108 TRANSACTIONS OF THE [jAN. 18, 



rential and of brief continuance. But the contour affords 

 ample evidence of surface erosion; rude planes occur on both 

 sides of Harrington sound, which can hardly be explained ex- 

 cept as the result of partial base leveling, while the north shore 

 of Hamilton island for three or four miles shows a gentle slope 

 with rivulet ways, which is distinctly due to surface erosion. 

 At the same time, it is positively certain that the general con- 

 tour of the surface, modified as it has been by surface erosion, 

 was determined originally by underground erosion, the removal 

 of matter by solution as well as by mechanical action and the 

 subsequent collapse of cavern roofs. Indeed, as will be shown 

 later on, it is difficult to avoid the conclusion that to the col- 

 lapse of cavern roofs must be ascribed the initiative of the 

 several sounds and harbors and even of the so-called lagoon en- 

 closed b}^ the outer reef or ledge flats. Let us examine first the 

 contour of the land and afterwards that of the submerged area. 



The Land. 



The characteristic feature of the islands is the " sink," the 

 characteristic feature of limestone regions ever3'where, but 

 especially notable here where the land strips are so narrow. 

 The space between Castle harbor and Harrington sound, one- 

 fourth to one-half mile wide, shows numerous sinks of all grades 

 from mere pits on the hilltops, five or ten feet deep, to broad 

 basins of six or eight acres reaching to within a few feet of tide 

 level or even holding a pond of water. 



A similar condition exists between Harrington sound and the 

 lagoon, a strij) one-tLiird to somewhat more than one-half mile 

 wide, where several ponds were seen, into some of which fishes 

 have passed from the sound. Other sinks do not reach the 

 water and are filled with rich soil. A small basin near the Crawl 

 on the north shore, shows the process by which ba^^s and even 

 large sounds are formed. This, containing several acres and re- 

 sembling the other ponds and sinks in all the ordinary features, 

 is at a few yards from the shore. The waves breached the 

 wall and opened a narrow, shallow strait. A few years ago 

 the roof of the subterranean channel gave way and the strait 

 became deep. Alread}^ cutting between tides is shown around 

 the border. A line of ponds and sinks lies near the road on the 

 other side of Harrington sound, separated from each other by 

 very low divides. The last of them, near the Devil's hole, is at 

 but a rod or two from the sound and in a short time it will be 

 opened, as undercutting is well advanced on the sound side. 

 Sinks are numerous in St. George and St. Davids, but only one 

 was seen which extends below water level. The shore cutting 



