An Introduction to the Study of South African Rainfall. 15 



plete records for not less than seventeen or eighteen years, have 

 been computed and arranged in Table 25. In the few instances 

 made use of where the record is for a less number of years, the 

 station is inserted because the area is badly represented or because 

 of some importance in the site. The material comes chiefly from 

 the pick of the annual reports of the Meteorological Commission for 

 the years 1880-1901. Of the rest, the Kimberley record, as already 

 mentioned, is from private registers ; the Bloemfontein record was 

 taken mostly from a register printed in a local newspaper ; the 

 Natal records are from the very excellent annual reports issued 

 by the Durban Observatory, supplemented by some MS. monthly 

 totals kindly sent me by Mr. Nevill. 



In forming Table 25 I have availed myself largely of Buchan's 

 work, simply combining in most of the cases his tea-year normals 

 with the results for twelve additional years. This course was justified 

 by an occasional test of the same normals. The rainfall areas I., 

 II., III., IV., ... of the Meteorological Commission's arrangement 

 have been retained for the sake of continuity, although it is not 

 •easy to see why they were ever adopted." The only attempted 

 improvement has been to divide all the areas which seemed to 

 extend too far east and west into two. A comparison between the 

 Commission's map and its modification given at the end of this 

 paper, together with an examination of the rainfall of some of the 

 Karroo areas, will show the necessity for such a plan. Our sub- 

 division raises the number of areas from fifteen to twenty. For 

 each of these the mean rainfall has been computed, and approximate 

 angles and coefficients in the sine series determined — only approxi- 

 mate because the months vary in length. These last appear in 

 Table 26, wherein V x , V 2 , V 3 are the epochs ; u If u 2 , u 3 the ampli- 

 tudes ; u\, u' 2 , u 3 the amplitudes in parts per thousand of the 

 mean monthly rainfall. We shall return to a consideration of 

 these constants presently. 



So far as these 160 stations go, Section I. (the Cape Peninsula) 

 has the most copious fall, with an average of over 40 inches per 

 annum ; Section XV. (Natal) coming next with nearly 36 inches. 

 'The western division of Section IX. (Northern Border) is the most 

 arid, with an average of less than 6 inches. It is to be re- 

 membered, however, that the sectional average results depend to 

 a great extent upon the number of gauges in a given area, and 

 their position. Generally the number of good records throughout 



* The latitudes, longitudes, and altitudes, are from the latest Eeports of the 

 Met. Com. Some of these are doubtful : Uniondale, Kleinpoort, and Glenconnor, 

 in Section V. , for example. 



