Gravitational observations of free vibrations of the earth due to the Alaskan earthquake of March, 1964 were compared with corresponding results for the Chilean earthquake. In the graver spheroidal modes of periods greater than 7 minutes which were compared, the periods generally agreed within a few parts per thousand. The freemode spectrum has been used by Anderson, Landisman, and others to improve models of the earth. To fit both the free mode periods and the seismic data, Anderson's modification of classical earth models introduces a more rapid increase of density in the upper mantle, and less rapid increase near the core. In addition to the precise set of numbers characterizing the free periods, the free mode observations supply another independent set of criteria in the form of the “Q” values for the different modes. The large Q value of more than 25,000 for the purely radial mode excited by the Alaskan quake is noteworthy. In gravity observations of the earth's oscillations, of period one hour or more, and especially in observations of the semidiurnal and diurnal tides, the vertical component is observable with greater reliability than is the horizontal component. Improved instrument sites, probably drill-hole sites, seem to be required. A laboratory model for a simple type of drill-hole gravimeter incorporating a new efficient transducer due to R. Havens was shown. The instrument has a 20-sec period, and in preliminary tests shows a sensitivity of a few microgals.