ßQ K. TSURUTA. 



(See the upper one of the annexed figures). 



The results obtained remained on the whole 



exactly the same as before. Still another 



mode of making the junctions was tried 



(See the lower of the annexed figures). A 



small cylinder of icon was divided unequally into two lengthwise, and 



• in the larger segment a groove was cut for the 



^^£i£\ wires to lodge in. The segments with the wires 



^^S^^ between them were firmly tied together by fine 



iron wire and then by hempen thread. I could 



not find anything that might be ascribed to this change in my 



arrangements. 



Believing thus that the question as to the junctions may be set 

 aside, I next sought for other causes. One thing which I think 

 should be thoroughly examined is the iron-copper junctions (§ 4), 

 for even when they are only very slightly different, (of course suppos- 

 ing them kept at the same temperature), an E. M. F. would yet be 

 produced which could not be neglected in comparison with the one 

 we have to measure. A series of observations was, therefore, made 

 under the usual conditions, but with the junctions completely sur- 

 rounded by powdered ice. Steam being then cut off and the wires 

 allowed slowly to cool during the night,, another series of observations 

 was on the next day carried out, keeping the junctions in the usual 

 condition. These series are represented in Fig. 3, PL IV, the curve 

 a representing the former and the curve b the latter. Then, proceed- 

 ing exactly in the same way but with less time intervening, I 

 obtained another pair of curves a' and //. Thin soft iron, 0*5 m.m. 

 in diameter, was chosen for the. experiments, because I thought it 

 well not to lose much time in going through a cycle of loading and 

 unloading and thereby allow the temperature of the iron-copper junc- 



