2. What are the problems encountered during calibration?

The available input data x (standards) and associated measured values y are not perfectly known.

• On the variable x , there is always an uncertainty on the values of x .

• For the variable y , the measurement process is not perfect, and its repeatability and other uncertainties lead to different y values for the same x .

Furthermore, during a calibration, the measurement process is carried out with the same operator under environmental conditions that vary little over the course of the calibration. These conditions generate covariance phenomena between errors on x , between errors on y and between errors on pairs ( x , y ).

The aim of modeling is to determine the model that best describes the observations.