29() BELL SYSTEM TECIISICAl. JOIRS.U. 



slreani flowing sp()iuancniisl\' out <it an intaiulrsri'ni wire-; \\\v\ are 

 three segments of one single cur\e, plotted on dilTerent scales as the 

 numerals show. This curve bends so gradually around towards 

 tangency with the axis of abscissae, that one can hardly avoid the 

 inference that it is really approaching that axis as if to an asymptote, 

 and that if the electrometer at any point ceases to declare a current, 

 it is because the electrometer is too insensitive to respond to the 

 smaller currents, and not because there are no faster electrons. Look 

 instead at the curves of Fig. 2, which refer to the electrons emerging 

 from an illuminated surface of sodium. These curves slant so sharjily 

 towards the axis of abscissae, they bend so slightly in the portions of 

 their courses where the data of experiment determine them, that the 

 linear extrapolation oxer ihe Utile interval into the axis commends 

 itself as natural and ine\ilable. Because the curves for the thermi- 

 onic electrons approach the axis so geiitK-. it is agreed that their 

 \elocities are distributed continuously o\er an unlimited range; be- 

 cause the curves for the photoelectrons cut into it so acuteK', it is felt 

 that their \elocities are confined below a definite maximum \alue. 



This therefore is the photoelectric efl'ect : waves of light inundate 

 the surface of a metal, and electrons pour out with \arious velocities, 

 some nearly attaining and none exceeding a particular topmost 

 \alue. I will designate this maximum speed, or rather the corre- 

 sponding maximum kinetic energy, by £max- Analyzing the process 

 in the classical manner, one must imagine the waves entering into the 

 metal and setting the indwelling electrons into forced oscillations; 

 the oscillations grow steadily wider; the speed with which the electron 

 dashes through its middle position grows larger and larger, and at 

 last it is torn from its moorings and forces its way through the surface 

 of the metal. Some of the energy it absorbed during the oscillations 

 is spent (converted into potential energ>') during the escape; the 

 rest is the kinetic energy with which it flies away. Even if the electron 

 were free within the metal and could oscillate in response to the 

 waves, unrestrained by any restoring force, it wdiild still ha\e to 

 spend some of its ac(|iiired energy in passing out through the l)oimdar\- 

 of the metal (the laws of thermionic emission furnish evidence enough 

 for this). It is natural to infer that £,„ax 's the energv' absorbed by 

 an electron originally free, minus this amount (let me call it P) which 

 it must sacrifice in crossing the frontier; the electrons which emerge 

 with energies lower than /Jnmx nM\\ be supposed to ha\e made the same 

 sacrifice at the frontier and others in addition, whether in tearing 

 themselves away from an additional restraint or in colliding with 

 atoms during their emigration. This is not the only conceivable 



