81 tissues" were not. Since the "t" test is weakened by non- normality of sample distributions and differences between the sample variances, non-parametric methods were employed to determine whether these factors could control the decisions based on the "t" tests. A test suggested by White (1952), based on rank, was employed to test the significance of differences "between areas" where the sample sizes were unequal. "between tissues-within areas" test, pairs, signed ranks test was used. For the the Wilcoxen (1945) matched: The results of both paramet- ric and non-parametric tests are summarized in Table ll. The correspondence of both methods strongly suggests that the specif activity of zinc-65 differed significantly in both eyes and live: between the Kabelle and Rongelap samples, but not between the ey: and livers of fish from the same location. The graphical plot of the descriptive statistics in Fig. and the tests of significance in Table 11 indicate that the ratios of zinc-65 to stable zinc in goatfish tissues in 1959 were significantly higher at Kabelle than at Rongelap. provides further evidence that effective mixing of This fallout radiozine in the waters of the lagoon had not occurred between the time of the initial contamination of the atoll and the time the samples were collected. Inspection of the levels of total zinc in the eyes of the Rongelap sample of 20 goatfish (Table 4) suggested a negative 6

Select target paragraph3