1892.] Relative Densities of Hydrogen and Oxygen. 451 



brings into play a fresh quantity of zinc, with its accompanying con- 

 tamination. Moreover, the supply of acid that can be included in 

 one charge of the generator is inadequate, and good results are only 

 obtained as the charge is becoming exhausted. These difficulties are 

 avoided when zinc is discarded. The only material consumed during 

 the experiments is then the water, of which a large quantity can be 

 included from the first. On the other hand, the hydrogen liberated 

 is necessarily contaminated with oxygen, and this must be removed 

 by copper contained in a red-hot tube. In the experiments to be 

 described the generator was charged with potash,* and the gases 

 were liberated at platinum electrodes. In the case of a hydrogen 

 filling the oxygen blew off on one side from a mercury seal, and on 

 the other the hydrogen was conveyed through hot tubes containing 

 copper. The bulk of the aqueous vapour was deposited in a small 

 flask containing strong solution of potash, and the gas then passed 

 over solid potash to a long tube packed with phosphoric anhydride. 

 Of this only a very short length showed signs of being affected at the 

 close of all operations. 



With respect to impurities, other than oxygen and oxides of 

 hydrogen, which may contaminate the gas, we have the following 

 alternative. Either the impurity is evolved much more rapidly than 

 in proportion to the consumption of water in the generator, or it is not. 

 If the rate of evolution of the impurity, reckoned as a fraction of the 

 quantity originally present, is not much more rapid than the corre- 

 spondingly reckoned consumption of water, the presence of the im- 

 purity will be of little importance. If on the other hand, as is 

 probable, the rate of evolution is much more rapid than the con- 

 sumption of water, the impurity is soon eliminated from the residue, 

 and the gas subsequently generated becomes practically pure. A 

 similar argument holds good if the source of the impurity be in the 

 copper, or even in the phosphoric anhydride ; and it applies with in- 

 creased force when at the close of one set of operations the generator 

 is replenished by the mere addition of water. It is, however, here 

 assumed that the apparatus itself is perfectly tight. 



Except for the reversal of the electric current, the action of the 

 apparatus is almost the same whether oxygen or hydrogen is to be 

 collected. In the latter case the copper in the hot tubes is in the re- 

 duced, and in the former case in the oxidised, state. For the sake of 

 distinctness we will suppose that the globe is to be filled with 

 hydrogen. 



The generator itself (fig. 1) is of the \J-form, with unusually long 

 branches, and it is supplied from Grove cells with about 3 amperes 

 of electric current. Since on one side the oxygen blows off into the 



* At the suggestion of Professor Morley, the solution was freed from carbonate, 

 or nearly so, by the use of baryta, of which it contained a slight excess. 



