THE ALLEGED DESICCATION OF EAST AFRICA 5 
above sea-level, but at the entrance to Kilindini harbour 
(Mombasa) there is more than that : on the island, some 200’ 
feet back from the edge of the present coral cliff, there is another 
cliff which evidently marked the width of the channel in, say, 
Pleistocene times ; in the opposite direction a third cliff is 
just emerging from low water level. 
Now, I believe that it can be safely asserted that the width 
of a channel through a coral barrier usually depends on the 
amount of fresh or semi-fresh water discharged through it, 
the mixture of fresh water with the salt checking the growth 
of the reef. I therefore believe that when the cliff marked C 
was the boundary of the tidal channel, the volume of fresh 
water discharged into the upper waters of the harbour was 
much greater than it is to-day ; similarly, when the cliff B 
bounded the water-channel, the volume of fresh water was 
greater than at present, when a new cliff A is being formed 
which can be clearly seen at low tide. The greater volume of 
fresh water would also bring down mud which is generally 
recognised as being a factor that checks the growth of corah 
This theory is borne out by the large amount of erosion wdiich 
is visible in the canons of the Mwachi and Manolo rivers, which 
even now in flood-time discharge a considerable amount of 
fresh water into the head of Port Reitz. The amount of fresh 
water, however, at any season bears a very small proportion 
to the enormous volume of salt water which flows into and 
out of Port Reitz and Kilindini harbour. The Andromache 
reef and the reef to the north of the port correspond in age 
to the semi-submerged cliff A, which is just showing above 
sea-level. 
At the time when cliff C formed the north boundary of 
Section of entrance to Ki find inLHarboui* 
