Abstract: A sampling system includes a plurality of sensing means (3) mounted on one or more beams (2) which reciprocate, over a moving web (1) of material. The sensing zones of adjacent sensing means (3) trace out scanning paths which periodically overlap in a sampling region. One of the sensing means (primary) scans over a known reference to update its calibration periodically or continuously and this is transferred to the other sensing means (secondary) effectively via the travelling material (due to the overlap). Calibration is performed by signal processing means, storing and comparing monitoring signal levels or corresponding digital values derived at positions of overlap and making computations for the purpose of calibration. With sufficient overlap of the scanning paths any one of the sensors can be made redundant. Alternatives include sensing means which move relative to other sensing means as well as rotary arrangements.

Abstract: In the method and apparatus, optimum correlation between measured and known values of transmittance or reflectance determines the property or identity of a sample. Three or more discrete radiation components can be selected which are transmitted or reflected by relatively different amounts due to interference, absorption or scatter, thereby enabling fewer radiation components to be used to provide an accurate result. The known values can be derived by analogue or digital techniques, the digital technique employing a model or models taking account of absorption, interference and/or scatter. Techniques are disclosed for reducing computing time by range ajustment, identification of a zero correlation value in a differential function of the model or models and computation of correlation coefficients for two or more groups of radiation components.