200 



extent and this applies in the same manner to indol-positive 

 and indol-negative strains; other sera agglutinate numbers of 

 strains irrespective of whether they are indol-positive or indol- 

 negative. 



Of the 11 ,, indol-positive" sera 9 gave on an average better 

 group agglutination with indol-positive strains than with indol-nega- 

 tive, but with two sera the converse was the case. All the 3 „indol- 

 negalive" sera agglutinated indol-positive strains to a greater extent 

 than indol-negative.* If there is anything to be learned from the 

 table it is that on the whole, indol-positive strains are agglutinated 

 more readily than indol-negative both by „indol-positive" and ^indol- 

 negative" sera, but any correlation between agglutination and indol 

 formation cannot be presumed from it. 



We might think that both the agglutination and the indol 

 reaction would form an adequate basis for a rational classifi- 

 cation. But as no relation between them can be demonstrated, 

 at least one of them must be rejected for such a purpose. 

 It is probable that neither of them is sufficient for a natural 

 grouping (that is to say, one that is founded on all the cha- 

 racters). 



A connection between indol formation and the slight diffe- 

 rences in the degree of haemoglobinophilia could not be found; 

 nor between the former and the temperature minimum for 

 growth or resistance against heating. 



But a certain relation could be shown to exist between 

 the indol reaction and typical or atypical growth, in that 

 almost all the atypical strains which were tested for indol 

 formation gave a negative reaction. 



This was the case with H127, 151, 158, 161, 171, 176, 178, 179, 

 181, 205, 314a, c, 318, 319, 322, 330, 332a, b, 335, 336, GP1, 2, 3, 4a, 

 b, 5; only H314b and d and 324 gave a positive reaction. 



In contrast to this nearly 90o/ of the typical strains formed 

 indol. The connection between kind of growth and the indol 

 reaction is therefore very marked. Since the distinction between 



* It would not cause any change here if the error introduced 

 by the incomplete titration of some of the samples was corrected 

 (see pp. 186—187). 



