Revision of ISO 19229 to support the certification of calibration gases for purity

The second edition of ISO 19229 expands the guidance in its predecessor in two ways. Firstly, it provides more support and examples describing possible experimental approaches for purity analysis. A novelty is that it describes how the beta distribution, or some other suitable probability distribution can be used to approximate the distribution of the output quantity, i.e., the fraction of a component. It also provides guidance on how to report coverage intervals in those cases, where the usual approximation from the GUM (Guide to the expression of Uncertainty in Measurement) to use the normal or t distribution is inappropriate because of vicinity of zero. Coverage intervals play an important role in conformity assessment, and it is also customary to report measurement uncertainty in the form of a coverage interval, notwithstanding that ISO/IEC 17025 does not explicitly require it. ISO 6141, which sets requirements for certificates of calibration gas mixtures, does require the statement of an expanded uncertainty, which has been interpreted that in the case of a non-symmetric output probability distribution, a coverage interval should be stated, along with the value and the standard uncertainty. This paper gives a brief background to the choices made and examples in ISO 19229.