next up previous
Next: Profile variations Up: Results Previous: Results

Radial velocities

The radial velocity curves were computed using a gaussian fit on the line. However, because many lines are blended (Fig. [*]), especially the CaiiK line which is very broad (FWHM$\simeq$8Å), we computed the gaussian fit only for the core of the line (the 5% deepest part of each line).


  
Figure: Averaged spectrum, over one pulsation period. Are indicated the CaiiK line, the two Fei lines and the Ali line. The horizontal bar shows the domain which is represented in Figs.3 and 4
\begin{figure}
\resizebox {\hsize}{!}{\includegraphics{fig1.ps}}\end{figure}

Then, due to the well-marked sinusoidal shape of the velocity curves (Fig.[*]), a sine-fit was done, providing the mean-velocity $\gamma$ and the projected semi-amplitude K. Table[*] summarises the results obtained for the best 4 lines.


 
Table: Mean-velocity $\gamma$and semi-amplitude K together with their respective uncertainty [km.s-1] for the four measured absorption lines.
line $\gamma$ K
$\lambda\lambda$ 3933 CaiiK $51.0 \pm 0.2$ $4.7 \pm 0.3$
$\lambda\lambda$ 3944 Ali $48.0 \pm 0.1$ $4.8 \pm 0.2$
$\lambda\lambda$ 3920 Fei $48.0 \pm 0.1$ $4.8 \pm 0.3$
$\lambda\lambda$ 3922 Fei $48.3 \pm 0.1$ $4.7 \pm 0.3$

The $\gamma$-velocity is close to 48km.s-1 for two lines ($\lambda\lambda$ 3944 Ali and $\lambda\lambda$ 3920 Fei). For the $\lambda\lambda$ 3922 Fei line, the velocity is slightly larger, but this may be due to the uncertainty on the laboratory wavelength (Mathias [1994]) since the difference is very small, corresponding to about 4mÅ. On the other hand, the $\gamma$-velocity associated to the CaiiK line is appreciably larger by about 3km.s-1. It may be due to a blend, but the synthetic spectrum computed with a Kurucz's model does not show any significant lines close to the core of the CaiiK line. It should also be noted that the $\gamma$-velocity is slightly larger than the one obtained in PaperI by about 1km.s-1, which was already globally larger than previous values reported in litterature. This could be due to an additionnal velocity related to a binary system, as suspected by HIPPARCOS.

Concerning the pulsation amplitude, the value is foun- ded to be 14% larger than the values obtained in PaperI, which can be attributed to a multiperiodicity effect. Note that the different values are compatible with each other, which implies that the projection factor (set to 1 for the computation of the rest frame velocity) is similar for the 4 lines.


  
Figure: Stellar rest frame velocity (RFV) associated to $\lambda\lambda$3933CaiiK (circle), $\lambda\lambda$3944Ali (dot), $\lambda\lambda$3920Fei (triangle) and $\lambda\lambda$3922Fei (cross). Note that we shifted the curve associated to the CaiiK by about 3km.s-1 in order to have the same zero-velocity. The uncertainty is less than 0.5km.s-1
\begin{figure}
\resizebox {\hsize}{!}{\includegraphics{fig2.ps}}\end{figure}

Fig.[*] represents their radial velocity in the stellar rest frame. The four velocity curves have the same behavior, very close to a sine-wave. This is a completely different of the one obtained by DLS where a bump is present around the pulsation phase $\varphi \sim 0.55$, and where the CaiiK line velocity curve differ from the other curves at two phases: $\varphi \sim 0.2$ and $\varphi \sim 1.4$ by about 4km.s-1. In our data, no small variations larger than 0.5km.s-1 are visible. Thus, contrary to DLS, we conclude that the radial velocity variation of the CaiiK line is identical to that of the other considered metallic lines.


next up previous
Next: Profile variations Up: Results Previous: Results

8/13/1998