I picture the action as follows: Suppose two fairly clean pieces of metal in light contact—say two pieces of brass or of iron—connected to a single voltaic cell; a film of what may be called oxide intervenes between the surfaces so that only an insignificant current is allowed to pass, because a volt or two is insufficient to break down the insulating film, except perhaps at one or two atoms.[14] If the film is not permitted to conduct at all, it is not very sensitive; the most sensitive condition is attained when an infinitesimal current passes, strong enough just to show on a moderate galvanometer.
Now let the slightest surging occur, say by reason of a sphere being charged and discharged at a distance of forty yards; the film at once breaks down, perhaps not completely—that is a question of intensity—but permanently. As I imagine, more molecules get within each other’s range, incipient cohesion sets in and the momentary electric quiver acts somewhat like a flux. It is a singular variety of electric welding. A stronger stimulus enables more molecules to hold on, the process is surprisingly metrical; and as far as I roughly know at present, the change of resistance is proportional to the energy of the electric radiation, from a source of given frequency.
It is to be specially noted that a battery current is not needed to effect the cohesion, only to demonstrate it. The battery can be applied after the spark has occurred, and the resistance will be found changed as much as if the battery had been on all the time.
The incipient cohesion electrically caused can be mechanically destroyed. Sound vibrations, or any other feeble mechanical disturbances, such as scratches or taps, are well adapted to restore the contact to its original high resistance sensitive condition. The more feeble the electrical disturbance the slighter is the corresponding mechanical stimulus needed for restoration. When working with the radiating sphere ([Fig. 19]) at a distance of forty yards out of window, I could not for this reason shout to my assistant to cause him to press the key of the coil and make a spark, but I showed him a duster instead, this being a silent signal which had no disturbing effect on the coherer or tube of filings. I mention 40 yards, because that was one of the first outdoor experiments; but I should think that something more like half-a-mile was nearer the limit of sensitiveness for this particular apparatus as then arranged. However, this is a rash statement not at present verified.[15] At 40 or 60 yards the exciting spark could be distinctly heard, and it was interesting to watch the spot of light begin its long excursion and actually travel a distance of 2 in. or 3 in. before the sound arrived. This experiment proved definitely enough that the efficient cause travelled quicker than sound, and disposed completely of any sceptical doubts as to sound waves being, perhaps, the real cause of the phenomenon. Signals were obtained across the full width of the college quadrangle, and later, with larger apparatus, between the college tower and another high building half-a-mile away.
Fig. 19.— Radiator used in the library of the Royal Institution, exciting the Coherer ([Fig. 17]) on the lecture table in the Theatre. I also used a radiator with two or with three large spheres between two knobs, and described it in Nature, Vol. 41, p. 462, 1890. This is the radiator which Prof. Righi has improved and made in a compact form with oil between the two middle spheres.
Invariably, when the receiver is in good condition, sound or other mechanical disturbance acts one way, viz., in the direction of increasing resistance, while electrical radiation or jerks act the other way, decreasing it. While getting the receiver into condition, or when it is getting out of order, vibrations and sometimes electric discharges act irregularly; and an occasional good shaking does the filings good. I have taken rough measurements of the resistance by the simple process of restoring the original galvanometer deflection by adding or removing resistance coils. A half-inch tube, 8 in. long, of selected iron turnings ([Fig. 18]) had a resistance of 2,500 ohms in the sensitive state. A feeble stimulus, caused by a distant electrophorus spark, brought it down 400 ohms. A rather stronger one reduced it by 500 and 600, while a trace of spark given to a point of the circuit itself ran it down 1,400 ohms.
This is only to give an idea of the quantities. I have not yet done any seriously metrical experiments.
Added later.—My assistant, Mr. E. E. Robinson, early noticed that when a telephone was used as receiver, say with a single-point coherer ([see illustration on opposite page]), which is a very sensitive arrangement, every disturbance of the coherer due to received waves is accompanied by a crackle or tick in the telephone, without any tapping back being necessary. This is, indeed, the easiest mode of receiving signals, and we often practised it. If a suitable, well-damped galvanometer, such as a Thomson marine speaking-galvanometer, is included also in the circuit (a more sensitive one is sometimes necessary—and we frequently used a D’Arsonval—but it must be well damped), the meaning of these ticks is recognised; each represents a minute change in the resistance of the coherer—not at all the full change usually employed, but little subsidiary changes, sometimes up and sometimes down, barely sufficient to affect a galvanometer, but quite adequate (being so sudden) to disturb a telephone. This method of receiving, which at first is very sensitive, after a time becomes less so; the point shows signs of fatigue, probably due to too perfect cohesion having been gradually established, and a mechanical tap back is desirable to restore it to its original condition.
If all the signals received were precisely of the same strength, I doubt if these superposed crimples of resistance would occur; but signals depending on quality of sending spark never are of the same strength, and accordingly the sudden slight variations of resistance do occur. Usually an ordinary high resistance telephone was employed, and it was joined to the coherer circuit through one of the usual small transformers—a plan which has many obvious advantages.