40 Messrs. K. Honda and T. Terada on the Change of 



removed from it, and the distances of the knife-edges were 

 adjusted, until neither the addition nor removal produced any 

 rotation of the mirror. Since the elastic elongation was not 

 exactly a linear function of the tension applied, a compen- 

 sated system for a given initial and additional weight caused 

 a rotation of the mirror for a different initial weight and the 

 same additional weight. Hence the compensation was made 

 anew for every initial weight, 1 kilogram always being used 

 for the applied weight. It was, however, a matter of no 

 great ease to obtain an exact compensation. Hence the 

 slight deflexion in no field was always recorded, and taken 

 into account in the calculation of the deflexion due to the 

 change of elasticity by magnetization. To prevent shocks 

 in putting on the additional weight, the face of the initial 

 weight was covered by a mat of cotton-wool. In this way, 

 the reading of the deflexion was greatly facilitated. 



In the above arrangement, the rotation of the mirror 

 caused by a change of temperature is very small, as it is 

 affected only by the relative expansion of the specimen and 

 the compensating wire. But to avoid the effect of heating 

 as much as possible, the coil was water jacketed. 



To increase the sensitiveness of the arrangement, a long 

 scale-distance of 7*23 m. was used. The scale was graduated 

 on a ground glass, and illuminated from behind by a mantled 

 gas-flame. The telescope was placed in front of the apparatus 

 and near the rotating mirror; consequently the line joining 

 the centres of the scale and the mirror deviated slightly from 

 the line of sight of the telescope. In this way, the scale at 

 such a long distance could be read to a tenth of a millimetre 

 with the observing telescope. 



If the specimen undergoes a differential elongation 8h a 

 rotation &(f> of the axis is produced, such that 8<f> = 81/2>\ 

 where r is the radius of the axis. If the scale-distance be d 

 and the reading on the scale corresponding to the elongation 

 hi be n, w T e have 



Sd> = — 7 ; hence SZ=--.-. 



T 2d a 



In our case, 2^=1*61 mm., so that a deflexion amounting 

 to one division of the scale in the field of the telescope cor- 

 responds to an elongation of 4*47 x 10~ 8 per unit length of 

 the specimen. The sensitiveness of the apparatus was there- 

 fore about ten times as great as that of the apparatus used 

 by many previous investigators. 



The magnetizing current supplied by a set of secondary 



