2 Observations and data reduction

The observations reported here were carried out between May 1994 and January 1996. The 4-feed receiver system installed in the secondary focus of the 100-m telescope was used in a multi-beam mode. Each horn feeds a 2-channel receiver with an IF polarimeter providing full Stokes information simultaneously. The system temperature was $\sim$80 K on the sky (zenith), the effective bandwidth was 300 MHz. In the beginning of 1995 the receiver's band centre had to be moved from 10.55 GHz to 10.45 GHz in order to avoid the new ASTRA 1D satellite. This is only a change of 1% in frequency, which will not have any noticeable influence on the observed source properties (for a source with spectral index $\alpha = -1$ this implies a 1% change in flux density; $S_{\nu} \sim \nu^{\alpha}$). The nominal half-power beam width is 69$^{\prime\prime}$. The total number of sources observed was 1050.

2.1 Cross-scans

Almost all the sources ($95\%$) were observed with cross-scans, with the main beam scanning a distance of 7$^{\prime}$ to ensure adequate baselines. The scanning speed was 30$^{\prime}$/min. The offset feeds were used to efficiently remove atmospheric noise. For sources less extended than 30$^{\prime\prime}$ the cross-scans were oriented in right ascension and declination. In the case of more extended sources, the cross-scans were oriented with one scan direction along the sources' major axes (e.g. along double or triple components), with the scan length increased to 10$^{\prime}$. This orientation was taken from the VLA maps of Vigotti et al. (1989). Depending on the expected flux density, the total number of such cross-scans was chosen to be between 8 and 64. Individual subscans were checked for interference or residual atmospheric fluctuations and discarded prior to averaging if necessary. We evaluated the differential signals between the main horn and two of the reference horns, which have beam throws of +3$^{\prime}$ and $-5^{\prime}$ in azimuth. This allowed a proper judgement of the data quality and enabled us to recognize confusing sources that had been accidentally scanned across by the reference feeds, thus causing a negative response: the probability of picking up unrelated background sources in both reference beams simultaneously is very small so that such a negative response only shows up in one of the two recorded differential signals. The data in the final cross-scans were sampled at 18$^{\prime\prime}$ intervals. With the above scanning speed, this implies an integration time of 0.6 second in each subscan. Averaging also the two scanning directions we obtain a nominal rms noise in the final cross-scans between $\sim$2.4 mJy/b.a. (8 scans) and $\sim$0.9 mJy/b.a. (64 scans).

  

Figure 1: Examples of cross-scans at $\lambda$2.8 cm: a) a 4 mJy source, b) a 20 mJy source, c) a 100 mJy source, d) a double source

The actually measured rms noise values were generally somewhat higher because of residual atmospheric noise. A fit with one Gaussian was applied to the final scan yielding the amplitude, the width, and the position of the centroid of the Gaussian for Stokes parameters I, Q and U. For the double sources with diameters larger than 40$^{\prime\prime}$ we applied a fit with two Gaussians only to the scan along the major axis of the radio source. The decomposition was successful for 74 sources, whose data have S/N larger than 10. As mentioned in the introduction, the linear polarization of the sources will be presented in a future paper. Figure 1 displays some template plots of the cross-scans for sources with different flux densities.

Standard calibration sources were cross-scanned at regular intervals (about every two hours, with two cross-scans each) to check the telescope pointing and flux density scale. For the latter purpose the primary calibrators were 3C286 and 3C295, with 3C48 and 3C138 being used as secondaries. The pointing accuracy was found to be stable to within $\sim\!3^{\prime\prime}$, sufficiently good to ensure reliable flux density measurements. The exact flux density scale for each target source was applied by checking two subsequent observations of calibration sources. The calibrated flux densities are on the flux density scale of Baars et al. (1977).

In order to recover the total flux the source extension can be used. Therefore we computed the FWHM obtained from the Gaussian fit of the data for point-like sources obtaining the following results: a mean value of $70^{\prime\prime}\, \pm4\, ^{\prime\prime}$ for sources stronger than 50 mJy, and for fainter sources $71^{\prime\prime}\, \pm\, 10^{\prime\prime}$. The spread of FWHM found permits to recover the total flux with an error of up to 15%. We then decided to determine the integrated flux using a simulation program. The correction factor has been applied to all double and diffuse sources with an angular extent between 20$^{\prime\prime}$ and 40$^{\prime\prime}$ and, in addition, to more extended sources where the deconvolution could not be done because of low S/N.

Our simulation program was built using two point-like components with a flux density ratio R₂₀ obtained from the VLA 1.4-GHz maps convolved with the Effelsberg beam (HPBW = 69$^{\prime\prime}$). We had to use R₂₀ instead of the unknown flux density ratio of source components at 10.6 GHz (R_2.8); however, Fig. 4b (see below) will show that R_2.8 changes by up to a factor of 2. Our simulation shows that this introduces an additional average error of less than 4$\%$.

Another simulation, which takes into account the extended brightness distribution, was used to compute the correction factor for the diffuse sources. The factor for double and diffuse sources is in the range between $\sim$10$\%$ and $\sim$35$\%$.

2.2 Maps

A total of 53 sources were mapped: they have complex structures larger than 70$^{\prime\prime}$. The map sizes were adjusted such as to ensure sufficient baseline areas and accounting at the same time for the maximum beam throw of the four-feed system, which is 17$^{\prime}$. The standard mapping technique with this receiver system was described by Gregorini et al. (1992). All four horns were employed by observing in the multi-beam mode and applying the restoration algorithm of Emerson et al. (1979). Depending on the strength of the sources, between 2 and 14 coverages were obtained by scanning the telescope in azimuth and separating subsequent scans in elevation by 20$^{\prime\prime}$. After restoration to the equivalent single-beam map the individual coverages were averaged (in I, Q, and U) and then interpolated onto a grid in right ascension and declination. Also in the course of the mapping campaigns standard calibrators were observed and processed in the same way as the target sources. All maps were numerically integrated to yield total flux densities which were also brought to the flux density scale of Baars et al. (1977) by comparison with the mapped flux calibrators. Some of the maps show considerable detail. These will be displayed and briefly described in Sect. 3.