TWO STUDIES IN VARIABILITY FOR SOIL CONCENTRATIONS: WITH ALIOQUOT SIZE AND WITH DISTANCE P. G. Doctor and R. O. Gilbert Battelle Memorial Institute, Pacific Northwest Laboratory Richland, Washington ABSTRACT Two sources of variability encountered in radionuclide field studies include within-sample (between aliquot) variability and variability in concentrations over distance in the field. This paper describes the results of two studies conducted by the Nevada Applied Ecology Group (NAEG) to investigate these sources of variability on the Nevada Test Site (NTS). The first study reported here investigates the variability in 241 Am concentrations in <l0-mesh, ball-milled soil aliquots of size 1, 10, 25, 50, and 100 g drawn from a single, large composite sample collected at Nuclear Site (NS)-201 in a region where the average 24lam concentration is about 1.9 nCi/g dry soil in the surface 5 cm. The standard deviation (s) between aliquots withdrawn from this field sample was found to decline from a high of 1.52 nCi/g for the l-g size to 9.17 nCi/g for 100-g aliquots. The coefficients of variation also declined from 79% to 9% for these aliquot sizes. These results imply that many more l1-g aliquots are required to estimate the true mean concentration of an individual soil sample with specified precision than is required for, say, 100-g aliquots. The estimated median and geometric mean concentrations were found to increase with increasing aliquot size. This results from an observed decrease in skewness of the underlying distribution of aliquot concentrations as aliquot weight increases. However, the arithmetic mean concentration did not systematically increase with aliquot size, a result 405