2 Scientific Proceedings, Royal Dublin Society. 
most important, and must occur if air has not been completely 
expelled from the liquid. Let us consider first what will occur if 
we start with saturated vapour and slowly condense it. At first the 
permanent gas will be evenly distributed through the large volume, 
and its partial pressure will be very small. As condensation 
proceeds, very little of the permanent gas will be dissolved by the 
liquid, the greater part remaining diffused through the residual 
vapour. As the volume diminishes, the pressure of the permanent 
gas rises, slowly at first, then more and more rapidly, until, when 
condensation is nearly complete, it may become very considerable, 
and the final bubble, as pointed out by Andrews, may require to 
be greatly compressed before the gas in it is dissolved by the 
liquid. 
Suppose, now, that the vapour has been completely condensed, 
and the liquid left at rest long enough for the dissolved gas to 
diffuse evenly through it; and that, on increasing the volume 
again, a bubble of vapour flowin at the top of the tube, no ebulli- 
tion taking place from below. The vapour will then contain very 
little more permanent gas than that which was originally contained 
in the liquid which has evaporated, because the diffusion of dissolved 
gas through a liquid is a slow process. The partial pressure of 
the gas will, therefore, be little higher than it would be if the 
whole of the liquid were evaporated and the gas evenly distributed 
through the vapour. If ebullition from below can be prevented 
when the volume is again increased, the residual liquid will retain 
most of its dissolved gas, and the very low partial pressure of the 
gas in the vapour will remain nearly constant. If, on the other 
hand, ebullition from below takes place, the dissolved gas will 
escape into the ascending bubbles, and will quickly be carried up 
into the vapour. In practice, it is found that ebullition does not 
readily take place at the lower temperatures, whereas it is quite 
impossible to prevent it at the highest temperatures. 
Tf, then, a liquid is nearly, but not perfectly, free from dissolved 
air and if the readings of pressure are taken during evaporation, 
it may generally be expected that the errors will be very small or 
negligible at the lower temperatures, but noticeable at the higher 
ones. It may be remarked also that if the film of gas which 
adheres so obstinately to glass has not been removed during the 
filling of the tube, the film may remain on the glass during the 
