THE SCIENTIFIC WORK OF C. J. DAVISSON 789 



times called the '^thermionic work-function" of the metal. If a given set of 

 data is fitted first by the T'^ law and then by the 7^ law, different values of 

 b and therefore different values of the thermionic work-function are ob- 

 tained. Which is right? 



This question can be answered if the thermionic work-function can be 

 measured with adequate accuracy by some other method. Such a method 

 exists: it is called the ''calorimetric" method. Suppose an incandescent wire 

 surrounded by a cylindrical electrode. If the latter is negative with respect 

 to the former, the emitted electrons will return to the wire, and there will 

 be no net thermal effect due to the emission. If, however, the cylinder is 

 positive with respect to the wire, the electrons will be drawn to it, and the 

 wire will fall in temperature: this is the ''cooling-effect" due to the emis- 

 sion. The resistance of the wire will decrease, and if the current into the 

 wire is held constant, the voltage between its terminals will be lessened. 



The experiment may sound easy, and so it might be if all of the current 

 flowed within the wire from end to end; but the bleeding of electrons through 

 the entire surface makes the current vary from point to point along the wire, 

 and complicates the test enormously. Others elsewhere had tackled this 

 difficult problem of experimentation; but Davisson and Germer found a 

 better way to handle it, and their results for tungsten were presented at a 

 meeting at the end of 1921 and published fully the following year. From 

 their data they calculated the thermionic work-function of the metal, which 

 when thus determined we may denote by e(f). It agreed with the value of 

 kb obtained from the newer form of "Richardson's law," disagreed with the 

 other. Thus Davisson was in the position of having confirmed the Fermi- 

 Dirac distribution-law before it had been stated! 



It remains to be said that, years later, Davisson and Germer repeated 

 this experiment upon an oxide-coated platinum wire. Here they came upon 

 a complication from which clean metal surfaces are fortunately exempt. 

 The character of the oxide-coated wire changed with the temperature; and, 

 since the measurement of the "constant" b requires a variation of the tem- 

 perature, its value did not provide a reliable measure of the work at any 

 single temperature, whereas the "calorimetric" measurement did. 



Now at last we are ready to attend to the early stages of the studies 

 which were destined to lead to the discovery of electron-waves. These were 

 studies of what I shall call the "polycrystalline scattering patterns" of 

 metals: the name is descriptive rather than short. A beam of electrons is 

 projected against a metal target which is in the condition, normal for a 

 metal, of being a complex of tiny crystals oriented in all directions. Some 

 of these electrons swing around and come back out of the metal with un- 

 diminished energy: these are the electrons that are "elastically scattered," 



f 



