291 
as the defence seems, this shingle beach completely prevents erosion 
though at the same time accretion is probably very slow. In the case 
of the highest and most advanced coral reefs usually the same shino-le 
beach occurs ; from which fact we might conclude that as the initial 
stage of any fringing-reef must have been that of a simple submerged 
sandstone ledge of greater or less extent, we see here the original shingle 
beach, thrown up where this ledge originally became subaerial, to which 
the waves have eaten back over the present raised reef until all the 
sandy soil formed during the earlier “ embankment and pool ” stage has, 
with the vegetation it supported, been swept into the sea. This shingle 
having been reached the erosive action has been checked, and the surer, 
if slower process of shingle accumulation has been initiated or, at all 
events, renewed. From this account of these bays it will be seen that 
the fringing-reef exhibits in some parts a phase more advanced than it 
exhibits in others. But it does not therefore follow that these more 
advanced “ platform ” portions are older than the earlier “ embankment 
and pool ” portions. They cannot, in one sense, be so old, for we must 
suppose that all these reefs commenced contemporaneously, and the 
“ embankment and pool ” reefs are still growing, whereas the “ platform ” 
reefs have now no living coral. The different stages therefore merely 
indicate that the sandstone reefs running out from the headlands in 
which the various ridges end are in different parts of the islands situated 
at different depths, and the condition of the reefs indicates that the 
sandstone ledges are shallower, and that deep water is further from the 
shore towards the south than towards the north end of the islands. At 
quite the southern extremity of Little Coco bare sandstone reefs, too 
shallow for the growth of a coral fringing-reef, stretch away south¬ 
eastward in much the same way as the well-known Alguada reefs extend 
southward off Cape Negrais. On the east coast of Little Coco are high 
coral reefs exposed at low-tide, fringed by a coral-shingle beach, while 
towards the north end of the island are similar high reefs fringed by a 
shore of sandy soil which, with the beach-forest growing on it, is being 
washed away by the sea. On the west coast, where the reefs are high, 
and, though still in the “ pool ” stage appear from their jagged edges 
to be approaching the “ platform ” stage, a line of low sand-dunes, per- 
' haps the highest development of the epoch of sand-accretion, have been 
thrown up; these at present protect the shore and have actually closed 
up, at the south-west corner, the mouth of a mangrove-creek. 
Similarly, in Great Coco, near the southern extremity and between 
the main island and Jerry there is a lai’ge bare sandstone reef which ex¬ 
hibits very well the arrangement and dip of the strata; further up the 
east coast denudation is going on, still further north the site of a beach- 
101 
