The following stars were included in the HIPPARCOS list of RR Lyraes but are probably not RR Lyraes.
V753 Cen and LS Her:
V753 Cen has a period of 0.221 days and LS Her a period of 0.231 days. These
are too short for RRc Lyraes and we did not observe them.
V429 Ori:
Using the GCVS period of 0.5017 days we took two spectra, a third of a cycle
apart, in December 1993 and a further spectrum in December 1994. These
showed, firstly, a constant velocity of 
km 
and, secondly, a
narrower H
than found for RRab Lyraes. The width of the H
profile suggests the star we observed was of mid-G spectral type. At the
HIPPARCOS coordinates of this star: ((2000.0) = 04
56
12.0
;
(2000.0) = 
3124)
a search of the HST Guide Star Catalogue shows a star (V = 11.2) at exactly
the same position. There are several fainter stars, 
,
approximately three arcmins away, but the nearest star of similar brightness
is more than 6 arcmins away. The most probably explanation is that there
is a variable in this region but either the coordinates or the magnitude
are seriously in error.
ET Hya:
Using the GCVS period of 0.685519 we took a spectrum in March
1994 and then three well-phased spectra in December 1994. These showed a
constant velocity of 
km . The H profile was consistent
with other RRab Lyraes. A search of HST Guide Star Catalogue shows a star
at exactly the HIPPARCOS coordinates: ((2000.0) = 
;
(2000.0) = 
)
with another star of similar brightness only 43 arcsecs away. It seems likely
the two stars have been confused, in particular we note that Layden (1994), on
the basis of six velocity measurements, derived a systemic velocity of
20 km and [Fe/H] = -1.69.
NSV 5394:
HIPPARCOS included this object as an RR Lyrae. It has
coordinates ((2000.0) = 
;
)
and V = 9.0. The possibility of such a bright variable being hitherto
undiscovered seemed improbable and indeed on our first attempt to observe
this star it was clear from the TV Acquisition screen that it was a double
star with approximately 2 - 3 arcsecs separation between the two
components. Probably the presence of such a close companion has led to a
spurious identification as a variable.
KN Per and BB CMi:
Our spectra showed a broad and shallow H
which was clearly different to the H profile of other RR Lyraes.
These stars have been studied as part of the Behlen Observatory
Variable Star Survey (Schmidt 1991) and on the basis of those observations they
have been classified as Eclipsing Variables (KN Per (Schmidt 1991) and
BB CMi (Schmidt 1995)).
XZ Cet:
This star
has the period of a fundamental mode RR Lyrae but the light curve shape and
amplitude resemble more closely an overtone pulsator. Teays & Simon (1985),
from a comparison of both the light curve and physical characteristics of the
star with pulsation models, concluded that XZ Cet was probably an anomalous
Cepheid rather than an RR Lyrae. Anomalous Cepheids are more luminous than
RR Lyrae and are believed to have gained mass, presumably from a companion.
We took several spectra of the star and interestingly we find a significant
difference between our systemic velocity, 
km , and that of
Layden (1994) who found 
km . There was no evidence of line
doubling.
Teays and Simon show the star has similar temperatures, during the
pulsation cycle, to an RRab Lyrae but a lower gravity. Using our spectra taken
on the descending branch then for 
K and
we find 
, in reasonable
agreement with the value of 
found by Layden (1994).
V363 Cas and AT And:
The radial velocity measurements for these
stars showed relatively small amplitudes. This is consistent with the light
curve amplitudes which, according to the GCVS, are also unusually small
(0.43 mags for V363 Cas and 0.50 mags for AT And). Thus the stars appear to be
similar to XZ Cet (period implies fundamental mode pulsation but the light
curve is more consistent with overtone pulsation). We have included the
stars amongst the RR Lyraes in the Appendix but clearly there is some
uncertainty in the classification.
V1719 Cyg and SS Psc:
There is some confusion in the literature
regarding the classification of certain stars as RRc Lyraes or large amplitude
Scutis. Large amplitude Scutis are A-F stars in the
early post-main sequence stage of evolution (e.g. Mc Namara & Feltz
1978; Breger 1980). Amongst Pop. II stars the periods of the large
amplitude Scutis (also known as SX Phe stars or Variable Blue
Stragglers) are typically 
0.10 days and thus they are clearly
separated from RRc Lyraes, which have periods 
0.25 days. However,
amongst Pop. I the large amplitude Scutis (also known as dwarf
Cepheids or AI Vel stars or RRd stars) have periods of 
0.12 days
but the tail of the distribution extends to longer periods hence the
confusion with the RRc Lyraes. The simplest method of discriminating
between the two groups is in term of surface gravity since long period,
large amplitude Scutis will have larger masses,
, than RRc Lyraes, 
.
Most of the stars with periods 0.25 days that we observed are
metal-deficient and therefore the RRc Lyrae identification is secure,
however, three of the stars we observed (DE Lac, P = 0.25 days; V1719 Cyg,
P = 0.27 days; SS Psc, P = 0.29 days) have metallicities approximately solar.
This is confirmed by the published Strömgren photometry of McNamara &
Feltz (1978) and Johnson & Joner (1986) from which the following
parameters were derived by the authors: 
, 
,
[Fe/H] = 0.2 for DE Lac; 
, 
, [Fe/H] = 0.4 for
V1719 Cyg and 
, 
, [Fe/H] = 0.0 for SS Psc. In
particular, the high gravities found for these stars suggest they are more
probably Scutis. It may be questioned whether the calibration of
Strömgren photometry is valid for pulsating stars; however Siegel (1982)
obtained gravities from Strömgren photometry of 3 RRab Lyraes and these
gravities are in good agreement with the value adopted by us in Sect. 4.1
(Siegel found a mean value for the three stars of 
; we adopt
log g = 2.75).
BB Vir:
BB Vir has a period, amplitude and light curve shape that imply
it is a fundamental mode pulsator. However, in a recent study Kinman
& Caretta (1992) found its B-V colour was too blue compared to other fundamental
mode RR Lyraes of similar period and metallicity. They suggested the star
may have a Blue Horizontal Branch companion. Fernley (1993b) compared IUE
spectra of BB Vir and RR Lyrae itself and this showed that BB Vir has a
large ultraviolet excess, compatible with the presence of a BHB companion.
We took two spectra of BB Vir on 19 March 1994 at 6.32UT and 11.10UT which,
using the ephemeris of Fernley (1993b), corresponds to phases 0.34 and 0.75
respectively. This phase interval covers very nearly the full radial velocity
amplitude of the star and we found 
km from the two
spectra. However, we detected no obvious line doubling in either H
or the OI triplet, both of which should be stronger for the proposed
companion than for BB Vir itself.
An alternative explanation is that the star is an unusually long period ``c'' type; however, both the amplitude and the light curve shape argue against this, as does its position in the period - temperature diagram (Fernley 1993a).
We note that BB Vir may have both variable amplitude (the GCVS lists it as ``Blazhko Effect?") and a variable period (see discussion in Kinman & Caretta 1992). It may be that BB Vir is a fundamental mode RR Lyrae that has, for some reason, evolved a long way beyond the fundamental blue edge without yet changing mode. This "hysterisis" effect is well-known in globular clusters and it is possible BB Vir is an extreme example of it. We have included the star amongst the RR Lyraes in the Appendix but clearly it may not be a ``normal" RR Lyrae.
Acknowledgements
Our thanks to Enrique Solano for his help with the A&A macros. The authors acknowledge the financial support of NATO through Grant Number CRG 940746.