OF ARTS AND SCIENCES. 231 



and it would lead us to expect that the specific gravity of oxygen gas 

 referred to hydrogen gas would be slightly greater, certainly not I 

 than the corresponding half molecular weight. Now we have from 

 the results published in this and our preceding paper, — 



Half molecular weight of carbonic acid gas 21.82 



Sp. Gr. of carbonic acid gas (referred to hydrogen gas) . 21.96 



Half molecular or atomic weight of oxygen gas . . . 15.87 

 Sp. Gr. of oxygen gas (referred to hydrogen gas) . . . 15.88 



It has been thought to discredit the low value of the atomic weight 

 of oxygen we have found by the very easy but wholly gratuitous as- 

 sumption that the hydrogen gas on which we experimented was im- 

 pure, and we had intended in this investigation to demonstrate the 

 identity of the gas from the several sources employed by a comparison 

 of their specific gravities. We are under great obligations to Lord 

 Kayleigh, who has relieved us from this necessity by anticipating the 

 work. He has experimented on hydrogen gas, not only from all the 

 sources we used, but also on gas which had been occluded by palla- 

 dium, and has obtained the same result in all cases ; namely, the value 

 1-3.S84,* essentially identical with that which we have found in this 

 investigation with the hydrogen gas from our electrolytic apparatus.! 



This point, however, had not been overlooked in the previous paper. 

 It was there shown that the results obtained, in three distinct series of 

 experiments, with hydrogen gas prepared by three different processes 

 and with three different forms of apparatus, were essentially identical, t 

 On the doctrine of chances, such an agreement would have been prac- 



* 



tically impossible had there been an appreciable amount of accidental 

 impurities, in the gas from either of the sources. We say accidental 

 impurities, for there may be inherent impurities common to the 

 from all sources, of which we as yet know nothing; and, as we wrote 

 before, " The question still remains, Is the hydrogen gas thus prepared 

 the typical hydrogen element? But this is the same question which 

 must arise in regard to any one of the elementary substances ; and :ill 

 that we can say is, that the evidence in regard to the purity of the 

 hydrogen we have used is as good as any that can be adduced in regard 

 to any one of the elementary substances whose atomic weight has been 

 most accurately determined." § 



* Proceedings of the Royal Society, vol. xlv. p. 426. 

 t These Proceedings, vol. xxiii. p. 108. 

 t See table, these Proceedings, vol. xxiii. p. 173. 

 § These Proceedings, vol. xxiii. p. 171 



