It must be said, however, that the box must be exceedingly well closed for the screening to be perfect. The very narrowest chink permits their entrance, and at one time I thought I should have to solder a lid on before they could be kept entirely out. Clamping a copper lid on to a flange in six places was not enough. But by the use of pads of tinfoil and tight clamping, chinks can be avoided, and the inside of the box becomes then electrically dark. If even an inch of the circuit protrudes, it at once becomes slightly sensitive again; and if a mere single wire protrudes through the box, not connected to anything at either end, provided it is insulated where it passes through, the waves will utilise it as a speaking-tube, and run blithely in. And this happens whether the wire be connected to anything inside or not, though it acts more strongly when connected. In careful experiments, where the galvanometer is protected in one copper box and the coherer in another, the wires connecting the two must be encased in a metal tube (Figs. 19C and 21), and this tube must be well connected with the metal of both enclosures, if nothing is to get in but what is wanted. [Illustration: FIG. 20.--Spherical Radiator for emitting a Horizontal Beam, arranged inside a Copper Hat, fixed against the outside of a metal-lined Box, which contains induction coil and battery and key. One-eighth natural size. The wires pass into the box through glass tubes not shown.] Similarly when definite radiation is desired, it is well to put the radiator in a copper hat open in only one direction (Fig. 20), and in order to guard against reflected and collateral surgings running along the wires which pass outside to the exciting coil and battery, as they are liable to do, I am accustomed to put all the sending apparatus in a packing case lined with tinfoil, to the outside of which the sending hat (Fig. 20) is fixed, and to pull the key of the primary exciting circuit by a string from outside, so that not even key connections shall protrude, else exact optical experiments are impossible. [Illustration: FIG. 21.--General arrangement of experiments with the Copper “Hat,” showing Metal Box on a Stool, standing outside the Theatre. The Box is not exactly represented, but inside it the Radiators were fixed with a graduated series of apertures; the Copper Hat containing the Coherer is seen on the Table with the Metal Box on the left of the Table containing Battery and Galvanometer Coil connected to it by a compo pipe conveying the wires, as in Fig. 19C; the Lamp and Scale barely indicated at one side of the Table; a Paraffin Prism; and a Polarising Grid of copper wires stretched on a frame. (This figure is from a thumbnail sketch by Mr. A. P. Trotter, taken at the Lecture in 1894.)]