Uncertainty due to primary sampling of 222Rn in analyses of water

Measurement uncertainty is an important variable, to be accounted for when decisions have to be made based on measurement results. Measurement uncertainty is composed of two main components; one is related to the primary sampling, the other to the sample preparation and the subsequent analysis of the sample. The component related to the sample preparation and the analysis is commonly well evaluated in proficiency testing while there is generally no straightforward similar approach to evaluate sampling uncertainty. ISO 17025:2017 explicitly requires that testing laboratories performing sampling and analyses determine the uncertainty related to the primary sampling. In order to determine uncertainty arising in the primary sampling of 222Rn in water destined for human consumption, three laboratories IRE (BE), DiSa (LU) and SCK CEN (BE) initiated a joined sampling and measurement campaign. The dual split sample method in combination with ANOVA was used to evaluate the primary sampling uncertainty (precision) of the different methods. The tests showed that sampling bias is very probably, but that with good laboratory practice the sampling uncertainty precision and respectively bias can be kept below 5%. •Primary sampling of 222Rn in water was organized.•Sampling uncertainty was determined by dual-split-sample concept and ANOVA.•No correlation between sampling bias and sampling precision was observed.•Sampling uncertainty for 222Rn in water is of the order of 5%.