Hydroacoustic arrivals detected by a hydrophone array on the Northern Mid-Atlantic Ridge (MAR) are used to better understand T phases from oceanic crust earthquakes. T-phase events are selected from two study areas: the inside corner and ridge transform intersection at the eastern ends of the Kane (2330N) and Atlantis (30N) transform faults. Both are regions of high relief (5000 m) and events are located throughout the massif and transform valleys. We investigate the spatial distribution of T-phase locations as a function of water depth. We examine the characteristics of T phases as a function of both distance from the event to each hydrophone and event location water depth. Finally, we use a ray-trace model to test for bathymetric blockage along the propagation path of the T phase. We observe that acoustic magnitudes of T phases show no dependence on water depth of event location. This is opposite to predictions from current T-phase generation models which show water depth dependence. There may be a correlation between acoustic magnitude and propagation path topography, but results are sensitive to the sound velocity profile used in the model. Our results underscore the complexity of T-phase generation and propagation.