The average concentration of 239-2405, in A Site over all three strata is 11.0 + 8.07 pCi/g (X + S.E.) which is close to the 21 + 6.6 (1975, Table 16), pCi/g reported in Gilbert et al. for the <5,000 cpm The plutoregion in Area 11, of which A Site is a part (Fig. 1). nium levels in A Site are low compared with those at B, C, and D Sites in Area 11 (Table 16, Gilbert et aZ., vegetation mean to soil mean for 1). 1975). The ratio of 239-2405, in A Site is 0.095 (Table If one considers only the 14 paired soil, vegetation Pu concen- trations, the ratio of vegetation mean to soil mean is 0.064 + 0.096, which is similar to that reported in Gilbert et al. for Area 1l. (1975, Table 17) The paired vegetation-soil data for 2354 and 2381 for all strata combined are plotted in Figs. 3 and 4. The estimated linear correlation between soil and vegetation is 0.93 for 2354 but only 0.35 for 2387 The higher correlation for 2354 could be related to the preponderance of this isotope in the device, whereas the 2385 present is primarily natural uranium. Three models have been fit to these data in Figs. 3 and 4. 1. oy, = 8x, +e; A n n , where 8, = on Yi! de Ky 2. yy = Box, + ey , where 8, = Pa (y,/x,)/n 3. y, = By + 8x, +e; , A n and the usual linear regression model (see Snedecor and Cochran, 1967, p. 167, for discussions on Models 1 and 2). inappropriate for both 2354 and 2381 It is clear that Model 2 is that Model 1 is considerably better, but that Model 3 is the best of the three for 2351, it too being inadequate for 238y. The poor fits using the commonly used ratio estimates 8, and 85 suggest that calculation of such ratios without first plotting the data and checking for adequate fits can give misleading results. Gilbert et al. (1975) report similar poor fits using By and B5 for 239-2405, concentrations in adjacent soil and vegetation samples at the other safety-shot sites. Model 3 was also fit to the logs of the soil and vegetation pairs, but the 128