An algorithm is developed to derive optimized calibration procedures for practical thermometers.
The algorithm, utilizing a set of calibration data for one kind of practical
thermometer, analyses the characteristics of the sensors so as to obtain the optimum
locations of the calibration temperatures and relevant deviation functions. Following
the procedure, one can estimate the calibration accuracy for a given number of calibration
points with their best allocation. As an example, the algorithm is applied to
industrial grade platinum resistance thermometers. It is shown that five calibration
points are enough to realize a calibration accuracy better than 10 mK for a 20-273.16 K
temperature range.