107 and is generally confined to counting error. Estimates of the (laboratory) within-sample variance, and the (field) within-station sampling variance have been made below; in summary, the errors were found to be variable but small enuugn to prevent yross misinterpretation of the data presented. A complete schematic of the procedural aspects of sampling and subsampling have been outlined in Figure 6. The samples collected in the field were subjected to two subsamplings which may have introduced errors into con- centration later associated with "whole sediment" concentrations. The first subsampling involved aliquoting portions of the whole dried sediments into containers for gamma spectroscopy or for homogenization (grinding). Since the aliquots of surface sediments taken for grinding were much smaller than the aliquots for core samples, an estimation of the errors introduced in the former aliquoting provides an upper limit to the errors that might be encountered in aliquoting the larger core subsamples. To estimate the variance introduced by aliquoting portions of dried surface sediments, the three major types of sediments (crater fires, naturally coarse grained, and mixtures of each} were subdivided and analyzed for 2394240 Pu. The results of these tests were expressed as coefficients of variation, as computed below, and are shown 5.D. where: S.D. = one standard deviation; X = mean The Jargest variance is shown by the mixture of fines and unpulverized Halimeda at 35%. This value is an upper limit since the test used aliquots tto to three times smaller (due to available sample) than those used for the actual samples of this consistency.