A series of interferometry experiments has been conducted between two antennas at the Goldstone Deep Space Communications Complex near Barstow, California. The primary objective of these experiments was to construct and demonstrate on a short base line independent‐station instrumentation capable of making three‐dimensional base line measurements with accuracies of a few centimeters. To meet this objective, phase stable instrumentation was constructed to precisely measure time delay by means of two‐channel bandwidth synthesis. Delay measurements produced by this instrumentation led to three‐dimensional earth‐fixed base line measurements with a precision (reproducibility) of 2–5 cm for the components of a 16‐km base line. When this interferometry base line vector was compared with a base line vector obtained by a conventional survey, the two measurements differed by 2–15 cm in the various components. Probable causes of this vector discrepancy are discussed, with particular emphasis on the difficulties encountered in the comparison of a purely geometric interferometry base line vector with a geoid‐referenced survey base line.