794 On the Motion of Electrons in Solids. 



with what it would be if governed by the temperature. The 

 atoms now form a framework of obstacles through which 

 the electrons move. This framework is continually set into 

 vibration by collisions with the electrons, but its motion is 

 dissipated into radiation so rapidly that the atoms never 

 acquire a motion comparable with temperature-motion. 



This view compels us to suppose that each atom has always 

 two electrons associated with it *. The whole of these electrons 

 need not be free at all temperatures. In some substances 

 which are poor conductors, only a few electrons may be free, 

 and in raising the temperature a large part of the energy 

 may be used in setting free new electrons. For such sub- 

 stances the atomic heat need not be near 5*88, but it should 

 approximate to this value as we reach temperatures at which 

 all the electrons have become free. Weber's researches f 

 show that such an asymptotic limit exists. Here, for instance, 

 is the variation of the atomic heat of carbon (diamond): — 

 Temperature... 0° 50° 100° 150° 200° 606° 806° 985° 

 Atomic heat .., 112 172 2'28 2 81 3-33 5-26 5"36 549 



The energy of motion of the atoms, although small, need 

 not be altogether negligible, so that the limit reached when 

 all the electrons have been set free, may be somewhat greater 

 than the exact value 5*88. As the melting-point is approached, 

 the energy of the atoms will increase greatly, so that this 

 value may be much exceeded. Thus PionchonJ finds for 

 the atomic heat of iron at 500° the value 9 # 84. 



24. This view of the matter is still not completely satis- 

 factory, for it leaves it difficult to understand the process of 

 conduction of heat between a solid and a gas in contact with 

 it. It seems as if we must suppose the atoms to form into 

 large clusters, which themselves take up the temperature- 

 motion from the electrons and might convey it to the mole- 

 cules of a gas in contact. The motion of these clusters as 

 they are bombarded by the electrons, w r ould be similar to 

 that of the solid particles in the Brownian movements. The 

 energy of motion of the clusters, namely, 3RT or 6RT for 

 each, would only contribute imperceptibly to the value of the 

 specific heat. 



If this last conjecture is correct there would be another 

 way of definiug the temperature of a solid in addition to the 

 two already given ; we might define it by the mean kinetic 

 energy of its molecular clusters. 

 January 7, 1909. 



* Except perhaps in elements whose atomic weight has been calculated 

 solely from the specific heat. 



t Pogg. Ann. cliv. p. 575. J C. JR. cvi. p. 1344. 



