$ Transactions of the South African Philosophical Society. 



•statistics. To do so successfully would require all the accumulated 

 records of a weather bureau, although it may be possible later on to 

 introduce some facts of interest in this connection. At present it 

 may perhaps be taken as a working basis that the greater part 

 •of our rain falls in thunder-squalls and secondary cyclones, no 

 great quantity coming in cyclones proper nor with straight 

 isobars. 



Table 17 gives the diurnal variation of rainfall at Kenilworth, in 

 quantity and duration, both in whole totals for six years (1897- 

 1902), month by month, and in annual averages. There was not a 

 separate recording rain-gauge at this station during this period, and 

 therefore the values have been mostly reduced from the automatic 

 records of the evaporation tank. The yearly average quantities 

 show a principal maximum at 4.30 p.m., and a principal minimum 

 at 9 a.m. — being thus very nearly the inverse of the barometric 

 phases. There are points of relationship with the diurnal oscillation 

 of the barometer, as well as with the horary dewpoints. The 

 phases of frequency are interesting because the maxima tend to 

 fall earlier, and the minima later, than those of quantity. The last 

 line of Table 17 gives the average fall for each hour of rain at 

 different times in the day. This indicates that the rate of fall is 

 lightest an hour or so after sunrise, and heaviest just before' sunset, 

 the increase of rate during the daylight hours being gradual. 

 During the night hours there are considerable irregularities, with a 

 strong tendency to a long-drawn-out maximum between XXIII. 

 and III. 



Table 18 shows the diurnal variation of rain for different seasons 

 in portions of the total fall in six years ; the total hours ; and the 

 average rate in each rainy hour. At the foot of the Table is given a 

 summary showing how the seasonal variation is distributed between 

 day and night, i.e., from VI.-XVIIL, and from XVIIL-VL, respec- 

 tively. This proves that, saving the spring months (August- 

 October), the greatest quantity of rain falls at night. In the 

 summer it rains as often by night as by day; but in the winter 

 there are half as many rainy hours again at night as there are by 

 day. The rate of fall per rainy hour, however, is pretty much the 

 same night and day, excepting in the spring. Putting all the 

 seasons together, we find the quantity rather the greater at night, 

 but the number of hours still greater, so that on the whole the rate 

 of fall is greater by day. 



Table 19 is inserted for comparison with Tables 17 and 18. It 

 .shows : — 



1. Thunder and Thunderstorms. 



