The energy distribution in the optical may be best understood as a combination of an M4.5III + B5-A0 star. Zhekov et al. (1996) have shown that the ``luminosity class" of the hot source in this binary changes. During the state of high ejection activity (January-April 1990, see below) it resembled a supergiant while since late 1990 it is more like a main sequence star.
The line spectrum of MWC560 can be also considered as a combination of two different components: i) absorption line systems with different radial velocity (RV), and ii) emission spectrum with almost constant look (Kolev & Tomov 1993). The behaviour of both types of spectra is quite different.
The Balmer lines dominate the absorption spectrum of MWC560. A complete
view of the MWC560 spectrum is published earlier (Kolev & Tomov
1993). In Fig. 1 (click here) we present the evolution of 
using all the plate
spectra obtained in the period 1990-1993. 
was chosen
because it is relatively free of strong blending. The panels of the
figure are arranged according to the annual observing seasons of MWC560\
but in fact this arrangement follows also the different behaviour of
the absorption spectrum.
The distinctions between the absorption components of 
in the
different observing seasons are clearly visible in Fig. 1 (click here). The
intensity and radial velocity variations of these components during
the seasons after April 1990 are less appreciable in comparison with
the changes during January-April 1990. The observed behaviour of the
strong shifted Balmer absorption components in the whole period
1990-1993 is in a good agreement with the supposed discrete and
quasi-stationary regimes of matter ejection (Tomov et al.
1992; Kolev & Tomov 1993).
The spectral observations of MWC560 since 1990 give us reasons to
assume that in addition to the high-velocity matter ejection a
continuous low-velocity mass-loss takes place as well. On all
spectra, with the exception of these obtained in 1990/1991, a weaker,
relatively wide absorption component with a velocity of about
- 280kms
is present in all Balmer lines (Figs. 1 (click here)-3).
The Balmer absorption components show a very complicated picture and for the sake of clarity the nomenclature of the various absorptions is given in an additional figure inserted in the rightmost panel in Fig. 1 (click here). The strong shifted Balmer absorptions (SSBA) are connected with the discrete and quasi-stationary matter ejection, while the slowest Balmer absorptions (SBA) indicate the continuous low-velocity mass-loss.
Other absorption lines that are permanently present or appear in different moments in the spectrum of MWC560 are the components of CaII K, HeI and some ions and neutral metals (mainly Fe and Ti). Figures 2 (click here) and 3 (click here) show the behaviour of these lines on two typical spectra obtained in each season. The components of CaII K have the same RVs and shapes as those of Balmer lines. The K-line was weakest during January-April 1990 and in the first half of the 1991/1992 season. During the rest of the time these absorptions were very strong, often comparable with the Balmer ones (Fig. 2 (click here)). There are two HeI lines - 4471Å and 4026Å, confidently present in most of the spectra of MWC560. The absorption components of these lines are usually very weak. Their violet-shifts are of the same order as these of the hydrogen and CaII absorptions. In Fig. 3 (click here) the position of the weak HeI4471Å absorption is indicated by an arrow in the spectra in which this component is present.
Figure 3: The spectral region around 
in the spectrum of
MWC560. Each observing season is represented by the same two
spectra used in Fig. 2. The weak HeI absorptions are
marked by thin arrows and the MgII absorption is
marked by thick arrows
A noticeable absorption of MgII4481Å appeared in the spectrum
of MWC560 in 1990/1991 only (thick arrows in Fig. 3 (click here)). This absorption
shows considerably small violet-shift. We measured an average radial
velocity - 0.9
2.8kms
for this line.
The emission spectrum of MWC560 is dominated by the lines of the singly ionized metals, mainly FeII and TiII (cf. Kolev & Tomov 1993). The singly ionized metals show strong absorption components during the observing season 1990/1991 only (Figs. 2 (click here), 3 (click here)). These components are equally violet-shifted as the SSBA. The strongest one is a TiII absorption blend at 3760Å which exceeds in intensity the nearest Balmer lines H11 and H12. This TiII blend is the only absorption (in addition to the hydrogen, CaII and HeI ones) which is intensive enough in the spectra obtained in 1992/1993.
Figure 3 (click here) shows an example of the emission spectrum of MWC560. The same
lines are present in the MWC560 spectrum all the time being very sharp
and positional stable. We chose the emission lines FeII4515Å\
and TiII4590Å, which do not show remarkable absorption
components, to make a rough estimate of their variations. The
equivalent width 
and the peak intensity 
in
continuum units are plotted in Fig. 4 (click here). One can note a small decrease
of 
(FeII) in 1990/1991 which may be explained by influence
of a weak absorption component. It is possible also to note a small
gradual increase of 
and 
for both lines but the
large data scatter prevents more definite conclusions.
Figure 4: Equivalent widths (Top) and peak intensities (
Bottom) of the FeII4515Å and TiII4590Å\
emission lines
Comparing the profile shapes and the behaviour of the Balmer lines in the period under discussion, it is obvious that they are far from the classical P Cyg-type (Figs. 1 (click here)-3). Their stationary emission components are much narrower and completely detached from the absorption ones. The observed conspicuous variations of the SSBA do not reflect in any way on the emission components. In particular, during the discrete ejections, it is clearly visible that the appearance and disappearance of the SSBA as well as the variations in their intensities, shapes and velocities are not connected with changes in the position, shape and intensity of the emission components (Fig. 1 (click here)). This indicates that the emission and strong absorption components of the Balmer lines in the spectrum of MWC560\ originate in different regions with different physical conditions.
The simplest way to explain the strong shifted absorptions is a highly collimated jet which has a significant covering factor to obscure the central source (Tomov et al. 1990, 1992; Shore et al. 1994). An important question is where the emission lines originate - around the hot component or in the M giant atmosphere? We suggested that the emission lines originate in the outer parts of an accretion disk, supposing a face-on geometry for the MWC560 system. The same was argued recently by Shore et al. (1994).
Near infrared photometry shows that the cool component behaves like a
normal M4-5 giant (cf. Buckley 1992; Zhekov et al. 1996). On
the other hand, if the emissions are formed in the M giant atmosphere
they must be superimposed on the composite spectrum. It is evident in
Fig. 3 (click here) that when the metal emissions are blended with the strong
absorption they are weakened. The emission components of
(Fig. 1 (click here)) and 
(Fig. 3 (click here)) in 1990/1991 are affected in
the same way by the red wings of their absorptions. This shows that
the outflowing gas in which the SSBA originate is projected on the
emission lines formation region and can absorb a part of its
radiation. Therefore, the emissions most probably arise around the
hot component and not in the M giant's atmosphere.
The hydrogen and ionized metals profiles in 1990/1991 closely resemble the P Cyg-type only because of the smaller ejection velocities.
We examined the changes in the intensities and velocities of the SBA
component in the period 1990-1993 with the exception of 1990/1991
observing season. Its equivalent widths (
) and radial
velocities are listed in Table 1 and examples of the variations of the
and RV are given in Fig. 5 (click here). The changes in the
of this absorption are not great during each season as
well as during all the time since 1990. The first members of the
Balmer series (
) show more negative RV in
comparison to the higher members (Fig. 5 (click here)). This difference is
probably caused by the blending with the Balmer emissions. The lines
after 
do not show emission components and the SBA
velocities are practically constant.
On our photographic spectra we did not find, even in 
,
(Fig. 3 (click here)) and 
(Fig. 1 (click here)), emission wings as wide as
the SBA. We suppose that the SBA components which are permanently
present in the spectrum of MWC560 arise in a persistent mass-outflow
with a velocity of about 280kms
. The lack of real P Cyg profiles
probably indicates that this outflow is not spherically symmetric.
Figure 5: Examples of the radial velocities (Bottom) and
(Top) of the SBA components in the
spectrum of MWC560
Table 2 presents the average heliocentric RV of the main absorption and emission features. Some of the spectra were measured independently more than once.
The Balmer emission components show more positive and largely scattered RV in comparison to the metallic ones (Table 2). This difference is not caused by a physical reason but it is a result of an influence of the red wing of the SBA component.
The RVs of the metallic emission lines are plotted in Fig. 6 (click here), where
the enlarged symbols signify the spectra being measured more than
once. No systematic changes can be noted during the four years'
monitoring of MWC560. The scattering of the points around the mean
radial velocity +35.5 kms
(Fig. 6 (click here)) is rather of the order of the
measurement errors. The periodogram analysis of the data, performed
using the Lafler & Kinman (1965); Deeming (1975);
Stellingwerf (1978) and Schwarzenberg-Czerny (1989)
methods, does not show any significant period in the range of 100 - 2000 days.
Figure 6: Radial velocities of the metallic emission lines in the
spectrum of MWC560. The long dashed line and the
short dashed lines mark the mean velocity of +35.5 kms
\
and the 
limits respectively
Chentsov (1994, private communication) has made high-accuracy RV
measurements of the emission and absorption lines on two CCD spectra
in the near infrared region obtained in 1992 and 1993. These
measurements include the emission lines of the same single ionized
metals, as the ones observed in the optical region and the absorption
lines of the neutral metals surely belonging to the M giant spectrum.
They show practically identical RVs with a mean value of +35.8 kms
.
The lack of noticeable variations in the RVs of the single ionized metals emissions and the coincidence of these velocities with those of the M giant absorptions do not contradict to a face-on location of the MWC560 system orbital plane.
The SSBA radial velocities, as well as these of CaII K show a picture
(Fig. 7 (click here)) in very good agreement with the two regimes of matter
ejection. During the discrete ejections period (January-April
1990) the strong shifted absorption RVs change with amplitude more
than 4000 kms
. These velocities cover an interval from - 1000 kms
\
to about - 6000 kms
around the star brightness maximum (cf.
Paper I) - the highest value ever observed. The RVs of the SSBA
change in very short time-scales. Even on spectra obtained in
two-three days interval, if the SSBA are present, their violet shifts
may differ by several thousands of kms
(Figs. 1 (click here) and 7 (click here), Table 2).
In the next observing seasons, when the matter ejection is
quasi-stationary, the RVs of SSBA show completely different
behaviour. The changes within each season are much smaller in
comparison to January-April 1990. The RVs observed in 1990/1991
remarkably differ from all the rest and are from - 130 to 320 kms
.
While, during the next two observing seasons (1991/1992 and 1992/1993)
the radial velocities range from - 1200 to 2000 kms
(Fig. 7 (click here),
Table 2).
As it was shown earlier (Tomov et al. 1994), the integrated ultraviolet flux changes in the same way as the RVs of the SSBA during the different observing seasons.
Figure 7: Radial velocities of the SSBA components and CaIIK
strong-shifted absorptions in the spectrum of MWC560. The
measurement errors are not shown here because they are very
small in comparison to the RV values (cf. Table 2)
The equivalent widths 
of the most intensive SSBA in all
spectra are presented in Table 3. The average values for the time
intervals distinguished by different regimes of matter ejection are
plotted in Fig. 8 (click here) in respect to the Balmer line numbers.
Figure 8: The normalized average equivalent widths of the
SSBA in the MWC560' spectrum in different observing seasons in
respect to the line numbers. The error bars for the lines
after H10 in the period January-April 1990 are not shown
because the measurements are made on the basis of a single
spectrum.The data for the normal supergiant 
Cyg are
taken from Leedjärv & Iliev (1989)
We chose the half-sum of the 
of 
and 
in
each plate as an internal normalization unit
because both lines are best measurable in all spectra.
The relation, in the case of discrete high-velocity ejections, is the steepest one and after n = 6 entirely coincides with the relation for normal main sequence A0-A2 star.
The relation for the 1990/1991 observing season, when a quasi-stationary ejection with low velocities was observed, resembles rather those for a supergiant of the same spectral type. The case of quasi-stationary ejection with high velocities, during 1991/1992 and 1992/1993 observing seasons, shows an intermediate picture. The resemblance of the relations for MWC560 in different moments with the main sequence or the supergiant stellar ones probably indicates the changes in the optical depth, the density and the velocities of the media where the absorptions arise.
Acknowledgements
We thank E.L. Chentsov who kindly provided us with unpublished radial velocities in the near infrared region. We also thank Ulisse Munari for the very constructive comments. This research has been sponsored by the Bulgarian National Foundation for Scientific Research under contract F-35/1991. We are grateful to the anonymous referee for critical comments.