Figure 1 (click here) shows that 82% of the sources are within
50
of the expected GB96 position and the
distribution peaks at an offset of 
, consistent with
the positional errors given in Neumann et al. (1994)
and Gregory & Condon (1991). Because the positional
uncertainty of the Green Bank positions (which greatly dominate the
uncertainties in the VLA positions) are flux dependent, we split the data
into two subsets: a bright and faint sample with a division at 75 mJy. We
find that the positional accuracies given in Neumann et
al. (1994) are then reliable for these two samples. For the
analyses that follow, we therefore exclude all bright sources offset from
the GB positions by more than 45
and all faint sources offset by
more than 120
. These criteria exclude 74 objects from the bright
sample and 47 objects from the faint. The number of spurious coincidences
which remain in the tables is quite small with only 
31 of the 757
remaining faint sources and 
10 of the 813 remaining bright sources
expected to be spurious. Flags are given in Table 2 to indicate the sources
we excluded.
Figure 2 (click here) shows the flux-flux diagram comparing the core VLA flux
density with the Green Bank measurement. For those 
10% of the fields
for which more than one source was detected, we have chosen as the ``core''
component that source which is closest to the RASS position. We also
considered using the brightest source on the field or the source closest to
the GB position. In all the following analyses, the differences between the
results obtained using these three different criterion are well within the
uncertainties in the data. Only when we completely excluded the fields with
more than one detected source did any of the results change significantly,
increasing the median core-to-lobe parameter (Sect. 5) by 
20%. While
source variability and uncertainties in both flux density measurements
produces a few points in Fig. 2 (click here) where 
, it is clear that nearly all of the sources in the GB catalog
contain emission which is significantly resolved at the arcsecond-scale.
Figure 3: a) Distribution of 
, defined as the (VLA core
flux density)/(Green Bank flux density - VLA core flux density). This
parameter is a rough measure of the radio core-to-lobe ratio. The last bin
contains the 118 sources for which the Green Bank flux density is less than
the VLA core flux density. b) Distribution of the core-to-lobe ratio
for radio galaxies. Bins containing undetected cores are denoted with a
left-facing arrow. c) Distribution of the core-to-lobe ratio for BL
Lacertae objects. The hatched histogram distinguishes the radio-selected from
the X-ray-selected objects