When lightning itself strikes a guard, it has indeed often been found that the opposite sides of the protective air gap are fused together. This, no doubt, may be partly due to a straightforward melting or welding by heat, but it is probably not solely that. Molten metals without a flux do not so readily weld. It is almost certainly due to a cohesive action also, the difference of potential between the molten terminals resulting in adhesion and amalgamation, a phenomenon also observed in the frequent locking of an electric arc formed between two metallic electrodes. However this may be, certainly the phenomenon occurs on a small scale, for if the pair of knobs or points placed as a shunt to protect a galvanometer or other telegraphic instrument from lightning (or what is easier experimentally and essentially the same thing, from a Leyden jar discharge) be set too close together, the galvanometer will be found to be short-circuited after a spark, and the knobs will be found, both by mechanical and electrical tests, to be feebly united at a single point.[23] Not only, however, is the galvanometer short-circuited by the metallic junction so formed, but at the instant of the formation of the joint it experiences a very perceptible kick, indicating a momentary current, coincident no doubt with the electric discharge, but one from which it would have been protected had not the junction occurred. The galvanometer kick is clearly an effect due to the uniting metals, but it has not yet been fully elucidated; it seems to have been first observed by Mr. Stroh in his excellent researches on microphonic action, related in the _Journal_ of the Society of Telegraph Engineers, 1883 and 1887, and it may possibly be thermo-electric, as Prof. Hughes, who also observed it, thinks likely; but it may be electro-chemical, or it may be connected with an effect observed later by FitzGerald in his galvanometer mode of detecting Hertzian waves, which he published at the Royal Institution in 1890. The point of present interest is the cohesion which sets in between the knobs when the spark occurs: an extremely feeble spark was found sufficient to produce the effect, provided the surfaces were already almost infinitely close together, _i.e._, provided they were already in what would be called contact, with the merest imperceptible film of (probably) oxide separating them, just the kind of film which a chemical flux is useful in removing. The electrical stimulus appears to act as such a flux, and the adhesion of the two surfaces was demonstrated by an electric bell and single cell in circuit. Every time the spark occurred the bell rang, and continued ringing, until the table, or some part of the support of the knobs, was tapped so as to shake or jar them asunder again.[24] The arrangement constitutes a convenient detector in the syntonic Leyden jar experiment, depicted in Fig. 4, p. 6 (_see_ also p. 21). [23] “Modern Views,” second edition, p. 359.