and Gases near their Critical Temperatures. 153 



rise to some remarkable refraction figures at the surface of 

 the liquid. By substituting the test-tube of alcohol and water 

 for the tube of ether, corresponding figures are obtained — 

 heat in one case, and diffusion in the other, causing these 

 figures to pass through the same changes. It appears possible 

 that the black surface may be due to the mixing of the liquid 

 ether with its vapour when they are of nearly equal density. 

 This view receives some support from the fact that just before 

 the defined surface of the liquid is lost, the convection- 

 currents become so rapid and violent as to disturb and 

 apparently almost break through it into the vapour above. 



The above conclusion as to the form of the surface receives 

 confirmation from some experiments with an external tube en- 

 closing two glass plates 0*15 millim. thick, and 11 millims. 

 wide, and separated from each other by a short distance. Not 

 the least rounding of the surface of the liquid could be 

 detected either between or around the plates. That no de- 

 pression of the liquid was observed between the plates, corre- 

 sponding to that in the capillary tubes, may be explained by 

 the very small expansion of the liquid in this tube at that 

 point at which the depression usually takes place. Under 

 the same conditions the depression is absent also from a 

 capillary tube. The expansion, as already stated, is dependent 

 upon the amount of liquid in the tube and upon the rate at 

 which it is heated. On cooling, just as the cloud appeared, 

 and before the liquid line had made its appearance in it, an 

 interesting action was observed taking place on the surfaces 

 of these glass plates; for, extending downwards into the cloud 

 and considerably above its upper limit, a liquid film could 

 be seen running over their surfaces. This probably affords 

 evidence of a surface-action preceptibly influencing the position 

 at which the condensation of the liquid takes place. 



It remains only to briefly notice a class of facts to which 

 reference has not yet been made, but which includes certain 

 conditions capable of modifying some of the results described 

 in a part of this paper. It has been very frequently observed 

 that when a tube is heated for the first time, the depression is 

 smaller than it is when the tube is again heated within a short 

 time of the first experimnnt. In a few capillary tubes the 

 liquid is seen to disappear at the same level as the liquid out- 

 side them; but reheating shortly after the liquid has condensed, 

 the usual depression is observed. When such a tube has been 

 left a sufficient length of time in contact with the liquid on 

 heating, the same result is obtained as at first ; this may be 

 repeated indefinitely. For two tubes this time has been de- 

 termined, and in both cases found to be about 20 hours ; a 



