Gravimetry has had a long history, using pendulums, torsion balances, and static spring gravimeters. Relative accuracy adequate for many geophysical problems was already attained by 1900, but it took another half century to build readily portable gravimeters. Calibration and datum definition remained problems until the 1970s when free-fall absolute gravimeters were developed that now have a precision of 10-3 mGal. The problems of geographic inaccessibility and field party costs (notably in areas of greatest tectonic interest) and now being overcome by airborne gravimetry that has already achieved accuracies of 1--3 mGal with resolutions of 10 to 20 km. Satellite techniques are the best way to determine the long-wavelength variations of the gravity field. The resolution of the models has steadily improved with the number of satellites and the precision of the observations. The best current model includes tracking data from more than 30 satellites, satellite altimetry, and surface gravimetry and has a resolution of about 290 km (harmonic degree 70) with the most recent improvements coming from Doppler orbitography and radiopositioning integrated by satellite (DORIS) tracking of the SPOT 2 satellite and satellite laser ranging (SLR), DORIS, and Global Positioning System (GPS) tracking of the TOPEX/POSEIDON satellite. Meanwhile, radar altimetry has become the dominant technique to infer the marine geoid with a resolution of tens of kilometers or shorter. Similarly, the gravity fields of the Moon, Venus, and Mars have been determined to harmonic degrees 70, 75, and 50, respectively, although tracking limitations result in variations of spatial resolution.

Modeling Earth's gravity field from the abundance of precise data has become an increasingly complex task, with which the development of computer capacity has kept pace. Contemporary solutions now entail about 10,000 parameters, half of them for effects other than the fixed gravity field of Earth. Temporal variations arising from tides have long been well modeled, and nontidal changes are now being identified. The improvements in gravitational models engendered corresponding advances in geophysical interpretation. Isostatic models were refined and expanded to account for regional thermal and tectonic histories. Interpretation of the long-wave-length gravity field determined by satellite techniques has been mainly in terms of plate tectonics as a manifestation of mantle convection. Gravity has been significant in inferring that there must be a large increase in viscosity with depth (most strongly, from the apparent slow sinking of subducted slabs). The prospects for increasing accuracy and resolution in the determination of Earth's gravity field rest primarily with the development of new measurement systems. Airborne gravimetry is taking promising new steps using GPS, but significant global model improvements awaits a dedicated satellite gravimetry system, and future satellite altimeter missions will do more for ocean dynamics studies than geoid improvement. Advances in interpretation will occur through the development of other data, such as seismic tomography, and larger-scale computer modeling of tectonics and convection. ¿ American Geophysical Union 1995