Mode density and frequency extraction in the δ Scuti star HD 50844

We consider the high mode density reported in the delta Scuti star HD 50844 observed by CoRoT. Using simulations, we find that extracting frequencies down to a given false alarm probability by means of successive prewhitening leads to a gross over-estimate of the number of frequencies in a star. This is due to blending of the peaks in the periodogram due to the finite duration of the time series. Prewhitening is equivalent to adding a frequency to the data which is carefully chosen to interfere destructively with a given frequency in the data. Since the frequency extracted from a blended peak is not quite correct, the interference is not destructive with the result that many additional fictitious frequencies are added to the data. In data with very high signal-to-noise, such as the CoRoT data, these spurious frequencies are highly significant. Continuous prewhitening thus causes a cascade of spurious frequencies which leads to a much larger estimate of the mode density than is actually the case. The results reported for HD 50844 are consistent with this effect. Direct comparison of the power in the raw periodogram in this star with that in delta Scuti stars observed by Kepler shows that HD 50844 has a typical mode density.