EFFECTS OF PRESSURE ON NITRATES. G15 



be computed from the data of Bellati and Romanese appears to be 

 due to an error of Tammann's in converting Lussana's data, which 

 are given in atmospheres, to kilograms. Lussana himself states that 

 the agreement of his experimental values with the computed values 

 is almost perfect, and I have verified this, at least within one of 

 two percent. The apparent discrepancy between Tammann's and 

 Lussana's observations may furthermore be at least partly explained 

 by the difference in the pressure range, because of the fact that the 

 slope of the transition lines III-IV and III-II changes markedly with 

 pressure. The slope of the line II-III decreases algebraically, the 

 normal direction of variation, but the slope of the line III-IV increases 

 with rising pressure, a somewhat unusual effect. These two equili- 

 brium lines are so short that it would be a matter of considerable 

 experimental difficulty to accurately measure the variation of slope 

 with pressure. That there is a variation with pressure was first 

 suggested to me by an examination of my own equilibrium points. 

 Furthermore, on making the computations, it appeared that such a 

 variation was imperatively demanded by the conditions at the triple 

 point. With the help of the known relations at the triple point it 

 was possible to get a fairly good idea of what the actual variation of 

 slope must be. The values given in the Table were computed so as 

 to satisfy the conditions at the triple point, and with no other condi- 

 tions in mind. In fact, I did not know of the discrepancy between 

 the values of Lussana and Tammann until after I had made the 

 computations. All of the discrepancy between these observers 

 cannot be explained in this way, however. Lussana gives for the 

 slope of the transition III-IV over a range of 2.50 kgm. 0.0287, Tam- 

 mann giAcs over the range up to the triple point, that is up to 860 kgm., 

 the average slope 0.0346. I find for the average slope up to the triple 

 point 0.0364, and for the average over the first 2.50 kgm. 0.0318. The 

 variation of slope with pressure is fairly large and will explain a large 

 part of the discrepancy between Tammann and Lussana, although it 

 will evidently not account for all of it. The agreement of Tammann's 

 and my results is better than that of Tammann and Lussana. It 

 must be said that Tammann did detect the curvature of the transition 

 lines, and states that the II-III curve is concave downwards and the 

 III-IV concave upwards, but he makes no allowance for this fact in 

 comparing his results with those of Lussana. In spite of the sur- 

 prisingly good agreement found by Lussana between his results and 

 those computed from the thermal data of Bellati and Romanese, this 

 agreement must be recognized as accidental, and not necessarily 



