426 Transactiofis of the South African Philosophical Society. 



of so curious a fact is likely to be. It is certainly difficult to believe 

 that the temperature of the earth down to 6 feet below the surface at 

 least should fall 2°, or thereabouts, all at once, and remain 

 at the lower point thenceforward. 'Although, for that matter, the 

 yearly averages of the other meteorological elements given in 

 Table 2 show something of the same sort of thing. It is worth 

 noting, moreover, that the averages of certain of the elements 

 at Durban show a corresponding drop in the last five years as 

 compared with the first five, e.g. : — 



TABLE 7. 

 Annual x\verages of Temperature at Durban. 





Black Bulb 

 in vacuo. 



Max. Temp, 

 in Shade. 



Min. Temp, 

 in Shade. 



1897 to 1901... 

 1902 to 1906... 



130-2 

 129-3 



o 

 79-7 

 78-9 



o 

 61-7 

 61-3 



It is possible, of course, that the earlier site of the earth 

 thermometers at Kenilworth may have been somewhat warmer 

 than the later ones ; but even if so it is curious that the 

 difference was not apparent in 1901, the first year of the change 

 of site. 



Perhaps the most interesting features in these earth temperatures 

 are obtained when the annual variation at each depth is resolved 

 into its harmonic constituents. Table 8 gives the epochs and 

 amplitudes as far as the third harmonic term computed on the 

 assumption that each month is equal in length. Some amount of 

 error is naturally involved in such an assumption ; nevertheless, 

 because my object is to show how the various quantities compare 

 iizter se, no very useful purpose would be served by dividing the 

 year into equal aliquot parts. It is further to be noted that 

 the surface temperatures are six-year averages, whereas the 

 deeper temperatures are ten-year. The corresponding elements 

 for the air temperatures are taken over from a previous paper 

 in this series." The time is counted from January 16. As 

 before, the constants for depths of 1 and 2 feet are computed 

 from the averages of three observations a day. 



* " Some Kesults derived from the Constant Values in the Periodic Formulae," 

 Trans. S. A. Phil. Soc, 1903. 



