begin to ask 'has it rained this year?' or even 'how many years ago did you have rain 

 here?'. 



I have noticed a tendency for Western civilisation to limit consideration of deserts 

 to areas where some though inadequate rain falls every year, and to neglect the rest. 

 This is convenient for those who try to measure the degree of aridity in terms of mean 

 annual rainfall, but it leaves the general picture of desert biology, and even of human 

 desert ecology, very incomplete. I think we have been mislead by rainfall maps. Lack 

 of data forces them to lump all desert regions into one final omnibus category of say 

 250 m/m to zero mean annual rainfall. Whereas if we had enough data to spread this 

 category on a logarithmic scale we should see biology stretching out far beyond the 

 limit of annual rain. 



For the more arid areas rainfall data is both inadequate and unreliable, and must 

 remain so until we have 10- year automatic recorders. For it is against human nature 

 to look conscientiously at an empty rain gauge for several years on end. By the time 

 rain does come the gauge has probably been put to some other use, or the observer is 

 elsewhere. It is the rule in some more rain- favoured countries for the gauges to be 

 stored during the dry season and put out on a fixed date, and it is not unknown for a 

 single widespread rain storm exceeding a whole year's mean to remain unrecorded, be- 

 cause it fell too soon. Moreover desert recording stations coincide with human habi- 

 tation which needs permanent water, i.e. with spots of least elevation. Hence their 

 recorded rainfall is probably considerably lower than elsewhere around. 



Beyond the limit of annual rains the biological significance of mean precipitation 

 dwindles rapidly. I suggest that the dominant factor which takes its place is the mean 

 period between effective storms. I would define an effective storm as a fall of such 

 magnitude that some water remains availably stored in favoured spots such as sand, 

 mud pans and rock fissures after immediate surface evaporation has ceased. 



The mean rainless period, in years, unlike the mean annual rainfall which needs 

 careful quantitative measurement under very adverse conditions, already exists as a 

 clear estimate in the minds of nomads. Their lives depend on it. And this estimate 

 could be extracted by careful questioning. A fair estimate of the quantity of rain from 

 an effective storn could also be made from descriptions of the degree of flooding. 



It is just possible that the mean rainless period, which we could get, might be 

 linked approximately with the mean annual rainfall, which we cannot get, and the lat- 

 ter, though insignificant, as thus obtained indirectly, could then still be used for the 

 sake of continuity of the measuring scale. Various scrappy bits of information rather 

 suggest that the precipitation from a mean effective storm remains fairly constant from 

 one part of a given desert region to another, provided due allowance is made for the 

 effect of ground elevation. For N.E. Africa which includes the most arid areas in the 

 world I would put this constant at 15 to 20m/m. Allowing 50% run- off concentration 

 this figure agrees with the precipitation needed to soak sand to a depth of 20 to 30 cm. 

 Similarly a guess can be made of the proportion which effective rain bears to the total 

 rain. We might put this at \ and assume that % of the total rain falls as light showers 

 and can be neglected. 



10 



