The large discrepancy reported by Schmieder et al. in 1968 between their value for the electric quadrupole coupling constant of the 5 2P3/2 state of 39K and an earlier value measured by Ritter and Series in 1957 is investigated.

By means of an approximation method formulated for this purpose by Pegg in 1969, it is shown that the discrepancy can be attributed to the shift induced by the strong oscillating field used of necessity in the Ritter and Series experiment. The positions of the resonances can be expressed in terms of effective hyperfine coupling constants, where the electric quadrupole term is strongly dependent on the radio-frequency field strength, while the magnetic dipole term is independent of the field strength.

It is shown that the rotating-field approximation is not appropriate for the calculation of shifts in double-resonance experiments in zero magnetic field.