Astron. Astrophys. 325, 954-960 (1997)

2. General description of the survey

A detailed description of the survey will be presented in the paper containing the whole data catalogue (Vettolani et al. 1997, Paper III, in preparation). Here we briefly summarize the main characteristics of the ESP project.

2.1. The photometric sample

The main area of the ESP survey is a strip long in right ascension and thick in declination (hereafter strip A). In order to make full use of the allotted nights, we were able to survey also an area of (hereafter strip B), five degrees west of strip A. Both strips are located in the region of the South Galactic Pole. The position was chosen in order to minimize the galactic absorption effects (). The right ascension limits are from to for strip B and from to for strip A, at the mean declination (1950).

The target galaxies, with a limiting magnitude , were extracted from the Edinburgh-Durham Southern Galaxy Catalogue (EDSGC, Heydon-Dumbleton et al. 1988, 1989), which has been obtained from COSMOS scans of SERC J survey plates. The EDSGC has a completeness at and an estimated stellar contamination (Heydon-Dumbleton et al. 1989).

Preliminary analysis of CCD data, obtained with the 0.9m Dutch/ESO telescope for about 80 galaxies in the magnitude range in the region of our survey, shows a linear relation between (EDSGC) and (CCD), with a dispersion () of about 0.2 magnitudes around the fit (Garilli et al. in preparation). Since the CCD pointings cover the entire right ascension range of our survey, this includes both statistical errors within single plates and possible plate-to-plate zero point variations.

We do not have enough information to assess the reliability of magnitudes for the 71 ESP galaxies with . It is known that in this bright range various problems may affect the measures, as saturation, ghosts and spikes on plates, the presence of substructure in bright galaxies, or contamination from overlapping bright stars (e.g. Loveday 1996). With this caveat in mind, we simply note that presently we do not have evidence for large errors on the magnitudes of bright galaxies in our catalogue, on the basis of a few cases we could check.

2.2. Observations and data reduction

The observations have been obtained during six observing runs (in the period 1991-1993) with the multifiber spectrograph OPTOPUS (Lund 1986, Avila et al. 1989) at the Cassegrain focus of the ESO 3.6m telescope at La Silla. The geometry of the survey (see Fig. 1) is a regular grid consisting of two rows of adjacent circular fields. Each field has a diameter of 32 arcmin, corresponding to the field of view of the OPTOPUS spectrograph. The two rows of fields are separated by 15 arcmin and slightly overlap each other. The total solid angle is steradians (i.e. 23.3 square degrees).

Fig. 1. Survey geometry.

The instrument OPTOPUS has 50 fibers, each with a diameter projected on the sky of arcsec. The fibers are manually plugged into holes drilled in aluminum plates at the galaxy positions. For each field we reserved at least 5 fibers for the measurement of the spectrum of the blank sky. On average, there are 35 galaxies brighter than per OPTOPUS field and of the fields have fewer than 45 galaxies. When the number of galaxies in a field exceeded the number of available fibers, we selected at random the galaxies to be observed from the total galaxy list. Whenever possible, we re-observed overdense fields in order to reach a higher completeness.

Another set of observations has been obtained in October 1994 with the multifiber MEFOS spectrograph (Felenbok et al. 1997). Thanks to these observations we have been able to observe some of the galaxy spectra that had no redshift determination after the OPTOPUS runs. In particular, we have used these observations to reach a completeness as uniform as possible over all the OPTOPUS fields in strip A.

The spectra cover the wavelength range from 3730Å to 6050Å, with an average pixel size of 4.5Å. We measured the redshifts of the galaxies by cross-correlating the sky-subtracted spectra with a set of 8 template stars. The template stars have been observed with the same instrumental set-up used to obtain the galaxy spectra. We also measured redshifts from the emission lines, whenever present. The median internal velocity error is of the order of km/s. From a comparison of our 8 templates with three SAO radial velocity standard stars we estimate that the zero-point error should be smaller than km/s. We will give a full description of our observations and our reduction procedure in Paper III.

2.3. Sample statistics

We observed a total of 4044 objects, corresponding to of the parent photometric sample of 4487 objects. Out of the 4044 observed objects, 493 turned out to be stars and 208 have a too low signal-to-noise ratio to provide a reliable redshift (failed spectra). In the end, our final sample consists of a total of 3342 galaxies with reliable redshifts ( one quasar). About half of the galaxies in our sample have detectable emission lines (see Sect. 3). In Table 1 we report a summary of the basic numbers relative to strip A and strip B.

Table 1. Sample statistics

The mean completeness of our galaxy sample is . We derive this value by assuming that all the failed spectra correspond to galaxies and that the percentage of stars among the 443 objects that were not observed is the same as in the spectroscopic sample (i.e. ). The completeness level is significantly different in strip A and in strip B, being for strip A and for strip B. The different completeness level between the two strips is due to our choice of repeating observations only for fields in strip A. Fig. 2 shows the completeness for each OPTOPUS field: panel a) refers to the northern row (field numbers ), panel b) to the southern row (field numbers ).

Fig. 2. Redshift completeness for each OPTOPUS field. Field numbers increase with right ascension. a Northern row. Field numbers from 1 to 9 correspond to region B; field numbers from 21 to 65 correspond to region A. b Southern row. Field numbers from 101 to 109 correspond to region B; field numbers from 121 to 164 correspond to region A.

A detailed understanding of the completeness and the selection effects is extremely important in any analysis of our data. We therefore performed a number of tests to assess the statistical properties of the galaxies for which we did not measure the spectrum or we did not get a measurable spectrum.

First of all, in the fields where the number of objects is higher than the number of available fibers, the objects we observed are a random subset of the total catalog with respect to both magnitude and surface brightness. The set of observed galaxies departs from randomness only with respect to the selection of the position of the target objects, due to the instrumental constraint that two fibers can not be put closer than about 25 arcsec; obviously this means that when dealing with pairs of galaxies at small projected separation, we could observe only one galaxy (except when the other galaxy was selected for a second exposure).

As far as failed spectra are concerned, we find that the fraction of failed spectra increases at fainter magnitude. However, the fractional increase of failed spectra at fainter magnitudes is not strong, reaching % at the limit . This effect is taken into account in our statistical analysis of the survey (see for example Paper II).

© European Southern Observatory (ESO) 1997

Online publication: April 28, 1998

helpdesk.link@springer.de