EMISSION VELOCITIES OF PHOTO-ELECTRONS. 
215 
the ratio of the effect due to reflected light to the direct effect is 1'5 to 85. Consider 
what happens at the point A' where the curve cuts the axis (fig. 4). Only the 
shortest wave-length has any tendency to cause the plate to cliarge up positively, 
while it receives negative electricity due to the whole range of wave-lengths reflected 
to the sides. When using monochromatic light the latter effect would l)e much less. 
A zinc plate charged up to 2‘02 volts when illuminated by the unresolved light from 
the mercury arc. When the shortest wave-length X 1849 was isolated and directed 
to the plate it charged up to 2’06 volts. It is evident therefore tliat 2'OG volts was 
very nearly the true velocity with which the fastest electrons were emitted. 
From these considerations it seems quite justihahle to take the potential to which 
the disc rises as being the actual maximum velocity of emission. 
7. Experiments with MonocliromeUic Ligld. —In the following pages experiments 
are described from which it is concluded that the maximum energy of emission of 
photo-electrons, and not the maximum velocity, is proportional to the frequency of 
the incident light. It seemed the best course to work not with a large number of 
different wave-lengths, but with three selected wave-lengths, and having fixed on 
these, to determine the corresponding emission velocities with the liighest possible 
accuracy. Observations were sometimes made with other wave-lengths, 1)ut the 
proof rests largely on the experiments made with X 2537, X 2257, and X 1849. 
Nearly all the elements distilled in these experiments were pure. In most cases 
the distilled portion was tested chemically at the end of the experiment. 
Cadmium .—The total photo-electric currents from a surface of distilled cadmium 
corresponding to the lines X 2537, X 2257, and X 1849, were approximately proportional 
to 40, 20, and 1. The times taken for the disc to attain its final potential were in the 
inverse order. The rate at which the final potential was attained, when using X 1849, 
was so slow that it always required about an hour to determine the final potential, 
even when the system was initially charged to a value '1 volt below its final value. It 
was this excessive feebleness of the light tliat made the experiments difhcidt, even 
when the electrical conditions were at their best. 
In 'Table II. the experimental results are given. The velocities, corresponding to 
the three wave-lengths, were determined three times each in rotation. For reasons 
discussed in section 4, the lines X2967 and X 3340 were obtained by the transparency 
limits of mica and glass. In the fourth and fifth columns the theoretical values are 
given on the energy law and Ladenburg’s law respectively, taking the values for X 2537 
and X 1849 as standard. 
The velocity predicted by the energy law for X 2257 is in much better agreement 
with the experimental value than that predicted by Ladenburg’s law. The experi¬ 
mental values for X2967 and X3126 are lower than the theoretical values, on account 
of the earth’s magnetic field curling the paths of the photo-electrons so much that 
none of them impinge on the case normally. Though the experimental velocity for 
X 3126 is given as 0 in the table, a well-marked current was obtained with an 
