Effect of correlations of the experimental data on the errors in evaluated neutron cross sections

This work concerns the underestimation of the errors in the recommended neutron cross sections obtained by using strict statistical methods to analyze measurements. It is shown that one reason for this is inadequate representation of the initial experimental data, specifically, correlations of the experimental data are neglected. The effect of correlations of the experimental data on the error in the evaluated neutron cross sections is studied using realistic models and practical examples. Special attention is focussed on correlations of measurements performed by different authors. Analysis of correlations of this kind substantially increases the errors in the evaluated cross sections. Analysis of the model examples shows that the correction can be large (∼70%) even for cases with relatively weak correlation of the measurements performed by different authors. For the evaluated cross sections of the reactions¹⁰³Rh(n, n′)^103mRh and²³⁷Np(n, f) the corrections reach 41% and 25% for certain energy ranges. 5 figures, 1 table, 22 references. Federal Science Center of the Russian Federation—A. I. Leipunskii Physics and Power-Engineering Institute. Translated from Atomnaya énergiya, Vol. 86, No. 1, pp. 40–46, January, 1990.