Atmospheric Origin of the Aurora, 6fc. 159 



ced before them all ; and find, in the relative temperatures re- 

 quired for them, a corroboration of the conclusion drawn from 

 the time of the succeeding storm in relation to their relative 

 heights in the air. 



The absolute temperature is, for the seasons of their occur- 

 rence, low for all, and of itself affords evidence of the existence 

 of crystals. From semi-monthly observations for five years, on 

 two springs at Schenectady, I have inferred, that the mean tem- 

 perature of the earth there is 48.8°,- and this accords nearly v/ith 

 the mean temperature of the air in that vicinity for the last ten 

 years. Should we make allowance for the daily mean, and for 

 the mean seasons of the year in which the aurora occurs, we 

 should have a still more just and striking view of the cold usu- 

 ally required for its production.^ The barometer rises and the 

 thermometer falls before an aurora, and the mean length of time 

 is about two days; and consequently these changes commence 

 about four days before the storm, or about three and a half days 

 when there are not two auroras in succession. 



This affords one of the earliest and surest prognostics of the 

 storm, and is more to be relied on than even the subsequent de- 

 pression of the barometer, which, in modern times at least, seems 

 solely to have attracted attention. It would be curious, (though 

 it is perhaps improbable, and I have not seen the original, ) if this 

 early ascent of the barometer were that alluded to in the long 

 since banished rule of Pascal. Though this patriarch of this 

 branch of science may, as is alledged, have fallen into a grave 

 error in regard to this, yet there will be revived a certain modifi- 

 cation of his rule, that the barometer rises before a storm ; and 

 perhaps he may be acquitted of the error and prove to be the ori- 

 ginal discoverer. 



That the changes of pressure and temperature commence be- 

 fore the am'ora, accords with the above theory. They are to be 

 regarded as among the causes rather than the effects of the au- 

 rora. Yet that they continue a little beyond the time of it, I 

 have long since observed, and expressed it- by the rule, that the 

 barometer is usually rising, and the thermometer falling, on the 

 evening of an aurora. 



Within a few years, an interesting confirmation of the above 

 theory, so far, at least, as to the fact of a connexion between at- 

 mospheric vapor and magnetism, has been presented -in many in- 



