Diurnal Meteorological Co7iclitions at Kimherley. 159 



the resultant direction being 4,942 hours, almost exactly from E. 

 by N. : its magnitude being therefore rather less than 14 per cent, of 

 the whole period. But in the whole six years of observation, out of 

 52,560 hours of wind (of which 465 were variable), the unbalanced 

 components were — 



From the north 888 hours ; 



From the east 3,134 hours, 



the resultant direction being 3,258 hours from a rather more 

 northerly direction ; its actual magnitude being therefore a shade 

 over 6 per cent, of the period, or, say, an hour and a half ]jcr 

 diem. 



It has been shown, in the paper referred to, that an inverted 

 diurnal temperature curve and a curve of east component variation 

 look very much alike, and that if they be drawn to suitable scale one 

 will overlie the other for some hours of the day. The same fact is 

 brought out in another way by Table 36, where the increase in bulk 

 of the east component is compared with the decrease of temperature 

 hour by hour. The heavier type signifies a positive decrease of tem- 

 perature, or increase of east component respectively, and ordinary 

 type signifies negative quantities. The temperature change is for 

 the hour ending with the stroke ; the wind hours are in reality half 

 an hour later. It is seen that after the quantities first change sign 

 (about VI.), the wind variation is more gradual than the temperature 

 variation, but more abrupt at the second change of sign (about XVI.). 

 Apart from this the gradients are remarkably similar, and have their 

 maxima about the same time, whether the sky be clear or cloudy. 

 There is an exception to the rule in the hours on one side or the 

 other of midnight, which has been explained as due to the influx of 

 easterly winds. 



The north and east components of Table 35 have been submitted 

 to analysis, the result as far as third harmonic term appearing in 

 Table 37. The angular quantities are reckoned from zero at mid- 

 night. The amplitudes u^, u^, and u^ are total, not average, and it 

 might therefore be anticipated that they would be greater for the 884 

 cloudy days than for the 576 clear days ; and a fortiori still greater 

 for the whole four years. They are reduced to yearly proportional 

 and comparable magnitudes in the quantities «i, z4> ^^^ ^4- 



The first and second amplitudes of the north component are 

 decreased under cloudy skies, the second noticeably so ; the third 

 amplitude is slightly increased. The east component shows the 

 decrease very plainly in each amplitude ; the second cloud amplitude 



