M. Járó - L. Költő szerk.: Archaeometrical research in Hungary (Budapest, 1988)

Analysis - BALLA Márta, BÉRCZI János, KEÖMLEY Gábor, ROSNER Gyula, GABLER Dénes: Provenance studies of ceramics by neutron actiwtion analysis

Comparison of the relative error calculated from the statistics of decay with the relative standard deviation of the mean value (per cent) Element Relative error (%) Stand, dev. of mean value (%) Sc 0.5 1.6 Cr 5.5 7.9 Fe 1.0 1.3 Co 2.9 4.4 Rb 7.0 8.6 Cs 4.5 6.7 La 0.6 0.8 Ce 6.0 7.2 Eu 1.8 3.5 Yb 10.0 133 Lu 7.1 95 Hf 5.0 7.0 Ta 10.0 12 3 Th 35 53 U 6.8 11.4 The accuracy and reproducibility of our measurements were controlled by a standard reference material named Standard Pottery (prepared by Perlman and Asaro); this material has been analysed by a great number of laboratories. Table 2 shows the mean values of our measurements of standards and reference data for comparison. The results and errors are the average and standard deviation of the five measurement series carried out over about three years. The 4th column of the table shows the differences of the measured and reference values. It can be seen that for most of the elements the deviation of the mean value of our measurements from the reference data is smaller than the standard deviation of the mean value or scarcely exceeds it. There is a systematic error only in the case of iron, though this does not cause a problem during comparison studies. On the basis of our experience gained during other reference material measurements we suggest that the concentration value of europium the literature must be corrected. As is shown by the tabular data our measurements are sufficiently correct and accurate. Investigations on homogeneity The cardinal question of chemical analyses of finds is whether it is possible to extend the information obtained from a small sherd to bigger potteries, i. e. what is the confidence or error of stating that the composition of the sherd is similar to that of the whole original pottery. This leads us to conclude on what deviation can be considered significant between two samples. In the literature there are relatively few examinations of this kind [8]. Judging from these it cannot be indicated about the examined elements