that their means were equal except for 2:31Pa where the F ratio was barely significant.Additional data were obtained on precision by applying t h e procedure t o 500-ml samples of a uranium mill process waste known to contain about 3 mg of cerium and four times the 100-pg limit for the electrodeposition of 2,'3nTh. This sample was chosen to represent one of the worst cases for t,he determination of thorium. T h e results of the five replicate analyses are summarized in Table V, and an alpha spectrum of the thorium fraction is shown in Figure 4. Alpha spectra were obtained under the conditions described by Sill and Olson (31 ) to minimize recoil at~om contamination of the solid state detector. T h e counting efficiency for the ""Ac fractions was reduced to correct for the additional precipitate formed from cerium present in the sample. Thorium recoveries were determined with ','j4Th tracer and ranged from only 8 t o 25% because of large losses in the electrodeposition.In spite of the large thorium losses. the relative deviation of an individual result from the mean was only 4')/0 for r:ioTh, but 9, 16, and 5% for 2:iaTh, ?.STh, and "25Th, respectively. A major factor in the variance of the ?;'"Th results was the large correct,ion applied for the contribution of degraded ':ioTh alpha particles in the ';"'Th energy region. T h e relative uncertainties of just the numbers of counts were approximately 8 to 12% and 2 to 4% in the ' "Th and 22TTh energy regions, respectively. T h e wide variance of the results for '2,'i1Pa was due to its large and erratic losses on silica from the process waste sample and also in the electrodeposition. In such cases, thorium recoveries are not reliable estimates of ',jlPa recoveries. and the results were included in Table V merely to emphasize this point. T h e relative deviations of individual results about (31) C. W . Sill and D. G. Olson, Anal. Chem. , 42, 1596 (1970) the means for 2 2 7 A~, 228Ra, and 2zsRa were 2 to 4%, about the same as the precision for 2 2 7 A~ and 22sRa in the series of pitchblende ore analyses.Sensitiuity. T h e following detection limits for 500-ml liquid and 100-m" air samples were calculated for paired observations (32) and are merely intended t o illustrate the sensitivities of the procedure under the conditions specified.T h e blank activity for the determination of 2zsRa by either gross-alpha counting' or the emanation method is about 25 cph, which results in detection limits for gross counting and emanation counting, respectively, of 1 X 10-9 and 3 X 10-lo pCi/ml for liquid samples and 4.5 X 10-15 and 1.4 X pCi/ml for air samples. T h e 228Ra blank of about 195 counts per 50 minutes is equivalent to a detection limit of 2.5 X 10-9 pCi/ml for a liquid sample and 1.2 X pCi/ml for a n air sample. However, decay of the 6.13-hour 228Ac used for the indirect determination of 2L'xRa results in a practical detection limit 1.5 to 2 times these values. T h e blank for ";Ac is about 7 cph, and the detection limit is 6 X pCi/ml for air samples. Gross alpha blank...