A view of uncertainty at the bench analytical level

 The problem with which analytical laboratories are confronted, after traceability of their results has been demonstrated, is correctly estimating their uncertainty– to which traceability is also to some extent subject. While the general principles for calculating the uncertainty of physical measurements are applicable to chemical metrology, some refinements are needed, especially careful selection and planning the level at which uncertainty will be estimated by each laboratory in accordance with its capacity and required demands. Depending on the particular decision to be made, the mechanism to be used to estimate the uncertainty varies markedly; also, the rigour of the estimation increases with increasing stringency of the demands. This paper describes the primary sources of uncertainty in chemical metrology and discusses different approaches to its estimation in relation to the type of analytical laboratory concerned. The view presented tries to be close to the bench analytical level, in order to be practical and flexible for laboratories, although it could sometimes be considered slightly heterodox. Received: 25 March 1997 · Accepted: 20 September 1997