Within the 1979-1982 time frame, at Delft University's Department of Aerospace Engineering some numerical experiments were performed by processing a limited number of multiday global tracking data arcs. Each arc consisted of laser ranging data acquired in the 1978-1980 period. These arcs were selected especially for their relative abundance of European observations of the satellites LAGEOS, Starlette, and GEOS 3. An evaluation of the orbit errors due to errors in the gravity models was performed both in terms of apparent range and timing biases of each pass over a tracking station and of orbit differences between solutions in which different gravity models were applied. In addition, from LAGEOS and Starlette tracking data, separate solutions were obtained for the positions of the European laser ranging stations at Kootwijk, Wettzell, Grasse, and Mets¿hovi. In spite of the use of the most accurate tailored gravity models available at the time, the results, both in terms of absolute coordinates and in baselines, often differ considerably from more recent solutions by other authors. It is hypothesized that intermittent data-taking problems at some of the European stations are responsible for these differences. Their distorting effects are relatively strong because of the limited number of passes on which our solutions are based.