264 MR. W. R. BOUSFIELD AND DR. T. M. LOWRY ON THE ELECTRICAL 



determining the total loss in weight of the apparatus during the action. In this we 

 were unsuccessful ; the quantity of hydrogen to be weighed was about 6 grammes, 

 and under atmospheric pressure would have occupied some 70 litres ; working under 

 reduced pressure the volume would be increased to about a thousand litres, and our 

 drying apparatus proved to be insufficient to hold back the moisture contained in 

 this large volume of gas. As, however, we considered the possible presence of 

 hydrogen in the metal to be the only serious error to which our method of preparing 

 standard solutions was liable, we were led to devise a further method of testing for 

 this impurity. According to the measurements of TROOST and HAUTEFEUILLE 

 (' Comptes Rendus,' 1874, vol. 78, p. 807), sodium hydride begins to dissociate at a little 

 above 300C. ; at 330 C. its vapour pressure is 28 millims. and rises to 752 millims. 

 at 420 C. We therefore took a test-tube of hard glass, placed in it 3 grammes of 

 sodium and exhausted by means of a mercury pump. After any occluded gases had 

 been liberated by fusing the metal, the temperature was raised, by means of a bath of 

 fusible alloy, to 550 C, in order to dissociate the hydride, the temperature being 

 recorded by means of a high-pressure mercurial thermometer. In this experiment no 

 gas was liberated. In the second experiment, with 5 grammes of metal, a bubble of 

 gas was obtained which occupied about four-tenths of a cubic centimetre, but was 

 probably air rather than hydrogen. This experiment was conclusive, for the presence 

 of O'Ol per cent, of hydride-hydrogen would have produced 5 cub. centims. of gas, 

 and at the temperature used the whole of this must have been liberated into the 

 vacuum produced by the pump. 



We consider, therefore, that whatever impurity may have been present, none was 

 in sufficient quantity to influence our measurements, and that if the atomic weight of 

 sodium be taken as 23 > 05 ; t 0*005, the solutions used as standards were accurate 

 within approximately the same limits, that is about 1 part in 5000. 



The solutions required for the measurements of density and conductivity at 18 C. 

 were for the most part stored in platinum vessels until required, but, by taking 

 advantage of the relative inertness of the most concentrated solutions, we were 

 enabled to make use of glass bottles for storing the solutions required for determining 

 the temperature coefficients. In consequence, perhaps, of its extreme viscosity and the 

 relatively small amount of water present, a 50-per cent, solution of sodium hydroxide 

 has far less action on glass than a dilute solution. Its density did not alter by a 

 single unit in the fourth place when stored during ten weeks in a glass bottle. A 

 further advantage in the use of concentrated solutions is found in the fact that the 

 carbonate is almost insoluble, and, if produced by exposure to air, separates as a 

 precipitate at the bottom of the bottle, producing a small change in the concentration 

 of the solution, but hardly any depreciation of its purity. 



