162 Transactions of tJie South African Philosophical Society. 



the amplitude of the curve of wind-velocity will vary with the 

 variation of cloud. And thus the solution we have obtained from 

 the physical problem is not the one we sought. It is important, 

 though, to observe that the result introduces a very substantial 

 reason why there should be a diurnal cloud period. 



Table 42 gives the annual average miles of wind from each of 

 the principal directions during each hour of the four years. This 

 Table proves in reality to be only an emphasised epitome of Table 34. 

 The points of dissimilarity between them will be best appreciated 

 after comparing together the mechanical resolutions of each given 

 respectively in Tables 35 and 43. The only material dissimilarity 

 revealed is that the maximum of the east component of velocity is 

 more than four hours earlier than the maximum of direction. In all 

 the other cases the phases of the wind-direction components are 

 half an hour or so earlier than those of the wind-movement. The 

 range in magnitude of the resultants is much the same in each 

 curve, the maximum being about three and a half times the minimum. 

 The minima occur near sunset ; but the maximum velocity- resultant 

 is at noon, while the maximum direction-resultant is an hour before 

 sunrise. There is, likewise, a second, small, maximum of resultant 

 direction at noon, and a second maximum of resultant velocity at 

 midnight. The vectorial angle (f) made between OE and OE 

 {i.e., the angle EOE) is pretty much the same in either case, there 

 being a relative increase in the angular velocity of the velocity 

 resultant just after sunrise, and a corresponding decrease just after 

 sunset. 



The harmonic constants of the diurnal curve of wind-velocity are 

 given in Table 44, where the angles are reckoned from midnight. 

 Both in amplitude and epoch the curves have strong temperature 

 traits. When all days are considered the amplitudes are seen to 

 follow the same order of magnitude as the amplitudes of temperature. 

 The epochs in V^, Vg, and V3, are each an hour (more or less) earlier 

 than the corresponding epochs of temperature, the epoch in V^ 

 being, however, only half an hour earlier. Also a change from a 

 clear sky to a cloudy varies the first three epochs in the same 

 direction as, though by a greater amount than, those of temperature. 

 As compared with a clear day the effect of a cloudy sky is to 

 accelerate V^ and V2, and to retard V4 by half an hour or so. But 

 in the case of V3 we have again the remarkable cloud feature noted 

 before, the variation here carrying the epoch more than two hours 

 later. This further instance must be regarded as completing the 

 proof that the most potent perturbing periodicity under a clouded 

 sky has a wave-length of eight hours. There are not any pressure 



