The unfavorable effect of intercurrent infection on diabetes has long been observed. Joslin1 drew attention to the presence of infection in nearly every case of coma in the pre-insulin era. At present, a diminution of the effectiveness of insulin during infection is common experience.2a Although many studies have been devoted to the effects of infection on diabetes, less attention has been given to the underlying cause of the apparent increased susceptibility of diabetic patients to infection. As the number of deaths from coma have decreased within the past decade, the relative incidence of deaths from various infections has increased. In one series,1 death from infection during diabetes has increased from 8.5 per cent in the Naunyn era to 26.8 per cent at present. Infection, not including tuberculosis and syphilis, was the cause of death in 126 of Warren's2b 300 fatal cases. He remarked that whereas arteriosclerosis