1. Whole cultures of Bact. coli grown under carefully controlled conditions have been subjected to the action of heat at temperatures ranging from 47 to 55° C. and the survivor curves determined.

2. The occurrence of excessive variation between replicate plates of counts made when the mortality exceeds 95 % was very much less evident than when phenol was the lethal agent.

3. At temperatures of 35° C. and below, disinfection of the cultures was never complete, as a permanent population of cells became established. In some cases the numbers of heat-resistant cells reached nearly 0·01 % of the original population, but great fluctuations were observed.

4. The death-rate during the active part of the disinfection was not constant but, in general, increased with time. In the faster disinfections this increase was difficult to detect and the logarithmic death-rate appeared to be virtually constant, but the increase is nevertheless believed to have been present in all cases.

5. The data for all the experiments have been combined to give a composite disinfection curve from which it was concluded that the death-rate was low at first but rose to a maximum at which it remained constant until the mortality had reached at least 99·99%.

6. This finding led to the decision to treat all the experiments as if the regression of log survivors on time had been linear between mortalities of 90 and 99·99%. The standard errors of the calculated regression coefficients were small, so that this method of treating the experimental data appears to be justified.

The authors wish to express their gratitude to Messrs I.C.I. (Pharmaceuticals) Ltd. for their valuable financial assistance, which has enabled one of us (H. E. F. D.) to devote his full time to this work.