this figure and assuming (a) that caesium-137 is trans- ferred from the diet of cattle to milk about ten times as readily as strontium-90 (ref. 24), and (b) that the average calcium content in milk is about 1-1 g/l., a soil factor for caesium-137 of about 0-03 can be derived from the value for strontium-90 calculated by equation (2) (Table 1). On this basis, the cumulative deposit of caesium-137 would on average have been expected to account for less than 1 per cent of the caesium-137 which has hitherto entered milk from world-wide fall-out. The analysis of survey results for past years could not be expected to reveal this component and at the present time the soil factor can thus be estimated only from laboratory experiments. The concentrations of caesium-137 in milk which would be predicted on the basis of equations (3) and (4) are compared in Table 3 for the two models of fall-out which have already been outlined; no allowance has been made for losses of caesium-137 from the soil by processes other than the decay of radioactivity because its movement in soil and absorption by crops are much smaller than those of strontium-90. Integrated over the first 10 years, equation (4) leads to predicted values 20-30 per cent higher than those given by equation (3) for both models; over 50 years under conditions of decreasing fall-out the inte- grated value given by equation (4) is 50 per cent higher. Over this period the soil would contribute about 20 per cent of the total. Conclusions Evidence reviewed in this article indicates that the annual average level of strontium-90 in milk in the United Kingdom is influenced to an appreciable extent by the rate of fall-out in the last half of the previous year as well as by the current rate of fall-out and cumulative deposit. An improved basis for predicting future levels of strontium-90 in milk is therefore provided by an equation which includes a proportionality factor for this ‘lag-rate’ effect in addition to the rate and soil factors used in former assessments, equation (2). The same type of equation is applicable to caesium-137 (equation 4) though with that nuclide the contribution from the cumulative deposit has been so small that it can be ignored under conditions of fall-out experienced hitherto. The levels of both nuclides in milk which would be predicted under conditions of fall-out which may be expected during the next decade as a result of weapon testing are not greatly altered when allowance is made for the lag-rate effect. In any year the greatest difference is unlikely to exceed about 20 per cent and averaged over 10 years predictions for strontium-90 would be altered by a smaller amount. This degree of agreement between different methods for prediction encourages confidence 8