226 Mr. K. Tsuruta on some Effects of 



twisting and untwisting, the effects anticipated would come 

 out considerably complicated, though otherwise simple. 



Heat was applied to the second junction by means of 

 steam sent into a hollow vessel of brass surrounding it. 



To begin with, the experimental wire was well annealed, 

 and at the same time straightened. It was cut into three 

 parts, each of a proper length. The ends of two of these — 

 one the longest and to be twisted, and the other to remain 

 unaffected — were first fixed to the second junction-piece, and 

 put in position. Then the outer end of the part to be twisted 

 was passed through the first junction-piece, and was firmly 

 held there, as explained above, with the help of two small 

 spanners, so as to avoid as much as possible any arbitrary 

 initial twisting. One end of the remaining third part, which 

 was to remain unaffected, was secured in position as ex- 

 plained above. The free ends of the two idle wires were 

 dipped in mercury contained in two test-tubes, and sur- 

 rounded with a large quantity of water, the whole being then 

 put in a box with thick wooden walls. 



The galvanometer used was of low resistance, and made 

 highly sensitive with a controlling magnet or two. The 

 ordinary arrangement of rocker and six mercury pools 

 served as commutator, which it was necessary to have on 

 account of the high sensitiveness of the galvanometer with 

 incessant changes of zero. The commutator seems to have 

 been a considerable source of disturbance, as different metals 

 came here into contact (I had not at that time that form 

 of commutator which was used in my other investigations). 

 Absolute measurements of the current produced were, to my 

 extreme regret, reserved to later occasions ; in the curves given 

 below the current-strength is indicated only in scale-divisions. 



in my arrangement there are, it will be seen, many things 

 which must be improved or replaced by others, some of 

 which in fact I have put on record in my journal. All these 

 untried things must be omitted, and of the results I obtained 

 only the following three points seem to deserve place here. 



(a) As it was thought that thermo-electric effects of twist, 

 if any, would show themselves more conspicuously in the case 

 of iron than with other metals, as had been the case with refer- 

 ence to longitudinal tension, itwas soft iron that I examined first 

 of all. The experimental wire was, to begin with, subject to the 

 tension of some leaden weights, and then it was twisted and 

 untwisted two or three times by way of drilling ; definitive 

 observations were then made. The deflexion given as corre- 

 sponding to a given amount of twist must be in reality the 

 result of the superposed deflexions due to longitudinal tension 

 and twist applied. The pure effect of twist can properly be 



