crease of 1° Fahr. produces a change of angle amounting only 

 to + 0'-05 ; so that a = + -000015, and the relative change of 

 the force of the bar = + -000029. 



"If we assume that the induced magnetism of the iron bars 

 is proportional to the inducing force, the coefficient p may be 

 found by inverting the bars, and observing the angles of de- 

 flection in the direct and inverted positions. For, these angles 

 being denoted by u and u, it may be readily shown that 



2 



P = = ; • 



sin u + sin u 



It was by this method that I originally proposed to deter- 

 mine the constant of the preceding formula. The assumption 

 upon which it rests is the same as that which Poisson has taken 

 as the basis of his theory of induced magnetism. It is, how- 

 ever, as Dr. Lamont has shown, not strictly in accordance with 

 fact ; and it is therefore necessary to seek another mode of 

 determining the constant. It is obvious that this quantity 

 will be known, if we can alter the inducing force artificially, 

 by a small but known amount, and observe the change of angle 

 thereby produced. This is the principle of the method de- 

 vised by Dr. Lamont for the purpose ; it is practised in the 

 following manner. 



" A magnet is placed at a considerable distance above or 

 below the suspended magnet, their centres being in the same 

 vertical line ; and it is so arranged as to be capable of rotation 

 round a horizontal axis parallel to the suspended magnet in 

 its deflected position. Let this magnet be first placed verti- 

 cally, in which position it exerts no direct action upon the 

 suspended magnet, but only on the iron bars. Then, if R and 

 R' denote the forces exerted by the auxiliary magnet upon 

 the two bars, g U= VR, SU = VR' ; so that if kn denote the 

 corresponding change of angle, expressed in scale-divisions of 

 the instrument, we have 



VR + VR'=X cosu kn. 



