Another of Du Moncel's researches[2] relates to the effect of polar projections or shoes--movable pole pieces, if you like--upon a horseshoe electromagnet. The core of this magnet was of round iron 4 centimeters in diameter, and the parallel limbs were 10 centimeters long and 6 centimeters apart. The shoes consisted of two flat pieces of iron slotted out at one end, so that they could be slid along over the poles and brought nearer together. The attraction exerted on a flat armature across air gaps 2 millimeters thick was measured by counterpoising. Exciting this electromagnet with a certain battery, it was found that the attraction was greatest when the shoes were pushed to about 15 millimeters, or about one-quarter of the interpolar distance, apart. The numbers were as follows: Distance between shoes. Attraction, Millimeters. in grammes. 2 900 10 1,012 15 1,025 25 965 40 890 60 550 [Footnote 2: "La Lumiere Electrique," vol. iv., p. 129.] With a stronger battery the magnet without shoes had an attraction of 885 grammes, but with the shoes 15 millimeters apart, 1,195 grammes. When one pole only was employed, the attraction, which was 88 grammes without a shoe, was _diminished_ by adding a shoe to 39 grammes! CONTRAST BETWEEN ELECTROMAGNETS AND PERMANENT MAGNETS. Now I want particularly to ask you to guard against the idea that all these results obtained from electromagnets are equally applicable to permanent magnets of steel; they are not, for this simple reason. With an electromagnet, when you put the armature near, and make the magnetic circuit better, you not only get more magnetic lines going through that armature, but you get more magnetic lines going through the whole of the iron. You get more magnetic lines round the bend when you put an armature on to the poles, because you have a magnetic circuit of less reluctance with the same external magnetizing power in the coils acting around it. Therefore, in that case, you will have a greater magnetic flux all the way round. The data obtained with the electromagnet (Fig. 42), with the exploring coil, C, on the bend of the core, where the armature was in contact, and when it was removed are most significant. When the armature was present it multiplied the total magnetic flow tenfold for weak currents and nearly threefold for strong currents. But with a steel horseshoe, magnetized once for all, the magnetic lines that flow around the bend of the steel are a fixed quantity, and, however much you diminish the reluctance of the magnetic circuit, you do not create or evoke any more. When the armature is away the magnetic lines arch across, not at the ends of the horseshoe only, but from its flanks; the whole of the magnetic lines leaking somehow across the space. Where you have put the