594 BRIDGMAN. 



ful attempt was made before even the existence of the line II-III was 

 estabUshed. Because of the unusual difficulties, a special procedure 

 was necessary in getting the points of the II-III curve. After the 

 reaction had started to run, and had been perhaps half completed, 

 pressure was artificially changed in the direction of the suspected 

 equilibrium, and after this change the direction of secondary pressure 

 change noted. After each change of pressure it is necessary to wait 

 only long enough to be sure of the direction of spontaneous pressure 

 change before making another trial. It is thus possible to obtain 

 results very much more quickly than by waiting for pressure to be 

 brought back to the equilibrium value by the progress of the reaction, 

 and in those cases in which Av is small, it is possible to shut the equili- 

 brivmi pressure within much narrower limits. With the galvanom- 

 eter and the bridge used the sensitiveness was great enough so that 

 a change of j kgm. could be certainly detected. This made it possible 

 to shut the equilibrium pressure within values differing by about 

 50 kgm. on the I-II curve, and about 100 kgm. on the II-III curve. 

 One must be careful, when working with such small pressure changes, 

 not to be deceived by the effects of heat of compression, which under 

 proper conditions simulate the appearance of a reaction. 



The values of equilibrium pressure and temperature, shown in 

 Figure 7 are fairly consistent. The values of AV, on the other hand, 

 which are shown in Figure 8, were never satisfactory, although I used 

 every wile at my command to get good values. The difficulty, which 

 would be great enough in any event because of the small values of A^, 

 is increased by the great sluggishness of the reaction between the two 

 phases in each other's presence, as well as by the large superpressure 

 required to start the reaction. If one tries to complete the reaction 

 by running the pressure far beyond the transition point, error from 

 hysteresis is introduced. Part of the difficulty is doubtless connected 

 with the chemical instability of the salt itself. After the series of runs, 

 some of the salt in the lower part of the cylinder was found changed to 

 a brick red color. Thallium, of course, forms two series of salts, 

 univalent and trivalent, and the trivalent form of some salts is the 

 more stable. If there were a change from a thallous to a thallic salt, 

 the reacting volume of the thallous salt would be less, and the change 

 of volume less. If this change did take place, it must have been 

 toward the end of the experiment, and would suggest a reason for the 

 low points found for Av on the II-III curve at 11200 and 11700 kgm., 

 and on the I-II curve at 7000 kgm. These points were measured 

 with particular care, and there seems no doubt that they give accu- 



