800 HEMOPHILUS 



capsulated strains appear to possess a relatively specific protein component P, and a 

 protein component M, which is common to most strains of H. influenzae. It may be noted 

 that several earUer workers found that complement-fixation tests gave evidence of the 

 presence of an antigenic relationship that was not revealed by agglutination (WoUstein 

 1919, BieUng and Weichbrodt 1920, Kristensen 1922). 



It is clear that the change that Pittman describes as occurring in her smooth 

 strains falls within the definition that we adopted for the S -> R, or the 

 M -^ S -> R, variation in Chapter 8, the loss of the surface antigen which deter- 

 mines type-specificity in the normal, virulent form of a bacillus. Pittman's capsu- 

 lated forms are apparently the most virulent pathogenic type of H. influenzce ; 

 they are isolated commonly from the meninges in one of the severest of human 

 infections caused by the organism ; and Gordon, Woodcock and Zinnemann (1944) 

 report that meningeal infections with strains not among Pittman's types are less 

 severe. They are also the most mouse-virulent of the observed varieties (Fothergill 

 and Chandler 1936, Chandler et at. 1937, 1939, Raettig 1940). The implication 

 that most of the strains isolated from the nose and throat of normal persons are 

 in an intermediate or a "rough" phase, must, therefore, we think, be accepted. 

 We must also, it would appear, accept the view that the antigens that dominate 

 the rough forms in this bacterial species are more heterogeneous than those present 

 in the normal smooth phase, a finding that differs from that recorded for most 

 other groups. 



It will be noted that, under this definition, all the strains referred to in preceding 

 sections as " atypical " and many, probably the great majority, of those referred to 

 as " typical," would be classed as rough variants. The antigenic structure of the 

 para-influenza bacilli has not yet been submitted to any special study ; but it 

 seems very unlikely that any of them would fall into Pittman's " smooth " category. 

 Whether they are in any way antigenically different from the non-haemolytic rough 

 forms we do not know. Miles and Gray (1938) found an antigenic relationship 

 between a proportion of the strains of non-hsemolytic H. para-influenzce they 

 studied. 



The Koch-Weeks bacillus, in its serological relationships as in all its other characters, 

 appears to be indistinguishable from H. influenzce. Ivnorr (1924) has found that different 

 strains of this organism show marked antigenic heterogeneity, while some strains are 

 identical with certain strains of the influenza bacillus. 



A small sample of strains of H. influenzce-suis exammed by Lewis and Shope (1931) 

 showed the same type of antigenic heterogeneity that is encountered among the ordmary 

 strains of human influenza bacilli. Comparison with a few strains of H. influenzce of 

 human origin did not reveal any example of antigenic identity, though there was some 

 overlapping in cross-agglutination tests. Similar findings are recorded by Ivirchenbauer 

 (1934). These observations were made before the publication of Pittman's findmgs, so that 

 there was no differentiation between smooth and rough strains. 



We have as yet no information in regard to the antigenic relationships of H. cants. 



H. pertussis differs from H. influenzce in that all recently isolated smooth strains 

 appear to belong to a single antigenic type. Moreover, it would seem that all strains, 

 when first isolated from the body on an optimal medium, are in the smooth state. 

 The behaviour of these strains on artificial culture raises points of considerable 

 interest. 



Bordet and Sleeswyk (1910) noted that recently isolated strains of ^. pertussis, grown 

 on the Bordet-Gengou medium, all agglutinated with a serum prepared against any one 



