3 The Balmer emission

The general appearance of the Balmer emission lines in the quiescent phase 1987-89 has already been described (Kotnik-Karuza et al. 1992). $\rm H \alpha$ and $\rm H \beta$ are the most intense and easily measurable members of the series with line profiles varying in time. They are easily distinguished from the highest points of the pseudocontinuum. There are no remarkable absorption features to disturb the neighbourhood of the both Balmer lines. In order to estimate the influence of the cool star on the symbiotic spectrum a comparison with the spectra of 30 Her and $\beta$ And has been made. No Balmer lines in emission were observed in their spectra (Kotnik-Karuza & Jurdana-Sepic 1997). The smooth continuum in the $\rm H \beta$ region in 30 Her and $\rm H \alpha$ and $\rm H \beta$ Balmer lines in absorption in $\beta$ And suggest that these lines in the spectrum of CH Cyg are not augmented by the possible disturbance of the cool component.

  

Figure 2: Balmer emission line profiles normalized to the continuum level from June 87-Jan. 89

  

Figure 3: The intensity ratio V/R in the double-peaked Balmer emission lines $\rm H \alpha$ (circles) $\rm H \beta$ (squares) -our data, $\rm H \alpha$ (asterisk) -Leedjärv 1989

The time evolution of Balmer emission line profiles normalized to the local continuum is given in Fig. 2. Their predominantly expressed double-peaked structure with asymmetry in peak heights imply that these lines could originate either in a rotating accretion disc or are a result of superposition of an absorption feature on to a single emission component. In order to test the former assumption we measured the intensity ratio of the violet and red wing of $\rm H \alpha$ and $\rm H \beta$ and extrapolated these results to the previous measurements of $\rm H \alpha$ (Leedjärv 1989) as shown in Fig. 3. Our values of V/R of $\rm H \alpha$ in the observed period 1987-89 did not exceed 1, which would exclude an eclipse of the hot component by the cool giant. It is true that V/R of $\rm H \beta$ was slightly higher than 1 in January 1989. Also the ratio V/R of Leedjärv (1989) of $\rm H \alpha$ exceeds 1 in the period JD 244 6850-244 7050, i.e. at the begining of our time interval. Anyway, the discrepancies are not bothering since none of the known models for CH Cygni predict an eclipse at that time. Unfortunately, there are no observations in the $\rm H \alpha$ and $\rm H \beta$ spectral region just in the time when an eclipse could have happened according to the triple star model of Hinkle et al. (1993) and with the assumption of high orbital inclination. However, there are objections to this model (Munari et al. 1996), as well as to the possibility to take an eclipse as a probe for the existence of an accretion disc in binaries with the hot component being less massive than the cool one (Robinson et al. 1994). An eclipse in the far more reliable long period orbit could not have happened at that time (Mikolajewski & Mikolajewska 1988). The different intensity and time evolution of the central depression in $\rm H \alpha$ compared to that of $\rm H \beta$ argues in favour of the latter explanation, which means that the compact object, most likely a white dwarf, accretes matter via the M giant's wind resulting in an accretion complex being different from a substantial accretion disc. In this picture the lines arise in an ionized region of the red giant's wind around the hot component and are then self absorbed in the neutral regions of the same wind. Bode et al. (1991) fitted the $\rm H \alpha$ line profiles during the quiescent phase: out of 8 profiles only one taken in June 1986 lent itself readily to testing the accretion disc hypothesis. This suggests that the accretion disc can be considered as a transient phenomenon associated only with outbursts of the system.

The most intriguing phenomenon in the time evolution of both lines is the appearance of an one component asymmetrical emission profile without any noticeable absorption or additional emission in July 1988. After that, a gradual decrease in intensity of $\rm H \alpha$ and $\rm H \beta$ toward the end of 1988 was recorded. At the beginning of 1989 the double-peaked structure was re-established. We believe that these changes in line profiles reflect the variable rate of the mass flow via stellar wind onto the hot component, which influences the optical thickness of the gas along the line of sight. The enhanced mass transfer from the cool component could be considered as a probable sign of a renewed activity. The time coincidence of intensity decrease with approach to apoastron in the long period orbit cannot be taken as the only reason of the reduced mass transfer. If this were true, typical irregular appearance of activity phases of variable duration could not be explained.

The more negative radial velocities of the single emission and central absorption of the Balmer lines by about 10 km s^-1 with respect to the systemic velocity in our case suggest that the Balmer region is a slowly expanding shell of material ejected by the cool object.