The ScSn phase-equalization and stacking algorithm of Jordan and Sipkin (1977) has been applied to an extensive set of HGLP and ASRO data to obtain regionalized estimates of QScS. Tests of the algorithm using synthetic data reveal no significant sources of bias. The low value of QScS previously obtained for the western Pacific (156 ± 13) is corroborated by additional data, and QScS observations in other regions correlate with variations in crustal age and tectonic type. A representative value for the ocean basins sampled by our data is 150, with the best estimates being somewhat lower (135 to 142) for younger oceanic regions and somewhat higher (155 to 184) for older regions. The two subduction zones sampled here, Kuril-Japan and western South America, are characterized by larger QScS estimates than the ocean basins (197 ± 31 and 266 ± 57, respectively), and the difference between them is qualitatively consistent wit the contrasts in upper-mantle attenuation structure proposed by Sacks and Okada (1974). Continental regions are poorly sampled in this study because the signal-generated noise in the vicinity of the ScSn phases is generally larger for continental paths, but a representative value is inferred to be QScS = 225. For paths crossing China, QScS is observed to be lower (∼180), providing additional evidence for a high-temperature upper mantle previously inferred from surface-wave and travel-time measurements. Our best estimate for the average Earth is QScS = 170 (±20 per cent), which appears to be significantly lower than that predicted by normal mode data, suggesting some frequency dependence. Q− 1ScS correlates with ScSn-ScSn−1 travel time along a line given by
where ΔTScS is the JB residual in seconds; this correlation favors a thermal control on the ΔTScS variations. It is inferred from the tectonic correlations that much, if not most, of the heterogeneity expressed in the QScS and ΔTScS variations is confined to the upper mantle. Substantial differences in the attenuation structures underlying continents and oceans are implied. In fact, the average quality factor for the upper mantle beneath stable cratons may not be much less than that for the lower mantle.