1: \documentclass[11pt,preprint]{aastex}
2: %\documentclass{../../emulateapj}
3: \usepackage{psfig}
4: %\usepackage{emulateapj5}
5:
6: %%%%%%%% Begin custom definitions %%%%%%%%%%%%%
7:
8: \newcounter{species}
9: \def\ion#1#2{\setcounter{species}{#2}#1$\;${\sc\roman{species}}\relax}
10: \def\lion#1#2#3{
11: \setcounter{species}{#2}#1$\;${\sc\roman{species}}$\;\lambda${#3}\relax
12: }
13: \def\llion#1#2#3{
14: \setcounter{species}{#2}#1$\;${\sc\roman{species}}$\;\lambda\lambda${#3}\relax
15: }
16:
17: \def\kms{\ifmmode{~{\rm km~s^{-1}}}\else{~km s$^{-1}$}\fi}
18: \def\lsim{\lower0.3em\hbox{$\,\buildrel <\over\sim\,$}}
19: \def\gsim{\lower0.3em\hbox{$\,\buildrel >\over\sim\,$}}
20: \def\hst{{\it HST}}
21: \def\lam{$\lambda$}
22: \def\llam{$\lambda\lambda$}
23:
24: \makeatletter
25: \newenvironment{inlinefigure}{%
26: \def\@captype{figure}%
27: \noindent\begin{minipage}{0.999\linewidth}\begin{center}}
28: {\end{center}\end{minipage}\smallskip}
29: \makeatother
30:
31: %%%%%%%% End custom definitions %%%%%%%%%%
32:
33: \begin{document}
34:
35: \shorttitle{Variable UV Absorption Lines in Quasars}
36: \shortauthors{Wise et al.}
37:
38: \title{Variability of Narrow, Associated Absorption Lines in
39: Moderate- and Low-Redshift Quasars}
40:
41: \author{John H. Wise, Michael Eracleous, Jane C. Charlton, and Rajib
42: Ganguly\altaffilmark{1}}
43:
44: \affil{Department of Astronomy and Astrophysics, 525 Davey Laboratory,
45: The Pennsylvania State University, University Park, PA 16802}
46:
47: \email{jwise, mce, charlton, ganguly@astro.psu.edu}
48:
49: \altaffiltext{1}{Current address: Space Telescope Science Institute,
50: 3700 San Martin Drive, Baltimore, MD 21218; e-mail {\tt
51: ganguly@stsci.edu}.}
52:
53: %\date{Accepted 2002. Received 2002}
54: %\pagerange{\pageref{firstpage}--\pageref{lastpage}}
55: %\pubyear{2002}
56: %\begin{document}
57: %\maketitle
58: %\label{firstpage}
59:
60: \begin{abstract}
61: We present the results of a search for variability in the equivalent
62: widths (EWs) of narrow, associated ($\vert\Delta v\vert \leq
63: 5,000\kms$) absorption lines found in the UV spectra of $z \leq 1.5$
64: quasars. The goal of this search was to use variability as a means of
65: identifying absorption lines arising in gas that is intrinsic to the
66: quasar central engine. We have compared archival \hst/FOS spectra of
67: quasars with recent spectra obtained as part of our own snapshot
68: survey of the same objects with STIS. The intervals between
69: observations are 4--10 years. We primarily focused on the \ion{C}{4}
70: absorption lines, although we also studied other lines available in
71: the same spectra (e.g., Ly$\alpha$, \ion{N}{5}, \ion{O}{6}). Our main
72: result is that 4 out of 15 quasars, or 4 out of 19 associated
73: absorption systems, contained variable narrow absorption lines, which
74: are indicative of intrinsic absorption. We conclude that a minimum of
75: 21\% of the associated absorption-line systems are variable. Because
76: not all systems will have necessarily varied, this is a lower limit on
77: this fraction and it is consistent with previous estimates based on
78: variability, partial coverage analysis, or statistical arguments. If
79: we interpret the upper limits on the variability time scale as upper
80: limits on the recombination time of the absorber, we constrain the
81: density of the absorber to be $n_e > 3000~{\rm cm}^{-3}$ and its
82: distance from the ionizing source to be $R\lsim 100$~pc. Moreover, we
83: are now able to pick out specific intrinsic absorption-line systems to
84: be followed up with high-dispersion spectroscopy in order to constrain
85: the geometry, location, and physical conditions of the absorber. We
86: briefly illustrate how followup studies can yield such constraints by
87: means of a simulation.
88:
89: \end{abstract}
90:
91: \keywords{galaxies: active -- quasars: absorption lines -- quasars:
92: general -- accretion}
93:
94: \section{Introduction}
95:
96: The spectra of quasars often show a plethora of absorption lines, some
97: of which arise in gas that is somehow associated with the quasar,
98: while others arise in systems that are unrelated to the quasar (e.g.,
99: galaxies along the line of sight at a much lower redshift). A recent
100: summary of the properties of intrinsic absorption lines (and of their
101: nomenclature) can be found in \citet{HaSa04}. The properties of
102: absorbing gas that is intimately associated with the quasar are of
103: particular interest in studies of quasar central engines because they
104: provide constraints and insights on the accretion process that powers
105: the central engine (such gaseous systems are referred to hereafter as
106: ``intrinsic systems'' and the lines they produce as ``intrinsic
107: absorption lines'') . The predominance of {\it blueshifts} among
108: intrinsic absorption lines, suggests that the gas has the form of an
109: outflowing wind, which in turn makes it an important component of the
110: accretion process. More specifically, the wind can be an important
111: mechanism for extracting angular momentum from the accretion flow,
112: allowing accretion to proceed. Thus the study of intrinsic absorption
113: lines can provide information on how common winds are, as well as
114: estimates or limits on the mass outflow rates, which can serve as
115: constraints on models for the accretion flow. Studying the dynamics
116: and conditions in the absorber also allows us to explore its connection to
117: the broad-emission line gas. In addition, intrinsic absorption lines
118: trace the evolution of quasar outflows, and presumably the cosmic
119: evolution in quasar fueling rate \citep[see, for example, the
120: discussion by][]{Ga_etal01}.
121:
122: Before any detailed study of the physical conditions in intrinsic
123: absorption systems can be carried out, one must identify intrinsic
124: absorption lines with certainty, a task that can be rather difficult
125: in practice. In the case of {\it broad} absorption lines (BALs), whose
126: widths exceed 2,000\kms\ by definition and often exceed 10,000\kms\ in
127: practice \citep[e.g.,][]{Tu87,Tu88,We91}, a good case can be made that
128: the absorbing gas has the form of a fast accretion-disk wind
129: \citep[see, for example,][]{Mu95,Pro00}. Similarly, narrow absorption
130: lines (hereafter NALs; those whose widths are small enough that the
131: strong UV resonance doublets, such as \ion{C}{4}, \ion{Si}{4}, and
132: \ion{N}{5}, are resolved, i.e., FWHM$\lsim$300\kms), with $z_{\rm a}
133: \ll z_{\rm e}$ (where $z_{\rm a}$ and $z_{\rm e}$ are the absorption
134: emission redshifts, respectively), or more specifically with
135: velocities of 18,000\kms\ or more, relative to the quasar, are likely
136: to arise in gas that is distant from and not related to the quasar
137: \citep[see][]{We79a}. This is not guaranteed to always be the case,
138: however, since a number of NALs with $z_{\rm a} \ll z_{\rm e}$ have
139: been found to vary, which suggests that the absorber is closely
140: related to the quasar \citep*[e.g., in Q2343+125 and
141: PG~2302+029;][]{Ha97b,Jan02}. Moreover the statistical study of
142: \citet{Ri99} found that $dN/dz$ (the number of absorption-line systems
143: per object per unit redshift) in the range $1.6<z<3.5$ depends on
144: quasar optical and radio properties. More specifically, they find a
145: statistical excess of absorbers in the velocity range $-15,000\kms$ to
146: $-65,000\kms$ in optically luminous quasars ($M_{\rm V} < -27.0$),
147: relative to quasars of lower luminosities.
148:
149: In the velocity range $\Delta v \lsim -18,000\kms$, $dN/dz$ is
150: significantly higher than the expected density of intervening
151: absorbers \citep{We79a}. In fact, at $\Delta v < -5,000\kms$ as many
152: as 2/3 of the absorption systems could be intrinsic to the quasars
153: \citep*[see, for example,][and references therein]{Ald94}. This excess
154: of absorbers could be attributed to a combination of galaxies in a
155: cluster surrounding the quasar and parcels of gas ejected from the
156: quasar in the form of a wind \citep{We79a,AWFJ87}. The NALs of the
157: latter type are particularly interesting and deserving of detailed
158: study in the context of quasar central engines because they appear to
159: be ubiquitous in quasar spectra \citep[e.g.,][]{AWFJ87} and their
160: frequency appears to evolve with redshift \citep{Ga_etal01}. However,
161: judging which particular NALs arise in gas intrinsic to the quasar
162: is rather challenging.
163:
164: This leads us to undertake this study, whose goal is to identify
165: intrinsic NALs in moderate- and low-redshift quasars ($z\leq 1.5$)
166: based on their variability. This is a relatively economical technique
167: discussed by \citet{BaSa97}. We focus on absorption lines that are
168: within 5,000\kms\ of the quasar redshift; such absorption-line systems
169: are conventionally regarded as ``associated'' with the quasar,
170: following \citet{Fo86}. Although many statistical studies have been
171: conducted to determine the frequency of intrinsic absorbers relative
172: to all NAL systems, only a handful of specific absorption systems have
173: been identified as truly intrinsic. Moreover, variability information
174: is sparse. \citet{Ba97} estimated that 30\% of NALs are variable to
175: some unspecified level, while specific variable systems in
176: high-redshift quasars have been studied in detail by \cite{Ha95},
177: \citet{Ha97b}, and more recently by \citet{Na04}.
178:
179: To search for variability among associated NALs in $z\leq 1.5$ quasars
180: we have carried out a snapshot survey with the Space Telescope Imaging
181: Spectrograph (STIS) on the {\it Hubble Space Telescope} (\hst). We
182: chose quasars which had already been observed with the \hst's Faint
183: Object Spectrograph (FOS) and were known to have associated
184: \ion{C}{4}, \ion{N}{5}, or \ion{O}{6} NALs. The specific goals of
185: this survey were to:
186: \\
187: (a) Determine the fraction of variable NALs among
188: the associated NALs observed in low-redshift quasars. This provides a
189: {\em lower limit} on the fraction of associated NALs that are truly
190: intrinsic, since not all intrinsic NALs will have varied while we
191: were observing them. The results obtained for high-redshift quasars,
192: observed from the ground, need not apply at low redshifts, since the
193: frequency of associated NALs evolves with redshift \citep{Ga_etal01}.
194: \\
195: (b) Increase the number of confirmed intrinsic absorbers so that more
196: detailed followup studies can be carried out on {\it specific} systems.
197:
198: We describe the observations and data analysis methods in \S2 while in
199: \S3 we present the results. In \S4 we discuss the implications of NAL
200: variability, we comment on individual objects, and suggest future,
201: followup work on NAL quasars.
202:
203: \section{Observations and Data Analysis}\label{S_obs}
204:
205: A sample of 15 quasars with associated NALs were surveyed for
206: variability primarily in the following UV absorption lines:
207: \llion{O}{6}{1032,1038}, \lion{H}{1}{1215}, \llion{N}{5}{1239,1243},
208: and \llion{C}{4}{1548,1551}, but also in the \llion{Si}{2}{1190,1193},
209: \lion{Si}{2}{1260}, and \lion{Si}{3}{1207} lines. The journal of
210: observations is given in Table~\ref{tab_obs}, which also includes
211: information for both the first-epoch FOS observations and the
212: second-epoch STIS observations. We have paid particular attention to
213: collecting accurate redshifts from the literature, since these are
214: important in determining the velocities of absorption lines relative
215: to the rest-frame of the quasar, and whether a specific absorption
216: system is blueshifted or redshifted relative to the quasar. Thus, we
217: have examined the original redshift reports for all of our target
218: quasars and considered their uncertainties before adopting the
219: reported values. In \S\ref{S_indiv}, we give a brief discussion of our
220: adopted redshift for specific objects. In Table~\ref{tab_obs}, we
221: list the redshifts we have adopted along with their uncertainties,
222: where available, and the source of the information. In cases where an
223: uncertainty is not quoted, we expect that the uncertainty is of order
224: a few in the last decimal place given. An uncertainty in the redshift,
225: $\delta z$, translates into an uncertainty in the velocity of an
226: absorption system relative to the quasar rest-frame of $\delta v\sim
227: c\;\delta z/(1+z) < c\;\delta z$. Redshift uncertainties of $\delta z
228: \lsim 10^{-4}$ are comparable to redshift errors arising from
229: heliocentric corrections; since the papers reporting the redshifts
230: rarely report the application of heliocentric corrections, an
231: additional uncertainty of up to 30\kms, may have to be taken into
232: account.
233:
234: As Table~\ref{tab_obs} shows, not all desirable lines were observed in
235: all objects, although the \ion{C}{4} line (the primary target) was
236: covered in the spectra of most objects. The first-epoch FOS spectra
237: of these objects (through the G190H and G270H gratings) were collected
238: in the early to mid 1990's and were kindly provided, fully and
239: uniformly reduced, by Sophia Kirhakos and Buell Jannuzi
240: \citep[see][]{Sc93}. A large fraction of them were part of the \hst\
241: Quasar Absorption Line Key Project (hereafter, KP). Part of the
242: reduction process was to correct the absolute wavelength scale; we
243: have checked these corrections by comparing the observed wavelengths
244: of securely-identified Galactic lines with their expected values and
245: found that the agreement to be better than 0.1~\AA\ half of the time
246: and never worse than 0.27~\AA. In our followup snapshot program, the
247: quasars were observed with STIS using the G230L grating and a
248: $52\arcsec\times 0.\!\!\arcsec2$ slit. This grating obtains a
249: resolution of $R \sim 1000$ and a dispersion of 1.58~\AA\ per pixel
250: with two pixels per resolution element \citep{Proffitt02}. This
251: spectral resolution corresponds to a velocity resolution of
252: 300--600\kms\ in the STIS G230L spectral range, compared to a
253: resolution of about 230\kms\ in the FOS spectra with four overlapping
254: pixels per resolution element \citep{Keyes95}. The spectral range for
255: the STIS G230L is 1570--3180~\AA, whereas the FOS G190H and G270H
256: spectral ranges are 1573--2330~\AA\ and 2221--3301~\AA, respectively.
257: These gratings from different instruments conveniently overlap so that
258: a variability analysis can be conducted on the chosen objects.
259:
260: To reduce the STIS data, we used the {\sc calstis} pipeline
261: \citep{Brown02} in the Space Telescope Science Data Analysis System,
262: within the Image Reduction and Analysis Facility
263: (IRAF/STSDAS)\footnote{IRAF is distributed by the National Optical
264: Astronomy Observatories, which are operated by the Association of
265: Universities for Research in Astronomy, Inc., under cooperative
266: agreement with the National Science Foundation.}. We did not
267: attempt the correct the zero-point of the STIS wavelength scale,
268: although we did determine its shift by comparing the wavelengths of
269: Galactic lines with those of their counterparts in FOS spectra; we
270: found that the STIS spectra were shifted by an average of 1.5~\AA\
271: (with a dispersion of 0.5~\AA). Normalized spectra were created by
272: dividing the calibrated spectra with an {\it effective} continuum,
273: which includes both the true continuum and the broad emission lines.
274: The continuum fits were created by selecting points along the
275: spectrum, and fitting a cubic spline to those points. We manually
276: selected the points so that the fit would effectively interpolate
277: across the NALs while conforming to the spectrum unaffected by the
278: NALs. We made no effort to optimize the continuum fit at the ends of
279: the spectra where the signal-to-noise ratio (hereafter $S/N$) was
280: low. In some cases, the absorption lines are broad and/or strong
281: enough that they irrecoverably distort the profiles of the emission
282: lines on which they are superposed. In such cases the placement of the
283: continuum can constitute a significant source of uncertainty in the
284: measured EW. To include this uncertainty in our analysis, we used
285: two extreme continuum fits, to the FOS and to the STIS spectra, in
286: addition to the best fit. The line identification procedure and EW
287: measurement were repeated for each of the fits. An example of two
288: extreme continuum fits to the \ion{C}{4} region of the STIS spectrum
289: of PG~1309+355 is shown in Figure~\ref{fig_cont}. In
290: Figure~\ref{fig_spec} we show the STIS spectra of all objects with the
291: best continuum fit superposed as a solid line. The extreme continuum
292: fits are also shown as dotted lines, but these are not readily
293: discernible unless the differ considerably from the best fit.
294:
295: After normalizing the spectra by dividing by the continuum fit, a
296: two-step process was carried out to detect and measure absorption
297: lines. First, the unresolved line method \citep{Sc93}, used in the KP
298: project, was employed to detect lines in the spectra at the $5\sigma$
299: level \citep[i.e., at 5 ``significance levels'' in the formalism of
300: Schneider et al. 1993 and ][]{Jan98}. On a second pass, the EWs of the
301: identified lines were measured from the {\it observed} spectra by
302: integrating the data directly. The FOS and STIS spectra were searched
303: independently and the results were compared in the end. We measured
304: all lines in the spectral regions of interest, irrespective of their
305: origin. Many of the targeted doublets that were resolved in the FOS
306: spectra were not resolved in the STIS spectra, therefore we cataloged
307: the two members of the doublet separately in the former case and
308: compared the sum of their EWs with the EW of the corresponding blends
309: in the STIS spectra. Similarly, in cases where lines of interest in
310: the STIS spectra were blended with lines from unrelated systems, we
311: did not attempt to de-blend them but measured the total EWs of the
312: blends and compared them with the sums of EWs of resolved lines in the
313: FOS spectra in the end.
314:
315: A significant number of lines detected in the FOS spectra were not
316: detected in the STIS spectra, which typically had a lower $S/N$ than the
317: FOS spectra. In such cases, we determined and cataloged the local
318: photon noise level and the uncertainty resulting from the placement of
319: the continuum, which can be combined in quadrature to yield an upper
320: limit on the EW of the line in the STIS spectrum.
321:
322: The results of the measurements are listed in Table~\ref{tab_var},
323: where we provide the following information:
324:
325: \begin{description}
326:
327: \item[\it Columns 1--4:] The observed wavelength and observed EW along
328: with the uncertainty in the EW, for lines measured in the STIS
329: spectra. The uncertainty is broken up into two parts, which should
330: be added in quadrature to give the final error bar on the EW: an
331: uncertainty due to Poison noise, $\sigma_{ph}$, and an uncertainty
332: due to a range of possible continuum fits, $\sigma_{cont}$
333: (discussed above and illustrated in Figure~\ref{fig_cont}). In cases
334: where the continuum fit was unambiguous, producing a negligible
335: uncertainty in the EW, we set $\sigma_{cont}=0$. If a line was
336: detected in a FOS spectrum but not in the corresponding STIS
337: spectrum, the EW column gives an upper limit, namely
338: $W_{\lambda}^{\rm max}({\rm STIS})=5\sigma_{ph}({\rm STIS})$.
339:
340:
341: \item[\it Column 5:] A set of flags indicating how blended lines in
342: the STIS spectra are compared with the corresponding resolved lines
343: in the FOS spectra. A ``$\Sigma$'' indicates that the line listed in
344: the STIS column is a known blend and the EW given is the total EW
345: of the blend. In the same row, under the FOS column, we give the sum
346: of the EWs of the individual lines making up the blend, as
347: measured in the FOS spectrum. The rows below the total EW row in
348: the FOS column give the EWs of the individual lines. These lines
349: are identified with increasing index numbers in column 5. The
350: corresponding rows in the STIS column are left blank.
351:
352: \item[\it Columns 6--9:] The same information as in columns 1--4, but
353: for lines measured in the FOS spectra. We note that the there are only
354: two lines detected in STIS spectra but not in FOS spectra; for these
355: lines we provide upper limits, analogous to the STIS upper limits.
356:
357: \item[\it Columns 10 and 11:] The difference in equivalent width
358: between the FOS and STIS spectra: $\Delta W_{\lambda}\equiv
359: W_{\lambda}({\rm FOS}) - W_{\lambda}({\rm STIS})$, and this
360: difference normalized by its error bar, $\Delta
361: W_{\lambda}/\sigma$. The error bar was computed by adding all the
362: relevant error bars in quadrature, i.e., $\sigma^2 =
363: \sigma^2_{ph}({\rm STIS}) + \sigma^2_{cont}({\rm STIS}) +
364: \sigma^2_{ph}({\rm FOS}) + \sigma^2_{cont}({\rm FOS})$. In cases
365: where a line is detected in the FOS spectrum and not in the STIS
366: spectrum, we adopt $W_{\lambda}({\rm STIS})=W^{\rm
367: max}_{\lambda}({\rm STIS})\equiv 5\sigma_{phot}({\rm STIS})$ and
368: {\it vice versa} for a single case of a line detected with STIS but
369: not with FOS. Entries corresponding to variable absorption lines are
370: underlined in column 11 for easy identification.
371:
372: \item[\it Columns 12--15:] The line identification, consisting of the
373: ion, the rest-wavelength of the transition, the difference between
374: the observed and expected wavelength of the line in the observer's
375: frame, $\Delta\lambda$, and the redshift of the absorption-line
376: system, $z_{a}$. The redshift of the system is determined from its
377: Ly$\alpha$ line, or from its strongest line, if Ly$\alpha$ is not
378: available. The wavelength differences of other lines are determined
379: as $\Delta\lambda=(1+z_{a})\lambda_{rest}-\lambda_{obs}({\rm
380: FOS})$. In the case of blended doublets, we give identifications
381: for the constituent lines, rather than for the doublet. Entries
382: corresponding to variable absorption lines are underlined in column
383: 12 for easy identification.
384:
385: \item[\it Column 16:] The velocity offset of an absorption line for
386: cases where this is less than 10,000\kms. A positive velocity
387: offset indicates a redshift of the absorption line relative to the
388: quasar. These velocities are based on wavelengths measured from the
389: FOS spectra whenever possible. Their uncertainties can be estimated
390: by considering the following: (a) typical uncertainties in the
391: absolute wavelength scales of the FOS spectra, which amount to
392: approximately 15\kms, (b) the fact that an uncertainty in the
393: redshift of order $\delta z\sim 1\times 10^{-4}$ results in an
394: uncertainty in velocity of no more than 30\kms, and (c) the
395: possibility that an additional heliocentric uncertainty of up to
396: 30\kms\ may be needed. In summary, in cases where redshift
397: uncertainties are $\delta z \lsim 10^{-4}$, we estimate that
398: uncertainties in velocity offsets are of order 20--30\kms, while in
399: cases where redshifts uncertainties are $\delta z \sim 10^{-3}$, we
400: estimate that uncertainties in velocity offsets are of order
401: 100--200\kms. Entries corresponding to variable absorption lines are
402: underlined for easy identification.
403:
404:
405: \end{description}
406:
407: To check our methodology, we compared the our EW measurements from the
408: FOS spectra with those measured by \citet{Be02} from the same data and
409: found the two sets to be in very good agreement. This comparison was
410: facilitated by the fact that the results of \citet{Be02} are available
411: in a convenient electronic form\footnote{See {\tt
412: http://lithops.as.arizona.edu/$\sim$jill/QuasarSpectra/}.}. We
413: show the results of this comparison graphically in
414: Figure~\ref{fig_comp}, where we plot (a) the EW measured here against
415: the EW measured by \citet{Be02} and (b) the distribution of {\it
416: normalized} differences between our own measurements and those of
417: \citet{Be02}, namely $\Delta W_{\lambda}/\sigma \equiv
418: \left[W_{\lambda}({\rm theirs}) - W_{\lambda}({\rm ours}) \right]/
419: (\sigma^2_{\rm theirs} + \sigma^2_{\rm ours})^{1/2}$. This histogram
420: comprises 247 absorption lines, it appears symmetric about zero and
421: contains five outliers at $|\Delta W_{\lambda}/\sigma|>3$ (one is out
422: of the range of the plot). After examining closely the outlying cases
423: we can attribute all of them to differences in fitting the continuum.
424:
425: \section{Results\label{S_res}}
426:
427: In the FOS spectra of the 15 quasars in our sample we have detected
428: and identified the following absorption lines: 62 associated
429: absorption lines (9 of which are slightly {\it redshifted} relative to
430: the quasar redshift), 56 non-associated lines, and 102 Galactic lines
431: (doublet members are counted separately in this census). If we assign
432: the identified lines to {\it systems} according to their velocities,
433: we obtain 19 associated and 12 non-associated systems (4 of the
434: associated systems are redshifted relative to the quasar). In
435: addition, we detected 67 unidentified lines in FOS spectra, the vast
436: majority of which are found at wavelengths blue-ward of the Ly$\alpha$
437: emission line in the spectra of the four highest-redshift quasars in
438: our sample. Thus, we attribute these lines to the Ly$\alpha$ forest
439: and related metal-line systems.
440:
441: Closely spaced lines that are resolved in FOS spectra, are sometimes
442: blended into unresolved complexes in the STIS spectra because of the
443: lower spectral resolution and sampling rate of the latter. Moreover,
444: since the $S/N$ of the STIS spectra is lower than that of the FOS
445: spectra, many lines detected in the latter spectra are not detected in
446: the former. These two effects led to us finding 39 associated and 121
447: non-associated lines or complexes with FOS counterparts in the STIS
448: spectra (the complexes are identified by the flags in column 5 of
449: Table~\ref{tab_var}). We also find one associate and one
450: non-associated line in STIS spectra with only upper limits from FOS
451: spectra. Moreover, for 86 lines/complexes in the FOS spectra we were
452: only able to obtain EW upper limits from the STIS spectra. Finally, we
453: also detected 21 lines in regions of the STIS spectra not covered in
454: the FOS spectra.
455:
456: To detect NAL variability, we compared the EW difference between
457: observations with the uncertainty resulting from photon noise and from
458: the placement of the continuum. In particular, we computed the
459: normalized EW difference, $\Delta W_{\lambda}/\sigma$, as defined in
460: the previous section and tabulated it in column 11 of
461: Table~\ref{tab_var}, for all lines or blended complexes detected in
462: both the STIS and FOS spectra (in the case of blended complexes we
463: formed the difference between the EW of the complex from STIS and the
464: sum of EWs of constituent lines from FOS). We show the results of
465: this comparison graphically in Figure~\ref{fig_scatter}, where we plot
466: (a) the STIS and FOS EWs against each other, and (b) the distribution
467: of the normalized EW difference. We divide the lines into two
468: categories, for which we make separate plots: associated lines,
469: defined by $|\Delta v|<5,000$\kms \citep[relative to the quasar
470: redshift; following][]{Fo86}, and non-associated lines (including
471: Galactic and unidentified lines). Included in this figure are the two
472: lines with STIS detections and FOS upper limits, bringing the total
473: number of lines to 40 associated and 122 non-associated. It is
474: noteworthy that out of 122 non-associated lines the difference in EW
475: between the FOS and STIS spectra never exceeds $3\sigma$.
476:
477: The distribution of normalized EW differences in the upper panel of
478: Figure~\ref{fig_scatter}b shows a number of outliers at $|\Delta
479: W_{\lambda}|/\sigma > 3$, which we interpret as variable lines. We
480: find 3 such outliers, when the expectation value, assuming a Gaussian
481: probability distribution, is 0.11. These are the \lion{C}{4}{1549}
482: doublets in PKS~2135--14, MRC~2251--178, and PG~2251+113 ($\Delta
483: W_{\lambda}/\sigma =3.4$, 3.6, and 5.0, respectively). Absorption
484: lines that are deemed to be variable are identified in
485: Table~\ref{tab_var} by underlining the value of $\Delta
486: W_{\lambda}/\sigma$ as well as the line identification and velocity
487: offset.
488:
489: To search for additional variable lines, we compared the 5$\sigma$ EW
490: upper limits of the 86 lines not detected in the STIS spectra with the
491: EWs of their counterparts measured from the FOS spectra. The results
492: of this comparison are illustrated in Figure~\ref{fig_limits}, where
493: we plot the distribution of $[W_{\lambda}({\rm FOS})-W^{\rm
494: max}_{\lambda}({\rm STIS})]/(\sigma^2_{\rm FOS} +\sigma^2_{\rm
495: STIS})^{1/2}$. This is analogous to the distribution of normalized
496: EW differences plotted in Figure~\ref{fig_scatter}, but with
497: $W_{\lambda}({\rm STIS})$ replaced by the upper limit $W^{\rm
498: max}_{\lambda}({\rm STIS})\equiv5\sigma_{ph}({\rm STIS})$. The plot
499: shows one clear outlier with a 3.3$\sigma$ deviation, which is the
500: Ly$\alpha$ line in PG~1718+481 at $\Delta v=+469$\kms, which we take
501: as a variable line.
502:
503: We have scrutinized the EW measurements of apparently variable NALs to
504: make sure that their variability is not caused by systematic errors
505: such as continuum placement or line blends in STIS spectra. We also
506: compared our FOS EW values of the apparently variable lines with those
507: of \citet{Be02} and found them to agree to ``$1\sigma$'' or better in
508: the cases of PG~1718+481, MRC~2251--178, and PG~2251+113. As a case
509: in point, we note that the variability of the \ion{C}{4} doublet of
510: MRC~2251--178 has already been reported and discussed by
511: \citet*{Ga01}. The analysis of this paper was carried out
512: independently of the that paper and yielded a similar result. In the
513: case of PKS~2135--14, our measurement differs from that of
514: \citet{Be02}, but we believe that our own measurement is the one that
515: should be adopted. We discuss this case in detail and present our
516: arguments in \S\ref{S_indiv}. To provide a visual demonstration of
517: variable lines we overlayed the normalized spectra from the FOS and
518: STIS. Before carrying out this exercise, we convolved the FOS spectra
519: with a Gaussian of width equal to one STIS resolution element. To
520: account for the sampling differences between the two instruments, we
521: discarded every other pixel in the FOS spectra to reduce the sampling
522: rate to two pixels per resolution element. We then interpolated the
523: FOS spectra using a cubic spline so that the final wavelength bins
524: matched those of the corresponding STIS spectra. Finally, we aligned
525: the wavelength scales of corresponding spectra by cross-correlating
526: regions containing Galactic absorption lines. The results of this
527: exercise are shown in Figure~\ref{fig_overp}, where we plot segments
528: of the original STIS and FOS spectra as well as the normalized spectra
529: from the two instruments (after smoothing, resampling, and alignment).
530:
531:
532: \section{Discussion and Conclusions}
533:
534: \subsection{Summary of Results and Immediate Implications\label{S_summ}}
535:
536: We can summarize our main observational results as follows:
537:
538: \begin{enumerate}
539:
540: \item
541: We have identified four variable associated absorption lines. None of
542: the non-associated lines were found to vary. The EWs of the variable
543: lines span the entire range of EWs in the absorption-line sample.
544:
545: \item
546: Each of these lines traces a different absorption {\it system},
547: implying that at least 21\% (4/19) of associated systems are
548: intrinsic.
549:
550: \item
551: The variable lines are very close to the quasar emission redshift:
552: four out of the five have $|\Delta v|<500$\kms. Moreover, one of the
553: variable systems is redshifted relative to its host quasar.
554:
555: \end{enumerate}
556:
557: Since not all intrinsic lines will have necessarily varied between
558: observations, the fraction of variable systems represents a lower
559: limit on the fraction that originate in gas in the immediate vicinity
560: of the quasar central engine. Our result is comparable to that of
561: \citet{Na04} who find that at 23\% of the associated NALs systems that
562: they have monitored from the ground in $z\sim 2$ quasars are variable.
563: Moreover, \citet{Ba97} report a similar fraction of variable NALs
564: ($\sim 30$\%) in higher-redshift ($z\gsim 2.5$) quasars. This
565: estimate of the fraction of intrinsic NAL systems is also compatible
566: with the estimate of 50\% for $z_{\rm e} \approx 2$ quasars by
567: \citet{Ga99}, who used partial coverage as their diagnostic tool. We
568: do note the caveat that the studies of \citet{Ga99} and \citet{Na04}
569: were based on a small number of objects (6 in the former and 8 in the
570: latter), therefore their exact statistical results are somewhat
571: uncertain.
572:
573: The variability we observe could result from one of the following two
574: processes.
575:
576: \begin{enumerate}
577:
578: \item{\em Fluctuations of the Quasar Light. --} Fluctuations of the
579: quasar's ionizing continuum induce the ionization structure of the
580: gas to vary. One can test this hypothesis by studying the
581: variability of lines from ions with different ionization potentials,
582: as we discuss further below. Absorption features should have a
583: delayed reaction to the variations of the quasar's ionizing
584: radiation due to the finite recombination time of the absorbing gas.
585: In our study and in the work of \citet{Ha95}, \citet*{Ald97}, and
586: \citet{Na04}, NAL variability time scales are typically about 3--5
587: years in the quasar rest frame. If we take these to be upper limits
588: to the recombination time, we can estimate a lower limit of the
589: electron density of the absorber clouds by these time scales,
590: following \citet{Ha97b}. The recombination time is $\tau_{\rm
591: rec}\sim 1/\alpha_{\rm r} n_{\rm e}$, where $\alpha_{\rm r}$ is
592: the recombination rate coefficient and $n_{\rm e}$ is the electron
593: density. For the \ion{C}{4} ion, $\alpha_{\rm r} = 2.8 \times
594: 10^{-12}~{\rm cm^{-3}}~{\rm s^{-1}}$ \citep{Ar85}, which leads to
595: $n_{\rm e}\gsim 3,000~{\rm cm^{-3}}$. Using this density
596: constraint, we can place limits on the distance of the absorber from
597: the ionizing source. We adopt equation (3) of \citet{Na04} for the
598: distance of the absorber from the ionizing source, which we recast
599: as $R\approx 95\; L_{44}^{1/2}\; n_4^{-1/2}\; U_{-2}^{-1/2}$~pc,
600: where $U_{-2}$ is the ionization parameter in units of $10^{-2}$,
601: $L_{44}$ is the bolometric luminosity of the quasar in units of
602: $10^{44}$~erg~s$^{-1}$, and $n_4$ is the density in units of
603: $10^4~{\rm cm^{-3}}$. Following \citet{Na04}, we estimate the
604: bolometric luminosity as $L_{\rm bol}\approx 4.4\lambda L_{\lambda}$
605: at $\lambda=1450$~\AA; we obtain $L_{44}\approx 1$ for the quasars
606: with variable \ion{C}{4} lines. Finally, if we assume the optimal
607: ionization parameter, $U\approx 0.02$, from the models of
608: \citet{Ha95} and use the above density constraint, we obtain $R\lsim
609: 100$~pc.
610:
611: \item{\em Bulk or Internal Motion of the Absorber. --} Bulk motion of
612: the absorber across the quasar line of sight will cause the
613: absorption features to vary for one or both of the following
614: reasons: (a) a change in the column density of the absorber, and/or
615: (b) a change in the coverage fraction in the case of a patchy
616: absorber that covers the background source only
617: partially. Similarly, density changes along the line of sight to the
618: continuum source (resulting form the passage of pressure waves, for
619: example) would have the same effect for the same reasons. The
620: observed variability time scales could thus be equally well
621: interpreted in this context. However, for a unique interpretation,
622: we need more information about the location and physical conditions
623: of the absorber. As a specific example, we consider a scenario in
624: which the absorber is a small parcel of gas in the broad
625: emission-line region (BELR). Assuming the continuum source is the
626: UV-emitting region of the inner accretion disk, its size is of order
627: $D_{\rm cont}\sim 10\; GM/c^2 = 1.5\times 10^{14}\; M_8$~cm, where
628: $M_8$ is the mass of the black hole in units of $10^8~{\rm
629: M}_{\odot}$. If a parcel of gas crosses the cylinder of sight to
630: this region at the dynamical speed of the BELR, $v_{\rm dyn}\sim
631: (GM/r)^{1/2}\sim 3,600\; M_8^{1/2}\; r_{17}^{-1/2}$\kms\ \citep[where
632: $r_{17}$ is the radius of the BELR in units of $10^{17}$~cm;
633: see][]{Kaspi00}, the variability time scale will be of order
634: 5~days. If, on the other hand, a pressure wave crosses the cylinder
635: of sight at the speed of sound ($c_{\rm s} \approx 10\;
636: T_4^{1/2}$\kms, with $T_4$ the temperature in units of $10^4$~K),
637: the variability time scale will be of order 5~years. If we assume
638: that the absorber is further away from the quasar central engine
639: (e.g., in the narrow-emission line region or in the host galaxy)
640: the above estimates can be modified accordingly.
641:
642: \end{enumerate}
643:
644: The moral of the above discussion is that the origin of
645: absorption-line variability can be diagnosed with the right
646: observations. In particular, if the variability time scales are
647: constrained to be on the order of a few months or less, internal
648: motions (i.e., waves) in the absorber will be called into question.
649: Such stringent constraints have, in fact, been obtained for some
650: high-redshift quasars \citep[see][and references therein]{Na04}.
651: Another test can be performed using high-dispersion spectra covering
652: transition from a wide range of ionization states, as we discuss in
653: detail in \S\ref{S_fut}, below.
654:
655: \subsection{Discussion of Individual Objects\label{S_indiv}}
656:
657: \begin{description}
658:
659: \item{\em EX 0302--223. --} The redshift quoted for this object in
660: quasar catalogs is 1.400. This value was taken from \citet*{Cha81}
661: and is incorrect. The correct redshift is 1.409 and is reported by
662: \citet*{Marg85}, who also note that the former value had a
663: typographical error. The difference in the two redshifts translates
664: into a velocity difference of about 1100\kms, which would have led
665: to a significant error in the velocities of Table~\ref{tab_var}. In
666: fact, the absorption system at $\Delta v \approx -450$\kms\ would
667: have appeared at a positive velocity if the former value of the
668: redshift were used.
669:
670: \item{\em QSO J0909--095. --} Although the \ion{C}{4} doublet is
671: detected in the FOS spectrum, the poor $S/N$ and lower sampling of
672: the STIS spectrum lead to a large and uninteresting upper limit to
673: the EW at the second epoch. We do, however, detect the Ly$\alpha$
674: line in the STIS spectrum. The redshift of this object comes from a
675: low-resolution spectrum by \citet{Kne98} in which no narrow,
676: forbidden lines were detected. Therefore, it is unusually uncertain
677: and is quoted only to two decimal places.
678:
679: \item{\em QSO 0957+561A. --} A damped Ly$\alpha$ absorber (DLA) at
680: $z=1.391$ along this line of sight produces many NALs \citep[see,
681: for example,][]{Dol00}. As a result, continuum fitting is uncertain
682: throughout the spectrum, which restricts our conclusions about the
683: nature of the associated absorber. This object is the prototypical
684: gravitationally-lensed quasar, discovered by \citet*{Wal79}. It is a
685: radio-loud quasar according to its 5~GHz power \citep[$P_{\rm 5\;
686: GHz} =6\times 10^{25}~{\rm W~Hz^{-1}}$ based on the flux reported
687: by][]{Has81}. The redshift reported in quasar catalogs is 1.4136,
688: which is the value measured from the broad \ion{Mg}{2} emission line
689: by \citet{We79b}. We believe that this value does not correctly
690: reflect the redshift of the quasar for the following reasons. There
691: are several measurements of the redshift based on the broad
692: \ion{C}{4}, \ion{C}{3}] and \ion{Mg}{2} UV emission lines
693: \citep{Wal79,We79b,Wil80,You81}. Measurements from the same emission
694: line by different authors agree with each other but redshifts from
695: the \ion{C}{4} and \ion{C}{3}] lines are systematically lower than
696: the redshift from the \ion{Mg}{2} line by about $3\times
697: 10^{-3}$. This is consistent with a well known trend in radio-loud
698: quasars, in which the centroid of the broad \ion{C}{4} line is at
699: the same redshift as the narrow, forbidden lines, while the the
700: centroids of the broad \ion{Mg}{2} and H$\beta$ lines are slightly
701: redshifted \citep[see, for example,][]{Bro94,Mar96}. Therefore, we
702: believe that the redshift obtained from the \ion{C}{4} emission line
703: is more likely to represent the systemic redshift of the quasar
704: and we adopt the value of 1.4093
705: from \cite{You81}. If the former redshift were adopted, the
706: velocities listed in the last column of Table~\ref{tab_var} would
707: increase by 510\kms.
708:
709: \item{\em PKS 2135--14. --} The \ion{C}{4} absorption line of this
710: object has been the subject of detailed study by several authors.
711: It is an interesting case because the absorption lines are {\it
712: redshifted} relative to the peaks of the broad emission lines and
713: relative to the systemic redshift, as defined by the narrow,
714: forbidden lines. In particular, \citet{BK83} compared {\it IUE}
715: spectra taken approximately 2.6 years apart and found a 25\% change
716: in EW (from 2.0~\AA\ in 1979 to 2.5~\AA\ in 1981) as well as a shift
717: of the absorption-line centroid by 300~\kms\ between the two
718: epochs. We find a significant change in EW between the STIS and FOS
719: observations (from 4.7~\AA\ in 1992 to 3.2~\AA\ in 2000; see
720: illustration in Fig.~\ref{fig_overp}), as well as a significant
721: change between the {\it HST} (1992 and 2000) and {\it IUE} (1979 and
722: 1981) observations. However, the EW we measure from the FOS
723: spectrum is higher than the values measured by \citet{Be02} and
724: \citet{Ha97c} (4.7~\AA\ by us {\it vs} 3.1~\AA\ and 3.6~\AA\ by
725: them). The difference is a result of the placement of the {\it
726: effective} continuum (i.e., the emission line profile). We have
727: experimented with several different fits to the peak and blue side
728: of the \ion{C}{4} emission-line profile including using the profile
729: of the \ion{C}{3}]$\;\lambda$1909 line as a template. In
730: Figure~\ref{fig_pksfits} we show the \ion{C}{4} profile of this
731: object as observed with the FOS, with our two extreme continuum fits
732: superposed as smooth, solid lines. The fit based on the \ion{C}{3}]
733: profile falls between the two extremes and is the one we adopt as
734: the optimal. We obtained the EW by integrating the normalized
735: spectrum directly between the two extreme points where the fitted
736: profile meets the observed profile. In contrast, the fit used by
737: \citet{Ha97c} resembles the shape of our lowest acceptable fit but is
738: placed somewhat lower (see their Fig.~3); it suffers from the
739: drawback that the resulting peak of the \ion{C}{4} line falls a few
740: {\AA}ngstrom short-ward of its nominal wavelength. Those authors
741: obtained the EW by fitting the absorption profile with 2 or 4
742: components. The fit used by \citet{Be02} was a simple interpolation
743: over the primary absorption trough (illustrated schematically as a
744: dotted line in Fig.~\ref{fig_pksfits}), which also led to a low
745: value for the EW. \citet{BB86} and \citet{Ha97c} have studied the
746: ionization conditions in this particular absorption system and
747: concluded that the ionization parameter is likely to be high
748: ($U\approx 0.25$), although lower values (as low as $U\approx
749: 3\times 10^{-3}$) could not be ruled out. They discussed a number of
750: possible scenarios for the absorber, including galaxies within the
751: cluster harboring the quasar, the quasar host galaxy itself, and gas
752: intrinsic to the quasar central engine, but they were not able to
753: select one of these scenarios as an obvious favorite. The
754: variability results we present here, especially when combined with
755: the earlier results of \citet{BK83}, make a strong case that this
756: absorber is intrinsic to the quasar central engine. This finding is
757: particularly interesting in view of the fact that this
758: absorption-line system appears to be redshifted relative to the
759: quasar. Such a redshift does not necessarily imply infall of the
760: absorbing gas toward the quasar central engine because other
761: scenarios can be found that produce such a redshifted absorption
762: line. For example, a parcel of gas in a rotating accretion-disk wind
763: passing in front of an {\it extended} continuum source at the center
764: of the disk can easily produce a redshifted absorption line.
765:
766:
767: \end{description}
768:
769: \subsection{Future Prospects\label{S_fut}}
770:
771: The variability of lines from ions of different ionization potentials
772: can be used to distinguish between variations in the ionizing
773: continuum and variations of the column density (due to transverse
774: motion of the absorber). Ideally, high-resolution ($R \gsim 40,000$)
775: spectra should be used for such a test so that the absorption lines
776: are fully resolved. Under these conditions one can determine whether
777: the background source is partially covered and infer the column
778: density of the absorber, thus constraining its location \citep[see,
779: for example][]{BaSa97,Ha97a,Ga99}.
780:
781: To illustrate and evaluate the method, we carried out the following
782: exercise. We used the photoionization code CLOUDY \citep{Fer96} to
783: simulate the the ionization structure of the absorber, assuming that
784: the incident ionizing spectrum is that of \citet{MaFe87}. Input
785: parameters for the models are the total hydrogen density, the
786: ionization parameter, and the total hydrogen column density for each
787: component, assuming Solar abundances and a metallicity of 1/3 the
788: Solar value.
789: % $\log (Z/Z_{\odot}) = 0.5$.
790: Throughout this exercise we assumed that the coverage fraction is
791: unity for all transitions and that it did not change as the column
792: density or ionization parameter changed. The results for simulated
793: profiles are not sensitive to the hydrogen density in the optically
794: thin regime; values from $10^5$ to $10^9~{\rm cm}^{-3}$ give similar
795: results. For our particular example we have assumed that the absorber
796: consists of three kinematic ``components'' at relative velocities
797: $-75$, 0, and $20\kms$, with column densities of $\log N_{\rm H}
798: =19.5$, 20, and 19, and broadening parameters $b = 8$, 12, and
799: 10\kms. The ionization parameter is taken to be $\log U =-2$. In
800: Figure~\ref{fig_synth} we display the synthetic spectra resulting from
801: this exercise. The middle set of panels show a simulated spectrum
802: ($R=23,500;~S/N=20$) of the \ion{C}{4} profile along with a variety of
803: other transitions. The left-hand set of panels of
804: Figure~\ref{fig_synth} demonstrate the effect of bulk motion, which we
805: represent as a decrease of the column density of each absorption
806: component by an order of magnitude. The result is that all of the
807: absorption lines become weaker, regardless of ionization state. The
808: right-hand set of panels demonstrates the effect of increasing the
809: continuum strength and hence the ionization parameter by an order of
810: magnitude. The lines from the lower ionization species are not
811: detected, while the lines from the higher ionization species, such as
812: \ion{N}{5} and \ion{O}{6} become stronger. A decrease in the total
813: column density and an increase in the continuum strength have a
814: similar effect on the \ion{C}{4} profile, but the difference between
815: the two scenarios can be diagnosed if multiple transitions from a
816: variety of ionization states are observed.
817:
818: With the above scientific questions and technical considerations in
819: mind, the next logical step in our systematic study of intrinsic NALs
820: is to target specific quasars whose NALs are demonstrably intrinsic.
821: If the quasars cooperate, we can infer the physical conditions and
822: location of the absorber through repeated observations. Thus we will
823: be able to assess the role of the NAL gas in the overall accretion
824: flow. Alternatively, significant progress can also be made by single
825: epoch observations of intrinsic (and other, associated) NALs at high
826: spectral resolution. Using high-resolution spectra that encompass
827: transitions from a wide range of ionization states, we can determine
828: the ionization structure of the absorber and constrain its distance
829: from the ionizing source.
830:
831: \acknowledgments
832:
833: This work was supported by grant HST-GO-08681.01-A from the Space
834: Telescope Science Institute, which is operated by the Association of
835: Universities for Research in Astronomy, Inc., under NASA contract
836: NAS5-26555. We also acknowledge support from NASA grant NAG5-10817.
837: We have made use of the NASA/IPAC Extragalactic Database (NED) which
838: is operated by the Jet Propulsion Laboratory, California Institute of
839: Technology, under contract with the NASA. We thank the anonymous
840: referee for many valuable technical comments. We are grateful to Sofia
841: Kirhakos and Buell Jannuzi for providing us with the uniformly and
842: fully reduced spectra of quasars observed by the {\it HST}/FOS. We
843: also thank Fred Hamann for useful discussions and especially for
844: reminding us of the importance of accurate redshifts.
845:
846: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
847: %%% R E F E R E N C E S
848: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
849:
850: %\clearpage
851: \begin{thebibliography}{}
852:
853: \bibitem[\protect\citeauthoryear{Aldcroft, Bechtold, \&
854: Elvis}{Aldcroft et al.}{1994}]{Ald94} Aldcroft, T. L., Bechtold, J.,
855: \& Elvis, M. 1994, ApJS, 93, 1
856:
857: \bibitem[\protect\citeauthoryear{Aldcroft, Bechtold, \&
858: Foltz}{Aldcroft et al.}{1997}]{Ald97}Aldcroft, T., Bechtold, J., \&
859: Foltz, C. 1997, in Mass Ejection from Active Galactic Nuclei,
860: eds. N. Arav \& R. J. Weymann, ASP Conf. Ser. 128 (San Francisco:
861: ASP), 25
862:
863: \bibitem[\protect\citeauthoryear{Anderson et al.}{1987}]{AWFJ87}
864: Anderson, S. F., Weymann, R. J., Foltz, C. B., \& Chaffee, F. H., Jr.
865: 1987, AJ, 94, 278
866:
867: \bibitem[\protect\citeauthoryear{Arnaud \& Rothenflug}{1985}]{Ar85}
868: Arnaud, M. \& Rothenflug, R. 1985, A\&AS, 60, 425
869:
870: \bibitem[\protect\citeauthoryear{Bahcall et al.}{1993}]{Bah93}
871: Bahcall, J. N. 1993, ApJS, 87, 1
872:
873: %\bibitem[\protect\citeauthoryear{Bahcall \& Spitzer}{1969}]{Bc69}
874: %Bahcall, J. N. \& Spitzer, L. J. 1969, ApJL, 156, 63
875:
876: \bibitem[\protect\citeauthoryear{Barlow}{1993}]{Ba93} Barlow, T. A. 1993,
877: Ph.D. Thesis, University of California, San Diego
878:
879: \bibitem[\protect\citeauthoryear{Barlow}{1995}]{Ba95} Barlow,
880: T. A. 1995, BAAS, 27, 872
881:
882: \bibitem[\protect\citeauthoryear{Barlow \& Sargent}{1997}]{BaSa97}
883: Barlow, T. A. \& Sargent, W. L. W. 1997, AJ, 113, 136
884:
885: \bibitem[\protect\citeauthoryear{Barlow, Hamann, \& Sargent}{Barlow et
886: al.}{1997}]{Ba97} Barlow, T. A., Hamann, F., \& Sargent, W. L. W. 1997, in
887: Mass Ejection from Active Galactic Nuclei, eds. N. Arav \&
888: R. J. Weymann, ASP Conf. Ser. 128 (San Francisco: ASP), 13
889:
890: \bibitem[\protect\citeauthoryear{Bechtold et al.}{2002}]{Be02}
891: Bechtold, J., Dobrzycki, A., Wilden, B., Morita, M., Scott, J.,
892: Dobrzycka, D., Tran, K.-V., \& Aldcroft, T. L. 2002, ApJS, 140, 143
893:
894: \bibitem[\protect\citeauthoryear{Bergeron \& Boiss\'e}{1986}]{BB86}
895: Bergeron, J., \& Boisss\'e, P. 1986, \aa, 168, 6
896:
897: \bibitem[\protect\citeauthoryear{Bergeron \& Kunth}{1983}]{BK83}
898: Bergeron, J., \& Kunth, D. 1983, \mnras, 205, 1053
899:
900: \bibitem[\protect\citeauthoryear{Bergeron et al.}{1983}]{Ber83}
901: Bergeron, J., Boksenberg, A., Dennefeld, M.,\& Tarenghi, M.\ 1983,
902: \mnras, 202, 125
903:
904: \bibitem[\protect\citeauthoryear{Boroson \& Green}{1992}]{Bor92}
905: Boroson, T. A. \& Green, R. F. 1992, \apjs, 80, 109
906:
907: \bibitem[\protect\citeauthoryear{Boiss\'e et al.}{1998}]{Boi98}
908: Boiss\'e, P., Le Brun, V., Bergeron, J., \& Deharveng, J.-M. 1998,
909: A\&A, 333, 841
910:
911: \bibitem[\protect\citeauthoryear{Brandt, Laor, \& Wills}{Brandt et
912: al.}{2000}]{Bra00} Brandt, W. N., Laor, W., \& Wills, B. J. 2000, ApJ,
913: 528, 637
914:
915: \bibitem[\protect\citeauthoryear{Brotherton et al.}{1994}]{Bro94}
916: Brotherton, M. S., Wills, B. J., Steidel, C. C., \& Sargent,
917: W. L. W. 1994, \apj, 423, 131
918:
919: \bibitem[\protect\citeauthoryear{Brown et al.}{2002}]{Brown02}
920: Brown, T., et al. 2002, in {\it HST} STIS Data Handbook, Version
921: 4.0, ed. B. Mobasher (Baltimore: STScI)
922:
923: \bibitem[\protect\citeauthoryear{Chanan, Downes, \& Margon}{Chanan et
924: al.}{1981}]{Cha81} Chanan, G.~A., Downes, R.~A., \& Margon, B.\ 1981,
925: \apjl, 243, L5
926:
927: \bibitem[\protect\citeauthoryear{Dolan et al.}{2000}]{Dol00} Dolan,
928: J. F., Michalitsianos, A. G., Nguyen, Q. T., \& Hill, R. J. 2000, ApJ,
929: 539, 111
930:
931: \bibitem[\protect\citeauthoryear{Eracleous \& Halpern}{2004}]{Era04}
932: Eracleous, M. \& Halpern, J. P. 2004, \apjs, 150, in press
933:
934: \bibitem[\protect\citeauthoryear{Foltz et al.}{1986}]{Fo86}
935: Foltz, C. B., Weymann, R. J., Peterson, B. M., Sun, L., Malkan,
936: M. A. \& Chaffee, F. H., Jr. 1986, ApJ, 307, 504
937:
938: \bibitem[\protect\citeauthoryear{Ferland}{1996}]{Fer96} Ferland,
939: G. 1996, Hazy, University of Kentucky Internal Report
940:
941: \bibitem[\protect\citeauthoryear{Ganguly et al.}{1999}]{Ga99} Ganguly,
942: R., Eracleous, M., Charlton, J. C., \& Churchill, C. W. 1999, \aj,
943: 117, 2594
944:
945: \bibitem[\protect\citeauthoryear{Ganguly et al.}{2001a}]{Ga_etal01} Ganguly,
946: R., Bond, N. A., Charlton, J. C., Eracleous, M., Brandt, W. N., \&
947: Churchill, C. W. 2001a, \apj, 549, 123
948:
949: \bibitem[\protect\citeauthoryear{Ganguly, Charlton, \&
950: Eracleous}{Ganguly et al.}{2001b}]{Ga01} Ganguly, R., Charlton, J. C.,
951: \& Eracleous, M. 2001b, \apj, 556, L7
952:
953: \bibitem[\protect\citeauthoryear{Hamann et al.}{1995}]{Ha95} Hamann,
954: F., Barlow, T. A., Beaver, E. A., Burbidge, E. M., Cohen,
955: R. D. Junkkarinen, V. \& Lyons, R. 1995, ApJ, 443, 606
956:
957: \bibitem[\protect\citeauthoryear{Hamann et al.}{1997a}]{Ha97a} Hamann,
958: F., Barlow, T. A., Junkkarinen, V. \& Burbidge, E. M. 1997a, ApJ, 478,
959: 80 % Variable NAL in UM 675
960:
961: \bibitem[\protect\citeauthoryear{Hamann, Barlow, \&
962: Junkkarinen}{Hamann et al.}{1997b}]{Ha97b} Hamann, F., Barlow, T. A.,
963: \& Junkkarinen, V. 1997b, ApJ 487, 87
964: % High-ejection speed NAL in Q2343+125
965:
966: \bibitem[\protect\citeauthoryear{Hamann et al.}{1997c}]{Ha97c} Hamann,
967: F., Beaver, E. A., Cohen, R., Junkkarinen, V., Lyons, R. W., \&
968: Burbidge, E. M. 1997c, ApJ, 488, 155 % NAL in PKS 2135--14
969:
970: \bibitem[\protect\citeauthoryear{Hamann \& Sabra}{2004}]{HaSa04} Hamann,
971: F. \& Sabra, B. 2004, in AGN Physics with the Sloan Digital Sky
972: Survey, eds. G. T. Richards \& P. B. Hall (San Francisco: ASP), in
973: press (astro-ph/0310668).
974:
975: \bibitem[\protect\citeauthoryear{Haschick et al.}{1981}]{Has81}
976: Haschick, A. D., Moran, J. M., Reid, M. J., Davis, M., \& Lilley,
977: A. E., \apj, 243, L57.
978:
979: \bibitem[\protect\citeauthoryear{Mathews \& Ferland}{1987}]{MaFe87}
980: Mathews, W. G. \& Ferland, G. J. 1987, ApJ, 323, 456
981:
982: \bibitem[\protect\citeauthoryear{Jannuzi}{2002}]{Jan02} Jannuzi,
983: B. 2002, in Extragalactic Gas at Low Redshift, eds. J. S. Mulchaey \&
984: J. T. Stocke, ASP Conf. Ser. 254 (San Francisco: ASP), 13
985:
986: \bibitem[\protect\citeauthoryear{Jannuzi et al.}{1998}]{Jan98}
987: Jannuzi, B. et al. 1998, ApJS, 118, 1
988:
989: \bibitem[\protect\citeauthoryear{Kaspi et al.}{2000}]{Kaspi00}
990: Kaspi, S., Smith, P. S., Netzer, H., Maoz, D., Jannuzi, B. T., \&
991: Giveon, U. 2000, \apj, 533, 631
992:
993: \bibitem[\protect\citeauthoryear{Keyes et al.}{1995}]{Keyes95}
994: Keyes, C. D., et al. 1995, FOS Instrument Handbook, Version
995: 6.0 (Baltimore: STScI)
996:
997: \bibitem[\protect\citeauthoryear{Knezek \& Bregman}{1998}]{Kne98}
998: Knezek, P.~M.~\& Bregman, J.~N.\ 1998, \aj, 115, 1737
999:
1000: \bibitem[\protect\citeauthoryear{Le Brun et al.}{1997}]{LeB97} Le Brun
1001: V., Bergeron J., Boiss\'e P., \& Deharveng J. M. 1997, A\&A, 321, 733
1002:
1003: \bibitem[\protect\citeauthoryear{Margon, Downes, \& Chanan}{Margon et
1004: al.}{1985}]{Marg85} Margon, B., Downes, R.~A., \& Chanan, G.~A.\ 1985,
1005: \apjs, 59, 23
1006:
1007: \bibitem[\protect\citeauthoryear{Marziani et al.}{1996}]{Mar96}
1008: Marziani, P., Sulentic, J. W., Dultzin-Hacyan, D., Calvani, M., \&
1009: Moles, M. 1996, ApJS, 104, 37
1010:
1011: \bibitem[\protect\citeauthoryear{Murray et al.}{1995}]{Mu95} Murray,
1012: N., Chiang, J., Grossman, S. A., \& Voit, G. M. 1995 ApJ, 451, 498
1013:
1014: \bibitem[\protect\citeauthoryear{Narayanan et al.}{2004}]{Na04}
1015: Narayanan, D., Hamann, F., Barlow, T., Burbidge, E. M., Cohen, R. D.,
1016: Junkkarinen, V., \& Lyons, R. 2004, \apj, in press (astro-ph/0301668)
1017:
1018: \bibitem[\protect\citeauthoryear{Netzer et al.}{1995}]{Net95} Netzer,
1019: H., Brotherton, M. S., Wills, B. J., Han, M., Wills, D., Baldwin,
1020: J. A., Ferland, G. J., \& Browne, I. W. A. 1995, ApJ, 448, 27
1021:
1022: \bibitem[\protect\citeauthoryear{Proffitt et al.}{2002}]{Proffitt02}
1023: Proffitt, C., et al. 2002, STIS Instrument Handbook, Version 6.0
1024: (Baltimore: STScI)
1025:
1026: \bibitem[\protect\citeauthoryear{Proga, Stone, \& Kallman}{Proga et
1027: al.}{2000}]{Pro00} Proga, D., Stone, J. M., \& Kallman, T. R. 2000,
1028: ApJ, 543, 686
1029:
1030: \bibitem[\protect\citeauthoryear{Richards et al.}{1999}]{Ri99}
1031: Richards, G. T., York, D. G., Yanny, B., Kollgaard, R. I.,
1032: Laurent-Muehleisen, S. A., \& Vanden Berk, D. E. 1999, ApJ, 513, 576
1033:
1034: \bibitem[\protect\citeauthoryear{Schmidt \& Green}{1983}]{Sch83}
1035: Schmidt, M.~\& Green, R.~F.\ 1983, \apj, 269, 352
1036:
1037: \bibitem[\protect\citeauthoryear{Schneider et al.}{1993}]{Sc93}
1038: Schneider, D. P., et al. 1993 ApJS, 87, 45
1039:
1040: \bibitem[\protect\citeauthoryear{Turnshek}{1987}]{Tu87} Turnshek,
1041: D. A. 1987, in QSO Absorption Lines: Probing the Universe
1042: ed. J. C. Blades, D. A. Turnshek, \& C. Norman (Cambridge: Cambridge
1043: Univ. Press), 17
1044:
1045: \bibitem[\protect\citeauthoryear{Turnshek et al.}{1988}]{Tu88}
1046: Turnshek, D. A., Foltz, C. B., Grillmair, C. J., \& Weymann,
1047: R. J. 1988, ApJ, 325, 651
1048:
1049: \bibitem[\protect\citeauthoryear{Walsh, Carswell, \& Weymann}{Walsh et
1050: al.}{1979}]{Wal79} Walsh, D., Carswell, R. F., \& Weymann
1051: R. J. 1979, Nature, 279, 381
1052:
1053: \bibitem[\protect\citeauthoryear{Wampler, Bergeron, \&
1054: Petitjean}{Wampler et al.}{1993}]{Wa93} Wampler, E. J., Bergeron,
1055: J., \& Petitjean, P. 1993, A\&A, 273, 15
1056:
1057: \bibitem[\protect\citeauthoryear{Weymann et al.}{1979a}]{We79a}
1058: Weymann, R. J., Williams, R. E., Peterson, B. M., \& Turnshek,
1059: D. A. 1979, ApJ, 234, 33
1060:
1061: \bibitem[\protect\citeauthoryear{Weymann et al.}{1979b}]{We79b}
1062: Weymann, R. J., Chaffee, F. H., Davis, M., Carleton, N. P., \&
1063: Carswell, R. F. 1979, \apj, 233, L43
1064:
1065: \bibitem[\protect\citeauthoryear{Weymann et al.}{1991}]{We91} Weymann,
1066: R. J., Morris, S. L., Foltz, C. B., \& Hewett, P. C. 1991, ApJ, 373,
1067: 23
1068:
1069: \bibitem[\protect\citeauthoryear{Wills \& Wills}{1976}]{Wil76}Wills,
1070: D.~\& Wills, B.~J.\ 1976, \apjs, 31, 143
1071:
1072: \bibitem[\protect\citeauthoryear{Wills \& Wills}{1980}]{Wil80}Wills,
1073: D.~\& Wills, B.~J.\ 1980, \apj, 238, 1
1074:
1075: \bibitem[\protect\citeauthoryear{Wills et al.}{1999}]{Wil99} Wills,
1076: B. J., Laor, A., Brotherton, M. S., Wills, D., Wilkes, B. J., Ferland,
1077: G. J., \& Shang, Z. 1999, ApJ, 515, L53
1078:
1079: %\bibitem[\protect\citeauthoryear{York et al.}{1986}]{Yo86} York,
1080: %D. G., Dopita, M., Green, R., \& Bechtold, J. 1986, ApJ, 311, 610
1081:
1082: \bibitem[\protect\citeauthoryear{Young et al.}{1981}]{You81} Young,
1083: P., Sargent, W. L. W., Boksenberg, A., \& Oke, J. B. 1981, \apj, 249,
1084: 415
1085:
1086: \end{thebibliography}
1087: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1088: %%% T A B L E S
1089: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1090: %\clearpage
1091: %%% Table 1: Observation Dates
1092: \begin{deluxetable}{lllcccll}
1093: \tabletypesize{\footnotesize}
1094: \tablecolumns{8}
1095: \tablewidth{0in}
1096: \tablecaption{Journal of Observations\label{tab_obs}}
1097: \tablehead{
1098: \colhead{} & \colhead{} & \colhead{} & \multicolumn{2}{c}{Obs. Date (UT mm/dd/yyyy)} & \colhead{}
1099: & \colhead{$z$} & \colhead{Pub} \\
1100: \noalign{\vskip -6pt}
1101: \colhead{} & \colhead{} & \colhead{} & \multicolumn{2}{c}{\hrulefill} & \colhead{Interval} & \colhead{} \\
1102: \colhead{Object} & \colhead{$z$\tablenotemark{a}} &
1103: \colhead{Targeted Lines} & \colhead{FOS} & \colhead{STIS} & \colhead{(years)} &
1104: \colhead{Refs\tablenotemark{b}} & \colhead{Refs\tablenotemark{c}} }
1105: \startdata
1106: EX 0302--223 & 1.409 & \ion{N}{5}, Ly$\alpha$, \ion{O}{6} & 12/07/1995 & 02/22/2001 & 5.22 & 1 & 1 \\ % 1904 days
1107: QSO J0909--095& 0.63 & \ion{C}{4} & 10/29/1996 & 03/04/2001 & 4.35 & 2 & none \\ % 1587
1108: PG 0953+414 & 0.2341(4) & \ion{C}{4} & 11/05/1991\tablenotemark{d} & 01/21/2001 & 9.22 & 3 & 2,3 \\ % 3365
1109: & & & 06/18/1991\tablenotemark{d} & & 9.60 & & \\ % 3505
1110: QSO 0957+561A & 1.4093(1) & \ion{N}{5}, Ly$\alpha$, \ion{O}{6} & 01/26/1995 & 09/08/2000 & 5.62 & 4 & 4 \\ % 2052
1111: MRC 1118+128 & 0.685(1) & \ion{C}{4}, Ly$\alpha$ & 02/26/1995 & 03/28/2001 & 6.09 & 5 & none \\ % 2222
1112: PG 1241+176 & 1.273 & \ion{O}{6}, Ly$\alpha$ & 12/08/1992 & 03/19/2001 & 8.28 & 6 & 2 \\ % 3023
1113: PG 1309+355 & 0.184 & \ion{C}{4} & 05/20/1996 & 04/04/2001 & 4.88 & 6,7 & 5 \\ % 1780
1114: PG 1351+640 & 0.0882(2) & \ion{C}{4} & 09/05/1991 & 05/03/2001 & 9.67 & 3 & 5,6 \\ % 3528
1115: PG 1411+442 & 0.0896(5) & \ion{C}{4} & 10/03/1992 & 02/12/2001 & 8.37 & 3 & 2 \\ % 3054
1116: PG 1425+260 & 0.366 & \ion{C}{4}, \ion{N}{5} & 06/29/1996 & 04/26/2001 & 4.83 & 6,7 & 5 \\ % 1762
1117: 3C 351 & 0.37194(4) & \ion{C}{4} & 10/22/1991 & 08/18/2001 & 9.83 & 8 & 2,7 \\ % 3587
1118: PG 1718+481 & 1.084 & \ion{N}{5}, Ly$\alpha$, \ion{O}{6} & 05/13/1993 & 03/09/2001 & 7.82 & 7 & 2 \\ % 2857
1119: PKS 2135--14 & 0.20036(4) & \ion{C}{4} & 09/13/1992 & 09/20/2000 & 8.02 & 9,10 & 8 \\ % 2929
1120: MR 2251--178 & 0.06398(6) & \ion{C}{4} & 08/02/1996 & 11/05/2000 & 4.26 & 11 & 9 \\ % 1556
1121: PG 2251+113 & 0.3255(3) & \ion{C}{4} & 12/04/1992 & 05/01/2001 & 8.41 & 3 & 7,10 \\ % 3070
1122: \enddata
1123: \tablenotetext{a}{See \S\ref{S_obs} and \S\ref{S_indiv} of the text for a
1124: discussion of the adopted emission redshift values and related velocity
1125: uncertainties. The figure in parethesis gives the uncertainty in the
1126: last digit whenever this is available.}
1127: \tablenotetext{b}{{\em References to adopted redshift measurements. --}
1128: (1) \citet*{Marg85};
1129: (2) \citet{Kne98};
1130: (3) \citet{Mar96};
1131: (4) \citet{You81};
1132: (5) \citet{Wil76};
1133: (6) \citet{Sch83};
1134: (7) \citet{Bor92};
1135: (8) \citet{Era04};
1136: (9) \citet{Ha97c};
1137: (10) \citet{BB86};
1138: (11) \citet{Ber83}.
1139: }
1140: \tablenotetext{c}{{\em References to previously published FOS spectra}. --
1141: (1) \citet{Boi98};
1142: (2) \citet{Jan98};
1143: (3) \citet{Wil99};
1144: (4) \citet{Dol00};
1145: (5) \citet*[][the spectra were discussed but not displayed]{Bra00};
1146: (6) \citet{Mar96};
1147: (7) \citet{Bah93};
1148: (8) \citet{Ha97c};
1149: (9) \citet{Ga01};
1150: (10) \citet{Net95}.
1151: }
1152: \tablenotetext{d}{The earlier date corresponds to the FOS G190H spectrum and the later date the FOS G270H spectrum.}
1153: \end{deluxetable}
1154: %------------------------------------------------------------------------
1155: \clearpage
1156: %%% Table 2: Associated and Galactic Absorption Lines
1157: \input tab2.tex
1158: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1159: %%% F I G U R E S A N D C A P T I O N S
1160: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1161: \clearpage
1162: \begin{figure} %% FIGURE 1
1163: \centerline{\includegraphics[angle=-90,scale=0.55]{f1.eps}}
1164: \caption{An example of extreme continuum fits to the STIS spectrum of
1165: a broad \ion{C}{4} emission line. We incorporate these uncertainties
1166: into our selection criteria of variable lines. An additional
1167: illustration using a FOS spectrum can be found in
1168: Figure~\ref{fig_pksfits}.
1169: \label{fig_cont}}
1170: \end{figure}
1171: %------------------------------------------------------------------------
1172: \clearpage
1173: \begin{figure} %% FIGURE 2
1174: \centerline{\includegraphics[scale=0.4,angle=90]{f2b.eps}
1175: \includegraphics[scale=0.4,angle=90]{f2d.eps}
1176: \includegraphics[scale=0.4,angle=90]{f2f.eps}}
1177: \centerline{\includegraphics[scale=0.4,angle=90]{f2a.eps}
1178: \includegraphics[scale=0.4,angle=90]{f2c.eps}
1179: \includegraphics[scale=0.4,angle=90]{f2e.eps}}
1180: \caption{\small The STIS spectra of quasars observed in this survey with the
1181: best {\it effective} continuum fit superposed (the fit is not
1182: optimized in low-$S/N$ regions near the ends of some of the
1183: spectra). The extreme continuum fits (see \S\ref{S_obs} of the
1184: text) are also shown as dotted lines. The lower trace in each panel
1185: shows the error bar spectrum. The tick marks show the detected
1186: absorption lines listed in Table~\ref{tab_var} \label{fig_spec}}
1187: \end{figure}
1188: %
1189: \clearpage
1190: \begin{figure} %% FIGURE 2
1191: \centerline{
1192: \includegraphics[scale=0.4,angle=90]{f2h.eps}
1193: \includegraphics[scale=0.4,angle=90]{f2j.eps}
1194: \includegraphics[scale=0.4,angle=90]{f2l.eps}}
1195: \centerline{
1196: \includegraphics[scale=0.4,angle=90]{f2g.eps}
1197: \includegraphics[scale=0.4,angle=90]{f2i.eps}
1198: \includegraphics[scale=0.4,angle=90]{f2k.eps}}
1199: \bigskip\centerline{Fig.~\ref{fig_spec} {\em (continued)}}
1200: \end{figure}
1201: %
1202: \clearpage
1203: \begin{figure} %% FIGURE 2
1204: \leftline{\includegraphics[scale=0.4,angle=90]{f2n.eps}}
1205: \leftline{\includegraphics[scale=0.4,angle=90]{f2m.eps}
1206: \includegraphics[scale=0.4,angle=90]{f2o.eps}}
1207: \bigskip\centerline{Fig.~\ref{fig_spec} {\em (continued)}}
1208: \end{figure}
1209: %------------------------------------------------------------------------
1210: \clearpage
1211: \begin{figure} %% FIGURE 3
1212: \centerline{\includegraphics[scale=0.6,angle=-90]{f3.eps}}
1213: \caption{Comparison of the EWs of 247 lines that we have measured in
1214: FOS spectra with measurements of the same lines by \citet{Be02}.
1215: (a) Our EW measurement plotted against that of \citet{Be02}. The
1216: dashed line has unit slope and illustrates the good agreement
1217: between the two sets of measurements. In the interest of clarity,
1218: this plot shows only the range of observed EWs up to 6~\AA; three
1219: lines with $6\;{\rm \AA}<W_{\lambda}< 20\;{\rm \AA}$ are omitted.
1220: (b) The distribution of the normalized deviation between the two
1221: measurements, namely $\Delta W_{\lambda}/\sigma \equiv
1222: \left[W_{\lambda}({\rm theirs}) - W_{\lambda}({\rm ours}) \right]/
1223: (\sigma^2_{\rm theirs} + \sigma^2_{\rm ours})^{1/2}$. This
1224: histogram, which comprises all lines, appears symmetric about zero
1225: and includes five outliers with $|\Delta W_{\lambda}/\sigma|>3$ (one
1226: is out of the range of the plot at $|\Delta
1227: W_{\lambda}/\sigma|\approx 8$). For comparison, we overplot as a
1228: solid line a Gaussian of unit standard deviation, which represents
1229: the {\it expected} distribution of $\Delta W_{\lambda}/\sigma$.
1230: \label{fig_comp}}
1231: \end{figure}
1232: %
1233: %------------------------------------------------------------------------
1234: \clearpage
1235: \begin{figure} %% FIGURE 4
1236: \centerline{\includegraphics[scale=0.65,angle=-90]{f4.eps}}
1237: \caption{Distribution of differences between the EWs of the same lines
1238: as measured by the STIS and the FOS in two velocity bins relative to
1239: the quasar redshift: $|\Delta v|<5,000$\kms\ (associated lines; top
1240: panel) and $|\Delta v|>5,000$\kms\ (including Galactic and
1241: unidentified lines; bottom panel). See \S\ref{S_obs} of the text for
1242: an explanation of this division. Included in this figure are the two
1243: lines with STIS detections and FOS upper limits, bringing the total
1244: number of lines to 40 associated and 122 non-associated. (a) A plot
1245: of $W_{\lambda}({\rm STIS})~ vs ~W_{\lambda}({\rm FOS})$ restricted
1246: to EW up to 5~\AA\ for clarity. The dashed line in each panel has
1247: unit slope. (b) The distribution of the EW differences normalized
1248: by the uncertainty, $\left[W_{\lambda}({\rm FOS}) - W_{\lambda}({\rm
1249: STIS}) \right]/ (\sigma^2_{\rm FOS} + \sigma^2_{\rm
1250: STIS})^{1/2}$. The error bars ($\sigma_{\rm STIS}$ and $\sigma_{\rm
1251: STIS}$) include contributions from photon noise uncertainties and
1252: continuum placement uncertainties. The solid line overplotted on
1253: each histogram is a Gaussian of unit standard deviation, which is
1254: the distribution expected {\it a priori}, if the normalized EW
1255: differences are a result of measurement errors only.
1256: \label{fig_scatter}}
1257: \end{figure}
1258: %
1259: %------------------------------------------------------------------------
1260: %\clearpage
1261: \begin{figure} %% FIGURE 5
1262: \centerline{\includegraphics[scale=0.6]{f5.eps}}
1263: \caption{Distribution of the normalized EW differences of lines
1264: measured in FOS spectra but not detected in STIS spectra. This is
1265: analogous to the plot in the Figure~\ref{fig_scatter}b, but with
1266: $W_{\lambda}({\rm STIS})$ replaced by the 5$\sigma$ upper limit
1267: $W^{\rm max}_{\lambda}({\rm STIS})$. The 3.3$\sigma$ outlier on the
1268: far right is a Ly$\alpha$ line in PG~1718+481 at $\Delta
1269: v=+469$\kms.
1270: \label{fig_limits}}
1271: \end{figure}
1272: %
1273: %------------------------------------------------------------------------
1274: \clearpage
1275: \begin{figure} %% FIGURE 6 a-d
1276: \hbox{
1277: \hskip -0.3truein
1278: \includegraphics[angle=-90,scale=0.37]{f6a.eps}
1279: \hskip -0.3truein
1280: \includegraphics[angle=-90,scale=0.37]{f6b.eps}
1281: }\vskip -0.15truein
1282: \hbox{
1283: \hskip -0.3truein
1284: \includegraphics[angle=-90,scale=0.37]{f6c.eps}
1285: \hskip -0.3truein
1286: \includegraphics[angle=-90,scale=0.37]{f6d.eps}
1287: }
1288: \caption{ Visual illustration of the the variable NALs found in our
1289: variability survey. The top panel in each set shows STIS and FOS
1290: spectra overplotted. The FOS spectrum can be distinguished by its
1291: higher sampling rate and by the fact that it has been shifted
1292: upwards for clarity. The absorption lines of interest are identified
1293: by tick-marks. The bottom panel in each set shows the
1294: continuum-normalized spectra from the FOS (dashed line) and STIS
1295: (solid line). The FOS spectrum has been smoothed to the STIS
1296: resolution and resampled to the STIS wavelength scale to provide a
1297: fair comparison. The wavelength scales of the normalized spectra
1298: were aligned by comparing the observed wavelengths of Galactic,
1299: interstellar lines.\label{fig_overp}}
1300: \end{figure}
1301: %------------------------------------------------------------------------
1302: \clearpage
1303: \begin{figure} %% FIGURE 7
1304: \centerline{\includegraphics[scale=0.5,angle=-90]{f7.eps}}
1305: \caption{The \ion{C}{4} emission-line spectrum of PKS~2135--14 with
1306: two extreme {\it effective} continuum fits superposed as smooth,
1307: solid lines. The dotted line is a linear interpolation over the main
1308: absorption trough; it approximates the fit used by \citet{Be02} to
1309: measure the EW of the same absorption line.
1310: \label{fig_pksfits}}
1311: \end{figure}
1312: %------------------------------------------------------------------------
1313: \clearpage
1314: \begin{figure}
1315: \centerline{\includegraphics[scale=1.1]{f8.eps}}
1316: \caption{ Results of simulations demonstrating the effect of bulk
1317: motion and changing continuum strength on a series of atomic
1318: transitions of a hypothetical NAL system. The details of the
1319: simulations are described in \S\ref{S_fut} of the text. The middle
1320: set of panels shows the initial state of the absorber. The set of
1321: panels on the left show the effect of decreasing the column density
1322: by an order of magnitude. The set of panels on the right show the
1323: effect of increasing the ionization parameter by an order of
1324: magnitude.
1325: \label{fig_synth}}
1326: \end{figure}
1327: %
1328: \end{document}
1329: