Detonations at the depth of the sound channel axis off Perth, Australia were recorded on Bermuda hydrophones at a 178°.2 range (180° is antipodal). The analysis by Shockley et al. of this 1960 transmission experiment allows for the geographic variation in the sound speed profile along the great circle path. The agreement between measured and computed travel times is within 10 s.
We have modified the Shockley et al. analysis by allowing for Earth flattening and lateral refraction. The appropriate path on an ellipsoidal Earth is the geodesic, and this differs significantly from the put circle for nearly antipodal ranges. The southernmost point of the geodesic is at 52°1 S as compared to 47°.3 S for the great circle, and the geodesic travel time lags the great circle travel time by an unacceptable 34 s on account of the cold, slow waters at high southern latitudes. The effect of lateral refraction is in the opposite sense: the appropriate refracted ray path is northward of the great circle; in fact it intersects the African continent so that there is no direct Perth to Bermuda path: Bermuda is in the shadow. We find that a scattered path just rounding the tip of South Africa, thus consisting of one refracted ray from Perth to Cape Agulhas and another ray from Cape Agulhas to Bermuda (with a 3° kink at Cape Agulhas), has a travel time within a Few seconds of the measured 13 000 s travel time. There remains the question of how enough acoustic energy gets into the shadow zone to account for the measured intensity at Bermuda (which is poorly known). Diffraction is too weak. Bottom scatter and bottom refraction off Cape Agalhas may be significant. A continuous diffusion into the acoustic shadow by internal wave inhomogeneities is a more likely candidate. We favor, as an explanation, perturbations in lateral refraction associated with the intense mesoscale activity in the area of the Agulhas retroflection. The conclusion is that global acoustic transmissions are very sensitive to oceanographic conditions.