W. M. Davis — Shaler Memorial Study of Coral Beefs. 241 



glacial valley. This special condition might obtain on some 

 islands, but not on all ; yet in no case that came under my 

 observation were the erabayments continued inland by " valley- 

 in-valley " forms of two-cycle origin. Every embayment that 

 I saw — and their total number must be many hundred — occu- 

 pied merely the submerged distal part, C, C, fig. 1, of a 

 maturely-opened, one-cycle valley or valley system, of which 

 the branching heads reached up to the mountain crest without 

 any indication of revival by a relative depression of baselevel. 

 In no case was a narrow incised valley prolonged upstream 

 from a bayhead, as should happen if the lowered sea-stand 

 had not endured long enough for the waves to cliff the spur 

 ends; and it must not be supposed that the incised valley is 

 filled again with alluvium, for that result could not be reached 

 until the whole bay is filled with a delta plain, sloping at the 

 same angle as the preglacial floor. 



If it be supposed, on the other hand, that the glacial period 

 was relatively long, then the deepened valleys might have 

 everywhere been widened so far as to destroy all traces of pre- 

 glacial forms ; but in this case again the spur-ends of narrow- 

 lagoon islands must have been truncated in wave-cut cliffs, 

 D, D, fig .5. The assumption that the sea would have time to 

 cut cliffs around a narrow-lagoon central island while the small- 

 stream valleys of the island were becoming maturely opened, 

 seems to be well supported by general principles, for the 

 widening of the valleys of small streams is a slow process com- 

 pared to the cutting back of shore cliffs by sea waves ; but the 

 assumption is further warranted by the special cases of Tahiti 

 and New Caledonia, inasmuch as the high sea cliffs of those 

 islands, the production of which involved a strong recession of 

 the shore line in very resistant rocks, appear to have been 

 developed in the same period of time (while the sea surface 

 stood relatively lower than now) that witnessed the mature 

 widening of the large valleys in the same resistant rocks. 

 Furthermore, volcanic islands outside of the torrid zone, such as 

 Tristan d'Acunha in the South Atlantic, have had their cliffs 

 cut back faster than small valleys are cut down and hence 

 much faster than such valleys are widened, so that their streams 

 cascade into the sea ; the same is true on the northeast side of 

 the island of Hawaii. All this seems to show that if the gla- 

 cial lowering of sea level endured long enough for an island, 

 then undefended by living reefs, to suffer mature dissection by 

 small streams, its shore must at the same time have been cut back 

 in mature cliffs by the waves. Wave attack would be much 

 stronger than valley weathering ; hence if drowned valleys are 

 one or two miles wide, the spur ends between them ought to be 

 cut far back in high cliffs : this must be true whether the island 



