... as Einstein describes it at first in his 1917 paper, allows for such oscillations - to use again the later wave picture - which need more than one revolution in order to result in a closed wavetrain. They are thus characterized by two integer numbers - that of the periods and that of the revolutions. According to Einstein's sketch of the principle and to present-day terminology, this is a torus quantization, also termed Einstein-Brillouin-Keller (EBK) quantization.

At the same time, he works out the apparently small but decisive distinction to such electronic orbits which arbitrarily often and arbitrarily close return to every point, but never really close. This type of motion (which Keller does not consider at first) corresponds to a fundamental assumption of classical statistical mechanics (ergodic hypothesis), i.e. is a model view of the complex motion of classical many-particle systems (or of Bohr's atomic model with many electrons alike) - which are not comprised by the EBK quantization rule.