7 The frequency f$_\mathsf{2}$ = 0.46 c d $^\mathsf{-1}$

Figure 18: Greyscale plots of all spectra taken before HJD 2451150 for the He I 6678 lines, normalized by its mean profile, folded in phase using the frequency f2 = 0.46 c d-1. The velocity range corresponds to the one determined by the variance in Sect. 5.

In spite of the perturbation caused by the frequency f₁ (travelling patterns on a timescale of about half the period), the frequency f₂ can be distinguished in the strong He I lines displaying emission in their outer wings such as He I 5876 and 6678 (Fig. 18). It is also suspected in other He I lines weakly affected by emission such as He I 4713 and 4921, but does not appear in the purely photospheric lines.

Because the frequency f₂ is weak, the phase diagrams are not as clear, but it seems that the phase is constant. However, the power is strong in the emission wings of the He I lines and the phase is coherent in these regions. Because f₂ does not appear in the purely photospheric lines but rather in the emission wings of the He I lines, f₂ is more likely to be associated with orbiting material.

Looking at the V/R data as a function of time (Fig. 13), we see that the frequency f₂ is not always as strong: it is very strong during the first week of the campaign, then it suddenly fades on a timescale of 2 days between HJD 2451148 and 2451150. A summed power spectrum of the He I 6678 line for three sets of data (HJD between 2451139 and 2451144, 2451144 and 2451150, 2451150 and 2451161) is shown in Fig. 19. As the datasets are smaller, the accuracy of the frequencies is less good, but one can see that the frequency close to 0.5 c d^-1 disappears in the last dataset. This suggests that a cloud of material ejected in an outburst orbits around the star during a few periods. Transient periods with similar explanations have been observed in other Be stars (Stefl et al. 2000).

Figure 19: Summed power obtained with RLC method for the He I 6678 line for 3 datasets: HJD between 2451139 and 2451144, 2451144 and 2451150, 2451150 and 2451161. The peaks are smoothed by a Gaussian filter. The accuracy of the frequency determinations for these datasets (between 0.1 and 0.2 c d-1) is less good than for the whole MuSiCoS 98 dataset.

Figure 20: Ratio of the Violet over Red emission peaks of the He I 5876, 6678 and 7065 lines, showing emission wings, folded in phase with the frequency f2 = 0.46 c d-1. Best sine fits are overplotted. See symbol caption in Fig. 12.

In Fig. 20, we see for the He I 5876, 6678 and 7065 lines that the V/R ratio varies in phase with the frequency f₂. Dividing the data into 2 parts: the spectra before and after HJD 2451150, clearly shows the difference in strength for this frequency. As an example the He I 6678 line is shown in Fig. 21 with best sine fits. Looking at the ESO data (triangles) in particular assures that this is not due to site-to-site effects.

Figure 21: Ratio of the Violet over Red emission peaks of the He I 6678 line folded in phase with the frequency f2 = 0.46 c d-1, before and after HJD 2451150. Best fit sinusoids are overplotted as solid lines. The best-fit curve from the top panel is also reproduced as a dotted line in the lower panel. See symbol caption in Fig. 12.