GLACIAL-CONTROL THEORY OF CORAL REEFS. 227 



islands in the western Pacific are explained by the sinking of those 

 islands. However, the dating of that subsidence is not yet established, 

 and the actual bays may be due to the Pleistocene cleaning-out of 

 unconsolidated sediments which had been deposited in valleys, 

 drowned during the Tertiary fragmentation of the Australasiatic 

 continent. Secondly, the Glacial period was long enough for some 

 further deepening of the Tertiary valleys by subaerial erosion. Well 

 marked "edges" of the resulting valley-in-valley topography should 

 not appear generally in the central islands of the present day, if the 

 " edges " were formed early in the Glacial period. Thirdly, the narrow 

 rock gorges cut at the heads of the bays should be more or less com- 

 pletely covered by post-Glacial alluvium. 



The drowning of stream valleys is not the only cause for embay- 

 ments. In each case it must be determined whether the bay is due to 

 irregular accumulation of volcanic products, to faulting, to volcanic 

 explosion, or to erosion. In many instances the bays are clearly 

 en axe with valleys cut by streams and are so located because of pre- 

 liminary subaerial erosion. However, such bays may not all represent 

 river valleys submerged by change of sea-level. Ocean waves usually 

 tend to smooth continental coast-lines, faced by broad submarine 

 shelves. The shelves have a double office. They furnish shallow 

 platforms on which coastal detritus may be quickly aggraded to sea- 

 level; and they lower the erosive energy of the surf by partly wearing- 

 out waves from the open ocean. Smoothing of a coast-line is a direct 

 function of offshore aggradation. As the latter is delayed because 

 of great depth of water, the waves have a longer time to search out 

 the weak places in the land mass attacked. The more steep-to the 

 coast, the more powerful are the attacking waves. In both respects 

 undefended volcanic islands, with very deep water close to their 

 shores, are subject to specially rapid searching by the waves. Now 

 the very existence of a main valley in a Pleistocene island implies 

 that its flooring rocks were already somewhat weakened by weather- 

 ing. The volume of rock above sea-level, per unit length of shore 

 line, was smallest at the intersection of the shore line with the 

 valley floor. For two reasons, therefore, the surging breakers must 

 have tended to cut bays in the Pleistocene islands, just as they are 

 now cutting baj's in the Algerian coast, in some parts of the North 

 Atlantic coast, in Christmas island (Indian ocean), and elsewhere. ^° 



60 See T. Fischer, Petermann's Geog. Mitt., p. 1 (1887); C. W. Andrews, 

 Geog. Jour., 13 (1899) (map of Christmas island). 



