i 7a] 
Mr. Berthollet* took 288 grs. of nitre and 72 of fulphur ; 
after expofing them to a fuflicient heat he found 12 grs. of 
fulphur fublimed, and in the retort 228 grs. of tartar vitriolate. 
He already proved in a preceding memoir}, that nitre is _ 
entirely decompofed by ¢ of its weight of fulphur, which is 
the proportion here employed, therefore we muft fuppofe that 
288 grs. of nitre contained as much alkali as 228 of tartar 
vitriolate, and no more. Now by my computation 288 grs. 
of nitre contain 132,96 of mere alkali, and 228 grs. of tartar 
vitriolate contain 125,4. So that if I am miftaken, it is in 
attributing too much and not too little alkali to nitre, and my 
error is only 7,6 grs. in 288 of nitre, or 2,6 per cent. 
By:Mr. Bergman’s computation 288 grs. of nitre contain 
141,12 of alkali, and 228 grs. of tartar vitriolate contain only 
118,56; fo that his error amounts to 22,56 grs. in 288 of nitre, 
or 7,8 per cent. 
I now proceed to examine the accuracy of our refpective 
computation of the proportion of acids in thefe two falts, ftill 
taking Mr. Berthollet’s experiment as the criterion. 
In this experiment the fum of the ingredients was 288+72 
=360 grs., but the fum of the products was only 288+12 
=240 grs.; therefore 120 were loft or diflipated in air. And 
Mr. 
* Memoirs, Par. 1782, p. 603. + Mem. Par. 1781, p. 232. 
