Dominant frequency (DF) computed from action potentials is a key parameter for investigating atrial fibrillation in animal studies and computer models. A recent clinical trial reported consistent results computing DF from 30 Hz to 400 Hz bandpass filtered bipolar electrograms in humans. The DF (< 15 Hz and, thus, filtered out) was recovered by rectifying the signal, while the theoretical background of this approach was left uncommented. It is the focus of this paper to provide this background by a Fourier analysis. We demonstrate that it is mainly the timing of the narrow deflections (local activation at the catheter tip) which contribute to the DF peak in the frequency spectrum. Due to the typical signal morphology pronounced harmonic peaks occur in the spectrum. This is a disadvantage when computing the regularity index (RI) as a parameter for local organization and signal quality. It is demonstrated for synthetical and patient data that at low DF the RI is far below the optimal value one even for high underlying organization and good signal quality. The insight obtained promotes the development of better measures for organization. The finding that mainly timing of activation contributes to DF might promote the development of powerful realtime signal processing tools for computing DF.