The core-mantle boundary (CMB) sets boundary conditions for processes occurring within the core. Thus the history of the geomagnetic field is intimately connected with the history of the CMB: information about one can often provide information about the other. It is a simple matter to separate sources of external origin from those of internal origin, but there is no unique way to specify the location of the internal sources. Indirect arguments suggest that sources within Earth's outer core strongly dominate the field for spherical harmonic terms with degree less than about 14 and that crustal sources play an increasing role in harmonics with higher degrees. Building on the pioneering studies of Roberts and Scott and of Backus in the 1960s, the velocity field of the core fluid at the top of the outer core can be estimated from geomagnetic secular variation data. However, even with appropriate assumptions about location of the magnetic field source, there is a serious nonuniqueness in inversion of the secular variation data to velocity field because magnetic field lines are not changed by movement of conductive fluid parallel to those field lines.

Additional assumptions about the flow are therefore required, sometimes leading to quite different estimates in the pattern of fluid flow. The resulting velocity field estimates appear to be sensitive to conditions (e.g., small lateral variations in temperature) at the CMB. Consideration of mantle dynamics suggests that large changes in the CMB conditions probably occur on a 10- to 100-m.y. timescale. Secular variations with periods shorter than a million years, but longer than several years, almost certainly originate from processes operating in outer core; unfortunately, there is not yet consensus is to what those processes are. Longer-period variations in the paleomagnetic record, including nonstationarities in the rates of magnetic field reversals, in paleointensities, and in paleosecular variation, may reflect changes in CMB conditions. Because these boundary conditions are controlled primarily by mantle dynamics, there have been several speculations regarding causal links between changes in Earth's lithosphere and changes in Earth's magnetic field. Finally, the evidence linking lateral variations at the CMB to perceived systematics in polarity transition data is found to be intriguing, but insufficient. ¿ American Geophysical Union 1995