2 The construction of the catalogue

Our survey is based on a set of UKST plates in the U, $B_{\rm J}$ and R bandpass (see Blair & Gilmore 1982 for a definition of the various systems). UKST plates subtend $6\hbox{$.\!\!^\circ$ }4\times 6\hbox{$.\!\!^\circ$ }4$. Details of the photographic plate material are listed in Table 1. The plates have been scanned in Mapping and Imaging Mode on the COSMOS microdensitometer (MacGillivray & Stobie 1984). The mapping mode used a spot size of 16 $\mu$m (FWHM), corresponding to 1 arcsec. For each band-pass the digitized Mapping Mode data have been added together in order to increase the signal to noise ratio. The reliability of the co-adding technique has been verified by many tests on UKST plates (Hawkins 1991). The resulting coadded digitized data has been analyzed by the COSMOS crowded-field analysis software (Beard et al. 1990). The resulting tables contained the instrumental magnitudes, the area above the threshold, the intensity weighted x and ypositions, and other useful parameters. We discriminated between point-like and extended sources using the COSMOS image parameters, that is, on morphological grounds. We defined a plane in which stellar locus was well determined, and drew the separation line near the locus. For typically $B_{\rm J}<19.5$, we used the log(isophotal area) versus magnitude plane. At fainter magnitudes, where stellar peak surface brightness ( $I_{{\rm peak}}$) are not saturated, the log (I_peak-I_sky)/I_skyversus magnitude plane provided a better separation. The two separation lines produced the same star-galaxy ratio at the overlap magnitude. Relative photometry, using instrumental magnitudes, was produced by COSMOS. Calibration was based on published sequences (Hawkins & Bessell 1988; Warren et al. 1991a).

 

 

Table 1: UKST photographic plates

emulsion	filter	number	date	exp.
				time
				(min)
IIaO	UG1	U 2639	1976-09-28	180 IIIaJ

The candidates were extracted from a rectangle in the sky centered at $\alpha$(1950): $00^{{\rm h}}$ $50^{{\rm m}}$ $34.3^{{\rm s}}$ and $\delta$(1950): $-28\hbox{$^\circ$ }$ $10\hbox{$^\prime$ }$ $08\hbox{$^{\prime\prime}$ }$ with limits $\Delta\alpha_{{\rm min}}=-2.170^{\circ}$, $\Delta\alpha_{{\rm max}}=2.715^{\circ}$, $\Delta\delta_{{\rm min}}=-2.520^{\circ}$, $\Delta\delta_{{\rm max}}=2.520^{\circ}$ (i.e. the limits in right ascension are the two lines distant $\Delta\alpha$ from the meridian at $\alpha$(1950): $00^{{\rm h}}$ $50^{{\rm m}}$ $34.3^{{\rm s}}$; see Cristiani et al. (1995), their Table 1). All the objects inside a radius of 0.15 degrees centered at $\alpha$(1950): $00^{{\rm h}}$ $50^{{\rm m}}$ $17.43^{\rm s}$, and $\delta$(1950): $-26\hbox{$^\circ$ }$ $51\hbox{$^\prime$ }$ $25.2\hbox{$^{\prime\prime}$ }$have been excluded from the catalogue as this region includes a crowded globular cluster. It results a total area of 24.55 deg².

We have selected as candidates all the UVx "not extremely extended'' objects with $B_{\rm J}\leq 20.5$, satisfying a type of modified Braccesi less-restricted two color criterion (La Franca et al. 1992; Cristiani et al. 1995). The completeness of the selection has been tested against the QSOs already known in the field with redshift in the range $0.3<z\leq2.2$. We have selected $92\%$ of the 176 blue QSOs already known in the field with $B_{\rm J}\leq 20.5$. The completeness becomes $89\%$ if the "red'' QSOS from Warren et al. (1991a) are included in the comparison.