164 J. STANLEY GARDINER. 
At the west end this belt runs out into a horn, and its continuation is shown by a series 
of rocks which form a line round the sand-flat, the only open part, never at any time 
enclosed by such rocks, being to the 
north-east. The central part of the island 
is entirely of sand-formation and now 
washing away. In the past it was clearly 
formed by the piling up of sand through 
the north-east opening, a horn now ex- 
tending in this direction. Between the 
sand and rock a narrow pool was enclosed, 
stretching right along the island from east 
to west. represented now by a mangrove 
swamp with a pool of water (barachois or 
kul) to the west. The line of rocks 
around the sand-flat is not situated on 
the reef itself, which is formed of great 
coral colonies with deep pools and channels Fic. 30. Maduwari Island and reef. The rocky area is diagonally 


3-4 Feet of 

Water 
of water between, but well within it. The shaded. 4. Mangrove swamp. JB. Pool of water or kuli. 
reef is h c oa 380 oma C. Boat-channel through reef. D. Rocks represented in 
ree Is suc as 1S grow ing up anyw ere Plate XII. 
within the bank, and of recent formation, 
while these rocks are undoubtedly the remains of the original reef which was elevated. 
Their foundation is a flat of rock, fairly bare of mud and sand, so that corals may grow 
on it. The masses stand up to a height of 7 or 8 feet on this. They are formed of 
coral rock of recent formation, and are undermined and pitted all over. In the drawing 
(Plate XII.) the end 48 feet of the longest mass (66 feet) is represented; its top is never 
covered even at the highest tides. Here is an undoubted case of an island having originally 
been a nearly-closed crescentic reef. Its old lagoon was largely if not completely filled in 
with sand, but, owing to the erosion of the rocky protective part, most of this has now 
been removed. The reef has grown out on every side, and a pool is being formed in the 
centre by solution, the old reef having been dissolved away to a depth of a few feet. 
In North Mahlos there is no case of any island which so strikingly tells its own 
history, but there are many similar islands and reefs to the east of Miladumadulu. Yet 
the islands of the east side of North Mahlos show stages in the washing away of the land 
and the formation of great broad pools between it and the reef. Round Hurudu and 
Raskateen the reef closely fringes the land. Inguradu has a sand-flat to the north-east, 
just covered with water at low tide. Kenurus has a broad sand-flat on either side of the 
neck and a pool to the north-west, each slightly hollowed out below the level of the 
reef-flat. Fainu has a channel 3 feet deep along the whole of the north side, and off 
Rasmadu a similar one extends nearly completely round the land. Duwafuri has the channel 
cut across in places by lines of beach sandstone but otherwise perfect, varying up to 
2 fathoms in depth in a pool to the north-west. Wahdu again has a great broad flat to 
the west with 1 to 2 fathoms or more of water, but this reef shows a much later stage 
in the presence of a deep velu (8 fathoms) in the centre. Nevertheless only a tendency 
to form deep pools of water pari passu with the washing away of the land is clear. 
The changes taking place are better exemplified on the reefs within the bank, where 
