OVERLAPPING RESONANCES 975As expected, when one of the channels is switched off, Eq. (24) reduces to Eq. (8). By integrating out e and one of the total widths in Eq. (24) we find that the distribution of the single total width r=r\/(rx) is given by p(r)=(M37r~1)[exp( / zi)][r(2-r)-M2]-1/2 X{MIM3-T#I(MI)+#O(MI)] -2[r(2~r)-/x 2 Ko(Mi)} Xexp{-2 M 3Cr(2~r)i u 2 ]}, (25) where Mi= (This should be compared with the total dimensionless width y of the R matrix which is given byTo compare the distribution of the width V given by Eq. (25) with the one given by Eq. (26), we again calculate the mean-square deviations of the quantities T and y. As in Sec. II B, we find that the distribution of the width T is always broader than the one given by Eq. (26), except when the quantities D C i/(S) and Dc2/(S) are very small.The spectroscopic factors S n of bound neutron states are usually found from (d,p) stripping reactions. An alternative method of finding S n for medium-to-heavy nuclei is to analyze isobaric analog resonances observed in (p,p) scattering from these nuclei. The present analysis uses a modified i?-matrix theory in which boundary matching is done within the optical-model potential region rather than directly onto the Coulomb potential region. A resonance mixing phase and an optical penetrability are introduced. Both single-and multilevel resonances are treated. The effects of compound elastic scattering and the energy dependence of the level shift are investigated. Formulas for the spreading width are obtained. The variation of S n with the value of the matching radius and the best choice of this radius are discussed. As examples of the method, analyses of the s-wave resonance in 92 Zr(p,p) 92 Zr near 6.0-MeV bombarding energy and of s-and d-wave resonances in 90 Zr(p,py°Zr near 5.8 and 6.8 MeV are presented. The values of S n obtained are compared with those from (d,p) experiments, and the reliability of the two methods is discussed.