354 M. Bertkollet's Experiments on the Proportions [Nov 

 Proportions very different from those of Berthollet, namely, 



Nitric acid 49-9 



Potash 50-1 



100-0 



It is obvious that in Berthollet's experiments 4-5 gr. of the 

 nitric acid were lost, and this loss seems to have fallen chiefly, 

 if not entirely, upon the oxygen. Let us deduce the weight of an 

 atom of nitric acid from my experiment. We have 45*6 : 54-4 

 :: 6 : 7-158. The weight of an atom of nitric acid comes out 

 7*158, which I believe to exceed the truth somewhat. It 

 seems obvious, therefore, that my mode of experimenting has a 

 tendency to increase the quantity of acid somewhat above the 

 truth ; so that some of the potash must have been sublimed. 

 Probably the error might be diminished by employing sulphate 

 of ammonia instead of sulphuric acid to decompose the nitre. 

 It is obvious that a much lower heat at least would be requisite. 

 I intend to repeat the experiment in this way ; and may state the 

 result hereafter. 



In the mean while, as the errors in the two experiments lie on 

 different sides, the probability is, that the mean of the two will 

 come very near the truth. Such a mean will give for the com- 

 position of nitre, 



Nitric acid 52*15 



Potash 47-85 



100-00 



Now 47-85 : 52-15 :: 6 : 6-539. This weight, 6-539, though 

 not absolutely correct, is undoubtedly much nearer the truth than 

 either the number derived from Berthollet's experiment or mine. 



Dr. Wollaston made an experiment which was susceptible of 

 very considerable precision. He saturated a given weight of 

 bicarbonate of potash with nitric acid, and ascertained the 

 weight of the nitre resulting. Nitre turned out a compound of 

 100 acid -f 86-764 potash. This is equivalent to 



Nitric acid 53-543 



Potash 46-457 



100-000 



The error in this experiment lies upon the same side as in 

 mine, but it is smaller. Both errors are not more than sufficient 

 to counterbalance Berthollet's error on the opposite side. There- 

 fore, if we take the mean of all the three, we must obtain a 

 result which will approach still nearer accuracy. This mean 

 gives us nitre composed of 



4 



