Is there a first-order discontinuity in the lowermost mantle?

In 1983, Lay and Helmberger [Geophys. J. R. Astron. Soc. 75 (1983) 799-837] reported the detection of a precursor to the seismic phase ScS. They attributed this precursor to a sharp seismic discontinuity located several hundred kilometers above the core-mantle boundary. Such a lowermost mantle discontinuity implies the existence of a sharp phase change or a chemical boundary. Precursors to ScS and, less frequently, PcP have since been observed in numerous locations, but are not a global phenomenon. Frequently, PcP precursors are weak or absent when ScS precursors are observed in the same location, and vice versa. There can be significant variations in the amplitude and arrival time of the precursor relative to the main phase. The presence or absence of these precursors has led to speculations about the nature of the lowermost mantle. Here we demonstrate that ScS or PcP precursors may be produced by gradients in seismic wave speed associated with large-scale lowermost mantle heterogeneity. Rather than a phase or chemical boundary with substantial topography, such gradients require lateral variations in temperature and, close to the core-mantle boundary, composition.