An investigation of some sources of error in applied geochemical data from South Devon, England.

Various aspects of analysis by AAS have been investigated. Digestion of 0.5g of minus 125 micron material in 4M HNO3 at 95°C for 4 hours seems to be the optimum for such elements as Cu, Pb, Zn, Fe and Mn. Tin determination by AAS, using an air-acetylene flame, is possible, but problems with the ammonium iodide digestion used, remain to be solved. However, the selection of minus 190 micron material as optimum for Sn may be of only local significance. Some interference effect was observed when different weights of the same samples were analysed, and a spectral interference between Cu and Ag has been detected under certain conditions. Some losses of reproducibility result from evaporation from aqueous digestions after storage for only a few hours in unsealed containers. The postulated presence, in stream sediments, of HNO3-soluble ferric arsenate, and the above digestion, permit as determination by conventional AAS. Standardised, systematic techniques for such processes as drying and sieving of samples are shown to be necessary in order to achieve good reproducibi1ity. Partition of error variance has demonstrated that, unless analytical detection limits are approached, sampling error dominates sub-sampling and analytical errors. Various aspects of sampling error have been investigated. The lowering of AAS detection limits, by the use of flame-less atomisation techniques, means that a case may now be made for the use of acidified natural water samples as a regional as well as follow-up, sampling medium. The dependence of metal partition between sediment and water on physicochemical properties of the stream has been demonstrated. Therefore, a complete understanding of secondary dispersion can only be achieved if water, as well as sediment, is analysed. Significant seasonal variations in metal content of both sediment and water samples have been detected and related to the groundwater/surface run-off system. The data from the main field area, at Yarner Wood, has been evaluated in the context of the mineralisation believed to be present.