AKI-AKI. 117 



sand. The island slopes slightly from the south end towards the north 

 side, where the shingle is coarser, and coarsest at the northeast point. Ac- 

 cording to Mr. Townsend, the island is slightly dished towards the centre, 

 occupied in part by a small sink of fresh water a little way in from the 

 beach. The central part of this so-called lagoon is covered by low brush 

 (PI. 69, fig. 1), and surrounded by large trees on the outer edge close to the 

 beach ridge, a short distance inland. 



The reef platform is from 75 to 100 feet wide, very flat, with a compara- 

 tively low outer edge. On the southeast side there is a continuous beach 

 rock ledge with a wider reef flat platform than on the north side ; it is from 

 150 to 200 feet wide. 



The inner part of the island is about two feet lower than the outer edge. 

 A part of it is bare or is covered with fragments of broken coral gradually 

 passing into a rich black soil covered with grass, and large hard-wood trees 

 (PL 69) as well as Pandanus, together with bushes similar to those edging 

 the outer rim of the island. Part of the island is also covered with small 

 coral-boulders thrown into the interior while the island was building up. 



Recent beach rock and conglomerate as well as parts of the old ledge crop 

 out in the interior, where the natives have dug water holes. A shallow sink 

 exists (PI. 69, fig. 2), formed undoubtedly, as in all limestone areas, by solution, 

 which has carried off more or less of the material cemented together as beach 

 rock. The island undoubtedly owes its shape to the throwing up of the 

 beaches from different directions upon the original plateau underlying the 

 island, as the soil of the interior is made up of vegetable matter and de- 

 composed coral fragments or boulders, among which can be recognized 

 Madrepores, large masses of Porites, Pocillipores, and Astrseans. In the 

 middle of the island are found the largest trees and the best soil. As is 

 shown by the structure of this island, the existence of alternate layers of 

 beach I'ock and of recent conglomerate or of coral rubble does not by any 

 means indicate a talus ; it merely indicates a reef flat over which the sea 

 has either been building up beaches, or cutting down ledges, or adding masses 

 of coral sand or shingle. This may explain much of the character of the 

 successive layers of rock met with by boring through the upper strata at 

 Funafuti, said to have been due to subsidence. 



