190 WEBER OX THE CONNEXION OF DIAMAGNETISM 



add, — 1st, that it is very easy to prove if this compensation ac- 

 tually exists, by effecting, not small changes of intensity, but by 

 interrupting or commutating the entire current ; 2nd, that if it 

 should appear that this compensation is not complete, it is very 

 easily rendered so by winding an end of the induction-spiral 

 once or oflener round the end of the spiral of the electro-dia- 

 magnet. In this way it is easy to rescue the diamagnetic induc- 

 tion from every foreign influence. 



Tfie remaining portions of the Induction-apparatus. 



With regard to the arrangement of the remaining portions, 

 which are more or less left to the discretion of the observer, I 

 would make the following remarks. To move the bar of bismuth 

 to and fro in the induction-spiral, I connect it with a crank, 

 attached to the wheel B, fig, 5, Plate I. To cause the current 

 excited in the induction-spiral by the forward motion of the 

 bismuth to have the same direction through the multiplier as 

 that excited by its backward motion, a commutator dd is at- 

 tached to the wheel, which moves with the latter, and by which 

 the connexion of the ends of the spiral with the wire of the mul- 

 tiplier is reversed at the end of every semi-revolution (at the 

 moment when the bismuth attains the commencement or end of 

 its path). The uniform direction of all induced currents through 

 the multiplier would be followed by a uniform deflection of the 

 needle towards the same side. To enable the observer to pro- 

 duce a deflection towards the other side, a second commutator, 

 ee, fig. 5, is placed beside the telescope E, which can be changed 

 by the observer himself, and which may be called the subsidiary 

 commutator. The following must be particularly attended to, — 

 1st, that the induction must be increased by increasing the velo- 

 city of the wheel, rather than by lengthening the path traversed 

 by the bismuth ; 2nd, that no thermo-magnetic current shall be 

 generated at the rotating commutator ; it must be so arranged 

 that metals of the same kind only rub against each other. The 

 influence of such currents, when very feeble, may be readily 

 eliminated by suitably combining the observations. Finally, to 

 obtain an approximate idea of the strength of the galvanic cur- 



