CH. v] VAEIATIONS IN FERMENTING POWER 67 



proach, the reactions are riot necessarily constant. There is 

 a great deal of evidence to show that spontaneous variations 

 frequently occur. 



5. In any case, the classification of bacteria according to 

 their action on certain ca,rbohydrates is a very artificial one. * 

 Twort emphasises the fact that the difference between one 

 organism which produces such slight acidity that the alkaline 

 reaction of the medium completely masks it, and another 

 organism which produces a slight but definite acid reaction, 

 is no greater than the difference between the latter organism 

 and a third which produces marked acidity. In the same way 

 the difference between an organism which yields acidity in 

 24 hours and one which requires 48 or even 72 hours to do so, 

 cannot be regarded as a fundamental one. It is merely a matter 

 of degree. 



6. Further, the decision as to which group of carbohydrate 

 compounds (the sugars, the glucosides, the starches, etc.) shall 

 constitute the test substances is a purely arbitrary one, 

 depending not infrequently upon their cheapness and the 

 facility with which they can be obtained. If one group of 

 carbohydrates be chosen a certain classification will follow ; if 

 another group be selected an entirely different classification 

 may result. 



The only justification for founding a classification upon 

 one series of experiments rather than upon the other is the 

 fact that the classification so obtained corresponds more closely 

 to differences brought out in other ways, such as differences 

 in agglutination or pathogenicity. 



If these other differences are inconstant and distinctions 

 based on them have been found to be unreliable, then a series 

 of fermentation tests designed to correspond with them is at 

 once suspect and cannot be trusted as a final appeal. If on 

 the other hand these other differences (in agglutination, patho- 

 genesis, etc.) are constant and have been found to justify a 

 certain division into "species," a series of fermentation tests 

 which correspond to them may afford a very much simpler 

 method of deciding to which of these species a certain 

 organism belongs, and also of separating one species from 



5—2 



