164 Br. Thomson on the [Sept. 



Now the theoretical specific gravities are : 



Oxygen I'llll 



Azotic 0-9722 



Both our results are higher than theory indicates ; oxygen by 

 , I f , , , or little more than ^-J_-th part, and azotic gas by 

 ■9T.V(7) or little more than .^JL^th part. Now, though every 

 possible pains was taken to render these experiments as accu- 

 rate as possible, I have not a doubt that both ,T-j;Virt^^ ^"^ TsVoth 

 are considerably within the limits of error. The gases were 

 weighed in a glass flask capable of holding about 50 cubic 

 inches ; so that the weight of the gas subjected to experiment 

 (supposing it the same as that of air) did not difter very much 

 from 15 grs. The flask was first weighed with great care. It 

 was then exhausted of air, and weighed again. It was then 

 screwed to a gas holder, and filled with the gas whose specific 

 gravity was required and weighed again. The specific gravity 

 was obtained by dividing the weight of the gas thus ascertained 

 by the weight of the same bulk of air previously taken out of the 

 flask by the air-pump, and its weight determined by weighing. 

 Now in each of these three processes a stop-cock was first 

 screwed to the flask, and afterwards unscrewed from it. This 

 stop-cock was furnished with a collar of leather which I was 

 under the necessity of greasing in order to make air-tight. I 

 was at pains after each unscrewing to wipe the brass part 

 attached to the flask quite clean ; but it is difficult to be sure 

 that no alteration whatever is made in the absolute weight of 

 the flask. An error of the thousandth part of a grain would be 

 sufficient to occasion all the difference which exists between the 

 numbers obtained by experiment and the theoretical numbers. 

 Now I am afraid that there is scarcely any care that will be suffi- 

 cient to prevent the chance of an error amounting to T-oWth of a 

 grain ; a quantity which certainly does not exceed the millionth 

 part of the whole weight of the flask and its contents. 



But this is not the only source of error to which we are 

 exposed. Unless the gases examined be absolutely pure, or 

 unless we have the means of determining the absolute quantity 

 of foreign gaseous matter mixed with the gas tried, it is obvious 

 that the specific gravity of the gas which we obtain cannot be 

 the true one. The azotic gas, the specific gravity of which was 

 ascertained in my laboratory, was obtained by abstracting the 

 oxygen gas from it by means of nitrous gas. Now had an excess 

 of nitrous gas been added amounting to no more than — air^h part 

 of the azotic gas, or, in other words, had the azotic gas sub- 

 jected to experiment been a mixture of 299 volumes of pure 

 azote and one volume of nitrous gas, this small proportion of 

 impurity would have been sufficient alone to account for the 

 increase of specific gravity observable in our trials. 



The specific gravity both of the oxygen and azotic gas, as 



