148 C.-E. A. WiNSLOw AND Anne F. Rogers 



ber of constant characters and easily identified by mutual fertility. From one point 

 of view each distinct race might be considered a species; but to apply a name for every 

 grade of difference in each varying character would be impracticable; and such names 

 could have no true specific value. The best solution of the difficulty is the establish- 

 ment of certain types around which the original organisms may be more or less closely 

 grouped; but it must be clearly recognized that the groups thus formed are defined by 

 relation to the type at their center and are not sharply marked off at their extremities 

 from the other groups adjacent.' 



For these reasons the science of systematic bacteriology has 

 remained in a notably undeveloped state. A score of large groups 

 of bacteria have been more or less satisfactorily recognized by Fliigge 

 (1896) and others. Certain of these groups, like the aerobic spore- 

 formers, the colon bacilli, and the diphtheria bacilli, doubtless 

 represent true natural famihes or genera. In one such group, that 

 of the aerobic spore-formers, where appreciable morphological dif- 

 ferences exist, the species and varieties have been carefully worked 

 out by Chester (1904). Far too many specific names among the 

 bacteria however, mean less than nothing. The incomplete descrip- 

 tion of a vast number of identical or minutely differing forms has led 

 to a confusion quite disheartening to the student of such systematic 

 works as those of Migula (1900) and Chester (i 901). Among the 

 Coccaceae we have compared the published descriptions of 445 

 species and found evidence for only 31 distinct types (Winslow and 

 Rogers, 1905). These are defined mainly by arbitrary combina- 

 tions of the three characters of acid production, chromogenesis, and 

 the liquefaction of gelatin. It is small wonder that most bacteri- 

 ologists have abandoned any attempt at a natural classification, 

 and have sought refuge in such frankly arbitrary schematic group- 

 ings as those of Fuller and Johnson (1899), Weston and Kendall 

 (1902), and Jordan (1903). The same tendency carried to its 

 extreme is shown in the decimal systems of Gage and Phelps 

 (1903), and Kendall (1903), and in the modifications recently adopted 

 by the Society of American Bacteriologists. 



These systems are most valuable for a routine descriptive work, 

 and for arranging and cataloguing records of cultures. They may, 

 however, lead to error, unless used with due caution. In the first 

 place, the determinations on which such schemes are based are usually 

 qualitative only and not quantitative. In the second place, the 



' Winslow and Rogers, 1905. 



