two means present themselves:  (1) to cause a, favorable magnetic current and act upon the armature, and (2) to suppress such portions of the current in the spirals as are injurious in effect.  The first solution was thought of by Gramme in 1871, and is represented diagramatically in Fig. 2.  The second is due to Prof.  Pfaundler, and dates back to 1870.  The core is cut through the center (Fig. 3), and the portion to the right is suppressed; the current is interrupted between da and cd’, and is closed only between a and c (v, Fig. 1).  It results from this arrangement that, under the action of the current, the polarity due to remanent magnetism does nothing but increase.  It suffices then for but little remanent magnetism to prime the machine; the polarity of the shell continues to increase, and the energy of the magnetic field, and consequently of the current, has for a limit only the saturation of the soft iron.  If, now, we curve the core, the spirals, and the armature into a circle, we have a Gramme or a Pfaundler machine, according as we consider Fig. 2 or Fig. 3.

[Illustration:  FIG. 2.]

[Illustration:  FIG. 3.]

This latter apparatus has in this case the form shown in Fig. 4.

[Illustration:  FIG. 4.]

The spiral, s m b, is movable, and the core, N o s, is kept in a position of equilibrium by virtue of its weight, and is provided with rollers.  For the sake of greater clearness, the front part of the armature is supposed to be removed.  The current does not circulate in the spirals to the right of the diameter, W O, which latter is not absolutely vertical.  The position of the rubbers and armature is regulated once for all.  We do not know just what were the means devised by Kravogl to suppress the current in the spheres to the right.  At all events, it is probable that the system has grown old since Gramme invented his collector.  In the application of the Kravogl motor to the generation of continuous currents, Professor Pfaundler now proposes to ingeniously utilize the Gramme collector.  In such a case the arrangement shown in Fig. 5 would be adopted.  Let us suppose an ordinary collector having as many plates as there are sections in the ring, these plates being connected as usual with the entrance and exit wires of the sections.  The diametrically opposite touches that are in the line, W O, are divided, and one of the halves is connected at the entrance, c a’ (Fig. 4), with the corresponding section, while the other communicates with the exit, c’ a, of the neighboring section.  Each of these halves is prolonged by a piece of metal bent into the form of an arc of a circle and embracing a little less than a semi-circumference.  Between these prolongations there is an insulating part.  In the rotary motion of the spiral, at least one of the touches is always outside of the arc comprised between the brushes, R. In order to secure a continuity of the circuit in the effective arc, W S_ o_, it is only necessary to arrange a rubber, M, in such a way as to establish a communication between the two parts of the divided touch as soon as this latter enters the arc under consideration.