Generic Characters in the Coccaceae 197 



in every aggregate, some less constantly. Other organisms never 

 show packets, or only very rarely. Some cocci always stain, and 

 some always decolorize, by Gram, while intermediate forms tend 

 more or less strongly toward either type. In surface growth two 

 distinct types, the faint to meager and the abundant to very heavy, 

 are manifest. In acid production there appear to be three centers 

 of distribution corresponding to organisms which fail to ferment, 

 those which ferment slightly, and those which produce large amounts 

 of acid. In relation to nitrate reduction, three types appear, accord- 

 ing as the cocci fail to reduce, or form nitrites or ammonia. On 

 gelatin the organisms studied group themselves either as liquefiers 

 or as non-liquefiers, and in color production four distinct centers 

 appear, in which the pigment is white, yellow, orange, or red. 



Our estimate of the value of these type-centers is greatly in- 

 creased when we find that the central points for the different characters 

 do not vary independently, but are correlated together to a remark- 

 able degree. Again, we should expect, and we actually find, in some 

 cases, the correlation of single characters varying, those properties 

 generally correlated appearing in certain organisms in exceptional 

 combinations. If, however, we consider, not the single character — 

 not the individual organism — but the aggregate of the correlations of 

 various properties as manifested in a considerable series of indi- 

 viduals, certain well-defined systematic units appear, marked by the 

 association of a number of independent characteristics. Such an 

 association can be explained only on the ground of relationship, and 

 the types marked by the simultaneous occurrence of a number of 

 properties may rightly be taken as the centers from which other, 

 more aberrant individuals have varied. 



The fact that correlation exists shows that, on the average, the 

 fluctuations of these characters do not occur independently, but 

 are so closely bound up with those of other properties as to vary 

 together with them. This may be because the selective action 

 of the environment produces a parallel change in each, or because 

 the two characters are so closely bound together, in the physiological 

 balance of the organism, that a change in one leads to a corresponding 

 variation in the other. In either event it is clear that the larger 

 systematic units (families or genera) must be marked by these pro- 



