On the Choice of Regression in Linear Calibration. Comments on a paper by R. G. Krutchkoff