1885.] S. A. Hill — Olservations of tlie Solar Thermometer, 37 



From Table III., it appears that the variation of the solar heat from 

 year to year has been similar to that deduced from the Allahabad obser- 

 vations, while the range indicated is even greater. The highest annual 

 mean is that for 1878, when the sun spots were at a minimum, and the 

 lowest, that for 1881 or 1882, when the spots were probably at a maxi- 

 mum. 



Such a very distinct variation in the sun spot period must, I think, 

 be the effect of a real variation in the emission of solar energy, but the 

 great range of the observed inequality is probably due in part to other 

 causes. It is evident from the means at the foot of Table II. that all 

 the terrestrial causes of variation have not been eliminated, for, whilst 

 these means are nearly constant for the months of September, October, 

 November, December, and January, they are much more variable and 

 considerably greater in the dry hot months of the year. If the excess 

 temperature of the solar thermometer above that of the air were a true 

 measure of the solar radiation, it should, when the observations are 

 corrected for atmospheric absorption, give somewhat lower results for 

 May and June than for December and January, on account of the greater 

 distance of the sun in the former months. The opposite variation which 

 is observed must be due to heat reflected from the bare hot ground and 

 from the dust particles suspended in the air. 



Since the bareness and hardness of the ground-surface under the 

 instrument and the quantity of dust in the air are due to the same 

 cause, and increase pan passu ; it is, perhaps, justifiable to assume that 

 the increase of the observed effect which is due to reflexion may be 

 taken proportional to the quantity of dust. That is to say, we may put 



A = -2 (1 + ad). In this formula A represents the mean value for any 



month at the foot of Table III.; S, the value this mean would have if 

 the ground were moist and grassy, the air free from dust, and the earth 

 at its mean distance from the sun ; r the radius vector of the earth at 

 the middle of the month ; d the proportion of dust in the air ; and a a 

 coefficient which remains to be determined. The proportionate num- 

 bers for dust which I have assumed are : — 



Jan. Feb. Mar. Apl. May June Sept. Oct. Nov. Dec. 

 8 6 7 9 10 6 1 2 

 These differ somewhat from those already adopted for Allahabad, 

 but, perhaps, represent the facts more justly. By means of the formula, 

 with these values for d, we find : — 



S = 73-0^ 

 a = -0157 

 In the month of May, therefore, the observed radiation is 15 or 16 per 



