Uncertainty justification of neutron cross section standards

Abstract
The existence of general measures for the covariance matrices of the evaluated data uncertainties obtained in different model and non-model least-squares fits of the same experimental data is considered. It is shown that some quantities (called measures here) calculated from the elements of covariance matrices are exactly and others are approximately conserved quantities independent from the model used in the fit. The use of these measures allows a simple comparison of uncertainties obtained in different evaluations of the same cross sections. The measures were used for the justification of the uncertainties of the neutron cross section standards evaluated in least-squares fits of a bulk volume of the experimental data. It is shown that the uncertainties obtained in the standards evaluation are realistic and determined completely by the uncertainties of the experimental data when taking into account all known correlations between the different experiments.