2o8 SCIENCE PROGRESS. 



explicable on the impurity-difTiision theory developed above. 

 The critical phenomena become very simple, and in ac- 

 cordance with theory, even for gases in which the presence 

 of some impurity is revealed by a not quite constant 

 vapour-pressure. The temperature at which the liquid- 

 surface disappears is found independent of the volume and 

 is therefore easily ascertained ; the best method is to apply 

 a slowly rising and falling temperature alternately. If we 

 assume the critical isothermal between certain limits to 

 have a similar shape on both sides of the critical point, the 

 true critical density will be found by calculating the average 

 density when the liquid surface disappears half way up 

 the column. Very concordant values for the critical 

 density are found by this method, and I believe that 

 the optical method (if thus applied) does not deserve all 

 the severe things that have been said about it. There is 

 only one point that leaves some uncertainty in the method ; 

 one or two tenths of a decree above the critical tem- 

 perature a very characteristic blue mist fills the tube, 

 probably due to liquid particles floating in the substance. 

 But in his best experiments the writer of this article saw 

 this mist disappear at less than '2 above the critical tem- 

 perature as measured by the disappearing of the liquid 

 surface, and the uncertainty is therefore very small indeed. 

 If no sudden changes of volume or temperature take place 

 all the other critical phenomena, like the striai described by 

 Andrews, hardly show themselves, if they do so at all. 



There are other methods for determining the critical 

 constants. Chappuis, for instance, measured the refractive 

 index of the two phases by an interference method, and 

 calls critical temperature the temperature where the two 

 coincide. But if impurities are present, and retardation in 

 the process of mixing, the refractive indices will not be 

 equalised at the critical temperature, and the method is bound 

 to give a result which is too high. For carbonic acid he 

 found 3 1 "6° C. (Andrews 30*90^ Amagat 31 '35°). Amagat 

 determined the critical temperature of CO2 by plotting the 

 densities as far as possible and completing the curve. A 

 direct method would seem preferable, though in the hands 



