GLACIAL-CONTROL THEORY OF CORAL REEFS. 225 



area, where evidences of crustal uneasiness during the later stages of 

 the Cenozoic are independent of the drowned-valley criterion. 



As shown by the charts, island embayments of the class here con- 

 sidered have everywhere depths no greater than those anticipated if 

 the bays are due to subaerial erosion during the Pleistocene time of 

 lowered sea-level. In spite of aggradation by waves and currents, 

 it seems inevitable that a few of the larger valleys should have greater 

 depths, if their drowning were caused by crustal sinking in the Recent 

 period. The failure of these greater depths tends from the first to 

 weaken Dana's criterion. 



According to the subsidence theory, the still unsubmerged portions 

 of the valleys should be of the one-cycle type, unless post-Glacial 

 sinking were too slow for the complete submergence of the sides of the 

 inner valleys which were excavated below the Tertiary floors because 

 of the lowering of base-level in the Glacial period. According to the 

 Glacial-control theory, the drowned valleys in general should be of the 

 two-cycle type, or should have been of that type at some stage since 

 the first climax of glaciation. 



In attempting to test the two theories by the use of this principle, 

 it must be remembered that in volcanic islands a broad erosion valley 

 is not necessarily as old as an equally broad valley cut in non-volcanic 

 rocks. Observers in the Hawaiian islands, for example, cannot fail 

 to be impressed with the breadth, as well as depth, of obviously youth- 

 ful valleys cut in the volcanic formations. An abundant development 

 of amphitheatres and broad troughs is an early result of subaerial 

 attack on the typical volcanic island. The causes for such abnormal 

 physiographic development are not wholly understood. One of them 

 may well be a special tendency to caving or landsliding along the walls 

 of young valleys cut in volcanic masses. The latter are characteristi- 

 cally interrupted by zones of weak ash-beds or weak scoriaceous phases 

 of the lava flows — layers liable to water-soaking, with consequent 

 danger of landslides. A second cause for rapid slumping is found 

 in the unusual prevalence of vertical joints in massive lavas. Widen- 

 ing of the Pleistocene gorges cut in the central islands may therefore 

 be more pronounced than in the case of gorges simultaneously cut in 

 continental rocks. Whatever be the reasons, valley-making in vol- 

 canic islands gives results somewhat different from those usual in 

 typical areas of the continents. The physiographic processes operat- 

 ing on the oceanic volcanoes certainly need thorough study. At least 

 until that is accomplished, it is not advisable to apply the continental 

 chronometer to the islands. There, many broad valleys seem to be 



