Twenty-four adjacent standard NAEG surface soil samples (5-in. diameter ring to a depth of 5 cm) were collected on a 6 by 4 grid (see diagram, Each sample was Appendix A) approximately 150 ft north of GZ at NS-201. The <10-mesh dried, ball-milled, and sieved through a 10-mesh screen. fractions from all samples were composited (pooled) and mixed by kneading for approximately 10 minutes. The aliquoting procedure for the <10-mesh fraction proceeded as follows: the entire mixed sample was divided into four quarters, each quarter divided into fifths, and each fifth placed in a separate bottle. This gives a total of 20 bottles. was taken which the according described One aliquot each 1, 10, 25, 50, and 100 g from each bottle for a total of 100 aliquots. The order in different size aliquots were taken was systematically varied to a "latin-square" design (Federer, 1963, Chapter 6) as in Table Al in Appendix A. This was done to mitigate any confounding of concentration for a particular aliquot size with the order in which that size aliquot was drawn from the bottle. Each aliquot was Ge(Li) counted for Am by REECo* personnel at Mercury, Nevada. Counting times were 1,000 minutes or until the counting error (20 level) was <10%. Actual counting errors were about 7%. Special samples were also collected in the vicinity of NS-201 for the purpose of calibrating the Ge(Li) counting system for the different size aliquots. Details are given in Appendix A. The >10-mesh fraction was handled similarly to the <10-mesh fraction except that only 10 ten-gram aliquots were analyzed for comparison with ten-gram aliquots of the <10-mesh soil. The average Am concentration for the >10-mesh aliquots were 0.00748 + 0.00213 nCi/g (x + standard error). This mean is about three orders of magnitude less than the mean obtained for the <10-mesh soil (see next section). RESULTS The Am concentrations in nCi/g for the five aliquot sizes are given in Table 1 in the form of "stem-and-leaf" displays. Figure 1 is a plot of the data that also gives the median, arithmetic mean (x), geometric mean (GM), standard deviation (s), standard error (SE), and coefficient of variation (c) of the 20 aliquots for each aliquot size. In Figure 1, the two outer solid lines delineate the ranges of the observed data and should not be construed as confidence limits. The center line connects the arithmetic mean for each aliquot size. The stars locate the medians and the parentheses denote the mid 50% (interquartile range) of each data set. It is clear that the variability between aliquots is highest for the l-~g Size and decreases as the aliquot size increases. The data are skewed *Reynolds Electrical and Engineering Co. 409