This data set is clearly asymmetric with a very long tail as a result of the This stem-and-leaf display gives the same information as extreme datum 305. the histogram in Figure 6, but in addition retains the actual numerical values of the data for use in latter calculations. The depth column, which counts the number of data points from each end, is useful for computing summary statistics such as the minimum, 25th percentile, median (50th percentile), 75th percentile, and the maximum. For this data set we find = = = = = minimum 25th percentile median 75th percentile maximum 0.5 3.5 7.6 11.0 305 [see Conover (1971) for discussions of the median and other percentiles]. These statistics can be displayed as a "box" plot as follows (the AM is also shown for comparison): ARITHMETIC MEAN MEDIAN N = 47 yo 7/ *K * 0 2 4 6 8 e 10 e 12 e 14 e 16 e e 304 306 The box area contains the middle 50% of the data (called the interquartile range) and tells us where the bulk of the data lies. Taken as a whole, the plot above gives us information on the center of the distribution, its variability, symmetry (or asymmetry), and the occurrence of outlying values. The stem-and-leaf display in conjunction with this summary plot conveys additional information over that provided by the AM and SE. Additional information such as the GM and other percentiles could also be displayed on the above plot if desired. To further illustrate the usefulness of the above approach, we display box plots in Figure 7 for the rodent (Dtpodomys microps) data that were summarized in Figure 3. The AM, GM, SE, and range are displayed in column [A] and the box plots in column [B]. The GMs are seen to be considerably smaller than the AMs, and slightly less than the medians. This was also true for the soil data in Figure 6. The tendency of the GM to be smaller than the median for these data illustrates our feeling that the GM may not give sufficient weight to the larger observations, at least if we take a conservative approach of not wanting to underestimate "average" concentrations. The interquartile ranges in Figure 7 give us additional information on the possible downward shift in Pu concentrations in the GI tract relative to those in pelt. Since the number of samples (n = 11) is so small for these three data sets, a stem-and-leaf display does not give as much information as we would like and the estimates of the interquartile range and other parameters are not precise. It is still useful, however, to go through the procedure if for no other reason than it requires a look at each individual datum to see how it relates set s 258 to the other data in the