418 J. STANLEY GARDINER, 



by a piling-up of sand, such as has formed the kuli of Midu-Huludu. Further their serial nature 

 along the centre of a very narrow island is scarcely consistent with the second view. Lastly, in 

 the centre of Putali the sea has broken its way through the outer line of rock into two of the 

 pools; the damage, however, had been repaired by the natives with large stones. 



Wiringili. The two islands marked on the chart have become joined by a sand-bank, the 

 trees on which are still small. The island is stated by the natives to have a small kuli behind 

 its E. point. The seaward reef is much strewn with boulders, which have accumulated in a bank 

 with three green shrubs at the S. end. The seaward points of the reef are notably shallow for 

 some distance beyond the reef edge. 



Bushy islet to the north of the atoll now consists of a single island covered with shrubs 

 and two sand-banks. 



The east islands show five in the series instead of four as represented. This is due to 

 the presence of a small island between Maradu and Ruehi-hura, which was probably accidentally 

 joined to the latter in the original survey. The islands are all formed of sand on their lagoon 

 side, which in some places shows an extension, and in others a washing-away. All are slightly 

 lower in the centre but without definite kuli, although mangroves grow in some. All have more 

 or less stony seaward beaches, but the presence of a rocky area depends on the distance of tiieir 

 shores from the reef edge. Off Hangadu the reef is about 150 yards broad, and the island 

 has a narrow, rocky line. N. of this the reef narrows, becoming first quite similar to the reef 

 off Hera ; and then off' Hitadu the land extending to the inner edge of the boulder zone. The 

 reef-flat off these parts has no pinnacles or other rocks, being quite smooth; the rocky area in 

 the land is always well-defined. The timber, with which the latter is clothed, rises for some 

 distance back from the beach, suggesting that it is growing outwards by additions from its reefs. 

 At the W. point of Hitadu this is undoubtedly the case, as also along the N. side, but it is 

 less marked to the S. A characteristic feature of Addu, most noticeable here, is the large amount 

 of sand found with the rocks on the beach in places, the whole appearing to be mainly formed 

 by it. On this side of Hitadu, however, we saw no trace of a beach sandstone formation. Off 

 the reef edge the slope is extremely gentle, the sea breaking in a gale over 100 yards outside 

 the reef edge off the W. point of the atoll. 



Maradu is separated from Hangadu by a pool of water 2 fathoms deep, although both to 

 the E. and W. their reefs are continuous. This pool is evidently filling up, consequent, I think, 

 on the reef against the lagoon having reached the low-tide level. There are traces here of a 

 former passage into tiie atoll-lagoon in a less consolidation of both the seaward and lagoon edges 

 of the reef, the latter being still very imperfect and the former having practically no boulder 

 zone. The island of Maradu itself shows growth along its reefs, as do Faidu and Gan. This 

 should ultimately join all tliese islands together, but the growth may only be temporary. Further 

 N. by Ruehi-hura there were signs of an irruption of the sea across the land, but no definite 

 gap was created. 



The reef off Maradu has a high crest, nuUipore-covered edge, but relatively small fissures. 

 The reef-flat is about 40 yards broad, and behind this there is a boulder zone about half as 

 broad, with a boat channel, having everywhere at least a breadth of 90 yards to the land. The 

 boulder zone is low, covered by small loose rocks, and is simply the inner part of the reef-flat. 

 It has here no connection with any elevation of the atoll. Further in, the island of Maradu is 

 all of sand formation, a little beach rock indicating a former washing-away of its eastern shores, 

 which now seem almost stationary. 



Faidu and Gan resemble Maradu, but Gan runs out into a great flat, half-a-mile broad, 

 across which lines of rock extend out to the boulder zone. The latter is in this part of the 



