on the Temperature of the Sun. 



543 



found out that this single formula is not sufficient to represent 

 the phenomenon for one entire day. I then tried to employ 

 other empirical formulae, and in particular the two more com- 

 plex exponential formulae 



q=zabv+ e , 



q = u + m t 8 € + ny e , 



a, b, m, n, a, /3, 7 being empirical constants ; but not one of the 

 formulas employed is able to represent with sufficient precision 

 the whole phenomenon. 



In using the following three couples of values with the for- 



mula q- 



= abv+ e , 



6 = 1-5, 2 = 203*8 



) 





€=2-25, 9=174-0, 





6 = 3-00, 2 = 152-0, 



we obtain 







y= 7-06180, 





log& = 8-966537-10, 







a = 309-27. 





E. 



q observed. q calculated. 



Difference. 





15 



203-8 







175 



192-6 19276 



016 





200 



182-8 182-88 



008 





225 



1740 







2-50 



1660 165-98 



-0-02 





275 



158-9 158-71 -017 





200 



1521 





4 » 



1327 134-42 +1-72 



In this case the formula answers pretty well. 

 By applying the same formula to the curve of 28th Sep- 

 tember for the interval 6 = 1'4 and e = 2'6, we obtain 



a = 292-25. 

 In the curve of 1st November, interval 6 = 2*0 and 6 = 3*2, we 

 have 



« = 333-24; 

 For the 10th October, 



tf = 339*95. 

 The other curves would give results as much smaller as the 

 interval taken for the calculation of the constants was greater. 

 21st October, interval e = l*8 and e = 10, 



« = 260-9. 

 21st October, interval 6=1*8 and e = 6, 

 a = 272'2. 

 But between the calculated values and the observed values 

 the differences are too great, as the following Table shows: — 



