SECT. 1] SUB-OCEANIC STRUCTTJRAL EXPLORATION BY SEISMIC SURFACE WAVES 119 



Data on long-period Love waves affected primarily by the mantle are avail- 

 able from a variety of sources. The term "G-wave" frequently appears in this 

 connection and is used to describe a long-period Love wave of large amplitude 

 and little dispersion propagating over oceanic jaaths. There is no intent, how- 

 ever, in using the term G-wave to imply a type of propagation mechanism 

 other than that of the Love wave. 



Sato (1958) analyzed multiple long-period Love-wave arrivals from several 

 large earthquakes recorded on the Pasadena strain seismometer and found 

 Fourier components with periods between 50 and 300 sec, all with group 

 velocities of about 4.4 km/sec. Sato's results and other studies strongly suggest 

 that Love waves with periods between 150 and 300 sec are dispersed very little 

 in the cases of both oceanic and continental paths. Press (1959), Kovach (1959), 

 Bath and Vogel (1957), and Ewing, Jardetzky and Press (1957, p. 212) published 

 data on continental Love waves indicating that oceanic and continental Love- 

 wave dispersion curves generally differ at all jDeriods less than at least 100 sec. 



Long-period oceanic Love-wave dispersion was analyzed by Landisman, Sato 

 and Ewing (1958) using a theoretical model consisting of two homogeneous 

 surface layers underlain by a layer with constant gradient in rigidity. The 

 upper layer was taken to represent the oceanic crust and the intermediate 

 layer the upper mantle. The model giving the best fit to Sato's data and other 

 data is shown in Fig. 7 as curve C. The indicated decrease in shear velocity at 

 a depth of 50 km is shallower and of greater magnitude than that derived by 

 Gutenberg and by Lehmann (Fig. 7, curve A) for various continental areas on 

 the basis of body-wave travel times. 



Results on the existence and the great extent of the low-velocity layer under 

 oceans were confirmed and elaborated by Dorman, Ewing and Oliver (1960), 

 who found a structural configuration consistent with the observed Rayleigh- 

 wave group-velocity dispersion curve. This work was based on machine calcula- 

 tions according to the multilayer method (see Dorman, 1962). The crust and the 

 mantle to a depth of 2800 km were represented by about 35 homogeneous 

 layers in these calculations. The experimental study was limited to obtaining 

 a solution for the shear- velocity distribution in the mantle. The effect of density 

 variations was fixed by using Bullen's model A for density and the effects of 

 variations in crustal structure were eliminated by using the same crustal layers 

 based on seismic refraction results throughout the calculations. The oceanic 

 Rayleigh-wave dispersion curve of Sutton (in litt.) for periods less than 140 sec 

 and the mantle Rayleigh-wave dispersion data of Ewing and Press for longer 

 periods were used as data for comparison with theoretical calculations. The 

 final approximation of mantle shear- velocity distribution is shown in Fig. 7, 

 curve B. The dispersion curve computed from this model is shown with the 

 observed data in Fig. 6. The quality of agreement between the experimental 

 and theoretical curves over a broad spectrum lends weight to this solution. 

 Deviation of the theoretical curve from the experimental curve for periods 

 greater than about 200 sec is of the sort expected from the use of a flat-earth 

 calculation. Fig. 7 shows by comparison of the oceanic solution (curve B) with 



