On the Constitution of aqueous Ammoma. 19.3 



We tluis see that the statement which Dr. Thomson gave in 

 his 5th edition, and which is repeated in the 6th, is still wider 

 of the truth than I have represented ic in the Dictionary. He 

 says, Water condenses 7S0 times its volume of the gas, and be- 

 comes of specific gravity 0-900. I allowed him inadvertently 

 two-thirds of the quantity, or .520 volumes, which is too much. 

 At the head of the 3d column, for " 520 volumes, " it should 

 have been 4-13 volumes. It is also said, "Sir H. Davy's table 

 differs very little from that of Mr. Dalton, the truth probably 

 lying between them." Now, if v/e consider the enormous error 

 of Dr.Thomson's number 36 per cent., corresponding to his 780 

 volumes ; Mr. Dalton's number 22-2 is but an inconsiderable 

 deviation from Sir H. Davy's number 26, which is undoubtedly 

 very near the truth. 



Glasgow, Feb. 20, 1821. 



Glasgow, March 21, 1S2I. 



Since writing the preceding account, the prosecution of my 

 new system of chemical analysis, announced in the introduction 

 to the Dictionary, of which you have been pleased to give so 

 kind a report, has -led me to construct the following table of 

 the aqueous combinations of ammonia. The experiments, on 

 which it is founded, were conducted with every refinement of 

 precision which I could think of. 



It will be observed that the initial quantity, 26-5, or the real 

 ammonia in water of density 0900, is somewhat greater than 

 that formerly deduced. This increase is owing to the prevention 

 of exhalation of ammonia during the process of experimenling; 

 a pretty dilute water being employed, and its exact neutralization 

 with acid being rapidly accomplished. If, on the other hand, 

 one hunflred grains of water, whose density is 0-9455, be exposed 

 during its neutralization, in a capsule for a few minutes, half a 

 grain of ammoniacal gas may escape, which will cause a defi- 

 ciency of one per cent, in the estimate of alkali in water of 

 0*900. The terms marked with the asterisk are experimental : 

 the others are interpolated. The error in any of them cannot 

 amount, I believe, to more than a small fraction of one per 

 cent. 



B b 2 Table 



