The parity nonconserving (PNC) resonant scattering of neutrons at epithermal energies is discussed within the optical model. We show that the empirical optical-model potential with real PNC interaction does not account for the appreciable resonance-averaged difference between positive and negative helicity total p-neutron cross sections (Ao. ) for 3 Th, but reproduces the very small (Ao. ) for 2 U. It is suggested that the 2p-1h local doorway, previously proposed to account for the sign correlation in Th, is the main source for the large (Atr) seen in this system. It is further proposed that if an optical potential is used to account for (Atr), a complex PNC interaction is required in cases involving strong sign correlation. PACS number(s): 25.40.Ny, 11. 30.Er, 24.10.Ht, 24.80.Dc In a recent paper [1],we proposed a model that explains the sign correlation seen in the data on parity-nonconserving (PNC) resonant neutron scattering from Th at epithermal energies [2]. Our model emphasizes the role of a local 2p-1h doorway state which is responsible for the weak coupling between the p and s waves. In Th, this doorway state is reasonably well separated from other doorways, whereas in U, where no sign correlation is seen [2], there are several 2p-1h doorways present in the energy range of interest. In this paper we address another question also relevant to the PNC neutron reaction, namely to what extent the averaged total cross-section difference (ho. q) is accounted for by an optical model which contains an appropriate one-body PNC interaction. The quantity 50. q is just 0.~~-0.~~, where 0 q is the cross section at the q th p resonance and the superscripts + ( -) refer to positive (negative) helicity neutrons.In the following we give arguments that indicate that the same local 2p-1h doorway responsible for the large positive value of the resonance averaged longitudinal asymmetry in n+ Th (the sign correlation effect), also has bearing on the optical-model analysis. We should mention here that our local 2p-1h doorway states are statistical in nature, in contract to the collective 0 doorway (giant monopole) state considered in Refs. [3,4] and to the high-lying single particle "doorway" of Ref. [5]. Both of these alternative doorwaytype approaches to the sign correlation problem require an unrealistically large parity violating matrix element about 2 orders of magnitude larger than the normal (-1 eV). Two optical-model calculations related to the TRIPLE [2] data have been reported [6,7]. These calculations purported to analyze the resonance-averaged longitudinal analyzing power Permanent address: Instituto de Fisica, Universidade de Sa™o Paulo, C.P. 20516, Sao Paulo, S.P. , Brazil. Et was concluded in both of these references that the onebody PNC part of the optical potential has to be more than 2 orders of magnitude larger than usual estimates in order to reproduce (Pq)q [6,7].From the optical-model point of view, the quantity (Pq)q is rather cumbersome to analyze. A more natural quantity to discuss is the ratio of energ...