Design of linear calibration experiments

This paper discusses the design of calibration experiments, focusing on approaches that minimise the uncertainties associated with the use of the instrument. For this problem, the optimal design depends on the actual response of the system and so cannot be implemented without prior information about the actual response. We review some of the standard approaches in optimal design applied to calibration problems and then consider how they can be extended to optimise the in-use uncertainty. We also consider issues when the model for the calibration function is only partially known.