QUAD and NEAR appear to result in contours with less bias than obtained using a polynomial fitting routine called TREND. It is suggested that Universal Kriging should be evaluated for its applicability for estimating plutonium contours and, possibly, inventory. INTRODUCTION Considerable effort has been expended in sampling soil, vegetation, small vertebrates, large vertebrates, and air for plutonium, americium, uranium, and other radionuclides in connection with the NAEG environmental sampling program on the Nevada Test Site (NTS) and the Tonopah Test Range (TTR). The results of this work to date have been presented primarily in the NAEG progress reports edited by Dunaway and White (1974), and White and Dunaway (1975), and the recent IAEA/ERDA Symposium on Transuranium Nuclides in the Environment.! would appear useful, however, if these data could be pulled together in a It concise manner so that the reader could obtain a better grasp of the total data picture to see how the soil results relate to those for vegetation, small mammals, etc. In this paper, we attempt the first step in such a synthesis by displaying many of the data collected at Area 13 (Project 57) on a single graph. This area was chosen for our initial effort primarily because of the cattle grazing study currently under way in Area 13 (Smith, 1974, 1975, 1976; Smith et al., 1976a). A limited number of data on plutonium concentrations in cattle tissue are available for comparison with the pelt, gastrointestinal (GI) tract, and carcass of small vertebrates as well as with soil and vegetation concentrations. While we are concerned entirely with Area 13 in this paper, there is an obvious need to graphically pull together the data from the other nine safety-shot sites being studied. The information plotted for Area 13 consists of simple arithmetic means (AM), Standard errors (SE) and ranges (maximum minus minimum observation) of the data. We suggest, however, that these summary statistics often do not adequately describe the underlying structure in the data (they may even hide it). This is due in part to the skewed (lognormal-type) distributions often observed in data sets. Consequently, we discuss some additional ways that data sets can be quickly summarized and displayed graphically that convey additional informa- tion. Different methods of estimating the "center" of a skewed distribution are considered relative to how they "weight" the largest observations in a data set. lHeld in San Francisco, November 17-21, 1975. 2Clean Slates 1, 2, and 3 on the TTR; Double Tracks on the Nellis Air Force Bombing and Gunnery Range; Area 5 (GMX Site) and Area 11 (Sites A, B, C, and D) on NTS. 238