Towards an uncertainty paradigm of detection capability

Abstract

In this editorial we review the idea of detection limit. The rationale for the currently preferred definition is clear but complex. As an outcome, a practical implementation of the definition can be difficult in chemical measurement and usually falls short of its intent, making the added value of the complexity questionable in practice. Attention is further drawn to important shortfalls in the current definition, such as the unspecified significance levels, the conditions for estimating the dispersion of results, and the assumption of homoscedasticity. More fundamentally, the establishment of imaginary ‘limits’ where none exist in reality seems unjustified and can be misleading to analysts and their customers alike. We suggest that for many purposes detection limit and related limits are made redundant simply by reporting the measurement result and its uncertainty.