Classical justification of the Franck-Hertz experiment