Spectra, position, magnitudes and colors are presented for 485 faint (B<20.5) emission line objects selected with the ultraviolet-excess (UVX) criterion on a area of 24.6 sq. deg in the South Galactic Pole. The objects were selected from the analysis of pixel-to-pixel stacking of COSMOS scans of UKST U, J and R plates. The candidates were observed with the Meudon-ESO Fiber Optics System (MEFOS) at the ESO 3.6m telescope. 429 type 1 AGNs have been identified (373 in the redshift range 0.3<z<2.2). This sample has allowed the measure of a difference on the QSO clustering evolution in comparison with that found for galaxies (La Franca et al 1998). The region is part of the ESO Imaging Survey (EIS) and of the 2dF QSO redshift survey.

This is the first paper in a series aimed at defining a statistically significant sample of QSOs in the range $ 15 < B < 18.75$ and $ 0.3 < z < 2.2$. The selection is carried out using direct plates obtained at the ESO and UK Schmidt Telescopes, scanned with the COSMOS facility and searched for objects with an ultraviolet excess. Follow-up spectroscopy, carried out at ESO La Silla, is used to classify each candidate. In this initial paper, we describe the scientific objectives of the survey; the selection and observing techniques used. We present the first sample of 285 QSOs ($M_B < -23$) in a 153 deg$^2$ area, covered by the six ``deep'' fields, intended to obtain significant statistics down $B \simeq 18.75$ with unprecedented photometric accuracy. From this database, QSO counts are determined in the magnitude range $ 17 < B < 18.75$.

The sky subtraction performances of multi-fiber spectrographs are discussed, analysing in detail the case of the OPTOPUS system at the 3.6 meter ESO telescope at La Silla. A standard technique, based on flat-fields obtained with a uniformly illuminated screen on the dome, provides poor results. A new method has been developed, using the [OI] emission line at 5577 \AA\ as a calibrator of the fiber transmittance, taking into account the diffuse light and the influence of each fiber on the adjacent ones, and correcting for the effects of the image distortions on the sky sampling. In this way the accuracy of the sky subtraction improves from $2-8 \%$ to $1.3-1.6 \%$.