radiological follow-up and, in the case of cancer diagnosis, of consensus opinion of pathologists. The Appendix is provided to allow others to perform different analyses of the data, recognizing that the data base is incomplete. Verifying the data over the last seven years has resulted in changes in age, identification number, assigned dose, and diagnosis. Several independent groups reported age at exposure, and the Adams et al. (6) version was used here. Different ages at exposure influences the age distribution of cancers, which in turn impacts strongly on the risk coefficient for a given age group. The external thyroid dose was due to gamma exposure from the fallout cloud and fallout on the ground, and was taken as equal to the external whole-body dose reported by Lessard et al. (3), i.e., 190 rad at Rongelap, 110 rad at Ailingnae, and Ll rad at Utirik. These external doses were estimated for a point which was 1 meter above the ground, thus some variation in external thyroid dose with a person's height may have occurred. To a first approximation external thyroid dose is inversly proportional to height above the ground. We derived this proportionality by neglecting photon attenuation and buildup, and by limiting the height above ground to between 0.5 and 1.5 meters. The impact on the risk coefficient estimates, relative to assuming that external thyroid dose was height dependent, was minimal, since the person-rad from external exposure was much much lesa than the person-rad from {nternal exposure. The data for the unexposed comparison groups are indicated in Table 1. In the age- and sex-matched comparison group used for this study, two papillary carcinomas have been observed. The summary is completed through 1983. To apply the data for risk coefficient determination, we modified the matched group results by the ratio of 31/29, which corrects for the difference in the number of reported observation years. The larger, less defined comparison population studied by Conard et al. (7) is shown in the first half of Table l to show that spontaneous cancer risk is not a strong function of group age for the Marshallese people. The comparison data indicated a spontaneous rate of 3x10 cancers per person-rad-years at risk. A lower spontaneous rate has been reported for the U.S. population, Lxl0 * per person per year (2). The Marshallese comparison data were used in the risk coefficient computations made here. A summary of data in the Appendix appears in Tables 2 through 4. Note that out of 9 papillary cancers listed in the Appendix, only 2 were observed in males. This male to female ratio is similar to that reported in other studies (1,2,8). Tables 2 through 4 contain the tnput data which we used with Eq. (1). The data were grouped in the same manner as in other reports dealing with cancer and radiation exposure of the thyroid. The age groups were the Same as that used by Conard et al. (7) and Adams et al. (6). To determine the average years post-exposure to onset of carcinoma, we set onset of carcinoma as the time of clinical observation of a thyroid nodule; thus, a latent period was assumed, but a period of several years could have elapsed before a nodule became large enough for detection by routine palpacion by che physician. Therefore, the true latent period could be shorter than that assumed here. Tables 2 through 4 include the expected carcinomas, computed from the age- and sex-matched comparison group, and a summary of the total person-rad from manmade internal and external sources. 23