462 Relative Densities of Hydrogen and Oxygen. [Feb. 18, 



third series of hydrogen weighings is very good, but that the mean 

 from the second series is decidedly lighter. This may have been in 

 part fortuitous, but it is scarcely probable that it was so altogether. 

 Under the circumstances we can hardly reckon the accuracy of the 

 final results as closer than -3-^0 o"- 



A word should perhaps be said upon a possible source of systematic 

 error, viz., mercury vapour. There is no doubt that hydrogen 

 passed over mercury takes up enough to cause a slow and superficial, 

 but quite distinct, discoloration of sulphur over which it subsequently 

 flows. In the experiments here recorded, the gas did not, indeed, 

 flow over mercury in mass, but, inasmuch as mercury was used to 

 secure the tightness of some of the joints, it is difficult to feel sure of 

 its absence. Again, in evacuations conducted with a mercury pump 

 can the vacuum be regarded as free from mercury vapour, which, it 

 must be remembered, would not show itself upon the gauge of the 

 Toppler? If both the hydrogen and the " vacuum " were saturated 

 with mercury vapour, the result of the weighings would, according to 

 Dal ton's law, be free from its influence. The same may be said of 

 any volatile impurity arising from the grease* upon the stopcocks. 

 As the matter stands, the results must, I think, be regarded as affected 

 with a possible error amounting to a fraction of the weight of mercury 

 vapour at the temperatures employed. But this is probably a very 

 small quantity. 



According to Hertz,f the vapour-pressure of mercury at 15 C. 

 would be about O'OOl mm. If this be correct, the weight of mercury 

 vapour in an atmosphere of hydrogen would be as a fraction of the 

 latter 



0-001 _ 1 



~76CT X20() "3800" 



It appears that in an investigation of hydrogen aiming at an accu- 

 racy of 1/10,000 the question of mercury vapour requires very careful 

 consideration. 



The accompanying table of results found by various experimenters 

 may be useful for comparison : 



* Composed of vaseline anl beeswax, 

 f ' Wied. Ann.,' vol. 17, p. 199. 



