82 BRIDGMAN. 



ing in this connection, as it is in many others; the width of the region, 

 shown in Figure 13, passes through a minimum. On the equihbrium 

 curve, temperature rises as pressure falls, as for NEUSCN, so that 

 one would expect a steady increase of width with rising pressure. The 

 minimum seems to have some connection with the locality of the sharp 

 bend of the equilibrium line. 



The fact that there is an indifferent region introduces a possible 

 error into determinations of the equilibrium values of pressure and 

 temperature. Of course the fact that there is a region within which 

 the reaction does not run does not prevent our attaching a definite 

 meaning to the equilibrium coordinates. These are to be defined 

 thermodynamically as the points at which the thermodynamic poten- 

 tials of the two phases are equal. To apply this definition demands 

 that at least at one point the reaction run without sticking. The error 

 in the equilibrium coordinates due to the region of indifference is also 

 of course operative at atmospheric pressure. In most cases it is only 

 possible to shut the equilibrium temperature between an upper and a 

 lower limit. The limits vary greatly for different substances; there 

 are many solid transitions which are apparently as sharp as a melting 

 point. 



The actual point of equilibrium, defined thermodynamically as 

 above, may be situated, as far as we can tell, at any point within the 

 region of indifference. In all the preceding work, the equilibrium 

 point has been assumed to be in the middle of the indifferent band. 

 In most cases this can lead to only very small error, because the width 

 of the band is small compared with the total pressure. But in a case 

 like that of AgNOs at 0° the error from this cause may become appre- 

 ciable. An attempt to correct the equilibrium point by displacing it 

 from the center in proportion to the transition velocities from above 

 and below might lead to better results, but we could not be sure of 

 them, because we have seen that there is no necessary connection 

 between the width of the band and transition velocity. It is conceiv- 

 able that the true equilibrium point might lie on that side of the center 

 of the band toward the smaller velocity'. 



Before proceeding to the final part of the discussion, which will be 

 occupied with an attempt to find the implications as to mechanism 

 of these new facts, it will pay to very briefly recapitulate the nature 

 of the facts. There are two cardinal facts; in the first place there is a 

 region of indifference surrounding the equilibrium point within which 

 the transition does not run even when the phases are in contact, and 

 secondly the transition velocity at points equally distant from the 



