August 5, 1920] 



NATURE 



717 



motion may be compared with that of a swarm of 

 bees in which each insect is flying hither and 

 thither rapidly, whilst the whole swarm is being 

 blown by a gentle breeze slowly down a road. If 

 the electrons merely surge to and fro, \i gives 

 rise to a form of current we describe as an 

 alternating current, and if they execute this 

 motion very rapidly we call it an electric oscilla- 

 tion. 



The reason an electric current produces heat in a 

 conductor is because the drift energy of the electrons 

 IS then being continually converted into additional 

 irregular-motion energy in the free electrons and 

 atoms by collisions of electrons with the atoms of the 

 conductor. If, then, the temperature becomes very 

 high— that is, if the irregular electronic motion 

 becomes very great— certain electrons may acquire 

 such velocities fhat they are flung out from the surface 

 of the wire even against the attraction of the positive 

 atomic ions left behind. If there is no electric force 

 tending to make the electrons move away from the 

 neighbourhood of the hot wire, these electrons con- 

 stitute a space charge around it, and the repulsion 

 they exercise on each other tends to keep other elec- 

 trons from getting out into the space. Suppose, how- 

 ever, that the incandescent wire is placed in the axis 

 of a highly exhausted glass tube, and is surrounded 

 by a metal cylinder which is kept positively electrified, 

 the electrons move to it, and others then' make their 

 exit from the wire. Such a tube with incandescent 

 wire cathode and cold metal plate anode is now called 

 a thermionic tube. The steady emission of electrons 

 is called a thermionic current. In the case of a 

 tungsten wire brilliantlv incandescent in vacuo and 

 under sufficient electric force, this current may 

 amount to as much as an ampere per square centi- 

 metre of surface. This means that electrons are being 

 flung or pulled out at the rate of millions of billions 

 per second per square centimetre. So soon as Sir Joseph 

 Thomson had proved by experiment that this elec- 

 tronic emission was taking place the explanation of 

 the effects observed in incandescent electric lamps 

 bv Edison, Preece, and myself became clear. For 

 in the Edison experiment we have a slow drift of 

 electrons through the carbon filament superimposed 

 on a very tapid and erratic motion, and multitudes 

 of these electrons are escaping from the filament on 

 all sides — just like steam escaping from a porous or 

 leaky canvas steam pipe. If the plate in the bulb 

 is connected to the positive pole of the filament- 

 heating battery, it is positively electrified and it 

 attracts these escaped electrons, and thev enter it and 

 drift through the external wire, forming the observed 

 Edison current. 



Suppose, then, that we connect the collecting plate 

 by a wire external to the bulb with the negative ter- 

 minal of the filament, and that we insert in this 

 circuit a battery of a number of cells which can be 

 altered so as to vary the potential of the plate, the 

 said battery havin^r its negative terminal connected 

 to the filament, we then find that a thermionic cur- 

 rent flows which can be measured by an amperemeter 

 inserted in the circuit. If we vary the voltage from 

 zero upwards we shall find that the thermionic cur- 

 rent increases, but not indefinitely. It soon reaches 

 a value, at which no further increase of voltage raises 

 the current. The reason the current does not increase 

 indefinitely is because for each particular tempera- 

 ture of the filament there is a certain maximum 

 possible rate of electronic emission. The electrons are 

 drawn awav from the filament at a rate which in- 

 creases with the potential of. the plate up to that 

 point nt which the maximum emission rate is reached. 

 NO. 2649, VOL. 105] 



The thermionic current then becomes stationary and 

 is said to be saturated. 



It is remarkable that although this emission of 

 electricity from incandescent substances had been 

 studied for more than a quarter of a century, none 

 of them made any practical application of it prior to 

 1904. At that date I was so fortunate as to discover 

 a totally unexpected application of this thermionic 

 emission in wireless telegraphy. Before 1904 only 

 three kinds of detector were in practical use in wire- 

 less telegraphy, viz. the coherer, or metallic filings 

 detector, the magnetic-wire detector, and the elec- 

 trolytic detector. The coherer and the electrolytic 

 detectors were both rather troublesome to work with 

 on account of the frequent adjustments required. 

 The magnetic detector was far more satisfactory, and 

 in the forrn given to it by Senator Marconi is still 

 used. It is not, however, very sensitive, and it 

 requires attention at frequent intervals to wind up the 

 clockwork which drives the moving iron-wire band. 



In or about 1904 many wireless telegraphists wer« 

 seeking for new and improved detectors. I was 

 anxious to find one which, while more sensitive and 

 less capricious than the coherer, could be used to 

 record the signals by optical means. Our electrical 

 instruments tor detecting feeble direct or unidirec- 

 tional currents are vastly more sensitive than any we 

 have for detecting alternating currents. Hence it 

 seemed to me that we should gain a great advantage 

 if we could convert the feeble alternating currents in 

 a wireless aerial into unidirectional currents which 

 could then affect a mirror galvanometer or the more 

 sensitive Einthoven galvanometer. There were 

 already in existence appliances for effecting this con- 

 version when the alternations or frequency was low, 

 namely, one hundred or a few hundred per second. 

 After trying numerous devices my old experiments on 

 the Edison effect came to mind, and the question 

 arose whether a lamp with incandescent filament and 

 metal collecting plate would not provide what was 

 required even for extra high frequency currents, in 

 virtue of the fact that the thermionic emission would 

 discharge the collecting plate instantly when posi- 

 tively, but not when negatively, electrified. Accord- 

 ingly I appealed to the arbitrament of experiment, and 

 the following arrangement was tried. 



Two coils of wire were placed at a distance, and 

 in one of them electric oscillations were created by 

 the discharge of a Leyden jar. The other coil had 

 one terminal connected to the filament of a lamp, and 

 the collecting plate to one terminal of a galvano- 

 meter, the second terminal of the latter being con- 

 nected to the second terminal of the coil. I found, 

 to my delight, that my anticipations were correct, 

 and that electric oscillations created in the second 

 coil by induction from the first were rectified or 

 converted into unidirectional gushes of electricity 

 which acted upon and deflected the galvanometer. 



I therefore * named such a lamp with collecting 

 metal plate used for the above purpose an oscillation 

 valve, because it acts towards electric currents as a 

 valve in a water-pipe acts towards a current of water. 

 I soon found that" for the purposes of wireless tele- 

 graphy quite a small low-voltage lamp with a metal 

 cylinder placed round a carbon or metal loop filament 

 was a verv effective rectifier, and could be used for 

 converting the feeble alternating currents in a wire- 

 less receiving aerial into unidirectional currents 

 capable of affectinf a telephone or galvanometer. It 

 was almost immediately adopted in practical wireless 

 telegraphy as a simple and ^asily managed detector, 

 and the intermittent rectified currents were passed 

 through a telephone. Some time after the introduc- 

 tion of this oscillation valve T found that another 



