36 



Report S.A.A. Advancement of Science. 



Before closing, there is one point not witliout interest : The 

 periodic formulas for temperature are always computed for the diurnal 

 \'ariation — as if we expect that the barometric pressure should rise 

 and fall as the atmospheric planes of equal pressure rise and fall (or 

 fall and rise). Such a process, though quite orthodox, seems to me 

 to be not necessarily the best. For there is no evidence that the 

 annual phenomenon of simultaneous high pressure and low 

 temperature, and vice versa, has its counterpart in the diurnal process. 

 If tension rather than weight should be in question, perhaps harmonic 

 formulce for the diurnal variability, rather than for the diurnal 

 variation of temperature, are likely to lead more quickly to a result. 

 At any 'rate, 1 have been employed lately, in spare intervals, in 

 computing the monthly constants for a number of places whose hourly 

 temperatures are known to me. Whatever view one may take of the 

 method the results are certainly striking. The following are some 

 annual values : — 



1. TREVANDRUiVI. 



Barometric Pressure.. 

 Temp. Variability 

 Temp. Variation 



2. KiMBERLEY. 

 Barometric Pressure.. 

 Temp. Variability 

 Temp. Variation 



3. Greenwich. 

 Barometric Pressure.. 

 Temp. Variability 

 Temp. Variation 



With regard to these comparative values it may be said at once that 

 although the constant angles and co-efficients in the barometric 

 formulae shew a much better agreement with the formulas for 

 temperature variability than they do with those for temperature 

 variation during the cour.se of the year, the epoch of the second 

 harmonic term of temperature variability changes from one side to the 

 other of the epoch of the second harmonic term of barometric pressure, 

 being earlier in summer and later in winter by half an hour or so. I 

 am annexing some Tables, which you will have an opportunity of 

 examining in full. One point is especially worthy of notice, namely, 

 the constancy of the epoch of the first term of temperature variability 

 not only throughout the year at each station, but all over the world. 

 Thus, e.g., the epoch of the first term of temperature variation at 

 Greenwich averages nearly two hours later than it does at 

 Trevandrum, whereas the epochs of temperature variability are less 

 than a quarter of an hour apart. The immutability of the epoch 

 month by month at any station is specially remarkable if it be 

 remembered that the time of sunrise is nearly two hours earlier in 

 summer than it is in winter at Kimberley, and 4I hours earlier at 

 Greenwich. Remembering the importance of the second harmonic 



