618 RICE 



AET. L 



on that account. We may,for simplicity of statement, consider 

 a system of two homogeneous masses with one dividing surface; 

 the statement can easily be extended to cover wider cases. Let 

 us suppose the system is varied to a state in which the condi- 

 tions in the phases and dividing surface are not conditions of 

 equilibrium as regards temperature and potentials, and the 

 dividing surface is changed in position ; also let it be found that 

 this is a state of smaller energy than the unvaried state, the 

 total entropy and total masses however being the same as 

 originally. Now imagine that the dividing surface is "frozen," 

 as it were, in the varied position. (This is equivalent to the 

 postulate of Gibbs as to constraining the surface by certain 

 fixed lines.) If left alone, the system in this "frozen varied" 

 state would tend to a new state of equilibrium; we are conceiv- 

 ing that its total energy is not altered from the varied value, 

 nor, of course, the individual volumes of each phase; the total 

 masses are not to vary either, but there may still be passage of 

 components through and into or out of the dividing surface (its 

 rigid condition is not to interfere with that). In this third 

 state (second varied state) the entropy will of course have 

 increased above that of the first varied state and so above that 

 of the original state of equilibrium. Now by the withdrawal of 

 heat (the rigidity of the system being still preserved) we can 

 arrive at a third varied state, which is also one of equilibrium, 

 in which the total entropy, etc., will be as originally, but the 

 energy less than that of the second varied state and therefore 

 less than that of the original state. Of course, on imagining the 

 surface now to be "thawed out," that is, the constraint on it 

 removed, we cannot be sure that the varied pressures established 

 in the phases and the varied tension in the surface will be con- 

 sistent with the curvature of the dividing surface, which must of 

 course remain in the same varied position all the time (for if it 

 moves from this the volumes and therefore the potentials will 

 change from the values arrived at in the last state and might 

 not be in equilibrium in the two phases in the final state). The 

 point, however, is that if there is a non-equilibrium state 

 infinitesimally near the original state which is one of less energy, 

 there is also a quasi-equilibrium state infinitesimally near which 



