136 Transactions of the South African Philosopliical Society. 



hourly temperatures ever lies evenly upon a re-entering curve, the 

 columns have each been extended to include the second midnight, 

 and 1 a.m. ; these, however, have not been used in forming the mean 

 values of the last line. The time of minimum temperature evidently 

 accompanies the time of sunrise pretty closely month by month. 

 But the time of maximum temperature falls irregularly anywhere 

 between half-past one o'clock p.m. and a quarter to three. Fortu- 

 nately we shall be in a position later on to give some explanation of 

 this circumstance. It appears that a period of four years is not 

 sufficient to eliminate all irregularity in the monthly means ; for the 

 maximum temperature of February happens in the present case to 

 be greater than that of January. Yet the shape of the diurnal curve 

 will be found true enough. 



Table 4 gives the mean temperature variation in each hour, i.e., the 

 rise or fall of temperature from one hour to the next. The heavier 

 type indicates that the temperature is rising ; in all other cases it is 

 falling. A glance is sufficient to show us that the former operation 

 lasts for about eight hours, only, of the twenty-four, month after 

 month, with a small extension of time in the spring months. It 

 would appear, then, at first sight, that the temperature shows a 

 disposition to rise for a given length of time after sunrise — a quite 

 fallacious conclusion, however, as we shall see presently. From a 

 consideration of Table 4 we may distinguish the following epochs of 

 temperature for the year : — 



1. The transition from a falling temperature to a rising {i.e., the 

 minimum) about the time of sunrise, say at half-past five o'clock 

 a.m. for the year. 



2. The time when the temperature is rising most rapidly, about 

 two hours after sunrise, or about eight a.m. on the mean of the year. 



3. The transit from a rising temperature to a falling {i.e., the 

 maximum) between XIII. and XV., or say at about a quarter past 

 two o'clock p.m. for the year. 



4. The time when the temperature is falling most rapidly, a little 

 before sunset, or not much later than half-past five p.m. for the year. 



Table 5 gives us the hourly and monthly temperatures under clear 

 skies. The last column (giving the year) is the average of the twelve 

 monthly columns. By reckoning it in this way the greater number 

 of clear days in the winter is not allowed to unduly influence the 

 final result." We see that under clear skies the average shade tem- 

 perature is, upon the whole, less than the normal, though slightly 

 greater in the middle of summer. The total diurnal range is, how- 



* It is equivalent to the supposition that there was an equal number {/.{',, 12) of 

 clear days in each month. 



