1: \documentclass{aastex}

2: %\documentclass[12pt,preprint]{aastex}

3: \usepackage{emulateapj5,psfig}

4: %\usepackage{emulateapj5,natbib,psfig}

5: \usepackage{natbib,psfig}

6: \usepackage{graphicx}

7: %% If you want to create your own macros, you can do so

8: %% using \newcommand. Your macros should appear before

9: %% the \begin{document} command.

10: %%

11: %% If you are submitting to a journal that translates manuscripts

12: %% into SGML, you need to follow certain guidelines when preparing

13: %% your macros. See the AASTeX v5.0 Author Guide

14: %% for information.

15: \renewcommand{\floatpagefraction}{0.9}

16: \renewcommand{\textfraction}{0.00}

17: \newcommand{\vdag}{(v)^\dagger}

18: \newcommand{\myemail}{kmb@astro.ox.ac.uk}

19: \newcommand{\gtsim}{\mbox

20: {{\raisebox{-0.4ex}{$\stackrel{>}{{\scriptstyle\sim}}$}}}}

21: \newcommand{\ltsim}{\mbox

22: {{\raisebox{-0.4ex}{$\stackrel{<}{{\scriptstyle\sim}}$}}}}

23: \newcommand{\mc}{\multicolumn}

24: %% You can insert a short comment on the title page using the command below.

25: \slugcomment{Accepted by ApJ Letters}

26: 

27: \shorttitle{A relativistic jet in a radio-quiet quasar}

28: \shortauthors{Blundell, Beasley \& Bicknell}

29: 

30: \begin{document}

31: 

32: \title{A relativistic jet in the radio-quiet quasar PG\,1407+263}

33: 

34: \author{Katherine M.\ Blundell\altaffilmark{1}, Anthony J.\

35: Beasley\altaffilmark{2} and Geoffrey V. Bicknell\altaffilmark{3}}

36: 

37: \altaffiltext{1}{University of Oxford, Astrophysics, Keble Road, Oxford,

38: OX1 3RH, UK}

39: \altaffiltext{2}{Owens Valley Radio Observatory, California Institute

40:    of Technology, PO Box 968, Big  Pine, CA 93513}

41: \altaffiltext{3}{Research School of Astronomy \& Astrophysics,

42: Australian National University, Canberra, Australia }

43: 

44: 

45: \begin{abstract}

46: We present the results of a multi-epoch radio monitoring campaign

47: measuring the milliarcsecond structure of the jet in the radio-quiet

48: quasar PG\,1407+263.  This is the highest-sensitivity,

49: highest-resolution multi-year study of a distant active galaxy.  The 

50: observations are naturally explained in terms of a beamed relativistic

51: jet, some of whose fluctuations in flux density can be ascribed to

52: interaction with the narrow-line region of the quasar. The optical

53: properties of PG\,1407+263, in particular the low equivalent widths of

54: the emission lines, may be related to the fact that we are viewing

55: this quasar almost pole-on, giving us a direct view into its

56: broad-line region.

57: \end{abstract}

58: 

59: \keywords{galaxies: jets --- quasars: individual (PG1407+263) ---  radio

60: continuum: galaxies}  

61: 

62: \section{Introduction}

63: \label{sec:intro}

64: 

65: In recent years, the weak radio core emission associated with the

66: class of so-called `radio-quiet' quasars \citep[for

67: example,][]{Kel89,Mil90,Mil93} has been shown by a succession of Very

68: Long Baseline Interferometry (VLBI) imaging experiments usually to be

69: compact on milliarcsecond (mas) scales \citep{Blu96,Fal96,Blu98}.  The

70: typical physical size of \ltsim\ a few cubic parsecs for these radio

71: luminosities and corresponding brightness temperatures ($> 10^6$\,K)

72: strongly argues against star-formation as the origin of this radio

73: emission, despite the striking continuity of the correlation of

74: far-infra-red and radio emission from star-forming galaxies to

75: radio-quiet quasars \citep{Sop91a,Sop91b}.  Instead, a mechanism

76: resembling that which powers the core emission in the more

77: radio-luminous `radio-loud' quasars is suggested.

78: 

79: One quasar in particular, PG\,1407+263, identified in a mas imaging

80: survey of radio-quiet quasars \citep{Blu98} clearly has multiple

81: components and seemed an important target to monitor subsequently with

82: VLBI techniques.  It was first discovered as a quasar in the

83: Palomar-Green survey \citep{Sch83} with $z = 0.94 \pm 0.02$

84: \citep{McD95}.  With a Hubble constant of $70 \> \rm km\,s^{-1}

85: Mpc^{-1}$, and $(\Omega_{\mathrm M},\Omega_{\Lambda}) = (0.3,0.7)$,

86: the luminosity distance $\approx 6.2 \> \rm Gpc$ and 1\,mas $\approx$

87: 7.9\,pc.

88: 

89: \section{Measurements}

90: \label{sec:meas}

91: We observed PG\,1407+263 at 8.4\,GHz using the Very Long Baseline

92: Array (VLBA) with the Very Large Array (VLA) in phased array mode on

93: 1996 June 15 (VLBA only), 1997 April 27, 1998 January 31, 1999 March

94: 20 and 1999 July 20.  We usually recorded a 64-MHz bandwidth (a 32-MHz

95: bandwidth was used for epochs 4 \& 5) with 2--bit sampling and dual

96: circular polarization.  The observations were phase-referenced

97: \citep{Bea95}, i.e.\ regular observations of a bright extragalactic

98: source close on the sky to the target were made to derive

99: residual antenna-based and atmospheric phase corrections, which were

100: interpolated to calibrate the target data and provide accurate

101: astrometric referencing.  The calibrator source used was OQ\,208,

102: separated from PG\,1407+263 by 2.2$^{\circ}$.  We typically used a

103: switching cycle of two minutes on the calibrator and three minutes on

104: the target.  The total on-source time for OQ\,208 was 0.5\,hrs (epoch

105: 1) and 2.8\,hrs (epochs 2, 3, 4 and 5).  Throughout the three-year

106: monitoring, a pair of \ltsim\ mJy components were persistently detected

107: (Figure\,1), albeit with changes in structure and flux density

108: (Figure\,2).

109: 

110: 

111: \section{Light curves and structural changes}

112: \label{sec:lightcurves}

113: One striking change seen between successive epochs is that the

114: structure of each component appears to vary: this is particularly

115: marked in the final three epochs (Figure\,1).  Real changes in shape

116: represent an intriguing aspect of the parsec-scale evolution of the

117: source: since 1~mas corresponds to 7.9~pc, real expansion between

118: epochs of one tenth of a synthesized beam would correspond to apparent

119: motion of a few times the speed of light.  The components appear both

120: resolved and unresolved at different epochs.  Component B appears to

121: be significantly resolved (with a high total flux density) in earlier

122: epochs, while both components appear essentially unresolved (with low

123: flux densities) in the final epoch.  Unmodeled residual spatial

124: gradients in tropospheric (and ionospheric) delays lead to

125: interferometer phase errors which are cancelled towards the bright

126: calibrator source OQ\,208, but remain in the target data.  Thus, apart

127: from the above general assertions above about component B, it is

128: difficult to draw solid conclusions from the structural variations:

129: simple centroid fitting to the components, while technically possible,

130: can be ambiguous in interpretation.

131: 

132: The rms variation of the position of component A is $\approx

133: 300$\,$\mu$-arcsec.  Typical estimates of the effects of tropospheric

134: delay errors on phase-referenced VLBI data with a calibrator 2.2\,deg

135: away are 100--200\,$\mu$-arcsec. In addition, the typical

136: signal-to-noise (S/N) of the detections is 5--10 to 1; this would

137: introduce a positional variation roughly equal to the synthesized beam

138: divided by the S/N -- i.e.\ 100 to 250\,$\mu$-arcsec.  Therefore, the

139: position wandering largely results from a combination of unmodeled

140: tropospheric delay errors and S/N considerations.  Given the weak

141: (\ltsim\ mJy) flux density levels in these images, it is not possible

142: to correct for these residual errors.  During the course of these

143: observations the relative position shift of components A and B

144: appeared at times to be systematic, possibly indicating superluminal

145: motion \citep{Blu98b}. Analysis of the existing data set does not

146: allow a definitive determination of superluminal motion in this

147: radio-quiet quasar, although we in any case infer it to have a highly

148: relativistic jet speed (Doppler factor $\gtsim 10$) in the analysis

149: below.  The rapid variations in component B's flux density are

150: unlikely to result from imaging coherence loss, which would affect

151: both components; A's relatively constant peak flux density implies

152: that B's flux density variations are intrinsic.  The dramatic changes

153: in B's flux density occur on timescales of 100--300 days, reminiscent

154: of those seen in the cores of radio-loud quasars \citep{Hou99}.

155: 

156: %\begin{figure}

157: %\epsscale{0.45}

158: %\plotone{f1.eps}

159: \psfig{figure=f1.eps,width=7cm,angle=0}

160: \figcaption{\label{fig:contours} VLBA images of PG\,1407+263 restored

161: with the same synthesized beam, 1.8 $\times$ 0.9\,mas with position

162: angle 170\,deg East of North.  The contours in each case are $\pm 0.3$,

163: 0.42, 0.6, 0.84, 1.2, 1.68, 2.4\,mJy/beam.  }

164: %\end{figure}

165: 

166: 

167: \section{Recent radio-loudness of PG\,1407+263}

168: \label{sec:radioloud}

169: 

170: In order to establish whether PG\,1407+263 is truly a radio-quiet

171: quasar, we made deep VLA images to search for evidence of extended

172: radio structure or hotspots.  These images revealed only independent

173: sources including an apparent double 53\,arcsec east of the quasar.

174: \citet{Kel89} commented that this double structure might be related to

175: the output of the quasar, since it lacks any optical identification on

176: the Palomar Observatory Sky Survey.  We made a more sensitive test by

177: observing the field in the near infra-red $K$-band on the UK Infra-Red

178: Telescope (UKIRT).  Figure\,3 indicates each part of the `double'

179: radio structure is identified with a galaxy and is thus unrelated to

180: the quasar.  Therefore, we do not believe that PG\,1407+263 has been

181: radio-loud for the last $10^{6 - 7}$\,years, where this timescale is

182: taken from the typical radiative lifetimes for synchrotron-emitting

183: particles in the lobes of radio-loud quasars \citep{Blu00}.

184: 

185: %\begin{figure}

186: %\epsscale{0.9}

187: %\plotone{f2.eps}

188: \psfig{figure=f2.eps,width=8cm,angle=0} 

189: \figcaption{\label{fig:fluxes}

190: Peak flux densities at 8.4\,GHz of components A \& B at each epoch

191: plotted against time.  The dotted line corresponds to the eastern

192: component (A) and the solid line corresponds to the western component

193: (B).  }

194: %\end{figure}

195: 

196: 

197: \section{Physical interpretation}

198: \label{s:interp}

199: 

200: It is feasible that the observed structural changes

201: in component B 

202: are related to at least one unresolved

203: sub-component. The much closer source 3C\,120 \citep{gomez00a} is a

204: good example of how jet knot components can appear and

205: fade.  G\'omez et al have interpreted this in terms of interaction

206: between the jet and the inhomogeneous ambient medium. Observed at

207: lower resolution, the jet in 3C~120 would present apparently

208: structurally changing partially resolved components. If we are indeed

209: observing such a phenomenon, then the offset between the two local

210: surface brightness maxima in component~B may give us some idea of the

211: scale of the jet diameter.  The perpendicular offset between the 

212: centroid of the two B

213: subcomponents and the line connecting the two main components, A and B

214: is about $ 0.7 \> \rm mas \approx 5.5 \> \rm pc$. Therefore, in the

215: following, we adopt a fiducial scale for the jet diameter, $D$ of $5

216: \> \rm pc$.  The causal limit on the diameter deduced from the

217: variability time scale, $\Delta t$, and corresponding to a Doppler

218: factor $\delta$ is:

219: $

220: D < c (1+z)^{-1} \, \Delta t \, \delta \approx 0.4

221: \left( \frac {\Delta t}{100 \> \rm days} \right) \,

222: \left( \frac {\delta}{10} \right) \> \rm pc.

223: \label{e:d_limit}

224: $ The variation from epochs 1 to 2 entails a factor $\sim 2$ decrease

225: in flux density over a timescale $\approx 400\> \rm days$. This

226: implies an upper limit on the diameter $\approx 1.6 \> \rm pc$ for a

227: Doppler factor of 10. This is intriguingly less than, but of the same

228: order as, the approximate scale size estimated from the VLBA images.

229: Of course, the discrepancy decreases with increasing Doppler factor.

230: [If the jet is pole-on so that $\delta \approx 2 \Gamma$ (where

231: $\Gamma$ is the bulk Lorentz factor), given the existence of

232: apparent proper motions $\sim 10 \, c$ in some quasars, then $\delta

233: \sim 20$ is not out of the question.]  Clearly, the reconciliation of

234: these two scales is in the direction of larger Doppler factors, not

235: smaller.  This is an indication that this source is highly beamed.

236: 

237: A second, more persuasive, indication for beaming comes from

238: estimating the intrinsic source power. The emitted power per steradian is:

239: $\int j^\prime_\nu dV^\prime = \, \delta^{-(3+\alpha)} 

240: \,

241: (1+z)^{-(1-\alpha)} \>  F_\nu \, D_L^2

242: \quad \rm W \> Hz^{-1} \> sr^{-1}$,

243: where the integral is that of the rest-frame emissivity,

244: $j^\prime_\nu$,  at the {\em observed} frequency $\nu$ over the proper 

245: volume ($V^\prime$) of each blob and $F_{\nu}$ is the flux density at 

246: $\nu$.

247: 

248: %\begin{figure}

249: %\epsscale{0.6}

250: %\plotone{f3.eps}

251: \psfig{figure=f3.eps,width=8cm,angle=0}

252: \figcaption{\label{fig:widefield} The top image is a seven

253: arcmin-square VLA image at 8.4\,GHz in D-array.  The lower image shows

254: a grey-scale of near-IR $K$-band on which are overlaid contours at

255: 8.4\,GHz from the VLA A-array, the lowest of which is 0.04\,mJy/beam.

256: The two radio sources are clearly identified with resolved

257: 18.5-magnitude galaxies, indicating they are independent radio

258: structures from the quasar.}

259: %\end{figure}

260: 

261: For unity Doppler factor, the total emitted power, at 8.4~GHz, of both

262: components at epoch~1 would be $1.1 \times 10^{24} \> \rm W \> Hz^{-1}

263: \> sr^{-1}$. This is comparable to the core powers of the most

264: powerful galaxies in the B2 sample \citep{deruiter90a}.  If

265: PG\,1407+263 were similar to the B2 sources then, using the

266: statistical relationship between core and extended powers derived by

267: \citet{deruiter90a}, its corresponding 1.4~GHz extended power would be

268: of order $10^{28} {\rm W \> Hz^{-1}}$ and the 8.4 GHz flux density

269: would be of order one Jansky.  However, since the flux density of any

270: extended emission from PG\,1407+263 is less than 0.1\,mJy, this is

271: another indication that the core is beamed.  The Doppler factor has to

272: be increased to $\sim 10$ in order to reduce the expected extended

273: flux density to the observed upper limit.

274: The additional parameters describing each knot are the spectral index

275: $\alpha$, the electron energy index, $a=2\alpha+1$, the volume,

276: $V^\prime$, of the knot in its rest frame, the Doppler factor,

277: $\delta$ and the ratio, $c_E$, of energy in relativistic electrons and

278: positrons to other particles. The rest-frame magnetic field that

279: minimizes the total energy density of a spherical blob in the rest

280: frame is: 

281: {\footnotesize

282: $B_{\rm min}^\prime = e^{-1} m_e \, \delta^{-1}(1+z)^{-\left(

283:    \frac{1-\alpha}{\alpha+3} \right)} \times$ \newline

284: 

285: $\left[ (a +1) \, C_2^{-1}(a) \, (1+c_E) \,  m_e^{-1} c

286:   \, f(a,\gamma_1, \gamma_2) \,

287:  F_\nu \nu^\alpha D_L^2{V^\prime}^{-1} 

288: \right]^{\frac{2}{a+5}}  \> {\rm T}.$}

289: Here $C_2(a)$ is a coefficient involving $\Gamma$-functions that

290: appears in the expression for the angle-averaged synchrotron

291: emissivity; $e$, $m_e$ and $c$ are the electronic charge, electronic

292: mass and speed of light respectively, and $f(a,\gamma_1, \gamma_2) =

293: (a-2)^{-1} \gamma_1^{2-a} (1-(\gamma_2/\gamma_1)^{2-a})$\footnote{See

294: http://www.mso.anu.edu.au/$\sim$geoff/HEA/HEA.html}.  We have

295: estimated the minimum-energy magnetic field and corresponding particle

296: energy densities, from the flux density of component B at epoch~1 and

297: for proper jet diameters of 0.375, 0.75 and 1.5\,mas as functions of

298: Doppler factor.  Figure\,\ref{f:energy} shows the associated jet

299: energy flux, $F_E \approx \left[ 4p \, \left( 1 + \frac

300: {\Gamma-1}{\Gamma} \frac {\rho c^2}{4p} \right) + \left( \frac

301: {{B^\prime}^2}{\mu_0}\right) \right] \, \Gamma^2 c \beta A_{\rm jet}$,

302: where $A_{\rm jet}$ is the cross-sectional area of the jet, $\rho$ is

303: the rest-mass density, and where the Poynting flux of a perpendicular

304: magnetic field has been included. We assume that the jet is pole-on

305: and purely relativistic, so that $\rho c^2/4p \ll 1$ and $c_E=0$.  The

306: almost constant value of the jet energy flux for Doppler factors $\ge

307: 2$ is the result of the $\delta^{-2}$-dependence of the magnetic and

308: particle energy densities canceling the $\Gamma^2$-dependence of the

309: energy flux.  If we assume that 0.75\,mas is a reasonable upper limit

310: to the jet diameter, then the jet is no more powerful than about $7

311: \times 10^{36} \> \rm W$.  For jet diameters less than the fiducial

312: value the jet is less powerful, albeit still relativistic.

313: We note that this is not the first radio-weak object for which

314: relativistic velocities have been inferred. \citet{brunthaler00}

315: inferred superluminal motion from their observations of the Seyfert

316: galaxy IIIZw2, directly implying relativistic motion.

317: 

318: %\begin{figure}

319: %\epsscale{0.9}

320: %\plotone{f4.eps}

321: \psfig{figure=f4.eps,width=8cm,angle=0} 

322: \figcaption{\label{f:energy}

323: Jet energy flux as a function of Doppler factor for three assumed

324: values of the jet diameter: the fiducial jet diameter and a factor of

325: two above and below this. }

326: %\end{figure}

327: 

328: \section{Discussion}

329: The low extended emission relative to the core in PG\,1407+263 is a

330: good argument that this jet is relativistically beamed. If our

331: inference of partially resolved structure in the jet is correct then

332: the case for relativistic beaming is reinforced by the variability

333: time scale ($\sim 400 \> \rm days$ for epochs 1-2).  If the jet

334: diameter is indeed $\sim 5 \> \rm pc$ then, given the typical

335: expansion rates of jets, the components are of the order of 100\,pc

336: from the core, placing them in the narrow-line region. These two

337: arguments point to a Doppler factor of $\ga 10$.  As

338: Figure~\ref{f:energy} shows, the dependence of energy flux on Doppler

339: factor and jet radius indicates a kinetic power no greater than about

340: $10^{37} \> \rm W$. Using theoretical \citep{bicknell98a} and

341: observational \citep{willott99a} relations between jet and radio

342: power, such an energy flux would be typical of a radio galaxy jet no

343: more than an order of magnitude above the FRI/FRII break

344: \citep{Fan74}.  The interest that attends this galaxy is the existence

345: of an FRI radio power associated with a bright quasar so it is

346: unsurprising that this quasar is classified as radio quiet

347: \citep{Blu01}.  It appears the only reason that we detect any

348: radio flux is that it is one of the few PG quasars that is beamed

349: towards us.

350: 

351: As well as its variation between epochs 1 and 2, component B also

352: exhibits a significant rise in flux density $F_\nu$ between epochs 3

353: and 4 then a sharp drop ($\Delta t = 122$ days) between epochs 4 and 5

354: with $\Delta F_\nu/F_\nu \sim 1$. This variability indicates a

355: diameter $\le 0.5 (\delta /10) \> \rm pc $ for component B around this

356: epoch -- much less than the indicative jet diameter $\sim 5 \> \rm pc$

357: inferred above. There are two possibilities: (1) the jet may not be as

358: wide as 5\,pc or (2) the variation is due to a small section of the

359: jet being affected by its interaction with clouds in the narrow-line

360: region of the quasar.  This second case is similar to what has been

361: observed in 3C\,120 by \citet{gomez00a}.  Our observations therefore

362: suggest a relativistic jet, of only moderate power, propagating

363: through the clumpy narrow-line region of a bright host quasar.

364: 

365: Does our consideration of the radio structure of PG\,1407+263 shed any

366: light on its curious optical properties e.g., the low equivalent width

367: of the emission lines \citep{McD95}?  If the Doppler factor of the jet

368: is as high as 10, then we are viewing PG\,1407+263 within a few

369: degrees of pole-on and looking directly into its broad-line region.

370: Thus, the emission line spectrum may be diluted by a direct line of

371: sight to the accretion disk continuum. An argument against this is

372: that the PG quasars show no evidence for orientation-dependent effects

373: and this is generally ascribed to the initial selection via

374: ultra-violet excess.  However, supporting evidence for PG\,1407+263

375: being close to pole-on is the large velocity dispersion of its

376: emission line gas $\approx 10,000\> \rm km \> s^{-1}$ \citep{McD95},

377: indicating that we may be looking directly into the inner part of the

378: broad-line region. 

379:  

380: \acknowledgments

381: 

382: KMB thanks the Royal Society for a University Research Fellowship.

383: AJB gratefully acknowledges support from NSF grant AST-0116558

384: (Combined Array for Research in Millimeter-wave Astronomy).  The VLBA

385: and VLA are facilities of the NRAO operated by AUI, under co-operative

386: agreement with the NSF.  We would like to thank the UKIRT service

387: programme, and Dr Paul Hirst especially, for help in obtaining the

388: UFTI data.

389: 

390: \begin{thebibliography}{}

391: 

392: \bibitem[Beasley and Conway(1995)]{Bea95}

393: Beasley A.J.\ and Conway J.E.\ 1995 in {\em Very Long Baseline

394: Interferometry and the VLBA}, Astr. Soc. Pac. Conf. Ser. Vol. 82,

395: Zensus, J.A., Diamond, P.J.\ and Napier, P.J.\ eds 328

396: 

397: \bibitem[Bicknell et al.(1998)]{bicknell98a}

398: Bicknell G. V.,Dopita M. A.,Tsvetanov  Z.I. \& Sutherland R.S., 1998,

399: \apj, 495, 680

400: 

401: \bibitem[Blundell \& Beasley(1998a)]{Blu98}

402: Blundell K.M.\ \& Beasley A.J., 1998a, MNRAS, 299, 165

403: 

404: \bibitem[Blundell \& Beasley(1998b)]{Blu98b} Blundell K.M.\ \&

405: Beasley A.J.\ 1998b, BAAS, 30, 1418

406: 

407: \bibitem[Blundell et al(1996)]{Blu96}

408: Blundell K.M., Beasley A.J., Lacy M.\ \& Garrington S.T., 1996, ApJ,

409: 468, {\sc l}91

410: 

411: \bibitem[Blundell \& Rawlings(2000)]{Blu00}

412: Blundell K.M.\ \& Rawlings S. 2000, \aj, 119, 1111

413: 

414: \bibitem[Blundell \& Rawlings(2001)]{Blu01}

415: Blundell K.M.\ \& Rawlings S.\ 2001, \apjl, 562, L5 

416: 

417: 

418: \bibitem[Brunthaler et al. (2000)]{brunthaler00} 

419: Brunthaler, A., et al,

420: \aap, 2000, 357, L45

421: 

422: \bibitem[Bridle and Perley(1984)]{Bri84}

423: Bridle A.H.\ \& Perley R.A. 1984, \araa, 22, 319

424: 

425: \bibitem[de Ruiter et al(1990)]{deruiter90a}

426: de Ruiter H., Parma P., Fanti C. \& Fanti R.,

427: 1990, \aap, 227, 351

428: 

429: \bibitem[Falcke, Patnaik \& Sherwood(1996)]{Fal96}

430: Falcke H., Patnaik A.R.\ \& Sherwood W., 1996, ApJ,  473, {\sc l}13

431: 

432: \bibitem[Fanaroff \& Riley(1974)]{Fan74}

433: Fanaroff, B.L.\ \&  Riley, J.M.\ 1974, MNRAS, 167, 31P

434: 

435: \bibitem[G\'omez et al(2000)]{gomez00a}

436: G\'omez J. L., Marscher A. P., Alberdi A., Jorstad S. G. \&

437: Garcia-Miro  C., Science, 289, 2317

438: 

439: \bibitem[Hough et al(1999)]{Hou99}

440: Hough D.H.\ et al., 1999, \apj, 511, 84

441: 

442: \bibitem[Kellermann et al(1989)]{Kel89}

443: Kellermann, K.I., Sramek, R., Schmidt, M., Shaffer, D.B. and

444: Green, R. 1989, \aj, 98, 1195

445: 

446: \bibitem[McDowell et al.(1995)]{McD95}

447: McDowell J.C., et al, 1995, \apj, 450, 585

448: 

449: \bibitem[Miller et al(1990)]{Mil90}

450: Miller L., Peacock, J.A. and Mead A.R.G. 1990, \mnras,

451: 244, 207

452: 

453: \bibitem[Miller et al(1993)]{Mil93}

454: Miller P., Rawlings S. and Saunders R. 1993, \mnras,

455: 263, 425

456: 

457: \bibitem[Schmidt \& Green(1983)]{Sch83}

458: Schmidt M.\ \& Green R.F., 1983, ApJ, 269, 352

459: 

460: \bibitem[Sopp and Alexander(1991a)]{Sop91a}

461: Sopp H.M. and Alexander P.\ 1991a, \mnras, 251, L14

462: 

463: \bibitem[Sopp and Alexander(1991b)]{Sop91b}

464: Sopp H.M. and Alexander P.\ 1991b, \mnras, 251, 112

465: 

466: \bibitem[Willott et al(1999)]{willott99a}

467: Willott C.J., Rawlings S., Blundell K.M.\ \& Lacy M.,

468: 1999, \mnras, 309, 1017

469: 

470: \end{thebibliography}

471: 

472: 

473: 

474: 

475: 

476: \end{document}