Harper-Dorn (HD) creep may be produced by inadvertent low-amplitude temperature cycling that is inevitable during high-temperature creep tests. A cyclic temperature change causes a cyclic change in the equilibrium point defect concentration. In turn, a cyclic chemical force on dislocations that produces climb motion is created because the actual point defect concentration can never exactly equal the instantaneous equilibrium value. A cyclic ¿ 1 K can produce a maximum chemical stress of ¿5 MPa. When the period of the cycling is neither very large nor very small and the chemical stress is larger than the applied mechanical stress it, rather than the mechanical stress, determines the dislocation density. In this case, the dislocation density is independent of the applied stress. Therefore, the creep rate is proportional to the applied stress and is independent of grain size, the characteristics of HD creep. If this explanation of HD creep is correct this type of creep will not occur in the earth's mantle and will not serve as a mechanism for Newtonian mantle viscosity.