165 strains of Citrobacter koseri isolated from clinical specimens were studied and their biochemical reactions determined. They were examined serologically by means of a scheme consisting of 14 O antigens. The sources of the clinical specimens were tabulated and the epidemiological information was summarized. The clinical significance of these findings is discussed.

The faecal flora of 55 children (aged 8 days to 8 years) and 16 adults was determined. All the children were artificially fed from birth. The faecal flora of the youngest age group was generally less complex and less predictable than that of adults. Some bacterial groups commonly found in adult stools, for example bacilli, lactobacilli and yeasts, were rarely found in the youngest infants. Most of the changes towards the adult pattern took place between 4 and 12 months. The faecal flora of children aged 1–4 years generally resembled that of adults, although lactobacilli were still infrequently isolated.

Sixty-three samples of the more solid material (sludge) separated from the effluent plants of dairy factories were examined for the presence of salmonellas and brucellas. Salmonellas were isolated from two samples (S. heidelberg. [1]; S. indiana [1]). No brucellas were isolated. None of the samples supported the growth of S. dublin. Salmonellas added to effluent sludge at a concentration of 106 organisms/ml, survived less than 70 days.

The sludge from dairy factory effluent plants does not appear to be a source for the spread of salmonellosis or brucellosis.

The serological relationship between Brucella spp., Yersinia enterocolitica IX, and the group N salmonella serotypes S. godesberg, S. landau, S. morehead, S. neusdorf, S. soerenga and S. urbana was examined using agglutination, antiglobulin, complement fixation, immunodiffusion and fluorescent antibody methods.

Antisera to the group N salmonella serotypes all reacted to significant titres in agglutination and complement fixation, but not antiglobulin or immunodiffusion tests with smooth brucella antigens. These antisera also reacted in agglutination, but not antiglobulin, tests with Y. enterocolitica IX. They did not react significantly in any tests with rough brucella antigens.

Conversely, antisera to smooth Brucella spp. agglutinated group N salmonellas to low titre and Y. enterocolitica IX to titres similar to those given against the homologous strain. Antiserum to Y. enterocolitica IX on the other hand reacted with smooth brucella antigens to high titre in agglutination, complement fixation and antiglobulin tests, and with the group N salmonella antigens to substantial titres in agglutination tests.

In direct fluorescent antibody tests, smooth Brucella strains and Y. enterocolitica IX reacted strongly with FITC-labelled antibody to Br. abortus whereas the group N salmonella strains reacted weakly.

In tests with monospecific antisera to the A and M determinants of Br. abortus and Br. melitensis respectively, Y. enterocolitica IX reacted only with the antiserum to the A determinant whereas group N salmonellas reacted to low titre with both A and M antisera.

The results of cross-absorption tests confirmed this relationship and suggested that the O30 antigens of group N salmonella serotypes contained antigenic determinants similar to, but not identical with, the antigenic structure shared by smooth Brucella spp. and Y. enterocolitica IX.