H. S. Carhart—One- Volt Standard Cell 63 



The coefficient is therefore a little less than T i ff per cent per 

 degree. A neglected variation of temperature of 10° C. can 

 cause an error of only T l o per cent. 



In another paper- I have given an analysis of the tempera- 

 ture coefficient of both the Clark and the Daniell cell when the 

 former contains no excess of zinc sulphate crystals, and have 

 shown that this coefficient is a purely thermo-electric phenome- 

 non at the contact of a metal and a liquid. Further investiga- 

 tions confirm this conclusion that the temperature coefficient of 

 a voltaic cell is the result of the superposition of the two thermo- 

 electromotive forces of the metal and the salt in contact with 

 it on the two sides of the cell, whenever this coefficient is not 

 complicated with oxidation effects and the solution and recrys- 

 tallization of salts. In the old Clark cell almost exactly half 

 the temperature coefficient is due to changes in density of the 

 solution by reason of the presence of an excess of zinc sulphate 

 crystals. 



These dissolve when the temperature rises and again slowly 

 crystallize out on a falling temperature. Time is required for 

 diffusion, and the cell is therefore slow in reaching its electri- 

 cal equilibrium after a temperature change. This time lag may 

 extend over several days. It is very appreciable when the 

 temperature change does not exceed one degree an hour. 



Whether a resultant temperature coefficient shall be positive 

 or negative depends upon the relative values and the signs of 

 the two thermo-electromotive forces at the two sides of the 

 couple. To measure these electromotive forces of thermal 

 origin I have made use of an experimental cell consisting of 

 two glass tubes 12 cm long and l cm to 2 cm in diameter, joined 

 near the top by a small tube 17 cm long, having near the middle 

 a narrow bore made by thickening the walls of the tube. When 

 it is desirable to use two liquids a plug of asbestos is placed at 

 the narrow part of the connecting tube to aid in filling with- 

 out admixture of the liquids and to diminish diffusion. When 

 the thermo-E. M. F. between a metal and a liquid is to be 

 determined the experimental cell is filled with the liquid and 

 wires of the same metal, cut from one piece, are inserted in the 

 two limbs. One limb is then kept in melting ice and the other 

 is raised to different temperatures, and the resulting E. M. F. 

 measured. For this latter purpose the most exact method, 

 giving practically an electrostatic measurement, is to balance 

 the E. M. F. of a standard cell against the fall of potential over 

 the requisite resistance in a circuit of constant high resistance 

 (10,000 ohms), as in the Kayleigh or compensation method of 

 comparing electromotive forces.f The experimental cell is then 



* London Electrician, June 12, 1891 ; Carhart's Primary Batteries, pp. 136-150. 

 fPkil. Trans., Part If, 1884, p. 441. 



