84 MEASUREMENT OF RAINFALL. 



minute period. And, since the object soug-ht in all these observa- 

 tions, automatically determined, is that the data should approach 

 as near as possible to the unit of measurement used in their 

 practical application, the twenty-four-hour chart is therefore pre- 

 ferable to the weekly one, and doubtless many of the eight-day 

 instruments in possession of the Government could be converted 

 to the twenty-four-hour type by a simple adjustment of the 

 clockwork. 



By the kindness of the Town Engineer of Johannesburs^ — 

 Mr. G. S. Burt-Andrews, M.I.C.E. — the Avriter has been per- 

 mitted to discuss the performances of the two types of instru- 

 ments belongino^ to his department. 



Until recently the 24-hour Richards .-Vuto-oauoe was 

 stationed at the Klipspruit Sewerag-e Farm, which is 7.4 miles 

 south-west of the 8-day Auto-gauge at tlie Burghersdorp Pipe 

 Yard, Johannesburg. The altitude of the former i< 5413 



M.S.L., and the latter is 5460, or a difference of 237 feet in 

 vertical height, each instrument being" 4 feet above tlie surround- 

 ing ground, and in open positions. 



Unfortunately it is only possible to take the ]:)eriod of 1909- 

 13 into consideration, and a large number of charts have been 

 studied. The larger chart gauge being stationed at Klipspruit, 

 the data therefrom is put forward with greater confidence than 

 the weekly chart of the Burghersdorp instrument. In both 

 cases, however, the results are only indicafiz'c not c(Miclusive; 

 not cnly the period is small, but the comparison is made on 

 different charts. Notwithstanding, however, it is claimed that 

 these indications arc important and justify the larger use of the 

 24-hour -Auto-gauge, and the more exact measiuxmcnt of rain- 

 fall. 



For the period inider review, rain fell at Klipspruit for 594 

 hours in 231 days, giving a total preci])itation of 75.19 inches 

 during the period, of which 16. t6 inches fell in 620 minutes, rv'.cr: 

 21.49% of the total rainfall fell in i-75% ' f the total time. Fur- 

 ther, it was found that for the season 1910-11, there were 18 

 falls of rain, which lasted from o to t hour, 24 from i to 2 

 hours, II from 2 to 3 hours. 6 from 3 to 4, 4 from 4 to 5, 6 from 

 5 to 6, 4 from 6 to 7. 3 from 7 to 8, i from 8 to 9, 3 from 9 to 

 10, I from 10 to II, I from 13 to 14, and i from 15 to 16 hours 

 of continuous rain; while for the season 1911-12, it was found 

 that the number of showers in duration of o to t hour \vas 20, 

 I to 2 hours 25, 2 to 3 hours 11, 3 to 4 hours 1. 5 to 6 liours 

 I, 5 to 7 hours I, and 10 to ii hours i. 



It was further noticed that rains of 2 hours and less in 

 duration contained spurts or spasmodic turns of intense fall, 

 where a few minutes of such rain dominated the rainfall for the 

 period. By combining these rains as against the 83 days of rain 

 for the season 1910-11. and the 68 days of rain for 191 r-12. we 

 get 57.6% of non-beneficial rainfall for the two seasons. 



In this connection it is interesting to note that this deter- 

 mination is in close agreement with one made by Mr. H. F. 



