Comparative experiments, in which either sample length or transducer configuration is varied, provide a reliable, accurate, and model-independent means of explicitly correcting ultrasonically determined elastic wave velocities, for transducer-bond phase shifts. The results of a numer of such comparative experiments, employing both lithium niobate and quartz transducers on fused silica and aluminium samples, indicate that an absolute accuracy of order ¿1 ns in travel time is routinely attainable. The use of these correction procedures in future ultrasonic studies should greatly improve the reliability of measured pressure and temperature derivatives of mineral elastic moduli and thereby facilitate more meaningful interpretation of seismic profiles for the earth's interior.