338 Transactions of the Royal Society of South Africa. 
more agglutinin than is required for complete agglutination of the organisms. 
It was found that if these agglutinating sera were absorbed (1) by the 
homologous strain, (2) another strain of the same type, and (3) another 
strain of a different type, and then tested with the homologous strain, the 
agglutinin could be almost completely absorbed by the homologous strain, 
but that the absorptive effect of the other organisms was relatively weak 
and equal in degree (Tables XII and XIII). 
For this purpose concentrated emulsions were prepared, and a certain 
dilution, 1 : 500 or 1 : 1000, of the antiserum was treated with the different 
organisms at 37° C. for two hours ; the mixtures were centrifugalised till the 
supernatant fluids were quite clear ; these were pipetted off and tested each 
with the homologous strain in various further dilutions. 
Absorption tests were also carried out with the antiserum which showed 
the paragglutination phenomenon. It was found that while treatment of 
the serum by the strain used for immunisation removed the agglutinin both 
for this strain and the heterologous strain, the heterologous strain was only 
capable of absorbing its own agglutinin (Table XIV). This is true for 
coagglutination effects generally (Castellani), and showed that the two 
strains were not serologically identical. 
It was thus quite impossible to establish any differentiation by agglutination 
tests betiveeti different types of these coliform bacilli, and no species specificity 
was displayed by their agglutinins; the specificity is restricted to the indi- 
vidual strain, and, by the ordinary agglutination reaction, and also by 
absorption tests, it is quite impossible to demonstrate that organisms of the 
same species (determined by cultural reactions) as the homologous strain, 
are more closely related to it than representatives of other types. Thus the 
individuality of the bacterial strain is most strikingly elicited. In this 
group strains are constantly assuming new characters {v. p. 360) and 
becoming highly specialised. Hence we must assume that a high degree of 
individuality is attained by each strain and that differences of cultural 
characters within certain limits are of little significance. 
Results observed with aiitisera to B. types 1 and 2. — In the foregoing 
observations regarding the action of B. coli agglutinins, the commoner 
types, B. MacConkey No. 71, B. coli communis, B. Griinthal, B. vesictdosus 
belonging to the indol forming sub-group (A) of coliform bacilli were 
studied ; further experiments were then carried out with agglutinating sera 
to certain of the B sub-group which are of less frequent occurrence. 
Antisera were obtained to strains of types Bl {B. MacConkey No. 74) 
and B2 whose characters are shown in Table III. A number of other 
strains identical in their characters to these types were tested with the 
respective immune sera. The results with an antiserum to strain 1, type Bl 
are shown in Table XV, It was noted that while the strain used for 
immunisation was agglutinated by a 1 : 8000 dilution of the serum, three 
