5 Discussion

Our observations of CH Cyg suggest that the phenomena recorded throughout its evolution phases are due to interaction between its cool and hot component well represented in the binary model with a 5700d period orbit. The complexity of spectroscopic phenomena and their changes in time reflect complex and variable conditions in both stellar atmospheres leading to the conclusion that even in the inactive phase the star has not been quiescent.

The shape and time evolution of the Balmer emission profiles do not suggest the existence of an accretion disk during the inactive state. Contrary to this, in outbursts a strong evidence of an accretion disk has been given. It means that the accretion disk in CH Cygni can be regarded as a transient phenomenon which disappears and reappears in different phases of star's evolution. This belief has been justified by simulations of Theuns & Jorissen (1993) which, apart from some discrepancies, proved to be applicable to CH Cygni. They, namely, discuss the formation of a disk in an unperturbed binary system where the primary does not fill its Roche lobe and is subject to a spherically symmetrical stellar wind around it. In the case of higher wind speeds, the gas is partly accreted by the secondary which is orbiting in the bubble of gas blown by the primary. The crucial question regarding the formation of an accretion disk concerns the amount of matter that can be accreted by the secondary.

The minimal value of the variable self-reversal in the double-peaked $\rm H \alpha$ and $\rm H \beta$ profiles observed with better resolution (Bode et al. 1991) and recognized as single line emission in our spectra would correspond to a reduced mass transfer from the red giant's atmosphere and to minimal optical thickness of the envelope around the hot companion. The assumption of an optically thick accretion disk in the Horne & Marsh model is obviously unsuitable for the quiescent period and responsible for the mismatch of the $\rm H \alpha$ and $\rm H \beta$ profiles taken by Bode et al. (1991) later than June 1986 with the accretion disk hypothesis (Robinson et al. 1994).

If the increase of optical thickness of the accreted matter toward the end of the observed period i.e. January 1989 were a result of an enhanced mass transfer rate, this could not have been caused by effects related to the eccentric orbit since the star was at that time far from periastron in the long period orbit (Mikolajewski et al. 1988). Variations of the mass transfer rate cannot be explained either by eccentricity of the short period orbit in the triple star model (Hinkle et al. 1993). Namely, the optical thickness of the envelope reached its minimum in July 1988, about 40 days after periastron i.e. shifted in phase only by 0.06.

On the other hand, the explanation of the gradual decrease of the excitation and effective temperature in the M giant's photosphere without influence of irradiation by the hot star in the detached binary system remains questionable.

According to the oblique rotator model in the inactive phase there is a lack of a noticeable accretion complex around the magnetic white dwarf (Tomov et al. 1996). Consequently an envelope of variable optical thickness cannot be explained by this model. Still, the main objection to the oblique rotator model follows from its prediction that this inactive phase of CH Cyg should last for several decades, necessary to accumulate enough material around the white dwarf to start a new propeller and accretor phase (Skopal et al. 1996b). The observations of a new outburst which announced itself already in 1990 contradict this prediction.

Because of the discrepancies of our results with the triple star and oblique rotator models we are tempted to propose a model acceptable at least for the observed inactive period 1987-1989. Since the origin of Balmer emission in an accretion disk seems to be improbable, we looked for it in a rotating envelope consisting of material supplied by the stellar wind from the cool component and being ionized by UV radiation emitted by the hot star. These mechanisms determine the physical conditions in the HII region, whereas a wind from the hot component as well as violent mass ejecting outbursts appear unlikely (Munari & Patat 1993). The HI region, which is influenced only by the red giant wind is responsible for the shape of the central reversal. Evident similarity of the peak-to-peak separation in the double peaked $\rm H \alpha$ profiles during the time evolution lead us to an attempt to establish a relationship beetwen these profiles and the shape of the rotating envelope. We adopted the simplest way already accepted for the description of the accretion disks (Robinson et al. 1994; Theuns & Jorissen 1993), according to which the particles in the envelope follow Keplerian trajectories, which represent curves of constant radial velocities. The double peaked emission originating from the physically longest constant radial velocity curve can be used to calculate the outer radius of the envelope. Similarly, the limits of the profile wings determine the corresponding inner radius.

  

Table 2: Inner and outer Keplerian radius of CH Cygni calculated from the $\rm H \alpha$ profiles

We measured peak-to-peak and wing limits separation in the $\rm H \alpha$ emission profiles at different epochs and, using our data together with those of Bode et al. (1991), calculated the corresponding inner and outer radii of the Keplerian accretion complex. These results together with the data of

  

Figure 5: Evolution of the inner (black symbols) and outer (white symbols) Keplerian radius in time derived in meters according to: our results (circles), Bode et al. (1991). (triangles) and Leedjärv et al. (1994). (squares)

Leedjärv et al. (1994) are given in Table 2 and Fig. 5. For the inclination of the system and the mass of the compact star we assumed the values used by Robinson et al. (1994): $i=78^\circ, M_{\rm C.S.}=1~M_{\hbox{$\odot$}}$. Nearly constant values of the inner and outer radii starting from June 1986 have been consistent with the inactivity of the hot component, though a slight increase of the inner radius in 1988 could possibly indicate a start of a process by which the infalling matter is expelled from the white dwarf.

In addition to the described influence of the hot and cool component on the shape of Balmer lines, their profiles can be affected by orbital phase, especially in cases with high inclination like CH Cygni. This dependence is expected to become larger in the case with increasing asymmetry of the Balmer region. No significant difference in the shape of the double peaked $\rm H \alpha$ profiles in the course of time, particularly in the depth of their central reversals suggests ultimately a spherical shell shape of the emitting region. This fits the model of the Strömgren sphere (Taylor & Seaquist 1984) which contains the HII region. The actual shape of this ionized envelope has been the subject of observational and theoretical research. The radio emission proved to be an important source of information since it originates exactly in the ionized portion of the red giant's wind. A model that accounts for the radio properties of the symbiotic stars (Seaquist et al. 1984) has further been developed by Taylor & Seaquist (1984) and is consistent with parameters from optical and IR data. It proves the shape of the ionization front to be determined only by the physical properties of the binary pair (mass loss rate from the cool giant, its wind velocity, separation of the components, luminosity of the ionized radiation). In the case of CH Cygni an ionization bounded region would be of an approximately spherical shape. Radio emission of CH Cygni was detected in quiescence and activity (Skopal et al. 1996b). The authors present a good fit of their radio observations during the latest outburst to the spherical shell model and even give the values of inner and outer shell radii.

Our idea of what is happening in CH Cygni in its transition between activity and inactivity, particularly during the apparently quiescent state, has been firmly supported by the already quoted calculations of Theuns & Jorissen (1993). Their three dimensional model of the flow pattern takes binary rotation and the same order of magnitude of the the wind and orbital velocity into account. The pattern differs substantially for the two values of polytropic index $\gamma$, reflecting an isothermic process for $\gamma =1$ and adiabatic for the assumed $\gamma =1.5$.

In both processes, adiabatic and isothermal, the gas that comes directly from the primary is compressed by the gravitational force of the accreting star toward the orbital plane. In the adiabatic case this compression heats up the gas, the pressure increases so much that the gas stream expands vertically and is not confined in a disc. Instead, it gives rise to a vortex structure in a plane perpendicular to the orbital plane.

In isothermal case the temperature and pressure are not high enough to dissipate the gas from the orbital plane and consequently, a thin, nearly Keplerian disk has been settled.

The transition from the isothermal case $\gamma =1$ with an accretion disk to the adiabatic case $\gamma =1.5$ without a disk reflects in this picture a variable thickness of the disk corresponding to $\gamma$ varying continuously between the given values.

Since $\gamma$ has been determined by the efficiency of cooling, this process will in its turn determine the formation of a disk. As an increase of cooling corresponds to an increase of density, and density is a function of mass loss rate, binary separation and the ratio of wind speed to orbital velocity, the probability of disk formation is expected to increase with an increased mass loss rate, decreased binary separation and with the wind speed of the same order of magnitude as the orbital velocity.

The disappearance of accretion disc in the transition to an inactive state is correlated with an increase of polytropic index $\gamma$ above 1 which corresponds to a decreased efficiency of cooling. The decrease of density associated with a decrease of optical thickness leads to reduced self-absorption in the outlying part of the accretion complex because of the smaller number density of absorbers along the line of sight. The double peaked line profiles with V/R<1 can be explained by absorption in the wind material whose radial velocity is directed predominantly toward the observer. A single line profile means that this process has achieved its minimum. It is the result of a decreased wind mass loss from the cool giant which is correlated with wind speed and binary separation. Since the physical properties of the cool giant determine how this material is lost and to what extent it interacts with the hot component, it means that the cool component ultimately determines the observable characteristics of this symbiotic star. The transient nature of the accretion disc in the detached CH Cyg system, with separation of components of about $1700~ R_{\hbox{$\odot$}}$ (Skopal 1995) is correlated with the variable degree of filling the red giant's tidal lobe (Iben & Tutukov 1996), which is significantly being remarkably underfilled in the inactive phase.

The question how to reconcile the complicated three dimensional transient structure around the hot component with the spherical shell model during the inactive state of the star remains open for further discussions.