TüERMO-ELECTRIC EFFECTS OF LONG. TENSION IN DIFF. METALS. 59 



as a whole arise ? That was the query which I set myself to solve, 

 for if it is not due to some flaw in my experimental arrangements, or 

 if it has root in the nature of the thing, then the labour of referring 

 the E. M. F. to absolute measure would be to no good effect. It 

 should be mentioned that I observed not merely a shifting towards 

 the negative side, but sometimes also, though not often, a shifting 

 towards the positive side. 



The shifting would take place, as already described, after a num- 

 ber of hours, during which the wires had been allowed to cool to the 

 atmospheric temperature and left undisturbed, or, as observed by me 

 once, even after a number of alternate heatings and coolings. Could 

 the standard cell have been really inconstant, the shifting might have 

 been due to that, but that the cell's E. M. F. should change by 1/10 

 or more seemed to me to be exceedingly improbable. Therefore, in 

 my eiforts to account for the difficulty my attention was at first 

 concentrated on the manner of making the junctions of the stretched 

 and un stretched portions. As described in my preceding paper, I 

 coupled them by putting their ends side by side and binding them 

 together round and round with fine iron wires as firmly as possible. 

 Lateral compressions thus introduced, one might suspect, could inter- 

 fere with the production of the thermo-electric E. M. F. due to longi- 

 tudinal tension. But seeing that the thermo-eJectric curves actually 

 obtained, one series of measurements after another, are nearly parallel; 

 that the points plotted on them are among themselves almost similarly 

 placed ; and that the shifting is sometimes towards the positive side 

 in the scale of E. M. F., I cannot attribute such great and varying 

 shifting to the interference of lateral compression. Wrong contacts 

 at the junctions were another thing to be looked after. I cut short 

 pieces from the experimental wires, laid them beside and along the 

 junctions, and bound all the four pieces together as above described 



