64 H. & Car hart — One- Volt Standard Cell. 



put in series with the standard cell, and the resistance in the 

 shunt of the main circuit required to maintain a balance shows 

 both the value and the direction of the E. M. F. of the experi- 

 mental cell as compared with the standard.* 



By this method the mean thermo-E. M. F. of Cu— CuS0 4 — 

 Cu per degree between 0° C. and 50° C. I have found to be 

 + •00073 volt. The positive sign indicates that the metal in 

 the warm solution is positive externally to the one in the cold, 

 or the metal in the cold plays the part of the zinc of a voltaic 

 couple. Boutyf gives the E. M. F. of Fe — FeS0 4 — Fe as 

 zero. It follows that a couple consisting of Fe — FeS0 4 — 

 CuS0 4 — Cu should have a positive temperature coefficient, and 

 this coefficient should be the same whether the whole cell is 

 heated or only the copper — copper sulphate side. Experiment 

 confirms the first conclusion at least. The E. M. F. of the cell 

 at 15° C. was found to be 0'625 volt, and at 29°4 C, 0*637 

 volt. The change is therefore +0-000833 volt per degree and 

 the coefficient is 0'000833 -r- 0-625 = 0-0013, or over T \- per 

 cent. 



Theoretically it should be 



0-00073^0-625 = 0-0012. 



The above combination is now used as an iron gravity bat- 

 tery, and the change in its E. M. F. with a sudden change of 

 the temperature of the room had attracted my attention before 

 I had investigated the temperature coefficient. After the 

 above iron-copper cell had been left standing for twenty-four 

 hours, with a piece of iron wire in the connecting tube of the 

 experimental cell to intercept the CuS0 4 which might diffuse 

 over, the E. M. F. was found to be 0*651 volt at 14°-4 C. ; and 

 for a range of 7*7 degrees the temperature coefficient was 

 0'0012. This is exactly the value obtained from theoretical 

 considerations. 



Another case in point is that of the voltaic series 

 Ni-NiSO -CuSO -Cu. 



4 4 



The E. M. F. was quite variable and only an approximate 

 measurement could be made. The temperature coefficient was 

 positive and equal to 0084, or over y 8 -^ per cent per degree C. 

 The actual change in E. M. F. was greater than this, for the 

 E. M. F. continued to fall on standing and the change by heat- 

 ing was thus in part masked. 



The thermal-E. M. F. of copper — copper sulphate is posi- 

 tive, while that of nickel — nickel sulphate is negative. Hence 

 the temperature coefficient of the above series should be posi- 

 tive and equal to the sum of the two thermal E. M. F.'s. The 

 result is positive but not equal to the sum of the two thermal 



* Carhart's Primary Batteries, p. 137. f Journal de Physique, vol. ix, p. 229. 



