Dr. Don Hendricks operation. -7- August 28, 1972 I rather suspect that this is true only if the source of contamination is such that the concentrations can be regarded as having a smooth gradient in both directions from some maximum -- one might thus visualizethree dimensional plot of concentrations as a hill of some regular shape, so that evenly spaced sampling locations make plotting contours somewhat easier and efficient. However, in view of thehistory of contamination* at Eniwetok I should doubt that there are many spots where this kind of pattern can be expected. Instead one might expect that the pattern would be sufficiently heterogenous that a "grid" pattern would give results that Took much like a random pattern. Thus the main advantage to a systematic pattern would be ease of locating plots. This can be a distinct advantage, but there are also some further disadvantages. One that concerns me is the aforementioned evidence that the frequency distribution will be skewed. I suspect that it will be important to have the best data we can get as to the shape of that distribution and such information ought to come from a random set of sample points. The shape of the distribution becomes important, it seems to me, if it becomes necessary to make exact statements about the fraction of the area (or volume) that will exceed a specified concentration. That is, suppose it is decided that concentrations of more than x picocuries per gram must be ‘tleaned up". One then would like to be able to fit a theo- retical curve and ‘use it to make probability statements about the fraction of the area (volume) that exceeds x picocuries per gram. How one actually goes about that calls for some more discussion, a topic we'll return to below. For the present, I only want to make the point that random sampling is indicated. The actual process of drawing a random sample has been discussed with Ollie Lynch. We recommend that a rather fine grid ?*say, 50 foot intervals be laid down on a map and numbered. The actual locations should then be selected from a table of random numbers and plotted on the map. The order in which the sample points are located should be recorded, so that any Tast minute decisions to change sample size can be easily handled (i.e., by dropping the last samples selected; or by adding more if needed). The arid interval can be ected on practical grounds--i.e., how accurately it is assumed that a particular location can be found by pacing. However, it js wise to use a finer grid than such considerations may indicate, and it is essential to leave no ambiguity in instructions for field work--that is field crews should have exact instructions as to distances and directions and be told to sample at the