20 



will begin to contract from diminishing temperature, and at 

 the moment of the most rapid acceleration of contraction, the 

 barometer will stand at its day minimum, which will probably 

 be late in the afternoon ; and it is found in fact to be from 

 four to five o'clock. From this time the rapidity of the down- 

 ward motion of the air from contraction begins to diminish, 

 and the barometer of course begins to rise ; and at the mo- 

 ment when it is most rapidly retarded in its contraction, the 

 barometer will be at its maximum night fluctuation, and will 

 again be above the mean, but not so much as the day max. 



" This max. is found to occur about ten or eleven o'clock, 

 r. M. The air will now go on contracting more and more 

 slowly, until about daylight, when it will be at rest, and the ba- 

 rometer will again be at a mean. 



" This theory was given by me to the Journal of the 

 Franklin Institute, and published ten or twelve years ago. 



" I ventured in that paper to predict, notwithstanding 

 some alleged observations at St. Bernard's Hospital to the 

 conti'ary, that it would be found by more careful observations 

 that the morning max. fluctuation would be greater in lofty 

 situations on the sides of mountains, provided they were not 

 very lofty, than on the plain below. 



" For it is manifest, that there will be not only a reaction 

 at these lofty situations, (a little less, it is true, than below,) 

 but some of the air will be lifted up, by the expansion of the 

 air below, above the upper place of observation ; which would 

 in all probability more than compensate the diminished re- 

 action at moderate elevations. 



" This prediction has been entirely verified by Lieute- 

 nant-Colonel Sykes's observations in India, and this verifica- 

 tion may be considered as a strong proof of the correctness 

 of the theory. It is quite probable, that max. day fluctua- 

 tion occurs later at considerable elevations than on the plain 

 below. 



" The theory would lead us also to suppose, that at very 



