Reliability and accuracy of environmental analytical data on moss samples : an interlaboratory comparison

An inter-comparison exercise was performed between two laboratories on mineralised moss samples. Ten samples, together with three certified reference materials (CRM 61, CRM 62 and CRM 482) and five blanks were mineralised independently in the two laboratories and then analysed by both with different techniques (AAS, ICP-MS, and INAA). The elements analysed were: Cd, Cr, Cu, Fe, Ni, and Pb. Accuracy tests showed that, for AAS, the relative standard deviations for five repetitions of the same sample were between 0.2 % for Cd and 2.9 % for Cr and that the concentration values found for reference materials were within one standard deviation from the certified values, with the exception of Cr and Cd for CRM 482. The statistical analysis was performed first with a simple linear regression and a Student t-test taking into account the possible influence of the mineralisation procedure and of the analytical technique, in order to evaluate their reliability. The results showed that more than 78 % of the regressions had a correlation coefficient higher than 0.750, and that 83 % of the slopes were not different from one with a level of significance of 0.01. Afterwards, the precision of the measurements were evaluated using maximum likelihood; the normalised errors showed that ICP-MS had a better performance on most of the elements analysed (with the exception of Fe) and that both the mineralisation procedures gave satisfactory results. The two laboratories were then found to produce reliable results