386 



ISLAND LIFE. 



[PAUT ir. 



it is much broader, and stretches out opposite Mozambique to a 

 distance of about eighty miles. The Mozambique Channel varies 

 from less than 500 to more than 1,500 fathoms, deep, the shal- 

 lowest part being where the Comoro Islands and adjacent shoals 

 seem to form stepping-stones to the continent of Africa. The 

 500-fathom line includes Aldabra and the small Farquhar 

 Islands to the north of Madagascar ; while to the east the sea 

 deepens rapidly to the 1,000-fathom line, and then more slowly, 

 a profound channel of 2,400 fathoms separating Madagascar from 

 Bourbon and Mauritius. To the north-east of Mauritius are a 

 series of extensive shoals, forming four large banks less than 100 

 fathoms below the surface, while the 1,000-fathom line includes 

 them all, with an area about half that of Madagascar itself. 

 A little further north is the Seychelles group, also standing 

 on an extensive 1,000-fathom bank, while all around the sea 

 is more than 2,000 fathoms deep. 



It seems probable, then, that to the north-east of Madagascar 

 there was once a series of very large islands, separated from it 

 by not very wide straits; while eastward across the Indian 

 Ocean we find the Chagos and Maldive coral atolls, marking 

 the position of other large islands, which together would form 

 a line of communication, by comparatively easy stages of 400 

 or 500 miles each between Madagascar and India. These sub- 

 merged islands, as shown in our map at p. 396, are of great 

 importance in explaining some anomalous features in the zoology 

 of this great island. 



If the rocks of Secondary age which form a belt around the 

 island are held to indicate that Madagascar was once of less 

 extent than it is now (though this by no means necessarily 

 follows), we have also evidence that it has recently been con- 

 siderably larger ; for along the east coast there is an extensive 

 barrier coral-reef about 350 miles in length, and varying in 

 distance from the land from a quarter of a mile to three or 

 four miles. This is good proof of recent subsidence ; while 

 we have no record of raised coral rocks inland which would 

 certainly mark any recent elevation, because fringing coral reefs 

 surround a considerable portion of the northern, eastern, and 

 south -western coasts. We may therefore conclude that during 



