statistical significance. Thus, there was no evidence that occupational expo- sure to radiation caused a detectable non-specific shortening in the expectation of life. The finding should not be interpreted to mean that radiologists have an increased life expectancy because of their profession: actually, arti- facts due to the possible absence of particularly unhealthy individuals in the group considered, or to insufficient corrections for social class and occupations, or to inadequacies of the standards chosen for comparison were discuss-— ed as possible sources of the slight deficiency of deaths due to causes other than cancer. 343. As to cancer mortality, a significant excess was found among radiolo- gists entering practice before 1921, the excess being confirmed to tumours of the skin and pancreas (and possibly to leukaemia). No excess mortality from cancer was found in those entering radiology after 1921, although the time elapsed up to the completion of the study was insufficient to ensure that the cancer hazard had been totally expressed. Buh. Seltser and Sartwell [S38] returned again to the problem of mortality of American radiologists, in comparison with other medical specialists. Their new study covered the period 1935-1958, during which the mortality experience of 33.616 members of several United States medical specialty Societies was analysed, in order to test the hypothesis of a possible increase in mortality due to occupational radiation exposure. The Societies were selected in such a way that their members had a postulated high (radiologists) intermediate (internists in general) or low (ophtalmologists and otorinolaringologists) rate of exposure to radiation. The mortality experience in these groups conformed to the hypothesis and the median age at death was about 5 years greater among the lowest-exposure than the highest-exposure groups. The method followed for the comparison was to determine person-years of exposure, specific for age and calendar time, and to relate these to mortality. There were three periods along which comparisons were made: 1935-1944; 1945-195) and 1955-1958. 345. Comparison of matched and paired subjects for low- and high-exposure groups were also carried out and they gave results consistent with those of group comparisons. The differences in mortality increased with age but de- creased with calendar time for all except the oldest age classes. There was no excess mortality of radiologists in the 35-49 classes of age over the per- iod 1945-1958, suggesting that by that time the hazards had been controlled. xe