]j 



Atmospheric Origin of the Aurora, ^'c. 



was in a single instance, and that on the occasion of two con- 

 secutive auroras, the latter tending to prolong and increase the 

 clearness. This instance being omitted — as it should be — the 

 decrements of clearness during the 24 hours succeeding the au- 

 rora are to the increments, as 113 to 1, the increment having 

 been in one instance one tenth, and the whole decrement in 

 thirty instances, 112 tenths. On none of the eight instances in 

 which there were auroras on two consecutive nights, had the 

 cloudiness increased on the evening of the second, as compared 

 with that of the first. The mean decrement of clearness for the 

 remaining 24 instances, was .46. Hence, to give a popular state- 

 ment, approximately true — the evening of an aurora is, on an 

 average, twice as clear as the succeeding evening, unless another 

 aurora occurs on the latter ; in which case, the sky continues 

 equally clear. As the forenoon succeeding an aurora is in gene- 

 ral unusually clear, this great decrement of clearness usually takes 

 place in a few hours, whilst the increments had required several 

 days. ' - 



The folloAving table, (abstracted from those on which the nine 

 propositions are founded, ) shows the mean temperatures at 9 P. M. 

 of the days of the different meteors, and on the evenings one and 

 two days previous ; also the mean number of days previous, when 

 the changes of pressure and temperature commenced.* 



The number of vertical beams is so small, as to forbid confi- 

 dence in mean results as to elapsed time. In the case of thfe 

 other meteors, we see a pretty near correspondence as to the 

 times when the thermometric and barometric changes commen- 



* Certain errors which, through the inadvertence of an assistant, had crept into 

 the tables, are here corrected. 



