We defined the sample to be observed from Table 3 (click here) and Table 4 (click here) of Hauck (1986). The reason why we considered Table 4 (click here) (containing stars classified spectroscopically as dwarfs but photometrically as giants) was that the photometric criterion of luminosity seems much more reliable than the spectroscopic luminosity class.
The selection criteria were:
)
data or unknown 
.
By insufficient , we mean that the star has less than three
published values in the literature. Applying these criteria, 50 A
and F giant stars were selected, both normal and metallic. It was judged
useful to have a good estimate of the binary freqency of normal A and F
giants for reference purposes. 40 stars in the sample are non-metallic,
while only ten are metallic according to the criterion 
(see Sect. 5.1). Thus, the estimate of binary frequency among
metallic giants will mainly rely upon old, published data. The
values of the sample stars range between -0.027 and 0.082.
They have been updated using a new reference sequence in the m2/B2-V1
diagram (Hauck et al. 1991), and sometimes by complementary
photometric measurements. Eight stars belong to the 
Scuti class,
and nine are spectroscopically classified as dwarf but have a large 
parameter ( is the luminosity parameter of Geneva photometry,
equivalent to Strömgren's 
) indicating that they very probably
are real giants. All stars are bright, with 
, implying that
interstellar reddening is insignificant.
The observations were performed at the Observatoire de Haute Provence (OHP)
with the Aurélie spectrograph attached to the 1.52 m telescope at the Coudé
focus (Gillet et al. 1994) in 1994. Four runs of a few nights'
duration each were made respectively at the beginning of May and June for 18
stars and at the beginning of November and December for 32 stars. These
measurements should allow to detect binary stars with a small period (
100 days).
The detector is a double barrette CCD Thomson TH7832 with 2048 pixels having
a size of 
. The spectra were obtained at a
reciprocal dispersion of 
in the spectral region
centred on H
[4780 Å, 5000 Å]. The reduction was made at the OHP
with IHAP procedures, using comparison spectra of thorium. Each stellar
measurement was preceded by a calibration exposure to compensate for
instrumental drift as much as possible. To normalise our spectrum, we simply
fit a straight line to the continuum. 291 stellar exposures were made, of
which 54 were devoted to standard stars. For most of the spectra a
signal-to-noise of 150 was achieved. The minimum number of exposures per
star was 2 and the maximum 11, with a mean value of 5 or 6 measurements per
star.