1: %\documentclass[12pt,preprint]{aastex} %rm \section{Appendix} & tbl \clearpages
2: \documentclass[12pt]{emulateapj}
3: \newcommand{\tcaption}{
4: \setlength{\abovecaptionskip}{0pt}
5: \setlength{\belowcaptionskip}{5pt}
6: \caption}
7: \newcommand{\jo}{SDSS J0242+0049}
8: \newcommand{\kms}{km\,s$^{-1}$}
9: \newcommand{\cmss}{cm~s$^{-2}$}
10: %\slugcomment{Submitted to ApJ Dec. 15, 2006}
11: \slugcomment{}
12: \journalinfo{Submitted to ApJ Dec. 15, 2006; Accepted April 27, 2007}
13: \shorttitle{Quasar Outflow Constraints}
14: \shortauthors{Hall et al.}
15:
16: \begin{document}
17: \title{Acceleration and Substructure Constraints in a Quasar Outflow}
18: \author{
19: Patrick B. Hall,\altaffilmark{1}
20: Sarah I. Sadavoy,\altaffilmark{1}
21: Damien Hutsemekers,\altaffilmark{2}
22: John E. Everett,\altaffilmark{3}
23: Alireza Rafiee\altaffilmark{1}
24: }
25: \altaffiltext{1}{Department of Physics and Astronomy,
26: York University, 4700 Keele St., Toronto, Ontario M3J 1P3, Canada}
27: \altaffiltext{2}{Senior Research Associate FNRS, University of Li\`ege,
28: All\'ee du 6 ao\^ut 17, Bat. 5c, 4000 Li\`ege, Belgium}
29: \altaffiltext{3}{Departments of Physics and Astronomy and Center for Magnetic
30: Self-Organization in Laboratory and Astrophysical Plasmas, University of
31: Wisconsin-Madison, 1150 University Avenue, Madison, Wisconsin 53706, USA}
32:
33: \begin{abstract}
34: We present observations of probable line-of-sight acceleration of a broad
35: absorption trough of \ion{C}{4} in the quasar SDSS J024221.87+004912.6.
36: We also discuss how the velocity overlap of two other outflowing
37: systems in the same object constrains the properties of the outflows.
38: The \ion{Si}{4} doublet in each system has one unblended transition
39: and one transition which overlaps with absorption from the other system.
40: The residual flux in the overlapping trough
41: is well fit by the product of the residual fluxes in the unblended troughs.
42: For these optically thick systems to yield such a result, at least one of them
43: must consist of individual subunits rather than being a
44: single structure with velocity-dependent coverage of the source.
45: If these subunits are identical, opaque, spherical clouds, we
46: estimate the cloud radius to be $r\simeq 3.9\times 10^{15}$~cm.
47: If they are identical, opaque, linear filaments, we estimate their width to be
48: $w\simeq 6.5\times 10^{14}$~cm. These subunits are observed to
49: cover the \ion{Mg}{2} broad emission line region of the quasar, at which
50: distance from the black hole the above filament width is equal
51: to the predicted scale height of the outer atmosphere of a thin accretion disk.
52: Insofar as that scale height is a natural size scale for structures originating
53: in an accretion disk, these observations are evidence that the accretion disk
54: can be a source of quasar absorption systems.
55: Based on data from ESO program 075.B-0190(A).
56: \end{abstract}
57: \keywords{quasars: general, absorption lines, individual (SDSS J024221.87+004912.6)}
58:
59: \section{Introduction}
60:
61: Absorption systems in active galactic nuclei (AGN) can be classified as
62: \emph{intrinsic}, which are
63: associated with the active nucleus and
64: often outflowing from it, and \emph{intervening}, which originate from clouds
65: external to the AGN's environment. Determining an absorption
66: system's classification can be difficult.
67: A reliable indication of intrinsic absorption is time variability,
68: such as a shift in the velocity of a given
69: feature or changes in its absorption strength as a function of velocity.
70: In the broad absorption line (BAL) troughs $\gtrsim1000$\,km\,s$^{-1}$ wide
71: which are most often found in the luminous AGN known as quasars,
72: reports of time variable absorption strengths have been relatively common
73: (\nocite{bea85,fol87,vsb87,sp88,tea88,bjb89,bjb92,bea92,bar94,hea95,mon96,hbj97,bhs97,hea97,sdss123,gea04,mect05}{Bromage} {et~al.} 1985; {Foltz} {et~al.} 1987; {Voit}, {Shull}, \& {Begelman} 1987; {Smith} \& {Penston} 1988; {Turnshek} {et~al.} 1988; {Barlow}, {Junkkarinen}, \& {Burbidge} 1989, 1992a; {Barlow} {et~al.} 1992b; {Barlow} 1994; {Hamann} {et~al.} 1995; {Michalitsianos}, {Oliversen}, \& {Nichols} 1996; {Hamann}, {Barlow}, \& {Junkkarinen} 1997b; {Barlow}, {Hamann}, \& {Sargent} 1997; {Hamann} {et~al.} 1997a; {Hall} {et~al.} 2002; {Gallagher} {et~al.} 2004; {Misawa} {et~al.} 2005; Lundgren et al. 2006).
74: Such variability can even include the appearance or disappearance of absorption
75: systems \nocite{kea96,gce01,ma02,gea05,lchg05}({Koratkar} {et~al.} 1996; {Ganguly}, {Charlton}, \& {Eracleous} 2001; {Ma} 2002; {Gallagher} {et~al.} 2005; {Leighly} {et~al.} 2005). In contrast,
76: velocity shifts in BAL outflows have been reported in only
77: Q~1303+308 \nocite{vi01}({Vilkoviskij} \& {Irwin} 2001) and
78: Mrk~231 (\nocite{rvs02}{Rupke}, {Veilleux}, \& {Sanders} 2002, and references therein),
79: although \nocite{gea03}{Gabel} {et~al.} (2003) have observed deceleration of a narrow absorber in
80: the Seyfert~1 NGC 3783.
81:
82: Acceleration must occur for AGN outflows to reach their observed velocities.
83: Nonetheless, velocity shifts in AGN outflows are seen quite rarely because
84: acceleration of an AGN outflow does not automatically translate into a change
85: in its observed velocity profile, and vice versa. For example, a fixed
86: mass loss rate into an outflow with a time-invariant driving force would yield
87: a time-invariant acceleration profile with distance in the outflow,
88: and thus produce unchanging absorption troughs.
89: \nocite{aea99}{Arav} {et~al.} (1999) illustrate how radial acceleration of gas crossing our line of
90: sight with a non-negligible transverse velocity produces an observed absorption
91: trough with a broadened radial velocity profile that does not change with time.
92: Since our lines of sight to AGN are essentially radial, and since
93: AGN are fed by accretion disks consisting of gas with predominantly
94: orbital velocities, most AGN outflows are expected to have non-negligible
95: transverse as well as radial velocities. Thus, most intrinsic absorbers
96: likely {\em are} exhibiting acceleration, disguised as a trough broader
97: than the thermal or turbulent velocity width of the gas.
98:
99: What are we then to make of cases where an outflow {\em does} exhibit a velocity
100: shift? First, note that when
101: our line of sight intersects the origin of an outflow, the
102: absorption trough can start at zero line-of-sight velocity in the AGN rest
103: frame, at least for ions present at the origin of the outflow.
104: Ions present only downstream in an outflow, or lines of sight intersecting an
105: outflow only downstream from its origin due to curvature in the flow lines,
106: will produce `detached' absorption troughs which do not start at zero velocity,
107: as will a shell of material ejected in an intermittent outflow.
108: With that in mind, consider possible explanations for
109: a velocity shift observed in a detached absorption trough.
110: Such a shift can be produced by changes in the ionization state as a function
111: of velocity in a fixed outflow, by changes in the acceleration profile
112: or geometry (or both) of such an outflow due to changes in the driving force or
113: mass loss rate, or by actual line-of-sight acceleration of a shell of material
114: from an intermittent outflow.
115: Observations of velocity shifts are therefore worthwhile because they may yield
116: insights into specific scenarios for quasar absorbers.
117:
118: Here we present multiple-epoch observations (\S 2) of a quasar in which a broad
119: absorption line trough of \ion{C}{4} increased in outflow velocity over
120: 1.4 rest-frame years (\S 3). We also discuss how two overlapping
121: outflows in the same quasar provide constraints on the properties of those
122: outflows (\S 4). We end with our conclusions in \S 5.
123:
124: \section{Observations}
125:
126: The Sloan Digital Sky Survey \nocite{yor00}(SDSS; {York} {et~al.} 2000) is using a drift-scanning
127: camera \nocite{gun98}({Gunn} {et~al.} 1998) on a 2.5-m telescope \nocite{gun06}({Gunn} {et~al.} 2006) to image 10$^4$\,deg$^2$
128: of sky on the SDSS $ugriz$ AB magnitude system
129: \nocite{fuk96,sdss82,sdss105,sdss153,ive04}({Fukugita} {et~al.} 1996; {Hogg} {et~al.} 2001; {Smith} {et~al.} 2002; {Pier} {et~al.} 2003; {Ivezi{\'c}} {et~al.} 2004).
130: Two multi-fiber, double spectrographs are being used to obtain resolution
131: $R\sim1850$ spectra covering $\simeq$3800-9200\,\AA\
132: for $\sim$10$^6$ galaxies to $r=17.8$ and $\sim$10$^5$ quasars to
133: $i=19.1$ ($i=20.2$ for $z>3$ candidates; \nocite{sdssqtarget}{Richards} {et~al.} 2002).
134:
135: The $z_{em}=2.062$ BAL quasar SDSS J024221.87+004912.6
136: \nocite{sdssedrq,sdssbalcat,dr3q,trump06}({Schneider} {et~al.} 2002; {Reichard} {et~al.} 2003; {Schneider} {et~al.} 2005; {Trump} {et~al.} 2006), hereafter referred to as \jo,
137: was observed spectroscopically three times by the SDSS (Table \ref{spec}).
138: We selected it for high-resolution spectroscopic followup because of the
139: possible presence of narrow absorption in excited-state \ion{Si}{2} and
140: \ion{C}{2} at $z=2.042$. A spectrum obtained with
141: the ESO Very Large Telescope (VLT) Unit 2 (Kueyen)
142: and Ultra-Violet Echelle Spectrograph (UVES; \nocite{uves}{Dekker} {et~al.} 2000)
143: confirms the presence of narrow, low-ionization absorption at that
144: redshift,\footnote{The weak, narrow character of that absorption
145: led to the classification of this object as a high-ionization BAL quasar
146: by \nocite{sdssbalcat}{Reichard} {et~al.} (2003) and \nocite{trump06}{Trump} {et~al.} (2006) based on its SDSS spectrum.}
147: analysis of which will be reported elsewhere.
148:
149: We observed \jo\ with UVES on the VLT UT2 on the nights of 4-5 September 2005
150: through a 1\arcsec\ slit with 2x2 binning of the CCD, yielding $R\simeq40000$.
151: The weather ranged from clear to thin cirrus, with $0.8-1.0$\arcsec\ seeing.
152: \jo\ was observed for a total of 5.75 hours in two different spectral
153: settings, yielding coverage from 3291-7521\,\AA\ and 7665-9300\,\AA.
154: Each exposure was reduced individually with optimum extraction \nocite{hor86}({Horne} 1986),
155: including simultaneous background and sky subtraction. Telluric absorption
156: lines were removed for the red settings using observations of telluric standard
157: stars. A weighted co-addition of the three exposures of each spectral setting
158: was performed with rejection of cosmic rays and known CCD artifacts.
159: Finally, all settings were rebinned to a vacuum heliocentric wavelength scale,
160: scaled in intensity by their overlap regions, and merged into a single spectrum
161: with a constant wavelength interval of 0.08\,\AA\ (Figure \ref{figtot}).
162: The SDSS spectra all share a common wavelength system with pixels equally
163: spaced in velocity, and so for ease of comparison we created a version of the
164: UVES spectrum binned to the those same wavelengths
165: but not smoothed to the SDSS resolution.
166:
167: \section{Broad Absorption Line Trough Velocity Shifts}\label{bal}
168:
169: The broadest absorption lines in \jo\ occur at a
170: redshift $z\simeq1.87988$ ($v=-18400$\,\kms\ relative to the quasar)
171: in Ly$\alpha$, \ion{N}{5}, \ion{Si}{4} and \ion{C}{4} (Figure \ref{fig:BAL}).
172: There is an offset between the peak absorption in \ion{C}{4} and \ion{Si}{4}.
173: The redshift $z=1.87988$ was determined from the deepest absorption in the
174: \ion{Si}{4} trough, and does not match the deepest \ion{C}{4} absorption.
175: This can be ascribed to a changing ionization state in the outflow
176: as a function of velocity.
177:
178: Comparison of the SDSS and UVES spectra suggested a shift in the position of
179: the \ion{C}{4} trough at this redshift. To investigate further, continuum
180: regions around that trough and the \ion{Si}{4} trough at the same redshift
181: were fitted and used to normalize all observed spectra.
182: (The Ly$\alpha$ and \ion{N}{5} troughs lie outside the SDSS wavelength range.)
183: For each epoch, the \ion{C}{4} and \ion{Si}{4} regions were fit separately
184: with third order Legendre functions using
185: {\sc splot} in IRAF.\footnote{The Image Reduction and Analysis Facility (IRAF)
186: is distributed by the National Optical Astronomy Observatories, which is
187: operated by AURA, Inc., under contract to the National Science Foundation.}
188: The continuum sample windows were selected to avoid
189: emission lines in the quasar rest frame \nocite{sdss73}({Vanden Berk} {et~al.} 2001).
190:
191: The extent of any shift can be measured by minimizing the $\chi^{2}$
192: between the normalized pixel-by-pixel fluxes in the spectra
193: when shifted by an integer number of pixels $m$
194: (assuming pixels equally spaced in velocity):
195: \begin{equation}
196: \chi_{\nu,m}^2 = \frac{1}{N-m} \sum_{i=1}^{N-m} \frac{(f_{2,i}-f_{1,i+m})^2}{\sigma_{1,i}^2+\sigma_{2,i+m}^2} \label{eqChi}
197: \end{equation}
198: where $f_{2,i}$ and $f_{1,i+m}$ represent the flux in spectra from epochs 1 and
199: 2 at pixels $i$ and $i+m$, respectively, $N$ is the
200: total number of pixels extracted from each spectrum for comparison
201: and $\sigma$ is the error for the flux at each pixel.
202:
203: The SDSS spectra from epochs 51821 and
204: 52188\footnote{Since the SDSS spectra from MJD 52177 and MJD 52199 are noisier
205: at short wavelengths than the SDSS spectrum from MJD 51821 and since visual
206: inspection of them revealed no obvious difference in their BAL troughs,
207: a weighted co-add of their spectra was made, with mean epoch 52188.}
208: were compared with the UVES spectrum from epoch 53619 (Table \ref{spec}).
209: A clear shift was found in \ion{C}{4} and a potentially
210: smaller shift in \ion{Si}{4}.
211: Neither trough shows a detectable shift between the SDSS spectra
212: from epoch 51821 and epoch 52188, and neither would be expected to do so if
213: the observed long-term shift was due to a constant acceleration (the shift
214: between those two epochs would be $\lesssim 0.5$ pixel for \ion{C}{4}).
215: In light of this, the $\chi^2$ test was conducted again, using
216: a weighted average of all three SDSS spectra, with mean epoch 52066.
217: From that comparison we conclude that the shift in \ion{C}{4} is
218: $3 \pm 1$ pixels
219: with 95.4\% confidence (2$\sigma$).
220: Zero velocity shift in \ion{C}{4} can be excluded with 99.9998\% confidence.
221: For \ion{Si}{4}, the shift is $1 \pm 3$ pixels at 95.4\% confidence.
222: Plots of these spectra are shown in the top two panels of Figure \ref{accNorm}.
223: It is important to note that there is no shift in the nearby narrow absorption
224: lines. Also, both troughs appear to keep a relatively constant intensity,
225: within the uncertainties. The bottom panel of Figure \ref{accNorm}
226: shows the excellent match to the epoch 53619 UVES spectrum
227: that results when the epoch 52066 average SDSS spectrum is shifted by 3 pixels.
228:
229: The middle panel of Figure \ref{accNorm} may suggest
230: that the long-wavelength end of the \ion{C}{4}
231: trough has a greater shift than the short-wavelength end. Splitting the
232: \ion{C}{4} trough into two sections, we find that $\chi^{2}$ is minimized at
233: a shift of $2^{+2}_{-1}$ pixels for the short-wavelength end and a shift of
234: $4^{+1}_{-2}$ pixels for the long-wavelength edge,
235: but that a uniform shift produces a marginally lower minimum overall $\chi^{2}$.
236: Thus, while there is possible evidence for a nonuniform velocity
237: shift of the \ion{C}{4}
238: BAL trough, the current data are of insufficient quality to prove its existence.
239: Many physical effects could produce a nonuniform shift (expansion of an
240: overpressured, accelerated shell of gas from an intermittent outflow,
241: to give one example).
242:
243: A shift of one SDSS pixel corresponds to a velocity shift of 69 \kms\ in the
244: observed frame or 22.5 \kms\ in the quasar rest frame ($z=2.062$).
245: A shift of $3 \pm 1$ SDSS pixels (2$\sigma$) over a rest-frame time span
246: of 1.39 years thus gives an acceleration of
247: $a = 0.154 \pm 0.025 \mbox{\ cm\ s}^{-2}$, where the error is 1$\sigma$.
248: Previously claimed accelerations for BAL troughs are much lower than that, at
249: $a = 0.035\pm 0.016$ \cmss\ over 5.5 rest-frame years in Q~1303+308 \nocite{vi01}({Vilkoviskij} \& {Irwin} 2001)
250: and $a=0.08 \pm 0.03$ \cmss\ over 12 rest-frame years for Mrk~231 \nocite{rvs02}({Rupke} {et~al.} 2002).
251: Our observation is more similar to that of \nocite{gea03}{Gabel} {et~al.} (2003), who determined the
252: deceleration of \ion{C}{4}, \ion{N}{5} and \ion{Si}{4}
253: in a narrow absorption system in a Seyfert galaxy and found
254: (for \ion{C}{4}) relatively large values of $a=-0.25\pm 0.05$ \cmss\ and
255: $a=-0.10\pm 0.03$ \cmss\ over 0.75 and 1.1 rest-frame years, respectively.
256: All of those observations involved much narrower troughs than is the case in
257: \jo. Also, the 1$\sigma$ relative uncertainty associated with the acceleration
258: of \jo\ is lower than the previous BAL measurements. These factors make \jo\ a
259: robust case for line-of-sight acceleration of a true BAL trough.
260: Still, it should be kept in mind that all these accelerations are
261: much smaller than the $a\simeq 100 \mbox{\ cm\ s}^{-2}$ predicted for
262: the main acceleration phase of a disk wind in the model of \nocite{mur95}{Murray} {et~al.} (1995).
263:
264: Furthermore,
265: BAL troughs can vary for several reasons. These include
266: acceleration or deceleration along the line of sight of some or all of the
267: absorbing gas,
268: a change in the ionization state of some or all of the gas, or
269: a change in $C(v)$ --- the covering factor of the gas as a function of the
270: line-of-sight velocity ---
271: due to the movement of gas into or out of our line of sight, for example
272: due to a change in flow geometry (see the introduction and
273: \S 3.3 of \nocite{gea03}{Gabel} {et~al.} 2003).
274: In many cases of variability all of the above origins are possible, but there
275: are cases where acceleration is very unlikely to be the cause (see below).
276: Because of this,
277: to be conservative we cannot assume that BAL trough variability is due to
278: acceleration
279: even though acceleration {\em could} be the cause of much of the
280: observed variability.
281:
282: Fig. 2 of \nocite{bjb89}{Barlow} {et~al.} (1989) and Fig. 2 of \nocite{bea92}{Barlow} {et~al.} (1992b)
283: are cases where observed time variability of BAL troughs is almost certainly
284: due to a change in the column densities of an ion at certain velocities
285: (whether due to a changing ionization or to bulk motion into the line of sight),
286: not due to a given ionic column density changing its velocity.
287: More ambiguous cases are illustrated by
288: \ion{C}{4} in Q~1246$-$057 (Fig. 3 of \nocite{sp88}{Smith} \& {Penston} 1988)
289: and \ion{Si}{4} in Q~1413+117 (Fig. 15 of \nocite{tea88}{Turnshek} {et~al.} 1988).
290: In both of those cases, a second-epoch spectrum shows more absorption
291: at the short-wavelength edge of the trough in question.
292: That could be because gas at lower outflow velocities in the trough was
293: accelerated to higher velocities. Yet in both cases, the
294: trough away from the short-wavelength edge is unchanged between the two epochs.
295: If acceleration was the cause of the variability,
296: a reduction in covering factor or optical depth, or both,
297: might be expected at the lower velocities where the gas originated.
298: No reduction is seen, arguing against the
299: line-of-sight acceleration hypothesis for these cases of trough variability.
300:
301: While every case for acceleration in a BAL trough will be ambiguous at some
302: level, comparing the variability we report in \jo\ to previous cases leads
303: us to believe that ours is the least ambiguous case seen to date of
304: acceleration in a true BAL trough ($\gtrsim 1000$\ \kms\ wide).
305: Monitoring the future behavior of the $z=1.87988$
306: absorption in this quasar would be very worthwhile, to see if the acceleration
307: was temporary, is constant, increasing, or decreasing, or varies stochastically.
308: The latter might occur if the velocity shift is due to a variable flow geometry
309: or to ionization variations as a function of velocity
310: caused by a fluctuating ionizing luminosity.
311: (Recall from Figure 2 that this system shows some evidence for ionization
312: stratification with velocity, in the form of an offset between the
313: velocities of the peak \ion{Si}{4} and \ion{C}{4} absorption.)
314: As this quasar is located in the equatorial stripe of the SDSS, which has been
315: repeatedly imaged over the past 7 years, it should eventually be possible to
316: search for a correlation between its ultraviolet luminosity and the
317: acceleration of this system.
318: (From the spectra alone, there appears to be a 5-10\% increase in the
319: luminosity of the object over the time spanned by the three SDSS spectra,
320: but no information is available on longer timescales since
321: the UVES spectrum is not spectrophotometrically calibrated.)
322: BAL trough velocity shifts are also expected if
323: BAL quasars are
324: a short-lived phase during which material is expelled from the nuclear region
325: \nocite{vwk93}({Voit}, {Weymann}, \& {Korista} 1993). In such a model the accelerating trough in \jo\ could be
326: interpreted as gas unusually close to the quasar,
327: currently experiencing an unusually large radiative acceleration.
328:
329: \section{Overlapping \ion{Si}{4} Troughs}
330:
331: There is a possible case of line-locking involving \ion{Si}{4}
332: in \jo. Stable line-locking in a given doublet occurs
333: when two conditions are met. First,
334: the velocity separation between two absorption systems at different redshifts
335: must be
336: very nearly equal to the velocity separation of the two lines of a doublet
337: seen in both systems \nocite{bm89}({Braun} \& {Milgrom} 1989). Second,
338: the reduction in line-driven acceleration of the shadowed system due to the
339: reduced incident flux in one component of the doublet must result in its
340: acceleration being the same as that of the shadowing system. This latter
341: condition may be difficult to meet in AGN outflows, where many lines
342: contribute to the radiative acceleration and there may also be substantial
343: non-radiative acceleration. Nonetheless, some spectacular examples of
344: apparent line-locking in AGN do suggest that it can in fact occur
345: (e.g., \nocite{splh02}{Srianand} {et~al.} 2002), even if only rarely.
346:
347: As shown in Figure \ref{fig:lambda}, in \jo\ there is narrow \ion{Si}{4}
348: absorption at $z=2.0476$ (hereafter system A$'$)
349: and a broad \ion{Si}{4} trough centered at about $z=2.042$ (hereafter system A).
350: \ion{Si}{4} line-locking of a third absorption system to system A$'$ or A
351: would result in absorption 1931 \kms\ shortward of those redshifts,
352: at $z=2.0280$ or $z=2.02245$ respectively.
353: What is observed in the spectrum, however, is broad absorption in between the
354: expected redshifts, centered at $z=2.0254$ (hereafter system B).
355: Both systems are observed in other transitions as well,
356: with system B having more absorption
357: in \ion{N}{5} and \ion{C}{4} but less in \ion{S}{4} and \ion{Mg}{2}.
358:
359: In this section we consider first the optical depths and covering factors of
360: these overlapping systems, with intriguing results.
361: We then consider whether they could be line-locked
362: or in the process of becoming line-locked.
363:
364: \subsection{\ion{Si}{4} Trough Optical Depths and Covering Factors}
365:
366: It is useful to determine if the \ion{Si}{4} troughs under consideration
367: are optically thick or not.
368: Figure \ref{fig:norm} shows the absorption profiles in velocity space relative
369: to $z=2.0476$ or to the corresponding line-locked redshift of $z=2.0280$.
370: System A+A$'$, seen unblended in the bottom panel, is free from
371: contamination in the blended trough (middle panel) at $-900<v<-650$ \kms.
372: At those velocities, absorption from the $\lambda$1402 component of System
373: A+A$'$ (bottom panel) appears so similar in shape and intensity to absorption
374: from the intrinsically stronger $\lambda$1393 component (middle panel)
375: that we can conclude system A+A$'$ is optically thick in \ion{Si}{4}.
376: For system B (seen unblended in the top panel) we
377: must
378: see how well various combinations of optical depth,
379: covering factor, and geometry \nocite{rvs05}({Rupke}, {Veilleux}, \& {Sanders} 2005)
380: can reproduce the profile of the trough composed of blended absorption from
381: system B and the optically thick system A+A$'$ (middle panel).
382:
383: For an unblended doublet, at each velocity $v$ the normalized residual
384: intensities $I_{1}$ and $I_{2}$ (in the stronger and weaker lines, respectively)
385: can be related to the optical depth in the stronger transition $\tau$ and the
386: fraction of the emitting source covered by the absorber along our line of sight,
387: the covering factor $C$ \nocite{sb2}(e.g., {Hall} {et~al.} 2003):
388: \begin{equation}
389: I_{1}(v)=1-C_v(1-e^{-\tau_v})
390: \label{eq:I1}
391: \end{equation}
392: \begin{equation}
393: I_{2}(v)=1-C_v(1- e^{-R\tau_v})
394: \label{eq:I2}
395: \end{equation}
396: where $R$ measures the relative optical depths of the lines.
397: For the \ion{Si}{4} $\lambda\lambda 1393,1402$ doublet, $R=0.5$.
398: In each absorption system we have only one unblended component,
399: but it can still be used to model the other component.
400: (For comparison, the two unblended troughs are overplotted
401: on the blended trough in the top panel of Figure~\ref{fig:norm2}.)
402:
403: First we test whether system B can be optically thin, with $C_v=1$.
404: Using this assumption and equations \ref{eq:I1} and \ref{eq:I2},
405: the optical depth $\tau_v(\lambda1402,B)$ was
406: calculated from the observed trough of \ion{Si}{4} $\lambda$1393 in system B.
407: The blended trough profile in this model should be
408: $\exp[-\tau_v(\lambda1402,B)]$
409: times the profile of \ion{Si}{4} $\lambda$1393 in system A+A$'$.
410: (The latter profile is taken as identical to the $\lambda$1402 trough profile
411: at $z=2.0476$ since system A+A$'$ is optically thick.)
412: The resulting model blended-trough profile is compared to the
413: observed blended-trough profile in the second panel of Figure~\ref{fig:norm2}.
414: Optically thin absorption from system B
415: falls short of explaining the depth of the blended trough.
416:
417: Next we test whether system B can be extremely optically thick, so that
418: the depth of its absorption is determined only by $C_v$. In this case,
419: we have two absorption systems absorbing at each $v$, but with different $C_v$.
420: The total absorption
421: is determined by $C_{v,blended}$, which
422: depends on what parts of the emitting source(s) are covered
423: by neither absorption system, by just one, or by both.
424: %(That is,
425: %the total absorption depends on the extent to which the two systems overlap
426: %transverse to our line of sight and cover the same parts of the source.)
427: %We can rule out the limit of maximum coverage
428: %of the source (maximum $C_{v,blended}$)
429: %where $C_{v,blended}=\min(C_A+C_B,1)$, since in that case
430: %$C_A+C_B>1$ at all $v$.
431: %Another limiting case is minimum coverage of the source,
432: %where $C_{v,blended}=\max(C_A,C_B)$.
433: That is, the total absorption depends on the extent to which the two systems
434: overlap transverse to our line of sight and cover the same parts of the source.
435: We can rule out the limit of minimum overlap, which yields maximum coverage of
436: the source: $C_{v,blended}=\min(C_A+C_B,1)$. In that case
437: $C_A+C_B>1$ at all $v$, but we do not observed $C_{v,blended}=1$ at all $v$.
438: Another limiting case is maximum overlap of the absorption systems, which
439: minimumizes the source coverage: $C_{v,blended}=\max(C_A,C_B)$.
440: The results of that model are shown in the third panel of
441: Figure~\ref{fig:norm2}.
442: It is not an improvement over the optically thin model.
443: However, at almost all velocities the maximum-overlap model has more residual
444: flux than seen in the data, while the minimum-overlap model has less.
445: Thus, overlap in $C_v$ which is less than the maximum possible
446: by a velocity-dependent amount can explain the data.
447: Such spatially-distinct, velocity-dependent partial covering
448: has been seen before in other quasars
449: (see the Appendix to \nocite{sb2}{Hall} {et~al.} 2003).
450:
451: The last case we consider is one where each covering fraction describes
452: the fractional coverage of the other absorption system as well as of the
453: continuum source, so that $I_{v,blended}=I_AI_B$
454: and $C_{v,blended}=C_A+C_B-C_AC_B$ (this is case 3 of \nocite{rvs05}{Rupke} {et~al.} 2005).
455: The results of this model are shown in the bottom panel of
456: Figure~\ref{fig:norm2}, again assuming A and B are both very optically thick.
457: The model reproduces the data reasonably well at almost all velocities,
458: and much more closely overall than the other models considered.
459:
460: The good fit of this model implies that the absorption in one or both of the
461: systems is produced by many small subunits scattered all over the continuum
462: source from our point of view. In that case, the amount of light transmitted
463: through both systems will naturally be $I_A(v)\times I_B(v)$
464: {\em at every velocity} $v$ (Figure~\ref{fig:paper}).
465: Deviations will only occur due to statistical
466: fluctuations, which will be greater the fewer subunits there are.
467: It is more difficult, though still possible, to explain the observations using
468: two `monolithic' systems; that is, systems in which absorption from the ion
469: in question arises in a single structure along our line of sight spanning the
470: range of velocities seen in the trough, but with physical coverage of the
471: source which varies with velocity (e.g., Figure 10 of \nocite{aea99}{Arav} {et~al.} 1999).
472: Two monolithic flows with unblended residual intensities $I_A(v)$ and $I_B(v)$
473: can produce any blended residual intensity from 0 to min($I_A(v),I_B(v)$)
474: essentially independently at each velocity $v$ (Figure~\ref{fig:paper}).
475: Thus, two monolithic flows can explain the observations, but only if they just
476: happen to overlap as a function of velocity in such a way as to mimic the
477: overlap of two systems of clouds.
478: Such an explanation is rather contrived, and we conclude instead that many
479: small subunits exist in one or both absorption systems. This conclusion
480: should of course be tested with observations of additional overlapping
481: absorption systems in other quasars, to ensure this case is not a fluke.
482:
483: Note that we have not considered the effects of different covering factors for
484: the continuum source and broad emission line region. As seen in
485: Figure \ref{fig:lambda}, line emission is a 10\% effect at best,
486: and is not a factor at all in the \ion{Si}{4}\,$\lambda$1393 trough of system B.
487:
488: \subsubsection{Constraints on the Outflow Subunits}
489:
490: The results above suggest that the absorbers A and B are composed of a number of
491: optically thick subunits.
492: We now discuss what we can infer about the parameters of these subunits, in the
493: limit that each subunit is so optically thick it can be treated as opaque.
494:
495: Assume that absorber A's residual intensity at some velocity, $I_A(v)$,
496: is created by $N_A$ subunits intercepting our line of sight, and similarly for
497: absorber B. When the two absorbers overlap along the line of sight, there will
498: be $N=N_A+N_B$ subunits along the line of sight. The average transmitted flux
499: $i$ in this case will be $<i>=(1-p)^N$, where $p$ is the average fraction
500: of the quasar's emission covered by an individual subunit.
501:
502: %(For example, a fixed $N$ would have $\sigma_N^2=0$, while a Poisson
503: %distribution with an average of $N$ would have $\sigma_N^2=N$.)
504:
505: If an average $N$ over all velocities is well defined,
506: the pixel-to-pixel
507: variations around the average value $<i>$ will be distributed with variance
508: $\sigma^2=\sigma_I^2+\sigma_i^2$, where $\sigma_I$ is the instrumental error
509: %(including continuum normalization uncertainty)
510: and $\sigma_i$ is given by
511: \begin{equation}
512: \sigma_i^2 = \sigma_{intrinsic}^2
513: +(1-p)^{2N}\left(\frac{N^2\sigma_p^2}{(1-p)^2} + [\ln(1-p)]^2\sigma_N^2\right).
514: \end{equation}
515: For example, fixed $N$ at all velocities would have $\sigma_N^2=0$, while
516: a Poisson distribution with an average of $N$ would have $\sigma_N^2=N$.
517: The intrinsic variance at fixed $N$ and $p$, $\sigma_{intrinsic}^2$,
518: is caused by the random overlap (or lack thereof) of $N$ subunits of
519: uniform projected fractional area $a$.
520: The relation between $p$ and $a$, and the form of $\sigma_{intrinsic}^2$,
521: depends on the shape of the subunits and of the quasar's emitting region.
522: In the Appendix we give formulae for the
523: cases of rectangular subunits of width $a$ and unit length and of circular
524: subunits of area $a$, under the approximation that the emitting region of the
525: quasar is projected on the sky as a square of unit area and uniform surface
526: brightness (see the discussion in the Appendix).
527: In both cases, $\sigma_p^2 \propto \sigma_a^2$. If $\sigma_a$ is negligible,
528: there are two unknowns ($a$ and $N$) and two observables ($<i>$ and $\sigma$)
529: which can be used to solve for them.
530: %However, we cannot usually assume $N$ is fixed at all velocities.
531:
532: More generally, we can constrain the subunit number and size as follows.
533: We have a predicted profile $i(v)=I_AI_B$ and an observed profile $I(v)$,
534: both of which depend on velocity.
535: In our case, the wide range of $i$ over the full trough
536: and the smooth pixel-to-pixel distribution of $i$
537: cannot simultaneously be reproduced at fixed $N$.
538: Reproducing the wide range of $i$ would require a small $N$,
539: which would not generate as smooth a velocity profile as observed.
540: Each subunit will probably have a velocity dispersion of
541: only $\sim$10\ \kms\ \nocite{pet97}({Peterson} 1997),
542: so for small $N$ strong variations in $i$ would be seen on that velocity scale.
543: Thus, the range in $i$ means either $N$ or $a$ varies with velocity, or both
544: do. To simplify the problem, we assume the subunits have a uniform size so that
545: $a$ is constant and $\sigma_a = 0$. (This should be an adequate approximation
546: if the subunits have a characteristic size scale.)
547: If we then assume a value for $a$, we can calculate
548: a predicted $N$ for each pixel as $N=\log i / \log (1-p)$,
549: using the expression for $p(a)$ appropriate to the chosen geometry.
550: The observed profile $I$ differs slightly from the predicted profile $i=I_AI_B$,
551: due to the intrinsic variance on the total covering factor of $N$ clouds
552: ($\sigma_{intrinsic}^2$)
553: and to the errors on $I_A$ and $I_B$ ($\sigma_A$ and $\sigma_B$, respectively).
554: Setting $\sigma_p\propto \sigma_a=0$ as discussed above and approximating
555: the variance on $N$ as $\sigma_N^2=N$, we have
556: \begin{equation}
557: \sigma_i^2 \simeq \sigma_{intrinsic}^2
558: +(1-p)^{2N}N[\ln(1-p)]^2 + I_B^2\sigma_A^2+I_A^2\sigma_B^2.
559: \end{equation}
560: The probability of observing a residual intensity $I\pm\sigma_I$
561: in a pixel, given a predicted value $i$ and associated $\sigma_i$, is
562: \begin{equation}
563: P(I\pm\sigma_I|i\pm\sigma_i) = \frac{1}{\sqrt{2\pi(\sigma_I^2+\sigma_i^2)}}\exp\left[-\frac{(I-i)^2}{2(\sigma_I^2+\sigma_i^2)}\right].
564: \end{equation}
565: Each pixel has a different $\sigma_i$ which depends on the adopted $a$.
566: To choose the best model, we find the value of $a$ that maximizes the likelihood
567: of the observations: $L=\prod_k P(I_k\pm\sigma_{I_k}|i_k\pm\sigma_{i_k})$.
568: Note that a systematic error in $I$ (e.g., due to a continuum estimate which is
569: too high or too low) will yield a systematic error in $a$.
570:
571: We use the velocity range $-700<v<-75$~\kms\ to calculate $L$, avoiding both the
572: narrow system A$'$ and the high-velocity edge of the trough from system A where
573: convolution with the instrumental line spread function may alter the true
574: relative absorption depths in the two lines of a doublet \nocite{gecc99}({Ganguly} {et~al.} 1999).
575: We find a best-fit relative filament width $w=0.0135$,
576: with a 99.994\% (4$\sigma$) probability range of $0.0014<w<0.0430$.
577: We find a best-fit relative cloud radius $r=0.081$,
578: with a 99.994\% (4$\sigma$) probability range of $0.029<r<0.143$.
579: There is no statistically significant difference between the likelihood
580: of the two fits.
581:
582: To convert these to physical sizes, we model the quasar's emission as being
583: from a \nocite{ss73}{Shakura} \& {Sunyaev} (1973) accretion disk with viscosity
584: parameter $\alpha=0.1$ radiating at the Eddington limit. (We discuss the issue
585: of coverage of the quasar's broad emission line region at the end of the
586: section.) For this quasar we estimate $M_{BH}=6.2 \times 10^8$ M$_\odot$ from
587: the second moment of its \ion{Mg}{2} emission line and its 3000~\AA\ continuum
588: luminosity, using the methods of Rafiee et al. (2007, in preparation).
589: For those parameters, 99\% of the continuum emission at rest-frame 1400\,\AA\
590: comes from $r<150R_{Sch}$, where $R_{Sch}=2GM_{BH}/c^2=1.8\times10^{14}$~cm
591: is the Schwarzschild radius of the black hole.
592: Since the relative sizes derived above were referenced to a square, not a
593: circle, we adopt the square that has the same area as a circle with radius
594: $150R_{Sch}$, which has sides of length $l=4.8\times10^{16}$~cm.
595: Thus, we find a best-fit filament width of $w=6.5\times10^{14}$~cm,
596: with a 4$\sigma$ range of $6.7\times10^{13}<w<2.1\times10^{15}$~cm,
597: and a best-fit cloud radius $r=3.9\times10^{15}$~cm, with a 4$\sigma$ range
598: of $1.4\times10^{15}<r<6.9\times10^{15}$~cm.
599:
600: These sizes, small on astronomical scales, suggest an origin for the subunits
601: in the accretion disk for either geometry. A plausible length scale for
602: structures originating in an accretion disk is the scale height $h$ of its
603: atmosphere (Equation 2.28 of \nocite{ss73}{Shakura} \& {Sunyaev} 1973).\footnote{If the accretion disk
604: has a strong magnetic field, the pressure scale height may be a less plausible
605: characteristic length. Numerical simulations of accretion
606: disks do not yet conclusively show if another characteristic scale is produced
607: by magnetohydrodynamic turbulence \nocite{arm04}({Armitage} 2004).}
608: At large radii,
609: $h\simeq 3R^3kT_s/4GM_{BH}m_pz_0$ where $R$ is the distance from the black hole,
610: $T_s$ is the disk surface temperature
611: and $z_0$ is the disk half-thickness.
612: (Though not obvious from the above, $h<z_0$ because the
613: disk surface temperature is lower than its midplane temperature.)
614: In this object, the best-fit filament width equals the scale height $h$ at
615: $r=5500R_{Sch}=9.9\times 10^{17}$~cm
616: and the best-fit cloud radius equals the scale height $h$ at
617: $r=25000R_{Sch}=4.5\times 10^{18}$~cm.
618: The various parameters for our two geometries are summarized in Table 2.
619:
620: Strikingly, the first of those distances from the central source is
621: equal to the distance the absorber must have to cover the
622: emission from the quasar's broad emission line region (BELR).
623: As seen in Figure 4, the line emission in the region of the absorption troughs
624: reaches at most 10\% of the continuum level, and at least system A covers both
625: the continuum emission region and the \ion{Si}{4}/\ion{O}{4}] BELR. In other
626: transitions, both systems at least partially cover the \ion{N}{5} and \ion{C}{4}
627: BELRs, and at least system A covers the \ion{Mg}{2} BELR. Since AGN BELRs are
628: stratified, with lower-ionization gas located farther from the quasar, to be
629: conservative we assume both systems lie exterior to the \ion{Mg}{2} BELR in \jo.
630: We use a relationship between $L_\lambda$(3000\AA) and $R_{\rm BELR,MgII}$
631: derived from reverberation-mapping data (\nocite{pea04}{Peterson} {et~al.} 2004; Rafiee et al. 2007,
632: in preparation) to obtain $R_{\rm BELR,MgII}=9.1 \times 10^{17}\ {\rm cm} =5000
633: R_{Sch}$ for \jo. Given the $\pm$25\% 1$\sigma$ scatter in this relationship,
634: this distance is in excellent agreement with the distance required for
635: filamentary absorber subunits to have widths matching the disk scale height.
636: Of course, the absorber could be located at any $R>R_{\rm BELR,MgII}$,
637: so spherical clouds of size equal to the disk scale height could still
638: match the data if the outflow arises at sufficiently large radii.
639:
640: We have outlined a consistent picture wherein systems A and B,
641: whether they consist of opaque filaments or clouds, are launched from
642: the accretion disk exterior to the \ion{Mg}{2} BELR with a subunit size
643: comparable to the scale height of the accretion disk atmosphere at that radius.
644: As a system accelerates, its typical density will decrease and its typical
645: ionization will increase, explaining the presence of high ionization
646: species in flows arising from a low-ionization emission-line region.
647: When the systems cross our line of sight, they have line-of-sight velocities of
648: $v_{los}=-2000$\,\kms\ for system A and $v_{los}=-3600$\,\kms\ for system B.
649: %comparable to
650: %and larger than the orbital velocities of $2900$\,\kms\ or $1400$\,\kms\
651: %expected at launch radii of 5500$R_{Sch}$ and 25000$R_{Sch}$, respectively.
652: For System A, $|v_{los}|$ is comparable to the $v_{orbital}=2900$\,\kms\
653: expected at its inferred launch radius of 5500$R_{Sch}$.
654: For System B, $|v_{los}|$ is larger than the $v_{orbital}=1400$\,\kms\
655: expected at its inferred launch radius of 25000$R_{Sch}$.
656: The spherical cloud dispersal time would be of order $\sim$110 years for
657: $T\sim 10^4$\,K, so the subunits will not disperse on their own between launch
658: and crossing our line of sight. However, partial shadowing of a subunit will
659: produce differential radiative acceleration of the subunit. Substantial
660: radiative acceleration could thus shorten the subunit lifetimes considerably.
661:
662: One potential complication is that the observed profile of the overlapping
663: trough deviates from the multiplicative prediction (Figure 6, bottom panel)
664: in a manner that is not random on velocity scales larger than $\sim$10\ \kms.
665: However, deviations on such scales should be random if, as expected,
666: the individual subunits have velocity dispersions of that order.
667: Instead, the deviations seem to be coherent on $\sim$100\ \kms\ scales.
668: It may be that the subunits do have velocity widths of that order
669: due to microturbulence (\nocite{bfbk00}{Bottorff} {et~al.} 2000).
670: Another possible explanation is that the outflow consists of filaments
671: wherein the material is accelerated so that its line-of-sight
672: velocity increases by $\sim$100\ \kms\ as it crosses the line of sight
673: (e.g., \nocite{aea99}{Arav} {et~al.} 1999).
674: Deviations from the expected profile should then persist for $\sim$100\ \kms\
675: instead of $\sim$10\ \kms. As compared to a model without line-of-sight
676: acceleration, there could be the same average number of filaments, but the
677: number would change more slowly with velocity (although other effects,
678: such as filaments not being exactly parallel, can affect that as well).
679: Observations of additional overlapping systems would be useful
680: for investigating this issue.
681: %The acceleration needed to produce an observed velocity change $\Delta v$
682: %in a time $t_{cross}=l/v_{transverse}$ is $a=v_{transverse} \Delta v /l$,
683: %where $l$ is the size of the emitting region.
684: %Adopting $v_{transverse}=v_{orbital}$, $\Delta v=100$\ \kms, and
685: %$l=4.8\times 10^{16}$\ cm, we find $a=0.06$ cm/s^2. If flows exhibit
686: %such small accelerations, this scenario would be plausible.
687: %%We defer further investigation to a future paper.
688:
689: We note that \nocite{goo03}{Goodman} (2003) have shown that thin accretion
690: disks without winds will be unstable to self-gravity beyond
691: $r_{Q=1} \simeq 2740 (10^8\alpha l_E^2/M_{BH})^{2/9} R_{Sch}$
692: where $l_E$ is the Eddington ratio;
693: using the parameters adopted herein, \jo\ has $r_{Q=1} \simeq 1100R_{Sch}$.
694: However, removal of angular momentum by a disk wind might help stabilize a thin
695: disk (\S 4.3 of \nocite{goo03}{Goodman} 2003), and there is reason to believe
696: such a process operates in AGN. Reverberation mapping places the BELRs of many
697: AGN at $r>r_{Q=1}$, and there is evidence that BELRs are flattened
698: \nocite{vwb00,sea05,aa05}({Vestergaard}, {Wilkes}, \& {Barthel} 2000; {Smith} {et~al.} 2005; {Aars} {et~al.} 2005)
699: as expected if they are located at the bases of accretion disk winds
700: \nocite{mur95}({Murray} {et~al.} 1995). Furthermore, quasar spectral energy
701: distributions are consistent with marginally gravitationally stable disks
702: extending out to $\sim 10^5R_{Sch}$ \nocite{sg03}({Sirko} \& {Goodman} 2003).
703:
704: Lastly, we note that
705: there is no contradiction in using the continuum source size to derive the scale
706: size of the subunits for an outflow the size of the BELR. This is because the
707: continuum source has a surface brightness $\simeq 2100$ times that of the BELR.
708: That number is the ratio of the continuum flux near 1400\,\AA\ in \jo\ to the
709: \ion{Si}{4}/\ion{O}{4}] flux, which we take to be $\simeq 9$,
710: times the ratio of the areas of the \ion{Si}{4}/\ion{O}{4}]
711: BELR and the 1400\,\AA\ continuum source.\footnote{The size of the
712: \ion{Si}{4}/\ion{O}{4}] BELR has been measured in only three AGN \nocite{pea04}({Peterson} {et~al.} 2004).
713: On average, it is comparable in size to the \ion{C}{4} BELR. We therefore use
714: the relationship between $L_\lambda$(1350\AA) and $R_{\rm BELR,CIV}$ given by
715: \nocite{pee06}{Peterson} {et~al.} (2006) to derive $R_{\rm BELR,SiIV}=4.1 \times 10^{17}$~cm for \jo.}
716: If $N$ subunits of the absorber each cover
717: a fractional area $a$ of the continuum source, $Nx$ subunits of the absorber
718: will each cover a fractional area $a/x$ of the BELR. For large $N$ and small $a$
719: the residual intensity of each region is equal, $i=(1-a)^N\simeq(1-a/x)^{Nx}$,
720: but the variance on $i$ from the BELR will be a factor $\simeq0.1/x$ smaller
721: than the variance on $i$ from the continuum source.
722: Thus, an absorber covering both the continuum source and BELR will have
723: essentially the same residual intensity $i$ and variance $\sigma_i^2$
724: (used to derive the absorber size constraints via Equation 6)
725: as an absorber covering only the continuum source.
726:
727: \subsection{Possible \ion{Si}{4} Line-Locking}
728:
729: We now return to the issue of whether systems A+A$'$ and B can be line-locked.
730: Line-locking occurs when the reduction in line-driving flux caused by the
731: shadow of one system decelerates the other, shadowed system
732: so that two systems end up with the same acceleration (which may be nonzero).
733: The two systems thereafter maintain
734: a constant velocity separation that keeps one system shadowed
735: \nocite{bm89}({Braun} \& {Milgrom} 1989). (However, there is some debate in the literature as to whether
736: line-driven winds are unstable to the growth of shocks \nocite{ocr88,poht04}({Owocki}, {Castor}, \& {Rybicki} 1988; {Pereyra} {et~al.} 2004).
737: If shocks can develop, they could
738: accelerate the wind
739: out of an otherwise stable line-locking configuration.)
740: For line-locking to occur in an accelerating flow,
741: there are two possibilities.
742: System B could have appeared
743: along a sightline linking the continuum source and system A+A$'$
744: at $2.0280<z<2.0476$ and accelerated
745: until it reached $z=2.0280$ and overlapped system A+A$'$ at $z=2.0476$.
746: Alternatively, system A+A$'$ could have appeared at $z>2.0476$ and
747: accelerated until it reached $z=2.02476$ and overlapped system B at $z=2.0280$.
748:
749: The latter scenario can be ruled out because the greatest deceleration of
750: system A+A$'$ would have occurred before it reached $z=2.0476$,
751: when it was shadowed by the deepest part of system B.
752: Instead, the deepest part of system B is observed to be shadowed
753: by the shallowest part of system A.
754: If line-locking was going to occur in this scenario it would have had to set in
755: when the shadowing was greatest (or earlier than that, if less than full
756: shadowing produced sufficient deceleration). If it did not happen then,
757: it will not happen with the observed, lesser amount of shadowing.
758:
759: The former scenario of an accelerating system B which has ended up
760: line-locked is plausible. The observed shadowing as a function of velocity
761: could in principle
762: have halted system B.
763:
764: One requirement of this former scenario, however, is that the narrow absorption
765: at $z=2.0476$ (system A$'$) should not be associated with system A,
766: the broad absorption immediately shortwards of it.
767: If they were associated, then some of the gas in system B at
768: $-350<z<-50$ \kms\ should have come to a halt at 0 \kms, where the shadowing
769: by system A$'$ would have been greater than the current shadowing by system A.
770: System A$'$ must be located farther from the quasar than either system A or B,
771: in this scenario.
772:
773: The optically thickest part of system A
774: is likely at $-650<v<-450$ \kms, where numerous low-ionization species are seen.
775: If any gas in system B was observed at $v<-650$ \kms, that gas
776: would have passed the point of maximum shadowing without becoming line-locked.
777: In fact, no gas in system B is seen at $v<-650$ \kms,
778: consistent with system B being line-locked.
779: One argument against this scenario is that if system B has been halted by the
780: observed shadowing, gas at different velocities in that system has been halted
781: by different amounts of shadowing. For example, gas at $-200$ \kms\
782: has been halted by shadowing of only $\sim$30\% of the continuum, while
783: gas at $-450$ \kms\ has been halted by shadowing of $\sim$95\% of the continuum.
784: It may be more physically plausible to suggest that gas at $-450$ \kms\ has
785: been halted, but that gas at $-200$ \kms\ has not yet been sufficiently shadowed
786: to become line-locked. In other words,
787: in this model system B is in the process of becoming line-locked.
788: However, comparison of the SDSS and UVES spectra shows no evidence for
789: variability in these \ion{Si}{4} troughs. The timescale for velocity changes
790: in this scenario could be longer than 1.4 years (rest-frame),
791: which would rule out line locking in a \nocite{mur95}{Murray} {et~al.} (1995) disk wind
792: in which the entire acceleration phase lasts $\sim1.6$~years,
793: or the line-locking could be occuring in a helical flow, stable on timescales
794: of years, in which
795: our sightline intercepts the flow before the gas becomes line-locked.
796:
797: Finally, note that the \ion{Si}{4} profiles in \jo\ are intriguingly similar to
798: some of the potentially line-locked \ion{N}{5} profiles seen in RX~J1230.8+0115
799: \nocite{gmcs03}({Ganguly} {et~al.} 2003). The $z=0.1058$ system in that object has a profile similar to
800: that of system A+A$'$ (strongest absorption at both ends of the profile),
801: and its $z=0.1093$ system is similar to that of system B (optically thick,
802: with the strongest absorption in the middle of the profile, at a velocity
803: corresponding to the weakest absorption in the other system).
804: Both systems have only about half the velocity widths of those in \jo,
805: however, and the relative velocities of the two systems are reversed ---
806: the weaker, single-peaked absorption profile has the lower outflow velocity.
807: It is also worth noting that the Ly$\alpha$ absorption profile in each object
808: appears to share the same covering factor as the species discussed above,
809: while at least one moderately higher-ionization species in each object
810: (\ion{N}{5} here, and \ion{O}{6} in RX~J1230.8+0115)
811: has a larger covering factor which yields nearly black absorption troughs.
812: Whether these similarities are just coincidences
813: will require data on more candidate line-locking systems.
814: (The line-locked systems in Q~1511+091 studied by \nocite{splh02}{Srianand} {et~al.} (2002) are much more
815: complex, but do not seem to include any profiles similar to those in \jo.)
816:
817: \section{Conclusions}\label{end}
818:
819: We find that the \ion{C}{4} BAL trough at $z=1.87988$ in the spectrum of \jo\
820: ($v=-18400$\ \kms\ relative to the quasar's rest frame) has likely undergone
821: an acceleration of $a = 0.154 \pm 0.025 \mbox{\ cm\ s}^{-2}$ over a period of
822: 1.39 rest-frame years. This is the largest acceleration yet reported
823: in a BAL trough $\geq$1000\ \kms\ wide.
824:
825: We also derive constraints on the outflow properties of two absorption systems,
826: overlapping and possibly line-locked in \ion{Si}{4},
827: at $z=2.0420$ and $z=2.0254$
828: ($v=-2000$\ \kms\ and $v=-3600$\ \kms\ relative to the quasar, respectively).
829: The overlapping trough in common to both systems indicates that
830: at least one of the systems must consist of individual subunits.
831: This contrasts with results strongly suggesting that the BELR itself consists of
832: a smooth flow, rather than a clumped one (\nocite{lbhf06}{Laor} {et~al.} 2006),
833: but agrees with results for a narrow intrinsic absorber in the gravitational
834: lens RXS J1131$-$1231 (\nocite{schs07}{Sluse} {et~al.} 2007).
835:
836: Assuming identical, opaque subunits, our data are consistent with
837: spherical clouds of radius $r\simeq 3.9\times 10^{15}$~cm
838: or linear filaments of width $w\simeq 6.5\times 10^{14}$~cm.
839: These subunits must be located at or beyond the \ion{Mg}{2} broad emission line
840: region.
841: At that distance, the above filament width is equal to the
842: predicted scale height of the outer atmosphere of a thin accretion disk.
843: Insofar as that is
844: a natural length scale for structures originating in an accretion disk,
845: these observations are evidence that the accretion disk is the source of
846: the absorption systems.
847: %Furthermore, the deviations of the overlapping trough's residual intensity
848: %profile from the product of the unblended troughs' profiles are coherent over
849: %velocity scales of $\sim$100\ \kms. This may indicate that the absorption
850: %systems consist of filaments or `flow tubes' within which the velocity of the
851: %outflowing material increases by $\sim$100\ \kms\ as the material crosses our
852: %line of sight.
853: %
854: It would be useful to obtain high-resolution spectra of additional cases
855: of distinct but overlapping intrinsic absorption troughs in quasar spectra
856: to determine if this case is representative.
857: If so, it would also be worth extending this work's analytic study of
858: the implications of the residual intensity variance to numerical studies
859: including a realistic quasar geometry, a range in absorber sizes
860: and optical depths, etc.
861:
862: \acknowledgments
863: We thank N. Murray for discussions, and the referee for helpful comments.
864: P. B. H. is supported by NSERC, and S. I. S. was supported by an NSERC
865: Undergraduate Summer Research Assistantship.
866: The SDSS and SDSS-II (http://www.sdss.org/) are funded by the Alfred P. Sloan
867: Foundation, the Participating Institutions, the National Science Foundation,
868: the U.S. Department of Energy, NASA, the Japanese Monbukagakusho,
869: the Max Planck Society, and the Higher Education Funding Council for England,
870: and managed by the Astrophysical Research Consortium for the Participating Institutions: American Museum of Natural History, Astrophysical Institute Potsdam, University of Basel, Cambridge University, Case Western Reserve University, University of Chicago, Drexel University, Fermilab, the Institute for Advanced Study, the Japan Participation Group, Johns Hopkins University, the Joint Institute for Nuclear Astrophysics, the Kavli Institute for Particle Astrophysics and Cosmology, the Korean Scientist Group, the Chinese Academy of Sciences,
871: Los Alamos National Laboratory, the Max-Planck-Institute for Astronomy,
872: the Max-Planck-Institute for Astrophysics,
873: New Mexico State University, Ohio State University, University of Pittsburgh, University of Portsmouth, Princeton University, the United States Naval Observatory, and the University of Washington.
874:
875: \appendix
876: %\section{Appendix}
877:
878: Consider the case of an absorber consisting of opaque subunits
879: of a uniform shape.
880: Suppose our line of sight to a quasar's emitting regions is intercepted by $N$
881: of these subunits, randomly distributed transverse to the line of sight.
882: Then the scatter possible in the covering fraction at fixed $N$ due to the
883: random overlap (or lack thereof) of the subunits with each other will depend on
884: the shape of the subunits. To obtain expressions for this variance, we
885: approximate the quasar's emitting regions as a square of uniform surface
886: brightness on the plane of the sky.
887: We do this solely because expressions for the variance
888: have been derived for the case of the unit square covered by two relevant
889: subunit geometries: circles of area $a$ and filaments of unit length and
890: width $a$.
891: We take the first case to represent a true cloud model, and the second to
892: represent a magnetically confined `filament' model.
893:
894: The case of the unit square randomly overlapped by filaments parallel to each
895: other and to two sides of the square, and of unit length and width $a$, is
896: treated by \nocite{gplin}Robbins (1944). The unit square is defined as the set of points
897: \{$0\leq x\leq 1;0 \leq y \leq 1$\}. The filaments that overlap the square
898: are centered at $y=0.5$ and distributed randomly in $x$
899: over $-\frac{a}{2}\leq x\leq 1+\frac{a}{2}$. Because of edge effects, the
900: average area covered by a filament is $p=\frac{a}{1+a}$, and the average
901: area uncovered by $N$ filaments is $i=(1-p)^N$.
902: The variance in the fractional area covered is
903: \begin{eqnarray}
904: %\eqnum{A1}
905: \sigma_{\rm filaments}^2=(1-a)^2(1-2p)^N -(1-p)^{2N}
906: +\frac{2a[(1-p)^{N+1}-(1-a)(1-2p)^{N+1}]}{(N+1)p}\nonumber\\
907: -\frac{2a^2[(1-p)^{N+2}-(1-2p)^{N+2}]}{(N+1)(N+2)p^2}
908: \end{eqnarray}
909: for $a<0.5$.
910:
911: In the case of the unit square randomly overlapped by circles of area $a$,
912: circles that overlap the square are distributed such that their centers
913: are within a distance $r=\sqrt{a/\pi}$ of the unit square.
914: Again the average area uncovered by $N$ circles is given by $i=(1-p)^N$,
915: but in this case $p=\pi r^2/(1+4r+\pi r^2)$.
916: The variance in the fractional area covered can be derived from expressions
917: given by \nocite{galp}{Kendall} \& {Moran} (1963), yielding
918: \begin{eqnarray}
919: %\eqnum{A2}
920: \sigma_{\rm circles}^2 =
921: \left[ \frac{1+4r-\pi r^2}{1+4r+\pi r^2} \right]^N
922: \left(1 -4\pi r^2 +\frac{64}{3}r^3 -8r^4 \right)
923: -\left(\frac{1+4r}{1+4r+\pi r^2}\right)^{2N}\nonumber\\
924: +2 \int_0^{2r}\left[1-\frac{2r^2 (\pi -\cos^{-1}\frac{q}{2r} +\frac{q}{2r}\sin\left(\cos^{-1}\frac{q}{2r}\right)}{1+4r+\pi r^2}\right]^N
925: (\pi q -4q^2 + q^3) ~dq
926: \end{eqnarray}
927: for $a<0.5$. The integral must be evaluated numerically for most $N$.
928:
929: For the same $a$ and $N$, $\sigma_{\rm circles}^2 > \sigma_{\rm filaments}^2$.
930: This can be understood by placing a subunit of either type in the center of the
931: square
932: and considering the probability that a second subunit of the same type
933: will overlap the first.
934: There is an area $2a$ in which a second filament can be placed to have
935: some overlap with the first (filament centers at $0.5-a<x<0.5+a$).
936: There is an area $4a$ in which a second circle can be placed to have
937: some overlap with the first (circles centered within $2\sqrt{a\over\pi}$
938: of $\{0.5,0.5\}$, for an area of $\pi (2\sqrt{a\over\pi})^2 = 4a$).
939: If $a$ is small,
940: the most likely value of $i$
941: is $i=1-2a$ for both geometries, but with circles there is a higher probability
942: of $i>1-2a$ and thus a larger variance.
943:
944: \clearpage
945: \begin{thebibliography}{}
946:
947: \bibitem[{Aars}, {Hough}, {Yu}, {Linick}, {Beyer}, {Vermeulen}, \& {Readhead} 2005]{aa05}
948: {Aars}, C.~E., {Hough}, D.~H., {Yu}, L.~H., {Linick}, J.~P., {Beyer}, P.~J., {Vermeulen}, R.~C., \& {Readhead}, A.~C.~S. 2005, \aj, 130, 23
949:
950: \bibitem[{Arav}, {Korista}, {de Kool}, {Junkkarinen}, \& {Begelman} 1999]{aea99}
951: {Arav}, N., {Korista}, K.~T., {de Kool}, M., {Junkkarinen}, V.~T., \& {Begelman}, M.~C. 1999, \apj, 516, 27
952:
953: \bibitem[{Armitage} 2004]{arm04}
954: {Armitage}, P.~J. 2004, {Theory of Disk Accretion onto Supermassive Black Holes} (ASSL Vol.~308: Supermassive Black Holes in the Distant Universe), 89
955:
956: \bibitem[{Barlow}, {Hamann}, \& {Sargent} 1997]{bhs97}
957: {Barlow}, T., {Hamann}, F., \& {Sargent}, W. 1997, in ASP Conf. Ser. 128: Mass Ejection from Active Galactic Nuclei, 13
958:
959: \bibitem[{Barlow}, {Junkkarinen}, \& {Burbidge} 1989]{bjb89}
960: {Barlow}, T., {Junkkarinen}, V., \& {Burbidge}, E. 1989, \apj, 347, 674
961:
962: \bibitem[{Barlow}, {Junkkarinen}, \& {Burbidge} 1992a]{bjb92}
963: {Barlow}, T., {Junkkarinen}, V., \& {Burbidge}, E. 1992a, in American Astronomical Society Meeting, Vol. 181, 1106
964:
965: \bibitem[{Barlow}, {Junkkarinen}, {Burbidge}, {Weymann}, {Morris}, \& {Korista} 1992b]{bea92}
966: {Barlow}, T., {Junkkarinen}, V., {Burbidge}, E., {Weymann}, R., {Morris}, S., \& {Korista}, K. 1992b, \apj, 397, 81
967:
968: \bibitem[{Barlow} 1994]{bar94}
969: {Barlow}, T.~A. 1994, \pasp, 106, 548
970:
971: \bibitem[{Bottorff}, {Ferland}, {Baldwin}, \& {Korista} 2000]{bfbk00}
972: {Bottorff}, M.~C., {Ferland}, G.~J., {Baldwin}, J., {Korista}, K. 2000,
973: \apj, 542, 644
974:
975: \bibitem[{Braun} \& {Milgrom} 1989]{bm89}
976: {Braun}, E. \& {Milgrom}, M. 1989, \apj, 342, 100
977:
978: \bibitem[{Bromage}, {Boksenberg}, {Clavel}, {Elvius}, {Penston}, {Perola}, {Pettini}, {Snijders}, {Tanzi}, \& {Ulrich} 1985]{bea85}
979: {Bromage}, G., {Boksenberg}, A., {Clavel}, J., {Elvius}, A., {Penston}, M., {Perola}, G., {Pettini}, M., {Snijders}, M., {et al.} 1985, \mnras, 215, 1
980:
981: \bibitem[{Dekker}, {D'Odorico}, {Kaufer}, {Delabre}, \& {Kotzlowski} 2000]{uves}
982: {Dekker}, H., {D'Odorico}, S., {Kaufer}, A., {Delabre}, B., \& {Kotzlowski}, H. 2000, in Proc. SPIE Vol. 4008, Optical and IR Telescope Instrumentation and Detectors, ed. M.~{Iye} \& A.~{Moorwood}, 534
983:
984: \bibitem[{Foltz}, {Weymann}, {Morris}, \& {Turnshek} 1987]{fol87}
985: {Foltz}, C.~B., {Weymann}, R.~J., {Morris}, S.~L., \& {Turnshek}, D.~A. 1987, \apj, 317, 450
986:
987: \bibitem[{Fukugita}, {Ichikawa}, {Gunn}, {Doi}, {Shimasaku}, \& {Schneider} 1996]{fuk96}
988: {Fukugita}, M., {Ichikawa}, T., {Gunn}, J.~E., {Doi}, M., {Shimasaku}, K., \& {Schneider}, D.~P. 1996, \aj, 111, 1748
989:
990: \bibitem[{Gabel}, {Crenshaw}, {Kraemer}, {Brandt}, {George}, {Hamann}, {Kaiser}, {Kaspi}, {Kriss}, {Mathur}, {Mushotzky}, {Nandra}, {Netzer}, {Peterson}, {Shields}, {Turner}, \& {Zheng} 2003]{gea03}
991: {Gabel}, J.~R., {Crenshaw}, D.~M., {Kraemer}, S.~B., {Brandt}, W.~N., {George}, I.~M., {Hamann}, F.~W., {Kaiser}, M.~E., {Kaspi}, S., {et al.} 2003, \apj, 595, 120
992:
993: \bibitem[{Gallagher}, {Brandt}, {Wills}, {Charlton}, {Chartas}, \& {Laor} 2004]{gea04}
994: {Gallagher}, S.~C., {Brandt}, W.~N., {Wills}, B.~J., {Charlton}, J.~C., {Chartas}, G., \& {Laor}, A. 2004, \apj, 603, 425
995:
996: \bibitem[{Gallagher}, {Schmidt}, {Smith}, {Brandt}, {Chartas}, {Hylton}, {Hines}, \& {Brotherton} 2005]{gea05}
997: {Gallagher}, S.~C., {Schmidt}, G.~D., {Smith}, P.~S., {Brandt}, W.~N., {Chartas}, G., {Hylton}, S., {Hines}, D.~C., \& {Brotherton}, M.~S. 2005, \apj, 633, 71
998:
999: \bibitem[{Ganguly}, {Charlton}, \& {Eracleous} 2001]{gce01}
1000: {Ganguly}, R., {Charlton}, J.~C., \& {Eracleous}, M. 2001, \apjl, 556, L7
1001:
1002: \bibitem[{Ganguly}, {Eracleous}, {Charlton}, \& {Churchill} 1999]{gecc99}
1003: {Ganguly}, R., {Eracleous}, M., {Charlton}, J.~C., \& {Churchill}, C.~W. 1999, \aj, 117, 2594
1004:
1005: \bibitem[{Ganguly}, {Masiero}, {Charlton}, \& {Sembach} 2003]{gmcs03}
1006: {Ganguly}, R., {Masiero}, J., {Charlton}, J.~C., \& {Sembach}, K.~R. 2003, \apj, 598, 922
1007:
1008: \bibitem[{Goodman} 2003]{goo03}
1009: {Goodman}, J. 2003, \mnras, 339, 937
1010:
1011: \bibitem[{Gunn}, {Siegmund}, {Mannery}, {Owen}, {Hull}, {Leger}, {Carey}, {Knapp}, {York}, {Boroski}, {Kent}, {Lupton}, {Rockosi}, {Evans}, {Waddell}, {Anderson}, {Annis}, {Barentine}, {Bartoszek}, {Bastian}, {Bracker}, {Brewington}, {Briegel}, {Brinkmann}, {Brown}, {Carr}, {Czarapata}, {Drennan}, {Dombeck}, {Federwitz}, {Gillespie}, {Gonzales}, {Hansen}, {Harvanek}, {Hayes}, {Jordan}, {Kinney}, {Klaene}, {Kleinman}, {Kron}, {Kresinski}, {Lee}, {Limmongkol}, {Lindenmeyer}, {Long}, {Loomis}, {McGehee}, {Mantsch}, {Neilsen}, {Neswold}, {Newman}, {Nitta}, {Peoples}, {Pier}, {Prieto}, {Prosapio}, {Rivetta}, {Schneider}, {Snedden}, \& {Wang} 2006]{gun06}
1012: {Gunn}, J., {Siegmund}, W., {Mannery}, E., {Owen}, R., {Hull}, C., {Leger}, R., {Carey}, L., {Knapp}, G., {et al.} 2006, \aj, 131, 2332
1013:
1014: \bibitem[{Gunn}, {Carr}, {Rockosi}, {Sekiguchi}, {Berry}, {Elms}, {de Haas}, {Ivezi{\'c} }, {Knapp}, {Lupton}, {Pauls}, {Simcoe}, {Hirsch}, {Sanford}, {Wang}, {York}, {Harris}, {Annis}, {Bartozek}, {Boroski}, {Bakken}, {Haldeman}, {Kent}, {Holm}, {Holmgren}, {Petravick}, {Prosapio}, {Rechenmacher}, {Doi}, {Fukugita}, {Shimasaku}, {Okada}, {Hull}, {Siegmund}, {Mannery}, {Blouke}, {Heidtman}, {Schneider}, {Lucinio}, \& {Brinkman} 1998]{gun98}
1015: {Gunn}, J.~E., {Carr}, M., {Rockosi}, C., {Sekiguchi}, M., {Berry}, K., {Elms}, B., {de Haas}, E., {Ivezi{\'c} }, {\v Z}., {et al.} 1998, \aj, 116, 3040
1016:
1017: \bibitem[{Hall}, {Anderson}, {Strauss}, {York}, {Richards}, {Fan}, {Knapp}, {Schneider}, {Vanden Berk}, \& {Geballe} 2002]{sdss123}
1018: {Hall}, P.~B., {Anderson}, S., {Strauss}, M., {York}, D., {Richards}, G., {Fan}, X., {Knapp}, G., {Schneider}, D., {et al.} 2002, \apjs, 141, 267
1019:
1020: \bibitem[{Hall}, {Hutsem{\'e}kers}, {Anderson}, {Brinkmann}, {Fan}, {Schneider}, \& {York} 2003]{sb2}
1021: {Hall}, P.~B., {Hutsem{\'e}kers}, D., {Anderson}, S.~F., {Brinkmann}, J., {Fan}, X., {Schneider}, D.~P., \& {York}, D.~G. 2003, \apj, 593, 189
1022:
1023: \bibitem[{Hamann}, {Barlow}, {Cohen}, {Junkkarinen}, \& {Burbidge} 1997a]{hea97}
1024: {Hamann}, F., {Barlow}, T., {Cohen}, R., {Junkkarinen}, V., \& {Burbidge}, E. 1997a, in Mass Ejection from Active Galactic Nuclei, ed. N. Arav, I. Shlosman, \& R. Weymann (San Francisco: ASP), 19
1025:
1026: \bibitem[{Hamann}, {Barlow}, \& {Junkkarinen} 1997b]{hbj97}
1027: {Hamann}, F., {Barlow}, T., \& {Junkkarinen}, V. 1997b, \apj, 478, 87
1028:
1029: \bibitem[{Hamann}, {Barlow}, {Beaver}, {Burbidge}, {Cohen}, {Junkkarinen}, \& {Lyons} 1995]{hea95}
1030: {Hamann}, F., {Barlow}, T.~A., {Beaver}, E.~A., {Burbidge}, E.~M., {Cohen}, R.~D., {Junkkarinen}, V., \& {Lyons}, R. 1995, \apj, 443, 606
1031:
1032: \bibitem[{Hogg}, {Finkbeiner}, {Schlegel}, \& {Gunn} 2001]{sdss82}
1033: {Hogg}, D., {Finkbeiner}, D., {Schlegel}, D., \& {Gunn}, J. 2001, \aj, 122, 2129
1034:
1035: \bibitem[{Horne} 1986]{hor86}
1036: {Horne}, K. 1986, \pasp, 98, 609
1037:
1038: \bibitem[{Ivezi{\'c}}, {Lupton}, {Schlegel}, {Boroski}, {Adelman-McCarthy}, {Yanny}, {Kent}, {Stoughton}, {Finkbeiner}, {Padmanabhan}, {Rockosi}, {Gunn}, {Knapp}, {Strauss}, {Richards}, {Eisenstein}, {Nicinski}, {Kleinman}, {Krzesinski}, {Newman}, {Snedden}, {Thakar}, {Szalay}, {Munn}, {Smith}, {Tucker}, \& {Lee} 2004]{ive04}
1039: {Ivezi{\'c}}, {\v Z}., {Lupton}, R., {Schlegel}, D., {Boroski}, B., {Adelman-McCarthy}, J., {Yanny}, B., {Kent}, S., {Stoughton}, C., {et al.} 2004, AN, 325, 583
1040:
1041: \bibitem[{Kendall} \& {Moran} 1963]{galp}
1042: {Kendall}, M.~G. \& {Moran}, P. A.~P. 1963, {Geometrical Probability} (New York: Hafner), 112
1043:
1044: \bibitem[{Koratkar}, {Goad}, {O'Brien}, {Salamanca}, {Wanders}, {Axon}, {Crenshaw}, {Robinson}, {Korista}, {Rodriguez-Pascual}, {Horne}, {Blackwell}, {Carini}, {England}, {Perez}, {Pitts}, {Rawley}, {Reichert}, {Shrader}, \& {Wamsteker} 1996]{kea96}
1045: {Koratkar}, A., {Goad}, M., {O'Brien}, P., {Salamanca}, I., {Wanders}, I., {Axon}, D., {Crenshaw}, D., {Robinson}, A., {et al.} 1996, \apj, 470, 378
1046:
1047: \bibitem[{Laor}, {Barth}, {Ho}, \& {Filippenko} 2006]{lbhf06}
1048: {Laor}, A., {Barth}, A., {Ho}, L., \& {Filippenko}, A. 2006, \apj, 636, 83
1049:
1050: \bibitem[{Leighly}, {Casebeer}, {Hamann}, \& {Grupe} 2005]{lchg05}
1051: {Leighly}, K.~M., {Casebeer}, D.~A., {Hamann}, F., \& {Grupe}, D. 2005, in Bulletin of the American Astronomical Society, 1184
1052:
1053: \bibitem[{Ma} 2002]{ma02}
1054: {Ma}, F. 2002, \mnras, 335, L99
1055:
1056: \bibitem[{Michalitsianos}, {Oliversen}, \& {Nichols} 1996]{mon96}
1057: {Michalitsianos}, A.~G., {Oliversen}, R.~J., \& {Nichols}, J. 1996, \apj, 461, 593
1058:
1059: \bibitem[{Misawa}, {Eracleous}, {Charlton}, \& {Tajitsu} 2005]{mect05}
1060: {Misawa}, T., {Eracleous}, M., {Charlton}, J.~C., \& {Tajitsu}, A. 2005, \apj, 629, 115
1061:
1062: \bibitem[{Murray}, {Chiang}, {Grossman}, \& {Voit} 1995]{mur95}
1063: {Murray}, N., {Chiang}, J., {Grossman}, S., \& {Voit}, G. 1995, \apj, 451, 498
1064:
1065: \bibitem[{Owocki}, {Castor}, \& {Rybicki} 1988]{ocr88}
1066: {Owocki}, S.~P., {Castor}, J.~I., \& {Rybicki}, G.~B. 1988, \apj, 335, 914
1067:
1068: \bibitem[{Pereyra}, {Owocki}, {Hillier}, \& {Turnshek} 2004]{poht04}
1069: {Pereyra}, N.~A., {Owocki}, S.~P., {Hillier}, D.~J., \& {Turnshek}, D.~A. 2004, \apj, 608, 454
1070:
1071: \bibitem[{Peterson}, {Bentz}, {Desroches}, {Filippenko}, {Ho}, {Kaspi}, {Laor}, {Maoz}, {Moran}, {Pogge}, \& {Quillen} 2006]{pee06}
1072: {Peterson}, B., {Bentz}, M., {Desroches}, L.-B., {Filippenko}, A., {Ho}, L., {Kaspi}, S., {Laor}, A., {Maoz}, D., {et al.} 2006, \apj, 641, 638
1073:
1074: \bibitem[{Peterson}, {Ferrarese}, {Gilbert}, {Kaspi}, {Malkan}, {Maoz}, {Merritt}, {Netzer}, {Onken}, {Pogge}, {Vestergaard}, \& {Wandel} 2004]{pea04}
1075: {Peterson}, B., {Ferrarese}, L., {Gilbert}, K., {Kaspi}, S., {Malkan}, M., {Maoz}, D., {Merritt}, D., {Netzer}, H., {et al.} 2004, \apj, 613, 682
1076:
1077: \bibitem[{Peterson} 1997]{pet97}
1078: {Peterson}, B.~M. 1997, Active Galactic Nuclei (Cambridge: Cambridge University Press), 71--89
1079:
1080: \bibitem[{Pier}, {Munn}, {Hindsley}, {Hennessy}, {Kent}, {Lupton}, \& {Ivezi{\' c}} 2003]{sdss153}
1081: {Pier}, J.~R., {Munn}, J.~A., {Hindsley}, R.~B., {Hennessy}, G.~S., {Kent}, S.~M., {Lupton}, R.~H., \& {Ivezi{\' c}}, {\v Z}. 2003, \aj, 125, 1559
1082:
1083: \bibitem[{Reichard}, {Richards}, {Schneider}, {Hall}, {Tolea}, {Krolik}, {Tsvetanov}, {Vanden Berk}, {York}, {Knapp}, {Gunn}, \& {Brinkmann} 2003]{sdssbalcat}
1084: {Reichard}, T., {Richards}, G., {Schneider}, D., {Hall}, P., {Tolea}, A., {Krolik}, J., {Tsvetanov}, Z., {Vanden Berk}, D., {et al.} 2003, \aj, 125, 1711
1085:
1086: \bibitem[{Richards}, {Fan}, {Newberg}, {Strauss}, {Vanden Berk}, {Schneider}, {Yanny}, {Boucher}, \& {Burles} 2002]{sdssqtarget}
1087: {Richards}, G., {Fan}, X., {Newberg}, H., {Strauss}, M., {Vanden Berk}, D., {Schneider}, D., {Yanny}, B., {Boucher}, A., {et al.} 2002, \aj, 123, 2945
1088:
1089: \bibitem[Robbins 1944]{gplin}
1090: Robbins, H.~E. 1944, Annals of Mathematical Statistics, 15, 70
1091:
1092: \bibitem[{Rupke}, {Veilleux}, \& {Sanders} 2002]{rvs02}
1093: {Rupke}, D.~S., {Veilleux}, S., \& {Sanders}, D.~B. 2002, \apj, 570, 588
1094:
1095: \bibitem[{Rupke}, {Veilleux}, \& {Sanders} 2005]{rvs05}
1096: ---. 2005, \apjs, 160, 87
1097:
1098: \bibitem[{Schneider}, {Hall}, {Richards}, {Vanden Berk}, {Anderson}, {Fan}, {Jester}, {Stoughton}, {Strauss}, {SubbaRao}, {Brandt}, {Gunn}, {Yanny}, {Bahcall}, {Barentine}, {Blanton}, {Boroski}, {Brewington}, {Brinkmann}, {Brunner}, {Csabai}, {Doi}, {Eisenstein}, {Frieman}, {Fukugita}, {Gray}, {Harvanek}, {Heckman}, {Ivezi{\'c}}, {Kent}, {Kleinman}, {Knapp}, {Kron}, {Krzesinski}, {Long}, {Loveday}, {Lupton}, {Margon}, {Munn}, {Neilsen}, {Newberg}, {Newman}, {Nichol}, {Nitta}, {Pier}, {Rockosi}, {Saxe}, {Schlegel}, {Snedden}, {Szalay}, {Thakar}, {Uomoto}, {Voges}, \& {York} 2005]{dr3q}
1099: {Schneider}, D.~P., {Hall}, P.~B., {Richards}, G.~T., {Vanden Berk}, D.~E., {Anderson}, S.~F., {Fan}, X., {Jester}, S., {Stoughton}, C., {et al.} 2005, \aj, 130, 367
1100:
1101: \bibitem[{Schneider}, {Richards}, {Fan}, {Hall}, {Strauss}, {Vanden Berk}, {Gunn}, {Newberg}, \& {Reichard} 2002]{sdssedrq}
1102: {Schneider}, D.~P., {Richards}, G.~T., {Fan}, X., {Hall}, P.~B., {Strauss}, M.~A., {Vanden Berk}, D.~E., {Gunn}, J.~E., {Newberg}, H.~J., {et al.} 2002, \aj, 123, 567
1103:
1104: \bibitem[{Shakura} \& {Sunyaev} 1973]{ss73}
1105: {Shakura}, N.~I. \& {Sunyaev}, R.~A. 1973, \aap, 24, 337
1106:
1107: \bibitem[{Sirko} \& {Goodman} 2003]{sg03}
1108: {Sirko}, E. \& {Goodman}, J. 2003, \mnras, 341, 501
1109:
1110: \bibitem[{Sluse}, {Claeskens}, {Hutsem{\'e}kers}, \& {Surdej} 2007]{schs07}
1111: {Sluse}, D., {Claeskens}, J.-F., {Hutsem{\'e}kers}, D., \& {Surdej}, J., \aap,
1112: in press (astro-ph/0703030)
1113:
1114: \bibitem[{Smith}, {Tucker}, {Kent}, {Richmond}, {Fukugita}, {Ichikawa}, {Ichikawa}, {Jorgensen}, {Uomoto}, {Gunn}, {Hamabe}, {Watanabe}, {Tolea}, {Henden}, {Annis}, {Pier}, {McKay}, {Brinkmann}, {Chen}, {Holtzman}, {Shimasaku}, \& {York} 2002]{sdss105}
1115: {Smith}, J., {Tucker}, D., {Kent}, S., {Richmond}, M., {Fukugita}, M., {Ichikawa}, T., {Ichikawa}, S., {Jorgensen}, A., {et al.} 2002, \aj, 123, 2121
1116:
1117: \bibitem[{Smith}, {Robinson}, {Young}, {Axon}, \& {Corbett} 2005]{sea05}
1118: {Smith}, J.~E., {Robinson}, A., {Young}, S., {Axon}, D.~J., \& {Corbett}, E.~A. 2005, \mnras, 359, 846
1119:
1120: \bibitem[{Smith} \& {Penston} 1988]{sp88}
1121: {Smith}, L.~J. \& {Penston}, M.~V. 1988, \mnras, 235, 551
1122:
1123: \bibitem[{Srianand}, {Petitjean}, {Ledoux}, \& {Hazard} 2002]{splh02}
1124: {Srianand}, R., {Petitjean}, P., {Ledoux}, C., \& {Hazard}, C. 2002, \mnras, 336, 753
1125:
1126: \bibitem[{Trump}, {Hall}, {Reichard}, {Richards}, {Schneider}, {Vanden Berk}, {Knapp}, {Anderson}, {Fan}, {Brinkman}, {Kleinman}, \& {Nitta} 2006]{trump06}
1127: {Trump}, J., {Hall}, P., {Reichard}, T., {Richards}, G., {Schneider}, D., {Vanden Berk}, D., {Knapp}, G., {Anderson}, S., {et al.} 2006, \apjs, 165, 1
1128:
1129: \bibitem[{Turnshek}, {Grillmair}, {Foltz}, \& {Weymann} 1988]{tea88}
1130: {Turnshek}, D.~A., {Grillmair}, C.~J., {Foltz}, C.~B., \& {Weymann}, R.~J. 1988, \apj, 325, 651
1131:
1132: \bibitem[{Vanden Berk}, {Richards}, {Bauer}, {Strauss}, {Schneider}, {Heckman}, {York}, {Hall}, {Fan}, {Knapp}, {Anderson}, {Annis}, {Bahcall}, {Bernardi}, {Briggs}, {Brinkmann}, {Brunner}, {Burles}, {Carey}, {Castander}, {Connolly}, {Crocker}, {Csabai}, {Doi}, {Finkbeiner}, {Friedman}, {Frieman}, {Fukugita}, {Gunn}, {Hennessy}, {Ivezi{\' c}}, {Kent}, {Kunszt}, {Lamb}, {Leger}, {Long}, {Loveday}, {Lupton}, {Meiksin}, {Merelli}, {Munn}, {Newberg}, {Newcomb}, {Nichol}, {Owen}, {Pier}, {Pope}, {Rockosi}, {Schlegel}, {Siegmund}, {Smee}, {Snir}, {Stoughton}, {Stubbs}, {SubbaRao}, {Szalay}, {Szokoly}, {Tremonti}, {Uomoto}, {Waddell}, {Yanny}, \& {Zheng} 2001]{sdss73}
1133: {Vanden Berk}, D.~E., {Richards}, G.~T., {Bauer}, A., {Strauss}, M.~A., {Schneider}, D.~P., {Heckman}, T.~M., {York}, D.~G., {Hall}, P.~B., {et al.} 2001, \aj, 122, 549
1134:
1135: \bibitem[{Vestergaard}, {Wilkes}, \& {Barthel} 2000]{vwb00}
1136: {Vestergaard}, M., {Wilkes}, B.~J., \& {Barthel}, P.~D. 2000, \apjl, 538, L103
1137:
1138: \bibitem[{Vilkoviskij} \& {Irwin} 2001]{vi01}
1139: {Vilkoviskij}, E.~Y. \& {Irwin}, M.~J. 2001, \mnras, 321, 4
1140:
1141: \bibitem[{Voit}, {Shull}, \& {Begelman} 1987]{vsb87}
1142: {Voit}, G.~M., {Shull}, J.~M., \& {Begelman}, M.~C. 1987, \apj, 316, 573
1143:
1144: \bibitem[{Voit}, {Weymann}, \& {Korista} 1993]{vwk93}
1145: {Voit}, G.~M., {Weymann}, R.~J., \& {Korista}, K.~T. 1993, \apj, 413, 95
1146:
1147: \bibitem[{York}, {Adelman}, {Anderson}, {Anderson}, {Annis}, {Bahcall}, {Bakken}, {Barkhouser}, {Bastian}, {Berman}, {Boroski}, {Bracker}, {Briegel}, {Briggs}, {Brinkmann}, {Brunner}, {Burles}, {Carey}, {Carr}, {Castander}, {Chen}, {Colestock}, {Connolly}, {Crocker}, {Csabai}, {Czarapata}, {Davis}, {Doi}, {Dombeck}, {Eisenstein}, {Ellman}, {Elms}, {Evans}, {Fan}, {Federwitz}, {Fiscelli}, {Friedman}, {Frieman}, {Fukugita}, {Gillespie}, {Gunn}, {Gurbani}, {de Haas}, {Haldeman}, {Harris}, {Hayes}, {Heckman}, {Hennessy}, {Hindsley}, {Holm}, {Holmgren}, {Huang}, {Hull}, {Husby}, {Ichikawa}, {Ichikawa}, {Ivezi{\'c}}, {Kent}, {Kim}, {Kinney}, {Klaene}, {Kleinman}, {Kleinman}, {Knapp}, {Korienek}, {Kron}, {Kunszt}, {Lamb}, {Lee}, {Leger}, {Limmongkol}, {Lindenmeyer}, {Long}, {Loomis}, {Loveday}, {Lucinio}, {Lupton}, {MacKinnon}, {Mannery}, {Mantsch}, {Margon}, {McGehee}, {McKay}, {Meiksin}, {Merelli}, {Monet}, {Munn}, {Narayanan}, {Nash}, {Neilsen}, {Neswold}, {Newberg}, {Nichol}, {Nicinski}, {Nonino}, {Okada}, {Okamura}, {Ostriker}, {Owen}, {Pauls}, {Peoples}, {Peterson}, {Petravick}, {Pier}, {Pope}, {Pordes}, {Prosapio}, {Rechenmacher}, {Quinn}, {Richards}, {Richmond}, {Rivetta}, {Rockosi}, {Ruthmansdorfer}, {Sandford}, {Schlegel}, {Schneider}, {Sekiguchi}, {Sergey}, {Shimasaku}, {Siegmund}, {Smee}, {Smith}, {Snedden}, {Stone}, {Stoughton}, {Strauss}, {Stubbs}, {SubbaRao}, {Szalay}, {Szapudi}, {Szokoly}, {Thakar}, {Tremonti}, {Tucker}, {Uomoto}, {Vanden Berk}, {Vogeley}, {Waddell}, {Wang}, {Watanabe}, {Weinberg}, {Yanny}, \& {Yasuda} 2000]{yor00}
1148: {York}, D., {Adelman}, J., {Anderson}, J., {Anderson}, S., {Annis}, J., {Bahcall}, N., {Bakken}, J., {Barkhouser}, R., {et al.} 2000, \aj, 120, 1579
1149:
1150: \end{thebibliography}
1151:
1152: \begin{deluxetable}{ccccrcc}
1153: \tablecaption{SDSS J0242+0049 Spectroscopic Observations and Inferences\label{spec}}
1154: \tablewidth{480pt}
1155: \tablehead{
1156: \colhead{}& \colhead{SDSS}& \colhead{SDSS}& \colhead{Epoch}& \colhead{$\Delta t_{rest}$}& \colhead{\ion{Si}{4}, \ion{C}{4} Shift}& \colhead{\ion{Si}{4}, \ion{C}{4} Shift}\\[0.2ex]
1157: \colhead{Source}& \colhead{Plate}& \colhead{Fiber}& \colhead{in MJD}& \colhead{(days)}& \colhead{vs. MJD 52188}& \colhead{vs. MJD 53619}
1158: }
1159: \startdata % src plate fiber mjd del_t 2188 3619
1160: SDSS~(1)& 408& 576& 51821& $-$80& 0, 0& 1, 4\\
1161: SDSS~(2)& 707& 332& 52177& 36& --- & --- \\
1162: SDSS~(3)& 706& 617& 52199& 43& --- & --- \\
1163: SDSS~Avg.~(2+3)& ---& ---& (52188)& 40& --- & 1, 3\\
1164: SDSS~Avg.~(1+2+3)& ---& ---& (52066)& 0& --- & 1, 3\\
1165: UVES& ---& ---& 53619& 507& 1, 3& --- \\
1166: \enddata
1167: \tablecomments{Epochs are given on the Modified Julian Day (MJD) system.
1168: The {\em rest-frame} time interval $\Delta t_{rest}$
1169: is given relative to MJD 52066.
1170: Velocity shifts of absorption lines are given in SDSS pixels (69 \kms);
1171: the \ion{C}{4} shift is the first number and
1172: the \ion{Si}{4} shift is the second number.
1173: }
1174: \end{deluxetable}
1175:
1176: \begin{deluxetable}{ccccccc}
1177: %\setlength{\tabcolsep} {0.01in}
1178: \tabletypesize{\scriptsize}
1179: %\rotate
1180: \tablecaption{SDSS J0242+0049 Subunit Parameters\label{model}}
1181: %\tablewidth{550pt}
1182: \tablewidth{525pt}
1183: \tablehead{
1184: \colhead{Subunit}& \colhead{Avg. Number}& \colhead{Best-fit Relative}& \colhead{Relative 99.994\%}& \colhead{Best-fit Physical}& \colhead{Physical 99.994\%}& \colhead{Atmospheric}\\[0.2ex]
1185: \colhead{Geometry}& \colhead{of Subunits $\bar{N}$}& \colhead{Width or Radius}& \colhead{Confidence range}& \colhead{Width or Radius}& \colhead{Confidence range (cm)}& \colhead{Scale Height Distance}
1186: }
1187: \startdata
1188: Filaments& $203\pm81$& 0.0135& $0.0014-0.0430$& $6.5\times10^{14}$~cm& $6.7\times10^{13}-2.1\times10^{15}$& $9.9\times10^{17}$~cm = 5500~$R_{Sch}$\\
1189: Spheres& $177\pm71$& 0.081& $0.029-0.143$& $3.9\times10^{15}$~cm& $1.4\times10^{15}-6.9\times10^{15}$& $4.5\times10^{18}$~cm = 25000~$R_{Sch}$\\
1190: \enddata
1191: \tablecomments{The average number of subunits $\bar{N}$ is the number
1192: of subunits responsible for absorption at each pixel, averaged over all pixels.
1193: The total number of
1194: subunits present depends on the unknown velocity width of each subunit.
1195: The atmospheric scale height distance is the distance from the black hole at
1196: which the accretion disk atmospheric scale height equals the best-fit width or
1197: radius of the subunit in question; see \S 4.1. $R_{Sch}$ refers to the
1198: Schwarzschild radius of a black hole with mass $6.2 \times 10^8$ M$_\odot$.
1199: }
1200: \end{deluxetable}
1201:
1202:
1203: \begin{figure}
1204: \epsscale{1.00}
1205: %\plotone{q0242loglamsm.eps}
1206: \plotone{f1.eps}
1207: \caption{VLT UT2 + UVES spectrum of \jo, smoothed by a 1\,\AA\ boxcar filter.}\label{figtot}
1208: \end{figure}
1209:
1210: \begin{figure}
1211: %\plotone{usebroadAbs.eps}
1212: \plotone{f2.eps}
1213: \caption{UVES spectra of
1214: BAL troughs in \jo\ vs. velocity (in \kms) in the $z=1.87988$ frame.
1215: Negative velocities indicate blueshifts and positive velocities indicate
1216: redshifts relative to that frame.
1217: Zero velocity corresponds to the long-wavelength members of doublets,
1218: and dashed vertical lines indicate all components of each transition.
1219: Contaminating narrow absorption lines are present near all troughs,
1220: but especially in those found shortward of the Ly$\alpha$ forest.}
1221: \label{fig:BAL}
1222: \end{figure}
1223:
1224: \begin{figure}
1225: \plotone{f3.eps}
1226: %\plotone{normComp-ave2.eps}
1227: \caption{Comparison of the $z=1.87988$ \ion{C}{4} BAL in \jo\ at the average
1228: SDSS epoch and the UVES epoch.
1229: Negative velocities indicate blueshifts and positive velocities redshifts,
1230: relative to $z=1.87988$.
1231: The solid line is a weighted average of all three SDSS spectra. The dashed
1232: line is the UVES spectrum binned into the same pixels as the SDSS spectra.
1233: Dotted vertical lines indicate the fitting regions used when conducting the
1234: $\chi^{2}$ test.
1235: The top panel compares the unshifted spectra for the \ion{Si}{4} trough,
1236: and the middle panel the unshifted spectra for the \ion{C}{4} trough.
1237: The bottom panel compares the \ion{C}{4} troughs after shifting the average
1238: SDSS spectrum toward shorter wavelengths by 3 pixels.}
1239: \label{accNorm}
1240: \end{figure}
1241:
1242: \begin{figure}
1243: %\plotone{pbhll_SiIV-lambda.eps}
1244: \plotone{f4.eps}
1245: \caption{Two broad, overlapping \ion{Si}{4} doublets in the unnormalized
1246: spectrum of \jo. Line
1247: identifications and redshifts for the different troughs are given on the figure.
1248: There is also narrow \ion{Si}{4} absorption at z=2.0314 which is not marked.
1249: }\label{fig:lambda}
1250: \end{figure}
1251:
1252: \begin{figure}
1253: %\plotone{pbhll_SiIV-norm1.eps}
1254: \plotone{f5.eps}
1255: \caption{Velocity plot of \ion{Si}{4} absorption after normalization by
1256: a fit to the total spectrum (continuum and weak emission lines).}
1257: \label{fig:norm}
1258: \end{figure}
1259:
1260: \begin{figure}
1261: \epsscale{1.25}
1262: %\epsscale{0.975}
1263: %\plotone{q0242si4norm2.eps}
1264: \plotone{f6.eps}
1265: \caption{Fits to the blended \ion{Si}{4} trough. The trough containing blended
1266: absorption from both redshift systems is shown as the solid line in all panels.
1267: The fits are shown as lighter lines with {\em total} error bars that include the
1268: observed errors on the flux in the blended trough, so that at each pixel the
1269: deviation between the actual trough and the fit can be directly compared
1270: to the total accompanying uncertainty.
1271: {\bf Top panel:} all three observed \ion{Si}{4} troughs are overplotted.
1272: The dashed line shows the unblended trough $\lambda$1393 trough, plotted in the $z=2.0280$ frame.
1273: The dot-dashed line shows the unblended trough $\lambda$1402 trough, plotted in the $z=2.0476$ frame.
1274: {\bf Second panel:} the fit and errors shown are for an optically thin
1275: lower-redshift system.
1276: {\bf Third panel:} the fit and errors shown are for an optically thick
1277: lower-redshift system with maximum overlap in covering factor with the
1278: optically thick higher-redshift system.
1279: {\bf Bottom panel:} the fit and errors shown are for the case where each
1280: system's covering fraction describes its fractional coverage of the other
1281: absorption system, so that the residual flux from both optically thick troughs
1282: can be multiplied together to give the residual flux in this blended trough.
1283: }
1284: \label{fig:norm2}
1285: \end{figure}
1286:
1287: \begin{figure}
1288: \epsscale{1.2}
1289: %\plotone{paperfig300.eps}
1290: \plotone{f7.eps}
1291: \caption{An illustration of how observations of optically thick absorption
1292: systems overlapping in velocity space can constrain absorber substructure.
1293: Each square is a schematic depiction of a quasar's emission region
1294: %at the same velocity $v$ relative to the quasar rest frame.
1295: at the same wavelength $\lambda$.
1296: %For simplicity, the quasar's emission region is depicted as a square.
1297: The hatched zones are the areas of the emission region covered by an absorber.
1298: %
1299: The leftmost column depicts absorption by system A only, which is assumed to
1300: produce a normalized residual intensity of $I_A(\lambda)=0.5$ at the wavelength
1301: shown. The value of $I_A(\lambda)$ is the same regardless of whether the
1302: absorption is due
1303: %have a covering factor of $C_A(v)=0.5$ at the velocity shown.
1304: %The value of $C_A(v)$ is the same regardless of whether the absorption is due
1305: to a monolithic flow (top row) or due to randomly placed subunits (bottom row),
1306: here shown as spheres in projection.
1307: %
1308: Similarly, the second column from the left depicts absorption by system B only,
1309: which is also assumed to produce a normalized residual intensity of
1310: $I_B(\lambda)=0.5$ at the wavelength shown.
1311: %assumed to have a covering factor of $C_B(v)=0.5$ at the velocity shown.
1312: %
1313: When systems A and B overlap, monolithic flows can produce any normalized
1314: %covering factor in the range $0.5\leq C_{A+B}(\lambda) \leq 1$ (top row, right)
1315: residual intensity in the range $0\leq I_{A+B}(\lambda) \leq 0.5$ (top row, right)
1316: depending on the specific areas covered by each flow at that wavelength
1317: $\lambda$. On the other hand, flows consisting of many randomly placed
1318: subunits will naturally yield an average value of
1319: $<$$I_{A+B}(\lambda)$$>$$=$$<$$I_A(\lambda)\times I_B(\lambda)$$>$$=0.25$
1320: {\em at every wavelength $\lambda$ where the systems overlap}
1321: (bottom row, right).
1322: %naturally yield $C_{A+B}(v)=0.75$.
1323: Deviations from this average will occur due to statistical fluctuations,
1324: which will be smaller the more subunits there are.
1325: Measuring the deviations thus enables us to constrain the size and
1326: number of the subunits (see \S 4.1.1).
1327: }
1328: \label{fig:paper}
1329: \end{figure}
1330:
1331: \end{document}
1332: