[en] The relation between random and deterministic features of the calibration problem for measuring devices with rectangular coordinate systems is considered. The numerical and methodological problems connected with the calculation of calibration coefficients are studied. The calibration transformation is defined as the best least squares approximation of some finite coordinate set by polynomials in two variables. Described are the GENCAL and ORCAL calibration programs, which use the orthogonal polynomials at the finite point set and the orthogonal polynomials at the unit interval of < -1.1 > x < -1.1 >. The results of computation by means of these programs for various values of input parameters and for various sets of the data measured are discussed