This paper describes studies of the travel times of P waves to North American stations from 20° to 96°. The study confirms the existence of the second major discontinuity in the travel-time curve at 24° and shows that there are no other major discontinuities in the travel-time curve up to 96°. The mean interval travel times can be fitted by a quadratic in Δ to within ±0.7 sec. On the average therefore d2T/dΔ2 is almost constant between 24° and 96°. When the times are fitted by a cubic in Δ the deviations are reduced by a factor of two to three, and the deviations are then of the same order as the standard errors of the mean interval times. Although there are no deviations from the cubic larger than would be expected, there are possibilities of minor discontinuities or irregularities in the travel-time curves which may be real, especially one between 53° and 59°.
The major outstanding problem is that the travel times for sets of data in which western events are dominant differ systematically from those for sets in which the distribution in azimuth is reasonably uniform. The differences might arise from bias in the analysis, but this does not seem probable. Thus the possibility exists that there are regional differences in the lower mantle as was suggested by Chinnery and Toksöz. P velocities were calculated using the Herglotz-Wiechert inversion method and are given in tables.