[Illustration:  Fig. 3.—­Disruptive discharge coil.]

It is contained in a box B (Fig. 3) of thick boards of hard wood, covered on the outside with zinc sheet Z, which is carefully soldered all around.  It might be advisable, in a strictly scientific investigation, when accuracy is of great importance, to do away with the metal cover, as it might introduce many errors, principally on account of its complex action upon the coil, as a condenser of very small capacity and as an electrostatic and electromagnetic screen.  When the coil is used for such experiments as are here contemplated, the employment of the metal cover offers some practical advantages, but these are not of sufficient importance to be dwelt upon.

The coil should be placed symmetrically to the metal cover, and the space between should, of course, not be too small, certainly not less than, say, five centimetres, but much more if possible; especially the two sides of the zinc box, which are at right angles to the axis of the coil, should be sufficiently remote from the latter, as otherwise they might impair its action and be a source of loss.

The coil consists of two spools of hard rubber RR, held apart at a distance of 10 centimetres by bolts c and nuts n, likewise of hard rubber.  Each spool comprises a tube T of approximately 8 centimetres inside diameter, and 3 millimetres thick, upon which are screwed two flanges FF, 24 centimetres square, the space between the flanges being about 3 centimetres.  The secondary, SS, of the best gutta percha-covered wire, has 26 layers, 10 turns in each, giving for each half a total of 260 turns.  The two halves are wound oppositely and connected in series, the connection between both being made over the primary.  This disposition, besides being convenient, has the advantage that when the coil is well balanced—­that is, when both of its terminals T_1 T_1 are connected to bodies or devices of equal capacity—­there is not much danger of breaking through to the primary, and the insulation between the primary and the secondary need not be thick.  In using the coil it is advisable to attach to both terminals devices of nearly equal capacity, as, when the capacity of the terminals is not equal, sparks will be apt to pass to the primary.  To avoid this, the middle point of the secondary may be connected to the primary, but this is not always practicable.

The primary pp is wound in two parts, and oppositely, upon a wooden spool W, and the four ends are led out of the oil through hard rubber tubes tt.  The ends of the secondary T_1 T_1 are also led out of the oil through rubber tubes t_1 t_1 of great thickness.  The primary and secondary layers are insulated by cotton cloth, the thickness of the insulation, of course, bearing some proportion to the difference of potential between the turns of the different layers.  Each half of the primary has four layers, 24 turns in each, this giving a total of 96 turns.  When