wrought-iron of high permeability. This core is fixed in a horizontal position, its ends being turned down to fit into massive castings, which form the continuation of the field-magnet core, and also the pole-pieces. The core is made to fit into the castings accurately, in order to avoid as far as possible the introduction of any additional magnetic resistance at the joints, and the three pieces are securely held together by two circular steel nuts, which screw on to the extreme ends of the wrought-iron core as shown.

As the field-magnet is fixed with the pole-pieces downward, it is necessary to magnetically insulate it, as far as possible, from the cast-iron bed-plate, in order to minimise the leakage of the lines of force. This is effected by supporting the field-magnet on two strong gun-metal brackets, which are bolted to the bedplate and field-magnet, as shown in figs. 200 and 201.

The spindle is of mild steel, the bearings being of gun-metal

lined with white metal; and it will be seen that the design is such that the centre of gravity of the moving parts is very low, thus securing freedom from vibration.

The sections given in figs. 201 and 202 illustrate the manner in which the armature is built up. The core-plates are of Swedish

charcoal iron, and a stout wrought-iron disc is placed at each end of the core. The plates are driven from the shaft by means of a gun-metal frame made in halves, and having three radial arms, with extensions at the ends. The edges of the arms fit into slots cut in the core discs, and the two halves of the frame are held firmly together by three steel bolts which pass though the arms. When the nuts on these bolts are screwed up, the extensions of the

arms press against the stout wrought-iron end discs, and hold the whole of the core discs tightly in position. The gun-metal driving frame is keyed to the shaft, thus securing rigid driving; and one end of the frame bears against a hub on the shaft, two steel nuts being screwed up against the other end.

The construction of the commutator is shown to the left in fig. 202. A phosphor bronze sleeve fits over the shaft, and it is provided at the right-hand end with a flange, coned to fit into the ends of the hard-drawn copper segments, while a similarly coned washer fits into their other ends, the washer being clamped up tight by means of a circular nut.

It will be seen that the machine is compact, and that the design is simple and readily admits of good mechanical construction throughout.

The particular machine illustrated is shunt-wound, and when running at 750 revolutions per minute it is capable of maintaining a current of 380 amperes at a potential difference of 105 volts.

We have remarked that it is not possible in practice to 'compound' a machine to give a constant current in the same manner that a constant potential can be maintained. Hence a variety of other devices, mostly mechanical, have been suggested for the purpose, one of which has already been described, and we will now direct attention to a machine, designed and constructed by Mr. J. G. Statter, to maintain a constant current at a constant speed when the external resistance is varied; or when the speed is varied and the resistance remains unaltered; or, within certain limits, when both speed and resistance vary.

Although the machine has not been used to any considerable extent, it probably best represents its particular class, and the considerations involved in the design will prove instructive.

When an armature is rotated in a simple field, with the brushes removed from the commutator, there are two points, at opposite extremities of a diameter, at one of which the potential is a maximum (positive) and at the other a minimum (negative), and it is at these points that the brushes should be set in order to obtain the greatest difference of potential. From the negative to the positive brush, either way round the commutator, the potential gradually increases in value, and if the negative brush were shifted,

say, 20° forward, it would touch at a point of higher potential, and consequently the difference between the two brushes would be reduced. A like reduction would follow if the positive brush were moved forward, because it would then make contact at a point of lower potential; and in the third case the difference of potential between the two might be decreased by giving them simultaneously a greater angle of lead, until they would be at nearly the same potential when moved through an angle. of 90°.

It follows that by merely shifting the brushes the potential difference at the terminals may be made what we please from the maximum downwards, and this method might be employed to vary the pressure, and therefore the current, to suit the requirements of the circuit. Thus the brushes might be set 20° ahead of their normal position of no sparking-that is, where the difference of potential between them is at a maximum-and then, if the current became too strong, the brushes could be shifted yet further ahead, thus reducing the potential difference and also the current; while by moving the brushes back towards their normal position the current could be increased in strength, should it fall below the desired value.

This would, however, be impracticable in an ordinary dynamo, on account of the terrific sparking which would ensue ; and one of the principal features in the machine under notice is the method by which it becomes possible to vary the lead of the brushes through a considerable angle, without causing this sparking.

Confining our attention at present to the case of ring armatures, it will be remembered that when a coil which is carrying the whole of the current generated in one half of the armature is short-circuited by the brush, the electro-magnetic inertia of the coil, due to its self-induction, prevents the instantaneous cessation of the current; and that even if the coil is commutated at the moment when its plane is exactly at right angles to the field, and therefore in itself inactive, yet there will be a considerable spark caused by what we may term the self-induction current in the coil. For this reason it is found to be necessary to give the brushes a slight extra lead, so that the coil may begin to cut lines of force, and have induced in it an opposing E.M.F. sufficiently strong to just counteract this self-induction effect, and stop the current in