2 The data

In the following two sections we recall the properties of the radiogalaxy sample and of the optical galaxy catalogue that have been used during this search. For more details on the radio source catalogue and the definition of the radiogalaxy sample, refer to Paper I.

  
2.1 The radiogalaxy sample

The radio source catalogue has been extracted from 31 maps of the 1.4 GHz NRAO VLA Sky Survey (Condon et al. 1998) and consists of $13\,340$ pointlike and 2662 double radio sources down to a flux limit of 2.5 mJy beam^-1  over an area of $\approx$550 sq degrees at the South Galactic Pole.

Optical identifications of NVSS radio sources have been made with galaxies brighter than $b_{\rm J} = 20.0$ in the EDSGC catalogue (Nichol et al. 2000) using a search radius of $15\hbox{$^{\prime\prime}$ }$, i.e. $\approx$$3\sigma$ positional accuracy for the faintest sources. The initial sample of optical counterparts consists of 1288 radiogalaxies, 926 of them having a pointlike radio morphology at the NVSS resolution of $45\hbox{$^{\prime\prime}$ }$.

As shown in Table 2 of Paper I, the contamination level due to spurious identifications varies according to the radio morphological classification, ranging from about $16\%$ for the lists of optical counterparts of pointlike radio sources and "close'' radio pairs (separation between components $D \le 50\hbox{$^{\prime\prime}$ }$), to about $28\%$ for the list of optical counterparts of "wide'' radio doubles ( $50\hbox{$^{\prime\prime}$ }< D < 100\hbox{$^{\prime\prime}$ }$).

In order to obtain a more reliable sample, the radiogalaxy data set used in the search of candidate clusters has been selected among these optical identifications on the basis of radio-optical distance and galaxy magnitude. The uncertainty in the optical identification sample is indeed the only source of contamination that can be limited when selecting cluster candidates by looking for excesses in surface galaxy density near the identified radiogalaxies. Other contamination terms - like the probability of detecting a candidate by chance coincidence of the radiogalaxy position with an optical density excess, or the possibility that the optical excess itself is intrinsically spurious, i.e. due to chance superpositions of galaxies along the line of sight - cannot in fact be reduced unless one knows the redshift distribution of the galaxies.

From the initial sample of optical counterparts we thus selected those radiogalaxies having $d_{\rm r-o} \le 7\hbox{$^{\prime\prime}$ }$. This constraint introduces a selection effect against faint sources in the radio sample, whose positional uncertainty is typically $\sim$ $5\hbox{$^{\prime\prime}$ }$.

Furthermore, as our aim is to select candidate clusters at intermediate redshifts, we discarded those radiogalaxies brighter than magnitude $b_{\rm J} = 17.5$. In fact, considering the magnitude - redshift relation typical of radiogalaxies obtained in the R band by Grueff & Vigotti (1977), and using color indexes for elliptical galaxies given in Frei & Gunn (1994), this cut in $b_{\rm J}$ magnitude corresponds to a redshift lower limit of $z \mathrel{\rlap{\lower 3pt\hbox{$\sim$ }} \raise 2.0pt\hbox{$>$ }}0.1$.

With these constraints, the final radiogalaxy sample that has been taken into account for the search of candidate clusters consists of 661 radiogalaxies, and the mean, expected contamination level due to spurious optical identifications has been lowered to about $10\%$.

  
2.2 The galaxy catalogue

The Edinburgh-Durham Southern Galaxy Catalogue (EDSGC, Nichol et al. 2000) lists $\approx$ $1.5 \times 10^6$ galaxies over a contiguous area of $\sim$1200 sq degrees at the South Galactic Pole. About one half of this area is currently covered by our radiogalaxy sample and has been considered for the search of cluster candidates.

The EDSGC has been obtained from COSMOS scans of IIIa-J ESO/SERC plates at high galactic latitude ( ${\mid b_{\rm II} \mid} \ge 20\hbox{$^\circ$ }$). The automated star-galaxy separation algorithm used for the EDSGC guarantees a completeness >$95\%$ and a stellar contamination <$12\%$ down to magnitudes $b_{\rm J} = 20.0$.

Magnitudes have been calibrated via CCD sequences, providing a plate-to-plate accuracy of $\Delta b_{\rm J}\simeq 0.1$ and an rms plate zero-point offset of 0.05 magnitudes.

The EDSGC incompleteness starts to exceed the $5\%$ only above $b_{\rm J} =20.5$ (Collins et al. 1992). When looking for candidate clusters we thus decided to make optical galaxy counts down to the magnitude limit $b_{\rm J} =20.5$: as the radiogalaxy sample reaches $b_{\rm J} = 20.0$, this choice makes it possible to point out also those regions of high galaxy surface density associated to the optically faintest radio sources in our sample.