Indeed, we do not fully know what contribution galaxies fainter than the photographic plate limit (cf. the photometry by Godwin et al. 1983) make to the overall light (and mass) of Coma, and whether this contribution can significantly increase the known baryonic mass in the cluster. Futhermore, the exact shape of the Coma luminosity function (Bernstein et al. 1995, hereafter B95, Biviano et al. 1995, Lobo et al. 1997, Trentham 1998) remains in doubt, with suggestions varying from a single atypical Schechter function with a steep faint end slope to a Gaussian at bright magnitudes and a power-law at the faint end. This confusion comes from the reliance on a 2D statistical correction to subtract the background contribution. Most papers analyzing cluster luminosity functions, especially in distant clusters, use this method (see e.g. Driver et al. 1994, De Propris et al. 1995), but such a procedure could induce severe correlated errors and biases. Although the large error bars on these predictions do not totally prohibit a conclusive statement on potential errors in the 2D statistical background subtraction, it is crucial to check carefully this method. Coma is an excellent first candidate for this purpose. In the present paper, we assume H0 = 100 km s-1Mpc-1 and q0 = 0.
The photometric sample on which our study is based is one of the deepest images presently available of the Coma cluster core taken by B95. This sample is described in detail by Adami et al. (1998). The size of the field is 7.5×7.5 arcmin and is centered on coordinates 12h 57m 17s; 28° 09' 35'' (equinox 1950, as hereafter). The data are roughly complete to the CCD magnitude R = 25. For 99 of the 105 galaxies of the range [19; 21.5], a bj magnitude is also available (Godwin et al. 1983).
Additionally, we have checked the magnitudes of B95 by using the photographic b26.5 and r24.75 faintest magnitudes of Godwin et al. (1983). We have found 12 common galaxies in the two samples. Even if the two systems are not directly comparable, we find well constrained relations, which support the B95 photometry:
bj = (0.95±0.09) b26.5 + (0.54±0.22) and
R = (0.80±0.10) r24.75 + (-1.12±0.19)
Spectroscopy: the three CFH runs
We have observed spectroscopically some of the targets using the 3.6m CFHT with the MOS spectrograph during 2 nights in March 1996, 3 nights in March 1997 and 1 night in May 1998. Because of the bad weather, only half a night was available in total, during which we have obtained spectra for 43 objects out of the 105 selected from the total B95 sample (see Adami et al., 1998).
The redshifts were deduced by fitting gaussian profiles on well identified lines as well as by using cross correlation techniques. If there were emission and absorption lines in the same spectrum, we selected emission lines to deduce the redshift by fitting them with gaussian profiles. We finally obtained 29 spectra (Figure 2) with more than two emission lines for the emission line galaxies (ELG hereafter), or H &K absorption lines and the 4000Å break for the galaxies with only absorption lines (ALG hereafter).
For 5 other spectra, we had only one line available. We could deduce a redshift for 3 of them by using the shape of this line or the absence of other lines. For the 2 last ones, we have different possibilities, but we are sure that these galaxies do not belong to Coma.
The redshift distribution observed along the Coma line of sight
A possible explanation for such a lack of faint Coma members in this particular region of the cluster could be their tidal disruption in the core of the cluster, as described by S97.
A physical structure at z 0.5 beyond the Coma cluster
Moreover, all these galaxies are ALG and represent 60 % of all the ALG in the spectroscopic sample. If we keep only the galaxies with H & K lines (and with a template velocity when available), the corresponding velocity dispersion is 660 km s-1 (calculated at z = 0), which is consistent with a cluster of galaxies, but too high for a group of galaxies (e.g. Barton et al. 1996). Futhermore, this structure does not correspond to any of the Rose (1977), Hickson (1982) or Barton et al. (1996) groups. We note that four of these six galaxies have a bj - R color equal to 2.25, only moderately inconsistent with the range of values calculated by Frei & Gunn (1994) for the elliptical galaxies between z = 0.4 and 0.6: [2.43; 2.5]. We may have observed the brightest members of a distant cluster at z 0.5 beyond Coma.
As presented above, 34 redshifts (32 secure) have been obtained in the direction of the Coma cluster core for faint galaxies (5 do not have measured magnitudes). Out of these 34 galaxies, only 1 may have a redshift consistent with the Coma cluster, all the others being background objects. Notice that the S97 independent photometric survey leads to the same result. The number of galaxies found to be in Coma is barely consistent at the 1 level with the expected number of objects in Coma predicted by B95 (6±5). Such a result obviously questions the validity of the statistical subtraction of background objects to estimate cluster luminosity functions. Tidal disruption can be proposed as an explanation of the apparent deficit of faint galaxies in this region of the core of the Coma cluster. Moreover, the presence of a distant cluster on the same line of sight increases the field counts locally and could explain the discrepancies between observations and statistical field counts.
2 Department of Physics, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213, USA
3 Institut d'Astrophysique, CNRS, Université Pierre et Marie Curie, 98bis Bd Arago, F-75014 Paris, France
4 DAEC, Observatoire de Paris, Université Paris VII, CNRS (UA 173), F-92195 Meudon Cedex, France
5 Department of Astronomy and Astrophysics, University of Chicago, 5640 S. Ellis Avenue, Chicago, Illinois 60637, USA
6 Osservatorio Astronomico di Brera, via Brera 28, 20121 Milano, Italia
Editor: Dr. Timothy M. C. Abbott, firstname.lastname@example.org Copyright © 1998, CFHT Corporation. All rights reserved.