In the method of extinction times as applied to disinfection of bacteria, the observational distinction is between samples which contain either none or one or more of active bacteria surviving after exposure to the disinfectant for given times. This proportion after any time is e-^ where ^ is the mean number of survivors after that particular exposure. The mean is expected to fall off sufficiently logarithmically with time. A rectilinear relation to time can thus be obtained by the loglog transformation Y = log(-logP). The loglog transformation differs from the probit transformation in not being symmetrical round P= 0.5. For the most informative observations P has the value 0.2032, but any observation where P lies between 0.2 and 0.3 will closely ap- proach this max. in the information it yields. A sample calculation is given and it is suggested that the place of the ED50 in probit analysis can be taken by the single mean survivor time, i.e., the time at which X = 1, Y = O, and P = 0.3679, in loglog analysis. The weighting coefficient is 90% of its max. value at this proportion.