642 BRIDGMAN. 



with pressure at constant entropy, and the change in the number of 

 free electrons, may be neglected. Griineisen compares his formula 

 with experiment with surprisingh' good results; in many cases the 

 agreement is within the limits of error, and the worst discrepancy is 

 for lead, where the difference is 50%. More recentl}^ Beckman^ has 

 extended the range of experimental material, and has applied Grii- 

 neisen's formula to all the available data. He concludes that the 

 formula cannot be considered exact, but must be regarded only as an 

 approximation. Griineisen himself certainly did not claim more. 



The new material of this paper allow a comparison with Griineisen's 

 formula over a somewhat wider range. Furthermore, since the 

 numerical values found in this paper often differ considerably from 

 those of Beckman, I have thought it worth while to recompute all the 

 data. The results are shown in Table XXV. In the recomputation I 

 have used my own values for the pressure coefficient and also for the 

 temperature coefficient, except in several cases where higher values 

 have been reported by other observers. In the recomputation I have 

 paid attention to several minor points. For the temperature coeffi- 

 cient of resistance I have used the coefficient of specific resistance 

 (which is strictly correct) instead of the coefficient of observed resist- 

 ance. The difference is only a fraction of a per cent. Furthermore, 

 I have used the instantaneous coefficient at 0° where the data are 

 available, instead of the average coefficient between 0° and 100°. 

 This again is the strictly correct procedure; for most substances it 

 makes little difference, but for iron the difference is 8.5%. It must 

 be recognized, however, that there are much greater uncertainties 

 in the fundamental data entering the equations than can be introduced 

 by the nicer points just mentioned. I have taken the fundamental 

 data, except pressure and temperature coefficient, from the last 

 edition of Landolt and Bornstein, selecting the values that seemed 

 most consistent. I have given the preference to the values of specific 

 heat of Jaeger and Diesselhorst at 18°. For zinc I have used the 

 value of the thermal expansion of the National Physical Labaratory, 

 O.O478. There is much discrepancy for this substance between 

 different observers; Griineisen's own experimental value, O.O45I, seems 

 certainly too low. The expansion of cadmium also is variously 

 given; I adopted O.O49O as a mean between Dorsey, O.O488I, and 

 Fizeau, O.O493. The expansion of molybdenum and tungsten were 

 taken from the Tables of the French Physical Society. It is to be 

 noticed that the Cj, in Table XXV is in kg. cm. per cm^ of material. 

 The compressibility of indium has been assumed to be 2.0X10"^ in 



