18 RELATIONSHIPS OF THE COCCACE^l 



single characters which have varied by themselves with- 

 out affecting other properties. 



If the Linnaean system is to be used among the 

 bacteria, however, it should be used correctly. Much 

 of the confusion in bacteriological literature results from 

 neglect of the simple rules of nomenclature. The prin- 

 ciple that a species should bear two Latin names, generic 

 and specific, and two names only, has been ignored by 

 many medical workers; and few bacteriologists, except 

 Migula and Chester, have respected the principle of 

 priority, which requires that a species shall bear the 

 name given to it in the first published description, suffi- 

 ciently full for identification. 



It was with the general views just outlined that we 

 undertook in 1905 a study of the systematic relationships 

 of the Coccaceae by biometrical methods. Five hundred 

 different strains of cocci were collected from various 

 sources and each was submitted to a series of eleven 

 definite and, in most cases, quantitative tests. The 

 results obtained were carefully analyzed with the follow- 

 ing objects in mind (Winslow and Rogers, 1906): "We 

 have first plotted the frequency curve for each character 

 in order to find whether the array varies about one 

 or several modes, and where these modes are situated, 

 with some measure of the extent of variation about these 

 centers. In the second place, we have calculated correla- 

 tion factors for the most significant pairs of characters. 

 Each mode on the curves of frequency may fairly be 



