302 BACTERIAL VARIATION 



occurs ill other species. It has been described, for example, in M. tetragenus (Reimann 

 1937a) and in Hcem. influenzce. (Chandler, Fothergill and Dingle 1939). In the influenza 

 bacillus, the M — >- 8 — >- R variation occurred spontaneously ; the reversion R — >- S, 

 but not S — y M, could be induced by serial passage through mice. There is also another 

 colonial variant, found in a number of bacterial species, which does not fall into line with 

 the S — >- R, or the extended M — > S — > R scheme of variation (see Morton 1940). It 

 is the D (dwarf colony) variant described by Hadley (1927, 1937), and must be distinguished 

 from his G (gonidial) variants, which give minute colonies that in some cases are said to 

 contain filtrable elements capable of growing into the normal, modal forms of the parent 

 bacteria. The D variant is characterized mainly by the small size of the colony, in com- 

 parison with the S variant. The variations S — >■ D and D — >- S have been observed 

 in a number of species. In many cases, colony size is the only conspicuous feature dis- 

 tinguishing D from S variants. Thus, the D and S variants of a strain of Salm. typhi 

 (Morris, Sellers and Brown 1941) and of a Group C streptococcus (Morton and Sommer 

 1944) differed little in virulence, or in biochemical and serological reactions. 



Nevertheless, in general, there is in each of the species we have described a 

 variation of the same essential character as those listed above — loss of normal, 

 type-specific, surface antigen associated with loss of virulence. There are, how- 

 ever, variations within the S form, both natural, and induced by exposure to 

 antisera or bacteriophage, which have no obvious connection with S — > R 

 variation by loss. For example, Takita (1937) records a change from " specific " 

 antigens to " group " antigens in the Sh. flexneri group, analogous to the diphasic 

 flagellar variation in the salmonellae ; and Kauffman (1941) records a variation 

 by loss of one of the sub-types of XII somatic antigen in certain strains oi Salmonella. 



There are other reasons also for rejecting the hypothesis of a necessary con- 

 nection between virulence and the cultural and antigenic characters of smooth- 

 ness. A loss of, or a considerable change in, virulence may occur apart from the 

 loss of the specific antigenic component characterizing the normal smooth form. 



Wilson (1928) observed variants of Salm. typhi-murium which were both smooth and 

 avirulent, and which must be regarded as true variants, as judged by their failure to 

 revert easily to the virulent parent type. Boivin (1939) found that two strains of Salm. 

 typhi-murium, one with a mouse M.L.D. of over 10,000 bacilM, the other with an M.L.D. 

 of under 200, both yielded approximately the same amount of " smooth " hpopolysac- 

 charide endotoxin, and that the two endotoxins were equally toxic and had equal immu- 

 nizing power. Thus virulence may vary independently of the antigen. Again, Shaffer, 

 Enders and Wu (1936) described two strains of Type III pneumococcus, which were both 

 fuUy capsulated, and antigenicaUy identical. One was avirulent for rabbits, and had 

 a greater tendency than the other, a virulent strain, to lose its capsule. The rough strains 

 of both could be induced to revert to the smooth forms by growth in the presence of smooth 

 kiUed cultures (see p. 305 below), and, whether the killed cultm-e was of the original virulent 

 or avirulent strain, both reverted to their original high or low virulence. The virulence, 

 in fact, was not an expression of the degree of " smoothness " of the two strains, but 

 dependent on some stable difference in physiological behaviour. Large changes in viru- 

 lence can in fact be induced in a strain without altering its smooth characters. Thus, 

 Hadley and Wetzel (1943), starting with a rough variant of an alpha-hsemolytic strepto- 

 coccus, raised its virulence by serial passage through mice. The total increase in virulence 

 during the transformation from rough to smooth was 140-fold ; subsequent passage of 

 the smooth form through mice raised its virulence 5,000-fold. It may be noted also 

 that Bhatnagar (1940) and Jawetz and Meyer (1944) could detect no difference in the 

 antigenic surface of virulent and avirulent strains of Past, pestis, though, as the latter 

 authors suggest, in this case virulence may have been associated with an antigenic com- 

 ponent that did not appear on the surface of the bacillus. 



