The application of the probit method to toxicity test data adjusted for mortality in the controls

When data from toxicity tests have to be adjusted for a mortality rate amongst untreated controls, three modifications to the usual method of probit analysis need to be considered. The full solution is more laborious than that ordinarily used, as it requires the estimation of this mortality rate from the whole of the data. When the rate is. not high (e.g. up to 20 %) a satisfactory approximation may be obtained by a simple alteration in die weighting coefficients, especially if the precaution is taken of using two or three times as many test organisms as controls as for any dosage of the poison, thus decreasing the relative value of the information on natural mortality contributed by the treated batches. The modified weighting coefficients are tabulated for values of the natural mortality from zero to 20 %. An example of the more complex calculations required for the full maximum likelihood solution is discussed in detail.