296 PROCEEDINGS OF THE AMERICAN ACADEMY 



be subtracted from the total magnetism of the armature when in con- 

 tact with the magnet; and when the two are nearly equal, as a small 

 error in observation will produce a large relative error in the result, 

 any small irreirularities should be removed from the curve. When, 

 however, there is any quantity of permanent magnetism at the junc- 

 ture, and of an opposite sign to that of the magnet, allowance should 

 be made for a portion of this magnetism spreading over the magnet. 

 The law by which this happens will be demonstrated by and by. It 

 is the same, nearly at least, as if the magnet were soft iron. About 

 half of this magnetism, therefore, spreads over the magnet. Unless 

 tliis counter-distribution be taken into account, it will be absolutely 

 impossible to obtain accurate results ; for no bar of soft iron used as an 

 armature will remain free from permanent magnetism, even if it were 

 perfectly non-magnetic before. 



I next investigated the change due to closeness of contact, and 

 found that a bar with perfectly flat ends varied in the amount of mag- 

 netism it drew from a magnet ten to twenty per cent, according to the 

 means taken to secure perfect contact of surfaces. 



The amount drawn off varies with the length. Some idea of the 

 proportion may be given by these numbers. 100 units being the 

 amount drawn off by a metre bar from a metre magnet, a bar a metre 

 and a half long would draw off about 1 35 units ; half a metre long, 

 80 units ; and 9 cm. long, about 30 units. A mass of iron in weight 

 equal to 11 6 cm. of iron rod of the size used above draws off about 

 G5 units ; so that the length makes a much greater difference than the 

 mass. When for a solid rod we substitute a bundle of wires of equal 

 length and weight, there is no sensible difference in the amount of 

 magnetism drawn off, as tested by the current induced in a helix. 



The old apparatus was now laid aside. The new one differed in 

 several respects. The bobbin was of brass ; thickness outside, 5^ mm. ; 

 inside, 3 mm. ; depth of cut was about 5J mm. ; and the diameter of 

 the rods was 1 2 mm. The magnet and armature had a small hole bored 

 in each end, and a small screw was fitted to draw them tightly together. 

 The surfaces of the ends were made perfectly flat in a lathe. Two 

 clamps served to limit exactly the motion of the bobbin to any distance 

 desired. The magnets and armature were placed east and west on a 

 table about eight metres from the galvanometer, whose deflections 

 were observed with a telescope. Therefore both the induced mag- 

 netism of the earth was eliminated, and the magnet had but little effect 

 upon the galvanometer. 



