Discussion of Statistical Basis Assumptions Several assumptions were used as the basis for developing the statistical approach for the solls erfort. As in any real situation, however, assump- tions hold only partially, and this would certainly be true for Eniwetok. First, it was assumed that, over the majority of the islands, the area sampled was rather unformally contaminated as far as surface distribution of radionuclides. Actually, activity levels would probably approximate a log normal distribution if the contamination was examined on a concentration versus frequency basis. However, over the area of any individual island any particu- lar activity level would be distributed with equal probability. bably true for the Phase I islands, even ELMER“is FRED. This is pro- For the Phase II islands on which there were no SGZ's, the assumption is approximate, and thus modified. Phase III islands, with SGZ's and evidence of construction activi- ties, presented some problems with this assumption, but, with appropriate modifications and careful interpretation for at least portions of these islands, the assumption could be approximate. The island of YVONNE, the single Phase IV island, was the exception. Uniformity was out of the question for the northern half. The southern half, on the other hand, was treated as uniform after considering previous survey data. The modification to save the uniformity assumption was to make another assumption. If the island's contamination was not uniform it was distributed on the surface with a slowly changing gradient. fication into assumed uniform surface groups. the Phase II, III and IV islands. This would then lead to stratiThis modification was applied to The data would, of course, verify or reject Cy these assumptions.