originally posted to sci.astro.research, in response to a question about radio galaxies from Ross Tessien
Ross is asking about unified models for radio galaxies and quasars (though he doesn't realise it.) The number of objects with observed superluminal motion is pretty small, actually, and this is because the measurements are difficult to make; you need more than one epoch of VLBI measurement. Some objects, such as 3C273 or 3C345, are very well studied, but in general the sample size is small and selection effects are a problem. So superluminal motion, while a good way of telling whether an object is relativistic and close to the line of sight, is not the most common diagnostic.
For convenience, let's divide the most powerful radio sources into (classical double) radio galaxies, lobe-dominated quasars and core-dominated quasars. When observed on kiloparsec scales in the radio, the first two classes look similar; they both have radio lobes, jets, and a central core or nucleus centred on the host object (an elliptical galaxy in the case of radio galaxies and a quasar in an elliptical in the case of radio-loud quasars). Lobe-dominated quasars invariably have brighter cores and brighter, more one-sided jets, however. (In fact, until recently very few jets had been detected in this type of radio galaxy at all. This has changed with better observations.) In the late eighties it was realised that this could be explained if radio galaxies and lobe-dominated quasars were actually the same sort of object, with the brightness differences in the radio cores and jets being explained as a relativistic Doppler beaming effect (radiation from a moving object is beamed in its direction of motion and suppressed transverse to the direction of motion). In this case the radio galaxies lie in or near the plane of the sky, and the lobe-dominated quasars lie at a smaller angle to the line of sight. Barthel (ApJ 336, 606, 1989) showed that for a well-defined sample of radio galaxies and quasars the number statistics and the linear sizes are consistent with a critical angle of about 45 degrees. Some care needs to be taken when looking at linear sizes, because of linear size (apparent) evolution and selection effects, and there is still some controversy over this, but the basic model seems OK. The core-dominated quasars, which as their name suggests are dominated in radio flux by their central components, can then be tacked on to this model as objects at a very small angle to the line of sight; their radio structures on the large scale seem to be consistent with this (i.e. you often see structures which could be the two lobes projected on top of one another). Various people have then looked at the statistics of cores and jets to try to determine what the velocities required are; the answers vary, but it looks like high velocities are needed in the core (v/c around 0.99) and that lower velocities, perhaps as low as v/c around 0.6, are adequate to explain the jets (e.g. Bridle et al. [AJ 108, 766, 1994]; Hardcastle et al, in prep).
Further evidence for this being the case can then be taken from VLBI measurements and superluminality; the jet seen on parsec-scales is always similar in sidedness and direction to the kiloparsec-scale jet, and the apparent superluminal velocities correlate with the degree of core-dominatedness. (But we are talking about very small samples in this last case.) Lobe-dominated quasars and, even more so, radio galaxies are difficult targets for VLBI because their cores are not all that bright. However, the nearby classical double Cygnus A has been observed with VLBI and shows _sub_luminal motion, as predicted by the model. There are various systematic programmes going on to improve the sample sizes, and with the VLBA this should get easier.