January 16, 1914] 



SCIENCE 



85 



to utilize a certain food stuff. Indeed bio- 

 chemical properties have a peculiar and 

 unique significance among the bacteria, 

 since it is precisely along the lines of metab- 

 olism that these organisms have attained 

 their most remarkable differentiation. The 

 higher plants and animals have developed 

 complex structural modifications to enable 

 them to obtain food materials of certain 

 limited kinds. On the other hand the bac- 

 teria have maintained themselves by ac- 

 quiring the power of assimilating simple 

 and abundant foods of varied sorts. Evo- 

 lution has developed gross structure in one 

 case without altering metaboli-sm; it has 

 produced a diverse metabolism in the other 

 case, without altering gross structure. 

 There is as wide a difference in metabolism 

 between the pneumococci and the nitrify- 

 ing bacteria as there is in structure between 

 a liverwort and an oak. The danger in 

 using physiological characters for classifi- 

 cation lies, not in their inherent unreliabil- 

 ity, but in the fact that so many physiologi- 

 cal properties are directly adaptive in 

 nature. Adaptive characters of similar 

 nature are likely to arise in different 

 groups under the influence of similar en- 

 vironmental conditions and may prove 

 altogether misleading as to true phylo- 

 genetic relationships. Professor Gadow 

 in his striking address before the British 

 Association has called the independent evo- 

 lution of a nearly identical character from 

 homologous material isotely. We have ex- 

 cellent examples among the bacteria. It 

 seems clear for example that we must 

 assume from the presence of liquefying and 

 non-liquefying types among so many of 

 the principal groups of bacteria that this 

 property has lain latent in a great many 

 independent lines of descent and has been 

 independently released in many of them, 

 perhaps by environmental forces. It is 

 particularly to avoid this danger of con- 



fusing independently acquired adaptive 

 characters with those which indicate real 

 community of descent that we must lay 

 stress on the significance of a number of 

 independent characters which occur in 

 correlation. If the correlation is due to an 

 essential dependence of one character upon 

 the other it is of course not particularly 

 significant ; but, when we find a number of 

 different characters, which have no neces- 

 sary connection, correlated together, the 

 presumption is warranted that conunon 

 descent is the connecting link which has 

 united them. 



It was in the study of the Coccaceae that 

 the full importance of emphasis on correla- 

 ted characters first impressed itself upon 

 me. It had long been the practise, follow- 

 ing the Migula system, to group all the 

 staphylococci of the skin and the saprophy- 

 tic cocci which divide in one and two planes 

 together in the genus Micrococcus and to 

 separate the packet-formers in the genus 

 Sarcina. The common cocci found on the 

 skin, all liable to assume at times patho- 

 genic properties, were usually classed as 

 merely three color varieties {aureus, albus 

 and citreus) of a single species Micrococcus 

 or Staphylococcus pyogenes. In the at- 

 tempt to apply statistical principles to 

 the classification of this group, ^ Mrs. 

 Winslow and I collected and studied 

 500 different strains of cocci, measuring 

 quantitatively so far as possible eleven 

 different characters of each strain. At once 

 a new and surprising set of relationships 

 manifested themselves. It was evident in 

 the first place that on the whole the cocci 

 living normally on the body surfaces, dif- 

 fered in almost every respect from the cocci 



1 First published in Biological Studies by the 

 Pupils of William Thompson Sedgwick, June, 1906, 

 and in the Journal of Infectious Diseases for the 

 same year and later elaborated in our book on the 

 "Systematic Relationships of the Coccaceae," 

 N. Y., 1908. 



