Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://star.arm.ac.uk/~ambn/rscvns.ps Äàòà èçìåíåíèÿ: Fri Aug 2 13:31:42 1996 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 21:12:27 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: annular solar eclipse

A&A manuscript no.
(will be inserted by hand later)
Your thesaurus codes are:
08(19.31.3; 19.36.1; 19.53.1; 19.08.1)
ASTRONOMY
AND
ASTROPHYSICS
27.5.1996
Environments of active close binaries II
GK Hydrae and TY Pyxidis
A. G. Gunn 1 , J. G. Doyle 1 , and E. R. Houdebine 2
1 Armagh Observatory, Armagh, BT61 9DG, Northern Ireland
2 Solar System Division, ESA Space Science Department, ESTEC Postbus 299, 2200 AG Noordwijk, The Netherlands
February 1996
Main Journal
Running title: Environments of TY Pyxidis and GK Hydrae
Abstract. High­resolution spectroscopic observations were
obtained for two eclipsing active close binary systems, GK Hy­
drae and TY Pyxidis. For GK Hya excess emission was ob­
served in the Balmer lines and Mg i b lines while TY Pyx
showed excess emission in the Balmer lines and Ca ii H &
K lines. The emission from GK Hya arises from a global phe­
nomenon associated solely with the secondary component. The
lack of significant He i D3 excess absorption and an analysis
of the flux ratio in the Balmer lines suggest that the emission
originates in prominence­like material seen off the limb of the
star. A marginal broad excess absorption band centered around
the Balmer emission peaks may be caused by this material ab­
sorbing the stellar continuum against the disk with high (350
km s \Gamma1 ) line­of­sight turbulent velocities. The inference that
plage­like regions and associated star­spots are not substantial
on this star is in agreement with the lack of spot modulation
in the photometric light curve. For TY Pyx the majority of the
Hfi excess and a larger proportion of the Ca ii core emission
could be attributed to a global phenomenon on the primary.
Analysis suggests that very large volumes are responsible for
the emission, however, there is no observational evidence for
the existence of discrete extended structures around TY Pyx.
Key words: star: binaries: spectroscopic ­ stars: RS CVn, BY
Dra binaries ­ stars: late type ­ stars: chromospheres of ­ stars:
magnetic activity
1. Introduction
Recently interest has been generated in the existence of cool
coronal material similar to solar prominences surrounding
either single active stars or the active components of bi­
nary systems. AB Dor is a well­studied single K0 dwarf for
which Robinson & Collier Cameron (1986) found evidence
of prominence­like condensations of mainly neutral material
trapped in corotation by the dynamo­induced magnetic field.
The variability in these transient Hff features (see also Col­
lier Cameron et al. 1990) is consistent with a series of clouds
Send offprint requests to: A. G. Gunn
crossing the stellar disk and these are thought to consist of
cool filamentary structures. Recently Byrne, Eibe & Rolleston
(1996) found prominence­like material embedded in the hot
corona of the rapidly rotating dMe flare star HK Aqr which
however appeared to form below the corotation radius. Other
single active stars also show clear evidence of circumstellar
prominences (Collier Cameron & Woods 1992).
Binary stars have also been the subject of intense scrutiny
for the existence of features analogous to solar prominences.
In such stars (for example RS CVn binaries; Hall 1976) ex­
cess absorption features are commonly seen in the continuum
of the non­active star as extended material from the active
companion overlies the stellar disk. Hall et al. (1990) reported
an unusual excess Balmer line absorption feature near pri­
mary eclipse in the SS Boo system which was attributed to
an extended region of 4­4.6R fi . Following this Hall & Ram­
sey (1992) surveyed 10 RS CVn systems and reported stable
prominence­like material corotating with either the primary or
secondary stars in eight of their targets. This study concluded
that amongst eclipsing RS CVn stars prominence material is a
common feature.
Some theoretical studies have been performed concern­
ing the observation of prominence­like material in active
stars. Buzasi (1989) developed an NLTE radiative transfer
model to derive values of the Balmer equivalent width ratio
(EWH ff /EWH fi
) in spectra corrected for the underlying pho­
tosphere. These results showed that low ratios (¸ 1­2) could
be achieved in both plage­like and prominence­like structures
viewed against a stellar disk but that values of ¸3­5 could
only be achieved in prominences seen off the limb of the star.
A similar conclusion was reached by Heasley & Mihalas (1976)
who considered detailed models of quiescent solar prominences.
Such predictions have been confirmed observationally by Land­
man & Mongillo (1979) who found high ratios in solar promi­
nences seen at the limb and by Chester (1991) who found ratios
! 2 in solar plage regions. In RS CVn binaries high ratios as­
sociated with prominences are often found (Newmark 1990;
Huenemoerder & Barden 1986).
The observational evidence for cool coronal material
around active stars is still very sparse but is significant in that
it is the remaining solar­like phenomenon to be confirmed in
such objects. Such stellar features are found to be much larger
than their solar counterparts (Hall & Ramsey 1994; Byrne,

2
Eibe & Rolleston 1996) and may well extend beyond the com­
ponent Roche lobes into regions where they are no longer mag­
netically confined. They therefore have important implications
for mass and angular momentum loss in active stars. They fur­
ther suggest that there exists cospatial plasma with significant
emission measure at both optical and X­ray wavelengths (since
active coronae are readily observed in the X­ray regime). The
details of prominence formation on this scale is an interesting
field of study, however, the unambiguous optical identification
of coronal structures in active late­type stars, and in particular
close binaries, is still required.
We have instigated a program of research in this field us­
ing a small sample of RS CVn binaries. In a recent paper
(Gunn & Doyle 1996, Paper I) we considered in detail the tech­
nique of spectral subtraction and its application in the search
for prominence­like material surrounding the active sub­giant
components of close binaries. In that paper we presented and
discussed our results for the highly active star ER Vulpeculae
where we found good evidence for substantial plage­like mate­
rial overlying the stellar surface and no evidence of extended
structures. In the present paper we present our analysis of two
further binary systems, GK Hydrae and TY Pyxidis. We sum­
marize observations to date for these stars, present our obser­
vational data, briefly review the spectral subtraction technique
and present our results and interpretation for these systems.
2. GK Hydrae and TY Pyxidis
GK Hydrae is an eclipsing binary of the RS CVn type with an
orbital period of 3.587 days and consists of an F8 primary and
a G8 IV secondary of radii 1.51R fi and 3.39R fi (respectively)
with a separation of about 6.8R fi . Note that in this paper
we term the presumed hotter component as the primary. This
corresponds to the primary star being eclipsed at phase 0.0.
Oliver (1974) noted Ca ii H and K emission from the cooler
component of GK Hya. Hall (1976) classed the system as an RS
CVn based on these observations and the wave distortions seen
in the optical light curve presented by Popper (1974). Popper
(1980) lists many of the properties of this system.
The available data for GK Hya is sparse and consists
mainly of optical photometry. uvbyfi photometry was provided
by Reglero et al. (1987) while Scaltriti et al. (1993) observed
anomalous polarisation in the R band in their survey of UB­
VRI linear and circular polarisation from RS CVns. The most
comprehensive study of the system is by Popper (1990) who
analysed spectroscopic data to determine the orbital param­
eters. Based on data provided by Hall & Kreiner (1980) and
Popper & Dumont (1977) he shows that the primary eclipse
is total, deduces the relative intensity weights of the system
and discusses the spectral types. Popper & Dumont (1977)
and Caton (1986) present photometry which shows intrinsic
variability although no clear photometric wave distortion is
found for the system. The details of surface features and the
circumstellar environment are therefore unknown and no de­
tailed spectroscopic study has yet been performed. As in other
RS CVn systems ultraviolet excesses are present for both com­
ponents of GK Hya, about 0.07 magnitudes for the primary
and 0.1 magnitudes for the secondary in U­B relative to B­V.
The colours are consistent with the spectral types of F8 and
G8 IV (Popper 1990).
The 6­cm radio continuum emission from GK Hya has been
measured at Ÿ0.4 mJy by Morris & Mutel (1988) and at 1.04
mJy by Drake, Simon & Linsky (1989). The radio emission
is therefore highly variable. The system was apparently not
detected in the Einstein IPC Slew Program (Drake, Simon &
Linsky 1992). However, Walter & Bowyer (1981) give an X­ray
luminosity of 7.9 10 30 erg s \Gamma1 based on their Einstein observa­
tions. More recently Dempsey et al. (1993) measured LX !1.54
10 30 erg s \Gamma1 using data from the All Sky Survey phase of the
ROSAT mission.
TY Pyxidis is an eclipsing RS CVn­type binary with an
orbital period of 3.199 days and consists of two almost identi­
cal G5 IV stars of radii 1.59R fi and 1.68R fi at a separation of
about 24.5R fi . TY Pyx is slightly evolved and the secondary
star is earlier than usual for RS CVns. The system was first
discovered to be an eclipsing variable by Strohmeier (1967).
Popper (1969) found Ca ii H and K emission lines as well as
stellar absorption lines from both components. There are few
detailed spectroscopic studies of TY Pyx other than that of
Xuefu & Huisong (1987) who observed Hff absorption from
the system and Andersen & Popper (1975) who presented the
first radial velocity curve of the system and determined the
physical properties. Montes et al. (1995a) recently observed
excess Hff emission from both components of the system with
the stronger emission from the primary component. Andersen
et al. (1981a) presented an excellent discussion of the prob­
lems associated with the evolutionary status of TY Pyx as well
as giving a four­colour photometric solution. Photometry for
TY Pyx has been provided, amongst others, by Surendiranath,
Rao & Sarma (1978), Hoffman (1978), Andersen et al. (1981b),
Rao & Sarma (1981), Caton (1986), Reglero et al. (1987) and
Allen et al. (1993). This binary appears to be a predictable sys­
tem with no apparent period changes observed. Caton (1986)
analysed UBV photometry which revealed a distortion wave
in the light curve with an amplitude of a few hundredths of a
magnitude which was however comparable with the errors and
therefore inconclusive. He also found little correlation between
the wave phases in different wave­bands. Rao & Sarma (1981)
also found no evidence for a distortion wave. It is therefore
unclear whether TY Pyx has observable surface features.
Many authors have attempted to measure the radio contin­
uum emission from TY Pyx. Vaughan & Large (1987) did not
detect the system above 5 mJy using the Molonglo Observatory
Synthesis Telescope (MOST) at 843 MHz and Collier Cameron
et al. (1982) did not detect it above 10 mJy at 5 GHz with the
Parkes 64­m dish. Slee et al. (1987) and Owen & Gibson (1978)
both failed to detect the system in their radio surveys at 6­cm.
Florkowski et al. (1985) gave an upper limit of 1 mJy while
Morris & Mutel (1988) detected it at the 0.7 mJy level with
the VLA. Stewart et al. (1988) gives the 8 GHz flux as 4 mJy.
Fox et al. (1994) studied the system at numerous frequencies
and found that the source showed no evidence of radio flares
on a time­scale of a few hours or less. However, they did find
flux variations of up to a factor of three on consecutive days.
TY Pyx has also received some attention in the X­ray
regime. Drake, Simon & Linsky (1992) report that it was not
detected in the Einstein IPC Slew Program. Dempsey et al.
(1993) measured the X­ray flux at LX = 4.63 10 30 erg s \Gamma1 for
each component using ROSAT All Sky Survey data. Pasquini,
Schmitt & Pallavicini (1989) modeled EXOSAT spectra and
determined a single component thermal plasma corona at a
temperature of 2.2 10 7 K. Using EXOSAT observations in the
photon energy ranges 0.05--2 keV and 1--30 keV Culhane et al.
(1990) modeled the X­ray light curve of TY Pyx. They found

3
differing temporal evolution at each wave­band which indicated
the presence of two distinct X­ray emission regions associated
with the bimodal temperature distribution. The high tempera­
ture component was found to be at least as large as the binary
separation while the lower temperature component underwent
periodic primary and secondary eclipses.
Simon & Fekel (1987) observed TY Pyx with the IUE
and used chromospheric and transition region UV emission
lines in a study of the rotation­activity correlation. Fernandez­
Figueroa, De Castro & Gimenez (1985) and Fernandez­
Figueroa, Sedano & De Castro (1986) presented measure­
ments and analysis of transition region emission line fluxes
(most notably the Mg ii lines) from IUE observations of TY
Pyx and found a good correlation with the X­ray luminosity.
These spectra show very well­defined emission features from
low­ionization (e.g. O i) to high­ionization (e.g. N v) species.
Danezis, Antonopoulou & Theodossiou (1993) present a com­
prehensive analysis of all IUE spectra for TY Pyx and demon­
strate the existence of a weak absorption shell and a variable
Mg ii emission structure which appears to be phase­dependent.
Antonopoulou (1983) studied TY Pyx using JHK infrared pho­
tometry in an attempt to find evidence of circumstellar ma­
terial and light curve distortions. Wave­like distortions were
found to be present which were attributed to a region of ex­
treme star­spot activity but no conspicuous infrared excess was
observed.
3. Observations and data reduction
High­resolution spectroscopic data were obtained during a 4­
night observing run in January 1994 carried out with the
University College London ' Echelle Spectrograph (UCLES) lo­
cated on the Anglo­Australian 3.9­meter Telescope (AAT) at
the Anglo­Australian Observatory at Siding Spring, Australia.
Descriptions of the telescope are given by Wampler (1975),
Gascoigne (1975) and Whelan (1976). The UCLES instrument
(Walker & Diego 1984; Walker & Diego 1985) is located at
the f/36 coud'e focus of the telescope giving good stability of
the spectrograph; random shifts in calibration over one hour
are expected to be no worse than 0.005 pixels or 15 m s \Gamma1 .
The 'echelle spectra were recorded on a Tektronix CCD with a
pixel size of 24¯m and dimensions of 525 \Theta 1024 pixels. The
resolving power of the spectrograph is approximately 50,000
which corresponds to a wavelength resolution of 0.13 š A or a
velocity resolution of 4.3 km s \Gamma1 at Hff (6562.852 š A). Observa­
tions were timed to encompass the eclipses of the targets but
due to scheduling constraints only secondary conjunctions were
observed. In total 52 spectra of GK Hya were obtained with
integration times in the range 50--300 seconds over the wave­
length range 4620--7020 š A. For TY Pyx 114 spectra were taken
with integration times in the range 60--180 seconds over the
range 3850--4970 š A. Figure 1 shows schematic representations
of the binary systems during the observations. The atmospheric
conditions remained reasonably good during the observations.
The estimated seeing generally remained in the range 1--3 arc­
seconds. On occasion the slit size was varied to account for the
extinction caused by thin cirrus cloud but this also increased
the background lunar counts for some exposures. As well as
the target binaries, a small set of standard stars were also ob­
served to act as spectral standard stars or reveal the extent of
atmospheric absorption lines; these stars were HR 2637 (HD
52619), HD 53590, HD 55685 and HD 63868.
GK Hydrae
Secondary occulted
F8
M 1.25
R = 1.51
G8 IV
M 1.34
R 3.39
phase = 0.465
phase = 0.516
TY Pyxidis
Secondary occulted
(phase 0.5)
G5 IV
M 1.22
R = 1.59
G5 IV
M = 1.20
R = 1.68
phase = 0.460
phase = 0.559
Fig. 1. Schematic representation of the geometry of the binary sys­
tems during the spectroscopic observations. For GK Hydrae (a) the
first spectrum was taken at phase 0.465 and the final spectrum at
phase 0.516 (dotted disk). The primary disk is completely within
that of the secondary during these observations. The mid­eclipse
spectrum occurs at HJD 2449381.0638. Phases were calculated using
the ephemeris and period of Hall & Kreiner (1980). For TY Pyxidis
(b) spectra begin at phase 0.460 and end at phase 0.559, well away
from eclipse contacts. At phase 0.5 the secondary disk is occulted by
the primary (dotted disk). The mid­eclipse spectrum occurs at HJD
2449378.0822 and phases were calculated using the period and time
of conjunction given by Andersen & Popper (1975) and Andersen et
al. (1981a) respectively. All masses and radii are in solar units.
CCD reduction and spectrum extraction and calibration
was performed with iraf 1 . After bias subtraction, image trim­
ming and flat­fielding, spectral apertures were defined by refer­
ence to bright star images and object spectra extracted. Spec­
tra for GK Hya consisted of 17 orders and for TY Pyx con­
sisted of 33 orders. Th­Ar calibration frames were extracted
and lines identified across all orders (typically 200 lines were
defined). Wavelength calibrations did not drift by more than
0.02 š A and the extracted spectra typically had signal­to­noise
ratios in excess of 50. Normalisation to unity of each spectrum
was performed individually by fitting a spline function across
pre­defined continuum points. This was to ensure an objective
normalisation to the entire data set. Some spectral orders were
corrected for the presence of atmospheric lines by reference to
rapidly­rotating B­star spectra.
1
iraf is distributed by the National Optical Astronomy Obser­
vatories, which is operated by the Association of Universities
for Research in Astronomy, Inc., under contract to the National
Science Foundation (USA).

4
6540 6550 6560 6570 6580 6590 6600
­.2
0
.2
.4
.6
.8
1
1.2
synthetic observed
excess emission
4840 4850 4860 4870 4880
­.2
0
.2
.4
.6
.8
1
1.2
excess emission
5850 5860 5870 5880 5890 5900 5910
­.2
0
.2
.4
.6
.8
1
1.2
5170 5180 5190 5200 5210 5220
­.2
0
.2
.4
.6
.8
1
1.2
excess emission
Fig. 2. Observed, synthetic and subtracted spectra for spectral orders containing activity­sensitive lines for a single GK Hydrae spectrum
(phase 0.465). Lines shown are (a) Hff, (b) Hfi, (c) Na i D and He i D 3 and (d) Mg i b. Bold lines indicate the observed spectra with overlying
thin lines showing the synthetic spectra. The zero­continuum data are the resulting subtracted spectra showing excess emission/absorption
features.
4. Spectral Subtraction
We have used the spectral subtraction procedure described in
detail in Paper I in the analysis of the spectroscopic data for
TY Pyx and GK Hya. This is a method of estimating the
chromospheric contribution in active lines and involves simu­
lating the inactive spectrum of the target star and perform­
ing a linear subtraction from the observed data. In practice
we synthesize the binary spectrum by combining two spectra
of stars which resemble the target components in all respects
other than in their apparent levels of activity. These essen­
tially non­active stars we refer to as standards. Any manifes­
tation of activity is then visible in the subtracted spectrum
as emission or absorption features above or below the zero
continuum level. Similar techniques have been used by Herbig
(1985), Young et al. (1989), Thatcher & Robinson (1993) and
Montes et al. (1995a,b). The reader is referred to Paper I for a
complete discussion of the application of spectral subtraction
in this study. We emphasize here that an iterative matching
technique (e.g. Barden 1984) is not used in this study to de­
rive radial/rotational velocities and intensity weights due to
the eclipse nature of the observations. Instead, geometrically­
corrected, Teff ­derived intensity weights and model radial ve­
locities are used directly. Radial velocities are provided for GK
Hya by Popper (1990) and for TY Pyx by Andersen & Popper
(1975). Gunn et al. (1996) recently confirmed the ephemeris
and velocity semi­amplitudes of these systems. The final sub­
tracted spectra were analysed by measuring excess absorption
or emission features in the active lines of the target stars.
As will be seen in the ensuing sections some problems were
encountered in matching the normalisations of target and tem­
plate stars (c.f. Figures 2 and 6). This is particularly prevalent
in the blue region and stems from the fact that common con­
tinuum points do not exist for two stars of greatly differing
rotational velocities, at high dispersion and with a profusion
of absorption features. However, the normalisation mismatches
simply reveal a smoothly varying background while detailed
features usually cancel well. We believe that the isolation of ex­
cess features is still possible in these cases although we stress
the risks of quantitative interpretation. We also concur that
spectral subtraction is a difficult technique to apply in the blue
region.
5. Results
5.1. GK Hydrae
The spectral standard stars chosen to form a match for the
GK Hydrae system were HD 52619 (F8 V) and HD 53590 (G8
IV). The velocities of these stars were checked by locating the
positions of selected stellar lines. Based on this analysis the
correction for the systemic motion of the spectral standards
is consistent with the observational data. Agreement between
the wavelength calibrations of spectral standards and target
spectra were checked by measuring the positions of several at­
mospheric lines visible in all spectra. The maximum deviation
was no more than 0.01 š A.
Spectral synthesis was performed for the active orders in
the GK Hydrae spectra. The lines analysed were Hff and Hfi,
the Na i D and He i D3 lines and the Mg i b lines. Figure 2
shows examples of the fits to these lines for a representative
spectrum. Bold lines are the observed spectra and thin lines the
synthetic fits. Also shown in these diagrams are the subtracted
spectra. In orders containing no activity­sensitive features good

5
0
.2
.4
.6 (a)
1
2
(b)
Phase
RV
(km/s)
.46 .48 .5 .52
­40
­20
0
20
40
(c)
F8
G8IV
Fig. 3. Results of the analysis of the subtracted Hff emission line
from GK Hydrae. The upper panel shows the variation in EW, the
middle panel the variation in FWHM and the bottom panel the
radial velocity of the emission compared to the RV curves of each
component.
cancellation of lines was achieved. Both Balmer lines and the
Mg i b lines show a clear excess emission feature. Hff also
appears to be embedded in a very broad excess absorption
background. Due to the slight normalisation mismatch for the
Na i D order, noise and small atmospheric lines it is difficult
to measure the emission feature in the sodium lines but it is
quite clear that excess emission is present. There is a slight
indication of some excess absorption in the He i D3 line but
this is not significantly above the noise. The feature is however
clearly present. Small atmospheric absorption profiles in this
order which were not removed well in the data reduction make
it difficult to draw any firm conclusions from the results.
Figure 3 shows the results of measurements of the sub­
tracted Hff emission profiles from GK Hya. There does not
appear to be much variation in the width or peak intensity
in Hff which suggests the emission is originating on a single
component of the binary. The equivalent width of the emis­
sion does not change during the observations suggesting that
the emission mechanism remains stable. The FWHM also ap­
pears to undergo no significant changes again implying that
the emission is from a single component. Measurement of the
radial velocity of the emission (Figure 3 (c)) confirms that it
is originating solely on the secondary component of the sys­
tem. Similar results for the Hfi and Mg i b lines (at 5172.68
š A and 5183.61 š A) are shown in Figures 4 and 5 respectively.
Hfi is in emission in the subtracted profile and also seems to
be embedded in a broad absorption background. The emission
0
.1
(a)
0
1
2
(b)
Phase
RV
(km/s)
.46 .48 .5 .52
­40
­20
0
20
40
(c)
F8
G8IV
Fig. 4. Results of the analysis of the subtracted Hfi emission line
from GK Hydrae. EW, FWHM and radial velocities are shown.
equivalent width and FWHM does not vary with phase (al­
though the FWHM may increase slightly with phase) and the
velocity of the emission is firmly associated with the secondary
component. For the two Mg i b lines the measured results have
a large dispersion but clearly the emission is again associated
with the secondary component.
5.2. TY Pyxidis
The spectral standard star chosen to form a match for both
components of the TY Pyxidis system was HD 55685 (G5 IV).
Checks on velocities and wavelength calibrations were made
on the spectral standard using the same procedures as for GK
Hya.
Spectral synthesis was performed for the active orders for
all TY Pyxidis spectra. The lines analysed were Hfi (–4861.33),
Hfl (–4340.47), Hffi (–4101.74) and the Ca ii H and K lines
(–3968.47 and –3933.66). Figure 6 shows examples for some
of these orders of the synthetic spectra formed for one TY
Pyx spectrum. Bold lines are the observed spectra and thin
lines the synthetic spectra. Also plotted on these diagrams are
the subtracted spectra. Some inactive orders in this part of
the spectrum gave imperfect cancellation of lines. It should be
noted immediately that all Balmer lines other than Hfi did not
show an excess absorption or emission feature above the noise
in the subtracted spectrum or which could be unambiguously
measured above other apparently uncancelled lines. A good
example of this is shown in Figure 6 (b) for the Hfl line. This
order also shows the severe effect of problems with continuum
definition. Hence for Balmer lines other than Hfi it is not pos­

6
GK Hydrae Mg b
0
.1
.2
.3
.4
.5
(a)
0
1
2
3
4 (b)
Phase
RV
(km/s)
.46 .48 .5 .52
­40
­20
0
20
40 (c)
F8
G8IV
Fig. 5. Results of the analysis of the subtracted Mg i b emission
lines from GK Hydrae. EW, FWHM and radial velocities are shown
for the –5172.68 line (filled squares) and the –5183.61 line (open
squares).
sible to make any statements other than that they are present
in absorption with rotationally broadened depths consistent
with the spectral types. As can be seen however the Hfi line
and the Ca ii H and K lines all show significant excess emis­
sion features. This indicates that Hfi has significant filling­in of
the global absorption profile over and above that expected for
standard stars, while the Ca ii lines display core reversals, the
definitive indicator of chromospheric activity. Although there
are normalisation mismatches for the Ca ii lines there may be
an additional source of excess absorption across the entire pro­
file. This is apparent for the K line which shows a continuum
apparently well matched in all parts of the spectrum other than
surrounding the emission core.
Figure 7 shows the results of measurements on the sub­
tracted Hfi emission line of TY Pyx. The Hfi line seems to be
embedded in a broad absorption component. Measurements
are difficult on this feature since it is not significantly above
the noise but its position would indicate that it is not coinci­
dent with the emission feature and therefore probably due to
normalisation problems. This absorption is present throughout
the observations which are not always of an eclipsing nature.
Therefore the absorption cannot be due to the atmosphere of
one of the stellar components. However, the absorption appears
to be symmetrical with a maximum displacement of about
1.5 š A, which corresponds to approximately 100 km s \Gamma1 at Hfi.
Another possibility for this feature is a mismatch in spectral
types giving unequal flux levels in the wings of the Balmer
lines.
Figure 7 (a) shows the equivalent width (EW) measure­
ments for the excess Hfi emission. There is some obvious vari­
ation in this quantity with phase and the line strength appears
to reach a minimum shortly before the conjunction and then
rises again before slowly decreasing once more. The FWHM
of the emission is