[ 148 ] 



ceeded the various refults of different chymifts with refped to 

 the proportion of it neceflary to alkalize nitre. 



It is almoft fuperfluous to add that the charcoal fhould be 

 projeiSled in very fmall portions. I feldom projeded more than 

 one or two grains at a time ; each operation lafted from twenty 

 to twenty-five minutes nearly. There is always fome portion of 

 nitre undecompofcd being protected by the furrounding alkali ; 

 this error is unavoidable but very fmall. Even the pofition of 

 the crucible in the furnace is not indifferent, for if it be near the 

 flue more coal muft be employed, which I attribute to the tor- 

 rent of air which in that cafe afFeds it and carries away more 

 than when the crucible is nearer to the anterior part of the 

 furnace. 



It may perhaps be fufpeded that this and fome other incidental 

 errors may be avoided by previoufly mixing the nitre and coal, 

 and projeding the mixture in fmall portions into a red hot cru- 

 cible; but not to mention that this method fuppofes the due 

 proportion of thefe two fubftances to be known, which cannot 

 be known till after the experiment, and alfo that every atom of 

 thefe fubftances is in perfed contad with the other fubftance, elfe 

 they cannot ad on each other; independently, I fay, of thefe 

 unfounded fuppofitions, this mode of experimenting is ftill more 

 fallacious than the former, as, during thefe projedions, a confide- 

 rable proportion of the nitre is fcattered and difperfed, and 



may 



