34 WiNSLow : A statistical criterion for bacteria 



and as adapted to definite quantitative measurement. When the 

 results obtained were analyzed from the statistical standpoint, 

 groups and subgroups of various grades were obviously apparent ; 

 and it was possible to draw up on this basis a general outline of a 



■ h 



classification for the group, which appears to correspond to true 

 phylogenetic relationships (Winslow, C.-E. A., & Winslow, A. R., 

 Systematic Relationships of the Coccaceae ; New York, 1908). In 

 the first place, two major divisions or subfamilies were distinguished, 

 one saprophytic and the other parasitic. The cocci from the hu- 

 man body generally occurred in chains or small irregular groups of 

 cells, stained by the Gram stain, formed a meager or only faii« sur- 

 face growth on media, produced acid in carbohydrates, and showed 

 no pigment or a white or orange one. The cocci from earth and 

 water, on the other hand, occurred in large cell groups or packets, 

 decolorized by the Gram stain, grew well on solid media, failed to 

 ferment carbohydrates and produced a red or yellow pigment. 

 Each character was sometimes found in the group where it did not 

 normally occur, but on the whole the average correlation was 



F 



very strong. Within each subfamily several groups of a second 

 grade of individuality were found, marked by the association of a 

 smaller number of characters than the subfamilies, but still defined 

 by the correlation of several independent properties. These were 

 called genera. Finally, within the genera, species were defined 

 by the distinct centers of variation which were manifest for single 

 independent characters, such as the liquefaction of gelatin and the 

 reduction of nitrates. 



Curiously enough, while we were at work upon the relationships 

 of the family of the Coccaceae as a whole, Andrewes and Horder 

 in England were applying almost exactly the same principles to 

 the classification of the subtypes included in the genus Strepto- 

 coccus, a genus which is so marked in its characters that it has 

 been one of the few groups of the Coccaceae long recognized as a 

 result of common observation. What species should be included 

 in the genus Streptococcus has been the subject of long and fruit- 

 less debate ; and the *' Vielheit '' or the ** Einheit " of the Strepto- 

 cocci seemed an insoluble problem. Andrewes and Horder, how- 

 ever, by studying the numerical frequency of occurrence of various 

 forms distinguished seven type centers which they could properly 



