 72 and 

 under ; orthocephalic, those with indices from 72 to 80 ; and brachy- 

 cephalic, those with indices of 80 and over. Broca's dolichocephalic 

 and brachycephalic groups correspond very closely with those of 

 Retzius, but he introduced a considerable number of intermediate 

 groups. 



All these subdivisions are purely arbitrary. There appears to be 

 no special reason for fixing the limits of the groups at one index more 

 than at another. 



Welcker and others following Retzius have greatly extended the 



