The observations were performed with the ESO 3.6 m telescope equipped with MEFOS
(see description below) during 6 nights on November 5-11, 1994 and 2 nights
on November 24-26, 1995. The grating used with the Boller & Chivens
spectrograph had 300 grooves/mm, giving a dispersion of 224 Å/mm in the
wavelength region 3820-6100 Å. The detector was CCD #32, with 5122
pixels of 
m.
The catalogue of galaxy positions used in this survey was obtained with the
MAMA measuring machine and is presented in a companion paper
(Slezak et al. 1997). This catalogue gives approximate magnitudes in the 
band,
which were used to select the galaxies to be observed spectroscopically.
CCD photometry of the central regions of the cluster in the V and R
bands was
later performed to recalibrate magnitudes and obtain V and R
magnitudes
for the entire photometric sample. We observed spectroscopically a total
number of 21 fields, with exposure times of 2
20 minutes for the two
fields with galaxies all brighter than 
, and 230 minutes
for the other ones.
We obtained 519 spectra in total (plus the same number of sky spectra).
Figure 1: Spatial distribution of the 551 galaxies with redshifts
Out of the 519 spectra obtained, we measured 421 reliable redshifts (the other
ones were discarded due to insufficient signal to noise).
Our catalogue includes these spectra, plus those previously published
by Beers et al. (1991) and Malumuth et al. (1992). For galaxies observed twice,
we chose the redshift with the smallest error (usually the Beers et al. data).
The positions of the objects for which we obtained reliable spectra are
shown in Fig. 1 (click here). These positions are relative to the following cluster
center: 
.
This center was chosen to coincide with that of the diffuse
X-ray gas component as
defined by Pislar et al. (1997).
MEFOS uses the big advantage of the prime focus for fibre spectroscopy. The 3.6 m ESO telescope has a prime focus triplet corrector delivering a field of one degree, the biggest at that time for a 4 m class telescope. This will no longer be the case once the 2dF project at the AAT reaches completion in a very near future. The focal ratio is F/3.14, well suited for fibre light input, leading to negligible focal ratio degradation. MEFOS (Guérin et al. 1993) is sitting on the red triplet corrector and is made of 30 arms that sweep the 20 cm diameter (one degree) field.
Figure 2: Photograph of the MEFOS instrument
Figure 2 (click here) shows the general arm display. In fact, only 29 arms are positioned on astronomical objects, one arm being used for guiding. The arms are displayed around the field as "fishermen-around-the-pond''. The arms are moving radially and in rotation, in such a way that each arm is acting in a 15 degree triangle with its summit at the arm rotation axis and its base in the centre of the field. So, all arms may access an object in the centre of the field and only one can reach an object at the field periphery. This situation changes gradually from the centre to the edge of the field. Each arm has its individual electronic slave board and all the instrument is under control of a PC computer, independent from the Telescope Control System (TCS). The arm tips carry two fibres 1 arcmin apart, each one intercepting 2.5 arcsec on the sky. One is used for the object, and the second one for the sky recording, and both go down to the spectrograph. Object and sky can be exchanged; this allows to cancel the fibre transmission effects.
Figure 3: Photograph of a galaxy field seen by the 29 windows on the CCD,
showing the position of the selected galaxies on the image bundles; this
allows to place the spectroscopic fibers accurately on each galaxy
Coupled firmly to the arm tip is inserted an image conducting fibre
bundle, that covers an area of 3636 arcsec2 on the sky. All the
image bundles are projected on a single Thomson 10241024 thick CCD,
Peltier cooled, connected to the same PC as the one driving the arms.
Figure 3 (click here) shows a
galaxy field as seen by the 29 windows on the CCD, corresponding to the
arm image bundles set on the object coordinates. This procedure, in
opposition to blind positioning, is the only one, to our knowledge, that
shows the objects on which the spectral fibre will be placed in a second step.
By analyzing the real position of the object in the image fibre, and
knowing the relative position of this image bundle and its connected
spectral fibre, a precise offset is computed and the arm is sent to its
working position. This offset takes care of all imprecisions due to the
telescope, the instrument and the coordinate inaccuracy. Given the poor
pointing of the telescope and the fact that the corrector and the
instrument are frequently dismounted, blind positioning would be
extremely dangerous. The positioning accuracy, as measured on stellar
sources, is 0.2 arcsec.
In the present stage, the spectral fibres, 135 mm in diameter and 21 m
long, are going down from the prime focus to the Cassegrain, where the B&C
ESO spectrograph is located. This spectrograph is fitted with a F/3
collimator to match the fibre output beam aperture, it has a set of
reflection gratings and a Tek 
thin CCD.
The spectra were reduced using the IRAF software. The frames were bias and flat field corrected in the usual way. Velocities were measured by cross-correlating the observed spectra with different templates: a spectrum of M 31 (kindly provided by J. Perea) at a velocity of -300 km s-1, and stellar spectra of the standard stars HD 24331 and HD 48381, which were each observed every night during our 1994 run. The cross-correlation technique is that described by Tonry & Davis (1979) and implemented in the XCSAO task of the RVSAO package in IRAF (Kurtz et al. 1991).
The positions of emission-lines, when present, were measured by fitting each line with a gaussian.
All the spectra were reduced by the same person (F.D.) in a homogeneous way. Redshifts of insufficient quality were discarded.