6 Periodicities in spectral line variations

Figure 14: Phase diagram of the radial velocity variations of the weak absorption lines, $P=2\hbox{$.\!\!^{\rm d}$ }7721$

Figure 15: Phase diagram of the radial velocity variations of the He I 5876Å narrow emission, $P=2\hbox{$.\!\!^{\rm d}$ }7705$

Figure 16: Power spectra: upper - EW of the narrow emission of He I 5876Å, centre - EW of the broad emission of He I 5876Å, lower - spectral window

Figure 17: Phase shifts between the radial velocities of the WALs, the radial velocities of the He I NELs and EWs of He I NELs, $P=2\hbox{$.\!\!^{\rm d}$ }7710$

Figure 18: Phase diagrams of the EW of the He I narrow emission (upper panel), the accretion components of Na I D1 (middle panel) and the veiling (lower panel), $P=2\hbox{$.\!\!^{\rm d}$ }7705$

Figure 19: Anti-correlation of the accretion components of Na I D1 and the EW of the broad emission of Fe II 5362Å

In Paper I, we have reported already the discovery of periodic variations in the photospheric and emission lines. Since then we have collected more spectra in 1999; the new spectra confirm that the period and the phase remain stable. All radial velocities were measured by cross-correlation with the template spectrum of $\gamma$ Cep (K1III-IV) in the regions 5550-5610Å and 6000-6050Å. The average radial velocity derived from the photospheric spectrum is +16kms^-1and is subtracted from all measured heliocentric velocities. The radial velocities of the WALs vary from -6 to +6kms^-1. A periodogram analysis indicates a period most likely in the range from $2\hbox{$.\!\!^{\rm d}$ }5$ to $2\hbox{$.\!\!^{\rm d}$ }9$. There is a number of almost equal, equidistant peaks, separated in frequency by 1/year. The strongest peak is at $P=2\hbox{$.\!\!^{\rm d}$ }7721$. A phase diagram with this period is shown in Fig. 14.

In addition to the WALs, we also analyzed the periodicities in other groups of spectral features: NELs, BELs and accretion components. We found that the WALs, NELs and accretion components all show about the same set of frequencies in the power spectrum, with a period in the range of $2\hbox{$.\!\!^{\rm d}$ }5$ to $2\hbox{$.\!\!^{\rm d}$ }9$. The BELs show a twice longer period in the range of 5 to 6 days. The period around $2\hbox{$.\!\!^{\rm d}$ }77$is also present in the variations of the BELs, but with less power than the 5-6 days period. This is illustrated in Fig. 16.

With our set of data we cannot prove that there is only one and the same period in the variations of the WALs, NELs and accretion components. For example, the power spectrum of the NELs shows that the most significant peaks are slightly shifted with respect to those of the WALs. The best period for the radial velocities of the narrow He I line is $P=2\hbox{$.\!\!^{\rm d}$ }7705$ (compared to $2\hbox{$.\!\!^{\rm d}$ }7721$ for the WALs). The corresponding phase diagram is shown in Fig. 15. Note the systematic shift in the mean radial velocity of the He I NEL by about +8kms^-1. The velocity amplitude is larger than that of the WALs.

The EW of the He I NEL varies from 0.15 to 0.60Å and shows the same periods as the radial velocity, with the strongest peak at $P=2\hbox{$.\!\!^{\rm d}$ }7705$. In order to check if the WALs and the He I emission vary in phase or with a phase shift, one must plot a phase diagram with a common period. A compromise period of $P=2\hbox{$.\!\!^{\rm d}$ }7710$was used for the diagram shown in Fig. 17. The result is that the radial velocity of the WALs and the He I NEL vary in anti-phase (phase shift 0.5), while the EW of the He I NEL shows a phase shift of about 0.25. Any other common period in the $2\hbox{$.\!\!^{\rm d}$ }5$ to $2\hbox{$.\!\!^{\rm d}$ }9$ interval shows the same phase shifts.

The He II 4686Å has only a narrow emission component, and varies in correlation with the narrow emission in He I 5876Å. However, the mean radial velocity is even more red-shifted, +20kms^-1.

The EWs of the accretion components vary in phase with the EW of the He I narrow emission. This is shown in Fig. 18 for the Na I D₁ red-shifted absorption between +200 and +400kms^-1. The stronger the He I NELs, the larger is the accretion absorption.

The EWs of the broad emissions of He I, Fe I, Fe II vary in correlation with each other. An anti-correlation exists between the variations of the EWs of the accretion components and the EWs of the broad emission lines of He I, Fe I, Fe II (Fig. 19). More about correlations between different spectral features is given in the next section, where the spectral line profiles formed in the accreting gas are discussed.

No periodicity was found in the veiling variations, and there is no obvious correlation between the veiling and any other spectral feature. Only a weak correlation exists between the veiling and the EW of the He I narrow emission. Conversion of EWs into fluxes does not improve the correlation.