The spectroscopic observations were made on the nights of April 10-13 1994
using the MEFOS spectrograph on the ESO 3.6 m telescope. MEFOS is a robot
controlled fiber instrument with 30 arms, disposed like fishermen around a
pond, about the 
prime focus field. The detector is a Tek
CCD, ESO 32. The dispersion on the CCD was
170 Å/mm and the spectra covered the range from 3800 to 5100 Å. The
resolution given by the combination of 
fibers and CCD
pixel with the standard Boller & Chivens spectrograph on the
3.6 m telescope was 3.8 Å. The fibers are disposed in
pairs, one for object and sky and one for sky. The arms bulk only permitted
a minimum angular distance of 6' between targets, a strong constraint.
Switching between the fibers permits precise sky substraction. We exposed
twice for 15-20 minutes. Further description of this equipment and
discussion of its performance can be found in Felenbok et al.\
(1996).
The MEFOS system permits one to obtain a CCD image of a square field 30'' in
size around each target, providing a precise centering of the fibers. The
original target selection was based on catalogs produced by the MAMA machine
from scans of the R ESO/SRC survey plates done for our project
(Slezak et al. 1996; Infante et al. 1996). Due to
the rather low galactic latitude of the fields, some 
of the
galaxies identified by the MAMA software proved to be confused or merged
objects. Such a rather high fraction is also related to the use of a
glass-plate copy on which the measured seeing on the stars can be evaluated
around 3 arcsec; confusion mostly occured between a faint or low surface
brightness galaxy and a nearby faint star. The MEFOS procedure allows the
fiber to be re-centred when necessary. This possibility led to the
observation of some faint galaxies, much fainter than the homogeneous
sample originally intended. Moreover, due to the severe angular limitation
referred above, we were prevented from observing many bright galaxies.
Therefore, the observed galaxies are far from been a magnitude limited
sample. In the galaxy selection process no consideration was taken of the
few previously known redshifts, thus generating some overlap for comparison
purposes.
The data reduction was carried out at Meudon using the standard IRAF package.
This software package efficiently corrects for bias, flat field, and sky
emission, combines the individual exposures and provides 1D spectra calibrated
in wavelength. Wavelength calibrations were done using a 5 min. exposure with
an 
or a 
lamp taken inmediately before and after each
observation. The spectra were rebinned with a scale of 1 Å/bin equally
spaced in wavelength. The velocities were determined from the
cross-correlation procedure with stellar and galaxy template spectra of
known radial velocity, according to Tonry & Davis (1979). We
used the RVSAO package developped at Harvard (Kurtz et al.\
1991; Mink et al. 1995).