WATER RESOURCES IN ARID REGIONS 97 
which observations of much longer duration are more necessary 
than in humid regions. 
As a consequence of the remoteness of the stations, the expenses 
are very high and cannot easily be borne by the arid regions, 
which are economically weak, without having recourse to aid by 
more humid regions. Apparatus runs the risk of being destroyed 
by nomadic peoples who, with no evil intent, consider a rain 
gage an object worthy of attention in the desert, interesting to 
dismount, and useful as a domestic apparatus. Some difficulties 
would be resolved if cheap and particularly strong automatic 
devices were used which did not require having observers in each 
place. Be that as it may, so great are the difficulties that the 
hydrology of arid regions, now in its infancy, will remain so as 
long as the network of observations remains largely undeveloped, 
especially in the mountains where the rains are most abundant. 
In Tunisia we have tried to fill the gaps in the pluviometric 
network by establishing an approximate map of the rainfall, on 
the following bases. 
The reports furnished by the stations of the network serve as 
reference marks. Between these stations isopluviometric lines 
have been interpolated by considering the variation in the rainfall 
according to altitude and exposure, and by complementing these 
means of evaluation by phytosociological information (g). The 
advantage of the phytosociological method is that it integrates 
the rainfall of a long period. It demands much care and time 
because annual vegetation can have a different appearance ac- 
cording to the year the survey is made. Therefore, perennial 
plants are of peculiar interest. Anyway one must not neglect 
any means of cross-checking. We consider, however, the result 
obtained satisfactory. It completes, at modest cost, the quite 
insufficient information on annual rainfall given by the clima- 
tological stations. 
The phytosociological method is not, after all, the only bio- 
logical method to determine the amount of rainfall. Another 
method consists of making, in the course of the seasons, evalua- 
tions of leaf surfaces and of the evaporating power of the soil and 
its vegetal cover (15). 
