. : http://star.arm.ac.uk/~ambn/261mzb.ps
: Tue Feb 11 15:04:27 1997
: Mon Oct 1 21:09:27 2012
:

: http astrokuban.info astrokuban
A&A manuscript no.
(will be inserted by hand later)
Your thesaurus codes are:
08 (08.01.1; 08.01.3; 08.03.2)
ASTRONOMY
AND
ASTROPHYSICS
23.12.1996
Properties of Herich stars. I Their evolutionary state and
helium abundance ?; ??
M. Zboril 1??? , P. North 3 , Yu. V. Glagolevskij 2 , and F. Betrix 3
1 Armagh Observatory, Armagh, College Hill, BT61 9DG, Northern Ireland
2 Special Astrophysical Observatory, Nizhnyj Arkhyz, Russia
3 Institute d'Astronomie de l'Universit'e de Lausanne , CH1290, ChavannesdesBois, Switzerland
Received date; accepted date
Abstract. A determination of the surface gravity and an abundance analysis of helium in a sample of 17 Herich
and 5 normal, reference stars is presented. These results are derived from low resolution CCD spectra, but each star
was measured at least 6 times in order to obtain a significant average spectrum for the spectroscopic variables. The
helium abundances derived from the models used are very close to 0:1 for normal, reference stars and are larger for
the others, clearly indicating the Herich phenomenon in them. NLTE effects, errors on the microturbulence value or
on the surface gravity do not influence the estimated helium abundances. Nevertheless, synthesized Geneva colours
are affected by the Herich peculiarity, especially the [UB] index which systematically changes by 0.025 mag per 0.1
of He abundance for the coolest stars in the sample.
We cannot confirm the correlation between the evolutionary state and the helium abundance reported previously
(Zboril et al. 1994), although we used a more reliable technique of log g determination. All Herich objects lie within
the main sequence: their surface gravities are all inside the range 4:1 ! log g ! 4:5, with no more than three objects
having log g ! 4:25. We find a significant spread of helium abundances in this range of surface gravities, from the solar
value 0:1 up to about 0:4. Some of the programme stars (including reference stars) present emission in their Balmer
lines and therefore some kind of stellar activity. Strong helium overabundance often coexists with emission and stellar
activity.
Key words: stars: abundances -- stars: atmospheres -- stars: chemically peculiar -- stars: fundamental parameters
1. Introduction
Herich stars are the most massive chemically peculiar (CP) stars: their spectral type is around B2, and they belong to
the main sequence. Their helium lines are anomalously strong for their colours, implying abundances n(He)=n(H) 0:5
in their atmosphere, instead of 0:1. Most of them are magnetic, like the cooler Ap Si and SrCrEu stars, but with fields
about 3 times stronger (Bohlender et al. 1987). They also show spectral, light and magnetic variability, like the other
magnetic CP stars (Pedersen & Thomsen 1977, Pedersen 1979). They are generally considered as belonging to the same
family as the magnetic Ap stars, because of the many characteristics they share with them. Their abundance anomalies
are interpreted as due to a diffusion mechanism involving a competition between the radiative and gravitational forces
in the atmosphere, in the presence of a wind (Vauclair 1975; Michaud et al. 1987; Vauclair et al. 1991). Vauclair (1975)
was the first to show that a wind is required to accumulate helium in the atmosphere because otherwise this element
Send offprint requests to: M. Zboril
? Based on observations collected at the European Southern Observatory, La Silla, Chile
?? Tables 1 is available only in electronic form at the CDS via anonymous ftp 130.79.128.5
??? On leave from Astronomical Institute,05960,Tatransk'a Lomnica,Slovakia

2 M. Zboril et al.: Properties of Herich stars. I. Evolutionary state.
would sink. Michaud et al. (1987) mention that their model predicts normal abundances of the CNO elements, which
indeed agrees with the observations, as a forthcoming paper will confirm.
The main review of the observational properties of Herich stars remains the work of Walborn (1983). He has
observed 19 such stars, taking photographic spectrograms at 39 mm \Gamma1 . He published equivalent widths of hydrogen,
helium and metallic lines as well as projected rotational velocities, which appear to be similar to those of normal B
stars. He also discussed the evolutionary state of these stars, but faced a problem with the scale of his equivalent
widths.
Since the work of Walborn, there has been as yet no systematic survey of the spectroscopic properties of these
stars using a linear detector. To fill this gap, and to tighten the constraints on the theoretical models of Herich stars,
we have observed a sample at both low and high dispersion with CCD detectors. This paper is the first of a series
and describes the abundance analysis of helium for 17 Herich stars and 5 normal reference stars. We give the helium
abundance and its dependence upon the surface gravity, interpreted as an age indicator on the main sequence. The
main purpose of this paper is indeed to confirm the correlation between the He abundance and surface gravity found
by Glagolevskij et al. (1992) and by Zboril et al. (1994) on smaller samples.
2. Observations and reduction
One of us (PN) obtained a total of 155 spectra for 24 stars in the spectral range 395.2 -- 493.8 nm using the 1.5m
spectrographic telescope and the Boller & Chivens spectrograph at the European Southern Observatory, La Silla,
Chile. The grating used was #20 (ESO numerotation) in the second order; it has a blaze wavelength of 455 nm and
1200 grooves per millimeter, giving a reciprocal dispersion of 32 Amm \Gamma1 , a resolution of 1.08 A (about 2 pixels per
A) and a resolving power R = 4150. The width of the entrance slit was 329 m, i.e. 3''. The detector was ESO CCD
#24, a Ford Aerospace 2048 \Theta 2048 chip with 15 \Theta 15m pixels. The integration times varied typically between 3 and
20 minutes according to the star's brightness.
The journal of the observations is given in Table 1. Six out of the 24 stars observed are defined as v sin i standards
by Slettebak et al. (1975). One different standard star was observed each night, while each Herich star was observed
in several nights, typically in 3 to 6 nights. The raw data were reduced by FB using the TACOS software developed at
Geneva Observatory. The procedure was the same as that used by North & Paltani (1994) and implied an extraction
of the spectrum using the Horne algorithm. The wavelength calibration was done using HeAr exposures done at the
beginning and at the end of each night. Since we were not interested in precise radial velocities, we did not make a
HeAr exposure after each observation. Therefore our wavelength scale may be slightly in error, in a different way
for each star, because of instrumental flexures. But this cannot affect our present analysis in a significant way. After
wavelength rebinning, we defined a continuum interactively, and the points chosen were linked by a thirddegree spline
function. The spectra were then divided by this continuum.
3. Analysis
The following transitions were identified and used in the spectra. Concerning neutral helium, we analysed the transitions
396.4, 400.9, 402.6, 412.0, 414.3, 438.7, 443.7, 447.1, 471.3 and 492.1 nm. The transitions 402.3 and 416.8 nm are
also visible but considerably weaker. The transitions of primary interest are 402.6, 414.3, 438.7 and 492.1 nm: their
equivalent widths are given in Table 2, allowing a comparison with Walborn's data (Figures 1). Other helium transitions
were either in hydrogen wings and blended (396.4, 412.0 nm) or blended (447.1 and 471.3 nm). In the case of some
(hot) stars transitions belonging to singly ionized helium were also identified: 410.0, 419.9, 433.8, 454.1, 468.5 and
489.5 nm. Since we primarily concentrated on helium abundance and its age dependence, we crosscorrelated and
coadded the spectra from individual nights to increase the S/N ratio at least in cases of nonvariables by using the
code Corel (written by MZ) for numerical correlation. Subsequently, the equivalent widths were measured by numerical
integration; the results are given in Table 2. The star HD 68450 has been removed eventually, since it is clearly a
very special case: it has been classified O9.7 IbII (e.g. Prinja et al. 1990). Indeed, hydrogen line profiles are extremely
narrow and have a peculiar shape in this object. This probably explains the 0.3 dex difference in helium abundance
found using high resolution CCD spectra in a previous study (Zboril et al. 1994). Likewise, the star HD 124448
(Popper's star) is not a member of the ``intermediate'' helium stars (as the Herich stars considered here are sometimes
called), but is extremely hydrogenpoor and is probably a postAGB star evolving towards the white dwarf state
(Schonberner & Wolf 1974).

M. Zboril et al.: Properties of Herich stars. I. Evolutionary state. 3
Table 1. Journal of the spectroscopic observations.The epochs are heliocentric julian dates corresponding to the middle of the
exposure. The stars HD 105435 and HD 121790 are rotational velocity standards from Slettebak et al. 1975.
Identification HJD texp Identification HJD texp Identification HJD texp
\Gamma2440000 [s] \Gamma2440000 [s] \Gamma2440000 [s]
HD 36485 8992.608 180 8996.760 1200 8996.609 300
=HR 1851 8993.541 240 8997.608 1200 8996.713 300
8993.630 240 8997.707 1200 8997.663 300
8994.540 240 8996.563 1200 HD 68450 8993.687 180
8994.620 240 COD\Gamma27 ffi 3748 8993.595 600 =HR 3219 8994.682 180
8995.540 240 =SAO 173185 8993.679 900 8995.679 180
8995.621 240 8993.811 900 8995.734 180
8996.536 240 8994.588 600 8996.676 180
8996.621 240 8994.671 1200 8996.745 180
8997.543 240 8995.587 900 8997.671 180
HD 37017 8993.548 180 8995.671 900 8997.744 180
=HR 1890 8993.635 180 8995.867 1200 COD\Gamma46 ffi 4639 8993.694 900
8994.545 180 8996.590 900 =CPD\Gamma46 o 3093 8993.784 1200
8994.624 180 8996.670 900 8994.689 900
8995.545 180 8996.823 900 8995.686 900
8995.626 180 8997.640 1200 8995.762 900
8996.541 180 8997.719 900 8996.685 1200
8996.626 180 8997.797 900 8996.771 1200
8997.548 180 HD 56139=HR 2749 8995.662 10 8997.680 1200
HD 37479 8993.583 180 HD 58260 8993.602 180 HD 92938 8993.728 30
=HR 1932 8993.640 180 =BD\Gamma36 o 3578 8993.728 180 =HR 4196 8993.818 30
8994.550 180 8994.595 180 8995.718 30
8994.631 180 8995.595 180 8996.730 60
8995.550 180 8995.721 180 8996.830 60
8995.631 180 8996.597 180 8997.729 50
8996.547 180 8996.733 180 8997.808 30
8996.630 180 8997.649 180 HD 96446 8992.837 180
8997.553 180 8997.732 180 =BD\Gamma59 o 3038 8993.732 180
HD 37776 8993.559 240 HD 60344 8993.610 300 8993.823 180
=BD\Gamma1 o 1005 8993.727 240 =BD\Gamma23 o 5673 8993.709 300 8995.722 180
8994.557 240 8993.822 300 8995.830 180
8995.555 240 8994.602 300 8996.734 180
8995.719 240 8994.704 300 8996.834 180
8996.551 240 8995.603 300 8997.733 180
8996.731 240 8995.701 420 8997.812 180
8997.558 240 8996.605 300 CPD\Gamma62 ffi 2124 8992.855 2100
8997.730 240 8996.724 300 =LSS 2394 8993.751 2700
HDE 260858 8993.584 600 8996.833 300 8996.791 2400
=BD+12 o 1223 8993.652 900 8997.657 300 8997.755 2400
8994.580 600 8997.786 300 HD 105435=HR 4621 8996.868 8
8994.641 900 HD 64740 8993.611 30 HD 108483=HR 4743 8997.804 30
8995.577 600 =HR 3089 8993.709 30 HD 110879=HR 4844 8993.842 10
8995.653 900 8994.603 30 HD 121790=HR 5249 8996.861 15
8996.578 900 8994.704 30 HD 122980=HR 5285 8993.844 20
8996.657 1200 8995.603 30 HD 124448 8993.830 1200
8997.627 900 8995.702 30 =BD\Gamma45 o 9033 8995.837 1200
HDE 264111 8993.572 900 8996.605 30 8996.848 1500
=BD+4 o 1447 8993.668 900 8997.658 30 8997.830 1800
8994.568 900 HD 66522 8993.617 300 HD 133518 8993.847 180
8994.657 1200 =BD\Gamma50 o 3111 8993.714 300 =BD\Gamma51 o 8745 8995.864 180
8995.566 900 8994.607 300 8996.833 180
8995.640 900 8994.708 300 8997.844 180
8995.751 1500 8995.608 300
8996.641 1200 8995.707 300

4 M. Zboril et al.: Properties of Herich stars. I. Evolutionary state.
Fig. 1. Comparison between Walborn's equivalent widths and ours for Hfl and three He I lines.The full lines represent the
equality, while the broken lines are the regression ones, assuming that our values are three times more precise than Walborn's.
Table 2. Hfl and helium 402.6 A, 414.3, 438.7 and 492.2 nm equivalent widths in A.
Star Hfl 402.6 414.3 438.7 492.2
36485 6.010 1.580 1.340 1.250 0.880
37017 5.580 1.570 1.170 1.220 0.902
37479 3.440 2.010 1.473 1.645 1.172
37776 3.740 1.830 1.432 1.520 1.202
260858 4.440 1.740 1.154 1.145 0.994
264111 3.945 1.980 1.540 1.310 1.125
27 4.801 1.590 1.140 1.080 0.921
58260 3.890 1.813 1.390 1.411 1.092
60344 4.110 1.662 1.320 1.250 0.950
64740 4.022 1.622 1.172 1.290 1.105
66522 4.250 2.030 1.610 1.540 1.264
46 3.820 1.535 1.126 1.210 1.118
92938 6.960 1.175 0.685 0.730 0.575
96446 3.780 2.440 2.150 1.975 1.540
62 2.702 2.204 1.865 1.940 1.334
108483 5.122 1.481 0.952 0.946 0.866
124448 --- 2.820 2.115 2.302 1.984
133518 4.825 1.863 1.290 1.294 1.073
56139 3.970 1.128 0.674 0.626 0.604
105435 2.602 1.040 0.672 0.548 0.340
110879 6.025 1.503 0.812 0.975 0.754
121790 4.960 1.470 0.910 0.932 0.808
122980 5.530 1.540 0.930 0.940 0.842

M. Zboril et al.: Properties of Herich stars. I. Evolutionary state. 5
3.1. Atmospheric parameters and modeling the atmospheres
To derive the atmospheric parameters of our stars we used the following criteria: H fi and H fl equivalent widths and
Glagolevskij's (1990) calibration, Geneva photometry calibrated by Kunzli et al. (1996), and theoretical H fi , H fl and
H ffi profiles using Kurucz's (1992) model atmospheres. Final estimates are given in table 4. They were obtained using
the following order of priority among the criteria adopted, namely: (1) Balmer line profiles, (2) Balmer line equivalent
widths and Glagolevskij's (1990) calibration, (3) Geneva photometry. The last 5 stars in Table 4 are reference ones.
Since UBV and Stromgren measurements are available only for less than half of the stars, we used as a photometric
method the Geneva photometry which is available for all programme stars but CpD \Gamma62 ffi 2124. Also since the spectral
region covered is wide (about 1000 A), we could use three hydrogen lines to derive the atmospheric parameters from
LTE Kurucz models: we modeled Hfi, Hfl and Hffi and fitted them to coadded observed profiles by means of a least
squares method. The Hfl line proved very suitable since the corresponding 2 sum remained small for all spectral
types. In this way, we got 6 estimates of effective temperature and surface gravity for each star because two best
estimates were considered for every hydrogen profile. As a first step, data from photometry were considered only as a
complementary method in some special cases. Even though emission was detected in the observed hydrogen profiles in
a few stars, only in the case of HD 56139 (a normal, comparison star) were the photometric and spectroscopic effective
temperatures different by as much as 7000 K; the spectroscopic method overestimates the temperature because it
underestimates the intensity of the H fi line, which is partly filled with emission, while Stromgren's c 0 parameter or
Geneva's X parameter is free from it. Figure 2 illustrates the fit in the case of HD 92938.
Fig. 2. H fl profile for HD 92938. Dots: coadded spectra from 7 nights, solid line: synthesized line profile from Kurucz (1992)
models with T eff = 15000K, log g = 4:0,dashed dotted line: T eff = 17000K, log g = 4:5, dashed line: T eff = 15000K,
log g = 4:5. Synthetic profiles were convolved with vsini = 130 kms \Gamma1 and a gaussian with FWHM = 1 A.
The following relations were found between three criteria for T eff :
T Glagol = 0:874 \Theta T spectro + 2790 (1)
T Geneva = 0:978 \Theta T spectro + 1308 (2)
T Geneva = 1:118 \Theta T Glagol \Gamma 1750 (3)

6 M. Zboril et al.: Properties of Herich stars. I. Evolutionary state.
Fig. 3. A part of H fl profile from Kurucz models (1992). Dots: T eff = 19000K, log g = 4:5, solid line: 19000K, 4.0 , dashed
dotted line: 19000K, 4.5, NLTE option+n(He) = 0:25, dashed line: 20000K, 4.0 For rest models solar abundances. No
rotational broadening.
with rms standard deviations of the residuals of 1194, 1021 and 871 K respectively, and where T spectro is the effective
temperature derived from the observed and synthetic Balmer line profiles. These relations where obtained with the
macro lsq2 of the supermongo package, which allows for errors not only on the y axis, but also on the x axis (here we
assumed similar errors on both axes). Thus, estimates from Geneva photometry are closest to the values derived from
the profiles of the hydrogen lines, in the sense that the slope is almost unity and the scatter of the residuals is minimum.
However, there is a systematic shift of 1300 K, the photometry giving higher temperatures than spectroscopy. This
systematic difference is difficult to explain, but it might be due to the nonstandard helium abundance. Unfortunately,
we do not have spectra of the normal, standard stars used by Kunzli et al. (1996) to calibrate the Geneva system in
T eff .
Strangely enough, the surface gravities derived from Geneva photometry are not well correlated with those derived
in a purely spectroscopic way from the Balmer line profiles (see open dots in Figure 4). Hydrogen line profiles are
sensitive to both effective temperature and surface gravity. On the other hand, the Balmer jump measured by Geneva
photometry is essentially sensitive to temperature, but is also affected by helium abundance. Therefore we have
estimated the effect of the helium abundance on the Geneva colours (see next subsection) and redetermined the surface
gravities from the Balmer line profiles, but using this time the effective temperatures from Geneva photometry (effect
of He abundance included) as a fixed parameter. We then obtain a much better agreement between the spectroscopic
and photometric estimates of the surface gravity (black dots in Figure 4). Here again, all log g values are larger than
4.0, as in the photometric estimates.
3.2. Revised photometric T eff and hydrogen profile
The Balmer jump may be affected by the enhanced helium abundance, due to both b \Gamma f and b \Gamma b transitions of
helium. This fact was established using fully consistent model atmospheres. Having taken passbands functions from
Rufener & Nicolet (1988), we computed theoretical Geneva colours which are presented in Table 3, by integrating the
flux at the stellar surface. Models have been computed with log g = 4:0 and T eff = 16000, 20000 and 25000 K, in the

M. Zboril et al.: Properties of Herich stars. I. Evolutionary state. 7
LTE and NLTE approximations an for the two helium abundances n(He)=0.1 and 0.4 (n(H)=1.0). Table 3 allows to
intercompare the effect of the overabundance of He on the Geneva colours.
Table 3. LTE and NLTE Geneva colours for log g = 4:0 models.
colour Effective temperature
index 16000 20000 25000
LTE NLTE LTE NLTE LTE NLTE
n(He) = 0.1:
UB 0.613 0.605 0.384 0.377 0.216 0.215
BV 1.122 1.124 1.170 1.171 1.226 1.227
VB1 0.322 0.324 0.389 0.391 0.464 0.465
B1B2 0.792 0.793 0.835 0.837 0.877 0.877
B2V1 0.225 0.226 0.248 0.248 0.279 0.279
V1G 0.524 0.525 0.533 0.533 0.545 0.545
n(He) = 0.4:
UB 0.514 0.486 0.309 0.294 0.176 0.165
BV 1.106 1.100 1.149 1.152 1.212 1.214
VB1 0.307 0.313 0.371 0.374 0.451 0.453
B1B2 0.794 0.800 0.838 0.841 0.878 0.880
B2V1 0.208 0.209 0.228 0.228 0.265 0.266
V1G 0.521 0.521 0.526 0.526 0.540 0.539
According to the table, the [U \Gamma B] index of the Geneva photometric system is most affected by the Herich
peculiarity. Other Geneva indices change by only a few mmag, which remains within the standard measurements
scatter. Therefore, the Herich peculiarity affects essentially the T eff estimate, while the estimate of the surface gravity
(essentially measured by the B1 \Gamma B2 index which is included in the reddeningfree Y parameter) remains unaffected
in practice.
Geneva colours were revised according to the above table applying twodimensional (He abundance, T eff ) fits; this
means they were corrected to a standard helium abundance n(He) = 0:1 so that they yield meaningful, unbiased T eff
estimates. Test runs indicate a much higher dependence of the fluxes to nonsolar abundances of helium and other
elements than to pure NLTE effect.
We can also see from Figure 3 that if hydrogen profiles are processed with solar helium abundance we may get
inconsistency. Therefore we processed profiles with expected abundances obtained with the LTE approximation and
computed NLTE line profiles. The NLTE option plays a role in line cores, as expected since it depends directly on level
populations. The deeper into the atmosphere, the closer the NLTE departure coefficient is to unity. Helium abundance
therefore should play a role in line wings. Hydrogen profiles were processed using the standard VCS broadening
theory. It is worth noting that in the temperature interval of the programme stars the hydrogen profiles are much
more sensitive to surface gravity. We may also expect better agreement with temperature from Geneva photometry.
In addition, this hydrogen profile effect maintains over the whole temperature range 1600025000K in which most
of our programme stars lie.
Finally, we quote in Table 4 the atmospheric parameters obtained with both methods, except for HD 56139 and
HD 105435 where values from photometry were preferred: the spectroscopic parameters would imply that these stars
are strongly Heweak, while emission is clearly present.
For the microturbulence, we adopted here the depthindependent value v turb = 6 km s \Gamma1 , an average of Kilian's
(1992) determinations for early type stars. Although Kilian (1992) found a loose correlation between v turb and log g, we
adopted here a constant value. Most helium lines lie on the third part of the curveofgrowth and microturbulence should
not be significant in this case; indeed, due to the large values of equivalent widths, the influence of microturbulence is
not crucial. A set of helium abundances for two values of microturbulence is computed in each case. The investigation
of microturbulent values and their behaviour together with the abundance of CNO and other elements as well as
rotational velocities (v sin i) will be presented in another paper.
Once the atmospheric parameters were derived we proceeded with model atmosphere computations in two ways:
first, we adopted Kurucz (1992) LTE model atmospheres and computed detailed helium line profiles and equivalent
widths. Thus, the model atmosphere was not actually computed. Second, since the Herich stars are early type stars
we computed several consistent NLTE model atmospheres following LTEGrayLTENLTECNLTEL sequence. The
LTEGray abbrevation stands for initial LTEgray model atmosphere while NLTEC is for NLTE model for continua

8 M. Zboril et al.: Properties of Herich stars. I. Evolutionary state.
Fig. 4. Photometric versus spectroscopic surface gravities for the programme stars. Open symbols: Purely spectroscopic
values for the Herich stars, T eff being also determined from the Balmer line profiles; + symbols are for normal stars. filled
symbols: Spectroscopic values obtained assuming the photometric T eff corrected for the effect of He abundance (Herich stars);
x symbols are for normal stars. The continuous line is the equality. The broken line is a regression line fitted through the black
dots, while the dotted line has been fitted through the open dots.
and NLTEL for NLTE model for lines, i.e. boundbound transitions.We used the NLTE code Tlusty (Hubeny and
Lanz 1992) from the CCP7 library to make these computations. Once an LTE model was computed, we computed in a
subsequent run a NLTE model to check the NLTE level populations for helium and their possible consequence on the
equivalent widths as well as NLTE model atmosphere for the computation of hydrogene line profiles. The criterion of
convergence was that the maximal relative change of any physical quantity in the model (temperature, total particle
number, electron density) and level populations do not exceed 0.001 at each depth point.
We performed fully consistent computations, starting with solar composition atmosphere models and modifying
iteratively the He abundance until the synthetic He profiles agreed with the observed ones. Besides typical continuous
opacities valid for BA type stars (H\Gamma, H +
2 , Rayleigh scattering) we considered 42 explicit levels: 5 levels of neutral
hydrogen, 1 level of ionized hydrogen, 30 levels of neutral helium (the states: 1 sing S, 2 trip S, 2 sing S, 2 trip P,
2 sing P, 3 trip S, 3 sing S, 3 trip P, 3 trip D, 3 sing D, 3 sing P, 4 trip S, 4 sing S, 4 trip P, 4 trip D, 4 sing D, 4
trip F, 4 sing F, 4 sing P, 5 trip S, 5 sing S, 5 trip P, 5 trip D, 5 sing D, 5 sing F, 5 trip F, 5 sing P, ! n = 6 ?,
! n = 7 ?, ! n = 8 ?), 1 level of ionized helium, 2 levels of neutral carbon (the states: 2 sing S, 2 sing D), 2 levels
of singly ionized carbon (2 dubl P 1=2 , 2 dubl P 3=2 ) and 1 level of doubly ionized carbon which contributes to the
continuum opacity due to boundfree transitions (abbrevation !? denotes averaged levels with principal quantum
number but not mixing singlet and triplet states). The transitions 396.4, 402.6, 438.7, 447.1, 471.3 and 492.1 nm for
He were processed as explicit transitions: radiative rates are linearized, the other helium transitions were considered in
detailed radiative balance, radiative rates are not evaluated but collisional rates are. Explicit energy levels chosen cover
the observed He i neutral lines satisfactorily. Here we stress that the attention is paid to investigations of potential
impact of NLTE on helium abundances and evolutionary state since NLTE level populations are both abundance and
temperature dependent. The procedure above was not successful for very Herich stars, i.e. with helium abundance
roughly 0.25 and more, due to problems with the convergence of the models. Some test runs indicate that even not
converging models (maximal relative change of order 0.10.01) might be applicable, but if they are not fully physically
consistent, they are not reported here.

M. Zboril et al.: Properties of Herich stars. I. Evolutionary state. 9
3.3. Helium abundances
Helium abundances were computed using Huben'y's code Synspec modified by Zboril (1996). Helium line profiles were
computed in the following way:
447.1 nm after Barnard, Cooper and Smith (1974), 402.6, 438.7 and 492.2 nm after Shamey (1969) and the other
neutral helium lines after Dimitrijevic and SahalBrechot (1984). Singly ionized helium lines were computed after
Schoening and Butler (1989). An example of coadded spectra and comparison with synthesized ones is given in
Figures 5 and 6. Synthetic spectra were computed with solar composition for all elements except for helium.
Fig. 5. He I 4471 A line profile of HD 92938 from 4 nights, and corresponding coadded spectrum (solid line).
Once the model atmosphere was chosen (Kurucz LTE model) and the helium line profiles and equivalent widths
computed, the evolutionary state of each star was defined by the ratio R=R z which was computed according to the
equation
log(R=RZAMS ) = 1
2 (log g ZAMS \Gamma log g) (4)
Using Tables 8 to 10 of Schaller et al. (1992), and applying the transformation
log g = \Gamma10:607 + log(M=M fi ) + 4logT eff \Gamma log(L=L fi ) (5)
one obtains log gZAMS = 4:291,4.265, 4.262, 4.251 and 4.234 for M = 5, 7, 9, 12 and 15 M fi respectively. Since
log gZAMS is nearly independant of the mass, we adopt here the constant value 4.26 when applying equation 4.
The evolutionary state is presented in Table 4; in this table are also listed the helium abundances obtained, and the
errors indicated represent in fact the rms scatter from the different transitions used. The errors on R/R z are evaluated
adopting the usual expression for propagation of errors, giving:
oe( R
R z
) = ln10
2
R
R z
oe(log g) = 1:15 R
R z
oe(log g) (6)
where oe(log g) is the error on surface gravity. The same goes for helium abundance where we consider as three main
error sources: equivalent width measureuments (typically 45 percents), T eff and log g estimates. In Table 4, the

10 M. Zboril et al.: Properties of Herich stars. I. Evolutionary state.
Fig. 6. He I 4471 A line profile of HD 92938, observed coadded spectrum (dots) and computed spectrum (solid line).
Synthetic spectrum computed with a Kurucz (1992) model, convoluted with a rotational profile having v sin i = 135 kms \Gamma1 and
with a gaussian having FWHM = 1 A(dashed line) which approximates the instrumental profile. The helium abundance was
set to 0.19 while a solar abundance was assumed for magnesium; v turb =6 kms \Gamma1 was adopted.
symbol new stands for new estimates using the corrected photometric T eff in the spectroscopic estimation of log g, as
explained above. HD 68450, which is very hot, and HD 124448 are both clearly atypical (as explained in Section 3) but
are nevertheless listed, although they will not be considered in the final analysis. An 'e' symbol denotes emission, while
the '(H)' symbol means that the emission is clearly present in the H fl profile; the 'e' symbol alone means that emission
in H fi is recognized from a comparison between the observed fi index of uvbyfi photometry and the fi c parameter
computed from Geneva photometry according to Cramer & Maeder (1979). The solar value adopted is 0.1 relative to
hydrogen (n(H)=1.0).
3.4. NLTE helium level populations
Generally, departure from local thermodynamic equilibrium is easily seen in hot stars since many quantitites (Saha
Boltzmann relation, radiative and collisional rates etc.) depend upon temperature and partly on the intensity of the
radiative field. However, a NLTE situation is also present in cool stars and the Sun in their upper atmosphere, because
of an anomalous temperature behaviour associated with population and depopulation processes. As mentioned already,
Herich stars occupy a wide interval of temperatures and have nonsolar helium abundances, so we decided to check
the helium abundances considering the NLTE option of the Tlusty and Synspec programmes. In the case of bound
free helium transitions, we used relations after Opacity Project (up to level 4); otherwise, we used the hydrogenic
approximation with exact Gaunt factors. Collisional rates (excitation and ionization) were calculated following the
special procedure by Hummer, which is already included in the programme, and following Mihalas et al. (1975).
Table 5 gives the quantitative NLTE effects on the equivalent widths, as well as the effect of the microturbulence.
A set of three models was computed, with log g = 4:0 and T eff = 15000 K, 20000 K and 30000 K. We adopted a
helium abundance of 0.25. Two microturbulent velocities were considered, namely 1 and 10 kms \Gamma1 but we assumed
that no macroscopic motions take place. The table gives the relative change in percent of the equivalent width. The
equivalent width increases when the NLTE effects are taken into account and when the microturbulence varies from 1
kms \Gamma1 up to 10 kms \Gamma1 . The behaviour of the bfactors for the transitions 492.2 nm (states 2 1 P 4 1 D) and 1083.3 nm

M. Zboril et al.: Properties of Herich stars. I. Evolutionary state. 11
Table 4. Atmospheric parameters and evolutionary state for Her stars obtained using Kurucz models. The quantities with the
``new'' superscript have been obtained assuming the corrected photometric T eff , while the others have been obtained from the
Balmer line profiles alone.
Star T eff log g T new
eff log g new R=Rz He abundance R=R new
z He new emis.
36485 18000 4.4 \Sigma 0.11 18400 4.5 \Sigma 0.27 0.85 \Sigma 0.11 0.14 \Sigma 0.07 0.76 \Sigma0.24 0.15
37017 18500 4.3 \Sigma 0.10 19200 4.4 \Sigma 0.16 0.95 \Sigma 0.11 0.14 \Sigma 0.08 0.85 \Sigma0.16 0.12
37479 22000 4.5 \Sigma 0.17 22200 4.3 \Sigma 0.29 0.76 \Sigma 0.15 0.22 \Sigma 0.09 1.01 \Sigma0.34 0.32 e
37776 22000 4.5 \Sigma 0.07 21800 4.3 \Sigma 0.32 0.76 \Sigma 0.06 0.19 \Sigma 0.09 0.95 \Sigma0.35 0.22
260858 18000 4.0 \Sigma 0.10 19200 4.3 \Sigma 0.15 1.35 \Sigma 0.16 0.24 \Sigma 0.10 0.95 \Sigma0.16 0.19
264111 22400 4.2 \Sigma 0.06 23200 4.3 \Sigma 0.35 0.85 \Sigma 0.06 0.20 \Sigma 0.11 0.95 \Sigma0.38 0.30
\Gamma27 ffi 3748 23000 4.5 \Sigma 0.11 22700 4.3 \Sigma 0.30 0.76 \Sigma 0.10 0.13 \Sigma 0.06 0.95 \Sigma0.33 0.18
58260 19000 4.0 \Sigma 0.19 19000 4.25\Sigma 0.16 1.35 \Sigma 0.29 0.27 \Sigma 0.10 1.01 \Sigma0.19 0.21 e
60344 21000 4.5 \Sigma 0.18 21700 4.2 \Sigma 0.27 0.76 \Sigma 0.16 0.16 \Sigma 0.10 1.06 \Sigma0.32 0.21
64740 21000 4.5 \Sigma 0.06 22700 4.3 \Sigma 0.30 0.76 \Sigma 0.05 0.13 \Sigma 0.08 0.99 \Sigma0.34 0.18
66522 18000 4.0 \Sigma 0.10 18800 4.3 \Sigma 0.16 1.35 \Sigma 0.16 0.38 \Sigma 0.12 0.99 \Sigma0.18 0.32
68450 32500 4.0 32300 4.05 --- --- --- --- e
\Gamma46 ffi 4639 22500 4.5 \Sigma 0.11 22000 4.3 \Sigma 0.40 0.76 \Sigma 0.11 0.13 \Sigma 0.08 0.95 \Sigma0.44 0.22
92938 15000 4.0 \Sigma 0.10 15000 4.1 \Sigma 0.14 1.35 \Sigma 0.16 0.20 \Sigma 0.08 1.20 \Sigma0.19 0.20
96446 20500 4.4 \Sigma 0.21 22000 4.5 \Sigma 0.70 0.85 \Sigma 0.21 0.38 \Sigma 0.12 0.76\Sigma0.60 0.37 e
\Gamma62 ffi 2124 26000 4.2 \Sigma 0.21 --- --- 1.07 \Sigma 0.26 0.81 \Sigma 0.15 --- --- e(H)
108483 19200 4.3 \Sigma 0.10 19100 4.2 \Sigma 0.16 0.76 \Sigma 0.09 0.12 \Sigma 0.05 1.07 \Sigma0.20 0.11
124448 --- --- 22000 3.7 \Sigma0.21 1.91 \Sigma 0.46 !0.8 ? --- !0.8 e(H)
133518 17500 4.0 \Sigma 0.10 18600 4.3 \Sigma 0.15 1.35 \Sigma 0.16 0.29 \Sigma 0.10 0.95 \Sigma0.16 0.23
56139 --- --- 18000 3.6 \Sigma0.18 --- --- 2.14 \Sigma0.44 0.12 \Sigma0.05 e(H)
105435 --- --- 26000 4.2 \Sigma0.21 --- --- 1.07 \Sigma0.26 0.08 \Sigma 0.05 e(H)
110879 18000 4.4 \Sigma 0.07 18200 4.2 \Sigma 0.15 0.76 \Sigma 0.06 0.10 \Sigma 0.05 1.07 \Sigma0.18 0.11
121790 19500 4.4 \Sigma 0.11 19600 4.1 \Sigma 0.16 0.76 \Sigma 0.10 0.10 \Sigma 0.06 1.15 \Sigma0.21 0.12
122980 18500 4.4 \Sigma 0.10 19600 4.2 \Sigma 0.16 0.76 \Sigma 0.09 0.11 \Sigma 0.06 1.10\Sigma0.20 0.12
(states 2 3 S2 3 P ) are given in Figures 7 and 8 for model atmospheres with an effective temperature of 15000 K and
30000 K respectively. The common feature is the decrease of bfactors in the outer atmospheric layers. Since the line
tansition 402.6nm is sensitive to NLTE for hotter atmospheres we need to stress that the transition is bound to 2 3 P
and 5 3 D states. For upper level, however, approximate relations (collisional rates and bf cross sections) are used and
therefore we may expect the dependence on input data.
Table 5. NLTE and Microturbulence effects. The figures listed are the relative increase in percents, of the equivalent width of
the indicated line, when NLTE effects are taken into account and V turb is increased from 1 to 10 kms \Gamma1 respectively.
line wavelength (nm)
T eff effect 402.6 nm 447.1 nm 492.2 nm
15000 NLTE 1.2 4.3 7.4
turb. 0.3 2.6 0.2
20000 NLTE 16.0 4.3 4.6
turb. 2.3 2.9 0.4
30000 NLTE 29.0 6.7 5.7
turb. 3.1 4.3 0.8
4. Helium abundance and evolutionary state
A correlation between the He abundance and the surface gravity was suspected by Glagolevskij et al. (1992) and by
Zboril et al. (1994), and one purpose of this paper is to confirm it on a larger sample. The results of Table 4 are shown

12 M. Zboril et al.: Properties of Herich stars. I. Evolutionary state.
Fig. 7. bfactors of 2 3 S (solid line), 2 3 P (dotted line), 2 1 P (dashed line) and 4 1 D (dash dot) neutral helium levels respectively
Fig. 8. bfactors of 2 3 S (solid line), 2 3 P (dotted line), 2 1 P (dashed line) and 4 1 D (dash dot) neutral helium levels respectively

M. Zboril et al.: Properties of Herich stars. I. Evolutionary state. 13
in Figure 9a and the suspected correlation is not confirmed. The regression lines have been fitted taking into account
errors on both axes, the errors on n(He) being on average three times smaller than those on R=R z . The statistical
significance of the correlation is not great: the Spearman rank correlation coefficients are 0.615 and 0.174 respectively
for the black and open dots, corresponding to a Student t test of 2.92 and 0.65, i.e. roughly 98% and 50% probabilities
that the correlation is real. The range in R=R z is rather small for the open dots (the most reliable in principle),
which represent the values obtained for the surface gravity by imposing the photometric T eff values, as explained
above. We have two extremely Herich stars, CPD \Gamma62 ffi 2124 and HD 124448 which, however, present emission at
least in their Balmer lines so that their helium abundances might be less reliable. HD 124448 being quite atypical
of the intermediate He stars, it has been excluded from the above regression. CPD \Gamma62 ffi 2124 was also excluded. On
the other hand, several other stars show some emission too but estimates of atmospheric parameters from Geneva
photometry and hydrogen profiles are quite consistent. Helium abundances of normal reference stars are quite typical
of the abundances in normal stars.
Our data do not allow us to tell definitely whether the ``Herich'' phenomenon is uniformly distributed on the whole
main sequence width, or not. If we rely entirely on the fits of the Balmer line profiles, then it seems the first term of
the alternative holds true, especially if we remember that the log g values are systematically too large by about 0.2
dex. If, on the other hand one relies on the photometric data, corrected for the effect of the He overabundance on the
photometric T eff determination, then Figure 9a indicates that Herich stars are confined in the vicinity of the ZAMS.
The latter conclusion depends on wether the synthetic colours computed in Section 3.2 are realistic or not.
The relation between He abundance and position of the star in the HR diagram is quite interesting with regard
to the theory of radiatively driven winds in B stars (Babel 1995, 1996). According to Figure 6 of Babel (1996), the
Herich stars have relatively large, homogeneous winds which permit He to be overabundant in their photospheres in
spite of the small radiative acceleration. Since the wind is larger when the luminosity is larger, one can qualitatively
expect the He abundance to be larger as well, so a correlation between n(He) and log L=L fi has been looked for (the
luminosity has been obtained from the ages.f code kindly provided by F. Figueras and C. Jordi, which interpolates
theoretical evolutionary tracks for given T eff and log g): Figure 9b shows that no such correlation exists. However, if
the wind is too large, He will leave the photosphere and will not be overabundant any more. The apparent lack of
Herich stars with small surface gravities may be significant in this respect, but can only be suggested with our data.
The sample is small and the uncertainties on the estimated surface gravities are large, making a clearcut conclusion
difficult.
5. Spectroscopic variability
The catalogue of Renson et al. (1991) gives the period of variability when it is known, with the notation 'S' indicating
the presence of spectroscopic variations (i.e. intensity of spectral lines), 'V' indicating radial velocity variations, 'M'
magnetic field and 'L' luminosity and/or colour variations. For HD 37479, the database reports VSLMtype variability
with a 1.191 days period; for HD 58260 there is S variability with a 1.66 days period; for HD 37776 there is SLMV
variability with a 1.539 days period and HD 37017 varies (SLM) with a 0.9012 days period. But, at least in helium
line profiles, we were unable to detect any variability in our programme stars.
6. Emission
Some of our programme stars display a kind of emission, preferably in the H fi line profile. This feature was even
detected in two reference standard stars and probably indicates the presence of a dense wind. Figure 10 displays the
emission in H fi for all stars where it is clearly present. Synthesized profile for HD 56139 was computed using LTE
Kurucz (1992) models with a chromospheric structure which basically follows the expression
T (dm) = TKur (dm) + T chrom: (dm) (7)
where dm is the mass depth variable and T the temperature. The chromosphere was added to the photospheric
model following roughly the solar model, i.e. for mass depth variable less than 0.01 there is a ``chromospheric'' tem
perature rise, followed by a plateau and by another, coronalike rise in temperature. The purpose was not to find the
temperature structure in such stars, but rather to give an idea how different the T () relation maybe in their high
atmosphere, compared with those stars showing no emission. The observed profile of HD 56139 indicates agreement
in the far line wings, while the line core suggests a different kind of atmospheric structure. There is also a redshift
observed here. Other stars indicate various levels of activity in the atmosphere. In the case of HD 105435 we probably
meet saturation and/or dynamical range effect since instrumentation was set up for absorption. Hydrogen line profile
of HD 37479 demonstrates that a kind of emission is recognized in Geneva photometry (from a comparison with the fi

14 M. Zboril et al.: Properties of Herich stars. I. Evolutionary state.
Fig. 9. a.Helium abundance versus evolutionary state for the Herich stars, with purely spectroscopic surface gravities (Full
dots). The open dots show the same, but with radii and helium abundances estimated on the basis of photometric values of T eff .
The solid line has been fitted to the full dots, allowing for errors in both axes. The dotted line is the same for the open dots.
b. Relation between He abundance and luminosity (same symbols as above).
index observed in uvbyfi photometry) while the line profile is almost regular. A model atmosphere study of these stars
with emission will be published elsewhere. We restrict ourselves here only by formulating a statement that hydrogen
beta line profile resembles more a Be phenomenon than a Bshell one. Studies of early type stars with emission indicate
these phenomena may even occur and disappear in one and the same star and are therefore time dependent: likewise,
we are able to detect marginal variability in Hfi in HD37479 at about the 15 percent level, the Betype emission being
always present. Possible variability in the case of HD 124448 is no more than 2 percent, if any. For other stars with
emission in Hfi we have unfortunately single records. The star CPD \Gamma62 ffi 2124 deserves a special comment, since the
Balmer lines are seen in emission not only in the star but also in the surrounding sky. Therefore the origin of the
emission lies in a nebula which not necessarily surrounds the star but may be in the foreground or in the background;
there is no reason to believe that the emission is taking place in any circumstellar material.
7. Conclusions
The abundance analysis of 24 (19 Her and 5 standard) stars was carried out using spectrum synthesis computations to
evaluate primarily helium abundance and its possible link with the evolutionary state, which was suspected in previous
papers. Helium abundances were obtained using current Kurucz models. The trend found in our previous papers, i.e.
an increase of the helium abundance with decreasing log g or with increasing radius, cannot be confirmed. If such a
trend existed at all, it would be extremely loose in any case. The radius was obtained from the surface gravity through
the formula R 1= p
g. Since we could use three hydrogen line profiles, we could determine the surface gravities with
a better reliability. However, an unexpected problem arose in the scale of the effective temperatures, which is 1300 K
larger for photometric than for spectroscopic estimates. Taking into account the effect of the helium overabundance
on the Geneva colours as well as on the computed Balmer line profiles resolves this discrepancy. The method we have

M. Zboril et al.: Properties of Herich stars. I. Evolutionary state. 15
Fig. 10. H fi line profile for stars in Table 4. Solid line: synthesized profile from a model atmosphere with ``chromosphere''
and ``corona'', and a photospheric structure corresponding to an LTE Kurucz model atmosphere (1992) of the star HD 56139.
dotted line: observed profile for the standard star HD 56139. dashed line: observed profile for HD 105435. dash dot dot
dot: observed profile for HD124448. long dash: observed profile for HD 37479. See text.
used here for determining the radius of the star relative to its ZAMS value is more precise than that used in our
previous papers, where we relied on the definition of the luminosity of a star: L=4R 2 oeT 4 resulting in the following
expression for the radius
log R = 8:46 \Gamma 2log T eff \Gamma 0:2M bol (8)
if corresponding solar quantities are taken account. However, to obtain the bolometric magnitude, we relied on the fi
index, which may be quite peculiar for many Herich stars (emission in H fi ). Furthermore, the Herich stars lie on the
boundary of the c1 \Gamma Hfi relation, where the calibration fi M v (fi) may not be valid. Besides, another calibration
step has to be done to obtain the effective temperature and bolometric magnitude. Finally, any error on M bol enters
linearly into the above formula.
The direct comparison with previous reports on the evolutionary state of the Herich stars (Zboril et al. 1994) is
not easy. The atmospheric parameters estimates were based on pure photometric methods and Geneva photometry
was only one of them. Thus both R/R z and helium abundances may have been blurred. We also used the relation
given above. Several stars observed here were analysed in our previous paper but they cover here a narrower R/R z
interval. This trend holds for all programme stars. The helium abundance interval, however, corresponds very well
with the previous paper. Only very active stars in this paper possess very high helium abundance ( 0.60.8) based on
Kurucz models. Synthesized Geneva colours for nonsolar He abundances show a systematic trend, especially for the
[U \Gamma B] index which decreases by about 0.025 mag per 0.1 He abundance. The effect decreases with increasing effective
temperature. This in turn leads to an overestimate of the effective temperature if standard photometric calibrations
(based on solarabundance Kurucz models) are used. We can expect also the hydrogen line profiles may be affected.
They depend nonlinearly on the three quantities T eff , log g, n(He) and since we do not know apriori n(He), this fact
may explain a systematic shift compared with theoretical evolutionary tracks.
The comparison of our equivalent width measurements with Walborn's, we found considerable discrepancies, which
may reach up to 40%. The S/N of Walborn's spectra is in the range 1020. Modelling hydrogen profiles and adding
this S/N to the theoretical profiles we met the following effect: the equivalent width is increased due to low S/N

16 M. Zboril et al.: Properties of Herich stars. I. Evolutionary state.
and increasing interval for locating the continuum level in such noisy data. This alone leads to a 3040% increase of
the equivalent width. If we add the scaling problem Walborn met, it is possible to explain the difference between our
measurements and Walborn's, at least in the case of hydrogen profiles shown in figure 1. As helium lines have a smaller
equivalent width, one may also expect a larger scatter (in relative value) for a given S/N.
Acknowledgements. MZ wishes to acknowledge gratefully an UK PPARC grant support. PN thanks Dr. Gautier Mathys and
the night assistants at ESO La Silla for their help, and the Swiss National Science Foundation for its support. Some photometric
measurements made in the Geneva system at La Silla by Dr. Michel Burnet are also gratefully acknowledged.
References
Babel J. 1995, A&A 301, 823
Babel J. 1996, A&A 309, 867
Barnard A.J, Cooper J., Smith E.W., 1974, J. Quant. Spectrosc. Radiat. Transfer 14, 1025
Bohlender D.A., Landstreet J.D., Brown D.N., Thompson I.B., 1987, ApJ 323, 325
Cramer, N., Maeder, A., 1979, A&A 78, 305
Dimitrijevic M. S., SahalBrechot S. 1984, J. Quant. Spectrosc. Radiat. Transfer., 31, 301
Glagolevskij Yu. V. 1990, in''Hot chemically peculiar and magnetic stars'', Proceedings of 8th subcommission ''Magnetic stars'',
eds. G. Scholz, PotsdamBabelsberg, Germany, p.62
Glagolevskij Yu. V., Topilskaya G. P., Kartashova T. A. 1992, in ''Stellar Magnetism'', eds. Yu. V. Glagolevskij and I. I.
Romanyuk, Russian Academy of Sciences, ''NAUKA'', SanktPetersburg, p.62
Hubeny I., Lanz T. 1992, A&A, 262, 171
Kilian J. 1992, A&A, 262, 501
Kurucz R.L. 1992, Rev. Mex. Astron. Astrofis. 23,45
Kunzli, M., North P., Kurucz, R.L, Nicolet, B., 1996, A&AS, in press
Michaud G., Dupuis J., Fontaine G., Montmerle T., 1987, ApJ 322, 302
Mihalas D., Heasley J. N.A, Auer L. H. 1975, NCARTN/STR104
North, P. and Paltani, S. 1994, A&AS, 288, 155
Pedersen H., 1979, A&AS 35, 313
Pedersen H., Thomsen B., 1977, A&AS 30, 11
Prinja R.K., Barlow M.J., Howarth I.D. 1990, ApJ, 361, 607
Renson P., Gerbaldi M., Catalano F.A., 1991, A&AS, 89, 429
Rufener F., Nicolet B., 1988, A&A 206, 357
Shamey P., 1969, PhD thesis, unpublished
Schaller G., Schaerer D., Meynet G., Maeder A., 1992, A&AS, 96, 269
Schild R., Peterson D. M., Oke J. B., 1971, ApJ 166,95
Schoening T., Butler K. 1989, Astron. Astrophys. Suppl. Ser., 78,51
Schonberner, D., Wolf, R.E.A., 1974, A&A 37, 87
Slettebak A., Collins II G.W., Boyce P.B., White N.M., Parkinson T.P., 1975, AJS 29, 137
Vauclair S., 1975, A&A 45, 233
Vauclair S., Dolez, N., Gough, D.O., 1991, A&A 252, 618
Walborn N.R., 1983, ApJ 268, 195
Zboril M., Glagolevskij Yu. V., North P. 1994, in ''International Conference on CP and Magnetic Stars'', Tatransk'a Lomnica,
eds. J. Zverko and J. Ziznovsk'y, p. 105
Zboril M. 1996, ASP Conf. Ser., in press
This article was processed by the author using SpringerVerlag L A T E X A&A style file LAA version 3.