value for m, among other things, depends on the within-laboratory variability (variability between replicate analyses) and the size of the actual difference between labs. Gilbert e¢ al. (1975) conclude that for soil samples from the safety-shot sites, the use of only three replicate analyses per sample by each laboratory is not likely They suggest that the ro detect even large differences between labs. use of less than 10 replicates might be largely a waste of time and money if one wants a fairly good chance of detecting even a ratio of Since environmental vegetation 2? between the high and low lab. samples are presumed to be less affected by the hot particle problem (the vegetation sample is dissolved, Major e¢ al., 1974), the within- lab variability should be less than for soil samples, allowing the use of fewer replicate analyses per sample. Some data is now becoming available for estimating this within-lab variability for vegetation. Hopefully, the hot particle problem in soil will not be as great at the nuclear study sites, allowing for a smaller number m of replicate analyses per lab. Since "standard" samples are prepared so that all aliquots should have the same plutonium concentration, an m of 3 per laboratory for these samples may not be unreasonable, although we have yet to examine the question with actual data. 113