GRF broadband array recordings provide strong support for the existence of anomalous P-coda phases along the Kuriles-European path first observed using WWSSN data. The data suggest that the phase, termed here PdP, arrives 3--5 s after the direct P-wave and has a slowness 0.7--0.8 s/deg smaller. The latter measurement precludes PdP's misidentification as a source stopping phase, and given the magnitude of the slowness difference, slab multipathing and diffraction are equally unlikely. Because the phase is clearly not observed for all Kurile events and from timing and geometric considerations, crustal reverberations are also untenable as an explanation for PdP's appearance. The hypothesis which best explains the PdP arrival time, arrival angle, and waveform is the presence of a 3% velocity jump in the lower mantle approximately 290 km above the core-mantle boundary. Because of the failure of other studies to observe PdP under favorable recording conditions, the reflector must be of limited regional extent beneath Northern Siberia. ¿ American Geophysical Union 1990