Catalogs of old magnetic measurements, contemporary publications, and ship's logs have been researched to compile a data set of geomagnetic observations of declination and inclination in the interval 1601 A.D. to 1694 A.D. This search produced 2766 measurements of declination, with quite good global coverage, and 37 of inclination. All inclinations were in the northern hemisphere and most in Europe. A literature search revealed many archeomagnetic and paleomagnetic measurements of field components for the seventeenth century: 40 of these were selected and used in the modeling procedure (intensity measurements were not found sufficiently reliable to allow any conclusions to be made about the overall strength of the geomagnetic field and declinations were too few to be of use). This data set was analyzed to produce a field model for the seventheenth century; the model was denoted epoch 1647 A.D. although no corrections were made for secular variation. The model exhibits many of the stationary features of the recent geomagnetic field and indicates absence of secular variation from the Pacific and a westward drift rate of about 0.25¿ per year in areas of the core-mantle boundary beneath the south Atlantic and Indian oceans between 1647 A.D. and 1715 A. D., the date of Bloxham's (1986) earliest field model. This is likely to remain the earliest field model because only three inclination measurements have been found in the sixteenth century; the data set will be used in the future to form a time-dependent model for the historical period. ¿ American Geophysical Union 1990