Our key assumption is that the bases of the two jets are intrinsically identical, antiparallel, axisymmetric stationary flows, and that the apparent differences between them result entirely from relativistic aberration.
This is an approximation in two important respects: we ignore small-scale
structure in the jets, and we assume that any intrinsic or environmental
asymmetries (clearly dominant on the largest scales in many objects) are
small compared with relativistic effects close to the nucleus. For any
individual source, it is difficult to be sure that this is the case,
particularly if the asymmetry persists on all scales. There are also a few
sources (e.g. 0755+379; Bondi et al. 2000 ) in which the counter-jet
appears much wider than the main jet, an effect which cannot be produced
by relativistic beaming. We argue that these cases are rare (Laing et al.1999)
and easy to recognize. If the jet/counter-jet ratio decreases with
distance from the nucleus, approaching unity on large scales, then
relativistic effects probably dominate, as most plausible intrinsic or
environmental mechanisms would generate asymmetries which stay roughly
constant or even increase with distance. A statistical study of a
complete sample of FRI sources with jets selected from the B2 sample
(Laing et al.1999) suggests that the median asymptotic jet/counter-jet ratio on
large scales 
1.1 for the weaker sources, so the assumption of
intrinsic symmetry for the jets is generally reasonable. We ensure that
this assumption is self-consistent by choosing an object whose jets are
straight, with similar outer isophotes on both sides of the nucleus.
We also require that the jets are bright, allowing imaging with high signal-to-noise ratio in total intensity and linear polarization, and that any effects of Faraday rotation can be accurately corrected. This led us to the choice of 3C31 as the first source to model.