456 



SCIENCE 



[N. S. Vol. XLI. No. 1056 



but without a central island, are given an- 

 other name. 



The testimony of the second group of wit- 

 nesses — massive elevated reefs such as occur on 

 certain Fiji Islands — convinced me that Dar- 

 win's theory of siibsidence gives the only satis- 

 factory explanation of the origin of such reefs 

 also; for their limestones rest unconformably 

 on the normally eroded surface of a preexistent 

 foundation. The erosion of the foundation 

 surface shows that it stood above sea-level 

 before the reef was deposited upon it; and the 

 occurrence of the reef shows that the eroded 

 foundation subsided to receive its marine cover. 

 Only after this subsidence was the compound 

 mass uplifted. The mere occurrence of ele- 

 vated reefs above sea level does not for a 

 moment prove that they were formed during 

 the emergence of their foundation. 



All the still-stand theories of barrier reefs — 

 that is, all the theories which involve a fixed 

 relation of the reef foundation to the sea level 

 during the formation of the reef mass — are 

 excluded by evidence of submergence found in 

 the embayed shore lines of the central islands 

 within barrier reefs. It may seem overbold 

 thus at a stroke to set aside several well-known 

 theories, accepted by experienced observers; 

 and so indeed it would be if these observers 

 had discussed the features of the embayed 

 central islands and had explicitly shown that 

 their embayments are not due to submergence, 

 but to some other cause. It is, however, a 

 regrettable fact that the observers who adopted 

 one or another of the still-stand theories took, 

 like Darwin himself, practically no account of 

 the embayed central islands, essential as the 

 testimony of these islands is in the solution 

 of the coral-reef problem. Such neglect is 

 all the more remarkable in view of the clear 

 statement, long ago published by Dana, regard- 

 ing the pertinence and the value of the testi- 

 mony afforded by the central islands of barrier 

 reefs. 



The glacial-control theory of coral reefs, re- 

 cently elaborated by Daly with special refer- 

 ence to the lagoons of atolls, will not hold for 

 barrier reefs. This theory assumes that no sub- 

 sidence of the reef foundations took place. 



and explains the lagoon floors of atolls as plat- 

 forms abraded across preglacial sea-level reef- 

 masses by the lowered and chilled sea of the 

 glacial period after the corals were killed; the 

 preglacial reef-masses having been formed by 

 upward or outward growth on their still-stand- 

 ing foundations. It then explains the en- 

 circling reefs which now surround the lagoons 

 as having been built up while the sea was 

 rising and warming in postglacial time. But 

 if the broad lagoons of large atolls, 20 or 30 

 miles in diameter, were thus formed, the cen- 

 tral islands within narrow-lagoon barrier reefs 

 should be cliffed all around their shore line, and 

 they are not. Furthermore, this theory ex- 

 plains the embayments of central islands 

 within barrier reefs as occupying new-cut 

 valleys that were eroded during the glacial 

 period of lowered sea level; but if this were 

 the case, the new-cut valleys should be pro- 

 longed upstream from the embayment heads 

 as incisions in the floors of preglacial valleys, 

 thus producing a " valley-in-valley " landscape ; 

 and this is not true in any one of the hun- 

 dreds of embayments seen during the past 

 year. Furthermore, many of the embayments 

 are so wide that, if they were opened by slow 

 subaerial processes, all the spur ends ought to 

 have been well cliffed by the sea; yet, as above 

 stated, they are not cliffed. Finally many of 

 the embayments are too wide to have been 

 eroded during the last glacial epoch, or even 

 during all the glacial epochs of the entire 

 glacial period, if the valleys of the formerly 

 glaciated volcanoes of central France are taken 

 as standards of the amount of erosion that 

 could be accomplished in such masses during 

 such intervals of time. The glacial-control 

 theory thus proves incompetent to explain 

 barrier reefs, and it is therefore held to be 

 generally incompetent to explain atolls also; 

 it may have more importance on the borders of 

 the coral zone, where the corals would most 

 likely have been killed during the glacial period : 

 the Marquesas Islands promise interesting 

 results in this connection. The glacial-control 

 theory has its greatest importance in conjunc- 

 tion with Darwin's theory of subsidence, for 

 submergence during subsidence may have been 



