On Certain Bacteria from the Air of New York City. 325 



pen precises." They overlook the fact that though the motility 

 ma}' var^', the flagella may be uniforml}' present as shown by 

 Moore in the case of B. coli-communis,* and further the varying 

 arrangement of these flagella when present, as shown b}^ the 

 classification of Messea.f It appears to the writer, that with a 

 perfectly satisfactorj' method of staining, the structures upon 

 which motilit}' depends should form the basis of a more nearly 

 natural classification than any in general use. At present, how- 

 ever, the practical difficulties are too great to allow of rapid and 

 accurate Avork on such a basis. 



So far, then, as generic characters go, these are, or maj' be 

 based upon morphology, as in the higher forms of life. But not 

 so with specific characters. Here morphology fails, for the 

 minor differences in structure, pattern or coloration, etc., which 

 we are accustomed to use, are either lacking, or so slight as to 

 be inappreciable. To separate species, we have to fall back upon 

 physiological characters. In other words, we place the several 

 species under approximately uniform abnormal conditions (arti- 

 ficial cultures) and note their behavior. Perhaps we may be 

 justified in assuming that the several individuals of a species 

 will behave somewhat in the same manner under these definite 

 abnormal conditions, but we have no criterion as to the extent of 

 variation to be expected. Physiological characters iiave not 

 been used to a sufficient extent among other groups of living 

 things to enable us to predict whether they are more variable 

 than the characters ordinarily used or not. Clearly it is a mis- 

 take to assume that the}' are less variable, though this assumption 

 seems to have been very often tacitl}' made. 



But granting that the physiological characters are sufflcientl}^ 

 constant for practical use, we must have enough of them to be 

 able to distinguish positively between every species. The lique- 

 faction of gelatin is not enough. The characters of growth forms 

 and color are often too variable, too similar, indefinite or difficult 

 to describe recognizabl}-. We should have more positive tests, 

 and as such are not wanting they should be universally applied. 

 It is impossible to say how many are needed ; this can only be 

 shown by experiment, and when we finally- get some idea of the 

 €xtent of the bacterial flora of the world. 



* V. A. Moore, ^Vilder Quarter Century Book, pp. 339-365. 



f Mesea, Rivista d'igieue e sanita publica, No. 14, p. 513 (1889). 



