116 BRIDGMAN. 



curve had to be obtained by lowering the temperature after every 

 reading far enough to reach the region where the reaction to III runs. 

 On increasing temperature again beyond the III-IV Hne, the reaction 

 from III to IV was certain to run, so that one could be sure in this way 

 of starting with IV. One consequence of the great lag of the reaction 

 from IV to VI is that the changes of volume determined from these 

 measurements are certain to be in error, unless the compressibility 

 of the two phases should by accident happen to be the same, because 

 these are not the differences of volume at the equilibrium point, but 

 at some other point. The measured difference is presumably too large. 

 The difficulty of measurement is further increased by the fact that the 

 difference of volume is excessively small. The discontinuity in the 

 piston displacement at the last point measured on this curve, at 137°, 

 was only 0.0018 of an inch. 



The transition from V to IV shows the same lag as that from VI to 

 IV, although the lag is not so obstinate. If temperature is raised 

 on the phase II to somewhere between 50° and 80°, the pressure being 

 at about 2000 kgm., and then pressure increased at this temperature, 

 the reaction from V to IV will not run, even if the pressure is increased 

 several thousand kilograms beyond the transition value. It is im- 

 possible in this way to obtain points on the IV-V line, but only the 

 II-V line will be found. This curious disappearance of the IV-V 

 line was the cause of much mystification before the explanation was 

 found. To be sure of getting points on the IV-V line, the phase IV 

 should be produced by first increasing pressure at low temperature, 

 so as to force the appearance of III, and then raising temperature 

 across the III-IV line. If the phase IV has once been formed, and 

 pressure is decreased so as to produce V, the transition may be made 

 to run in the reverse du-ection by increasing pressure again. That is, 

 the phase IV will be produced from V if IV has existed in the apparatus 

 only a short time previous. Under such circumstances there seem to 

 be some nuclei left about which the formation of IV can begin. A 

 precisely similar behavior has been found for some of the varieties of 

 ice. 



The points on the II-VI line could be obtained with fair definiteness, 

 and the values of i^v were also self consistent enough to warrant includ- 

 ing them in Figure 12. The concave side of the equilibrium line II-VI 

 is toward the temperature axis, which is the reverse of the case for 

 II-V. There are indications that the curve II-VI may split again at 

 its upper end, there being a modification VII ; an appearance at 200° 

 like that of two equilibrium points close together suggested this. The 



