1: \documentclass{mn2e}
2: \usepackage{psfig}
3: \usepackage{epsf}
4: %\documentstyle[psfig,epsf]{mn}
5: %\documentstyle[epsf,amstex,referee]{mn}
6: %\onecolumn
7: %
8: \newcommand{\ltaraw}{$\; \buildrel < \over \sim \;$}
9: \newcommand{\lta}{\lower.5ex\hbox{\ltaraw}}
10: \newcommand{\gtaraw}{$\; \buildrel > \over \sim \;$}
11: \newcommand{\gta}{\lower.5ex\hbox{\gtaraw}}
12: % UNITS:
13: \newcommand{\km}{{\rm\,km}}
14: \newcommand{\kms}{{\rm\,km\,s^{-1}}}
15: \newcommand{\kpc}{{\rm\,kpc}}
16: \newcommand{\mpc}{{\rm\,Mpc}}
17: \newcommand{\Mpc}{{\rm\,Mpc}}
18: \newcommand{\msun}{{\rm\,M_\odot}}
19: \newcommand{\lsun}{{\rm\,L_\odot}}
20: \newcommand{\hubunits}{{\rm km\,s^{-1}\,Mpc^{-1}}}
21: \newcommand{\cm}{{\rm\,cm}}
22: \newcommand{\Gyr}{{\rm\,Gyr}}
23:
24: \newcommand{\ffffff}[1]{\mbox{$#1$}}
25: \newcommand{\scnd}{\mbox{\ffffff{''}\hskip-0.3em.}}
26: \newcommand{\scmd}{\mbox{\ffffff{''}}}
27: \newcommand{\mcnd}{\mbox{\ffffff{'}\hskip-0.3em.}}
28: \newcommand{\mcmd}{\mbox{\ffffff{'}}}
29:
30: \newcommand{\apm}{APM08279+5255}
31:
32: \def\Ger#1{\noindent{\bf[$\diamondsuit$ #1]}}
33: \def\Rod#1{\noindent{\bf[$\heartsuit$ #1]}}
34:
35: \loadboldmathitalic
36: \title [STIS spectra of \apm]
37: {Spatially resolved STIS spectra of the gravitationally
38: lensed BAL quasar \apm: the nature of component C
39: and evidence for microlensing}
40: \author[G. F. Lewis et al.]
41: {Geraint F. Lewis$^{1}$, Rodrigo A. Ibata$^{2}$, Sara L. Ellison$^{3}$,
42: Bastien Aracil$^{4}$, \newauthor
43: Patrick Petitjean$^{5}$, Max Pettini$^{6}$, Raghunathan Srianand$^{7}$ \\
44: $^{1}$
45: Anglo-Australian Observatory, P.O. Box 296, Epping, NSW 1710, Australia:
46: Email \tt{gfl@aaoepp.aao.gov.au}\\
47: $^{2}$
48: Observatoire de Strabourg, 11, rue de l'Universite, F-67000, Strasbourg,
49: France:
50: Email \tt{ibata@pleiades.u-strasbg.fr} \\
51: $^{3}$
52: European Southern Observatory, Casilla 19001, Santiago 19, Chile:
53: Email \tt{sellison@eso.org} \\
54: $^{4}$
55: Institut d'Astrophysique de Paris -- CNRS, 98bis Boulevard
56: Arago, F-75014 Paris, France:
57: Email \tt{aracil@iap.fr} \\
58: $^{5}$
59: Institut d'Astrophysique de Paris -- CNRS, 98bis Boulevard
60: Arago, F-75014 Paris, France:
61: Email \tt{petitjean@iap.fr} \\
62: $^{6}$
63: Institute of Astronomy, Madingley Rd, Cambridge, CB3 0HA, U.K.
64: Email \tt{pettini@ast.cam.ac.uk}\\
65: $^{7}$
66: IUCAA, Post Bag 4, Ganeshkhind, Pune 411 007, India:
67: Email \tt{anand@iucaa.ernet.in}
68: }
69: \date{\today}
70: \begin{document}
71: \maketitle
72: \begin{abstract}
73: While gravitationally lensed quasars are expected to display an odd
74: number of images, invariably systems are observed with an even number
75: of quasars. For this, lensing galaxies must have very small core
76: radii; this provides strong demagnification of one of the images.
77: High resolution imaging of the gravitationally lensed BAL quasar,
78: \apm, reveals three point-like images. As these images possess similar
79: colours, it has been suggested that each represents a lensed image.
80: Here, spatially resolved spectra of the individual components,
81: obtained with STIS on the HST, are presented, clearly revealing that
82: each is an image of the quasar. This confirms that \apm\ represents
83: the first example of an odd-image gravitationally lensed system. The
84: implications for the properties of the lensing galaxy are discussed.
85: It is also found that the individual images possess spectral
86: differences indicative of the influence of gravitational microlensing
87: in this system.
88: \end{abstract}
89: \begin{keywords}
90: gravitational lensing -- quasars: individual: \apm
91: \end{keywords}
92:
93: \section{Introduction}\label{introduction}
94: The $z=3.911$ broad absorption line (BAL) quasar \apm\ was identified
95: serendipitously within a survey of carbon stars in the Galactic halo
96: (Irwin et al. 1998). The optical emission is coincident with a
97: $\sim1$Jy IRAS source at 100$\mu$m and was also found to be a
98: significant submillimetre source, with a flux of 75mJy at 850$\mu$m
99: (Lewis et al. 1998); the inferred bolometric luminosity is
100: $\sim5\times10^{15}L_\odot$, making \apm\ one of the most luminous
101: sources known. The discovery images revealed that \apm\ is not a
102: single point-like source, rather it is extended (Irwin et al. 1998).
103: Adaptive optics with the CFHT clearly displayed \apm's compound
104: nature, revealing a pair of point-like images separated by 0.4 arcsec
105: (Ledoux et al. 1998). Further observations, using NICMOS on the Hubble
106: Space Telescope (Ibata et al. 1999) and the Keck telescope (Egami et
107: al. 2000), uncovered a fainter third image between the brighter two,
108: the colours of which suggest that it represents a third image of the
109: quasar.
110:
111: Paradoxically, this conclusion is somewhat problematic. Although one
112: of the fundamental predictions of gravitational lens theory is that
113: there should always be an odd number of lensed images (e.g. Burke
114: 1981), in practice all lensed QSOs known to date exhibit an even
115: number of images. Narasimha, Subramanian, \& Citre (1986) have shown
116: that this is in fact expected if the lensing galaxies have very small
117: core radii; the core `captures' one of the images and strongly
118: demagnifies it. The fact that APM~08279+5255 apparently defies this
119: trend is reflected in the mass models for the lens which require
120: unphysically large cores to explain the brightness of image C, if the
121: latter is indeed an image of the QSO (Ibata et al. 1999; Egami et
122: al. 2000; Munoz et al. 2001). This led Ibata et al. (1999) to also
123: propose an alternative model where component C is actually the lensing
124: galaxy responsible for the observed configuration; this model
125: possesses more `typical' galaxy parameters. Recent observations of
126: the nuclear CO(1-0) emission in \apm\ reveals a complex morphology
127: which suggests that the lens in this system is a highly flattened
128: system, such as an edge-on spiral galaxy (Lewis et al. 2002). With
129: this model, component C is an image of the quasar and the ternary
130: configuration of the quasar source lying in the vicinity of a `naked
131: cusp' (e.g. Bartelmann \& Loeb 1998).
132:
133: \begin{figure}
134: \centerline{ \psfig{figure=Figure1.ps,width=5.5cm,angle=270.0} }
135: \caption{Flux distribution along the spatial direction of the
136: slit. From left to right the peaks correspond to images A, C and
137: B. The three component Gaussian fit is displayed with a dotted line,
138: and the dashed line shows the sum of the three Gaussians.}
139: \label{Figure1}
140: \end{figure}
141:
142: The situation is complicated by the presence of two very strong MgII
143: systems at z=1.062 and 1.181. A detailed study of the corresponding
144: absorption spectrum shows that these systems are damped with a very
145: high HI column density (${\rm >10^{21} cm^{-2}}$, Petitjean et al.
146: 2000). The objects associated with these two systems could both
147: contribute to the lensing of the quasar.
148:
149: Given the extreme apparent properties of \apm, an accurate lens model
150: is essential for the determination of its true intrinsic
151: properties. This paper presents new spatially resolved spectra of the
152: various components in this lensing system, with the goal of
153: determining the nature of component C. Section~\ref{obsred} presents
154: a description of the observing and data reduction, reviewing the
155: nature of component C in Section~\ref{discuss}. This section also
156: discusses the implications of this study. Finally, the conclusions
157: are presented in Section~\ref{conclusions}.
158:
159: \section{Observations and Reduction}\label{obsred}
160:
161: \subsection{Observations}\label{obs}
162: The aim of the observing program was to obtain a spatially resolved,
163: high resolution spectrum of the $z=3.911$ BAL quasar, APM~08279+5255.
164: While being triply imaged, the image separations in this system are
165: small, ranging from 0.15~arcsecs to a maximum of 0.38~arcsecs.
166: Ground-based observations of this bright source have revealed a rich
167: absorption spectrum caused by both intervening material and the
168: complex QSO environment (Ellison et al. 1999a,b; Srianand \& Petitjean
169: 2000; Petitjean et al. 2000).
170:
171: The Space Telescope Imaging Spectrograph (STIS) on-board the Hubble
172: Space Telescope (HST) was employed in this endeavor. The principal
173: aim was to probe the numerous intervening systems on sub-kpc scales,
174: to investigate the structure of intervening galaxy halos and metal
175: line systems on scales of $\sim$ 0.2 -- 1.6 kpc h$^{-1}$, as well as
176: their kinematics and spatial extents. A further goal was to obtain
177: multiple sightlines through the complex BAL flow on parsec scales,
178: yielding information on ionization, kinematics and metal enrichment;
179: these will be the topics of forthcoming articles. Here we present an
180: initial investigation of the spectra in the individual lensed images,
181: with the goal of establishing the nature of image C.
182:
183: The STIS G750M grating was used in five different (primary) wavelength
184: settings so as to achieve complete coverage from $\sim 6000$ -- 8600
185: \AA\ (see Table 1). Acquisition was achieved using the brightest
186: component A. The 52 $\times$ 0.2 arcsec slit was then oriented along
187: the major axis between the two brightest components (A and B) and then
188: offset perpendicularly by 0.02 arcseconds to achieve improved centring
189: on the faint component C. The STIS spatial pixel scale is 0.051
190: arcsecs/pixel, with a typical dispersion of 0.56\AA/pixel. In
191: addition to the standard calibrations (including spectra of the He-Ar
192: arc lamp for wavelength calibration), extra flat field images were
193: obtained for the three reddest settings in order to correct for
194: significant fringing at these wavelengths. Multiple exposures were
195: obtained for each setting, subsequent exposures stepped along the slit
196: by 1 arcsecond.
197:
198: Due to a combination of narrow scheduling windows and backlog from
199: previous cycles, only a subset of the orbits has so far been
200: completed, see Table 1. While we defer a comprehensive analysis of the
201: absorption spectra until the full data set has been acquired, there is
202: sufficient information in the first wavelength setting, at
203: $\lambda_{\rm central} = 6252$\,\AA\ (first line of Table 1) to
204: confirm the nature of component C. The data considered in this paper
205: consist of five individual spectroscopic exposures; each exposure was
206: offset by 20 pixels (one arcsecond) relative to the previous one, so
207: as to maintain optimal spatial resolution and minimise the effects of
208: CCD and camera artifacts.
209:
210:
211: \begin{table}
212: \begin{center}
213: \caption{Summary of Observations}
214: \begin{tabular}{lrr} \hline \hline
215: Wavelength &Total &
216: Completed \\
217: Range (\AA) & Integration (s) &
218: Integration (s) \\ \hline
219: 5965 -- 6538 & 14 900 & 14 900\\
220: 6482 -- 7054 & 11 800 & 0\\
221: 6997 -- 7569 & 11 800 & 8 700\\
222: 7509 -- 8081 & 14 900 & 14 900\\
223: 8025 -- 8597 & 21 100 & 6 200 \\ \hline
224: \end{tabular}
225: \end{center}
226: \end{table}
227:
228: \subsection{Data Reduction}\label{red}
229: The extraction of the individual spectra is slightly complicated by
230: the fact that the spectra are not completely separated, especially
231: components A and C, even with the superb resolution afforded by STIS.
232: The extraction was performed in a straightforward, but non-standard
233: way, which we now detail. The adopted procedure is similar to that
234: undertaken for the complex gravitational lens Q2237+0305 by Lewis et
235: al. (1998).
236:
237: The first step we took was to trace the spectrum of of component A
238: (the brightest QSO component) with a straight line fit; this gives the
239: spatial offset between the spectral images to $\sim 0.02$~pixels.
240: After correcting for the slope of the spectrum on the image, we obtain
241: the distribution of flux in the spatial direction, integrated over the
242: wavelength range, $5965$\AA\ to $6534$\AA (Figure 1). To aid the
243: visual interpretation of this diagram, we have overlaid a three
244: component Gaussian fit. The three quasar components are clearly
245: distinguished in this single spectroscopic exposure; also evident is
246: the peculiar asymmetric profile (seen as an excess over the Gaussian
247: fit to the left side of components A and B).
248:
249: We constructed a model point-spread function using an archival
250: spectrum of the star 51~Peg (HST root name O6IH50050) taken in
251: approximately the same wavelength range, only a day prior to the
252: observations described here. By fitting this PSF to the data in
253: Figure~1, we are able to define suitable extraction bands for each one
254: of the components. These bands are listed in Table~2, with positions
255: relative to the centre of component A. Estimates of the
256: cross-contamination from the other quasar components and the fraction
257: of the flux missed are also given in the table (assuming the model PSF
258: derived from the 51~Peg observation). The worst case is clearly image
259: C, where the extracted spectrum has a total of $\sim 10$\%
260: contamination from components A and B, and is missing $\sim 30$\% of
261: the flux.
262:
263: \begin{table}
264: \begin{center}
265: \caption{Summary of Data Reduction}
266: \begin{tabular}{lrrrr} \hline \hline
267: Image & Lower & Upper &
268: cross- & flux \\
269: & band edge & band edge &
270: contamination & missed \\ \hline
271: A & $-0\scnd15$ & $0\scnd05$ & 2\% & 7\% \\
272: B & $0\scnd25$ & $0\scnd50$ & 1\% & 1\% \\
273: C & $0\scnd10$ & $0\scnd20$ & 10\% & 30\% \\ \hline
274: \end{tabular}
275: \end{center}
276: \end{table}
277:
278:
279: He-Ar arc-lamp spectra were also extracted from the arc-lamp spectral
280: image from identical regions as the object spectra. These arc-lamp
281: spectra were used to wavelength calibrate the observed QSO spectra
282: which were then rebinned onto the interval $\lambda=5965$\AA\ to
283: $\lambda=6534$\AA\ with linear wavelength steps. Finally, the five
284: spectral observations of the three QSO images were combined using a
285: median-combining algorithm. Figure~2 displays the three spectra
286: obtained in this manner.
287:
288:
289: \begin{figure}
290: \centerline{ \psfig{figure=Figure2.ps,height=9.0cm,angle=0.0} }
291: \caption{The spectra of the three QSO images A, B and C (from top to
292: bottom). While differences in the individual spectra are apparent,
293: the lower panel confirms that image C is of the quasar. As well as
294: the prominent emission line structure due to ${\rm Ly_\alpha/N~V}$,
295: strong MgII $\lambda\lambda 2796,2803$ absorption at z=1.18 due to a
296: foreground galaxy is labelled. One striking feature is the presence
297: of MgII absorption at z=1.21 ($\sim6190$\AA), also labelled, which,
298: while being strong in image A, is weaker in C and non-existent in
299: image B, suggesting a non-uniform covering factor across the images
300: (Petitjean et al. 2000).
301: \label{spectra}}
302: \label{Figure2}
303: \end{figure}
304:
305: \section{Discussion}\label{discuss}
306: \subsection{The nature of component C}\label{compc}
307: A cursory examination of the individual spectra of the three images
308: presented in Figure~\ref{spectra} clearly reveals that each is of the
309: quasar source, and, therefore, image C represents a third image of the
310: quasar and is not the lensing galaxy. This confirms earlier
311: interpretations of the available photometric data (Ibata et al. 1999;
312: Egami et al. 2000; Munoz et al. 2001).
313:
314: \subsection{Spectral differences}\label{diffs}
315: A more detailed examination of the three spectra in Figure 2, however,
316: also reveals that they are not identical. Even accounting for the
317: strong Mg~II doublet lines near 6100\,\AA\ (at $z_{\rm abs} = 1.181$),
318: probably due to the presence of the lensing galaxy, and for the
319: Ly$\alpha$ and N~V BAL features below $\sim 6040$\,\AA, it is evident
320: that the equivalent width of the Ly$\alpha$+N~V emission line blend
321: near 6050\,\AA\ is different in each of the three images. Relative to
322: the continuum, the emission lines are strongest in image C and weakest
323: in B.
324:
325: These differences are summarised in Table 3 (where we have corrected
326: for the missing flux in the spectrum extraction procedure). For the
327: continuum level we adopted the mean observed flux in the wavelength
328: interval 6250--6350\,\AA\ which, at $z_{\rm em} = 3.911$, corresponds
329: to rest wavelengths 1273--1293\,\AA; this is a region free of
330: prominent emission lines and provides a good measure of the QSO
331: continuum (Francis et al. 1991). The peak emission line flux was
332: measured between 6040\,\AA\ and 6060\,\AA\ (1230--1234\,\AA\ in the
333: rest frame of APM~08279+5255) after subtracting the underlying
334: continuum. It can be seen from Table 3 that the contrast of image C
335: relative to A and B is more pronounced in the continuum than in the
336: emission lines; when viewed in the emission line light, images B and C
337: are of comparable brightness.
338:
339: \begin{table}
340: \begin{center}
341: \caption{Continuum to emission line flux ratios}
342: \begin{tabular}{lcc} \hline \hline
343: Image & Flux (continuum) &
344: Flux (Ly$_\alpha$ \& N~V) \\ \hline
345: A & 1.00 & 1.00 \\
346: B & 0.63 & 0.40 \\
347: C & 0.23 & 0.38 \\ \hline
348: \end{tabular}
349: \end{center}
350: \end{table}
351:
352: Such spectral differences are a natural consequence of differential
353: microlensing effects, with the small continuum emitting source being
354: more severely influenced by the action of microlensing than the larger
355: line emitting region (e.g. Saust 1994). In fact, it is generally
356: thought that the scale of the broad emission line region is sufficient
357: for it to be immune from significant microlensing influences (Nemiroff
358: 1988; Schneider \& Wambsganss 1990), and hence it reflects the
359: magnification due to the macrolens. While unresolved photometric
360: monitoring of APM08279+5255 reveals that this system has exhibited
361: pronounced variability over several tenths of a magnitude (Lewis, Robb
362: \& Ibata 1999; see also the continuing monitoring program at the Wise
363: Observatory at {\tt http://wise-obs.tau.ac.il/$\sim$eran/LM/}), the
364: spectroscopic evidence presented here points to the variation being
365: potentially due to gravitational microlensing.
366:
367: Hence, we conclude that the macrolensing magnifications of images B \&
368: C, based on the emission lines, are comparable, and it is these that
369: should be used in gravitational lens modeling of this system, rather
370: than the continuum flux. We note that a similar conclusion was reached
371: by Lewis et al (2002) who, based on extended CO emission in this
372: system, demonstrated that the likely lens is a highly flattened
373: object, such as an edge-on spiral. While their model predicted that
374: these images should be $\sim75\%$ the brightness of image A, they do
375: concede that currently there are not enough constraints to uniquely
376: tie-down a model. The magnifications presented in this paper,
377: therefore, will aid in the modeling of this unique ternary
378: gravitational lens system. It is important to note, however, that dust
379: in the lensing galaxy may result in extinction of some of the quasars
380: flux, especially image C which may be viewed through the disk of a
381: spiral system, and the values presented in Table~3 may not truly
382: reflect the magnifications of the macromodel. While dust can extinct
383: the quasar images, it cannot be responsible for the differing
384: line-to-continuum ratios observed in the STIS spectra, and hence the
385: conclusion of gravitational microlensing in this system is robust.
386:
387: It is also interesting to note that, compared to the emission line
388: flux, while the continuum in image B appears to be enhanced, the
389: continuum in image C appears to be depressed. Such a situation occurs
390: during a microlensing demagnification and is seen dramatically in
391: image D of the quadruple lens Q2237+0305 (Lewis et al. 1998). With
392: such a delineation in microlensing effects, an estimate of the size of
393: the emission regions can be made. Regions significantly smaller than
394: the gravitational microlensing scale-length, the Einstein radius,
395: (i.e. the continuum emitting region) can be significantly influenced
396: during microlensing, while regions significantly larger (e.g. the
397: broad emission line region) are not. The source plane Einstein
398: radius, $\eta$, is given by
399: \begin{equation}
400: \eta = \sqrt{ \frac{4 G M}{c^2} \frac{D_{os} D_{ls}}{D_{ol}} }
401: \end{equation}
402: where $D_{ij}$ are the angular diameter distances between an observer
403: (o), lens (l) and source (s). For \apm, $\eta \sim 0.01
404: \sqrt{M/M_\odot} h_{50}^{-\frac{1}{2}}{\rm pc}$ for an $\Omega_o=1$
405: cosmology, where $M$ is the typical microlensing mass. Of course,
406: confirmation of the gravitational lens nature of these spectral
407: features, and hence the applicability of this delineation
408: scale-length, requires further time resolved spectroscopy. While
409: spatially resolved spectroscopy, as presented in this paper, would be
410: ideal, variability in the equivalent widths of the emission lines
411: should be apparent in unresolved ground-based spectra.
412:
413: \subsection{Where is the lensing galaxy?}\label{wheregalaxy}
414: The spectra were examined for a signature of the lensing galaxy. The
415: models of Ibata et al. (1999), Egami et al. (2000) and Munoz et
416: al. (2001) all place the lensing galaxy in the close vicinity of the
417: quasar images, although to account for the relative image
418: brightnesses, this lensing galaxy possesses an implausibly large
419: core. Using a flattened potential and explaining the relative image
420: brightnesses as due to the influence of a `naked cusp', Lewis et
421: al. (2002) find that the lens galaxy is offset by $\sim0.5$ arcsec
422: from the quasar images.
423:
424: The data were examined for evidence of the lensing galaxy.
425: Unfortunately, it was found that it is not possible to place any
426: stringent constraints on the lensing galaxy from these
427: observations. The reason for this is that the PSF model we employed
428: does not give an accurate representation of our observations (the
429: spatial profile of the 51~Peg spectra is not as strongly asymmetric as
430: the profile shown in Figure~1), and it was not possible to subtract
431: off the bright quasar images to better than $\sim2\%$ using that model
432: PSF. By considering the residual luminosity in dark troughs in a Keck
433: HIRES spectrum of this system (Ellison et al. 1999a,b), Ibata et al.
434: (1999) showed that the lensing galaxy must be at least seven
435: magnitudes fainter than image A, so the present constraint contributes
436: no additional useful information.
437:
438: \section{Conclusions}\label{conclusions}
439: In this paper we have presented spatially resolved spectra of the
440: gravitationally lensed BAL quasar \apm\ obtained with STIS on the
441: HST. They clearly show that each of the three point-like sources are
442: images of the background quasar, confirming that \apm\ represent the
443: first truly odd-image lens system.
444:
445: An examination of the spectra reveals significant differences in the
446: equivalent widths of the Ly$\alpha + $N~V emission feature between the
447: images. Such differences are naturally explained by the differential
448: magnification influence of gravitational microlensing, which affects
449: the small continuum source but not the larger broad emission line
450: region of the QSO. While further spectroscopic monitoring is
451: necessary to confirm this microlensing hypothesis, unresolved
452: ground-based observations should be adequate for this purpose (as they
453: will still show variations in the equivalent widths of the emission
454: lines).
455:
456: Finally, the spectra where examined for evidence of the lensing
457: galaxy. While the HST resolution allowed us to extract the spectra of
458: individual quasar images, uncertainties in the point spread function
459: of the instrument limited the accuracy of the subtraction to $\sim
460: 2$\%. Available imaging and spectroscopy data show that the lensing
461: galaxy must be significantly fainter than 2\% of the QSO images, so
462: that we cannot detect it in the present set of observations.
463:
464: %\section*{Acknowledgments}
465:
466: \newcommand{\mnras}{MNRAS}
467: \newcommand{\nat}{Nature}
468: \newcommand{\araa}{ARAA}
469: \newcommand{\aj}{AJ}
470: \newcommand{\apj}{ApJ}
471: \newcommand{\apjl}{ApJ}
472: \newcommand{\apjs}{ApJSupp}
473: \newcommand{\aap}{A\&A}
474: \newcommand{\aaps}{A\&ASupp}
475:
476: \begin{thebibliography}{DUM}
477: %
478:
479: %\bibitem[Alloin et al.\ 1997]{1997A&A...321...24A}
480: %Alloin D., Guilloteau S., Barvainis R., Antonucci R., Tacconi L.,
481: %1997, A\&A, 321, 24
482:
483: %\bibitem[Bartelmann 2000]{2000A&A...357...51B}
484: %Bartelmann M., 2000, A\&A, 357, 51
485:
486: \bibitem[Bartelmann \& Loeb 1998]{1998ApJ...503...48B}
487: Bartelmann M., Loeb A., 1998, ApJ, 503, 48
488:
489: %\bibitem[Barvainis et al.\ 1997]{1997ApJ...484..695B}
490: %Barvainis R., Maloney P., Antonucci R., Alloin D.,
491: %1997, ApJ, 484, 695
492:
493: %\bibitem[Blain et al.\ 1999]{1999MNRAS.303..423B}
494: %Blain A.\ W., Moller O., Maller A.\ H.,
495: %1999, MNRAS, 303, 423
496:
497: \bibitem[Burke 1981]{1981ApJ...244L...1B}
498: Burke W.\ L.,
499: 1981, ApJ, 244, L1
500:
501: %\bibitem[]{}
502: %Carilli, C.L., Menten, K.M., and Yun, M.S. 1999, APJ,
503: %521, L25
504:
505: %\bibitem[]{}
506: %Carilli, C.L. and Holdaway, M.A. 1999, Radio Science,
507: %34, 817
508:
509: %\bibitem[Downes et al.\ 1999]{1999ApJ...513L...1D}
510: %Downes D., Neri R., Wiklind T., Wilner D.\ J., Shaver P.\ A.,
511: %1999, ApJ, 513, L1
512:
513: %\bibitem[]{}
514: %Downes, D. \& Solomon, P.M. 1998, APJ, 507, 615
515:
516: %\bibitem[Downes et al.\ 1995]{1995ApJ...453L..65D}
517: %Downes D., Solomon P.\ M., Radford S.\ J.\ E.,
518: %1995, ApJ, 453, L65
519:
520: \bibitem[Egami et al.\ 2000]{2000ApJ}
521: Egami E., Neugebauer G., Soifer B.\ T., Matthews K., Ressler M.,
522: Becklin E.\ E., Murphy T.\ W., Dale
523: D.\ A., 2000, ApJ, 535, 561
524:
525: \bibitem[]{aaaaa}
526: Ellison S.\ L., Lewis
527: G.\ F., Pettini M., Sargent W.\ L.\ W., Chaffee F.\ H., Foltz C.\ B., Rauch
528: M., Irwin M.\ J.,
529: 1999a, PASP, 111, 946
530:
531: \bibitem[]{asasasa}
532: Ellison S.\ L., Lewis G.\ F., Pettini M., Chaffee F.\ H., Irwin M.\ J.,
533: 1999b, ApJ, 520, 456
534:
535: \bibitem[Francis et al.(1991)]{1991ApJ...373..465F}
536: Francis, P.~J., Hewett, P.~C., Foltz, C.~B., Chaffee, F.~H., Weymann, R.~J., \& Morris, S.~L.\
537: 1991, ApJ, 373, 465
538:
539: %\bibitem[Hagen et al.(1992)]{1992A&A...253L...5H}
540: %Hagen, H.-J.~et al.\
541: %1992, A\&A, 253, L5
542:
543: \bibitem[]{dfdfdfd}
544: Ibata R.\ A., Lewis G.\ F., Irwin M.\ J., Leh{\'a}r J., Totten E.\ J.,
545: 1999, AJ, 118, 1922
546:
547: \bibitem[]{erere}
548: Irwin M.\ J., Ibata R.\ A., Lewis G.\ F., Totten E.\ J.,
549: 1998, ApJ, 505, 529
550:
551: %\bibitem[]{sdsd}
552: %Keeton C.\ R., Kochanek C.\ S.,
553: %1998, ApJ, 495, 157
554:
555: %\bibitem[]{sddsd}
556: %Keeton C.\ R.,
557: %2001, {\it astro-ph/0105200}
558:
559: %\bibitem[]{}
560: %Kneib, J.-P., Alloin, D., \& Pello, R. 1998, A\&A, 339, L65
561:
562: %\bibitem[Kneib et al.\ 1998]{1998A&A...329..827K}
563: %Kneib J.\ -P., Alloin D., Mellier Y., Guilloteau S., Barvainis R.,
564: %Antonucci R.,
565: %1998, A\&A, 329, 827
566:
567: %\bibitem[Kochanek et al.\ 2001]{2001ApJ...547...50K}
568: %Kochanek C.\ S., Keeton C.\ R., McLeod B.\ A.,
569: %2001, ApJ, 547, 50
570:
571: %\bibitem[Koopmans et al.\ 1998]{1998MNRAS.295..534K}
572: %Koopmans L.\ V.\ E., de Bruyn A.\ G., Jackson N.,
573: %1998, MNRAS, 295, 534
574:
575: \bibitem[Ledoux et al.\ 1998]{1998AL}
576: Ledoux C., Theodore B., Petitjean P., Bremer M.\ N., Lewis G.\ F.,
577: Ibata R.\ A., Irwin M.\ J., Totten E.\ J.,
578: 1998, A\&A, 339, L77
579:
580: \bibitem[]{erer}
581: Lewis G.\ F., Carilli, C., Papadopoulos, P., Ivison, R. J.,
582: 2002, MNRAS, 330, L15
583:
584: \bibitem[]{erer}
585: Lewis G.\ F., Chapman S.\ C., Ibata R.\ A., Irwin M.\ J., Totten E.\ J.,
586: 1998, ApJ, 505, L1
587:
588: \bibitem[Lewis et al.~1998]{1998MNRAS.295..573L}
589: Lewis G.~F., Irwin M.~J., Hewett P.~C., Foltz C.~B.,
590: 1998, MNRAS, 295, 573
591:
592: \bibitem[Lewis et al.\ 1999]{1999PASP..111.1503L}
593: Lewis G.\ F., Robb R.\ M., Ibata R.\ A.,
594: 1999, PASP, 111, 1503
595:
596: %\bibitem[Maller et al.\ 1997]{1997ApJ...486..681M}
597: %Maller A.\ H., Flores R.\ A., Primack J.\ R.,
598: %1997, APJ, 486, 681
599:
600: %\bibitem[Moller \& Blain 1998]{1998MNRAS.299..845M}
601: %Moller O., Blain A.\ W.,
602: %1998, MNRAS, 299, 845
603:
604: \bibitem[sd]{sdsdsdsdw}
605: Munoz, J.\ A., Kochanek, C.\ S., Keeton, C. R.,
606: 2001, ApJ, 558, 657
607:
608: \bibitem[Narasimha et al.~1986]{1986Natur.321...45N}
609: Narasimha D., Subramanian K., Chitre S.~M.,
610: 1986, Nature, 321, 45
611:
612: \bibitem[Nemiroff 1988]{1988ApJ...335..593N}
613: Nemiroff R.~J.,
614: 1988, ApJ, 335, 593
615:
616: %\bibitem[]{erererere}
617: %Papadopoulos P., Ivison R., Carilli C., Lewis G.,
618: %2001, Nature, 409, 58
619:
620: \bibitem[Petitjean et al.~2000]{2000A&A...359..457P}
621: Petitjean P., Aracil B., Srianand R., Ibata R.,
622: 2000, A\&A, 359, 457
623:
624: %\bibitem[Sanders \& Mirabel 1996]{1996ARA&A..34..749S}
625: %Sanders D.\ B., Mirabel I.\ F.,
626: %1996, ARA\&A, 34, 749
627:
628: \bibitem[Saust(1994)]{1994A&AS..103...33S}
629: Saust, A.~B.\
630: 1994, A\&A Supp, 103, 33
631:
632: \bibitem[Schneider \& Wambsganss 1990]{1990A&A...237...42S}
633: Schneider P., Wambsganss J.,
634: 1990, A\&A, 237, 42
635:
636: \bibitem[Srianand \& Petitjean 2000]{2000A&A...357..414S}
637: Srianand R., Petitjean P.,
638: 2000, A\&A, 357, 414
639:
640: %\bibitem[]{Yun}
641: %Yun, M.S., Scoville, N.Z., Carrasco, J.J., \& Blandford, R.D. 1997,
642: %APJ, 479, L9
643:
644: %
645: \end{thebibliography}
646: \end{document}
647:
648: