THERMO-ELECTRIC QUALITY UNDER PRESSURE. 271 



■we had only rough indications of the initial direction of variation of 

 Peltier heat with pressure and no indication whatever as to even the 

 sign of the change of Thomson heat. The present measurements 

 afPord fairly accurate data about the Peltier heat over the entire range 

 of temperature and pressure, and somewhat less accurate information, 

 but still accurate enough to give the essential features of the situation, 

 with regard to the Thomson heat. 



In one other particular these measurements have an advantage 

 over previous ones in this field. Recent improvements in moving 

 coil galvanometers, as well as the greater size of the effect, have made 

 it possible for me to use this type of galvanometer, instead of some 

 form of Thomson galvanometer. Every one knows how enormously 

 more convenient the moving coil galvanometer is than the Thomson. 

 It was possible to make these measurements like any others of physical 

 routine, at all times of the day, with no disturbance from outside 

 changes, instead of waiting for the exceptionally favorable conditions 

 of the early morning hours, as previous observers have been forced 

 to do. 



In addition to the effect of hydrostatic pressure on thermo-electric 

 quality, this paper contains measurements on the effect of tension. 

 In my previous work on resistance effects, it appeared that there is a 

 simple relation between the effects of hydrostatic pressure and tension 

 on resistance, which had theoretical significance. I hoped for a similar 

 state of affairs with respect to thermo-electric properties, but on 

 looking up previous data found such discrepancies between different 

 observers on the effect of tension, as to cast doubt on even the sign of 

 the effect in some cases. I therefore made a fresh experimental 

 examination with some of the metals of the pressure measurements. 

 It appears that the discrepancies between different observers is due to 

 the inherent nature of the effect. Results are not at all reproducible, 

 sometimes different lengths from the same spool give effects of differ- 

 ent sign. The situation with respect to tension is therefore quite 

 different from that with regard to hydrostatic pressure. The effects 

 produced by hydrostatic pressure on thermo-electric quality have a 

 definite physical significance; reproducible results are obtainable, 

 and different observers may agree, at least for normal metals, as is 

 shown by the quite unexpectedly good agreement between the results 

 of Wagner and myself over the range common to our work. But the 

 tension effects show no such regularity, and apparently depend more 

 on accidental incidents, such as difference in mechanical handling. In 

 the following, little space is devoted to this aspect of the question, 



