During February and March 1976 high-frequency Pn,Sn phases from several earthquakes in the distance range 18° to 29° were recorded on magnetic type from an array of bottom mounted hydrophones near Wake Island. Large discrepancies were found between the actual arrival sequences of these phases and those expected for uniform propagation of a plane wave front across the array. Furthermore, the signal envelopes of the wave trains were dissimilar on different hydrophones for any given Pn,Sn phase. Comparisons between arrival sequences on the different hydrophones and between the signal envelopes on a given hydrophone from groups of earthquakes with nearly identical hypocenters indicate that differences in velocity structure near individual hydrophones cannot explain these observations. They are more likely the result of either the mechanism of generation or of long-range propagation for Pn and Sn, or both.