THE epidemiology of Escherichia coli is poorly understood, and although serological characterisation of strains affords an excellent method of typing E. coli, it is time-consuming and can be performed only in specialised laboratories. Another method would be of value. Colicine typing as a method of differentiating strains has been used (Linton, 1960;McGeachie, 1965 ;Mustafa and Dunbar, 1970), but not so extensively as serotyping and its usefulness has not been defined.Colicine typing is simple to perform and has been found valuable in epidemiological studies on shigellae (Abbott and Shannon, 1958 ;Gillies, 1964;Hart, 1965), particularly on strains of Shigella sonnei because they are serologically and often biochemically homogeneous (Hart).This paper records our experience of colicine typing of E. coli and of the validity of the method as an epidemiological tool. Serological and colicine typing were used in this investigation to determine whether the combination of the two methods would provide a finer subdivision of strains, to show the frequency of colicine production and sensitivity among serologically defined strains, and to show any correlation between colicine types and antigenic structure of the organisms. This is particularly interesting because several workers attempting to purify colicines have observed a close association between colicines and the 0 antigen of the organism (Goebel and Barry, 1958;Hutton and Goebel, 1961;Barry, Everhart and Graham, 1963).An epidemiological application of the typing method is also reported.
MATERIALS AND METHODS
Specimensof faeces were collected from patients in a female, open medical ward as soon as possible after admission and thereafter at weekly intervals for 1 yr. The patients were those usually seen in a general medical ward and particularly included diabetics.