This paper presents an experimental test of the multi-stage patent race model of Harris and Vickers (Rev. Econ. Stud. 54 (1987) 1). Tied competitors invested more when nearer the end of the race, but this may have followed from a more general and unexplained pattern of a positive correlation between investment and progress in the race. The relationship between investment and gap between competitors was mostly not as predicted. Also, leaders did not invest more than (not heavily lagging) followers, and the race did not approach monopoly as the gap between competitors widened. Overall, the evidence brings only limited support to the theory.