148 
THE DESICCATION OE AFRICA 
are in places twenty feet high on the lower reaches. On the 
northern boundary of the plains are the Cholo Hills, down which 
several small streams flow from the plateau land, but these 
streams disappear into the ground after a few hundred yards 
on the flat. In one instance, to my certain knowledge, one 
stream sinks with very perceptible current into a small bed of 
reeds and does not reappear. It has often occurred to me that 
the underground water can here be traced by the position of 
the great belt of Borassus Palms (among which are a few 
Hyphaene Palms) which extends right across the flats in the 
Shire depression. These palms must have water, but they can 
go down twenty feet for it. 
Lake Tanganyika . — Despite the rumours of alleged tem- 
porary rises in the water level of this large lake with a 
small basin, the general study of the shores shows the old 
story of permanent and steady decrease in the level, which 
perhaps has been more rapid than in the case of Nvasa. Even 
allowing for the fact that the strong winds, which blow from 
the north, have piled the sand into long dunes, it must be 
admitted that the sand had to be exposed in the first instance 
before coming under wind action. I speak now of Cameron 
Bay at the beaches, Kasankalawe and Mbete. Old beaches 
can be traced back right into the tree-covered land. At the 
former place the regular formation of the old beaches is very 
noticeable. Furthermore, fragments of delicate mollusc shells 
are to be found in the most recent beaches, but are lost sight 
of in the older ones, doubtless owing to the dissolving action 
of the overlying vegetable humus. I refer to two shells which 
are only found alive in permanently submerged sand. One 
is a very delicate Pecten-like bivalve, name forgotten, very 
fragile. The other is a new (1906) species — Cleopatra Hargeri. 
These forms, with many others more robust, must have been 
left by the water, and could not have been blown back such 
distances, even in fragments. 
Such conjecture can be accepted when it be remembered 
that the old town of Ujiji is about three-quarters of a mile 
from the present lake shore, and so is the old beach where old 
natives, still living in 1906, remembered having dipped up 
water. 
