2 Sample pre-selection

We considered all the NVSS maps above $+60^{\circ}$ of declination (covering an area of $\pi$ steradians) and proceeded to a careful visual inspection of the contour map plots (at a 2$\sigma$ level). Each $4^{\circ}\times4^{\circ}$ NVSS map was divided into 16 $1^{\circ}\times1^{\circ}$ maps to facilitate the search for candidates. We then pre-selected those map features apparently related to a single source, with a total flux density $\ge$100 mJy and an angular extension larger than 4 $\hbox {$^\prime $ }$.

Flux density measurements on the NVSS maps were done using the task TVSTAT in AIPS, defining a polygonal area embracing all the source emission. For the computation of the angular extension we considered the maximum distance between contours at 3$\sigma$ level when the source was extended and diffuse with no evident sub-structure, or the distance between peaks of brightness when there were unresolved features (hotspots) at the source extremes. We took into account, when possible, curvatures in the radio structure of the sources so that a measured angular size corresponds to the length along the "spine'' of the radio source. A total of 122 sources were pre-selected for subsequent confirmation through higher resolution radio observations.

We note that due to the existence of "holes'' in the NVSS maps at the time of our search for candidates, a few sources fulfilling the requirements of the sample may have been skipped in the selection process and missed from the final sample. In fact, the known giant radio galaxies 4C+73.08 (J0949+732) and 4C+74.26 (J2042+751) were not selected for the pre-sample at this first stage, but were added later to the final sample. We note however that possible missing sources will not affect the statistical considerations derived from the sample.