SURFACES OF DISCONTINUITY 631 



turbance causing a slight growth in its size would cause the 

 first phase to encroach on the second; but, of course, finite energy- 

 would be required for the initial creation of the sphere before the 

 infinitesimal disturbance in the right direction is applied. 

 But if conditions were such that "zero globule" corresponded 

 exactly to "zero r," no finite energy would be required to create 

 the globule ; any infinitesimal impulse in the right direction pro- 

 ducing any globule however small would produce one larger 

 than the "critical globule," which in this case is "zero globule," 

 and at once the encroachment of the first phase on the second 

 phase would begin. This argument does not apply if the globule 

 does not vanish when r reaches zero, and the second phase is not 

 unstable in the strict sense. Gibbs clearly regards the second 

 case as the most general in nature. Doubtless he had in mind 

 the example of the formation of water drops in saturated vapor. 

 This instance is a good illustration of the application of the 

 abstract reasoning of these pages. When a drop of water is in 

 equilibrium with its vapor in a large enclosure, the vapor, over 

 its convex surface, is supersaturated as compared with vapor 

 over a plane surface; there is a tendency, on the slightest dis- 

 turbance in the right direction, for the drop to grow in size (as 

 we have frequently pointed out); as it does so its surface 

 flattens and the equilibrium vapor around it decreases in pres- 

 sure and density, as it naturally would do if it were being in part 

 condensed. Nevertheless, it is a commonplace physical fact 

 that it is next to impossible to start condensation in a mass of 

 saturated vapor quite free from dust particles or ions. 



48. The Possibility of the Formation of a Homogeneous Mass 

 between Two Homogeneous Masses 



We now pass on to the possibility of the formation of a fluid 

 mass between two other fluid masses. The latter are denoted 

 by the letters A and B. In the discussion on pages 258-261 

 they are supposed to be capable of being in equilibrium with 

 one another when meeting at a plane surface, so that the func- 

 tions p^it, n) and psit, ij) are to be equal to each other for all 

 values of t, ni, /X2, • • • On page 262 the problem is generalized, 

 but in the meantime this condition is to be kept well in mind. 



