after this period but before the end of December. During this period a 24-hour day, 7-day week counting schedule was maintained. The statistical distribution Distribution of Counts. of sample counts appeared to be of a logarithmic or log-log- arithmic nature. To further investigate the type of distri- bution two series of counts of 100 samples each of unashed post test Engebi sand were made. the oven and mixed. Sand ina jar was dried in Sampling cups for the two series held 6.1 + .05 mg (n « 100) and 2990 + 3 mg (n = 32), respectively. The small sample series was counted in the Nucleometer for 10 minutes per sample and the large samples in the end window counter for 1 minute each. The frequency distribution of the actual counts of the small samples was strongly skewed (Fig. 2, upper) but was approximately normal for the logar- ithms of the logarithms of these counts (Fig. 2, middle). For the large samples (Fig. 2, lower) the mode of the observed values was still to the left of the mean but the distribution was more nearly normal. It would appear that the distribu- tion of counts is strongly skewed to the left when the chance of occurrence of "speck" contamination (see Section 4.9.1) is small, but as the number of specks increases the distribution approaches the normal curve. For biological samples, espe- clally those with "surface" contamination, the distribution of counts could be expected to be similar to those of the sand samples. Unit of Measurement. The unit of measurement for re- cording radioactivity is disintegrations per minute per gram (d/m/g) of wet sample (unless otherwise noted) although -9- A]