Observations of the Earth's nutations are used to explore the possible presence of a conducting layer at the base of the mantle. The existence of such a layer is suggested by recent experiments on iron and silicate mixtures at high pressures and temperatures (Knittle and Jeanloz, 1986, 1989), which indicate that iron may react with silicate minerals to produce metallic compounds. Nutations provide an ideal tool for investigating the existence of such a layer because they are sensitive to the electrical properties within a thin magnetic boundary layer along the core-mantle boundary (CMB). Evidence in favor of a conducting layer is found in the effects of ohmic dissipation which cause the amplitude of the Earth's nutation to be out-of-phase with tidal forcings. Although out-of-phase components in the Earth's nutations are conventionally attributed to the dissipative influences of ocean tides and mantle anelasticity, these effects appear incapable of explaining nutation amplitudes determined from very long baseline interferometry (VLBI) observations. The additional effects of ohmic dissipation, enhanced by the presence of a thin conducting layer at the CMB, are sufficient to reconcile theory and observation.

The largest mismatch between observations and theory, taking account of ocean tides and mantle anelasticity, is a discrepancy of 0.39¿0.04 milliarcseconds in the out-of-phase component of the retrograde annual nutation. This discrepancy can be eliminated if the lowermost 200 m of the mantle has a conductivity of 5¿105 S m-1, and the magnetic energy in the spherical harmonic components of the field at the CMB in degrees l>12 is approximately 4 times greater than the energy inferred in degrees l≤12 from surface observations. Unfortunately, the VLBI observations do not uniquely resolve the mantle conductivity and magnetic energy, so that trade-offs exist between these two parameters. Nevertheless, a high mantle conductivity is favored on the grounds that the alternative, involving high levesl of magnetic energy, yields an excessive amount of ohmic heating. Other interpretations of the VLBI discrepancy are also discussed. One such alternative involves viscous dissipation at the CMB, requiring a kinematic viscosity of &ngr;=0.1 m2 s-1. ¿ American Geophysical Union 1992