samples are specially prepared in the laboratory so that all aliquots from a given sample are known to have the same radioisotope concentration. The purpose of both kinds of interlaboratory comparison samples is basically the same; namely, to see if all participating laboratories obtain analytical results that are "alike" to some acceptable degree for a given sample. If there are large, (biases) between laboratories, consistent differences then the use of "standard" samples is, in general, the better of the two methods to detect and estimate this bias. This is primarily due to the homogeneity of the "standard" soil or vegetation relative to the greater variability that can occur between aliquots of the same ''cross-check" sample due, perhaps, to the presence of hot plutonium particles or some other unknown environmental factor(s). The great variability between aliquots of the same environmental sample tends to obscure differences between labs. ing an interlaboratory comparison study, (1) the number of samples (n) to the laboratories and (2) Hence, when design- two important aspects are that need to be aliquoted and sent out the number of replicate determinations (m) that should be performed at each laboratory on each of the n aliquots. As concerns cross-check (environmental) reasonable to use enough samples (n) samples, it appears in order that a broad range of concentrations and field conditions (e.g., distance from ground zero, desert pavement as well as blow-sand mound samples, etc.) can be represented. This can be achieved by selecting cross-check samples according to some randomization scheme from the total number of field samples collected. The percentage of field samples chosen to be cross-check samples will depend on cost factors and the total number of field samples collected. Concerning "standard" samples, n should be large enough that a broad range of concentrations could be repre- sented in order to detect biases that may be dependent on concentration level (e.g., some labs could be biased on samples with large plutonium concentrations, but not for lower concentration samples). Gilbert et al. (1975) and Eberhardt and Gilbert (1972) discuss the selection of the number m of replicate analyses on cross-check samples that should be used to detect differences between labs. 112 The optimum