a plastic ring and disc. In each case the weight of the wet sample was estimated to the nearest 10th gram, on a Scripps electronic shipboard balance. The potassium activity of plankton is at most 5 d/min/g of wet material and should be considerably lower. No correction was made for this natural activity. One-half of the original sample was retained by Scripps for biological examination and one-quarter was taken by the University of Washington for activity measurement in their laboratories, Analyses of Fish Samples Fish samples were identified and dissected when necessary by the marine biologist. Samples were ashed at 600°C in nickel crucibles and weighed portions of the ash taken for analysis. The ash was slurried with about 500 ml of dilute sodium carbonate solution in an attempt to remove the natural potassium activity and the residue filtered off on fibre glass. The samples were covered with Mylar and mounted on plastic rings and discs for counting. When treated samples were analyzed by HASL by beta absorption techniques, it was apparent that the major portion of the activity was due to the natural potassium in most cases, This indicated poor separation by the method described and a number of samples were reanalyzed by dissolving the ash and reprecipitating carbonates. In addition, several samples were analyzed for individual fission products at the HASL laboratory. Samples of all reef fish, flying fish and other biological specimens were taken back to the Applied Fisheries Laboratory for activity measurement. There, individual tissues were ashed and counted, Counting Procedures The beta counters described had an average background count of 18 to 20 cts/min. The procedure on all samples consisted of running alternate periods of 30 min on background and 60 min on the sample. The average background for the period before and after the sample reading was subtracted from the sample count to give the net count. The efficiency of the counters was measured with half-gram samples of potassium carbonate mounted and measured under the same conditions as the sample ex‘cept that a 10-min counting period was used. The K49 activity was taken as 492 d/min/half-gram. This standardization is reasonable for mixed fission products, since the average energies of K#9 and of MFP are very close. The average efficiency of the counters was 20 percent. The counting statistics of all samples of one type are roughly equivalent. For water samples, the Poisson standard deviation is one count per minute per sample which becomes 25 d/min/liter. Plankton and fish samples were weighed, and in most cases, one gram samples were taken. This gives a standard deviation of 5 d/min/g unless stated otherwise in the results. The only possibility of improving the counting statistics is the development of low background counters. It is not possible to handle larger samples of material nor is a counting time greater than one hour practical for field use. If a 2-in. diameter Geiger tube with a background of 10 cts/min were available, the standard deviation on a water sample would be reduced to 15 d/min/liter while a 5-cts/min background would reduce the standard deviation to 10 d/min/liter. It is apparent that even a marked background improvement does not eliminate the problems of low level counting. A reliable estimate is obtained only by combining results ona relatively large number of samples, 4 ye