5 Equivalent widths

Aslanov & Barannikov (1989) claim that the equivalent width variability of emission lines is of a stochastic nature, while the one associated with absorption lines exhibits a period twice that derived from their radial velocities. More recently, Underhill (1994) found no periodicity in the variability of the EWs.

In our data, two different categories of lines emerge: most of the lines appear rather constant but several others have clearly variable EWs. In the first category, Si IV $\lambda$4088, He II $\lambda \lambda$4200, 4542 and He I $\lambda$4143 exhibit maximum EW variations below $\sim$20% and the scatter is quite low (see Table 4). Other lines fall also into this category, e.g. the O II and C II emission lines, the unidentified Of emissions and the N III absorptions. In agreement with Underhill (1994), we do not find any significant periodicity when considering the variability of the EWs of the He II absorption lines.

Turning now to the lines showing variable EWs, we will focus here on the behaviour of those lines for which we have the best coverage with our data sets. The Si III $\lambda \lambda$4552, 4568 and C III $\lambda \lambda$4647-50 emissions seem to decrease gradually during our observing campaign. The EWs are reduced by a factor 2 between 1987 and 2000. The He I $\lambda$4713 line also presents EW variations, up to 50% from the mean. But the most impressive changes occur in H$\gamma$, He I $\lambda$4471 and H$\beta$. Their total equivalent widths (absorption + emission) increase gradually (see Table 5). In Fig. 7, one can clearly see this continuous trend to higher EWs, as the emission component weakens and finally disappears (see Sect. 6).

This separation of the lines in 2 categories, the rather constant ones (e.g. He II emission and absorptions) and the clearly variable ones (e.g. Si III and the most variable He I and H I lines), points most probably towards different formation regions for each category.

Figure 6: Our mean He II data folded with Barannikov's (1999) period of 1627.6 d, superimposed on his orbital solution. 1-$\sigma$ error bars are shown: when no error estimates are shown, it means that only one spectrum was available.

 

 

Table 4: Mean EW for different lines, averaged over the entire observing campaign. The quoted EWs for H$\beta$ and H$\gamma$ are the total ones, including the He II Pickering absorption (and the DIB in the case of H$\beta$). The last two columns summarize the results of a TVS analysis (see Sect. 6). The $\sqrt {TVS}$ is given in units of the continuum flux, while the last column provides the wavelength interval over which the detected variations are significant at the 99% level.

Line	EW(Å)	$\sqrt {TVS}$	width (TVS)
	(Å)		(Å)
Si IV$\lambda$ 4088	$0.196 \pm 0.02$1
He I$\lambda$ 4143	$0.040 \pm 0.006$
He II$\lambda$ 4200	$0.538 \pm 0.033$
H$\gamma$	$0.506 \pm 0.385$
He I$\lambda$ 4471	$0.643 \pm 0.215$	0.065	4462.3-4473.3
He II$\lambda$ 4542	$0.672 \pm 0.057$	0.017	4535.5-4542.6
Si III$\lambda$ 4552	$-0.141 \pm 0.045$	0.016	4549.9-4555.3
Si III$\lambda$ 4568	$-0.110 \pm 0.030$	0.010	4565.4-4569.4
N III$\lambda$ 4634	$-1.040 \pm 0.059$	0.014	4630.7-4637.0
N III$\lambda$ 4641		0.018	4637.0-4644.4
C III $\lambda \lambda$ 4647-50	$-0.306 \pm 0.077$	0.013	4644.4-4652.7
He II$\lambda$ 4686	$-0.524 \pm 0.045$	0.013	4709.2-4714.2
He I$\lambda$ 4713	$0.134 \pm 0.040$	0.092	4849.0-4866.5
H$\beta$	$1.457 \pm 0.788$

 

 

Table 5: Average EW for each observing run for He II $\lambda$4200, H$\gamma$, He I $\lambda$4471 and H$\beta$. Standard deviations are also quoted: when no error estimates are given, it means that only one spectrum was available.

Run	EW He II	EW H$\gamma$	EW He I	EW H$\beta$
Jul. 1986	$0.559 \pm 0.003$
Jul. 1987	$0.524 \pm 0.019$	$0.153 \pm 0.038$
Aug. 1987			0.120	-0.566
Aug. 1991		$0.367 \pm 0.034$	$0.235 \pm 0.019$
Oct. 1993	$0.548 \pm 0.051$	$0.577 \pm 0.075$
Aug. 1994	$0.518 \pm 0.012$	$0.635 \pm 0.017$
Aug. 1996	$0.557 \pm 0.049$	$0.781 \pm 0.017$	$0.417 \pm 0.017$	$0.610 \pm 0.066$
Jul. 1997	$0.562 \pm 0.033$	$0.830 \pm 0.045$	$0.542 \pm 0.006$	$0.745 \pm 0.018$
Sep. 1998			$0.622 \pm 0.029$
Nov. 1998			$0.696 \pm 0.039$
Jul. 1999	$0.539 \pm 0.018$	$1.221 \pm 0.055$	$0.694 \pm 0.021$	$1.267 \pm 0.070$
Aug. 1999	$0.587 \pm 0.057$	$1.232 \pm 0.043$	$0.743 \pm 0.017$	$1.155 \pm 0.045$
Sep. 2000			$0.881 \pm 0.036$	$2.229 \pm 0.090$
Mean	$0.538 \pm 0.033$	$0.506 \pm 0.385$	$0.643 \pm 0.215$	$1.457 \pm 0.788$

Figure 7: Average EW for each observing run: He II $\lambda$4200 is represented by filled triangles, H$\gamma$ by open squares, He I $\lambda$4471 by crosses and H$\beta$ by open triangles. The errors are very small, roughly the size of the symbols used in the figure, and they are listed in Table 5.