In previous work, detonation propagation in tubes has been observed to occur at velocities well below the ideal Chapman-Jouguet velocity (UCJ). Such velocity deficits are thought to result from the presence of nonideal boundary conditions, which are not typically accounted for in theory. The growth of viscous and thermal boundary layers serve to remove energy from the reacting flow through frictional dissipation and heat losses to the cooler tube wall. In situations where the reaction zone length is small relative to the tube radius, these losses have a limited effect on detonation, presumably because the sonic surface quickly isolates them from the chemically reacting gas driving the front. However, as mixture sensitivity decreases, the spacing between the shock front and the sonic surface grows and allows the loss mechanisms increased time to act on the reaction zone. The resulting decreased detonation velocities, termed “sub-CJ detonations,” are observed in marginal conditions and range from 0.8–1.0UCJ before failure of the detonation occurs. But in extreme cases, “low velocity detonations” with average velocities as low as 0.5UCJ have been recorded [1, 2, 3, 4]. Other researchers have used microwave interferometry to obtain instantaneous velocity measurements showing that these low velocities are actually due to a combination of highly unsteady stuttering or galloping waves [2, 4]. However, due to the long wavelengths associated with these oscillatory modes, it is not clear if they are a persistent phenomena or simply a transient pathway to complete failure. Furthermore, the scaling of these modes is unknown since most measurements to date have been performed in tubes with identical 38-mm inner diameters (IDs). To address these issues, we have performed studies of detonation propagation under varying conditions. Results are presented for detonations in propane and hydrogen mixtures with tube IDs ranging from 1.27–6.35 mm and tube length-to-diameter ratios (L/D) ranging from 194–10,350. Average velocity measurements obtained from transducer transit times and high-resolution velocity measurements obtained from a high-speed camera are reported.