In choosing the final Subgroup sample sizes the statistician involved will need prior data on measurement variability and should be encouraged to apply U.S. Moss et al 2. (1969). Planning study of health effects. | There are several difficulties, tion, inherent in the situa- which this planning phase is designed to meet. Ologic methodology for countries the (e.g., in England, Japan). location of Epidemi- study of chronic disease has developed quite rapidly since 1950 in results such as the U.S. and a few other Such methods have not been pursued interest, however. And in addition, the specific conditions of this study are not well suited to epidemiologic field work. Our collaborator's and in "monitoring" interest leads, dotal medical records well in the health of unfortunately, residents to the sort of anec- known to result in serious bias, which can be dangerously misleading. It would be my hope that by focussing substantial energy and attention, funds, Gotal on planning a modern epidemiologic survey, these anec- collections can be either avoided or counterbalanced. An example study and some concerns choice of controls for a health (the control group in the PROPOSAL is adequate for a study of contamination levels) ning phase. Step one is the choice of parable community. Step two is k Moss, W.D. (1969). et al. 9009u3 I ii which can start in the plana "matched" or com- recommended to be mere Health Physics 17, 571-8.