666 M. Dubernet
For each analyte, 2–10 characteristic wavelengths are selected. The quality of the
calibration is then tested using a separate set of reference wine and must samples.
12.4.3.2 Routine Calibrations
It is necessary to re-calibrate the instrument periodically, notably when routine con-
trol indicates an instrument-related drift due to factors such as aging of the electronic
components, reparations, changing of spare parts, etc. Re-calibrating the instrument
does not modify the measurement wavelengths but does involve recalculation of the
K 0 and Kiconstants.
12.4.3.3 Routine Correction of Bias
Each time the equipment is used, control sample(s) containing a known concentra-
tion of the analyte are analysed. If the there is bias in the result by comparison with
the expected value, a correction may be carried out.
12.4.4 Usual NIR Methods
12.4.4.1 Determination of Ethanol in Wines and Musts During Fermentation
This is a very widely used determination, but suffers from the drawback of matrix
effects, for instance if the wine contains atypical concentrations of glycerol, as was
first reported by Bouvier (1986).
Characteristics of the Method
Inter-laboratory reproducibility: 0.19vol.%
12.4.4.2 Determination of Reducing Sugars in Musts
This is a less widely-used application. No inter-laboratory studies have been carried
out to describe characteristics of the method.
12.5 Fourier-Transform Infra-Red Analysers (FTIR)
According to Dubernet et al. (1999, 2000, 2001, 2006), Dubernet and Dubernet,
(2000), Bertrand and Dufour (2000), Patz et al. (1999), and Boulet et al. (2007a, b).