288 K. HONDA: on the jjuïual influence 



of tlie shorter side of the rectangular copper wire with 

 the brass plate B. The corresponding point on the opposite 

 side of the copper wire and the brass plate B' on the same 

 side were connected with the poles of an accumulator. The 

 current thus entered the brass piece B', then flowed through 

 the straight copper wire, and passed to two branches of the 

 wire. 



By this arrangement, tlie direct action of the current was 

 nearly eliminated ; the accuracy of the balance iu the branches 

 of the lead wire was tested before each experiment. The current 

 flowing symmetrically with respect to the central line along each 

 side of the rectangular wire produces nearly equal but opposite 

 field on the iron tube, and is 3 cm. apart from it ; hence the 

 disturbance of the circularly magnetizing field would be insensible. 

 This arrangement was found inconvenient for experimenting with 

 thick tubes. 



With larger iron and nickel tubes, I used another me- 

 thod of eliminating the heating by current. Each tube was 

 wound uniformly with an insulated copper wire around the wall 

 along its generating lines as to produce a nearly uniform 

 circular field by a current passing through the wire, as was 

 already done by Siemens and Schnitze. The coil was protected 

 by parafiin wax. It was then fixed in the central line of the 

 magnetizing coil, and cold water kept constantly flowing in the 

 tube. 



The magnetization and its change Avere mostly measured by 

 the magnetometric method. To determine them by means of the 

 ])allistic method, a secondary coil was wound on each tube. 

 Placing the tube inside the magnetizing coil, and passing a cur- 

 rent of cold water constantly round it, the change of longitudinal 



