ANB RATIO OF THEIR ArOMIC WEIGHTS. 89 



not accomplish anything till the nieivui y in which the eiidioineter stood was covered 

 with a laj'er of sul[)hurie acid, after which no more nitrogen was found. It was 

 therefore fair to assume that all the nitrosren found hiul been due to the entrance 

 of air at the last operation ; the volume of oxygen acconipanying it was therefore 

 computed and double this volume of hydi-ogen was assumed to have disai)[)eared by 

 c<)iid)inatioii with this oxygen derived from the air. 



12. DENSITY OF ELECTROLYTIC (JAS. LEAKAGE OF AIU BETWEEN MEKCUUY AND 



GLASS. 



It may be said that the contact of mercury with glass is proof against leakage 

 only when it is nearly as perfect as in a barometi'ie tube. A joint which leaks a 

 little may not be perceptibly improved by imraei'sion in mercury. I have seen a 

 Joint between a rubbei' and a glass tube, which leaked into a vacuum at the rate of 

 a bubble a minute, keep on at apparently the same rate Avhen immei'sed ten 

 centimetres in mercury, because the contact between the glass and the mercury Avas 

 not sufficiently })erfect. Since a clean surface is hard to maintain on the outside of a 

 tube, the only way in which mercuiy can be safely assumed I'eally to stop a leak is, 

 by using it to carry a layer of sulphuric acid, as was indicated by Crookes in 

 his experiments with high vacua. 



l;5. DENSITY OF ELECTROLYTIC GAS. OXIDATION OF MERCURY. LARGE EUDIOMETER. 



In some of the eudiometric analyses, some oxidation of mercury was noticed, as 

 might be expected. The amount of oxygen thus consumed was obviously neg- 

 ligible, but it was thought well to avoid possible objection by making it less. A 

 measured volume of hydrogen was therefore introduced into the eudiometer before 

 the admission of the electrolytic mixture. The results thus obtained were the 

 same as before within the errors of observation. 



I then constructed a large eudiometer iu which all but a hundredth or a 

 thousandth of the gas taken for analysis could be exploded out of contact with 

 mercury. It has a capacity of 3.2 litres. A stopper ground into its place and ex- 

 tending down through ninety centimetres of mercury is so arranged that it can be 

 raised or lowered at will, without danger of leakage. When an explosion was to be 

 made, the passage was stopped by this contrivance, so that the explosion did 

 not reach to the mercury. This eudiometer will safely bear the explosion of eight 

 hundred cubic centimetres of the mixed gases as measured at atmospheric 

 pi-essure. 



The admission of gas to the eudiometer was controlled by the stopper h, Fig. 

 30. When the desired pressure was reached, h was closed and then the similar 



