The number of electrons ejected by light engaged the attention of the early experimenters from rather a different aspect. The experiments were usually carried out with reflecting surfaces of the alkali metals, and an interesting dependence of the number of photoelectrons on the state of polarization of the light was observed. If the light was incident obliquely and was so polarized that the electric vector had a component perpendicular to the surface, the number of electrons ejected was found to be greater than if the electric vector lay wholly in the surface. This phenomenon was investigated in great detail by Pohl and Pring sheim, who found the variation of the emission with change of wave-length of the light when compared on the basis of equal amounts of incident energy. When the electric vector had a com ponent perpendicular to the surface they found a maximum oc curred at a certain wave-length, which was not present with the other type of polarization. This led them to effect a division of the whole effect into a normal and a selective effect, the latter showing the maximum.
The important bearing of this subject on the nature of light depends entirely on the surprising fact, expressed by Einstein's equation, that the velocity of the ejected electron is determined only by the frequency of the radiation, and is independent of the intensity of the radiation. This is entirely contrary to what we should expect from a wave motion, yet the phenomena of inter ference, and in fact the whole of ordinary optics, appear to show indubitably that radiation is a wave motion. This will be apparent if we imagine first a metal plate placed so close to an X-ray tube that the radiation is pouring on to it, and then removed so far away that the radiation is even difficult to detect. On the wave picture one would say that, in the second case, the electric vector in the radiation hitting the plate was very much weaker than in the first, and we would expect that not only would the number of electrons be less but also that the electric vector would be able to transfer much less energy to the electrons, and they would come out with much lower energies.
Whilst the number of electrons is certainly less in strict pro portionality to the intensity, the energy of the individual electrons will be the same in both cases. Weak or intense radiation, as long as the frequency remains constant, always gives the same energy to the electrons. This result is quite incompatible with our or dinary ideas of a wave motion, and the simplest way of expressing this is to notice how the alternative picture of light quanta ex plains the dilemma. If, instead of imagining the radiation to spread out from the source in waves, we imagine the energy to be emitted in bundles or quanta of magnitude hv then with in tense radiation there will admittedly be more quanta traversing a body, but even with weak radiation the few quanta which do arrive are still the same. They will naturally lead to the ejection of fewer electrons, but when a quantum is absorbed it always in volves the transference of the same amount of energy, and the photoelectron will have the same energy as if the density of quanta were far greater. Great difficulties are raised by any at tempt to explain interference by a light-quantum picture and the problem of reconciling these two different views has at present not been solved. A fuller discussion will be found in the articles
QUANTUM THEORY and ATOM, where it is pointed out that prob ably both views are correct. The point is that, if we insist on describing atomic processes in language and concepts used for large-scale phenomena, then inevitably a duality will appear. Radiation must sometimes be pictured as waves, sometimes as particles. How far-reaching is this duality may be seen in the fact that electrons, and other material particles also, behave some times in a manner only explicable on a wave basis.
The whole apparatus can truly be described as a photoelectric cell, since, when light enters the bulb, electrons will be ejected from the potassium, which will charge up positively, and will be collected by the central electrode rendering the latter negative. To measure light intensities precautions must be taken to collect the whole of the electron stream liberated, since it is the number of electrons which is directly proportional to the intensity of the light. The method of doing this is to apply a negative potential to the potassium surface as shown in fig. 4, the collecting elec trode being practically at the potential of the other end of the battery by connection through the measuring instrument. For considerable intensities of light this may be a sensitive galvanome ter, but in most cases some type of electrometer, such as an or dinary quadrant electrometer or the special quick period quadrant designed by Prof. Lindemann, is preferable. The most sensitive arrangement is to measure the rate at which the electrometer charges up, i.e., the rate at which the needle moves, but when it is possible the constant deflection method is far more convenient. It will be noted that while one terminal of the electrometer is connected directly to the collecting electrode, the other is con nected to earth, and the constant deflection method consists in connecting a very high resistance, of the order of io to io,000 megohms, across the electrometer. The latter then measures the potential drop across this resistance due to the current through the photoelectric cell. An important precaution is to have a guard ring (see fig. 3), consisting of a ring of metal or silvering, in the position shown, and to connect this to earth. This prevents charge creeping over the surface of the glass from the high po tential electrode to the measuring instrument. The general work ing of the cell is rendered far more regular by taking precautions to keep it in a dry atmosphere, as a film of moisture on the sur face of the glass renders it virtually a conductor.