Woods: What Is There in Physiognomy? 



3^5 



privilege of reproducing and distribut- 

 ing to newspapers and magazines. 

 Of this large collection I ordered forty 

 chosen at random by Mr. Livernois. 

 I told him I wished to make a study 

 of faces but did not tell him of any 

 theories that I had in mind, or of any 

 of the tests. He took these pictures 

 from his collection one after another, 

 just as it happened to be easy for him 

 to find a negative, and have it printed, 

 therefore the forty Canadian soldiers 

 ought to be a good random sample. 

 They came from all parts of Canada, 

 and were on their way overseas. The 

 forty proved to be forty-one, as one 

 accidental duplicate got included, 

 which was not discovered until after 

 some of the tests were made; but this 

 has been allowed to remain in the 

 collection, as it is amusing to see how 

 often people put the same face in two 

 different classifications. 



Ten persons, some more or less 

 expert in the art of judging faces, or 

 trained in habits of scientific observa- 

 tion, have classified these forty-one 

 pictures putting long or large noses 

 in one pile, short or small in another, 

 and reserving the middle pile for all 

 those that did not strike the eye as 

 belonging in either of the two outer 

 groups. No observer was aware of 

 the classification of any other observer, 

 as the record marks were placed on the 

 backs of the photographs. 



These tests show that whereas there 

 is considerable variation as to individ- 

 ual pictures selected for the three 

 grades, the proportions in the end are 

 sufficiently uniform to satisfy the pur- 

 pose at hand. 



Even for the individual pictures 

 selected, there is good and significant 

 correlation between the judgments of 

 any two observers. Roughly these 

 correlations are about r = .20 tor = .50.^ 



Here are three specimens, in the 

 accompanying squares, of my own 

 first classification compared with the 

 first three other independent judg- 

 ments. 



Opinion of 

 A. E. W. 



Large or 

 § Long 



'c 

 ■q, 

 O Average 



c 



O 



Small or 

 Short 



r = .39 



Opinion of 

 W. W. C. 



Large or 

 Long 



Q Average 



Small or 

 Short 



r = .35 



Check Opinion 



Large Small 



Long Average Short 



Large 

 ^ Long 



5 

 'c 



Q Average 



c 



O 



Small 

 Short 



24 



11 



24 



11 



24 



11 



.42 



* Method of rough approximation, given by Yule, Phil. Trans. A. CXCIV, 257-319. 



