THE species Escherichia coli is a large and heterogeneous group of organisms divided into types on the basis of their antigenic structure. There is some evidence that certain properties are found more commonly in strains isolated from disease processes than in those from normal faeces. These properties include haemolysin production, which has a high incidence in strains causing urinary-tract infections (Dudgeon, Wordley and Bawtree, 1921 ;Vahlne, 1945) ; necrotoxin production, which is associated with haemolysin production and is found in certain serotypes (Sjostedt, 1946); cytotoxicity in tissue culture, closely associated with the production of soluble haemolysin (Chaturvedi et al., 1969); production of large amounts of K antigen by strains infecting the kidney (Glyn, Brumfitt and Howard, 1971); and production of mucinase (ROSS, 1959) and of enterotoxin (Taylor, Maltby and Payne, 1958) by strains causing infantile gastroenteritis.Because of the evidence that carriage of E. coli round the external urethral meatus may be important in the pathogenesis of infection of the urinary tract (O'Grady et al., 1970), we have compared the properties of strains from this area which subsequently gave rise to infections of the urinary tract with strains which although colonising the urethral meatus did not give rise to infection. The association within strains of a number of different properties has also been studied. MATERIALS AND METHODSFifty strains of E. coli from urinary-tract infections, 397 strains from normal faeces and 30 strains obtained by swabbing the external urethral meatus (introital strains) were examined. Of the 30 introital strains, 15 subsequently gave rise to infections of the urinary tract in the patients from whom they were isolated. The patients from whom the remaining 15 strains were obtained either did not develop infections during the three months following the isolation of the strain or developed an infection due to a different strain. All these organisms were examined for the production of haemolysin and necrotoxin and serotypes were determined.To study the association in individual organisms of a number of different properties, 186 further strains of E. coli were examined. Fifty were from the faeces of ulcerative-colitis patients in remission, 50 from the faeces of ulcerative-colitis patients in relapse, 36 from the faeces of normal people and 50 from urinary-tract infections. These strains were not consecutive isolates but were chosen to include as far as possible equal numbers of haemolytic and

SUMMARY Strains of E. coli from the faeces of patients with active ulcerative colitis and with colitis in remission were examined for haemolysin and necrotoxin production. Cultures from 63 patients with active colitis grew haemolytic E. coli in 23 (37 %) as compared with 24 (21 %) from 1 15 patients whose disease was in remission (P < 0.05). The corresponding proportions for necrotoxin-producing strains were 22% and 12 %. Of 35 patients investigated both in relapse and remission of the colitis, 14 changed their carriage of haemolytic E. coli. Thirteen of these carried haemolytic organisms in relapse but not in remission, and one carried haemolytic organisms in remission but not in relapse (P < 0.01).Strains of E. coli were also examined from specimens of faeces obtained at weekly intervals for 28 weeks from 19 patients. The acquisition of haemolytic or necrotoxic E. coli strains tended to follow rather than precede the onset of the attack in the four patients who developed a relapse of the disease during this period. Among these 19 patients haemolytic and necrotoxic strains were found most commonly when blood was regularly present in the faeces. These facts suggest that conditions in the bowel during a relapse of colitis tend to favour the presence of haemolysin-and necrotoxinproducing organisms, rather than that these organisms cause the relapse.Of 50 strains of E. coli from patients treated with sulphasalazine tested for sensitivity to sulphapyridine 49 were resistant. There was no clear relation between the activity of the colitis and the presence of any of the 27 serotypes of E. coli examined.It has previously been shown in a small series of patients that the strains of Escherichia coli in the faeces of patients with ulcerative colitis differ from those of normal persons (Cooke, 1968). Two studies have been undertaken to extend this observation. In the first, the faecal coliform flora of patients with active colitis has been compared with that of patients whose disease was in remission; in the second, the time relations between a change in faecal flora and the activity of the disease were studied by examining the faeces of 19 patients at weekly intervals for six months. The opportunity has also been taken to assess the sensitivity to sulphapyridine of strains of E. coli from patients with colitis treated with sulphasalazine. 'Present address: