THE TANGANYIKA PROBLEM. 
J 9 
of water which occur on a great continental land mass, 
and throughout all the great continental land masses of 
the world. If we examine the great fresh-water basins 
and river systems of Africa, for example, the first fact 
which confronts us is the localisation of groups of varieties 
of the universal types. Thus, confining our attention, 
merely for the sake of simplicity, to the molluscs of Lake 
Nyassa, we find only one species — namely, Melania tnber- 
culata — existing in this lake which occurs in any of the 
other great lake basins.* In Tanganyika, excluding 
its peculiar halolimnic group, all the species of molluscs 
in it are peculiar to the district : they do not even occur 
in the adjacent lakes of Mwero, Rukwa, or Bangweolo, 
such types being either unrepresented in these lakes or 
represented by different species. So again in Kivu there 
is not even a Vivipara , and the only series of great 
African lakes that possesses a large number of identical 
species is that constituted by Albert Edward, the Albert, 
and the Victoria Nyanzas, which communicate with the 
Nile now, and were almost unquestionably once more or less 
connected together. These facts — and there are other 
similar ones connected with the distribution of species in 
the lakes and river systems of North America — seem to 
show in an unmistakable manner that many typical fresh- 
water forms have little if any capacity to migrate from 
their original centres of distribution, and, consequently, 
that the peculiar specific varieties which they present in 
these centres are due either to their original character or 
to natural selection, having operated since their origin, in a 
generally similar manner, in all these particular spots. In 
like manner, the distribution of the Characinid fishes in the 
American and African fresh waters is quite inexplicable 
* Shirwa is a lake in the same basin as Nyassa, and was at one time connected with it. 
