Correlation studies of 400 samples of sieved stream sediments and 325 samples of fluvial, concretionary Mn-Fe oxides from Maine resulted in the separation of elements into the following categories: (1) elements not scavenged by Mn-Fe oxides - B, Cr, K, Mg, Rb, Sc, Ti, V, and Zr; (2) elements probably not scavenged by Mn-Fe oxides - Ag, Be, Ca, Ga, La, Sb, and Y; (3) elements scavenged weakly by Mn-Fe oxides - Cu, Mo, Pb, and Sr; (4) elements scavenged strongly by Mn oxides - Ba, Cd, Co, Ni, Tl, and Zn; and (5) elements scavenged strongly by Fe oxides - As and In. Studies of the scavenged elements showed that the deviation from the mean is characteristically greater in oxide samples as compared to sieved sediments from the same locality. However, a significant increase in contrast between anomalous and background localities, when oxides are the sample medium, more than offsets the disadvantage of data scatter. The use of oxides as a sampling medium clearly and significantly accentuates anomalous localities. In general, non-ratioed data on oxides give very nearly the same results as data consisting of scavenged elements ratioed to Mn and Fe. However, ratioed data expand the geographic area of specific anomalies. Cd and Zn consistently show strong correlations with concretionary Mn-Fe oxides, but their concentrations in the oxides do not generally show as much contrast between anomalous and background localities as do Cu, Mo, and Pb. These latter elements are strongly scavenged where rocks are mineralized. ?? 1976.