Dr. Don Hendricks specified spots. -8- August 28, 1972 It should also be understood that there is no purpose in replacing or substituting sample locations. If there is some obstacle to sampling at a specified location (e.g., a concrete pad) then that fact should be recorded in the field and no sample taken. for this point. There are two grounds One, the heterogeniety of concentrations is such that shifting over a few feet prohibits calling the new location typical of the old, and, two, that obstacles to sampling are important in clean-up considerations -a chunk of concrete or a tree will offer problems, and the frequency of such problems ought to be estimated from the survey. While it is sometimes hard forfield workers to accept the policy of "no substitution", I think it is an important consideration and ought to be a hard and fast rule. If it seems desirable, some extra allowance in sample size can be made to get "enough" actual samp'tes, but the statistical evidence is simply "no sample point = no information" (or information of another kind, i.e., x% of the area has obstacles of a Specified kind). If there is reason to treat certain kinds of areas differently, then where possible, those areas ought to be set up in advance as specified sampling units. In particular, disturbed areas might fall into such a class. The above procedure may need to be cleared up in telephone conversations of other discussions. The essentials are to decide in advance what areas are to be treated as units, to set up maps of these units and to locate sample points at random on those maps. Where several islands are to be combined, the sampling should be at random over the group of islands, and not limited to a fixed number per island. If, however, a decision is made to fix the number of plots per island, we can handle this in the subsequent analysis. You will pay a price for this in«cthe sense of taking a somewhat larger sample than needed for statements about islands as a group {or by getting a wider confidence interval). Double-sampling The above proceeds as though wet-chemistry analyses for plutonium are to be made for all samples. Very likely this is not practical, and we are assuming that it may be possible to use a combination of wet chemistries and readings by a Ge/Li ("Gelly") detector. To do this within the limitations of statistically acceptable practice calls for some finesse, and constitutes the biggest source of worry for us in the entire survey. For one thing, there are some technical points we're not quite sure of, and these may take a fair bit of time and effort to sort out. Obviously there may be places where the Am/Pu ratio may be so variable as to make the technique unusable.