Number of Samples In using a random selection technique for choosing the sampling locations, the controlling factor is variability within samples and between samples, and not the number of samples per unit area. If the surface contamination is assumed to be uniform, to some degree, then the same concentration results would be obtained if a certain fixed number of samples were chosen at random as would be obtained if a certain number of samples per unit area were collected, regardless of the size of the area sampled. Obviously, the number of samples collected on a per-unit-~-area basis would be small for small areas and large for large areas. This could be too small for the small areas to be statisti- cally meaningful, and too large for large areas to be economically collectible. With a random sampling scheme, the precision, dependent on variability and the number of samples, is chosen for the area being sampled. This sets the minimum number of samples, regardless of the size of the area. Thus, the precision and the number of samples are predetermined and théir ratio maximized. For the random sampling portion of the soils program, the minimum number of sampling locations per island was fixed at five, even on the smallest island. There was an overall trend to develop the number of samples on a per-unit~area basis, but the actual numbers were optimized according to the desired precision based on an assumed coefficient of variation. This coeffi- cient was developed by experience and review of soils data from several different sources. Experience has shown that the coefficient of variation is a relatively constant value from group to group for radioactivity in soils |wever, as this was used as an assumption for the basis of the soil collection technique, this would have to be verified by the data for the effort to have merit.