Sampling variability of stream sediments in broad-scale regional geochemical reconnaissance

The errors resulting from field sampling, sample preparation and analysis are investigated for drainage sediments derived from homogeneous sandstone, shale, limestone, basic igneous and granite lithologies as a background to a primary reconnaissance survey of England and Wales. Emission spectrographic analyses of approximately 350 samples for Al, Ca, Fe, K, Si, Ba, Co, Cr, Cu, Ga, Li, Mn, Ni, Pb, Sc, Sn, Sr and V and atomic absorption analyses for Cu, Ni, Mn, Co, Zn and Pb have been treated statistically. For routine broad-scale surveys analysis of the unground -80-mesh sediment fraction proves satisfactory. Sampling error between localities in a stream is shown to be less than spectrographic analytical error for many of the elements considered here, except where mineralisation (contamination) occurs, but exceeds the atomic absorption analytical errors in most cases. Unless it it critical to reduce the amount of between-site variability (having regard to analytical errors) in the regional survey, there appears to be no advantage in the use of a more sophisticated sampling technique. The utility of moving-average maps in reducing the width of the confidence belt around the regional 'geochemical surface', and thus compensating to some extent for the sampling errors, is also demonstrated.