The phenomena of adsorption range from strong adsorption of a component from a dilute gas phase, or a dilute liquid solution, on to a solid surface, to mildly preferential adsorption of one component of a liquid mixture at the interface with the vapour. Theoretical adsorption isotherms (relationships between quantity of a substance adsorbed and its concentration in a bulk phase) were originally designed to apply to strong adsorption. They were empirical, e.g. the Freundlich isotherm, or they were based on simple statistical or kinetic theory, e.g. the Langmuir isotherm and isotherms which treated an adsorbed layer of molecules as a two-dimensional ideal or non-ideal gas. Modifications of such monolayer adsorption isotherms were also used to describe adsorption at the mercury-aqueous electrolyte interface, notably by Frurnkin and his co-workers 1'2 and by Parsons 3.
Obviously the ideal approach to a general theory of adsorption is a statistical calculation of molecular distribution in the region of the interface based on a detailed knowledge of the forces of interaction between all molecules present. Equally obviously the required information is not available. Even the calculation of molecular distribution at so simple an interface as that between liquid argon and its vapour 4,s can only be very approximate. Approaches that have been made to a statistical molecular treatment of multi-component interfaces have focused attention on a monolayer of molecules at the interface with simplified molecular interactions equivalent to "ideal" or "regular" mixing. Such calculations were carried out by Schuchowitzky 6, Belton and Evans 7, Guggenheim s and Everett 9. The monolayer model is discussed in some detail by Defay et al. (ref. 10, chap. 12). The extension to a multilayer model has also been considered 4'11 but in general the complexity and lack of information about molecular interactions preclude calculations for real systems. A quasi-thermodynamic approach to the problem was developed by Butler lz, Schuchowitzky ~3, Hoar and Melford ~4, Everett 15 and others, using the concept of chemical potentials in a surface zone which was generally assumed to be a monolayer.
The idea of a surface phase of finite thickness was specifically avoided in the rigorous thermodynamic treatment of surface phenomena by Gibbs a6. However, an alternative treatment using the concept of a surface phase was formulated by Verschaffelt ~ 7 and developed by Guggenheim 18. The latter showed that all the thermody-* We are indebted to a referee for bringing to our notice the papers by Eriksson.