We have obtained spectra of 20 objects classified
sdO(D) by Green et al. (1986) or He-sdB by Beers et al.
(1992) and Heber et al. (1996), and spectra of a few
objects classified sdO(A), sdO(B) and sdO(C) by Green et al. (1986).
The latter were obtained to provide overlap between our
sample and samples analyzed in other studies (e.g. Thejll et al.
1994; Saffer et al. 1994).
For objects with mV<14
spectra were obtained with the William Herschel telescope
(WHT) and Utrecht échelle spectrograph (UES)
in May 1995 covering the wavelength
interval 3860-5025 Å with a resolution 
.
Fainter objects were observed with the WHT and intermediate
dispersion spectrograph (ISIS) in
the wavelength intervals 4240-4974 Å and with a
resolution of 
. Additional spectra were
obtained with the 3.5 m telescope at Calar Alto (CA) in the
wavelength interval 4181-5070 Å with a resolution
. Owing to a problem in the image-derotator,
the S/N of the UES observations in 1995 is poor, and to
extract useful information the spectra have been binned to a
resolution 
. Further details
are given in Table 1 (click here). Moreover, with low S/N
in the UES spectra, and the substantial widths of important
absorption lines relative to individual échelle orders
in the UES spectra, there were rectification difficulties
which could only be resolved where a low-resolution CA or
ISIS spectrum was available. Thus, for example, the
blue wing of the H
profile in our UES spectrum of PG0909+276 is
incorrect (Fig. 5 (click here)).
It was noted that spectra of three objects did not resemble subdwarf spectra at all.
4471Å, H slightly less strong, and weak
HeI 4713. H
is just detectable. Our spectrum
is too noisy to show any weaker features. This star would
appear to be a DBA white dwarf.
For the remainder it became evident that several objects show
spectra substantially different from that expected from previous
observations of the prototype PG1544+488. Although the majority show
essentially no Balmer lines, weak HeII and possibly a
number of weak metal lines, in most cases HeII4686Å is
much stronger than in PG1544+488. Whilst it had been supposed that
several objects would indicate effective temperatures around
and surface gravities between those of the
extreme helium giants and the classical hydrogen-rich subdwarf B
stars, the observed strength of HeII4686Å is more often
similar to that seen in sdOB and sdO stars. It is therefore necessary
to review the classification of these stars.
To be properly useful, this review had to draw on a larger sample of
spectral types than was provided by the observations described above.
A number of 2.5 Å spectra were obtained in
a series of studies of hot subluminous stars at high galactic latitudes
(Moehler et al. 1990)
at the Calar Alto and the European Southern
Observatories. These have been used to provide both comparison spectra and
examples of hot subdwarf spectral types not already represented.
These spectra include the well-known subdwarfs Feige110 and
BD+28
4211. They also include PG1701+359 and PG2337+070
(Theissen et al. 1993), and PG1047+003, PG1559+048, PG1559+076,
PHL4, and PG1537-046, which have not previously been published.
Since B-type subdwarf spectra are also similar to those of
horizontal-branch B (HBB) stars, we have also examined a spectrum of
PG1705+537 from the Moehler et al. (1990) sample. Echelle spectra of
three helium stars obtained at ESO (LSS5121: Heber et al. 1986),
the WHT (LSIV+62) and the AAT (HD 144941:
Harrison &
Jeffery 1996) have also been used for comparison purposes. The
resolution of the échelle spectra was degraded to match the remainder
of the sample.
Figure 1: Schematic of classifications
for 10 Å resolution spectra of hot subdwarfs. The boxes are labelled
by Drilling (1996) nomenclature in bold type and their equivalents as described
in the text