FRYER— FORMATION OF ALDABRA, ETC. 439 



The lagoon of Astove is a primary lagoon, and that of Aldabra is secondary and 

 therefore the two atolls should perhaps be placed in different classes, though they are 

 in much the same stage. 



The vegetation on each is of a varied type. 



Stage 2. Cosmoledo Atoll is characterised by an imperfect land-rim, less than half 

 of its circumference being capped with land. The atoll circumference between the islands 

 is formed by a reef awash which consists of rubble and debris piled on elevated coral rock, 

 but in places the edge of this reef is covered with flourishing lithothamnia and is perhaps 

 beginning to grow seawards. 



The land, which once almost encircled the atoll, has been cut up by lagoon erosion 

 into the islands which exist at present. The latter consist of elevated coral rock, which 

 may be uncovered as at Aldabra — and in this case are obviously suffering great erosion on 

 both faces — or may be so piled with sand as to be almost hidden, in which case gain 

 or loss are difficult to determine*. 



The lagoon is deeper than those of atolls in Stage 1 , but contains little coral growth. 

 The true passes (with channels) are not more numerous than at Aldabra, for after they 

 reach a certain size a balance is struck between the power of growth of the calcareous 

 organisms on the sides of the pass, and the speed of the tidal current (which broadly 

 speaking determines both its erosive power and its power of preventing coral growth). 



The vegetation on rock is varied, but on the sand is typical of any sand cay. 



Stage 3. This stage is shown by Farquhar Atoll, which has little land on its 

 circumference, and this land is mainly in the form of sand cays, though enough elevated 

 coral rock is found to make it not improbable that Farquhar once had an almost complete 

 rocky land-rim. The reef between the islands is more flourishing than at Cosmoledo 

 and definitely shows signs of growth seawards. The lagoon is somewhat deeper than 

 those of the previous atolls, and is perhaps becoming more so. The vegetation is almost 

 entirely of a coral sand type. 



Among the non-atolls the series is not so complete, but Assumption and St Pierre 

 are an early state of Stage 1, while Providence corresponds to the sandy Stage 3. 



Considering the important features of the three stages, it does not seem to be 

 straining the imagination to predict in the case of Aldabra or Astove a gradual reduction 

 of the land-rim into small islands, which will be divided at first by stretches of barren 

 reef. The islands, as they are reduced in size, will become more sandy, until elevated coral 

 rock may appear quite absent, and at this period the vegetation will consist of the regular 

 coral sand plants. The lagoon will continue to grow deeper, and the "Farquhar" stage will 

 be the result. Beyond this it is impossible to foresee, for the ultimate fate of the atoll 

 depends on various conditions of which we have no data, though without further elevation 

 we can feel sure that no coral rock land will continue to exist. The atoll may continue 

 for ever with its reef awash and bearing a few sand islands, or on the other hand it 

 may be cut down fathoms beneath the sea to form the base on which corals can build 

 a fresh reef. 



* Loss is shown on Wizard Island by the presence of beach sand-stone. 



56—2 



