Treatment of Data Uncertainties

The generation and use of data covariance matrices are discussed within the context of the analysis of neutron-induced cross-section data via the R-matrix code SAMMY. Two complementary approaches are described, the first involving mathematical manipulation of Bayes' equations and the second utilizing computer simulations. A new procedure for propagating uncertainties on unvaried parameters will allow the effect of all relevant experimental uncertainties to be reflected in the analysis results, without placing excessive additional burden on the analyst. Implementation of this procedure within SAMMY is described and illustrated through the simulations.