142 Transactions of the South African Philosoi^hical Society. 



together with values found for Allahabad and Cordoba. All the 

 tabular matter refers to Kenilworth, except when it is stated other- 

 wise at the head of the column. Cordoba amplitudes are expressed 

 in Centigrade degrees. The outcome is to establish the great 

 punctuality of the first, second, and fourth periodic terms, whatever 

 the state of the sky, the phases of Yj being only twenty-five minutes 

 later under clear skies than they are under much cloud. On the 

 other hand, our investigation has accentuated the variation previously 

 found in the case of V3 ; for this angle (of 9° 10' under all cloudy 

 skies) is increased by 17° 44' under moderate cloud, and decreased 

 by 13° 27' under much cloud ; that is to say, there is a difference 

 between a clear day and a very cloudy day of upwards of thirty-nine 

 degrees, the clear falling earlier. The phase of V4 would seem to 

 be practically constant in magnitude in all weathers, and all over the 

 world. 



In view of the interest centred about the behaviour of the angle 

 Y3, it is worth while to inquire whether the fact that neither the 

 diurnal temperature curves for clear, nor those for cloudy skies, are 

 re-entering has any material influence upon its varying epochs. A 

 great influence is obviously not to be expected ; nor is it altogether 

 a fair question to Nature. Yet, as Lord Eayleigh has most pregnantly 

 remarked, " In order to introduce greater precision into our ideas 

 respecting the behaviour of the earth's atmosphere, it seems advisable 

 to solve any problems that may present themselves, even though 

 the search for simplicity may lead us to stray rather far from the 

 actual question." We shall, then, take the observed diurnal curve 

 of temperature under clear skies, considering the midnight value 

 fixed, but reducing every following hourly value proportionately to 

 its time-interval from midnight, so as to make the temperatures of 

 both the preceding and following midnights identical, thereafter sub- 

 jecting the deduced re-entering curve to the usual analysis. Some 

 such process as this is, or seems to be, made use of in reducing for 

 discussion the hourly observations taken four times a month in 

 India. The effect upon the amplitudes is not great ; the angular 

 magnitudes are changed each rather more than a degree, two 

 becoming greater and two less, by nearly equal amounts. In fact 

 we now have — 



p = 62°-4. 



Y, 229° 52' li, 14-252 



Y, 60 4 w, 3-427 



Y3 33 26 u^ 1-088 



Y, 224 14 u^ 0-997 



