198 C.-E. A. WiNSLow AND Anne F. Rogers 



found modifications of the whole center of gravity of the organism, 

 and the smaller groups by those characters which, though perhaps 

 showing sharper individual differences, vary by themselves without 

 affecting any other properties. 



Our object therefore has been, not to establish arbitrary boundary 

 lines, but to discover existing natural types distinguished by the 

 association of independent characters. In such a task it is obvious 

 that those characters are most important which show the most marked 

 correlations. What these characters are must be determined by study 

 in each particular group. Chromogenesis or gelatin liquefaction 

 may be of generic value in one family, or may mark only varieties 

 in another, as it is, or is not, correlated with a number of other 

 properties. In the Coccaceae, for example, the liquefaction of gelatin 

 and the reduction of nitrates appear, when judged by this standard, 

 to be of less importance than most of the properties studied. In some 

 cases they appear to be significant, but, in most of the groups indi- 

 cated, liquefying and non-liquefying forms, and reducing and non- 

 reducing forms, run parallel. Distinctions based on such a single 

 character alone may have specific, but certainly not generic, value. 

 On the other hand, we have been somewhat surprised to find that 

 such apparently fluctuating characters as chromogenesis and the 

 reaction to the Gram stain are strongly correlated with a number 

 of other properties. 



A general survey of the whole field of variation among the Cocca- 

 ceae indicates clearly the existence of two main sets of correlated 

 characters, corresponding to the subfamilies which we have suggested 

 in a previous communication (Winslow and Rogers, 1905). Habitat, 

 morphology, staining reactions, surface growth, acid production, 

 optimum temperature, and chromogenesis, all vary simultaneously 

 in one or the other of two directions, defining the two subfamilies 

 Paracoccaceae and Metacoccaceae. The first group, comprising most 

 of the forms from the body, shows, as a rule, chains and irregular 

 cell-grouping, stains by Gram, yields a meager or only fair surface 

 growth, forms acid in carbohydrates, and produces no pigment, or a 

 white or an orange one. The other group, from earth and water for 

 the most part, often shows packets, decolorizes by Gram, grows well 

 on artificial media, fails to ferment carbohydrates, and produces a 



