AG Dra (BD +67
922) is a yellow symbiotic star which during the last
16 years underwent many different phases of activity,
characterized by a luminosity increase of 
in the visual, and
by dramatic spectroscopic variation.
The December 1990 observation was made in the framework of the
campaign of coordinated IUE and ROSAT observations of symbiotic stars,
when the star was in quiescence (V=10.0, Mattei 1996).
The ROSAT PSPC observations indicate that AG Dra was in a high X-ray state,
while the ultraviolet UV spectrum was at minimum (Viotti et al.
1994; Greiner et al. 1997).
Our observation was made at spectroscopic phase 7.88 of the U-light curve
according to Mikolajewska et al. (1995) ephemeris (Fig. 1).
The only emissions present in the observed spectral region are H
and the He I 667.8 line. The first appear asymmetric, with extended wings.
The emission peak is red shifted, while the line barycentre at
0.1 
, like also the He I 667.8 line,
has the same velocity as the K-star absorptions.
Subsequently, we observed AG Dra with lower resolution at Loiano in
May 1992. The star was still in quiescence,
though slightly brighter than during the previous observing run (V=9.8).
H appeared stronger than in December 1990 with an indication of a
double peak structure.
In the NIR AG Dra displayed two emission lines of He II 1012.3 nm
and of He I 1083.0 nm, and much weaker P
(Viotti et al. 1994).
In January 1995 AG Dra was observed during the recent outburst phase
when it was declining (V=9.6) after the first light maximum,
and before the secondary light maximum of July 1995.
The
U-light curve phase was 10.56.
We found H emission much stronger than the previous observation,
but with the same shift as before with respect to the K spectrum.
ROSAT observations indicate that at this epoch the star has
nearly recovered its pre-outburst X-ray flux, after the remarkable
flux decrease during the optical maximum (Greiner et al. 1997).
The yellow and red spectra of the symbiotic, Mira-type variable R Aqr
(HD 222800) were obtained during the ROSAT-IUE campaign.
The yellow region only shows a great deal of
very narrow molecular absorptions of the M star,
and a deep and broad Na I D-doublet (see Viotti et al. 1994).
In the red spectrum H
is present as a narrow emission whose intensity and radial velocity are
strongly affected by the TiO band. The value of the equivalent width
shown in Table 1, refers to the continuum level on the blue side.
No other emissions are detectable in this spectrum.
CH Cyg (HD 182917)
is a late-type symbiotic star (M-type primary) especially known for the
recent large photometric variation (from the sixth to the ninth
magnitude, e.g. Skopal et al. 1996), and the radio jets.
In 1990 ROSAT observations made within the above cited RIASS campaign
indicated CH Cyg as a weak X-ray source (about 0.084
count s-1, Viotti et al. 1994). In addition,
the IUE ultraviolet spectrum taken on 1990 October 27,
was rather faint, and the energy distribution fitted rather well
that of the standard A0I star Cyg, supporting the hypothesis
of the presence of a warm photospheric-like (or shell-type)
cocoon envelope.
The high resolution spectrum in the NaD region shown in Fig. 3, was
obtained two months later, when the star had nearly the same visual
luminosity (
, Tomov & Yudin 1992).
It is noticeable the absence of the molecular absorption lines
of the M-stellar component, which normally dominate this spectral region.
The only detectable stellar feature is a broad absorption
which we attribute to He I 587.56 nm whose profile is strongly
contaminated by telluric lines.
In May 1992 the He I 587.56 nm line was strong in emission and the
M-type absorption spectrum was prominent (Viotti et al. 1994).
We observed again CH Cyg in January 1995 when the star was subject to
large photometric variability especially in the Balmer continuum
(V=8.3-8.9, U=8.1-10.0, Skopal et al. 1996). Skopal et al.
noted variations in the H and H
profiles between
spectrograms taken from many days to a few months apart.
Deep changes of the profile of H and He I 
447.1 were also
observed by Iijima (1995) between 1995 February 28.1 and March 1.2.
Our spectra show narrow emission lines of ionized metals overimposed to
a strong M spectrum and reveal profile variations in the hydrogen and helium
lines, on daily and shorter time scale.
Figure 3 (click here) shows the H, H
and H regions.
The Balmer lines have complex P Cygni profiles with
prominent emissions split in two by a central minimum whose position
remained stable with the same HRV for all the lines in all the exposures.
The intensity and position of the red peak of the emission component showed
minor variation. Major changes occurred on the blue side where the extended
absorptions largely varied from one exposure to another.
The H region is shown in Fig. 1 (click here).
In the intermediate resolution spectrum, which ends at 657.5 nm, the emission
peak resulted saturated, but the velocities of the absorption components are
very close to those of the other Balmer lines.
The same agreement has been found in the high resolution spectrum.
In the same exposure, on the red side of H is present a small
emission which we tentatively attribute to [N II] 658.4.
In the yellow region, He I 587.6 is strong in emission like in
1992, and the Na I resonance lines have a P Cygni profile (see Fig. 3 (click here)).
Profile variation similar to those of the hydrogen lines were also noted in
the He I lines, but the quantitative measurement is difficult because of
the blending with other lines.
Contrary to the hydrogen and helium lines, the other emissions remained
unchanged in all the exposures in intensity and
position with a mean HRV of -70 km s-1.
The NIR high resolution spectrum shows only a great deal of photospheric absorption lines of the M-star. In spite of the high S/N ratio and spectral resolution, no emission is detectable at the expected position of the main symbiotic features. Anyhow, it might be possible that these features are lost because they are too broad and shallow to be detected at our resolution (Fig. 2 (click here)).