110 BRIDGMAN. 



value 127° was adopted for the transition point at atmospheric pressure. 

 This could be safely done without making a redetermination of it for 

 this special sample, since this value is certainly as accurate as the 

 other values at high pressures. There are a number of measurements 

 at atmospheric pressure. For the transition point we have 127.2° 

 by Reinders,* 127° by Steger/^ 130° by Guinchant,22 and Schwarz^ 

 gives the limits 124.5° to 130.2° by a thermal method, and 126.3° 

 to 129.3° by an optical method. There is one value of the change 

 of volume, 0.00135 cm^ per gm. by Reinders,^^ against 0.00342 found 

 above. It would seem that the larger would be more likely to be 

 correct. For the latent heat there is the value 3.0 cal per gm. by 

 Berthelot, as quoted by Varet,^^ and 1.53 cal. by Guinchant,^^ as 

 against 1.2 cal. calculated above. If a smaller value of the change 

 of volume were used, the calculated value of the latent heat would be 

 smaller. 



We have already seen that at a maximum point the relations be- 

 tween the compressibilities and thermal expansions of the two phases 

 may be deduced from the curves for Av and All, the specific heat dis- 

 appearing from the relations at the maximum point. These values 

 at the maximum are included in the follow^ing. It has also been stated 

 that direct measurements of the difference of the compressibility 

 showed that above the maximum the phase of smaller volume has the 

 larger compressibility. The numerical value of the difference could 

 not, however, be determined with much accuracy. Within the limits 

 of accuracy, Aa would seem to be constant on the falling branch at 

 O.OeSS cm.^ per gm. per kgm. per cm^. Seven determinations of this 

 quantity were made; the deviations of the several values from 0.0633 

 are; +35%, -15%, 0%, -25%, 0%, -60%, and 0%. The devia- 

 tions are large, but still this is as accurately as one can hope to get 

 quantities as small as these by a method like the above. This value 

 is also not inconsistent with that deduced from the AV curve at the 

 maximum point, where we have already seen that Ace is determined 

 by the AV curve alone. It was not possible to make any measure- 

 ments of Aa on the rising branch of the curve. If, however, we make 

 the assumption that Aa is constant on both branches of the curve 

 (an assumption which is very doubtful) we may compute A/? and ACp 

 on the two branches. These values are given in Table XII, directly 



21 W. Reinders, ZS. phys. Chem. 32, 494-536 (1900). 



22 J. Guinchant, C. R. 145, 68-70 (1907). 



23 R. Varet, Ann. Chim. Phvs. 8, 79-141 (1896). 



