Hubble Space Telescope A-band images of a sample of 10 radio galaxies (RG), 10
radio-loud quasars (RLQ) and 13 radio-quiet quasars (RQQ) at redshift z ≃ 0.2 are
presented. The sources comprising the radio-loud sub-samples have been selected to
be indistinguishable in the P₅Gₕ₂ ~ z plane. The two quasar sub-samples have been
similarly selected to have indistinguishable distributions in the V — z plane. All three
sub-samples have been analysed with two-dimensional modelling software, designed to
accurately determine the host-galaxy parameters.
The modelling results show all of the RG, and all of the quasars with Mᵥ < -23,
to have bulge-dominated host galaxies, excluding the hypothesis that host morphology
is responsible for the radio-loudness dichotomy. Furthermore, the host galaxies are
found to be essentially identical to a de Vaucouleurs (ß = 0.25) law, with 30/33 objects
having ß parameters in the range 0.2 < ß < 0.3. The hosts of all of the objects studied
are found to be extremely luminous (L ≥ L*), with 25/33 objects having L ≥ 2L*, the
hosts of the RQQs are found to be typically 0.4 magnitudes fainter than their radio-loud
counterparts. The host galaxies of all three sub-samples are found to be larger than
average ellipticals (< rₑ >= 10.5 kpc), with a trend found for the RQQ hosts to be
some 20% smaller than those of the RLQs. For the first time it is demonstrated that
the hosts of quasars, as well as RG, display a Kormendy relation basically identical to
that of inactive ellipticals.
In combination with the previous K-band modelling of this sample (Taylor et
al. 1996) the first set of reliable optical-infrared colours for a significant sample of
AGN host galaxies has been determined. The R - K colours of the hosts in all three
sub-samples are shown to be consistent with each other, with the close agreement of the
median colour of R - K=2.5 with that expected from a ~ 14 Gyr-old stellar population
implying that the host galaxies of both RGs and quasars are consistent with having
formed at high redshift (z ≥ 3).
The nuclear colours of the RGs are shown to be perfectly consistent with those of
dust reddened RLQ nuclei. The combined luminosity, scalelength, Kormendy relation
and R - K colour results for the RG and RLQ sub-samples clearly demonstrate that
they are drawn from the same parent population, in strong support of the orientationbased
unification scheme of Barthel (1989). The corresponding results for the full 33
object sample show that, in terms of their global structural parameters, the hosts of
powerful AGN are no different to normal massive ellipticals.
The host galaxy-black-hole mass correlations of Magorrian et al. (1998) and Franceschini
et al. (1998) are combined with the AGN modelling results to investigate the
quasar radio-loudness dichotomy. All methods of black-hole mass estimation lead to
the conclusion that RLQs harbour black-holes of mass ≥ 10¹⁰M⨀. The RQQ blackhole
masses estimated via the Magorrian relation are typically smaller than those of the
RLQs (~ 5 x 10⁹M⨀), but show sufficient overlap to require the influence of another
physical parameter, possibly black-hole spin, to explain the radio-loudness dichotomy.
In contrast, the use of the Franceschini black-hole estimator predicts RQQ black-hole
masses a factor of two smaller than those of the RLQs, suggesting that black-hole mass
alone may be the crucial factor.
The application of the modelling technique to an existing HST survey of 3CR radio
galaxies at z ~ 1 shows them to have a scalelength distribution and Kormendy relation
indistinguishable from their z ~ 0.2 counterparts. The implied luminosity evolution
between z ~ 0.2 and z ~ 0.8 is shown to be consistent with that expected from passive
evolution alone in a low density Universe, with a high galaxy formation redshift (z ≥ 3).
The lack of evidence for significant merger activity in the redshift range 0.2 < z < 0.8
leads to the conclusion that the traditional passive evolution interpretation of the radio
galaxy K - z relation remains acceptable.