582 



BRIDGMAN. 



The data are presented in the order of compHcation of the phase 

 diagrams, and inehide the coordinates of the transition curves, the 

 change of volume, latent heat and internal energy, and in those cases 

 where the measurements were sufficiently accurate, an estimate of 

 the difference of compressibility, thermal expansion, and specific 

 heat of the different modifications. In the discussion of the relations 

 of the various forms it will appear that the three simplest nitrates, 

 those of Rubidium, Caesium, and Thallium, are much alike in their 

 behavior under pressure; the nitrates of Ammonium and Potassium 

 are much less simple in their relations, and Silver Nitrate is appar- 

 ently not connected with these at all. 



Detailed Data. 



Silver Nitrate. — Two batches of this substance were used. 

 The first was from Eimer and Amend, c. p., and was used both for 

 the run at atmospheric pressure and for the runs over the entire high 

 pressure range. For the determinations at high pressures it was 

 hammered dry into an open steel shell. The high pressure points 

 were found in the order of decreasing pressure and increasing tempera- 

 ture. For the measurement at low pressure, the AgNOs was fused 

 into a glass tube, in which it remained. The second batch was from 

 the J. T. Baker Chemical Co., "analyzed chemicals," and showed 

 by analysis less than 0.002% of impurity. This was fused into glass 

 moulds, from which it was removed for the measurements. The 

 moulds were of such a size that the sticks of AgNOs fitted loosely 

 the inside of the pressure cylinder. In all cases, pressure was trans- 

 mitted directly to the AgNOs by kerosene. With this second batch, 

 five of the high pressure points were repeated, between 6000 and 12000 

 kgm., six months after the original determinations. Every part of 

 the apparatus was different from the original, which had been de- 

 stroyed piece by piece in the meantime by various explosions. The 

 agreement of the p-t values of these two determinations is good. 

 This repetition was made necessary because of an accidental distribu- 

 tion of error in the two lowest Av points of the original measurements, 

 such that there appeared to be a discontinuity in the A^ curve, so that 

 there was consequently a possibility that another modification might 

 exist. 



No perceptible decomposition of the AgNOs could be detected 

 where it came in contact with the steel, although such might be 



