It is generally accepted that jets in FRI radio galaxies decelerate by picking up matter, but it is by no means clear whether the principal source of additional material is mass loss from stars (Phinney 1983 , ; Komissarov 1994; Bowman, Leahy & Komissarov 1996) or entrainment across the jet boundary ( Baan 1980; Bicknell 1984; Bicknell 1986; De Young 1996): both are expected to be important. Our models require significant transverse velocity gradients, in the sense that the edge of the jet is travelling about 30% more slowly than the centre. These gradients are prima facie evidence for interaction between the flow and the external medium. There is no reason why mass input from stars should generate such gradients (Bowman, Leahy & Komissarov, 1996), although a pre-existing gradient might be preserved as a jet becomes mass-loaded. The form of the transverse velocity profile in our best-fitting models varies surprisingly little as the jet decelerates, but the error analysis of Section 4.2 shows that the situation might be more complicated: a top-hat velocity profile at the inner boundary is consistent with the data, so the profile could still evolve significantly along the jet. The presence of large quantities of very slow material at the edges of the flaring and outer regions is firmly excluded, however (Table 7).
A second piece of evidence favouring deceleration by interaction with the external medium is the complex field structure in the flaring region, where we were forced to introduce a significant radial component, increasing towards the edge of the jet, in order to explain the low degree of linear polarization. This radial field component would not be expected from simple passive evolution of a mixture of longitudinal and toroidal field in the smooth velocity field we assume. The most natural way to generate such a radial field component is for the flow to have a disordered, turbulent character towards the jet edges such as might result from large-scale eddies. This is precisely the situation expected at the edge of the jet in the initial ``ingestion'' phase of the entrainment process (De Young 1996). The velocity field is then likely to have significant small-scale structure which is not included in our model, but our estimates of average bulk flow speed are unlikely to be seriously affected. Even if there is no dissipation or dynamo action in such a turbulent flow, there will be significant amplification of the magnetic field by shear, so the simplest adiabatic models, which assume laminar flow (Section 5.4), will be inappropriate.
Another way to distinguish stellar mass loading from entrainment across the jet boundary is to ask whether stellar processes can provide the mass input rate required to produce the observed deceleration. It is clear from the work of Komissarov (1994) and Bowman et al. (1996) that a jet which is decelerated purely by stellar mass loading will tend to reaccelerate on large scales, where the stellar density becomes low but the outward pressure gradient and buoyancy force are still appreciable. Our models require continuous deceleration in the outer region, favouring boundary-layer entrainment as the dominant mechanism there. We address this question via a conservation-law analysis in Laing & Bridle (2002), where we conclude that entrainment dominates after the beginning of the flaring region.
Little is known about the properties of turbulent relativistic shear layers, or of the viscosity mechanisms likely to predominate in magnetized relativistic jets. We cannot therefore relate the deduced velocity profiles to the internal physics of the jets. We note however that Baan (1980) computed steady-state models for viscous jets in constant-pressure atmospheres and estimated both the transverse velocity profiles and appearance of the jets (on the assumption that the emissivity is directly proportional to the viscous dissipation) for several forms of the viscosity. Baan's models generally predicted extended low-velocity wings that do not match our derived profiles. He did however discuss circumstances under which flat-topped velocity profiles such as those inferred here might arise, including that of an electron-positron jet.