%
%       Article coded in LATeX
%
%\documentstyle{l-aa}
%\documentstyle[referee]{l-aa}
\documentstyle[psfig]{l-aa}
\begin{document}
%
\appendix
\section{Discussion of individual systems}

\subsection{ADS 1315}

    Photoelectric magnitudes of AB and C components given in 
Table 2 were obtained by N.Shatskii with a scanning slit 
photometer. They agree well with the spectral type K1V. The 
equivalent widths of the CC dip are 0.5 $km/s$ less than expected
for the $B-V$ color index, probably  because of metal
deficiency. The spatial velocity of this system is 
typical for disk population rather than for spherical 
population, but disk population has considerable metallicity range
(Edvardsson et al. 1993).  

    The recent visual orbit of the pair AB (Baize 1994) was 
used to calculate dynamical parallax, and the corresponding 
absolute magnitudes agree well with adopted spectral type. 
Amplitude of relative radial velocities in the pair AB is 
estimated to be $7~ km/s$, hence the eventual determination of
orbital parallax from the radial velocity measurements of AB 
seems problematic. 


\subsection{ADS 1849 = HR 710}

     Babcock (1958) discovered the spectroscopic binary nature
of this magnetic Ap star and determined the orbital period. The
most thourough study of this system is that of Bonsack (1981)
who used his own observations and those of Babcock to derive the
elements of spectroscopic orbit. Spectroscopic elements based on
11 measurements were published by Abt \& Snowden (1973).
Surprisingly, neither of these orbits was included in the 8-th
catalogue of Batten et al. (1989). 

     The orbital elements given in Table 3 were obtained by
combining my 17 observations with 15 velocities of Bonsack
(1981) (his first measurement was not used due to unacceptably
high deviation). No constant
velocity corrections were applied, because the zero points 
of these data sets seem
to be in good agreement. Eccentricity is very small, but its
difference from zero seems to be significant (the $\chi ^2$ for
a circular orbit solution increases from 46.9 to 67.9).
The period is practically identical to 
Babcock's value  $2\fd997814$. The constancy of period and center of
mass velocity suggests that there are no other close visual
companions to ADS 1849A.

     Bonsack (1981) suggested that this system is viewed almost
pole-on ($\sin i=12\degr$), while axial and orbital rotations are
synchronized. The constant  negative polarity of magnetic field
and absence of convincing evidence for light variability are
additional arguments in favor of this interpretation. Our value
of $V\sin i$ leads to estimated inclination $i=14\degr$ and
estimated companion mass $\approx 1.0~M_{\odot}$. Its lines were
searched for and not found by Bonsack who concludes that the
companion is likely to be a white dwarf.

There is no doubt that the system AB is physical. The estimated radial
velocity difference between A and B is $1.7 ~km/s$ and agrees with the
observed difference of $0.7 ~km/s$. Distance to the system is
estimated from the spectroscopic parallax of the B component.

\subsection{ADS 3608}

    The period of Cab is almost exactly half year, so each year
the periastron passage can only be observed in December. Very 
faint lines of the secondary were detected in the autumn 1995, 
and the phase coverage for the secondary remains very poor. The 
reality of this detection is confirmed by the agreement of 
secondary mass from double-lined solution with the minimum 
secondary mass given by single-lined orbit, as well as by the 
good agreement of system model with normal parameters of dwarf 
stars (it must be noted however that the spectral types of 
components AB adopted in the model are earlier than observed).
 
    Radial velocity difference in the visual orbit of AB is estimated
to be $6.4 ~km/s$, in qualitative agreement with the measured
width of the CC dip. Modern interferometric measurements 
are needed for orbit improvement. The C component can be 
resolved with 10 m aperture although the magnitude difference 
of $2\fm3$ in the visible makes it a difficult target for speckle
interferometry.
 
    Orbit of Cab is highly inclined $(i \approx 54\degr)$, while the
inclination of the visual orbit is only $27\degr$ , so the sub-systems
AB and Cab are not coplanar. 

\subsection{ADS 3824 = HR 1706}

    The component C was found to be a spectroscopic binary with 
orbital period of almost exactly 3 days. The period is 
confirmed by radial velocity trend during the night.

    The F5V star (radius $1.2 R_\odot$) with a 3 day period has axial
rotation of $16.9 ~km/s$. When compared to the measured $V\sin i$, the
inclination $i \approx 64\degr$  can be obtained under the assumption of
synchronism which must be valid for a convective star with such 
a short period. So, secondary mass can be tentatively estimated 
as $0.3 M_{\odot}$, and non-detection of its lines is natural.
There is a good agreement between measured and estimated CC
equivalent width of Ca.

    The A component is itself a single-lined binary and also a
$\delta$ Scuti variable KW Aur. Physical relation between A and C
leaves no doubt: components remain fixed during 153 years since
their discovery. The difference of center of mass velocities
between Aab and Cab is $-1.4~km/s$. If it is real (i.e. not
caused by the difference of velocity zero points), it can be
accounted for by the motion in the wide system AC (the estimated
velocity difference is $1.7~km/s$). The visual component B
($11\fm1$ at $12\farcs6$) is optical as evidenced by its fast
relative motion.

    ADS 3824C is the X-ray source {\rm RE 0515+324} containing a
very hot DA white dwarf (Hodgkin et al. 1993). Its effective
temperature is estimated to be around $50\,000 K$, and it is among
the 10 brightest EUV stars in wavelength range $100 - 200 A$.
The visual magnitude of the WD is estimated to be $V=14\fm1$. An
attempt to identify the WD with the spectroscopic secondary Cb
meets some difficulty since the latter is not sufficiently
massive. I suggest that the WD is a close visual component to 
Cab and denote it by D in
Table 5. The period of CD systen is an order of magnitude
estimate which satisfies the observational constraints: CD
remains unresolved by conventional observing techniques, and the
velocity amplitude due to the motion in a 1000-year orbit is
less than 2 km/s. If, on the other hand, the WD is identical to
the spectroscopic companion Cb, a photometric variability of C
with the orbital period is expected due to the asymmetric
heating of the primary by the WD. Hodgkin et al. (1993) also note
the striking similarity of the spectra of A and C, which implies
that C may be a $\delta$ Scuti variable like A.

\subsection{ADS 3991 = HR 1782}

    The object is very difficult to observe because the angular 
separation between A and BC is only $2\farcs7$. Beavers \& Eitter
(1986) noted the velocity variability but ascribed it 
erroneously to the component B (probably because in the 
combined light of ABC the dip of A has lower contrast). I 
selected appropriate measurements from their data and found 
that they correspond to the orbit of Aa. These additional data 
are used to improve the period accuracy. All other orbital 
elements are computed only from my own measurements.
 
    The component A rotates at $6.9 km/s$, appreciably faster 
than synchronous 
($2.7 ~km/s$). It is evident that the system is neither
synchronized nor circularized. The CC dip of Ab is too small
(contrast 2\%) for reliable measurement of rotation.
 
    The visual system BC has two possible orbits (van den Bos, 
1962): circular orbit with $49.36 y$ period and eccentric orbit
with $24.68 y$ period. The combined CC dip of BC is significantly
broadened, implying difference of component radial velocities 
around $10 ~km/s$. 
The expected velocity difference in 1995 is only $5 ~km/s$ for
eccentric orbit and $10.7 ~km/s$ for circular orbit. Thus I
believe that circular orbit is more likely and adopt its 
parameters for the system model. The only speckle measurement 
of BC (McAlister et al. 1993) fits both orbits acceptably.

    There is a non-negligible difference of radial velocities 
between Aab and BC, $+3.8 ~km/s$. It may result from the motion of
BC in the visual orbit. The motion in the wide orbit A-BC can 
also explain this difference because an orbital period of 800 
years and total mass of $4.5 M_{\odot}$ correspond to maximum
velocity difference of $5.3 ~km/s$.

    There is no separate photometry of A and BC. In 
constructing system model I assumed that the combined magnitude 
of BC is $7.0^m$ and adjusted the magnitude of A accordingly. Then
the individual magnitudes of Aa and Ab were estimated from the 
ratio of CC equivalent widths. 

\subsection{ADS 6646}

    Radial velocities of this wide pair were measured on 
request from A.A.Kiselev. It is included in this publication 
because the component A was found to be a single-lined binary. The 
absence of photometry and MK spectral type makes the model of 
this system very uncertain. It is clear however that AB is a
physical pair.

\subsection{ADS 8861}

    Like the previous one, this is a wide pair from 
A.A.Kiselev's list. It is a nearby star Gliese 507 AB with
trigonometric parallax of $0\farcs0799 \pm 0\farcs0125$.
Variability of the
radial velocity of A was noted by Upgren \& Caruso (1989) and 
Stauffer \& Hartmann (1986) (hereafter SH86).
 
    Due to the faintness of the star and small contrast of the 
CC dip I had to iterate the orbital solution several times and 
to reject some aberrant measurements. The small magnitude 
difference $0\fm55$ and semimajor axis of $0\farcs051$
makes this
system an easy target for speckle interferometry. Speckle 
observations can be used to find orbital parallax, component 
masses and luminosities with good accuracy. The
displacement of the photocenter of Aab with a 200 day period 
can explain the significant scatter of trigonometric parallax
measurements as reported by Upgren et al. (1985). This is why a 
spectroscopic parallax is used in the model.
 
    Appreciable axial rotation of Aa and Ab conforms with the high
chromospheric activity of this star (cf. SH86). It was never 
listed as a flare star, however. Good agreement of observed and 
modelled equivalent widths of the CC dip indicates normal metal 
abundance, and there is no reason to consider Gl 507 as a 
subdwarf. Instead, the reduced depth of the $H_{\alpha}$ line noted
in SH86 can be explained by the contribution of Ab light.
 
    The CPM companion B is too faint for our instrument. 
Only one noisy measurement of B was made in exceptionally good 
conditions. Other radial velocity measurements given in the 
papers cited above are $-9.8$ and $-7.9 ~km/s$.

%
\end{document}
\bye