VARIETIES OF COLON BACILLI 199 



suggested by Andrewes and Horder (1906) and Winslow 

 and Winslow (1908) in the study of the cocci. The 

 essential point about this method is that the characters 

 of the organisms studied are not considered indepen- 

 dently, but in relation to each other. The individual 

 reactions are first studied quantitatively in a considerable 

 series of allied strains, so that those types of reaction 

 which are manifested by a large number of strains 

 may be distinguished from the rarer intermediate 

 variations. In the second place, the correlations be- 

 tween different characters are used as a basis for group- 

 ing the types on the assumption that a coincidence in 

 several characters indicates a closer relationship than 

 any single character alone. 



The statistical method has been applied to the colon 

 group in two extensive investigations, neither of which 

 has yet been published in full. Of the first by Howe, a 

 brief, abstract has appeared (Howe, 1912). The second 

 by L. A. Rogers, W. M. Clark, and B. J. Davis we 

 have had the opportunity of seeing in manuscript. 

 These two papers promise at last to lay a foundation 

 for a sound knowledge of the relationships of the colon 

 group. 



Howe (1912) in his investigation dealt with 630 

 strains of fresh intestinal colon bacilli. He concluded 

 from his exhaustive study that in bacilli of this type 

 isolated directly from stools, the characters of motility, 

 indol formation, ammonia production, nitrate reduction, 

 fermentation of mannite, dulcite, and starch were not 

 sufficiently correlated with each other or with other 

 characters to be of classificatory value. Dextrose, 



