An analysis of runs tests as indicators of nonstatistical structure in the compound nucleus

Runs tests were applied to computer-generated elastic-scattering excitation functions in order to examine their sensitivity to nonstatistical structure in the compound nucleus. The scattering amplitudes were either purely statistical or else they included contributions from additional resonances about 1 MeV broad. Our aim was to determine whether the runs tests could distinguish between the two cases. The tests were successful once trends producing correlations between the datum points were removed. Spurious nonstatistical effects were thus eliminated, demonstrating that if the trend-reduction problem is dealt with propertly, runs tests are effective in searches for nonstatistical structure beneath the compound-nucleus fluctuations in experimental excitation functions.