1822.] in modifying the Specific Gravity of Gases. 307 



Here the error is 0*0004, or about 1-1 73d part ; or the specific 



fravity of hydrogen gas, when taken in this way, comes out 

 -] 73d part below the truth. 



The method, which I am in the habit of adopting to obviate 

 this source of inaccuracy, is very simple ; and though it does 

 not annihilate the error ; yet it reduces it to so small a quantity, 

 that it may be neglected without any bad consequences. The 

 method is this : 



The flask in which the gas is to be weighed is exhausted by 

 the air pump, and then filled with common air, which has been 

 standing for some time upon the same water trough with the 

 gas whose specific gravity is to be taken. Thus filled with 

 common air, it is weighed very accurately by means of a balance- 

 made for me by Mr. Crichton, of Glasgow, which when loaded 

 with a pound in each scale turns sensibly with the l-200th of a 

 grain troy.* The flask is then exhausted and weighed again. 

 Let the loss of weight be m. Finally, the flask is filled with the 

 gas, whose specific gravity is wanted, and weighed again. Let 

 the increase of weight be n. It is obvious that the specific 



gravity of the gas is = -. 



Let us suppose that the specific gravity of pure hydrogen gas 

 is taken in this way at the temperature of 32 . We have seen 

 that at that temperature the hydrogen gas weighs l-173d part 

 less than it would do if it were dry, and that the common air 

 weighs l-150th less than the true weight. These two errors do 

 not indeed balance each other exactly ; but they reduce the 

 error to 1-7 th of what it would be, if we were to deduce the spe- 

 cific gravity of the hydrogen gas by comparing it with the 

 weight of dry air; so that the deviation from the truth is reduced 

 to -y-TT P^'^t- -^^^ by this the hydrogen gas will weigh more 

 than'it ought instead of less ; for 0-9933 : 1 :: 0-0690 : 0-06946 

 = specific gravity of hydrogen gas thus deduced. Now this 

 exceeds the true specific gravity of hydrogen gas by somewhat 

 less than 2 in the fifth decimal place. 



Now if we suppose the flask capable of holding 100 cubic 

 inches of gas (and this exceeds the size of my flask), the hydro- 

 gen weighed would not exceed two grains. My balance is only 

 capable of going to the l-200th of a grain, or to the 4000th part 

 of the weight of the hydrogen. Hence it is obviously incapable 

 of determining the weight of the hydrogen gas to the fifth 

 decimal place with accuracy. On that account I never go fur- 

 ther than four decimal places ; so that an error in the fifth is of 

 no consequence. 



Let us see what the error would amount to if we take the 

 'specific gravity of hydrogen gas in this way at the temperature 



. * This balance does Mr. Crichton's skill a great deal of credit. It is the bett 

 balance for chemicdl purposes which I have ever seen. 



X 2 



