The errors and the possibility to reduce them in the different steps of a quantitative analysis are discussed principally and with respect to the investigated Roman pottery. The concentrations of the elements were determined by x-ray fluorescence and in part by activation analysis. The variation of the values, which is due to the natural variation of the composition of the material, sampling and analysis, has been determined. Analytical results of the same material from different laboratories are compared and discussed with respect to already published data. Using these results, the problem of the significance of reference groups, secondary alterations of the composition and the advantage of main and trace elements is discussed. The importance of a methodological carefulness of all disciplines participating on a research project for the accuracy of reference groups and analytical data is pointed out. -- AATA