[ 448 ] 



3tio. Per table, looo cubic inches at 32'', barometer 30,3, weigh 

 333,1 grains, therefore 912,52 fhould weigh 303,96 grains. The 

 error of this coefficient is therefore 303,96 — 299,56 = 4,4 grains 

 in defed ; it retrenches therefore too little. 



I SHALL quote no more of thefe experiments, as, not having 

 made them with a view of fo nice a teft as the prefent, I cannot 

 anfwcr for their accuracy to half a grain, and the greateft nicety 

 would be requifite. Yet I believe, upon the whole, this to be a 

 better method of finding a coefficient than any manometrical ex- 

 periments. 



Method of finding the Coefficient. 



1°. The barometrical height being found to three decimal places, 

 and the temperature from 80® to 'r^^'^ true to one decimal place, 

 and alfo the hygrometer, find the weight of 1000 cubic inches 

 of air truly to -;V of a grain = W. 



2do. Find the weight of 1000 inches at 32'' at the fame ba- 

 rometrical height _ /, by the table, or ftill more nicely. Now 

 fince / denotes looo inches, /Falfoat 32° will denote a lower 

 number of inches, fince it is a lower weight; then ^ /. 1000:: 

 W. N. 



3tio. 





