produces immediate surface runoff. A small fraction, perhaps less than 52, of the annual precipitation of 460 mm infiltrates the waste trenches. The actual distribution of the waste solutions within the four trenches is not known, For modeling purposes, it was assumed that the trench from which the core was collected received ten percent of the total discharged. TABLE 1 COMPOSITION OF EFFLUENTS Date Volume (gallons) 1945 - 1.4 x 107 1951 - 1.0 x 10" 1950 1952 Chemical Composition Plutonium Concentration 160 ppm F_ 60 c/m/ml 13 ppm NH "Cone" ammonium 7 x 103 c/m/ml citrate _ 200 ppm F 1953 - 1967 4.0 x 106 0.1M Nano | Trace Cores were taken at six inch intervals to a depth of 20 feet. Drilling was accomplished without water for lubrication or cooling. A portion of each core was packaged in small polyethylene bottles and shipped to Argonne National Laboratory. Each sample was ground in a new mortar which was discarded after use to prevent cross contamination, dried four hours at 110 C., and loaded into a sample holder. Sample holders were constructed from 2 mm thick cardboard squares 4 cm on a side with a 5/8 inch hole in the center. A single strip of cellophane tape on each side formed a thin, wafer-shaped cavity which contained 0.4 ml of powdered material. The average weight of each sample was 0.35 g. Several analytical methods for determining actinide content were considered before choosing low energy photon spectroscopy. Chemical separation followed by alpha proportional counting was rejected because of difficulty in dissolving samples to insure complete recovery of the actinides. This method, however, offers the greatest sensitivity. Neutron activation analysis was also tried and, although promising, offers no greater sensitivity than our chosen method. The spectrometer consists of a planar lithium drifted germanium detector 2 em@ by 5 mm thick. Its shallow design results in a low background from high energy photons. Its small diameter is a compromise between high resolution, 350 eV in the L x-ray region, and good geometric efficiency, about 8%. The entire detector-cryostat assembly was contained in a large 6 inch thick steel background shield with a graded x-ray liner. The resulting background in the L x-ray region is 0.27 c/m. A multichannel analyzer coupled to a computer controlled real time data acquisition system completed the spectrometer. In operation, all samples were counted for a minimum of 15 hours. Data processing Was accomplished by initial subtraction of a normalized background count from a sample spectrum. 2*!Am then waa determined by integrating under the 60 keV photopeak and that value used to normalize a standard “4!Am spectrum which in turn was subtracted from the sample spectrum. This step eliminated the Am x-ray contribution from that portion of the spectrum due to the Pu x-ray. 739Pu was then determined by summing the integrals of the three L x-rays. The spectrometer efficiency was calibrated using synthetic standards prepared in an tdentical manner to the samples. A laboratory simulation of this site was prepared primarily aa a test of the validity of our modeling techniques (fig. 2). A specimen of Los Alamos tuff taken near the site was cut into a cylindrical plug 18 x 77 mm. The curved surface was waxed to restrict lateral flow of liquid. The cylinder of rock was attached to a glasa tube which served as a reservoir for liquids which were to percolate the rock, Waste solutions were synthesized from 2"lam and 237pu tracer; the latter in place of 239pu to facilitate low level determination, 119