The Am data suggest that the optimum number of clusters, samples within clusters, and aliquots per sample for estimating average concentrations will be different for areas near GZ, where concentration levels change rapidly as opposed to lower-level regions such as along Line 2. ACKNOWLEDGMENTS We would like to thank Ed Essington for his helpful suggestions concerning the design of these two studies. We also greatly appreciate the very competent performance of the REECo field sampling crews and members of the REECo counting laboratory. Special thanks also go to B. B. Vinson for her assistance in data reduction and analysis. Finally, we acknowledge Mary White for her support and interest in these studies. 436