Dr. Don Hendricks -9- August 28, 1972 Probably you can roughly identify those places in advance and proceed accordinjy (i.e., more wet chemistries). be as follows: {1) Determine what areas are to be treated as single sampling units (groups of islands, etc.). plots and collect the soils. tor on all samples. Beyond that, the procedure should (4) (3) (2) Take a large random sample of soil Determine Am concentrations by Ge/Li detec- Take a random subsample of all soil samples for the specified area and do wet chemistries on these. Our present thinking is that this sample should be at least 30 wet chemistries for each area. That is, if exact statements are to be made for some group of islands, or large island (or single area), these should each he based on at Teast 30 wet chemis- ties. I should expect that the southern islands (Phase I) could very likely be combined for one set of wet chemistries. However, if the soil samples are all kept and are available, then it should be possible to provide some modifications of the scheme as the results of the first set of wet chemistries become available--i.e., we start out with one set of analyses and do more as study of the data indicates. Ollie provided us with some advance data on the Phase IIA islands that indicates that the method should work fairly well there (high correlation between Am and Pu). Our recommendation is to take rather more samples for the "“Gelly" scans than the above criteria may indicate, since wehave based these rules roughly on wet chemistries ~- adding the double-sampling scheme will increase the variability. However, it should cut processing costs a good deal, and it seems prudent to take more samples than we think are needed and discard them if it turns out we don't need them--its much easier to collect them in the first pass than to send crews back again: Additional points The question of finding “burial grounds" by sampling has been mentioned. I should think that about al] we can say in general is that one could guess as to the size of a burial ground (width of a bulldozer blade, guess as to length) and calculate that finding one by blind search is about equivalent to the needle-in-a-haystack problem. No doubt there may be some surface clues, etc., which one could use to devise a scheme, but more information and some -