1885.] S. A. Hill — Observations of the Solar Thermometer. 33 



The differences between the numbers given in the table depend 

 primarily upon variations in the sun's incident heat and in the propoi-tion 

 of this which is absorbed before reaching- tlie instrument, tlic latter 

 being dependent upon the composition of the atmosphere and the ob- 

 liquity of the rays. Minor causes of variation depend upon the instru- 

 ment itself and the nature of its surroundings, and upon the reflexion of 

 heat from cloud, haze, or dust particles in the air ; the instrument being 

 designed to receive rays coming from all directions and not parallel rays 

 only. 



As regards the instrument itself, if its thermal capacity bo large, it 

 will be sluggish in responding to any change in the incident radiation. 

 This will cause the incident heat in the afternoon to appear greater than 

 in the forenoon. The Lucknow observations are not appreciably affecLed 

 with any error of this sort, since the thermometer is a small one with a 

 bulb not much larger than a pea and a tube so fine in bore as to make it 

 easy to estimate tenths of a degree Fahrenheit in reading it. It would, 

 therefore, respond almost instantaneously to any change in the incident 

 radiation, were it not that owing to friction in the narrow tube the 

 mercurial column seems to rise and fall by slight jumps and starts. 

 Observations made at equal hour angles before and after noon may 

 be expected, however, when combined, to eliminate any error due to the 

 sluggishness or jjer saltiun action of the thermometer. 



The effect of changes in the nature of the ground-surface beneath 

 the instrument and in other objects in the vicinity cannot be readily 

 eliminated. They have been reduced to a minimum, however, by placing 

 the thermometer in the centre of an open space on a stand 4 feet high. 



The antecedent probability that the variations in the absorptive 

 power of the atmosphere must be very considerable is great, for, even if 

 we have nothing else to go upon but the observations in Table I., these 

 indicate that the totpJ absorption is almost as great in June, when the 

 incident rays at noon are nearly vertical, as in December, when the sun 

 rises only 40° above the horizon. To estimate the absorbing power, it is 

 necessary to make some assumption regarding the manner in which it 

 varies with the thickness of the atmosphere traversed by the rays. The 

 only simple formula yet proposed which gives results in fair accord with 

 observations made on a clear day is that of Ponillet. This formula, it 

 is true, applies in strictness only to radiation of one definite kind, because 

 the atmospheric absorption is selective ; and Langley* has shown, by a 

 hypothetical example, that the approximate constancy of absori^tion indi- 

 cated by applying the formula to observations made on the same day 

 at the most various angles of obliquity may co-exist with an error of 



* Zeitsch. d. Oe»t. Gesellsoli. fur Met., B. xx, S. 86. 



