Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://star.arm.ac.uk/~ambn/300csj.ps Äàòà èçìåíåíèÿ: Tue Feb 9 17:46:10 1999 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 21:15:10 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: barnard 68

A&A manuscript no.
(will be inserted by hand later)
Your thesaurus codes are:
06 (08.01.1; 08.03.2; 08.09.2 LSS3184)
ASTRONOMY
AND
ASTROPHYSICS
Spectral analysis of the extreme helium star LSS 3184 ?
J.S. Drilling 1 , C.S. Jeffery 2 , and U. Heber 3
1 Dept. of Physics and Astronomy, Louisiana State University, Baton Rouge, Louisiana 70803, USA.
2 Armagh Observatory, College Hill, Armagh BT61 9DG, Northern Ireland.
3 Dr. Remeis­Sternwarte, Universit¨at Erlangen­N¨urnberg, D96049 Bamberg, Germany.
Received 17 October 1996; accepted 14 August 1997
Abstract. LSS 3184 is a hydrogen­deficient, early B­type
giant, recently found to pulsate with a period of 2.5 hours.
Its photospheric parameters have been derived from opti­
cal high­resolution spectra by the method of fine analysis.
The principal results are T eff = 23 300 K, log g = 3:35,
nH = nHe ! ¸ 0:00015, nC=nHe = 0:003, nN=nHe = 0:0005,
and nO = nHe = 0:0003. Hydrogen is extremely deficient.
The effective temperature is consistent with broad­band
visual and ultraviolet spectrophotometry and an extinc­
tion EB\GammaV ¸ 0:27. Its previous evolution is reflected in
the chemistry of the atmosphere, which contains enriched
nitrogen from CNO­cycle hydrogen burning, and carbon
from 3ff helium burning. Thus LSS 3184 is a true extreme
helium star with a composition similar to BD\Gamma9 ffi 4395.
With T eff , log g, and pulsation properties very similar
to the C­poor and N­rich helium star V652 Her, evolution­
ary mechanisms which can result in very different surface
compositions for the two stars must be examined.
Key words: stars: helium -- individual (LSS 3184) -- abun­
dances -- chemically peculiar
1. Introduction
The star LSS 3184 was found to be hydrogen­deficient by
Drilling (unpublished) during a spectroscopic survey of
OB + stars in the Southern Milky Way (Drilling 1980). It
was subsequently observed photometrically and the broad­
band colours were used to estimate the reddening, EB\GammaV =
0:26, and to derive an effective temperature T eff = 21 700 K
(Heber et al. 1986). A preliminary fine analysis by Heber
et al. (1986) yielded T eff = 22 000 K and log g = 3:2,
and showed nitrogen to be overabundant by a factor of
Send offprint requests to: J.S. Drilling
? Based on observations obtained at the European South­
ern Observatory, La Silla, Chile, and with the IUE satellite
retrieved from the IUE archive at the World Data Centre,
Rutherford Appleton Laboratory, UK
about 50. The purpose of the present paper is to give
the final results and details of the fine analysis, and to
present unpublished results on low­resolution ultraviolet
spectroscopy.
2. Observations
A blue­visual spectrum of LSS 3184 was obtained using
CASPEC, a Cassegrain 'echelle spectrograph, on the ESO
3.6m telescope at La Silla, Chile, at approximately UT
02:00, 1985 April 8. The 3600 second exposure covered
the wavelength range from 3800 to 4800 š A at a resolution
of 0.5 š A and a S/N of 30. The data reduction has been
described by Heber et al. (1986). Low­resolution spectra
(image numbers LWP 8966, LWP 8967, and SWP 29087)
obtained with IUE on 1986, August 30 and extracted from
the Uniform Low Dispersion Archive at the Rutherford
Appleton Laboratory have also been used in this investi­
gation. The SWP data have been corrected for sensitivity
degradation of the camera and geocoronal Lyff emission
has been removed. The LWP data have been filtered to re­
move noisy data. An inspection of the CASPEC spectrum
shows immediate similarities with the well­studied spec­
trum of the EHe star V652 Her = BD+13 ffi 3224 (Jeffery
& Hill 1986; Jeffery et al. 1986), and the line identifica­
tions for this star proved very helpful in completing the
line identifications for LSS 3184.
Equivalent widths ( W– ) were measured for all lines
identified. Measurements were made relative to a local
continuum defined by small regions of spectrum between
absorption lines within \Sigma20 š A of the line being measured,
this being of similar dimension to a single CASPEC 'echelle
order. Equivalent widths were measured both by direct in­
tegration and by integrating a Gaussian fitted to the line
profile after cleaning local defects such as cosmic rays or
weak blends in one or other wing. Both measures nor­
mally agreed to within \Sigma10%. For close blends due to the
same ion, the equivalent width of the blend was measured.
Where blends could be resolved into two or more compo­
nents, a multiple Gaussian was fitted and the equivalent

2 J.S. Drilling et al.: Spectral analysis of the extreme helium star LSS 3184
Fig. 1. The adopted model atmosphere for LSS 3184 is shown
relative to the loci of ionisation equilibria and He i line­profile
fits in the f T eff , log gg plane. The ionisation equilibria for
He i/ii, C ii/iii, N ii/iii, Si ii/iii and Si iii/iv are shown as
broken lines. The best fits to the HeI line profiles are shown as
dotted lines. Solid diamonds indicate the grid of line­blanketed
model atmospheres.
width of each Gaussian was recorded. This procedure was
successful in reproducing the equivalent width of the entire
blend and the relative strengths of the individual compo­
nents. After the analysis had been completed, we found to
our great embarassment that we had obtained a second,
similar spectrum at UT 03:38 on April 8. This spectrum
provided us with an external check, and measurements of
selected lines agreed to within \Sigma10%. The principal source
of error remains continuum placement. A few lines were
measured independently by two authors, again agreeing
to within \Sigma10%.
3. Photospheric parameters
The procedure adopted here has been to determine T eff
from the ionisation equilibria of prominent ions, along
with other atmospheric properties, and to verify subse­
quently that these are consistent with the continuous flux
distribution.
The LTE line­blanketed model atmospheres and the
line­formation codes used in this analysis are described
by M¨oller (1990) and by Jeffery & Heber (1992). The
important feature of these models is that line­blanketing
is treated using opacity distribution functions calculated
from the Kurucz & Peytremann (1975) list of 265 000 lines,
in the same manner as adopted in Kurucz's stellar atmo­
sphere program ATLAS6 (cf. Kurucz 1979). Previously,
Heber et al. (1986) had considered the blanketing effect
of 800 He i and metal lines. The continuous opacities are
nearly the same as those adopted by Kurucz (1979), with
the addition of carbon and nitrogen from Peach (1970).
A plane­parallel approximation is justified since the den­
sity scale height in the atmosphere is still small (! 2%)
Fig. 2. Theoretical models for observed He i line profiles in
LSS 3184 calculated for model atmospheres with nH = 0:0001
and with T eff = 24 000K; log g = 3:0; nC = 0:005 (top),
T eff = 23 300K; log g = 3:35; nC = 0:003 (middle), and
T eff = 22 000K; log g = 4:0; nC = 0:005 (bottom). Histogram:
CASPEC spectrum (1985 April 6). The theoretical profiles
have been convolved with an instrumental broadening profile
(FWHM = 0.2 š A).

J.S. Drilling et al.: Spectral analysis of the extreme helium star LSS 3184 3
Fig. 3. Theoretical flux distribution for T eff = 23 300K,
log g = 3:35 and EB\GammaV = 0:27 compared with spectropho­
tometric observations of LSS 3184. Histogram: IUE spectrum.
Filled circles: BV photometry.
Fig. 4. The log g \Gamma log T eff diagram for EHe ( ), stars, showing
the positions of LSS 3184, the hydrogen main sequence (H­MS),
and the classical (e \Gamma ­scattering) Eddington limit for radiative
stability. L/M loci (dashed) and pulsation instability bound­
aries (Saio 1995: dotted) are also shown.
compared to the stellar radius, even for the lowest gravity
models considered.
Since the hydrogen abundance ( nH ) in LSS 3184 is
small and T eff is to be obtained from the ionisation equi­
librium alone, the procedure for obtaining a model at­
mosphere for LSS 3184 involves only five free parameters,
being T eff , the surface gravity ( log g), the microturbu­
lent velocity ( v t ), the carbon abundance ( nC ), and the
nitrogen abundance ( nN ).
3.1. Microturbulence
The microturbulence ( v t ) determined by constraining the
abundances derived from numerous lines of C ii, N ii, and
O ii to be independent of equivalent width was v t =
15 km s \Gamma1 . The uncertainty in this figure is ¸ \Sigma5 km s \Gamma1 .
The adiabatic sound­speed in the model atmosphere at
the depth where the majority of these lines is formed is
12--13 km s \Gamma1 . v t would appear to be slightly supersonic,
but it is noted that similar and higher values were required
in analyses of other helium stars (Jeffery & Heber 1992,
1993, Jeffery 1993). This is a reminder that the notion
of microturbulence is useful in accounting for well­known
discrepancies between theoretical and observed stellar at­
mospheres, but has a limited physical significance. The
measured v t also accounts for the N abundance being
lower than that found by Heber et al. (1986) who adopted
v t = 5 km s \Gamma1 . After applying the measured instrumental
broadening of 0.2 š A (FWHM) to several strong O ii line
profiles, it was apparent that no further line­broadening
could be measured, such as might be due to rotation or
pulsation. Given that each spectrum was exposed for ¸ 0:4
cycles, a maximum line broadening of ¸ 15 km s \Gamma1 would
be superimposed by the radial pulsation (Jeffery & Hill
1996a). Since the question remains why such a large value
for v t is required in both BD\Gamma9 ffi 4395 and LSS 3184, the
fact that both are high luminosity pulsators suggests there
may be some correlation between pulsation and microtur­
bulence.
3.2. Ionisation equilibrium
The ionisation equilibria of C ii/ C iii, N ii/ N iii, and
Si ii/ Si iii/ Si iv were used to determine T eff as a function
of log g. In contrast to previous analyses (cf. Jeffery &
Heber 1992), the Si iii multiplet 2.00 lines gave results
consistent with Si iii multiplet 9.00, suggesting that at
the higher T eff and log g of LSS 3184, these lines can be
treated successfully in LTE.
The He ii –4685.7 š A line could be measured with an
equivalent width of 67 mš A. This was well below the strength
expected for the T eff indicated by other ions. The line
shape appears to be approximately triangular, and noti­
cably different from other isolated absorption lines and is
similar in both CASPEC spectra. However there is no sat­
isfactory explanation for this discrepancy in line strength.
3.3. Surface gravity
Line profiles were calculated for He i –4009 š A,
He i –4026 š A, He i –4121 š A, He i –4387 š A, and He i –4471 š A
in order to determine log g as a function of T eff . The non­
diffuse line He i –4121 š A is blended with nearby lines and
relatively insensitive to surface gravity, so was not con­
sidered further. The coincidence of the He i line fits and
the ionisation equilibria for C ii/iii and Si II/III/IV was
used to determine the overall model parameters (Fig. 1).
The N ii/iii equilibrium shown in Fig. 1 is based on four
weak N iii lines, with implied abundances spread over 1

4 J.S. Drilling et al.: Spectral analysis of the extreme helium star LSS 3184
dex, and is considered unreliable. The N ii lines, how­
ever, provide an excellent measure of the N abundance,
nN = n He = 0:0005. At T eff = 23 300 K and log g = 3:35,
the fits for He i –4471 š A, He i –4387 š A, He i –4009 š A and
He i –4026 š A (Fig. 2) are in excellent agreement with the
observations. However it is noted that at the gravity and
temperature of LSS 3184, the density in the atmosphere
where the helium lines are formed is at the upper limit of
the available line­broadening tables (Barnard et al., 1969,
1974, 1975). Publication of the excellent extension of He i
broadening calculations to higher and lower densities and
to additional lines by Beauchamp et al. (1996) is eagerly
anticipated.
3.4. Atmospheric parameters
The atmospheric parameters obtained for LSS 3184 are
thus T eff = 23 300 \Sigma 700 K, log g = 3:35 \Sigma 0:1, nC =
0:003 \Sigma 0:001, nN = 0:0005 \Sigma 0:0002, and v t = 15 \Sigma 5km=s.
These, with nH=0.0001 and solar abundances for other
atomic species, were adopted to calculate the final model
atmosphere used to derive the remaining properties. The
location of LSS 3184 in the fT eff , log gg plane is shown in
Fig. 4. For a star with this gravity, log(L=M ) = 3:52 \Sigma 0:15
(solar units). Assuming that the stellar interior consists
of a degenerate CO core surrounded by a luminous He­
burning shell, a core mass M ú 0:55 M fi may be estimated
(Jeffery 1988), but this is at the extreme lower­limit of
validity of the core­mass shell­luminosity relation.
3.5. Spectrophotometry
Having obtained the atmospheric parameters for LSS 3184
by the method of fine analysis, it remains to check that
T eff is consistent with the spectrophotometry. All IUE
low­resolution images were extracted from the Uniform
Low Dispersion Archive and an average spectrum weighted
by the respective exposure times was formed. There is no
significant variation in flux between images obtained with
the same camera at different times. The flux­calibrated
spectrum and the published Johnson B and V magnitudes
(Drilling & Hill, 1986) were compared with the model flux
distributions reddened by varying amounts. For T eff =
23 300K and log g = 3:35, a reddening of EB\GammaV = 0:27 \Sigma
0:02 gave the best agreement between model and observa­
tion, as shown in Fig. 3. A variation of \Sigma0:05 in EB\GammaV cor­
responds roughly to a variation of \Sigma2000 K in T eff . Small
discrepancies between theoretical and observed fluxes may
be due to insufficient line opacity (1600 š A), noise (2000 š A)
and a non­standard width in the 2175 š A interstellar band
(2400 š A).
3.6. Photospheric abundances
The line abundances for other ions were obtained by com­
paring equivalent widths with individual curves of growth.
In general, only lines found to be largely free from blends
are shown in Table 1. In addition, for carbon lines, a crit­
ical selection excluded all lines which were either very
strong (non­LTE), very weak (measurement errors), con­
taminated by any blend or cosmic ray, or for which there
were uncertainties in the atomic data. The individual line
abundances are shown in Table 1. The mean abundance
for each ion is given with the standard deviation about
the mean.
All three Balmer lines in our spectra are blended with
lines of S, Si, and C, respectively. An upper limit for the
hydrogen abundance of nH ! ¸ 0:0006 \Sigma 0:0003 is obtained
assuming no contribution from the blending lines. For Hfl,
this limit can be reduced to nH ! ¸ 0:00015 by considering
the S iii 4 component of the blend.
The final photospheric abundances are shown in Ta­
ble 2, alongside results obtained for V652 Her (Jeffery
et al. 1986), for BD\Gamma9 ffi 4395 (Jeffery & Heber 1992), for
HD144941 (Harrison & Jeffery 1996), and for the Sun (An­
ders & Grevesse 1989). Where data are available for more
than one ion of a species in Table 1, these have been com­
bined to form the mean values given in Table 2.
3.7. Time­dependent behaviour
The preliminary surface parameters derived by Heber et
al. (1986) for LSS 3184 were similar to those of the pul­
sating helium star V652 Her, prompting Saio (1995) to
predict that LSS 3184 should also pulsate with a period
of 0.1 days, as subsequently verified by Kilkenny & Koen
(1995). Since the pulsation period of LSS 3184 (¸ 155min)
is fairly short compared with the exposure times of our
spectra (60min), the effects of pulsation on the measured
line profiles and the assumption of hydrostatic equilibrium
in the model atmospheres need to be considered.
A preliminary measurement of the velocity curve (Jef­
fery & Hill 1996a) indicates that line broadening over 60
min. can not be much more than 15 km/s, which is equiva­
lent to the instrumental broadening. Further observations
to estimate the amplitude of the temperature variation
are in progress. Studies of the larger amplitude variable
V652 Her (Jeffery & Hill 1986) indicate that the hydro­
static approximation provides a sufficient description of
the stellar atmosphere through 90% of the pulsation cycle.
At minimum radius, rapid outward acceleration of the at­
mosphere increases the effective surface gravity by ¸ 1dex.
Such rapid acceleration is not seen in LSS 3184 (Jeffery &
Hill 1996a), and thus the atmosphere is assumed to be in
hydrostatic equilibrium throughout the pulsation cycle.

J.S. Drilling et al.: Spectral analysis of the extreme helium star LSS 3184 5
Table 3. Systematic errors in the mean ion abundances. The measured errors (ffi) in each of T eff , log g, v t , nC and W– are
shown first. The partial abundance derivative with respect to each of these variables is shown beneath, for each ion. Assuming
that the errors are independent and random, then the total error (ffi(log n)) shown on the right may be compared with the
standard deviation (oe(log n)) in the derived mean abundances.
ffi log T eff ffi log g ffi log v t ffi log nC ffi log W–
0.01 0.1 0.12 0.2 0.05
Ion @ log n
@ log T eff
@ log n
@ log g
@ log n
@ log v t
@ log n
@ log n C
@ log n
@ log W –
ffi(log n) oe(log n)
H 2.82 ­0.15 0.00 0.00 3.08 0.16 0.29
C ii 1.73 ­0.09 ­0.33 ­0.03 2.67 0.14 0.05
C iii ­15.84 0.90 ­0.61 ­0.19 ­8.97 0.49 0.16
N ii 0.16 0.06 ­1.12 0.02 2.12 0.17 0.25
N iii ­16.71 0.96 ­0.54 ­0.21 ­8.79 0.49 0.49
O ii ­5.20 0.34 ­0.65 ­0.05 ­2.16 0.15 0.35
Mg ii 4.76 ­0.17 ­0.66 0.03 4.84 0.26 0.11
Al iii 1.92 0.00 ­0.80 0.01 3.08 0.18 0.09
Si ii 13.10 ­0.57 ­0.65 0.15 11.24 0.59 0.16
Si iii ­3.01 0.24 ­0.84 ­0.02 ­0.49 0.11 0.18
Si iv ­16.41 1.04 ­0.82 ­0.18 ­9.66 0.53 0.25
P iii ­0.66 0.23 ­0.26 0.00 0.76 0.05 0.11
S iii ­4.63 0.42 ­0.47 ­0.04 ­2.03 0.13 0.25
Fe iii ­1.59 0.22 ­0.25 ­0.01 0.08 0.04 0.31
3.8. Errors
Uncertainties in the final abundances arise from a number
of possible sources including the placement of the con­
tinuum in measuring line profiles and equivalent widths,
errors in T eff , log g, v t and nC for the final model, errors
in atomic data used for the analysis of individual lines, a
failure to include all continuous and line opacity sources
from the model atmospheres, the neglect of convection and
the assumptions of local thermodynamic and hydrostatic
equilibrium and of plane­parallel geometry.
The derivatives of the mean ion abundances with re­
spect to T eff , log g, v t , nC and W– have been established
numerically and are shown in Table 3. Assuming that all
five variables are independent, which they are not, the
maximum error in each ion abundance can thence be es­
timated. For the majority of ions, these errors are smaller
than the standard deviation due to several lines belong­
ing to the same ion. It is noted that the largest errors
are due to temperature sensitive lines which themselves
define the measurement of T eff . Of other sources, errors
in atomic data contribute in the same way as errors in
W– , and are random errors associated with each multi­
plet. W– measurement errors are random errors associ­
ated with each line, except that continuum or background
definition may produce a systematic error whose contri­
bution is represented by @ log n
@ log W– in Table 3. Assumptions
made in constructing the model atmosphere may also lead
to systematic errors, since these affect where and how the
lines are formed. Quantifying these will require the con­
struction of new model atmospheres which have not yet
been successfully attempted.
4. LSS3184 and other EHe stars
The comparison provided by Table 2 demonstrates that
LSS 3184 has a composition which is similar to that of the
extreme helium star BD\Gamma9 ffi 4395. The H, He and CNO
abundance signature is indicative of material which has
been almost entirely CNO­processed (destroying H) and
partially 3ff processed (producing C). The magnesium,
aluminium, silicon, phosphorous and sulphur abundances
are subsolar by ¸0.4 dex, whilst iron is subsolar by ¸1
dex. The extraordinary surface abundances of extreme
helium stars have been discussed most recently by Jef­
fery (1996), where a comparison of all analyzed helium
stars may be found. The fact that in LSS 3184 nitrogen
is not enriched (by CNO processing) is superficially diffi­
cult to understand. Although nitrogen is destroyed by 3ff
burning, the admixture of carbon in LSS3184 is relatively
small, as indicated by the carbon abundance. Adopting a
subsolar primordial metallicity (­0.5 dex, say) provides a
natural solution since carbon is then enriched by 0.9 dex,
nitrogen by 0.7 dex and oxygen is depleted by 0.4 dex.
In one sense LSS 3184 is slightly unusual because sev­
eral EHes show a significant and inexplicable overabun­
dance of phosphorous (Jeffery 1996) which is absent here.
However, it should be remembered that amongst the EHes
there are large star­to­star variations in the abundances
of individual species. Therefore, in view of its abundances,
LSS 3184 may be properly considered to be a true extreme
helium star and carries significance as having the highest
gravity and lowest L=M ratio of any EHe studied so far.
If a core­mass luminosity relation can legitimately be in­

6 J.S. Drilling et al.: Spectral analysis of the extreme helium star LSS 3184
voked for EHe stars (cf. Jeffery 1988, Saio 1988), it must
be the lowest mass member of the group.
LSS 3184 has T eff and log g very similar to two other
stars with nH ! ¸ 0:01, namely V652 Her as mentioned
above (Lynas­Gray et al. 1984, Jeffery et al. 1986) and
HD 144941 (Harrison & Jeffery 1997) (Table 2). It had pre­
viously been supposed that these two stars were intrinsi­
cally different from other EHe stars, both because they had
comparatively high hydrogen abundances ( nH ¸ 0:01)
and because of their lower L=M ratios. Moreover their
CNO and other abundances are strikingly different from
those of other EHes. It was consequently proposed that
V652 Her (at least, cf. Jeffery 1984) had a different in­
ternal structure and evolutionary origin. However, with
LSS 3184 having similar T eff and log g to V652 Her and
HD 144941, and the R CrB star DYCen (Jeffery & Heber
1993) having a higher hydrogen abundance, such a hy­
pothesis ceases to be defensible. The evolutionary pro­
cess or processes (cf. Iben et al. 1996) which create EHes
must be able to create stars with luminosities ranging from
3:0 ! ¸ log L=M ! ¸ 4:5, hydrogen abundances ranging from
! \Gamma4 ! ¸ log nH ! ¸ \Gamma0:8, and a considerable variation in
the abundances of all major atomic species.
All three stars just discussed (LSS 3184, V652 Her and
HD144941) are located in a high­gravity finger in g \Gamma
T eff space where radial pulsations driven by He + ioniza­
tion occur because of high metal opacities around 10 6 K
(Saio 1995, Fig. 4). Such pulsations are predicted to oc­
cur only if the metal abundance is sufficiently high. In
the case of LSS 3184 our measurements of the metallic­
ity give Z ¸ 0:007, if we consider those species which
contribute most to the opacity at 10 6 K. Thus LSS 3184
provides a crucial test of the opacity calculations; with
Z ¸ 0:01, models state that LSS 3184 should pulsate,
with Z ¸ 0:004 it should not. Certainly the amplitude
of pulsations in LSS 3184 is much lower than in the more
metal­rich V652Her whilst the absence of pulsations in
HD 144941 (Jeffery & Hill 1996b) has already been demon­
strated to be due to its very low metallicity (Z = 0:0003,
Harrison & Jeffery 1997). Therefore a precise measure­
ment of the iron abundance from UV spectra and the con­
struction of theoretical pulsation models using opacities
tailored to the observed abundances would provide very
powerful diagnostics both of stellar structure and atomic
physics.
5. Conclusions
The analysis shows that LSS 3184 has a high carbon abun­
dance typical of other EHes. It is extremely hydrogen de­
ficient, hydrogen being present by less than one part in
10 4 by numbers. The evidence of the nitrogen and other
metal abundances implies that the primordial metallic­
ity of LSS 3184 was subsolar by ¸ \Gamma0:5 dex, but that
nitrogen and carbon have been enriched by CNO and
3ff processing. The low primordial metallicity may be re­
flected in the lower amplitude of LSS 3184 pulsations rela­
tive to V652 Her, which are driven by iron­group opacities
at ¸ 10 6 K.
Acknowledgements. The authors are grateful to the referee,
Bertrand Plez, for his remarks, and to UK PPARC's CCP7
for finance and software.
Appendix A: Additional line measurements
A list of measured equivalent widths for identified lines which
were not included in Table 1 is given in Table 1. Although ex­
haustive searches were made using published linelists, a signif­
icant number of lines, some relatively strong ( W– ¸ 100m š A),
eluded identification. A list of these, with laboratory wave­
lengths accurate to \Sigma0:1 š A, is given in Table 2. At least two
occur in the spectrum of LSE 78 (Jeffery 1993), and others also
occur in spectra of the helium stars BD+10 ffi 2179 and V348 Sgr
(Leuenhagen et al. 1994). Given that many lines which are nor­
mally not seen or are very weak in hydrogen­rich stars are ob­
served in LSS 3184, it is likely that many belong to unclassified
transitions of C ii, N ii, and O ii.
References
Allard N., Artru M.­C., Lanz T., Le Dourneuf M., 1990, AAS
84, 563
Anders E., Grevesse N., 1989, Geochim. Cosmochim. Acta 53,
197
Barnard A.J., Cooper J., Shamey L.J., 1969, AA 1, 28
Barnard A.J., Cooper J., Smith E.W., 1974, JQSRT, 14, 1025
Barnard A.J., Cooper J., Smith E.W., 1975, JQSRT, 15, 429
Beauchamp A., Wesemael F., Bergeron P., Liebert J. Saffer
R.A., 1996. Hydrogen­deficient stars, p. 295, eds. Jeffery
C.S., Heber U., ASP Conf. Ser. Vol. 96
Becker S.R., Butler K., 1989, AA 209, 244
Becker S.R., Butler K., 1990, AA 235, 326
Bell K.L., Hibbert A., Stafford R.P., McLaughlin B.M., 1994.
Physica Scripta 50, 343
Butler K. 1984, PhD Thesis, University College London
Drilling J.S., 1980, ApJ 242, L43
Drilling J.S., Hill P.W., 1986, Hydrogen­deficient stars and re­
lated objects, IAU Coll. 87, p. 499, Hunger K., Sch¨onberner
D. Rao N.K., Reidel, Dordrecht, Holland.
Hardorp J., Scholz M., 1970, ApJS 19, 193
Harrison P.M., Jeffery C.S., 1997, AA 323, 177
Heber U., Jonas G., Drilling J.S., 1986, Hydrogen­deficient
stars and related objects, IAU Coll. 87, p. 67, Hunger K.,
Sch¨onberner D. Rao N.K., Reidel, Dordrecht, Holland.
Holweger H., Heise C., Kock, M., 1990, AA 232, 510
Iben I. Jr., Tutukov A.V., Yungelson L.R., 1996. ApJ 456, 750
Jeffery C.S., 1984, MNRAS 210, 731
Jeffery C.S., 1988, MNRAS 235, 1287
Jeffery C.S., 1993, AA 279, 188
Jeffery C.S., 1996, Hydrogen­deficient stars, p. 152, eds. Jeffery
C.S., Heber U., ASP Conf. Ser. Vol. 96
Jeffery C.S., Heber U., 1992, AA 260, 133
Jeffery C.S., Heber U., 1993, AA 270, 167

J.S. Drilling et al.: Spectral analysis of the extreme helium star LSS 3184 7
Jeffery C.S., Hill P.W., 1986, MNRAS 221, 975
Jeffery C.S., Hill N.C., 1996a, Hydrogen­deficient stars, p. 396,
eds. Jeffery C.S., Heber U., ASP Conf. Ser. Vol. 96
Jeffery C.S., Hill P.W.: 1996b, The Observatory 116, 156
Jeffery C.S., Heber U., Hill P.W., 1986, Hydrogen­deficient
stars and related objects, IAU Coll. 87, p. 101, Hunger
K., Sch¨onberner D., Rao N.K., Reidel, Dordrecht, Holland.
Kilkenny D., Koen C.: 1995, MNRAS 275, 326
Kodaira K., Scholz M.: 1970, AA 6, 93
Kurucz R.L.: 1979, ApJS 40, 1
Kurucz R.L., Peytremann E.: 1975, A Table of Semiempirical
gf Values, SAO Special Report No. 362, Cambridge, Mas­
sachussets 02138
Leuenhagen U., Heber U., Jeffery C.S.: 1994, AAS 103, 445
Lynas­Gray A.E., Sch¨onberner D., Hill P.W., Heber U.: 1984,
MNRAS 209, 387
M¨oller R. U.: 1990, Diplom. Thesis, Universit¨at Kiel
Peach G.: 1970, Mem. R. astr. Soc. 73, 1
Saio H.: 1988, MNRAS 235, 203
Saio H.: 1995, MNRAS 277, 1393
St¨urenburg S., Holweger H., 1990, AA 237, 125
Wiese W.L., Smith M.W., Glennon B.M., 1966, Atomic Tran­
sition Probabilities, National Bureau of Standards, Wash­
ington
Wiese W.L., Smith M.W., Miles B.M., 1969, Atomic Transition
Probabilities, National Bureau of Standards, Washington
Yan Y., Taylor K.T., Seaton M.J., 1987, J. Phys. B 20, 6399
Table 1. Photospheric line abundances for LSS 3184. Refer­
ences for gf­values are given with the ion designation
Ion Ref.
Mult. – W– log gf \Gamma log n Notes
( š A) (mš A)
H i
fl 4340.4 60 3.41 bl S iii 4
ffi 4101.7 34 3.40 bl Si iii 8;10
ffl 3970.1 63 2.90 bl C ii 38
3.24 \Sigma 0.29
C ii Yan 87
12.01 4637.4 55 ­1.237 2.37
12.02 4306.3 27 ­1.678 2.33
17.08 4802.7 134 ­0.411 2.46
28 4313.1 95 ­0.378 2.44
4323.1 26 ­1.108 2.46
39 4413.3 39 ­0.629 2.40
42 4285.7 52 0.373 2.43
2.41 \Sigma 0.05
C iii All 90
1 4647.2 165 0.072 2.85
5 4659.1 11 ­0.652 2.50
4665.9 28 0.049 2.57
16 4067.9 50 0.827 2.77
2.67 \Sigma 0.16
N ii Bec 89
5 4601.5 204 ­0.385 3.31
4607.2 178 ­0.483 3.37
4613.9 169 ­0.607 3.30
4621.3 222 ­0.483 3.10
4630.5 320 0.093 3.00 bl C ii 49
4643.1 198 ­0.385 3.34
12 3995.0 296 0.225 3.38
15 4447.0 147 0.238 3.81
20 4774.2 42 ­1.056 3.27
4779.7 76 ­0.577 3.41
4781.2 41 ­1.036 3.30
4788.1 113 ­0.388 3.32
4793.7 67 ­1.056 3.01
4803.3 178 ­0.135 3.15
4810.3 60 ­1.036 3.09
30 3829.8 110 ­0.599 3.10
3842.2 97 ­0.695 3.10
3847.4 95 ­0.821 2.98
33 4227.7 177 ­0.089 3.02
39 4035.1 127 0.597 3.68
4041.3 165 0.830 3.66
4043.5 100 0.714 3.98
4056.9 67 ­0.461 3.07
48 4237.0 191 0.567 3.47 bl
4241.2 69 ­0.336 3.14
4241.8 166 0.728 3.51 bl N ii 47
4242.4 55 ­0.336 3.28
50 4160.5 24 ­1.125 2.93
4179.7 106 ­0.204 2.98
55 4427.2 116 ­0.004 3.04
4428.0 89 ­0.165 3.08 bl Mg ii 9

8 J.S. Drilling et al.: Spectral analysis of the extreme helium star LSS 3184
Table 1. contd.
Ion Ref.
Mult. – W– log gf \Gamma log n Notes
( š A) (mš A)
N ii contd.
4431.8 73 ­0.165 3.21
4432.7 139 0.583 3.47
4433.5 92 ­0.040 3.18 bl Mg ii 9
4442.0 82 0.312 3.61
58 4530.4 157 0.671 3.41 bl N iii 3
61 4678.1 124 0.079 3.00
3.27 \Sigma 0.25
N iii But 84
2 4640.6 32 0.140 3.58
3 4514.9 22 0.225 2.85
4523.6 11 ­0.353 2.72
4534.6 15 ­0.458 2.42
2.89 \Sigma 0.49
O ii Bel 94
1 4638.9 152 ­0.307 3.28 bl C ii 12:01
4650.8 86 ­0.331 3.75
4661.6 111 ­0.249 3.63
4676.2 74 ­0.360 3.82
2 4319.6 85 ­0.357 3.78
4345.6 102 ­0.330 3.66 bl N iii 10
4349.4 152 0.086 3.71
4366.9 61 ­0.320 4.03
5 4414.9 136 0.210 3.83
4452.4 52 ­0.757 3.57
6 3945.0 87 ­0.699 3.36
3954.4 159 ­0.396 3.12 bl Fe iii 120
3973.3 176 0.009 3.39 \Lambda bl C ii 37;38
3982.7 78 ­0.682 3.44
10 4069.9 137 0.365 3.83 bl C iii 16
4072.2 125 0.545 3.75
4078.8 45 ­0.272 3.68
4085.1 64 ­0.175 3.56
4092.9 15 ­0.306 4.21 \Lambda wk
4094.1 21 ­1.469 2.88
11 3907.5 35 ­0.924 3.19
12 3847.9 78 ­1.066 2.57 bl Mg ii 5
3850.8 43 ­1.076 2.93 bl Mg ii 5
3851.0 88 ­0.854 2.69
3863.5 14 ­0.971 3.61
3882.2 74 ­0.032 3.63
15 4591.0 83 0.346 3.81
4596.2 96 0.196 3.55
16 4351.3 93 0.228 3.65 bl N iii 10
17 3912.0 99 0.000 3.50
19 4132.8 82 ­0.122 3.39
20 4097.3 135 ­0.604 3.17 \Lambda bl O ii 48
4104.7 86 ­0.143 3.77
4119.2 89 0.453 3.90
21 4112.0 69 ­0.561 3.07
25 4705.4 76 0.519 3.89
4741.7 23 ­0.928 3.14
26 4395.9 61 ­0.183 3.40 bl Fe iii 4
Table 1. contd.
Ion Ref.
Mult. – W– log gf \Gamma log n Notes
( š A) (mš A)
O ii contd.
34 3821.5 25 ­1.027 3.08
36 4185.5 63 0.610 3.70
4189.8 95 0.723 3.51 bl S ii 44
40 4703.2 33 0.274 3.64
42 4192.5 27 ­0.325 3.24
48 4108.8 11 ­0.161 3.82
50 4062.9 19 ­0.148 3.57
67 4276.7 70 ­0.717 2.20 \Lambda bl O ii 54
4275.6 71 0.766 3.66
86 4489.5 15 0.587 4.30 \Lambda wk
4491.2 196 0.842 2.75
93 4609.4 35 0.709 3.93
97 4060.6 48 0.775 3.77
101 4253.9 165 0.929 2.47 \Lambda bl S iii 4
3.50 \Sigma 0.35
Ne ii Wie 66
56 4379.5 18 0.468 2.61 \Lambda bl N iii 18
56 4430.9 40 0.246 1.76 \Lambda bl Fe iii 4
Mg ii Wie 69
4 4481.1 0.568 e
4481.2 236 0.732 4.30 c
10 4384.7 9 0.166 4.31
4390.6 11 0.316 4.50
4.37 \Sigma 0.11
Al iii Kod 70
3 4512.5 88 0.405 5.57
4528.9 ­0.282 e
4529.1 143 0.670 5.47 c
5 4149.9 99 0.619 5.39
8 4479.9 0.894 e
4480.0 107 1.021 5.58 c
5.50 \Sigma0.09
Si ii Bec 90
1 3853.7 8 ­1.377 4.38
1 3862.6 19 ­0.682 4.69
3 4128.1 190 0.369 3.61 \Lambda bl
3 4130.9 58 0.545 4.62
4.56 \Sigma0.16
Si iii Har 70
2 4567.8 204 0.061 4.85
2 4574.8 165 ­0.416 4.61
3 4338.5 30 ­1.143 5.03
8.09 4716.7 80 0.491 4.55 \Lambda bl S ii 9
9 4813.3 56 0.702 4.83
9 4819.7 80 0.814 4.71
13 4683.0 18 0.185 4.55
4.76 \Sigma0.18
Si iv Bec 90
1 4116.1 81 ­0.103 4.51
5 4212.4 17 0.804 4.15
4.33 \Sigma0.25

J.S. Drilling et al.: Spectral analysis of the extreme helium star LSS 3184 9
Table 1. contd.
Ion Ref.
Mult. – W– log gf \Gamma log n Notes
( š A) (mš A)
P iii Wie 69
3 4222.2 45 0.190 6.66
4246.7 34 ­0.119 6.50
6.58 \Sigma0.11
S ii Wie 69
9 4815.5 51 ­0.050 4.15 ?
44 4217.2 41 ­0.150 3.62 \Lambda
4.15
S iii Har 70, Wie 69
4 4285.0 77 ­0.046 5.14
4361.5 44 ­0.724 4.81 \Lambda bl C iii 50
7 4354.6 53 ­1.602 3.84 \Lambda
8 3928.6 44 ­0.160 5.45 \Lambda bl He i 58
3983.8 75 ­0.438 4.82
3986.0 58 ­0.790 4.65
4.87 \Sigma0.25
Fe iii Kur 75
4 4419.6 36 ­2.301 5.50
118 4137.9 18 0.644 5.18
4139.4 16 0.553 5.14
4140.5 13 0.114 4.80
4164.8 78 0.940 4.67
4166.9 25 0.436 4.81
5.02 \Sigma0.31
Notes:
e . . .
c Line abundance calculated from blend
bl Blend with weak line from another species
\Lambda Line omitted from mean abundance
\Lambda bl Blended line omitted . . .
References:
All 90 Allard et al. 1990
Bec 89 Becker & Butler 1989
Bec 90 Becker & Butler 1990
Bel 94 Bell et al. 1994
But 84 Butler 1984
Har 70 Hardorp & Scholz 1970
Kod 70 Kodaira & Scholz 1970
Kur 75 Kurucz & Peytremann 1975
Wie 66 Wiese et al. 1966
Wie 69 Wiese et al. 1969
Yan 87 Yan et al. 1987
Table 2. Model parameters and photospheric abundances for
LSS 3184. Abundances are normalised to log
P
i ¯ i n i = 12:15.
Comparisons are given for V652 Her, BD\Gamma9 ffi 4395, HD 144941
and the Sun.
LSS V652 BD\Gamma9 ffi HD Sun
3184 Her 4395 144941
T eff 23 300 \Sigma 700 23 500 22 700 23 200
log g 3:35 \Sigma 0:10 3.7 2.55 3.9
v t 15 \Sigma 5 20
H Ÿ 7:72 9.5 8.74 10.28 12.00
He 11:54 11.54 11.54 11.54 10.99
C 9:02 \Sigma 0:23 7.03 9.17 6.80 8.58
N 8:26 \Sigma 0:25 8.9 7.97 6.5 8.05
O 8:05 \Sigma 0:36 7.9 7.90 7.0 8.93
Mg 7:17 \Sigma 0:11 8.1 7.25 6.1 7.58
Al 6:04 \Sigma 0:09 6.7 5.55 4.8 6.47
Si 6:91 \Sigma 0:24 7.7 7.83 6.0 7.55
P 4:96 \Sigma 0:11 5.8 6.21 5.45
S 6:67 \Sigma 0:25 7.4 7.83 7.21
Fe 6:52 \Sigma 0:31 7.4 6.57 6.4 7.48
References:
V652 Her Jeffery et al. 1986;
T eff , log g: Lynas­Gray et el. 1984;
O,Ne,Mg,Al,P,S,Fe: Jeffery 1996
BD\Gamma9 ffi 4395 Jeffery & Heber 1992
HD 144941 Harrison & Jeffery 1997
Sun Anders & Grevesse 1989;
Fe: Holweger et al.1990;
C: St¨urenburg & Holweger, 1990

10 J.S. Drilling et al.: Spectral analysis of the extreme helium star LSS 3184
Table 1. Other measured absorption lines in the spectrum of
LSS 3184
Mu. –( š A) W– Note Mu. –( š A) W– Note
He i
12 4713.2 473 16 4120.9 578
5 3964.7 462 50 4437.5 324
52 4169.0 275 54 4024.0 46
57 3935.9 236 61 3838.1 188 S iii 5
He ii
1 4685.7 67
C ii
1 4735.5 31 1 4738.0 18
1 4744.8 51 1 4747.3 18
4 3919.0 187 N ii 17 4 3920.7 182
6 4267.1 448 bl 12.02 4307.6 69
13 3831.7 68 13 3835.7 59
13 3836.7 15 27 4017.3 51
28 4317.3 123 O ii 2 28 4318.6 45 N iii 10
28 4321.6 54 N iii 10 28 4326.0 138 bl; O ii 2
32? 3952.1 113 33 3876.1 82 bl
33 3876.5 222 bl 33 3879.6 72
33 3880.6 73 33.01? 3868.9 51
35.01 4077.7 72 bl 36 4075.9 319 bl; N ii 38
36 4074.7 263 bl 37 3972.4 49 C ii 38
37 3977.3 49 bl 37 3978.8 66 bl
37 3980.3 48 39 4411.3 194 bl
40 4409.2 34 40 4410.0 96
41 4295.9 53 42 4291.8 51 C ii 41
45 4368.3 29 C ii 46 45 4369.9 14
45 4372.4 154 bl; C ii 46 45 4374.3 115
45 4375.0 47 N ii 16 45 4376.6 69
49? 4625.6 37
C iii
1 4650.3 128 5 4651.0 65 C iii 1
5 4652.0 7 5 4663.6 24
9? 4515.5 8 bl;cr 9? 4516.8 4 cr
14? 4382.9 26 16 4068.9 37
18 4186.9 21 21? 4162.9 58 S ii 44
N ii
6 3955.9 156 11 4654.5 50
11 4667.2 67 11 4674.9 62 O ii 1
14 4564.8 27 21 4459.9 34
21 4477.7 78 21 4488.1 32 O ii 104
21 4507.6 60 30 3838.4 64 S iii 5
30 3855.1 121 30 3856.1 124 O ii 12
38 4073.0 94 cr 38 4076.9 99 C ii 36
38 4082.3 97 38 4095.9 102 O ii 48
38 4087.3 67 O ii 48 39 4039.3 44
39 4044.8 99 43 4171.6 78
43 4176.2 164 43.01 4131.8 124
44 4110.0 76 44 4110.8 60
48 4247.2 25 N ii 21:02 49 4181.1 51
49 4196.0 94 N iii 6 49 4200.0 137 N iii 6
49 4201.4 30 50 4156.7 133 bl; C iii 21
50 4161.1 19 50 4173.6 211
54.01 4508.8 32 55 4417.1 1.61 O ii 5
Table 1. contd.
Mu. –( š A) W– Note Mu. –( š A) W– Note
N ii
57.01 4602.6 68 O ii 93 57.01 4608.1 55
58 4552.5 250 Si iii 2 61 4694.6 118
65 4124.1 80 65 4133.7 84
65 4145.8 30 S ii 44 68 4695.9 39 O ii 1
68 4698.6 93 O ii 25 68 4700.0 15
68 4702.5 11 68 4704.2 12
68 4706.4 3 cr 68 4709.6 38 N ii 25
68 4718.4 16 68 4721.6 9
72 3939.6 50 72 3940.7 44
72 3941.2 40 74 4206.3 100 bl
74 4207.5 92 73 4154.8 39
N iii
2 4634.1 19 cr 3? 4510.9 61 bl
3? 4518.2 16 cr 10? 4332.9 89 S iii 4
O ii
1 4641.8 154 N iii 2 1 4649.1 156 C iii 1
1 4673.8 27 C iii 5
2 4336.9 54 N iii 10 12 3864.6 13 bl
16 4347.4 90 bl? 19 4153.3 98 C iii 21
20 4103.0 62 N iii 1 21 4096.5 56 O ii 48
41 4327.5 31 N iii 10 41 4331.9 22 cr
48 4089.3 128 Si iv1 49? 4083.9 21 O ii 21
54 4294.8 29 S ii 49 58 4701.2 14 Al iii 6
77 4342.0 57 Si iii 46 92 4610.1 20
Ne ii
64? 4499.0 37
Ca ii
1 3933.7 545 1 3968.5 322
Al iii
9? 4364.6 30 Fe iii 4
Si iii
8.10? 4102.4 27 8.14 3924.5 57
8.16? 4800.4 38 10.08? 3947.5 45
15? 4554.0 30
P iii
9? 3904.8 42
S iii
7? 4439.9 12 8? 3986.0 58
Fe iii
121? 4273.4 34 121? 4304.8 26
121? 4310.4 30 4210.9 88
Notes:
Mu Multiplet
bl two or more components of same multiplet
cr cosmic ray
Table 2. Unidentified lines in the spectrum of LSS 3184. Wave­
lengths have been corrected for stellar motion to give labora­
tory wavelengths in air.
– W– – W– – W–
š A mš A
3897.0 183 3899.3 45 3915.4? 28
4015.9 41 4248.5 22 4259.6 55
4261.3 76 4292.3 137 4329.8 138