3.5. Some observational tests
None of the current models aiming to account for superlight expansion seems entirely convincing, and the fact that few can yet be decisively excluded is solely a consequence of the limited data available. We list here some questions which, if answered, would help to narrow down the field.
3.5.1. Does the position angle and/or the expansion speed vary during a single outburst? Models involving relativistic bulk motions can produce either accelerating or decelerating expansions. Indeed for those sources where the component separation is observed to expand by a factor ~ 2, it would be surprising if some change in the observed velocity were not seen. Blast wave models are generally decelerating, whereas free expansion of relativistic plasma (which is perhaps unlikely because adiabatic cooling would cause a rapid decline in flux density at frequencies where the source is optically thin) may accelerate. The expansion rates for "screen" models are determined mainly by the shape and orientation of the screen, but if the outburst or screen are not axisymmetric, the position angle may change. One interesting possibility is that the velocity may be accurately constant during an outburst. In this case the expansion would probably have to be attributed to one or more coherent objects moving inertially, although under certain conditions blast wave models can exhibit constant velocity phases. The present data seem consistent with uniform expansion during an outburst.
3.5.2. Are the position angle and expansion speed "remembered" from one outburst to the next? Successive outbursts should display the same alignment if they represent expansion along a preferred direction (e.g., the rotation axis of a central massive object or rotating gas cloud). A reproducible expansion speed would be a strong argument in favour of a screen model, but the absence of such would not necessarily rule it out because the signal velocity could vary between outbursts, which would alter the kinematics. In fact the two outbursts in 3C 120 seem to have had different speeds.
3.5.3. What are the absolute positions of the components and which ones are moving fastest? Clearly the development of phase-stable VLBI would require a considerable technical advance. This could, however, tell us which parts of the source are moving if the nucleus can be resolved into two or more components, one of which is unchanging. Alternatively, for those sources for which there is a sufficiently nearby unresolved radio source (e.g., 3C 345 / NRA 0512), one might obtain some absolute positional information.
3.5.4. How do the components of expanding doubles evolve? If the emission is due to synchrotron radiation, the component size must always be large enough to bring the brightness temperature below the plausible upper limit set by the inverse Compton restriction. It would be important to know how the source size or surface brightness distribution depends on wavelength and also whether or not a particular component faded out earlier. Some models (particularly those involving propagation effects) make specific predictions about wavelength dependence or the distribution of surface brightness. In simple screen models, the faster component would have lower brightness, whereas in models where the Doppler effect is more important, the reverse would be true. In sources showing successive outbursts, higher sensitivity observations would allow an old and fading component to be followed even when superposed on a young and bright source. This could considerably improve the interpretation of the kinematics. Some "phase-effect" models have a long recovery time and cannot readily account for repeated outbursts.
3.5.5. What are the polarisation properties of the individual components? If the emitting electrons are channelled along an effectively rigid magnetic field, then the observed radiation should be strongly polarised normal to the source axis it if is synchrotron radiation, and parallel if it is curvature radiation. If the field direction makes a small angle to the line of sight then perhaps a substantial circular polarisation could be observed (unless electrons and positrons are created in equal quantities). Induced Compton and plasmon scattering models also predict characteristic polarisation patterns related to the source geometry.
3.5.6. Can we identify the time when a given outburst was initiated? Brightness temperature limitations generally preclude the possibility of seeing the start of an outburst at radio wave-lengths. However, an unusual change in the optical or X-ray output may serve to identify the time origin (e.g., in 3C 273 which is particularly important in the case of screen models as this in principle allows a unique inversion of the geometry of the screen if the signal velocity is taken to be c.
3.5.7. Is the position angle of compact double structure aligned with extended radio structure? So far the evidence for this is mostly positive (although there are some known exceptions): this favours models in which the extended sources are being fuelled continuously by their central components. It is thus tempting to suggest that both sources share a common axis defined by the angular momentum of the central regions. This would imply that the directionality (if not complete collimation) was imprinted on a scale 1 pc. As already mentioned, accretion onto a black hole - whether disk-like or quasi-spherical - could establish a "twin exhaust" flow pattern even on a scale 100 Schwarzschild radii. 3C 273 is the only extended source known to display superliminal expansion, and it is very important to discover more examples where extended structure is also present. One can be fairly sure that the axis of a typical double sources does not make a particularly small angle with line of sight. So if rapid expansions with v0 >> c are found to be fairly common, it would rule out some models (e.g., "ballistic" models, which yield superluminous effects only when the motion is directed almost along the line of sight). Cygnus A is particularly important in this context, because there is fairly convincing evidence that the overall source axis lies almost in the plane of the sky.