evaporating in wispy tails to the clouds as it descends. When it c'oes reach the ground 

 it tends to be torrential and brief, the thunderstorm type, and comparatively local in 

 extent. 



Thus the two characteristics of desert rainfall are that it is accidental, being due 

 to a disturbance of equilibrium in the air, and that it is usually local in extent. In the 

 tropics it tends to be seasonal, precisely because the desert is in a rain- shadow area 

 and the air reaching it during the rainy season to windward is at least more saturated 

 than at other times of the year. 



Returning to the heavy insolation which is the cause of this instability rainfall, 

 we must note that it is much more effective on dark bare soil than on light- coloured 

 soil with some vegetation covering. This fact appears to me to be of considerable im- 

 portance in the biology of deserts, though of course it applies rather to the semi- 

 desert where vegetation cover can occur than to the utter desert where it cannot. 



In the western and drier parts of the Kalahari, for instance, the cover of low bush 

 and grass is quite considerable in the so-called rainy season. On the 'pans' how- 

 ever, which tend to occur in chains, the ground is either bare because of its salt con- 

 tent or has a short grass, kept shorter still by the herds of springbok. It was noticed 

 during my visit there that the thunderstorms tended to keep to the pans, that is to the 

 centres of the rising air currents. On one particular day a series of over a dozen 

 heavy thunderstorms passed along such a line of pans. At our camp, situated on a 

 sand- ridge half- a- mile from this chain, only one of the storms produced rain, though 

 it was nearly half- an- inch in half- an- hour. The natural deduction was that over the 

 pans the rainfall that day was very much greater, possibly several inches. This de- 

 duction was supported by the fact that the next day our lorries were badly bogged 

 crossing one of these pans, over which two days before we had driven at speed. 



In any area of instability rainfall we have to be very cautious about accepting 

 rain-gauge data, but it will be doubly so if there is ground for suspecting that storms 

 choose their paths with some consistency, in the way outlined above. 



If the tremendous insolation by day in a desert causes great vertical disturbance 

 in the atmosphere, the opposite occurs at night. The rapid cooling tends to cause an 

 inversion of temperature, so that air in contact with the ground becomes heavier and 

 remains there. All desert travellers are familiar with the experience of insupportable 

 heat by day and desperate cold at night. It is no exaggeration to say that a basin of 

 water outside one's tent may be frozen at six in the morning, thawed by eight and at 

 blood temperature by noon. 



These rapid changes of temperature must obviously affect plants and animals in 

 the desert, but at the moment we are concerned chiefly with the yield of moisture so 

 caused. Measured in inches, even if that were possible, the total derived from frost- 

 rime and dew would not be impressive, but the fact that it is in immediate contact with 

 leaves and branches is no doubt of biological significance. Certainly it is the case 

 that some antelope, notably the springbok, derive all their water-supply from the dew 

 on the grass in the early morning. Later papers will perhaps assess the value of this 

 source of moisture, particularly in the case of plants. 



