1. Moore's gauze swabs have, in our hands, been shown to be a reliable instrument for the survey of salmonella infection in the environment of two food premises.

2. A large bakery was demonstrated to be regularly contaminated with salmonellae. Thirty-one floor gully swabs out of 111 (27·9%) were positive for salmonellae. Little spread of infection to the general public was experienced.

3. Staff infection with salmonellae was shown to occur in this bakery. Fifteen positive specimens of staff sewage out of 93 were obtained (16·1%). There was little evidence of prolonged contamination of the sewers with salmonellae. Staff infections were probably, therefore, of short duration. The staff probably became infected from contaminated goods in the bakehouse. Rat contamination of sewers was excluded by water trapping.

4. It is not thought that regular examination of staff excreta would be a profitable means of discovering carriers or of preventing contamination of the products. The frequency with which such examinations would have to be performed to discover short-term excreters would be administratively impossible.

5. Of 274 swabs from gullies in abattoirs, 111 were positive for salmonellac (40·5%). Twenty-one different serotypes were isolated. Seventeen of these serotypes are known to occur in animal feedingstuffs.

6. Parallel examination of abattoir swabs and excreta from human infections for S. typhimurium, coupled with phage-typing of the strains, provided information of the geographically local nature of much sporadic illness due to this serotype. Of a total of 30 typable strains, 23 belonged to phage-types found in the abattoirs, or on local farms. Sporadic incidents were represented by 25 of these cultures, 18 of which corresponded in phage-type to those found in the abattoirs or in local farm animals. The phage-types of S. typhimurium isolated from the slaughter houses not infrequently corresponded in timing with the strains isolated from human infections.