The measurements of the magnetization vector t,rajectories in the read cycle revealed that the magnetization vector follows the variation of the applied field at rising time but is fairly delayed at falling time. It takes more than 200 ns to steady the vector of transient. About 10 percent of the hard-axis magnet,ization component remains after t,he first reading. This is explained by considering t.he torque curve and magnetic inhomogeneities. We may consider that the delayed magnet,ization rotation causes a hard-axis eflective field Heif. H,ff and Hzd are overlapped, and H,tf varies with time; this causes creeping. The solid lines in Fig. 2 show the read-write disturbance t.hreshold obtained from the magnetization vector trajectories, the switching curves, and the usual creep threshold curves. The calculated curves are in fair agreement with the direct measurements. Striped domains which make about 45” to the easy axis were observed in the Bitter pattern after the read cycle. The domain state may correspond to the remaining hard-axis magnetization component. When Hza is applied under this domain configuration, the magnet,ization reversal may occur, and it contributes to the creeping independent, of ti. The read-write disturbance becomes appreciable in the case of the NDRO mode and limits a cycle time in memory operations. These phenomena can occur not only in plated-wire memory but also in other magnetic film memories.