For decades, continental mantle earthquakes were classified by estimated depth — seismic networks triangulate the source and assign it to the crust or mantle based on whether the depth falls above or below the Moho discontinuity. The problem: depth estimates for shallow-to-intermediate earthquakes are imprecise enough that a quake near the boundary could be assigned to either side. The population of “mantle earthquakes” was partly an artifact of measurement uncertainty.
Stanford geophysicists Wang and Klemperer changed the classification method. Instead of estimating depth, they measured the ratio of two wave types: Sn waves that travel along the mantle's top surface and Lg waves that propagate through the crust. A mantle earthquake generates strong Sn waves and weak Lg waves. A crustal earthquake does the reverse. The diagnosis comes from wave propagation characteristics, not from an imprecise depth coordinate.
The result: 459 confirmed continental mantle earthquakes out of 46,000 examined since 1990. Roughly 1%. And they cluster — in the Himalayas and near the Bering Strait — rather than scattering uniformly. Before this method, mantle earthquakes were “rare and hard to detect.” Afterward, they are “rare and regionally concentrated.” The population changed because the method changed. The Earth's seismicity didn't shift; the filter through which it was viewed did.
The structural observation: the measurement method doesn't just refine the map — it produces a different map. Depth estimation creates a fuzzy boundary between crust and mantle events. Wave-type ratios create a sharp one. The two methods don't disagree about individual earthquakes in a way that could be resolved by more data. They disagree about what counts as evidence for location. When you change the evidence criterion, you change which events belong to which category, which changes the spatial pattern, which changes what the map says about the Earth's interior. The method doesn't describe the territory. The method partially constructs it.