The data presented in this paper complete the information about a sample of small-sized sources (Mantovani et al. 1992, 1994). The previous paper (Mantovani et al. 1994) presented the results on sources which were generally of small size and had high rotation measures. All but one of the sources discussed in that paper have a linear size <14 kpc. The source 0725+147 is larger with a projected linear size of 25 kpc, and was found to have rotation measures of about +545 and -2576 rad m-2 for the two lobes in the rest frame of the source. This object is associated with a quasar at a redshift of 1.387 and has an absorption-line system at a similar redshift (Hewitt & Burbidge 1993).
Among the sources presented here, 0320+053, 2033+187 and 2147+145 are the most compact with the former two being unresolved with the VLA A-array at 15GHz. This implies an angular dimension <100 mas, which corresponds to a linear size <0.3 kpc for z=0.5. All of these are not polarized at the detection levels of these observations.
Two sources, 0809-056 and 1239-044, have an overall structure that is similar to 0725+147. However, unlike 0725+147, they do not show evidence of large rotation measure. The source 0358+004 also has a triple structure, which is much more asymmetric compared to the above group of objects. All of them have a linear size in the range 20-25 kpc, a bit larger than the limit set by Fanti et al. (1995) for a steep spectrum source to be considered a member of the class of Compact Steep-spectrum Sources.
The two last sources in the sample, 1422+202 and 1741+279, each show a one-sided jet and an overall asymmetric structure, with a weak lobe of emission on the counter-jet side which is located closer to the nucleus. The jets are knotty and the one in 1741+279 has a wiggling structure which deserves further investigation. Moreover, in the case of 1741+279 there is a clear indication of a large bend in the jet direction close to the component which is likely to host the nucleus.
The above five sources, including 0725+147, have a linear size which is
somewhat larger (between about 21 and 43 kpc) than the "required" size
of 
kpc to be classified as CSSs. They can be interpreted as
medium sized objects which are the more evolved versions of the
CSSs, which are assumed to be young rather than frustrated
objects (Fanti et al. 1995; Readhead et al. 1996).
We have summarized in Table 3 (click here) the overall polarization parameters derived for the five polarized sources of the group. It is worth mentioning that the percentage polarization of these sources is rather high (in the range of about 6-13% at 8.4GHz) and that they depolarize going from higher to lower frequencies. The percentage polarization is significantly higher than that found for the compact sources with high RM presented in Mantovani et al. (1994). The four compact sources show a percentage polarization <3, while only 0725+147 has a percentage polarization comparable to the above values. The fact that the percentage polarization shown by some of the components is sometimes lower at 15GHz than at 8.4GHz is probably due to the fact that the sensitivity of the observations to the low brightness emission was insufficient in those cases.
Acknowledgements
The authors wish to think the referee, Dr. Ian W.A. Browne for his suggestions. FM thanks Miller Goss, Assistant Director, NRAO, Socorro, for his hospitality during period when part of the work was done. The National Radio Astronomy Observatory is operated by Associated Universities Inc., under cooperative agreement with the National Science Foundation; AIPS is NRAO's Astronomical Image Processing System.
