An Introduction to the Study of South African 'Rainfall. 11 



component. A miniature monsoon effect is suggested by this. For 

 the greater part of the year our rain seems to be associated with 

 surface winds having an easterly component ; but during midwinter 

 with surface winds from the south. This seems to furnish a link 

 between the summer rains of Durban and the winter rains of the 

 Cape Peninsula. At the same time two or three circumstances 

 must be borne in mind : First, that it does not follow that these 

 particular prevailing directions bring the rain, seeing that they may 

 be as much an effect as a cause ; second, that the cloud currents 

 generally spring from points lying somewhere between north and 

 west — clouds of the cirrus type coming more from the west, those 

 of lower levels more from the north, of the mean direction ; third, 

 that as a rule the particular wind-direction associated with abun- 

 dant rain, characteristic of any month, is the dominant normal 

 wind of that month — that is to say, with abundant rain the normal 

 prevailing wind prevails yet more ; or, in other words, the rainfall 

 decreases with the deviation of the vane from its normal position. 

 This is a curious commentary on a previous result, " that, relatively 

 to the normal curve, the cloudiness of the sky increases with the 

 deviation of the vane from its normal position." * The cloud 

 result, however, applies to the diurnal variation of the wind, while 

 the rain result applies to the mean direction for the month. A 

 noteworthy consideration is that whereas there were two months 

 having a rainfall less than the mean to every one having a rainfall 

 greater, in the six years under review, more deficient months have 

 been included in forming the wind-averages than abundant ones,, 

 and therefore a certain amount of preconceived bias towards dryness 

 has entered into the ratios. In passing it may be mentioned that 

 the same bias is some drawback to the greater number of 

 meteorological statistics in which one element has to be compared 

 with another. It is not unlikely that deviations from the 

 median, rather than from the mean, would give better compara- 

 tive results.! 



I have not yet made any experiments for the purpose of deter- 

 mining the temperature of falling rain, nor indeed do I yet know 

 of a likely method. I But it may be roughly estimated, in propor- 

 tionate numbers, from the consideration that in heavy falls the 

 temperature of air, rain, and dewpoint, must tend to a common 



* " Winds of Kimberley, Trans. 8. A. Phil. Soc, vol. xi., pt. 1, p. 92. 

 f See Harm, "Handbook of Climatology" (Ward's edition), p. 24; Merriman, 

 "Method of Least Squares," 8th ed., p. 210; Lupton, "Notes on Observations,'* 

 44. 

 J See, however, Hann, Lehrbuch der Met., p. 303. 



