and of the East African Mainland. 39 



the seaweeds Halimeda and Gaulerpa, which entirely cover the 

 surface of the outer slope, do no more than protect the under- 

 lying rock. This protection must be, however, quite complete. 



All these differences indicate that this reef is at an earlier 

 stage of formation than is that of Zanzibar. Whether younger 

 in point of time, or merely because erosion is retarded by the 

 greater hardness or more perfect homogeneity of the Pemba 

 rock, is unknown. The latter explanation is suggested by the 

 almost continuous line of cliffs which forms the east coast-line. 



In any case, as erosion proceeds, the state of things found in 

 Zanzibar will tend to be reproduced. As the reef widens by the 

 cutting down of the land, and by the formation of a boat channel, 

 a large amount of debris will come to be carried out by each tide. 

 The pollution of the water by this and by organic impurities 

 produced in consequence of the exposure of a broad reef to the 

 sun at low tide, will kill off all the organisms of the reef edge 

 except those which we find existing under these conditions in 

 Zanzibar. 



This latter flora can hardly be so completely protective as are 

 the corals and nullipores of Pemba. It seems probable if the 

 edge of the Zanzibar reef had depended for protection upon its 

 present flora during the whole of its existence that it might not 

 have so well resisted the eroding force to which it has been 

 exposed. 



In writing of the reefs of Zanzibar I cited the Bermudas as an 

 instance of results of erosion simulating the results of the growth 

 of reefs. The west coast of Pemba affords a much more direct 

 comparison, the bays included within the islands and reefs of the 

 Mesale-Kokota-Fundu-Njao barrier being entirely homologous 

 with the lagoon of. the Bermuda Pseud-atoll. The reduction of 

 these islands to reefs (a process already far advanced between 

 Mesale and Kokota) and the enlarging of the bays, converting the 

 present barrier reef into the edge of an atoll, may in time cause 

 the map of Pemba to bear a strong resemblance to that of the 

 Bermudas. 



The Mainland Reefs. 



These appear to fall into two groups which exist together, one 

 fringing, the other forming a barrier. Both are usually somewhat 

 irregular. The latter appears as a series of more or less elongated 

 reefs of old coral rock formed by the partial or complete cutting 

 down of islands to below tide level as described before in the case 

 of reefs and islands near Zanzibar. 



In the Zanzibar channel the barrier system is not trace- 

 able, though distinct in the Pemba channel. This anomaly is 



