Magnetisation in the Length of Metal Rods. 115 



In working with this apparatus, three possible sources of error were 

 .soon revealed. The first was due to the expansion of the rod in con- 

 sequence of the heating of the coil by the current. This effect could 

 be distinguished from the elongation resulting from magnetisation, by 

 the fact that the latter took place quite instantaneously, while the 

 expansion due to heat was gradual ; but it was likely to lead to un- 

 certainty in estimating the amount of permanent elongation accom- 

 panying the permanent or residual magnetism of the rod. This 

 uncertainty could be reduced to a minimum by taking care to close 

 the circuit only for a second or two when making an observation. 

 The second possibility of error resulted from the gradual yielding of 

 ihe magnetic rod, or, more probably, of some part of the base or 

 frame of the apparatus, under the pressure, small though it was, of 

 the brass lever. This may to a great extent be obviated by adjusting 

 the apparatus and leaving it for half an hour before making an ob- 

 servation ; but I am not quite sure that it ev^r entirely disappeared, 

 for even though the image of the wire remained perfectly steady upon 

 the scale so long as the apparatus was quite undisturbed, it is possible 

 that the shocks produced by magnetising and demagnetising the rod 

 might cause a sudden slight upward movement of the image, thus 

 making the permanent elongation of the rod appear to be somewhat 

 less than it in fact was. I think, however, that the error, after the 

 apparatus has been at rest for a sufficient time, must be very small. 

 In observing the purely temporary elongation resulting from so much 

 of the magnetisation as is purely temporary, no uncertainty what- 

 ever need arise from this cause, for the experiment can easily be re- 

 peated as often as may be necessary to obtain uniform . results ; but 

 an observation of the permanent extension cannot be repeated with- 

 out dismantling the apparatus and demagnetising the rod, after which 

 its condition will probably not be exactly the same as before. 



Lastly, errors may arise from the electromagnetic attraction 

 existing between the coil and the rod, which tends to draw a uniform 

 rod into such a position that the middle point of its axis shall coincide 

 with the centre of the coil. As at first constructed, the coil in my 

 apparatus was attached by means of screws to the under side of the 

 table, and the rod under examination passed freely through it, touching- 

 nothing whatever except the brass plate at the bottom and the lever at 

 the top. The length of the magnetic portion of the rod — that whicli 

 was the subject of the experiment — was in every case, as already 

 stated, exactly 10 cm., or 1'5 cm. less than the length of the coil. But 

 the distance from the brass base plate to the lever was 21 cm., and in 

 order to increase the experimental rods to this length, pieces of thick 

 brass wire were screwed or soldered to their two ends ; thus the rods 

 when prepared for the experiment were of a compound form, con- 

 sisting of iron, steel, or nickel in the middle, and brass at each end 



i 2 



