A fatigue crack is subjected to a single tensile overload and then returned to a lower cyclic load. In some metals, subsequent crack growth temporarily slows. Which explanation is most consistent with this observation?