118 



OIJVKR AND DOHMAN 



[chap. 8 



with periods as long as al)oiit 500 sec. Their data are show n in Fig. 6. Later, 

 Benioff and Press (1958) extended this range to waves with periods greater 

 than 600 sec. The large Chilean shock of 22 May, 1960, produced waves of even 

 longer periods. ])erhaps as great as about one hour. l)ut analysis is not yet 

 coni])lete. Other data on mantle Kayleigh-wave group velocities for purely 

 oceanic ])atlis were given by Sutton {in lift. ; see data in Fig. 6), Benioff and 

 Press (1958) and Kovach (1959). 







50 



100 



150 200 



Period (sec) 



250 



300 



350 



Fig. 6. Observed and theoretical Rayleigh-wave dispersion in the range of periods in- 

 fluenced primarily by the upper mantle. Solid dots are group -velocity data of Ewing 

 and Press derived from multiple paths around the earth. Open circles are data of 

 Sutton for paths across deep basins of the Pacific. The solid curves are computations 

 of Dorman, Ewing and Oliver on the basis of a layered model of the oceanic crust and 

 mantle (curve B, Fig. 7). The upper curve is phase velocity, the lower curve group 

 velocity. (Sood agreement between theoretical and observed dispersion is evidence 

 supporting the existence of a shear-velocity distribution beneath the oceans very 

 similar to curve B. 



Data on very long-period waves are usually obtained for very long paths 

 which include both oceanic and continental segments. Ewing and Press, as 

 well as subsequent investigators, detected no systematic differences in grouj) 

 velocities for these two kinds of paths. Preliminary measiu'ements of phase 

 velocity likewise showed no differences between continents and oceans in this 

 resi)ect (Nafe and Brune, 1960). More recent and more precise measurements of 

 phase velocity by Brune {in lift.) do indicate a small but measurable difference 

 in such cases, extending to jjeriods at least as great as about 400 sec. However, 

 the analysis of these data in terms of earth structure is not yet complete, and, 

 in any case, is not likely to affect significantly the results presented here. 



