HISTORY OF TUE ATOMIC THEORY. 2 1 



very improbable that the height of the atmosphere should be 

 subject to such fluctuations, or that it should be regulated in 

 any other manner than by the weekly or monthly mean tem- 

 perature of the lower regions; because the mean weight of the 

 air is so nearly the same in all the seasons of the year, which 

 could not be if the atmosphere was as high and dense above 

 the summits of the mountains in winter, as it is in summer. 

 However, the decision of this question need not rest on pro- 

 bability; there are facts which sufficiently prove, that the 

 fluctuation of density in the lower regions has the chief effect 

 on the barometer, and that the higher regions are not subject 

 to proportionable mutations in density. In the Memoirs of the 

 Royal Academy at Paris, for 1709, there is a comparison of 

 observations upon the barometer, at different places, and 

 amongst others, at Zurich, in Switzerland, in lat. 47° N., 

 and at Marseilles, in France, lat. 43° 15^ N. ; the former 

 place is more than 400 yards above the level of the sea; it 

 was found that the annual range of the barometer was the 

 same at each place, viz., about 10 lines; whilst at Genoa, in 

 latitude 44° 25' N., the annual range was 12 lines, or 1 inch; 

 and at Paris, latitude 48° 50' N., it was about I inch 4 lines. 

 In the same memoir it is related that F, Laval made obser- 

 vations, for ten days together, upon the top of St. Pilon, a 

 mountain near Marseilles, which was 960 yards high, and 

 found that when the barometer varie<l 2 j lines at Marseilles, 

 it varied but 1 } upon St. Pilon. Now had it been a law, 

 that the whole atmosphere rises and falls with the barometer, 

 the fluctuations in any elevated barometer would be to those 

 of another barometer below it, nearly as the absolute heights 

 of the mercurial columns in each, which in these instances 

 were far from being so. Hence then it may be inferred, that 

 the fluctuations of the barometer are occasioned chiefly by a 

 variation in the density of the lower regions of the air, and 

 not by an alternate elevation and depression of the whole 

 superincumbent atmosphere. How we conceive this flue- 



