[ 357 ] 



on one day 9^*, on another 8*^,9, on another 10^,5 ; whereas, 

 calculating from the above rule, it fliould always be 9^,9; fo on 

 the Mole he found the diminutions of heat on different days 

 7") S'*!^) 6^,6, 7*^,9 ; whereas, by the above rule, it flaould be 

 14^,8 every day. Here the error amounts to the double, or 

 more. 



To attain a clofer approximation, we muft refort to fome fixed 

 point above, whofe temperature is known, and fo diflant, that 

 its variations, if it undergoes any, fliould, with refpeit to that 

 diftance, be inconfiderable ; then,, this diffance being known, and 

 having another fixed point below, namely, the degree of heat 

 obferved, and framing a fet of equi-diftant intermediate terms, we 

 have the necelTary elements of adecreafing arithmetical progrelTion, 

 which all obfervations concur in eflablifhing as that obferved by 

 heat in its progrefs upwards, during the fummer months, and in 

 ferene weather. 



The fixed point in the fuperior region of the atmofphere to 

 which I here allude is the height of the upper term of congelation. 

 This, as well as the lower term of congelation, was firft announced 

 by Bouguer, fo long ago as Ann. 1749, though no ufe has fiiice 

 been made of either in meteorological enquiries. 



Vol. VIII. Yv The 



