88 S. A. 'Ri[\—Olservat'io7is of tJie Solar Tliermomcter. [No. 1, 



cent, greater than it would be if tlie ground were grassy and the air free 

 from dust, other things being the same. 



It follows from this that in a dry year the solar thermometer will 

 give higher indications than in a damp one when due allowance is made 

 for variations in atmospheric absorption. There can be little doubt that 

 part of the great excess of the results for the latter half of 1877, the 

 whole of 1878, and the first half of 1879, above those for subsequent years, 

 is due to this cause ; which is still better illustrated by comparing the 

 months of March, April, and May, 1877, with the same months of 1879. 

 In the former year the spring months were unusually showery, and, in 

 consequence of this, the ground-surface was covered with grass, whilst 

 in 1879 no rain fell and the ground was quite bare and dusty. 



The conclusion to be drawn from this investigation seems to be that, 

 while the results indicate a rather strong presumption in favour of the 

 hypothesis that the emission of solar heat varies inversely with the 

 number of sun spots, the hypothesis can only be definitely proved by 

 observations of some kind of actinometer which is protected from 

 reflexion and receives direct solar rays only. Probably, the form of 

 instrument which will be found most useful is a thermopile turned by 

 clock-work so as to face the sun and attached to a reflecting galvano- 

 meter by means of which the heating effect can be photographically 

 recorded. 



The absoption coefiicients given in Table IV. are least in the cold 

 weather months and greatest in the hot season and the rains. Since 

 these coefiicients are dependent upon the constitution of the atmosphere, 

 it may be assumed, as it has been in my previous paper, that the con- 

 stant p of Ponillet's formula is the product of three factors, a^, fS^, and y^, 

 where h is the barometric pressure, / the pressure of vapour, and d the 

 proportionate number for dust. In strictness, h should stand for the 

 pressure of the dry air only, but as the aqueous vapour thins out about 

 three times as fast on ascending as it would do on the hypothesis of an 

 independent vapour-atmosphere the pressure of the dry air is not (h — -/), 

 as some suppose, but something very little less than h. 



The mean values of the barometric pressure and tension of vapour 

 observed at noon in the days given in Table I., are the following :— 



Pressure. Vapour Tension. 



January 29+ '714 in. -304 in. 



February '660 '281 



March -529 -374 



April "415 -406 



May -293 -668 



June -171 -674 



