THE PAUMOTUS. 23 



from five to six fathoms deep. Some of the smaller islets we visited, among 

 which are Tikei, Aki-Aki, and Nukutavake, have no lagoons. The former 

 has a small shallow sink in which fresh water collects, but the rim is only 

 very slightly higher than the interior. The last two islands are apparently 

 depressed in the centre, three to four feet below the outer bank of sand which 

 forms the rim (about 10 to 12 feet high) of the basin of the island ; I was 

 at first inclined to look upon them as examples of islands which had been 

 cut down to the level of the sea and subsequently been built up by beach 

 rock and sand in the manner described above. The existence of extensive 

 sand dunes on two sides of the island at Pinaki, and of large dunes (esti- 

 mated to be 35 feet high) on the south shore of Nukutavake, seems to 

 indicate the possibility of there having been a shallow lagoon occupying 

 the centre of Aki-Aki and of Nukutavake, and that these lagoons were 

 gradually filled by the sand dunes, much as Pinaki is filling now. 



At Pinaki, as at other atolls and islets to the eastward, there are 

 fewer large trees than on the western atolls, tlie vegetation consisting 

 in great part of pandanus and putu trees and the usual coral-reef vege- 

 tation of the Paumotus ' and Fijis. 



The only atoll we have seen in the Paumotus the lagoon of which 

 is entirely shut off from the sea is Niau. In this case the old ledge 

 forming the land rim which surrounds the nearly circular lagoon is about 

 a third of a mile in width and sufficiently high, 15 to 20 feet, to pre- 

 vent any sea from having access to it except in case of a cyclone, as 

 that of 1878, when the sea washed into the lagoon. The lagoon is 

 shallow, of an average depth of about three fathoms, the deeper parts per- 

 haps five. The sea, however, undoubtedly finds its way into the lagoon 

 through the cavernous rim. 



Dana, and other writers on coral reefs, mention a great number of 

 lagoons as being absolutely shut off from the sea. I take it these 

 statements are due to their descriptions being taken from charts, many 

 of which, as in the case of the Paumotus, are very defective. For 

 nothing is easier than to pass unnoticed even at a short distance the wide or 

 narrow cuts which allow in so many cases the freest access to the sea to 



1 See Dana, Curals ami Coral Islands, 3d ed., p. 32G. 



