Generic Characters in the Coccaceae 199 



yellow or red pigment. It must always be remtmbLrtd that each 

 character may occasionally be found in the group where it usually does 

 not occur; but the association of these properties in the vast majority 

 of cases is very strong. We desire to extend our earlier definitions 

 of the two subfamilies by including the Gram reaction and chromo- 

 genesis; and the subfamilies as thus modified will be defined at the 

 end of this communication. It is a striking fact that these two chief 

 divisions among the Coccaceae correspond to the two markedly dilTerent 

 environments which exist in nature, the body of higher organisms, 

 and the outer world. A close correspondence with environmental 

 conditions should naturally be expected among such simple asexual 

 organisms as the bacteria, and it increases our confidence in the 

 reality of the groups established below to find each of thim localized 

 so sharply in one or other of the two main environments. 



Under the subfamilies we find a second grade of group-individu- 

 ality, marked by the association of a smaller number of characters 

 than the subfamilies, but still defined by the correlation of several 

 independent properties. Here morphology, surface growth, and 

 chrcmogencsis appear to be of greatest importance, acid production, 

 gelatin liquefaction, and nitrate reduction having special significance 

 in certain cases. Five distinct types have appeared with consider- 

 able clearness in the present study. It must be remembered that 

 the fundamental correlations which revealed these groups were 

 derived in an entirely impersonal way by measurements, made on 

 each character independently, generally by different observers, 

 and always without knowledge of the identity of the organism. 

 When individual races are considered, it is possible, by transferring 

 a few cultures on the border-line in a single character, to show that 

 the correlations are really closer than have appeared above. 



By this process we have attempted to group our 500 cultures 

 under the five subdivisions suggested by the correlation tables, and 

 have found the results so satisfactory as to confirm our confidence 

 in their reahty as natural groups. It seems to us that these groups 

 are of such importance as to deserve generic rank. Within each 

 there is ample room for the establishment of such a reasonable 

 number of species as detailed study may warrant. Good genera 

 must first be recognized, however. It is time that bacteriologists 



