134 Transactions of tlic South African Philosophical Society. 



competent computers are supposed to have at their fingers' ends. 

 It is sincerely to be hoped that such as they may be they are neither 

 important nor numerous enough to neutrahse whatever may be found 

 of interest here.* 



Although the labour entailed in the construction of the Tables at 

 the end has been very great, a period of four years is actually, for the 

 purpose of discovering normal means, rather short. The stated 

 values must, for that reason, be accepted as only first approxima- 

 tions, though correct enough to more than the limits of accuracy 

 of a preliminary statement. There is no question that a longer 

 period v^ould have yielded more valuable results ; but, on the other 

 hand, it seems safer to secure earlier publication of the somewhat less 

 valuable rather than take the risk of no publication at all. Which 

 risk is a very serious contingency : for depending entirely on private 

 enterprise, and perhaps also on a single life, the Kenilworth Meteoro- 

 logical Station cannot, in the nature of things, be assured a very long 

 lease of existence.! 



If the tangle of incidents lying between cause and effect in 

 meteorological phenomena was not already well known, there would 

 be no difficulty in recognising it in the results given in the following 

 pages. With the idea of simplifying the problem, most of the elements 

 are shown under three aspects : (1) in the ordinary way, the means 

 being arranged from a continuous record ; (2) when the sky is clear ; 

 (3) when the sky is cloudy. A clear sky is meant when the mean 

 cloudiness of the day does not exceed 20 per cent., nor the cloudiness 

 of a single observation exceed 30 per cent. All other days are cloudy. 

 The effect of this subdivision is to vary the conditions we are study- 

 ing through the agency of a most influential factor. In 1898 and 

 1899 only three observations of cloud per diem were made, namely, 

 at VIII., XIV., and XX.; in 1900 and 1901 three others were added, 

 namely at XI., XVII., and XXIII. It is possible, therefore, that the 

 two earlier years have a relatively larger number of ''clear" days 

 than the latter. 



In comparing the conditions prevailing on clear and on cloudy 

 days it is necessary to know at the outset that no systematic diffe- 

 rence exists between them depending upon some fortuitous tendency 

 to a grouping of one or the other at the beginning or end of a month. 



* Reasoning from previous experience, the most likely errors are reversed signs 

 and digits in the wrong order : + for — , 35 for 53, and so on. Tabular matter, I 

 find, begins to become blurred and indistinct to eye and sense after midnight. 



t Hann has expressed the urgent need for more material from the Southern 

 Hemisphere. See, e.g., the Quarterly Journal of the Hoyal Meteorological 

 Society for January, 1888. 



