160 Transactions of the South African Philosophical Society. 



being 55 per cent, of the clear and the thh^d amphtude less than 

 22 per cent, of the clear. 



The epochs of the north component show that under cloudy skies 

 Vj is 42 minutes earlier, V2 is 32 minutes earlier, and V3 is 21 minutes 

 later than under clear skies. The epochs of the east component 

 show that under cloudy skies Vi and V3 are practically unaffected, 

 while V2 falls nearly 73 minutes earlier. 



It can scarcely be claimed from this result that the north and 

 east components of wind-direction bear any very simple harmonic 

 relationship to the concurrent temperatures and pressures. The 

 east component epochs in which, bearing Table 36 in mind, they 

 might be looked for, are indeed less in agreement than the north ; 

 and whereas the cloud effect upon V2 is almost non-existent in both 

 temperature and pressure, in the case of the east component, 

 direction it is practically the only effect to be detected. Certainly 

 there is no trace of an eight-hour cloud period in the east component 

 of direction, so far as the phase-time is concerned : if anything, the 

 place where the cloud disturbance of an eight-hour period is exhibited 

 is in the amplitudes. In the north component epochs, on the other 

 hand, there are traces of a temperature effect, each angle varying in 

 the same direction, though not at quite the same rate, as those of the 

 temperature. 



The conclusions we may draw from all these circumstances are : — 



1. That the wind-direction is determined largely by the position 

 of the geographical area at which the temperature is highest irre- 

 spective of whether this maximum is high or low — the orbits of the air 

 particles, that is to say, depending upon the sun's apparent path ; 

 and — 



2. That clouds do not so much determine wind-direction as 

 accompany perturbations of the normal directions generated by 

 remoter influences. 



Table 38 gives the mean wind-velocity, hour by hour, during each 

 month, and for the year, from the average of the observations taken 

 during the four years 1898-1901. The subdivision into clear days 

 and cloudy, will be found in Tables 39 and 41, and the respective 

 deviations of these from the normal in Tables 40 and 41. In the 

 latter a wind-velocity greater than the mean is printed in heavier 

 type. 



Under clear skies the velocity of the wind at any hour in any 

 month is rarely greater than the mean ; in the case of the yearly 

 averages, never. Consequently, under cloudy skies, the yearly 

 averages are always greater than the mean. At midnight the 

 velocities under both clear and cloudy skies are appreciably equal 



