SURFACES OF DISCONTINUITY 585 



vapor is 41 dynes per cm. for an increase of water vapor pressure 

 of 1.75 cms. of mercury. There he leaves the calculation, but 

 had he known, as we know now from Iredale's and Micheli's 

 work, that the fall in tension is proportional to the increase of 

 vapor pressure, he could have finished the calculation for the 

 amount of the layer adsorbed just at saturation before actual 

 condensation into a genuine liquid water phase begins. For 



da - 41 



dp 1.75 X 981 X 13.6 

 and 



dff _ 17 A X 10-« X 41 

 ^ ~ ~ '^ dp~ 1.75 X 981 X 13.6 



= 3 X 10~^ (grams per sq. cm.), 



since y, the density of water vapor at 20°C, is 17.4 X 10~^ grams 

 per c.c. This is just the figure for a unimolecular film of water 

 molecules, but there is no doubt that no such conception was 

 in his mind. Indeed, the assumption he makes above shows 

 this. Iredale in one of his papers has some very interesting 

 remarks to make on the general theory of adsorption and con- 

 densation, but reference to them will be deferred until we have 

 commented on the subsection of Gibbs' work which deals with the 

 formation of a new phase at the interface between two phases, 

 since Iredale's comments involve the theoretical considerations 

 in that subsection. 



Another very interesting set of experiments were carried out 

 by Oliphant (Phil. Mag., 6, 422, (1928)). His apparatus was 

 adapted from one invented for another purpose by Schofield. 

 He found that an expanding mercury surface selectively adsorbs 

 carbon dioxide from a mixture of that gas with an excess of 

 hydrogen or argon, and that at all concentrations above 2 per 

 cent the carbon dioxide thus selectively adsorbed was constant 

 at a value about 6.5 X 10^'^ molecules per sq. cm. This very 

 nearly corresponds to a close-packed unimolecular layer. 

 Actually, Schofield's method does not involve the yda/dp rule 

 or the measurement of da/dp. It should be mentioned that 



