SELWYN (1975) described the carriage of antibiotic-producing organisms on human skin and showed how they might protect the carrier against colonisation with potential pathogens such as Staphylococcus aureus. Subsequent work by Selwyn and his colleagues (Marsh and Selwyn, 1977a and b;Milyani and Selwyn, 1978) has shown that these antibiotic producers can restrict the growth of susceptible strains in vitro. This paper reports a study of the carriage, in patients with skin disease, of "antibiotic"-producing cocci with an activity against S. aureus. MATERIALS AND METHODSSamples for bacterial culture were taken from the nose, the chest, the perineum and one or more lesions of patients admitted to the Inpatient Department of St John's Hospital for Diseases of the Skin; cotton-tipped swabs moistened with broth were used. When S. aureus was isolated from a skin lesion the opinion of a clinician, usually the senior house officer, was sought on whether the lesion was clinically infected or only colonised; these opinions were recorded. The swabs were inoculated on to Blood Agar Base (Oxoid CM55) used as a 'nutrient' agar and the plates were incubated aerobically for 24 h at 37°C and left on the bench for a further 24 h at ambient temperature. A semi-quantitative assessment was made of each colony type present.Representatives of all colony types were subcultured on to blood-agar-base plates under a serial code number and examined for production of inhibitors. In the initial studies on about 4000 isolates, the method used was a modification of a pyocine typing method (Darrell and Wahba, 1964) in which test strains were inoculated as a heavy streak across the diameter of a glass petri dish containing blood-agar-base medium. This plate was then incubated for 24 h, after which the growth was removed by scraping with a microscope slide and the plate was exposed to chloroform vapour for 10 min. After exposure to air to permit evaporation of residual chloroform, four indicator strains were inoculated at right angles to the site of the original culture. The plates were then reincubated for 24 h at 37°C. Antibiotic production was indicated by a failure of growth in the area from which the test strain had been removed.Another 1300 isolates were tested by a modification of thismethod in which, after removal of the test-isolate growth with a glass microscope slide, any remaining cells were spread over the surface of the medium with a cotton tipped swab ('Q-tip') moistened in broth. The plate was then exposed to ultraviolet light for 20 min. to kill remaining cells; four indicator strains were then applied as previously described.The indicator strains used were S. aureus PS80, the original epidemic strain which remains virulent for mice and is used for propagating phage 80; S. aureus PS 42E, the propagating strain for phage 42E; S. aureus Wood 46, the standard a-haemolysin producer; and Micrococcus luteus 27B, a wild-type isolate from a patient. RESULTSFrom 292 patients, 5282 strains were tested. Inhibitor producers active against S. ...