GEOLOGY AND PALEONTOLOGY OF THE CANAL ZONE. 313 



carried by the streams to the sea prior to the submergence after 

 which the hving reefs have formed ? Unless sediment was deUvered 

 to the sea so rapidly that a coastal plain pushed forward beyond the 

 interstream divides as to protect them from attack by the sea, their 

 seawiird ends should have been cliffed, should the flat have been 

 formed in the manner suggested. What are the submarine profiles 

 off the spur ends? Are there submerged cliffs at the divide tips? 

 One of Agassiz's illustrations* represents a cliff of considerable height 

 at one place on the shore of Maupiti. In my opinion sufficient evi- 

 dence is not available to establish how the reef flats of these islands 

 were formed, and they may be made to accord with whatever theory 

 of reef-flat formation an author may prefer. Should it ultimately 

 be proved that these barrier reefs accord with the Darwinian hypothe- 



FiG. 21.— Diagram to show how a unear reef lying across the wd^d is formed into a horseshoe. 

 (After hedley and Griffith taylok.) 



sis, a few instances in which that hypothesis applies will have been 



discovered. 



Atolls. 



There are two kinds of atolls:- Those of the first kind rise above 

 relatively shoal-water platforms, and are represented by the atolls of 

 the Great Barrier Reef of Australia, those of the Floridian reef-tract, 

 and the faros of the Maldives. That there was never any central land 

 area for these atolls is perfectly obvious. Hedley and Grifiith Taylor, 

 in their paper, cited on pages 245, 251, have shown how the atolls 

 along the Great Barrier have been shaped by the prevalent, mostly 

 wind-induced, currents; and I have shown in my papere on the Mar- 

 quesas and Tortugas atolls that precisely the same principles apply 

 to them. The principles involved are illustrated by the accompany- 



1 The Coral Reefs of the Tropical Pacific, pi. 101, fig. 4. 



