The analytical and economic importance of correctness in sampling

Abstract Sampling is part of quality evaluation as well as of the analysis. Probabilistic and non-probabilistic sampling methods are discussed, with examples of the severe financial consequences of incorrect sampling procedures. Only probabilistic methods (cross-stream and splitting) can be correct but their correctness depends on sample increment delimitation, increment extraction and integrity of increments and final sample. Recommendations are given for designing sampling procedures and designing sampling equipment. Common sources of error are described. The need for analytical specialists to be responsible for both sampling and sample processing is emphasized.