For drainage sediment to be used as a successful sampling medium in reconnaissance geochemical surveys, the length of any dispersion train from a typical target must be great enough for it to be detected by a widely spaced sampling pattern, and must also be homogeneous enough to avoid it being missed by the unfortunate selection of a random low value sample. Elements which occur in stream sediments in the form of discrete heavy minerals present two special problems in this regard. Since the element is concentrated into a few particles, the medium is extremely heterogeneous, which makes representative sampling impossible unless very large samples are taken, while the high specific gravity of such grains leads to their being concentrated in traps along the stream bed. Several ways of overcoming these difficulties have been suggested. Among these the "schlich" test, in which heavy material obtained by panning is taken for analysis is among the most popular. However, the results of the present study indicate that while this method improves the nature of the patterns obtained, they are still far from satisfactory. Alternative methods of sample preparation, including the taking of light material for analysis, appear to provide better answers to the problem. In the area studied this approach eliminated most of the "noise" associated with the data obtained by other methods, as well as giving an extended dispersion pattern.