108 
Transactions of the Boyal Society of South Africa. 
records, for every shower in the whole period exceeding '001 inch an hour. 
It summarises, therefore, rather more material than is used in Table Y. 
The hourly frequencies are smoothed in threes, but the quantities are 
unsmoothed. 
We find in Table VI what we should expect, namely, a considerable 
variation in the quantity of rain, hour by hour ; and what we should not 
expect, namely, a definite lunar curve of frequency with a maximum near 
moon rise and a minimum just after moon set. Quantities and frequencies 
are distributed as follows (neglecting the meridiem hour) : 
Inches. Frequency. 
Ante-meridiem . . 52'2 per cent. . 50*0 per cent. 
Post-meridiem . 43*5 ,, . 45* 7 
Thus, underlying the irregularity of the quantities there is a greater 
tendency to an ante-meridiem maximum than is shown by the more regular 
frequencies. In fact the hourly intensities of rainfall are : 
Ante-meridiem ...... -056 inch 
Post-meridiem ...... •051 ,, 
Hence it is evident that the vagueness of Table III is due to the peculiar 
incidence of rainfall. Rain coming most often, as it does, about the time of 
moon rise will naturally cause the evaporation to be less in the hours follow- 
ing that time than in the hours preceding. And so we get an explanation 
of the run of the hourly rates of Table V ; and these just happen to nearly 
counteract the run of the hourly rates of Table lY. 
If now we introduce corrections into Table III on the assumption that 
the mean evaporation curve for rainy days ought to be a straight line, we 
get the unsmoothed values of Table YII. 
Table YII. 
Corrected Hourly Mean Evaporation arranged according to the 
Lunar Day. 
Hour. 
Inch. 
Hour. 
Inch. 
Hour. 
Inch. 
I 
•0069 
X 
. -0069 
XIX 
. -0070 
II 
68 
XI 
70 
XX 
71 
III 
68 
XII 
71 
XXI 
70 
lY 
69 
XIII 
71 
XXII 
70 
Y 
69 
XIY 
71 
XXIII 
70 
YI 
69 
XY 
71 
XXIY 
69 
YII 
69 
XYI 
71 
XXY 
70 
YIII 
69 
XYII 
71 
IX 
69 
XYIII 
71 
The curve given by Table YII is in substantial agreement with Table lY, 
and considering that it is unsmoothed in any way is remarkably regular. 
