182 On the P/icBnomemn of I he Aurora Borealis. 



the different stations of observation by the observed angle of ele- 

 vation, these intersections will be t,he height of the arch at such 

 places. The above method would not apply so well to common 

 streamers; but when stationary for a time, circumstances might 

 be determined concerning them by those or similar observa- 

 tions. 



It is to be observed, that a circumstance has been noticed 

 in high northern latitudes, which cannot be accounted for other- 

 wise than by supposing an extraordinary degree of refraction to 

 exist in the polar regions. The circumstance is the appearing 

 -of the sun above the horizon, many days before that luminary 

 could be expected from the relative position of the earth at the 

 time, with only the quantity of refraction usually allowed. What- 

 ever is the cause of the increased refraction at the north pole, it 

 will greatly affect those phaenonieua, and must be ascertained &i 

 least before their height can be accurately found. I believe the 

 cause commonly, and perhaps truly, assigned for the great re- 

 fracting power above noticed, is the dense state of the atcnosphere 

 produced by extreme cold : there is however no certainty that 

 -the singular state which causes mock suns, landscapes, ike. in 

 the air, may not be concerned here. 



It may be asked. If the cause of the aurora lorealis is as above 

 explained, why are they not always visible at the two periods be- 

 fore meiitioned ? The answer is. So they would, were the atmo- 

 sphere always in that state of purity which is essential to their 

 exhibition. But the complete alisence of clouds' in so vast an 

 extent of a northern sky must be very rare. 



In a more limited space, the atmosphere must be more fre- 

 quently in such a state of serenity as to favour the exhibition of 

 those phaenoraena; which, with other causes before mentioned, is 

 no doubt the reason of the frequency of their appearance in more 

 northern latitudes. A conjecture has been offered concerning 

 the cause of the phaenomena in question, which being more ab- 

 surd than that already I think disproved, it is scarcely necessary 

 to take the trouble of refuting. It is, that tliose !ia;hts are pro- 

 duced by the combustion of hydrogen gas that has escaped from 

 -the earth, and accumulated in upper regions of the atmosphere. 

 It is evident that such combustion, were its existence not alto- 

 gether im|)robable, could neither be periodical nor local, nor ex- 

 hibit an appearance at all similar to streamers. Were this in- 

 flammable gas collected where the oxygen of the atmosphere had 

 access to it, and there kindled, they would instantly unite with 

 tremendous noise, and the water formed fall to the ground. 



But if it ascended above the atmosphere, it would be entirely 

 beyond the means of ignition; neither could combustion there 

 go on, from want of oxygen. 



In 



