30 



K. TSURUTA. 



leading wires. The wires, excepting A, were all wrapped in thin 

 Japanese paper and thus insulated from each other. 



§ 5. A reversing key introduced into the circuit, is necessary 

 for eliminating not only errors caused by the variation of the zero of 

 the galvanometer, but also any extraneous current found to accompany 

 the one which is to be measured. The choice of the form of key to 

 be used in connection with the present investigation was found to be 

 one of the most difficult things to do ; indeed, this kept me from get- 

 ting good results for a long time. Many forms were tried, and the 

 rocker of the usual construction together with six mercury pools in a 

 block of wood found to be the best of them all. My rocker was made 

 of copper and was very small. Great care is of course to be taken not 

 to expose the key to any considerable source of heat ; and in my own 

 case the whole reversing arrangement was put within a wooden box 

 and moved by means of a handle from without. 



§ 6. As I had to deal with a very feeble current, the galvano- 

 meter, which was of low resistance (0*26 ohms.), had to be made 

 highly sensitive and to be also well damped. The adjustment of a 

 controlling magnet, or two magnets in case of need, caused a good 

 deal of trouble and loss of time in every experiment. At the same 

 time, as a high degree of sensitiveness was attained, any small causes 

 of disturbance became proportionally pronounced and a series of read- 

 ings would come out somewhat irregular. I, therefore, usually did 

 not push the sensibility to a very high degree, but was content if it 

 was quite sufficient to show the variation of the thermo-electric E. M. F. 

 No reduction of readings taken to absolute measures was made, as the 

 aim of the present series of experiments was simply to study the 

 general nature of the thermo-electric hysteresis. 



§ 7. The curves showing the relation of the thermo-electric 

 E. M. F. to load, which I obtained in my earlier experiments on an- 



