DECREMENT OF TEMPERATURE^ ETC. 135 



change of season^ in which a greater difference of tem- 

 perature is mainly due to an increase of temperature at 

 the lower station. 



The amount of barometric oscillation at Geneva, with 

 a difference of temperature of i8°'25, is 36*214 in., and 

 on the Great St. Bernard 29*701 in.; and with a differ- 

 ence of 2i°*i8, the amounts are 42*284 in. and 33*892 

 in. respectively. The difference of the amounts at Geneva 

 is 6*070 in., and on the Great St. Bernard 4*191 in. 

 These results are also very remarkable, as it would 

 naturally be supposed that an increased disturbance of 

 the atmosphere would tend to produce a more equable 

 distribution of temperature ; but the occurrence of a ba- 

 rometrical pressure below the mean, with a difference of 

 temperature and amount of oscillation both above the 

 mean, the temperature of the lower station being also 

 slightly lower, is still more remarkable, as it is apparently 

 inconsistent with all the theories which have yet been 

 advanced to account for the irregular oscillations of the 

 barometer. 



The increase in the amount of barometric oscillation, 

 with increase of difference of temperature between the two 

 stations, appears to indicate that the disturbances of the 

 atmosphere which produce disturbances of barometrical 

 pressure take place chiefly in a horizontal, and not in a 

 vertical direction. 



The diminution of mean temperature with increase 

 of difference of temperature, and increased rainfall, points 

 clearly to the operation of a cooling agency sufficiently 

 powerful to neutralize completely the effects which, ac- 

 cording to the theory above alluded to, ought to be 

 produced by the latent heat of aqueous vapour when ren- 

 dered sensible by the condensation of the vapour into 

 clouds and rains. In my paper " On the Theory of Uain,^^ 

 I was led to conclude " that the formation of rain might 



