Compositional layering of the midmantle has been proposed to account for seismic and geochemical patterns [van der Hilst and Karason, 1999>, and inferred radiogenic heat source concentrations [Kellogg et al., 1999>. Compositional layering would require thermal boundary layers both above and below an interface. We construct a minimal 1-D model of a mid-mantle boundary consistent with the observed nearly adiabatic compressional velocity structure [Dziewonski and Anderson, 1981> and the proposed high heat flow from the lower mantle [Albarede and van der Hilst, 1999; Kellogg et al., 1999>. Ray tracing and reflectivity synthetic seismograms show that a distinct triplication is predicted for short-period P waves. Although topography on a boundary would cause uncertainty in the strength and the range of the triplication, many clear observations would be expected. We examine data from the US West Coast regional networks in the most likely distance range of 60¿ to 70¿ for a 1770-km-depth boundary, and find no evidence for P wave triplications. ¿ 2001 American Geophysical Union