nature of the radiation exposure between the Japanese survivors and in the United States radiologists. Since the latter hypothesis would be against a massive body of evidence collected on other mammals, one should logically be inclined to favour the first interpetation, particularly in view of the much more reliable experience from Japan, which is now based on about 20.000 deaths among about 80.000 A-bomb survivors with known T-65 doses. 37T.- Mention should also be made of an up-to-date report concerning the mor- tality experience of children exposed in utero to the A-bombs 203 deaths among 1923 subjects during 1945-1976. with dose in both cities. [K21]. There were The mortality ratio increased The increase was linear with dose in children dying within the first year of exposure; but there was no increase between 1 and 9 years and only the suggestion of a further increase after 10 years of age. The excess mortality was significant only for children exposed during the third trimester of pregnancy, but loss of the embryos and foetuses might have been present in an unknown percentage of pregnancies before term. Regarding the cau- ses of death, for 55 of 203 children no information was available and the sample Size is yet too small and too young for any conclusion about a possible excess death from non-specific causes. D. 378. CONCLUSIONS The data on occupationally-—exposed groups of workers do not lend them- selves to complete dose-effect analysis and, in the absence of precise dose evaluations, conclusions must rely on comparisons between groups of exposed and non-exposed individuals. Under these circumstances, the homogeneity be- tween the control and the test samples is critical because in many series the effects are marginal and their statistical significance may depend on the choise of controls. 379. The data on radiologists leave no doubt that particularly in the early days of radiology leukaemia and cancer were indeed induced in these persons. Observations in favour of this conclusion have been confirmed in all studies [D9, W2, W3, C27, S38, wh, M28, M29]. However, in some instances a higher incidence of neoplastic conditions was not accompanied by an increased death rate and shortening of life [D9, C27], while in others [W2, W3, S38, Wh] there was a true loss of life amounting to 5 to 6 years, following exposure for the