232 Mr. J. B. Hannay's Examination of Vacua. 



and filled up, and the residual air again imprisoned in H 

 by a new layer of frozen metal. After the first heating 

 the residual gas does not show itself, as the metal goe^ 

 right up into contact with that formerly frozen there. It 

 seems that if the quantity of gas is very small, it only 

 forms a layer or coating on the glass, and does not form 

 a bubble. This seems to show that the McLeod gauge only 

 measures the excess of residual gas over that required to 

 saturate the glass surface. When sufficient exhaustion was 

 supposed to have been attained, the bulbs E were sealed off; 

 and here it was found that the whole work had been futile, 

 because, on melting the tube through the metal just below 

 the barometric height, in order to detach the bulbs, the 

 appearances shown in figs. III. and IV. were seen. A large 

 amount of gas was given off whenever the glass became soft 

 enough for sealing. Fig. III. shows the sweating of the gas- 

 bubbles when just softening. Fig. IV. shows a tube melted 

 all round for some distance and the large amount of gas 

 driven off. It must be remembered that this tube had been 

 evacuated so far that it gave no residual gas on passing up 

 the metal. 



Some further attempts were made to get the bulbs eva- 

 cuated again, when it was intended to heat the tube at the 

 point of sealing till it was soft previous to sealing, allow- 

 ing it in fact to be sucked in till it was nearly closed, 

 pumping away the gases, so that none would come off on 

 melting for the final sealing. Experiments showed that when 

 once the glass was melted in a good vacuum no further gas 

 would come off on remelting. It was found, however, that 

 the stopcocks were all leak}', having warped by the long 

 heating (the operation takes a whole day) ; and after fighting 

 with this form of apparatus, it was abandoned for that shown 

 in fig. V. Here, instead of stopcocks, the tubes were bent 

 round like safety-funnels, and narrowed very much at the 

 bends : and the stopping was effected by a blast of cold air 

 kept blowing on the joint to be closed, and stopping the blast 

 whenever it was required to be opened — the heat of the bath 

 remelting the frozen plug. The transference of the metal 

 from bottom to top was effected as in the other apparatus. In 

 this the bulbs E were connected with the Sprengel pump by 

 HD, but were afterwards separated by passing metal up into 

 H, sweeping out residual air and freezing it in H, before the 

 final evacuation with the fusible metal was begun. This 

 form was adopted in order that the re.-idual air might be 

 subjected to no pressure, but have a free .-pace to pass out 

 into, thus avoiding condensation on the glass. Some progress 



