No abstract

60 geochemical laboratories from the world over, contributing in all 3687 bits of chemical data on 60 elements, haue willingly participated in the study of SOIL‐5 as a reference material. The preparation of SOIL‐5 in 45 kg and homogeneity teats are described; in addition, all the data compiled are presented. In evaluating the compiled data, the use of four concurrent tests for outlier‐rejection at the same level of significance, has been applied. After proper statistical treatment, “recommended values” could be established for 31 elements, most of them trace elements and “information values” for 29 elements. 23 of the recommended values could be given with a “relatively high” degree confidence whereas the other 18 recommended Values with a “reasonable” degree of confidence. An attempt is made to compare the analytical methods employed; the existence of possible sy3tematic differences is pointed ont.

Results of a recent International Atomic Energy Agency intercom parison test on the determination of 16 trace elements in simulated fresh water samples are reported and discussed from the standpoint of accuracy and precision. Despite the considerable spread of results obtained froni various laboratories, the over-all mean (after rejection of outliers) in most instances approximates closely to the true value. The importance o€ using sufficiently sensitive tests for the detection of outlying results is emphasised and the effectiveness of concurrent use of several criteria for that purpose is demonstrated. The relative frequency of employment of various analytical techniques is discussed and an attempt to compare their accuracy is presented.

No abstract

No abstract