Due to the ill-determined primary beam correction and the rapid loss of sensitivity far from the pointing center, we searched for sources only the area where the primary beam attenuation is less than a factor of 3. With this constraint, we covered an area of $\sim$ 0.620 sq. deg. at 6 cm, and $\sim$ 0.525 sq. deg. at 3.6 cm. In Fig. 1 we show all the pointing positions: the small grey circles represent the VLA primary beam HPBW at 3.6 cm (4.9 $^\prime$ ), while the larger black circles those at 6 cm (8.6 $^\prime$ ). The dotted line show the boundaries of the area covered at both wavelengths ( $\sim$ 0.493 sq. deg.), the ISOGAL sub-fields are included in this area, the dashed lines mark the boundary of the field observed either at 6 and/or 3.6 cm ( $\sim$ 0.652 sq. deg.).

$\begin{figure} { \psfig {figure=ds1710f01.eps,height=7cm} }\end{figure}$

Figure 1: At each pointing position a circle with diameter equal to the VLA primary beam FWHM is shown. Grey circles represent 3.6 cm pointings, black circles 6 cm pointings. The dotted line marks the boundaries of the area observed at both frequencies, the dashed line encompasses the area observed at either of the two bands. In both cases we considered only the area where the a primary beam attenuation is less than a factor of 3. Axes are galactic longitude and latitude (degrees)

Frequent observations of the quasar 1922+155 were used for gain and phase calibration, while the flux density scale was determined by observations of 3C 286. The calibration is expected to be accurate within 10%. We imaged all the fields using the AIPS IMAGR task with natural weighting, the resulting synthesized beam varied somewhat from field to field depending on the hour angle at which each field was observed, typical FWHM values are $\sim 6^{\prime\prime}$ at 6 cm and $\sim 3^{\prime\prime}$ at 3.6 cm.

2.1 Sensitivity

Due to the VLA primary beam attenuation and the different noise values in the various fields, the sensitivity of our observations is not expected to be uniform accross the observed region. Using our knowledge of the VLA primary beam attenuation pattern and the measured on-axis rms level in each of the observed fields, we computed the sensitivity maps for our survey at 3.6 and 6 cm (see also Zoonematkermani et al. 1990 and Becker et al. 1994). The measured on-axis noise level in the maps is generally $\sim 0.12-0.15$ mJy/beam at both frequencies, with the exception of some fields close to the bright complexes located at $l\sim 45\hbox{$.\!\!^\circ$}10$ , $b\sim0\hbox{$.\!\!^\circ$}13$ ( $\alpha(2000)=19^{\rm h}13^{\rm m}27^{\rm s}$ $\delta(2000)=10^\circ 53^\prime35^{\prime\prime}$ )and $l\sim 45\hbox{$.\!\!^\circ$}45$ , $b\sim0\hbox{$.\!\!^\circ$}06$ ( $\alpha(2000)=19^{\rm h}14^{\rm m}21^{\rm s}$ $\delta(2000)=11^\circ 09^\prime13^{\prime\prime}$ )which have a higher noise level (in the range 1-8 mJy/beam) due to residual phase and amplitude errors.

$\begin{figure} \begin{tabular} {cc} \psfig {figure=ds1710f02a.eps,height=9cm,an... ...} \psfig {figure=ds1710f02b.eps,height=9cm,angle=-90} \end{tabular}\end{figure}$

Figure 2: Computed noise maps for the 3.6 and 6 cm observations (left and right, respectively). Dotted and dashed lines as in Fig. 1. The on-axis noise level in the black areas can be as high as 8 mJy/beam

The computed rms maps are shown in Fig. 2, the area of each pixel ( $10^{\prime\prime}\times 10^{\prime\prime}$ ) corresponds to $\sim$ 3.5 beams at 6 cm and $\sim$ 14 beams at 3.6 cm. As seen from Fig. 2 most of the area covered by our survey has a rms sensitivity less than 0.3 mJy/beam at both frequencies. In Fig. 3 we show the cumulative distributions of the pixel values in the rms maps, more than 85% of the surveyed area has an rms value less than 0.5 mJy/beam.

$\begin{figure} { \psfig {figure=ds1710f03.eps,height=6.5cm} }\end{figure}$

Figure 3: Cumulative distributions of the noise values in the maps of Fig. 2. The grey line is for 3.6 cm data, the black line for 6 cm data

2.2 Source extraction

All images were inspected by means of contour plots and greyscale display to find sources. The images were then corrected for primary beam attenuation using the AIPS task PBCOR before source extraction.

The J2000.0 positions, peak and integrated flux densities and the sizes of the detected sources at both frequencies are listed in Table 1. In general, all the reported detections have a signal-to-noise ratio greater than five in at least one of the two bands. The names assigned to the sources have been derived from their galactic coordinates, as in Becker et al. (1994).

We arbitrarily divided the sources into two categories: 1) compact sources and 2) extended sources. In the first group are all the unresolved sources or the sources with deconvolved sizes of the same order or smaller than the synthesized beam FWHM. All the extended sources have sizes much greater than the synthesized FWHM, and thus they may be partially resolved out by the interferometer. The flux densities (and sizes) reported in Table 1 for these sources should be considered as lower limits. We shall see in the following that this arbitrary distinction based on observational considerations reflects also some intrinsic physical difference between the two groups.

At large distances from the pointing center, the correction factor due to the VLA primary beam attenuation may be large, and hence the source flux density could be ill-determined. In Table 1 the source that has the maximum correction factor applied is source #5, which is close to the edge of the surveyed area and has correction factors $\sim$ 2.1 at 6 cm and $\sim$ 2.5 at 3.6 cm. All other sources, with the exception of #22 and #29 at 6 cm, have correction factors lower than 2.0.

The positions of all the detected sources within our surveyed region are shown in Fig. 4a, where pluses represent 3.6 cm sources (21), and circles represent 6 cm sources (29), and larger symbols represent extended sources. Contour plots for all the detected sources are shown in the Appendix.

$\begin{figure} \begin{tabular} {ccc} \psfig {figure=ds1710f04a.eps,height=8.5cm... ... \psfig {figure=ds1710f04c.eps,height=8.5cm,angle=-90} \end{tabular}\end{figure}$

Figure 4: a) Positions of the detected sources at 3.6 cm (pluses) and 6 cm (empty circles), larger symbols represent extended sources; b) VLA 20 cm surveys: sources from Zoonematkermani et al. (1990; ZGS) are shown as pluses, grey contours show the NVSS image of the region; c) the position of the IRAS-PSC2 sources inside our extended survey area are shown as plus symbols, filled squares show the five sources which satisfy the Wood & Churchwell (1989a) color criteria

**Table 1:** Detected radio sources
$\begin{table} \vskip 0.3cm \noindent $^a$) Derived from galactic coordinates, as... ...ng flux has been assigned to source \char93 33.\\ $^f$) Not observed.\end{table}$

2 Observations and data reduction

2.1 Sensitivity

2.2 Source extraction