An Introduction to the Study of South African Bainfall. 23 



months. The two last columns show how many of these months 

 had a fall less than the mean. It appears that there were 126 

 months in which both places together had a fall less than the mean ; 

 53 in which the fall was greater than the mean at both ; 63 in which 

 the fall at Durban was greater, and at Kimberley less ; 70 in which 

 the fall at Durban was less, and at Kimberley greater. Is this any- 

 thing more than chance ? A perfectly chance distribution would give : — 



When both stations together are less than the mean 119 



When both together are greater 46 



Durban greater and Kimberley less.. 63 



Durban less and Kimberley greater 70 



— the difference between the numbers in the last two lines evidently 

 being a constant quantity. Thus there is an excess of 14 agree- 

 ments greater than the 165 allowed for by the matured chances, 

 which amounts to a deviation of both together in the same direction 

 of about one month in the year. So that even if there be a relation- 

 ship between these two stations it is not very decided. 



Belated more or less closely to the variation of rainfall is the 

 cloudy state of the sky. This last element is stated in monthly 

 averages in Table 28. Great accuracy must not be expected in 

 these numbers, although they may possibly show the shape of the 

 annual curves fairly well. It is a surprising fact that although the 

 observation for the amount of cloud demands less skill than the 

 reading of any instrument, it is done much more inaccurately. It 

 is surprising sometimes how the percentage of cloud at a given 

 station changes when the observer is changed ; it is still more sur- 

 prising how the observer himself sometimes changes. Kimberley 

 observers have not on the whole made very comparable observations,, 

 as may be, e.g., seen in the returns from Central Jones Street (G. J. 

 Lee), and from the Meteorological Commission Station (C. Aburrow), 

 for 1887." That they may easily agree well enough is proved in the 

 returns from Kenilworth, Kimberley (J. E. Sutton), and from Lennox 

 Street (H. R Harrison), for 1898. f Certainly the Kenilworth 

 averages seem to be fairly consistent, as tested with the sunshine 

 recorders. The following are the total percentages of sunshine plus. 

 cloud during daylight, year by year : — 



1898 108 per cent. 



1899 112 



1900 Ill 



1901 112 



1902 110 



Report of the Met. Com., 1887. f Ibid., 1898. 



