SLIDE 10 Analysis of samples. Soil samples were analyzed for 127 t, 129 25 149r, !45sp, 1376, , 1355p, , and 60¢9, The methods used for neutron activation analysis were described by Brauer (Br74) and Keisch (Ke65). JTodine was separated from soils according to the method of Studier (S$t62). Once separated, the iodine was irradiated with neutrons in a nuclear reactor, purified to reduce levels of interfering nuclides and then measured by gamma spectroscopy (Br80). In order to perform quality control, comparison samples containing known amounts of 125, 127, and 1295 were irradiated with each set of fodine samples isolated from Marshall Islands' soil. Following irradiation the fodine samples were further purified and then precipitated onto the detecting media. The number of atoms of activated nuclide were determined by gamma spectroscopy measurements of the radioactivity produced in the soil sample and in a comparison sample. The number of initial comparison atoms and resulting comparison activity were used to determine a production ratio. The production ratio was applied to the soil sample activity and the number of atoms of activated nuclide per gram of soil was estimated. Corrections to the soil activity were made based on results for soil sample blanks, comparison sample blanks and method yield. Individual counting errors were normally less than 5% although a few samples approached 20%. A least squares fitting was performed on results for nuclide soil activity per unit soil mass vs days post detonation using linear, exponential, logarithmic and power function models. Sample results varied from their best fit value by as much as a factor of 9 and by an average factor of 2.5 over the period 1955 to 1977. The best fitting function was determined based on a comparison, ef the coefficient of determination for each nod¢ hy Functions used with I results for Rongelap soil are plotted here. For I soil results, the best function was exponential. All four fitting functions were generally predicting soil activity per gram at times after 600 days post BRAVO nuclides. Significant departure between functions occurs during the fitting useful in for all period several hours out to one year post BRAVO. For example, at 0.5 days the difference between ghe exponential and power function estimates spans 5 orders of magnitude for 12 I. -