OF THE AURORA BOREALIS. 31 



Rome, on the aurora of 1726, estimated the height at 260! French leagues, or 

 about 735 miles; 1 and he concludes that the greater part give an elevation of 200 

 leagues, but that they range from 100 to 300 leagues. 



In 1790, Cavendish published, in the Philosophical Transactions, an estimate of 

 the height of an arch observed at different places on the 23d of February, 1781, 

 making it between 52 and 71 miles. Soon afterwards, Dalton wrote his celebrated 

 Essay on the Aurora Borealis, in which he calculated the height of an arch seen 

 in England, February 15, 1793, at 150 miles. But, in the Philosophical Transactions 

 for 1828, appears an article by the same author, in which he determined the height 

 of an arch, from what he deemed satisfactory data and more accurate than any 

 before used, to be nearly 100 miles. 



But the investigations of Professor Alexander C. Twining, on this subject, con- 

 tained in the 32d volume of the American Journal of Science, appear to me more 

 satisfactory and conclusive than an}' within my knowledge, and therefore, deserving 

 of particular consideration. He appears to have been equally fortunate in the 

 selection of his instances, and in the definiteness and variety of the observations 

 on which he founded his estimates. Not all the phenomena of an aurora are alike 

 favorable for this purpose. The corona, for example, being merely the effect of 

 perspective, it would be as useless to attempt to determine its height, as that of the 

 rainbow. The streamers, likewise, are objects too indefinite and evanescent for 

 such estimates. We must be sure that different observers are looking at the same 

 object, before we can attempt to determine its parallax. The boundary line of an 

 arch is such an object, and moving, as it frerpuently does, slowly from north to south, 

 the instant of time when it reaches a known star, furnishes to different observers 

 situated nearly in the same meridian an opportunity for marking its position at 

 a given moment, and thus supplies the requisite data for determining its parallax. 

 Another class of objects the most favorable for this purpose, are detached fragments 

 or patches of an auroral cloud, 2 which, presenting to observers in different places a 

 single insulated object, leaves no doubt that they are all looking at the same thing; 

 and its position with respect to particular stars at a given instant, affords oppor- 

 tunity for observations of the requisite degree of precision. It was from both these 

 classes of objects seen in the } r ear 1835 and 1836, that Professor Twining deduced 



1 Traite de Aur. Bor., p. 62. 



a My observations on the aurora borealis, continued for many years, leave no doubt on my mind of 

 the existence of clouds composed of a peculiar kind of matter, and properly denominated "auroral 

 clouds," not being formed of aqueous vapor, although, from their resemblance to certain forms of the 

 cirrus cloud, they are sometimes confounded with that, and are adduced by certain writers, not 

 personally conversant with these exhibitions, as proof that the aurora occurs in the region of the clouds. 

 Thus, Professor Secchi, in a late able and comprehensive review of the discoveries in Terrestrial 

 Magnetism {Phil, Mag., June, 1855, p. 445), remarks that, "at Rome, the perturbations of the 

 needle exhibit themselves in that particular state of the atmosphere in which there are slightly phos- 

 phorescent clouds, having at night the appearance of the rudiments of the aurora borealis." That 

 these were not ordinary cirrus clouds, composed of aqueous vapor, is evident from the description which 

 he proceeds to give of them, being from a very unusual quarter, being luminous, and being magnetic, 

 while the only proof that they were cirrus clouds was, that they looked like them. 



