108 TABLE 17. Coefficients of variation of 23942405, concentrations measured in Set_ ho.= Samnte descrintion A subsamples of dried Bikini sediments. No. Aliquots Crater fines Aver. Wt. Aliquots 5 5.16 g 4 of above 5 aliquots B Unpulverized crater area sediments 5 C Mixture of fines and unpulverized Hal‘meda etc. , 5 C.V. 13% - 5.1% 15.7 g 2.8% 5.23 g 35.% The second subsampling was of the homogenized surface sediments for the chemical analysis and total alpha radioactivity measurements. The variance which would arise from this step was not addressed experimentally. However, Nelson and Noshkin (op. cit.) reported analyzing the 23942405, concentration in duplicate subsamples of nine “homogenized" coral samples taken at Eniwetok Atoll. Before aliquoting the duplicate samples, the sediments were oven- dried and pulverized in a ball mill. of the samples (35A,a) the ©297440 Pu upper limit. No sample sizes were given, and in one concentration was reported only as an The relative standard deviation of the remaining eight sets of duplicate analyses ranged from 0.0 to 21.4%. Standard deviations is 12.7%. The mean of the eight relative This variance is similar to the 13% found in the "Set A" data shown in Table 17. The in situ sampling error was indirectly addressed by a comparison of the concentrations of radionuclides which were measured in both surface sediments and the surface 2 cm of the sediment cores collected at the same station. The differences in the concentration of the various radionuclides measured in Sediments collected by the two techniques result, of course, from both sampling (in situ) and laboratory subsampling bias. However, by assuming that