90 CAUSES which affeEl the ACCURACY 



fpaces, into the atmofphere ; and, in general, the differences 

 of temperature muft be proportional to the differences of eleva- 

 tion. 



IT is to be underftood, however, that this law is fubject to 

 certain anomalies, both annual and diurnal, and thofe inter- 

 mixed with other accidental irregularities, which it would be 

 difficult, perhaps impoflible, to afcertain. All that can be faid 

 of it is, that it is the law which nature tends to obferve, and 

 that the fum of the deviations from it, on the one fide, is pro- 

 bably equal to the fum of thofe on the other. In an effect that 

 is perpetually fubject to the action of accidental and unknown 

 caufes, the difcovery of a mean, from which the departures on 

 the oppofite fides are equal, is all that we can reafonably ex- 

 pect ; and it is fufficient for us to know, that, though any par- 

 ticular conclufion may involve an error, yet, if a multitude 

 of inftances be taken, the errors will certainly correct one ano- 

 ther. 



4. IF, therefore, H be the heat at the furface of the earth, 

 and h the heat at any given height a, above the furface, the 



heat, at any other height, as #, will be H . At a 



a 



medium, it is found, that FAHRENHEIT'S thermometer falls a 

 degree for every 300 feet that we afcend into the atmofphere ; 

 ib that, if x is expreffed in fathoms, the heat, at that height, 



. _ x 



5* 



5. BUT though we are thus led to conclude, that the decreafe 

 of heat in the fuperior flrata of the atmofphere is proportional 

 to their elevation, there is no reafon to fuppofe, that the con- 

 denfation produced by that decreafe is alfo uniform. Indeed, 

 the experiments of General ROY have placed it beyond all doubt, 

 that the variations in bulk of a given quantity of air are, by 

 no means, proportional to its variations of temperature. Thofe 

 experiments, though very numerous, are too few to afcertain 



exactly 



