independent of Changes of Temperature, 211 



of different percentages of carbon, were procured from a number of 

 different English, and Scotch firms. The size of these specimens was 

 in general about 15 cm. long and 1 or 2 thick, but as it was not 

 uniform the dimensions and weight in grams of each are given in 

 the accompanying table, columns I, II, and III. 



In column IV has been entered the dimension ratio, i.e., the ratio 

 of the length to the diameter or breadth, so that a comparison may 

 more consistently be made of the magnetic behaviour of any two 

 specimens. For thin rods, Cancani* finds that increase of this ratio 

 tends to diminish the temperature coefficient of a magnet. 



The course of an experiment was as follows: — The rod or bar in 

 its normal state, or after being hardened or annealed as occasion 

 required, was magnetised between the poles of a powerful electro- 

 magnet excited by a battery of twenty-six storage cells. The magnet 

 was then fixed rigidly in a horizontal tube, through which a stream of 

 cold water and steam could be alternately passed. The tube and its 

 contents were placed at a convenient distance from a sensitive dead- 

 beat magnetometer and at right angles to the magnetic meridian. 

 The deflections of the magnetometer needle were read by the usual 

 mirror and scale," the distance of the scale from the mirror being 

 1 metre, and from the readings were deduced directly the tempera- 

 ture coefficient and the total irreversible loss of magnetism. As the 

 deflections were never more than a few degrees of arc the angles and 

 their tangents were virtually equivalent. The intensity of magneti- 

 sation in C.Gr.S. units or magnetic moment per unit volume, although 

 not necessarily required, was approximately determined from the 

 formula 



I = H (ff-P)J tan 0.g 

 2d m 



in which the earth's horizontal force, H, was considered throughout 

 as constant and equal to 018 C.Gr.S. unit and also <r, the density, 

 was uniformly taken to be 7*8 ; m signifies the mass in grams ; d the 

 distance from the centre of the magnet to the magnetometer needle ; 

 I the half length of the magnet, and the deflection. 



The process of heating and cooling the magnet was continued until 

 the intensity fluctuated between two nearly constant values corre- 

 sponding to the temperatures of the cold water and steam. The 

 coefficient cc given in the eighth column was then calculated by 

 inserting these values in the equation 



It' — I t (l—a¥^i).f 



* R. Cancani, 'Atti della R. Acc. dei Lincei,' (4), 3, pp. 501—506, 1887; 

 ' Beibl.,' vol. 11, 1887. 



f I have followed the customary mode of writing this formula with a negative 

 sign preceding the coefficient, a; and, hence, a negative coefficient indicates an 

 increase of magnetic intensity with increase of temperature. 



