on the Atomic Weights of Bodies. 181 



If air be a combination of oxygen and azote, and, if tins 

 principle of Gay Lussac be true, it is clear that it cannot 

 be composed of 79 volumes azote and 21 oxygen, because 

 79 is not a simple multiple of 21. And, if air be not a 

 chemical compound, it is difficult to conceive how its com- 

 position should never vary under any circumstances what- 

 ever. Air, in this respect, is precisely similar to water, 

 which is always a compound of one volume of oxygen and 

 two volumes of hydrogen. But water is admitted on all 

 hands to be a chemical compound. 



A very slight alteration in the estimate of the constituents 

 of air would bring it under Gay Lussac's law. Were we to 

 consider it as a compound of 80 volumes of azotic and 20 of 

 oxygen gas, it would consist of four volumes of azotic and 

 one volume of oxygen gases, or of two atoms of azote and 

 one atom of oxygen. It is difficult to avoid suspecting 

 that a coincidence so very near does not hold exactly. 



In the year 1803, when the experiments of Humboldt and 

 Gay Lussac were made, chemical experimenting had not 

 reached that degree of precision which it has now attained. 

 And, whoever is acquainted with Volta's eudiometer, by 

 means of which their results were obtained, and will take 

 the trouble to read Mr. Cavendish's observations on eudio- 

 meters in the paper already referred to, will at once admit 

 that a greater error than ^tk might easily be committed, 

 even by very careful experimenters. 



Aware of these sources of uncertainty, I made a new set of 

 experiments on the composition of air, in the year 1824, with 

 every precaution that I could think of to ensure accuracy. 

 The air was collected in the middle of a green field, at 

 some distance from all houses, and from marshes or ditches. 

 And before examination it was carefully washed in caustic 

 potash ley. The hydrogen which was used was prepared 

 from a mixture of purified zinc and pure sulphuric acid 

 diluted with distilled water. The retort in which the <>-as 

 was extricated was quite filled with this dilute acid, and 

 a portion of the hydrogen gas was allowed to escape before 

 I began to collect it for use. The eudiometrical experi- 

 ments were made while the hydrogen gas was coming over 

 and it was taken for every experiment directly from the beak 

 of the retort, ami, consequently, without standing any time 



