THERMO-ELECTRIC QUALITY UNDER PRESSURE. 385 



in this paper has been that the ordinary thermodynamic argument, 

 by which the formulas have been deduced, allows conclusions only as 

 to the quantities of heat, and that no conclusion is justified as to the 

 local e.m.f s. corresponding to these heats. Complete diagrams and 

 tables are given for the e.m.f., Peltier, and Thomson heats of these 

 couples. 



The nature of the results was unexpectedly complicated. The 

 normal state of affairs is apparently a positive effect of pressure on 

 both Peltier and Thomson heats, but there are numerous examples 

 of negative effects, and almost none of the metals show regular varia- 

 tion of these quantities with pressure and temperature within the 

 range. Three metals, tin, iron, and aluminum, show complicated 

 variations of the e.m.f. of the couple, there being maxima and minima 

 with both pressure and temperature within the range. 



The effect of tension on thermo-electric quality of a few of the 20 

 metals was measured, and the results again were complicated. In 

 general the pure volume effect on thermo-electric quality is different, 

 according as the change of volume is brought about by a hydrostatic 

 pressure or a tension. This is the reverse of the behavior of electrical 

 resistance, which is determined primarily by the change of volume, 

 regardless of whether it is produced by tension or hydrostatic pressure. 



As a bye-product, the dependence of Thomson heat on temperature 

 gradient has been measured for mercury, and the possibility of such 

 an effect of gradient has been made more remote than before. 



The unexpected complications found makes these results disap- 

 pointingly meagre in their suggestions as to the nature of the thermo- 

 electric mechanism. The conclusions are mostly negative in char- 

 acter. The most unmistakable inferences may be drawn as to the 

 untenability of the old gas-free-electron theory of metals, but this is 

 not now new enough to be worth the experimental trouble. This 

 conclusion was drawn 10 years ago by Wagner from his data up to 

 300 kg., and the results of this paper can add nothing to the conclu- 

 sions of Wagner in this regard, since our results are in essential agree- 

 ment over ovu- common range. Further than this, the results suggest 

 most strongly that the thermo-electric mechanism must be compara- 

 tively complicated, that it cannot be at all of the simplicity imagined 

 by the free electron theory and that most likely the effects which we 

 measure are the resultant of different effects, which some times, at 

 least, work in opposite directions. What these effects may be, we 

 are not in a position at present to speculate. 



In the projected paper mentioned in the introduction on general 



