BACTERIAL CLASSIFICATION 29 



what is implied by these terms. Quantitative measure- 

 ments should be made wherever possible: if this grade 

 of exactness cannot be attained color charts or full 

 explanation of the meaning of each term should be 

 forthcoming. 



Finally, if our view is correct, bacterial types should 

 never be described on -the strength of an examination of 

 single individual strains; but only after a comparative 

 study of the numerical frequency of each particular 

 character in a considerable series of cultures. When 

 the measurement of such a property as gelatin lique- 

 faction in a given series of organisms shows distinct 

 bimodal curves (as in Fig. Ill on page 100), each 

 mode may be considered a distinct type. In the curves 

 of Fig. Ill, for example, it is clear that two centers of 

 variation exist in each of the three groups studied 

 one a liquefying and the other a non-liquefying type. 

 Often a quantitative comparative study is unnecessary, 

 in order that a type difference of this sort should be made 

 clear. For example, a survey of medical literature alone 

 is sufficient to show that the diplococcus of meningitis 

 and the diplococcus of pneumonia constitute two real 

 type centers in the general group to which they belong. 

 A vast majority of the organisms associated with one 

 pathological condition are flattened, biscuit-shaped pairs 

 of cells which decolorize by Gram and exhibit mutual 

 serum reactions. A vast majority of the cocci from the 

 other condition are elongated lanceolate cells which 



