In geochemical prospecting for gold a major difficulty is that many values are below the chemical detection limit. Tracers for gold thus play an important role in the evaluation of multivariate geochemical data. In this case study we apply geostatistical methods presented in Wackernagel (1988) to multielement exploration data from a prospect near Limoges, France. The analysis relies upon a metallogenetic model by Bonnemaison and Marcoux (1987, 1990) describing auriferous mineralization in shear zones of the Limousin. The aim of geochemical exploration is to find deposits of raw materials. What is a deposit? It is a geological anomaly which has a significant average content of a given raw material and enough spatial extension to have economic value. The geological body denned by an anomaly is generally buried at a specific depth and may be detectable at the surface through indices. These indices, which we shall call superficial anomalies, are disposed in three manners: at isolated locations, along faults, and as dispersion halos. These two definitions of the word "anomaly" correspond to a vision of the geological phenomenon in its full continuity. Yet in exploration geochemistry only a discrete perception of the phenomenon is possible through samples taken along a regularly meshed grid. A superficial anomaly thus can be apprehended by one or several samples or it can escape the grip of the geochemist when it is located between the nodes of the mesh. A geochemical anomaly, in the strict sense, only exists at the nodes of the sampling grid and we shall distinguish between: a pointwise anomaly defined on a single sample, and a groupwise anomaly defined on several neighboring samples. This distinction is important both upstream, for the geological interpretation of geochemical measurements, and downstream, at the level of geostatistical manipulation of the data. It will condition an exploration strategy on the basis of the data representations used in this case study. A pointwise anomaly, i.e., a high, isolated value of the material being sought, will correspond either to a geological phenomenon of limited extent or to a well hidden deposit.