STATISTICAL CALIBRATION AND PERIODIC PERFORMANCE CHECKS FOR SEGMENTED GAMMA SCANNING

The use of no intercept, Poisson weighted least squares regression for the calibration and ongoing performance checks of segmented gamma scanning is discussed. The lack of fit to a straight line which can result when process materials are used as standards is considered; it is included as a variation source in estimating gram uncertainty, but bias corrections are made in conducting instrument performance checks. Applications to actual calibration data are given.