There is some further discussion of the two regressions problem and implications for which way round to fit equations and how to select samples in:
2.
FearnT., “Flat or natural? A note on the choice of calibration samples”, in Near Infrared Spectroscopy, Bridging the Gap between Data Analysis and NIR Applications, Ed by HildrumK.I.IsakssonT.NæsT.TandbergA., Ellis Horwood, New York, p. 61–66 (1992).
