109 repeated sampling would yield samples whose measured concentrations of radionuclides would be normally distributed about some mean concentration, the relative standard deviation of measurements from such a set of collections wan be used 2S a measure of the cverall variance due to both collection and analysis. The variance found for each radionuclide measured were expressed as coefficients of variation, and are shown in Table 18. Using the convention that underlined values for core sediments are positive (and others are negative) deviations from a coefficient of variation of zero, the mean and the standard deviation of the mean were computed. The mean (coefficient of varia- tion) computed was found to be positive (6.86), indicating that the coring method of sample collection may have tended to collect higher radionuclide concentrations in surface sediments. This may reflect the fact that the "surface sediments" taken by the grab samples were 2.54 cm deep collections, and the “surface sediment” sections of the core samples were 2.0 cm deep. The standard deviation of the values is 22.3 percent of the mean. Although only eight duplicate samples were available for comparison, it is of significance to note that the coefficients of variation found for plutonium are not greatly different from those found for other radionuclides. The results and discussion above were developed to attempt to make manageable a very complicated subject from the data available. These data show that while cumulative errors for sampling and analyses are about + 20% for surface sediments, the order of magnitude of these errors is relatively small compared to range of concentrations found across the lagoon. 6. SUMMARY AND CONCLUSION The areal distribution of radionuclides presented in Section 5 develop a basic model of the radiological contamination of the Bikini sedimentary