JEFFREYS has raised an interesting objection1 to the possibility of convection or steady-state creep occurring in the upper mantle. Lomnitz has shown2 that the logarithmic stress–strain relationship observed in igneous rocks in the laboratory may be used to derive an attenuation coefficient for stress waves that is directly proportional to frequency (constant-Q) over a limited range of frequency. Jeffreys argues that as the logarithmic creep behaviour is associated with the constant Q behaviour of stress waves, and experimental data indicate that stress waves in the mantle exhibit constant Q, the mantle material cannot undergo steady state creep under a constant stress.