A Contribution to the Sttidi/ of the Rainfall Maj) of South Africa. 379 
and there ; and it is not always easy to determine whether an isohyet should 
pass round such a place to the right or to the left. Moreover, in areas 
where there are few gauges the isohyets are usually drawn too simply 
straight. So that there is always room for bias, and for criticism, upon the 
details of any rainfall map, whether those before us or any other. 
Looking at the annual map, we see that south of latitude 25° S. the 
isohyets trend mainly north and south, while north of latitude 20° the 
tendency of the trend is north -w^est to south-east. A remarkable feature is 
the dry belt near the tropic of Capricorn, stretching nearly all across the 
continent, and only broken by the wet oasis of the mountains in the Eastern 
Transvaal. The course of the isohyets inland from that part of the east 
coast lying between the parallels of 24° and 27° is to some extent tentative. 
When more rainfall records are available from the gauges Lieutenant 
Teixeira is establishing in Portuguese East Africa, it will probably be found 
that the isohyets meander more than they have been drawn to show. 
At first sight it might appear that the small closed isohyet of 60 in. 
on the tropic, and that of 50 in. near latitude 27°, ought to be in more 
intimate communication with one another, and one feels tempted at first 
to run the isohyet of 30 in. inland from about the place where the 
parallel of 27° nieets the coast, and below the 50-in. isohyet rather than 
above it. The intrusion of the isohyet of 20 in. from the west, nevertheless, 
in the immediate neighbourhood, as w^ell as the dry belt pushing westwards 
from Lourenco Marques, support the map. 
It is curious how relatively few conspicuous instances there are in which 
the mountain ranges show a pronounced influence upon the rainfall. It is 
evident that the distribution of South African rainfall is of general 
significance, and that local conditions are quite a secondary matter. Local 
influence can l)e traced nlong the terraces facing the Atlantic, along 
the Langebergen and Outniquas, in the Amatolas, in the northern end of the 
Brakensbergen, and, perhaps most strikingly of all, in the Zoutpansbergen, 
which, as Ave have seen, is the one obstacle to the uninterrupted course of the 
tropical dry belt across the continent. Lake Nyassa is an oddity : the Shire 
Highlands manage to catch the rain ; but if the map is to be trusted, the 
western side of the lake is much wetter than the eastern side in spite of the 
Livingstone Range. 
The most likely explanation of the lack of concordance between the 
distribution of rainfall and the physical features of the land is that most 
South African rain falls in thunderstorms, whose origins and tracks are 
largely independent of the geographical contours, and whose clouds are higher 
than most of the mountains. Now it is the almost universal custom to fill 
up by inference the gaps in our actual knowledge of the rainfall, and this 
usually resolves itself into an effort to make the isohyets and the land 
contours coincide. Thus if we liave a semicircular range of hills with only 
