2 , STATISTICAL METHODS. 



Met5:orto of Collecting Individuals for Meas- 

 urement, 



In collecting a lot of individuals for the study of the varia- 

 bility of any character undue selection must be avoided. The 

 rule is: 



Having settled upon the general conditions, of race, sex, 

 locality, age, which the individuals to be measured must fulfil, 

 take the individuals methodically at random and without possible 

 selection of individuals on the basis of the magnitude of tJie 

 character to be measured. If the individuals are simply not 

 consciously selected on the basis of magnitude of the character 

 they will often be taken sufficiently at random. 



The number ot variates to be obtained should be large; if 

 possible from 200 to 2000, depending on abundance and 

 variability of the material. 



Processes Preliminary to Measuring 

 Characters. 



Some characters can best be measured directly; e.g., the 

 stature of a race of men. Often the character can be better 

 studied by reproducing it on paper. The two principal 

 methods of reproducing are by photography and by camera 

 drawings. 



For photograpliic reproductions the organs to be measured 

 will be differently treated according as they are opaque or 

 transparent. Opaque organs should be arranged if possible 

 in large series on a suitable opaque or transparent back- 

 ground. The prints should be made on a rough paper so 

 that they can be written on ; blue-print paper is excellent. 

 This method is applicable to hard parts which may be studied 

 dry; e.g., mollusc shells, echinoderms, various large arthnx 

 pods, epidermal markings of vertebrates and parts of the 

 vertebrate skeleton. Shadow photographs may be made of 

 the outlines of opaque objects, such as birds' bills, birds' eggs, 

 and butterfly wings, by using parallel rays of light and inter- 

 posing the object between the source of light* and the photo- 



* A Welsbach burner or an electric light are especially good. Minute 



