ANTIGENIC VARIATION 719 



been accepted as defining separable varieties are very slight. Sometimes they 

 are revealed only by the use of a special medium or a special reaction. The 

 definition and labelling of these fermentative types has been carried out by a 

 few observers, particularly by Kauffmann, who would extend the process even 

 further than many other workers would be prepared to do (see Kauffmann and 

 Buron 1935). The significance of these types must, we think, remain very doubtful 

 until we have far more extensive data than are at present available. At the 

 same time the existence of the differences in euzymic activity can clearly not be 

 ignored, if only because the evidence that exists to-day, scanty as it is, suggests 

 that they are correlated with natural habitat and natural pathogenicity. 



To take, as an example, the fermentative varieties of Salm. enteritidis, Kauffmann 

 (19356) records the origin of small samples of these strains. Of 8 strains of Salm. enteritidis 

 examined by him, 7 were isolated from man, one from a guinea-pig. Of 22 strains of 

 Salm. enteritidis var. danysz, 13 strains came from human infections, 9 from rats or from 

 the Ratin virus used for the extermination of these vermin. The 8 strains of Salm. enteri- 

 tidis var. chaco all came from infected soldiers in the Chaco war. Of 7 strains of Salm. 

 enteritidis var. essen, some came from human infections, some from infections in ducks. 

 Again, strains of Salm. paratyphi B that ferment (Z-tartrate were found by Kristensen 

 and Kauffmann (1937) in patients suffering from enteritis ; strains causing typical para- 

 typhoid fever do not usually ferment £?-tartrate. Strams of Salm. typhi-murium that 

 infect European ducks and chickens do not as a rule ferment rhamnose in opposition to 

 those found in the United States, which are rhamnose -positive (Edwards and Bruner 1940a). 



If the fermentative abiUties described are found to be constant for any particular 

 strain, and if the varieties differentiated by them are found to show significant differences 

 in theu" natural distribution, there will clearly be a good case for providing them with 

 separate labels. At the moment it is wisest to regard these labels as provisional. 



The Bacteriophage Method of typing Sahnouella Strains. — Of far greater value, 

 though of more restricted application, is the method of subdividing antigenic 

 types into sub-types by means of the bacteriophage method of typing introduced 

 by Craigie and Brandon (1936a, b) and Craigie and Yen (1938). It was known 

 that salmonellae were susceptible to the action of bacteriophages, and that these 

 were closely related to the type of antigen present ; but owing to the wide 

 communal sharing of the same antigen, bacteriophages of this sort were useless 

 for the differentiation of sub-types. Soon after the discovery by Felix and Pitt 

 (1934a, b) of the Vi antigen of the typhoid bacillus, several workers established 

 the existence of bacteriophages acting specifically on the V form of the bacillus. 

 The special contribution of Craigie and his co-workers was their observation of 

 a peculiar adaptability possessed by one particular anti-Vi phage. When this 

 phage was grown on typhoid strains of different origin, races of bacteriophage 

 were obtained that were found to have developed a high degree of specificity 

 for the particular strain on which they had been propagated. By means of a 

 series of bacteriophages prepared in this way, Craigie and Yen were able to classify 

 nearly all of 592 strains of typhoid bacilli into six bacteriophage types. The 

 evidence obtained from a study of the origin of the strains revealed a high 

 degree of correlation between the bacteriophage type and the epidemic source. 

 These observations have since been abundantly confirmed, so that the bacterio- 

 phage method of typing is rapidly becoming a routine procedure in the investiga- 

 tion of outbreaks of typhoid fever. The degree of specificity seems to be almost 

 complete, and the results are more constant and reliable than those of typing 

 by the fermentative method. 



