Single-element distributions of geochemical exploration data have been portrayed by and interpreted according to the parameters of the (log) normal distribution. Its estimators, arithmetic mean and standard deviation are sometimes used to define outliers. Practical treatment of empirical geochemical data (with examples presented from regional stream sediment data) suggests, however, that the statistical assumptions often fail to describe the real data behaviour. Few measurements or small populations, deviating extremely from the main body of data can influence the estimators significantly and at least partly mask the inherent information. To overcome these problems resulting from conventional classical statistics, Exploratory Data Analysis (EDA) with its graphical and numerical techniques, independent of distributional assumptions, was introduced and successfully applied to geochemical data derived from a regional stream sediment survey in Austria. Resistant selected order statistics (5-number summary) are explained and discussed for their usefulness in outlier detection as well as for the comparison of different data sets. Great emphasis is given to graphical data summaries using one-dimensional scatter plots, quantile plots, boxplots and density traces.

EDA is also useful in geochemical mapping. This affects especially class selection, presentation of outliers and regional distributions of single elements as well as variables derived from detailed multivariate statistical analysis. Major advantages of EDA are the straightforward application of its techniques and the easily interpretable results. In addition, the geochemical contrast on maps can be enhanced considerably, resulting in a more objective outlier definition as well as a better resolution of the regional background.