138 Transactions of the South African Philosophical Society. 



winter. These effects, however, are largely due to the greater 

 number of clear days in the winter, whereby they have already 

 entered into the production of the normals. The Table shows 

 very strikingly how largely the intrusion of cloud warps the 

 diurnal curve of temperature. 



In Table 8 will be found the hourly temperatures on cloudy days. 

 The values are uniformly decreased throughout so as to bring them 

 into conformity with the supposition that the last column of Table 5 

 is the true average temperature of 576 clear days. If we were to 

 extend the diurnal series to include the midnight following, as well 

 as that preceding, we should find a temperature not exceeding 57°"0, 

 or about a whole degree less at the end of the day than at the 

 beginning. Now we have seen that the opposite is the case when 

 the sky is clear. It follows that the total effect of the clouds is 

 actually to permanently lower the air temperature, by preventing more 

 heat from entering than from leaving the lower reaches of the air. 

 Thus we see that the higher mean temperature on a cloudy day is 

 ■due more to the formation of cloud at the end of a hot period, than 

 to any advantage the clouds may be supposed to have as a heat-trap. 



The range of shade temperature under a clear sky runs along much 

 the same lines as when all weathers are reckoned, that is to say, it 

 is least in April and May, when dew is abundant by night, greatest in 

 the dryness of spring (Table 9). The monthly range for cloudy days 

 is more irregular, giving three maxima in the year. They are caused 

 ■entirely by the access of heat in the day and not by radiation at night. 

 Probably they might be levelled out in a longer series of observations; 

 but as they stand they indicate a partially clouded sky, in which 

 Cumulus prevails, admitting the solar heat with some freedom, but 

 checking, in part, its simultaneous radiative exit. 



Table 10 gives the constituents in the Harmonic Series to the 

 fourth periodic term, counting from midnight.''' For the mean tem- 

 perature monthly values are given ; the values of mean temperature, 

 temperature under clear, and temperature under cloudy skies, are 

 given for the year. 



The term V^, of twenty-four hours' period, is fairly uniform 

 throughout the year, the greatest difference of phase-time for any 

 two months not exceeding a quarter of an hour. The term V2, of 

 twelve-hours' period, has a difference of phase-time of upwards of an 

 hour between the earliest and latest months ; V3 of more than two 

 and a quarter hours ; and V4 of rather more than an hour. Each of 

 the four angles tends to its latest phase-time about midwinter, and 

 its earliest in November. There is also a tendency to a second 

 * Cape Colony Mean Time lor the meridian of 22 J'' E. is understood. 



