Possible Existence of Mutual Induction between Masses. 595 



A steel flywheel A (fig. 2), 56 cm. in diameter and 11 cm. 

 thick, is mounted on a vertical shaft, 0, supported by ball- 

 thrust bearing, P, and driven by an electric motor, R. Above 

 the flywheel is suspended a disk, B, 51 cm. in diameter, 

 made of very pure porcelain, weighing about 10 kg. The 

 first suspension was made of very thin steel strip, but the 

 final suspension is made of two round steel wires, each 

 0'025 cm. in diameter. The length of the suspension is 

 21 metres. The distance between the bifilars is about 

 - 15 cm. The torsional control on B is extremely small, 

 amounting to only 28 dyne-cm. for a deflexion of 1 radian. 

 The angular swing of B upon its principal axis has a natural 

 period of 2460 seconds. A mirror I (fig. 2) is mounted on 

 a stiff wire attached to the disk B, which enables the move- 

 ment to be accurately observed on a scale at 6 metres 

 distance. By means of a telescope a movement of 0*01 cm. 

 corresponding to a deflexion of 1/12,000 of a radian, can be 

 estimated, so that one can observe the effect of acting on the 

 edge of the disk with a force amounting to only 10~ n of the J' 



weight of the disk. The chief constants for this apparatus I 



are given in Appendix I. 



The experiment consists in rapidly accelerating the fly- 

 wheel A and observing the effect (if any) on B. In order ! 

 to isolate B from the effects of any air currents and vibration, 

 two screens, C and D, are interposed between A and B ; the 

 screen C is supported on a pillar of masonry to which the 

 bedplate of the motor is attached, and the screen D is 

 supported by a steel girder, S, fixed in the ground and 

 stayed by means of wires from distant walls, which were « 

 not appreciably affected by vibration. Two sheet-iron 

 covers, E and F, are placed over B to keep off currents 

 of air. The outer cover F is to protect the inner cover 

 from currents of air which might cause a difference in 

 temperature in parts of the inner cover. 



As might be expected, it was found to be very difficult to 

 screen B from accidental disturbances when the flywheel 

 was run up to speed. Many months were expended in 

 finding out the source of the disturbances and reducing 

 their magnitude. One cause of trouble was a very slight 

 swirling action of the screen D, which communicated a slight 

 swirl to the air inside, which caused B to rotate. To 

 eliminate this effect in the later experiments, B was sur- 

 rounded by an inner screen, shown at 8 in fig. 3. This inner 

 screen was completely suspended from the top of the building 

 by means of three wires. The torsional control upon it was 

 185 times greater than the control on B, so that it was hardly 



2 S2 



