Outlines a set of techniques that enable a researcher to discuss the "hidden structure" of large data bases. These techniques use proximities, measures which indicate how similar or different objects are, to find a configuration of points which reflects the structure in the data.