The submerged terraces within a few hundred feet below present sea level can probably best be explained as ancient beaches formed during the lower sea levels of the Pleistocene. The fact that the same levels are found along coasts of diverse geology and tectonic development supports the eustatic origin of terraces between sea level and 70-100 fathoms. The gradients of the continental shelf are so low and the benches are so persistent that block faulting and slump scars are excluded as general explanations. The benches of the continental slope extend to depths of 1500 fathoms and vary in depth from point to point along the continental slope. These cannot be Pleistocene eustatic levels unless we consider that they were formed prior to recent large crustal deformations. Again the persistence of the benches for many miles argues against a fault-scarp or slump-scar hypothesis. Thus, while the benches of the continental shelf are probably ancient beaches, particularly those traced at the same depth for thousands of miles, the benches of the continental slope are probably rock benches, while some may represent step faulting. SUBMERGED BEACHES ON THE CONTINENTAL SHELF: In Table 3 the depths of terraces or persistent levels of the continental shelf are listed for selected points in the North Atlantic. There is a remarkable uniformity in these data; the same levels are found near Newfoundland, in Florida, and on oceanic islands far from the glaciated areas. On the basis of data obtained in the North Atlantic it is not possible to date the different terraces, but probably most were formed in the period between 12,000 and 5,000 years B.P. when the sea rose in consequence of the melting of the Wisconsin glaciers. Coring and dredging on these submerged ancient beaches could probably produce material datable by the radiocarbon method. TABLE 3.--_Depth (in fathoms) of prominent continental-shelf terraces_ Each column based on only one nonprecision echogram