2 Observations

   
2.1 Imaging

We will use photometric results from two broad-band images in the analysis of the spectroscopic survey presented in this paper.

On 26 September 1995, we obtained an I-band image using the UH8k camera (Luppino et al. 1995) on CFHT. The I-band image was reduced chip by chip, i.e. the final result is in the form of eight individual images, one for each chip of the camera. The final images were obtained as the mean of 10 exposures of 1200 s each, using sigma clipping ($2.5\sigma$ above the mean level and $4\sigma$ below) to reject hot pixels and cosmic ray hits, and have very good seeing of $0\hbox{$.\!\!^{\prime\prime}$ }7$ FWHM; however, the background is marred by stray light, presumably due to the bright star 47 Psc (V=5.1, spectral type M3) at $\sim$ $50\hbox{$^\prime$ }$ distance from the centre of Cl0024+1654, although outside the field of view of the UH8k. The I-band photometric catalogue was obtained from the individually stacked chips using the SExtractor package (Bertin & Arnouts 1996) with a threshold of $1.5\,\sigma$ and a minimum detection area of 5 pixels. The catalogue contains more than 4 10⁴ objects over a field of about $28\times28\,{\rm arcmin}^2$, the limiting magnitude is $I\sim24$. Unless otherwise noted, we use total magnitudes as given by SExtractor's MAG_BEST. The internal errors on the I-band magnitudes (as given by SExtractor) are smaller than 0.05 for I<22.7 (0.01 for I<20.7). Note that due to the chip-wise reduction of the I-band image, the gaps between the chips (of typical width 5-10 arcsec) were not filled in during stacking and the photometric catalogue contains no objects from these regions.

On 15 November 1999 we obtained a 3600 s V-band image using the CFH12k camera (Cuillandre et al. 2000) on CFHT (Figs. 1 and 4a-4h). The six individual exposures were bias and flat-field corrected in the standard way using the MSCRED package under IRAF. The exposures were then registered onto the Digital Sky Survey (DSS) image of the field and median combined. The final image, a mosaic of all 12 chips with all the inter-chip gaps filled in, has $\sim$ $0\hbox{$.\!\!^{\prime\prime}$ }7$ seeing (FWHM). The V-band photometric catalogue obtained from this image contains $\sim3.7\times 10^4$ objects on a field of $\sim42\times28\,{\rm arcmin}^2$. The internal errors on the V-band magnitudes are smaller than 0.05 for V<23.3 (0.01 for V<21.1). The limiting magnitude is $V\sim25$.

In order to obtain colour information aperture magnitudes were measured in $14\,{\rm pixel}$ ( $2\hbox{$.\!\!^{\prime\prime}$ }8$) diameter apertures for the V- and I-band images. This diameter is sufficiently large compared to the seeing FWHM to enclose most of the light from the object under consideration and small enough to avoid contamination by neighbouring objects in crowded regions like the cluster centre. The objects were then matched up using a polynomial transformation of the I-band image coordinates onto the system of the V-band image. The centres of the two images coincide to within 40 arcsec, so that the overlapping region covers virtually the whole UH8k field. Note that the gaps from the I band image show up in the colour-magnitude catalogue as well. The resulting colour-magnitude catalogue contains $\sim$ $2.1\times 10^4$objects with errors on the colours of smaller than 0.05 for V<23.3 (0.01 for V<21.1). Star-galaxy classification over a sub-region of the colour-magnitude plane (as relevant for the present paper) will be described in Sect. 3.5.

A more detailed description of the photometric catalogue (including star-galaxy separation over the whole colour-magnitude plane) is given in Mayen et al. (2001) who use this catalogue to investigate the depletion of background galaxies due to the gravitational lens effect of Cl0024.

   
2.2 Spectroscopy

Spectra were obtained using multi-slit spectroscopy during three observing runs at CFHT and one at WHT. Table 1 shows the observing log.

Candidates for all the runs (except for run 1) were selected from a V-band mosaic obtained at the ESO New Technology Telescope (NTT) on 17/18 October 1993, with the primary selection criterion being $V_{\rm NTT}<23$. The seeing on this image was rather poor, $\sim1\hbox{$.\!\!^{\prime\prime}$ }7$ FWHM. In order to reduce contamination by stars, the preparatory shallow R-band images were carefully examined during all the CFHT runs; for runs 3 and 4, we took additional advantage of the excellent seeing of the deep UH8k I-band image.

Figure 1: $22\hbox {$^\prime $ }\times 25\hbox {$^\prime $ }$ section of the CFH12k V-band image showing the distribution of the objects in our spectroscopic sample. Expanded views of the marked regions are shown in Figs. 4a-4h as indicated in the image. The coordinates given are right ascension and declination relative to $\alpha_{2000}=00^{\rm h}26^{\rm m}35\hbox{$.\!\!^{\rm s}$ }70$, $\delta_{2000}=17\hbox{$^\circ$ }09\hbox{$^\prime$ }43\hbox{$.\!\!^{\prime\prime}$ }06$.

All the CFHT observations were done with the Multi-Object Spectrograph (MOS, Le Fèvre et al. 1994) with the O300 grism. The cameras used during each run and the corresponding pixel scales and dispersions are listed in Table 1. Two to five exposures per mask were obtained, depending on the magnitudes of the selected objects in each mask.

Band-limiting filters, chosen such that prominent spectral features (for instance the Ca I H/K lines blueward of the 4000 Å break) fall into the band at the redshift of interest, allow stacking of several rows of spectra on one mask, thus increasing the number of spectra observable in a given time. This strategy has been successfully employed by e.g. Yee et al. (1996, 2000). However, spectra covering a larger range of wavelengths provide more secure redshift determinations since more absorption and emission lines can be taken into account. This is particularly important in the presence of artifacts caused by insufficient removal of cosmic ray hits. In the case of Cl0024+1654 at z=0.395, the 4000 Å break roughly coincides with the strong sky emission line [O I]$\,\lambda$5577, so that insufficient subtraction of the sky line might cause a problem in the redshift determination if only a limited wavelength band were available. Also, covering a wide wavelength range is essential in order to derive spectral types for the galaxies and analyse in detail the spectral properties of the cluster members. For these reasons we did not use band-limiting filters and the usable wavelength range was typically 4500-8500 Å, depending on signal-to-noise ratio and the quality of the sky subtraction at the red end of the spectral range, where the sky emission is dominated by molecular bands.

The observations at WHT were made with the Low Dispersion Survey Spectrograph (LDSS-2, Allington-Smith et al. 1994), the med/blue grism and the Loral LOR1 detector. The usable wavelength range was 4000-7500 Å, somewhat bluer than for the MOS observations. All the LDSS-2 masks were covered by two exposures each.

We used $1\hbox{$^{\prime\prime}$ }$ wide slits throughout, resulting in a resolution of $\sim$13 Å, except for run 1 where the slit width used was $1\hbox{$.\!\!^{\prime\prime}$ }5$ with a correspondingly worse resolution of $\sim$20 Å.

 

 

Table 1: Log of the spectroscopic observing runs.

#	Date	Instrument	CCD	Grism	$N_{\rm masks}$	Exp. time	Pixel		Disp.	Std. star
						(ksec)	( $\mu{\rm m}$)	(arcsec)	(Å/pix)
1	24-27/08/92	CFHT/MOS	SAIC1	O300	2	4.5-7.5	18	0.377	4.31	Feige 110
2	24-27/08/95	CFHT/MOS	Loral-3	O300	6	6.6-15.6	15	0.314	3.69	Wolf 1346
3	12-15/09/96	WHT/LDSS-2	LOR1	med/blue	9	5.4-7.2	15	0.357	3.30	HZ4
4	11-13/11/96	CFHT/MOS	STIS-2	O300-1	3	6.6-8.1	21	0.440	5.03	Hiltner 600