152 Transactions of the South African Philosophical Society. 



Hann says »that it is almost undisturbed, despite its slight amplitude. 

 Under cloudy skies at Kimberley the epochs of V3 are much alike in 

 both temperature and pressure ; but while a clear sky makes the 

 temperature epoch 25° earlier, the pressure epoch becomes 39° later. 

 No doubt the true nature of the connection between the third com- 

 ponents of temperature and pressure is not by this made any less 

 obscure ; but it does, I think, show that there is such a connection, 

 and also that the cloud-period is sufficiently near to eight hours to 

 effectively disturb the phase-times of V3. The amplitudes, also 

 again, reflect the same phenomenon ; for while the clear-day ampli- 

 tude of temperature in 11^ is double that of the cloudy day, the cloudy^ 

 day amplitude of pressure is double that of the clear day. \ 



The imperturbable angle in the fourth component of temperature 

 is in striking contrast to the vacillation of the corresponding angle of 

 pressure. This matter may come up for discussion at some later 

 date. It will be sufficient here to suggest that failing a barometer 

 reading comfortably to four decimal places, a longer term of obser- 

 vation would be required to smooth out much of the pressure 

 irregularity in V^. 



That clouds constitute the disturbing factor in both first and third 

 components of pressure rather in virtue of the temperature they in- 

 terrupt than by the increased amount of moisture they accompany 

 seems made out by the subdivision (given in the first and second 

 columns of numbers) of clear days into sets of greater or less dew- 

 point than the monthly means ; and after, as said before, correcting 

 them for secular variation. Saving the doubtful fourth component 

 the amplitudes are not affected at all, nor the epochs very much. 

 Perhaps it may not be out of place to observe that one reason why 

 the column for all clear days is not the mean of the two subdivisions 

 is that each column is made up on the assumption that what is 

 called the mean yearly average is derived from the twelve sets of 

 results for each month. 



A periodic curve may be built up by the superposition of any 

 number of periodic curves. Thus the mean diurnal curves of tem- 

 perature, or pressure, are constructed from the monthly diurnal 

 curves ; the equation of time is built upon the separate irregular 

 motions of our worst timekeeper, the sun. Both classes of curve 

 represent physical facts as truly as the curves from whence they 

 sprang. But while the subordinate curves are most easily combined, 

 it is not so easy to separate them out again ; nor really certain that 

 the decomposition will necessarily yield any one of the original 

 curves. That is in general. The question here is, then, Are the 

 harmonic constituents, formed as above, actual realities, or mere 



