ve 2 _— Cancer Deaths among Radiation Workers theyare localized it would bedifficult to relate them to the uniform whole body doses experienced occupationally. Furthermore, individuals would be reluctant to cooperate in a survey which collected information which could prejudice their employment prospects. Medical exposures have to be considered as background and we must presume some cancellation since they are equally probable in exposed and control groups. Overa very long surveytime, the presumed cancel- lation of all backgroundeffects becomes more acceptable. 4. Results and discussion : In table 2 we have shown the numberof radiation-induced cancer deaths that would occur in the population (columns A and B) if the risk was 1074 per rad distributed in time as Aj, (column I) or A,, (column G). The longer latent period permits a greater influence of the normal death rate and results in a smaller number of radiation-attributed deaths. Since there is no striking distinction to be drawn between the effects of using Agg, Byy or Cy we have not reproduced the details here. Corresponding to the 7-7 total radiation-induced cancer deaths in column If for Aj, we calculated 8-23 for By) and 9-11 for C,,. The inerease through Ay», By) and C,, is due to the concentration of the risk into a shorter period hence allowing radiation-induced death instead of ‘natural’ death slightly more often. Comparing columns [2 and If we see that the 7:7 radiation cancer deaths are against a background of 1406 other cancer deaths andthis is a clear indication of the detection difficulties to be faced. A 30-year latent period is appropriate for leukaemia while death from all other forms of cancer may occur up to 50 years following exposure. Thus to find the number of radiation-induced leukaemia deaths expected we should seale the figures in column I using L (leukaemia) = 3-0 x 10-5 (see section 2.3), For example, how many deaths from leukaemias compared to all-cancers would we expect in a particular industry employing 3000 radiation workers cach receiving an avcrage dose of 0-5 rad/year? Using eqn (1) and table 2, we predict that the number of leukagmia deaths per year is 0:03-4 (distributed among 3000 workers and 5149 ex-workers) compared with 0-93 expected naturally; correspondingly there would be 0-08 cancer deaths compared with 42 naturally. Thus in a period of 25 years we would expect in the industry 24 leukacmia deaths (3 in-service) of which 0-8 (0-3 in-service) would be radiation-induced; in the same. time period we would expect 1056 (105 in-service) cancer deaths of which 2 (0-6 in-service) would be radiation-induced. Probably the most significant feature of table 2 is the balance of deaths between in-service and ex-workers. Any long time effects will be lost unless adequate provision for follow-up exists. For Agy risk 62% of all radiationinduced deaths will be among ex-workers. ‘Lable 3 shows all the quantitics described in table 2 but for a leaving rate of 10% instead of 5°. Comparison of the corresponding columns in tables 2 and 3 show the same number of age-specific radiation-induced deaths but distributed more heavily towards the cx-workers in the 10% case; for Ag, 81%