eject the electron completely from the normal state and thus make the atom into a positive ion, the Bohr theory gives a value of the same order of magnitude as the so-called “ionization potentials” which have been found by experiment for various gases. An exact correspondence between theory and experiment could for hydrogen not be attained with certainty, because the hydrogen atoms in hydrogen gas under ordinary conditions always appear united in molecules. In his very first paper, however, Bohr had studied Balmer’s formula also from another point of view, and had derived in this way an expression for the Rydberg constant K which agreed with experiment. These considerations have reference to the above-mentioned connection of the theory with the classical theory of electrodynamics. Such a connection had previously been known to exist in the fact that, for long wave-lengths, the radiation formula of Planck reduces practically to the Rayleigh Jeans Law which can be derived from electrodynamics. This is related to the fact that when ν is small (long wave-lengths), the energy quantum _h_ν is very small, and hence the character of the radiation emitted will approach more and more nearly to a continuous “unquantized” radiation. One might then expect that the Bohr theory also should lead in the limit of long wave-lengths and small frequencies to results resembling those of the ordinary electrodynamic theory of the radiation process. On the Bohr theory we get the long wave-lengths for transitions between two stationary states of high numbers (numbers which also differ little from each other). Thus suppose _n_ is a very large number. Then the transition from the orbit _n_ to the orbit _n_ - 1 will give rise to radiation of great wave-length. For in this case Aₙ and Aₙ₋₁ differ very little, and accordingly _h_ν is very small, as must ν be also. According to the electrodynamic theory of radiation, the revolving electron should emit radiation whose frequency is equal to the electron’s frequency of revolution. According to the Bohr theory it is impossible to fulfil this condition exactly, since radiation results from a transition between two stationary orbits in each of which the electron has a distinct revolutional frequency. But if _n_ is a large number, the difference between the frequencies of revolution ωₙ and ωₙ₋₁ for the two orbits _n_ and _n_ - 1, respectively, becomes very small; for example, for _n_ = 100, it is only 3 per cent. For a certain high value of _n_, then, the frequency of the emitted radiation can therefore be _approximately_ equal to the frequency of revolution of the electron in both the two orbits, between which the transition takes place. But even if this proved correct for values of n about 100, one could not be sure beforehand whether it would work out right for still larger values of _n_, for example, 1000.