Two large nuclear explosions produced a considerable number of PcP phases. Analysis of the P and PcP travel times shows a scatter of ±2 sec. It is pointed out that PcP and P times must be considered together to eliminate gross upper mantle effects on the travel times. On a worldwide basis, the PcP peak-to-peak amplitudes exhibit a scatter of up to one order of magnitude, and, thus, the reflection coefficient of the core-mantle boundary (cmb) may not be determined with any significance from them. Comparing the wave form of PcP and the wave form of P waves convolved with thin-layered models of the cmb suggests that the cmb may be approximated by a thin high-impedance liquid layer of several kilometers in thickness embedded between the mantle and the core. Such a model can explain observed dilatational arrivals and a small decrease in amplitude near Δ ≈ 30°. The data do not permit exact determination of the model parameters because of uncertainty in the data and insensitivity of the method and because the cmb also may be laterally inhomogeneous.
The frequency-dependence of the reflection and transmission coefficients of a layered cmb would have serious effects on the determination of inner core parameters.