5 Discussion

The main aim of these high resolution observations was the morphological classification of a group of CSS sources found in a larger, flux-limited sample mainly consisting of extended radio sources. Like the majority of CSS sources studied in previous works, the objects observed here show mainly a double-lobed structure, often asymmetric. The structures revealed by these observations are generally resolved and they may well represent the lobes of small-sized powerful radio sources. In fact, for the two sources observed at both 8.4 and 2.3 GHz, only steep spectrum components have been detected. The present observations do not reveal any definite core candidates, but in one case ( 1814-637) there is a weak third component between the lobes which could be suggested as a possible core. Weak tails are found pointing towards the source centre in a few cases, further suggesting that we are seeing two-sided emission from lobes and possibly hot-spots.

Figure 10: VLBI images of PKS 1934-638 at 2.3, 4.8 and 8.4 GHz (from Tzioumis et al. 1997). The top-right image is at a higher resolution because it includes Hartebeesthoek data. The other three images have approximately the same resolution of about 5 mas.

Generally, these images do not show many highly distorted or complex structures, which could be ascribed either to the interaction between the radio emitting plasma and the ambient medium or to projection effects. Even in the object which shows the most complex structure ( 1814-637), the basic double-lobed morphology is still clearly evident. Source linear sizes are in the range 0.08-2.9 kpc, and hence all sources in this sample can be classified as compact symmetric (CSO) or medium-sized symmetric (MSO) objects if we assume that an undetected core sits between the lobes revealed by the present observations.

Dallacasa et al. (1995) have investigated a sample of northern CSS sources (using MERLIN and EVN) selected from the Peacock & Wall catalogue (1981). As their sample and the present sample have the same selection and observing frequencies, it is expected that sources in both samples should have, on average, similar characteristics. CSS sources selected from the Wall & Peacock sample tend to have a higher turnover frequency and smaller size (e.g. Dallacasa et al. 1995) because of the higher selection frequency (i.e. 2.7 GHz) compared with the 3CR sample (178 MHz).

The size of the sources in the present sample range between $\sim$0.08 and $\sim$2.9 kpc, similar to the range found in the sample studied in Dallacasa et al. (1995), and smaller than the 3CR samples (Fanti et al. 1990 and references therein). Turnover frequencies are in the range around 0.1-0.2 GHz or below, with the exception of 1934-63 that peaks at $\sim$1.4 GHz. As expected from the correlation between the linear size of the source and the turnover frequency (Fanti et al. 1990), the smallest source in our sample is the GPS source 1934-638, that has the turnover at higher frequency.

Dallacasa et al. (1995) also find that most of galaxies their PW CSS sample show weak core candidates at $\la2$% of the total flux density, in contrast with the 3CR samples, where, however, the dynamic range was smaller and comparable to these achieved for most of the sources presented in this paper.

Of the seven CSS/GPS objects presented in this paper, four have been searched for HI absorption (0023-26, 1814-63, 1934-63 and 2135-20). The results have been summarised in Morganti et al. (2001) while the preliminary results from new VLBI observations for 1814-63 have been presented in Morganti et al. (2000).

Of the four objects, three are detected. The only one undetected is 2135-209, but this result is not too surprising given that it is classified as broad-line galaxy. It is interesting, however, that the most impressive HI absorption is detected in the object with the most complex radio structure in our small sample, 1814-63. The deep HI absorption appears extended and cover most of the source with the exception of the southern part of the southern lobe. The peak optical depth goes from $\sim$30% in the southern lobe to $\sim$10% in the northern region. The component with lower optical depth (ranging between 2 and 4%) is observed only against the northern region. Using the available systemic velocity ( $V_{\rm hel}=19~350$ km s^-1, Morganti et al. 2001), most of the H  I absorption appears to be blueshifted. This can be explained by the presence of extended gas, possibly surrounding the lobes, interacting/expanding with the radio plasma. Therefore, in the case of 1814-63, part of the radio morphology could be affected by a strong interaction with the ISM.

Overall, the properties of the sources discussed here are consistent with the results from the analysis of other samples of CSS sources. In a forthcoming paper we will discuss their properties in wavebands different from radio, compare them with those of extended radio sources and verify whether they support and "youth'' scenario.