February5,2008 PreprinttypesetusingLATEXstyleemulateapjv.11/12/01 RESULTS OF MONITORING THE DRAMATICALLY VARIABLE C IV MINI-BAL SYSTEM IN THE QUASAR HS 1603+38201 Toru Misawa2, Michael Eracleous2,3,4, Jane C. Charlton2, and Nobunari Kashikawa5,6 misawa,mce,[email protected],[email protected] February 5, 2008 ABSTRACT We presentsix new andtwopreviouslypublishedhigh-resolutionspectraofthe quasarHS 1603+3820 (z =2.542) taken over an interval of 4.2 years (1.2 years in the quasar rest frame). The observations em were made with the High-Dispersion Spectrograph on the Subaru telescope and Medium-Resolution 7 Spectrograph on the Hobby-Eberly Telescope. The purpose was to study the narrow absorption lines 0 (NALs). We use time variability as well as coverage fraction analysis to separate intrinsic absorption 0 lines, which are physically related to the quasar, from intervening absorption lines. By fitting models 2 to the line profiles, we derive the parameters of the respective absorbers as a function of time. Only n the mini-BAL system at z ∼ 2.43 (v ∼ 9,500 km s−1) shows both partial coverage and time abs shift a variability, although two NAL systems possibly show evidence of partial coverage. We find that all the J troughs ofthe mini-BAL systemvaryin concertandits totalequivalentwidth variationsresemble those 3 ofthecoveragefraction. However,noothercorrelationsareseenbetweenthevariationsofdifferentmodel 2 parameters. Thus, the observed variations cannot be reproduced by a simple change of ionization state norbymotionofahomogeneousparcelofgasacrossthe cylinderofsight. We proposethattheobserved 1 variationsarearesultofrapidcontinuumfluctuations,coupledwithcoveragefractionfluctuationscaused v 1 by a clumpy screen of variable optical depth located between the continuum source and the mini-BAL 6 gas. Analternativeexplanationis thatthe observedpartialcoveragesignatureis the resultofscattering 6 of continuum photons around the absorber, thus the equivalent width of the mini-BAL can vary as the 1 intensity of the scattered continuum changes. 0 Subject headings: quasars: absorption lines – quasars: individual (HS 1603+3820) 7 0 / 1. introduction 500km s−1)presentapowerfulwaytodeterminephysical h parameters of the accretion disk winds. Unlike the broad -p theQguaasseaorusshpavheasbeeseonfuasevdaraiestbyaocfkgorbojuecntdsstohuartcaersetolosctautdedy absorption lines (BALs; FWHM ≥ 2,000 km s−1; Wey- o mann et al. 1991) that are also associated with quasars r along our sight-lines to them. These objects include not outflows, NALs do not suffer from self-blending or from t only intervening absorbers such as intervening galaxies, s saturation, so that line parameters are more easily evalu- a the intergalacticmedium (IGM), clouds in the halo of the ated. BALsarethoughtto be associatedwithradiatively- : Milky Way, and the host galaxies of the quasars them- v driven outflows from accretion disks (e.g., Weymann et selves, but also intrinsic absorbers that are physically as- Xi sociated with the quasarcentralengines. One of the most al. 1991; Becker et al. 1997), while it is difficult to sep- arate intrinsic NALs from intervening NALs. Mini-broad r promising candidates for the intrinsic absorbers are out- a absorptionlines(mini-BALs)areanintermediatesubclass flowing winds from the quasars that could be accelerated between NALs and BALs, which are typically wider than by radiation pressure from the accretion disk (Murray et thoseofNALs,butnarrowerthanBALs. Mini-BALshave al. 1995; Arav et al. 1995; Proga et al. 2000) or by the advantages of both BALs (i.e., high probability of be- magnetocentrifugalforces(e.g.,Everett2005). Outflowing ing intrinsic lines) and NALs (i.e., line profiles can be windsareimportantcomponentsofquasarcentralengines resolved into individual components), which make them because they carry awayangular momentum from the ac- useful targets (e.g., Churchill et al. 1999; Hamann et al. cretion disk and allow the remaining gas in the disk to 1997a;Narayananetal. 2004). BALscouldprobethelow- accrete onto the central black hole. Quasar outflows are latitude, dense, fast portion of the wind, while the mini- also important for cosmology since they deliver energy, BALs and NALs may probe the lower-density portion of momentum, and metals to the interstellar and intergalac- thewindathighlatitudesabovethedisk. Thus,thestudy tic media, thus significantly affecting star formation and of mini-BALs and NALs complements the study of BALs galaxy assembly (e.g., Granato et al. 2004; Scannapieco becausethecorrespondingabsorbersresideindifferentre- & Oh 2004; Springel, Di Matteo & Hernquist 2005). gions of the outflow and allow us to sample different sets Intrinsic narrow absorption lines (NALs; FWHM ≤ of physical conditions. 1 Basedondatacollected atSubaruTelescope,whichisoperated bytheNationalAstronomicalObservatoryofJapan. 2 DepartmentofAstronomy&Astrophysics,ThePennsylvaniaStateUniversity,UniversityPark,PA16802 3 DepartmentofPhysics&Astronomy,NorthwesternUniversity,2131TechDrive,Evanston,IL60208 4 CenterforGravitational WavePhysics,ThePennsylvaniaStateUniversity 5 NationalAstronomicalObservatory,2-21-1Osawa,Mitaka,Tokyo181-8588, Japan 6 DepartmentofAstronomicalScience,GraduateUniversityforAdvancedStudies,2-21-1Osawa,Mitaka,Tokyo181-8588, Japan 1 2 Misawa et al. ToisolateintrinsicNALs,twotestsarecommonlyused: In the latter case, the crossing velocity is constrained to (i) partial coverage, i.e., trough dilution by un-occulted be v ≥8,000 km s−1, and the distance from the con- cross light (e.g., Hamann et al. 1997b; Barlow & Sargent 1997; tinuum source, r ≤ 0.2 pc. This is larger than the size of Gangulyetal. 1999),and(ii)variabilityoftheabsorption the continuum source, R ∼ 0.02 pc, but smaller than cont profiles within a few years in the quasar rest frames (e.g., that of the BLR, R ∼3 pc, estimated for this quasar. BLR Hamann et al. 1997a; Narayanan et al. 2004; Wise et On the other hand, the other C IV NALs found towards al. 2004). These effects occur for intrinsic absorbers that the quasar show no sign of being intrinsic. Further mon- are verycompactand verydense comparedto intervening itoring observations make it possible to (i) discriminate absorbers. The fractionofquasarsthathostintrinsicCIV between changesin the ionizationstate and motionof the associatedNALs(NALswithin5,000kms−1 ofthequasar absorbersacrossthe continuum source as the cause of the emission redshifts) has been estimated to be ∼25–27%at time variability seen in the C IV mini-BAL, and (ii) to z <2, using the time variability technique (Narayanan et confirm whether the other CIV NALs really show neither al. 2004;Wiseetal. 2004),and∼23%atz ∼2.5,basedon partial coverage nor variability. partial coverage analysis (Misawa et al. 2007). These sig- In this paper, we present the results of monitoring natures are a sufficient, but not a necessary, condition to HS 1603+3820 over 4.2 years (1.2 years in the quasar demonstratethe intrinsicnature ofanabsorber. Itis very rest frame) based on eight spectra. Six of these spectra likely that some NALs without time variability or partial were taken with the High Dispersion Spectrograph(HDS; coverage are also intrinsic. Noguchi et al. 2002) on the Subaru telescope, and two Both of the above methods have been applied to BAL withtheMedium-ResolutionSpectrograph(MRS;Horner, and mini-BAL systems. For example, partial coverage Engel, & Ramsey, 1998) on the Hobby-Eberly Telescope analysis has been carried out by Petitjean & Srianand (HET). The first two spectra in this time series were pre- (1999), Srianand & Petitjean (2000,) Yuan et al. (2002), sented and discussed by Misawa et al (2003, 2005). With andGangulyetal. (2003),whiletime-variabilityhasbeen the new spectra we are able to sample the variations of studied by Foltz et al. (1987), Turnshek et al. (1988), the absorption lines more densely and probe the internal Smith & Penston (1988), Barlow et al. (1992), Barlow structure of the absorber. (1993),Vilkoviskij& Irwin(2001),Ma (2002),Narayanan In §2, we describe the observations and data reduction. et al. (2004), and Lundgren et al. (2007). It is quite Themethodsusedformodelfittingareoutlinedin§3. The difficult (oralmostimpossible)to deblendkinematic com- propertiesofthe mini-BALandofthe otherNAL systems ponents in those heavily blending absorption features us- are examined in §4. The possible origins of the time vari- ing low/intermediate resolution spectra. However, some ability seen in the mini-BAL system are discussed in §5, authors successfully deblended narrower absorption com- and a summary is given in §6. We adopt z = 2.542 as em ponents from each other in such systems using high res- the systemic redshift of the quasar, which was estimated olution spectra of R > 40,000 (e.g., Yuan et al. 2002; fromits narrowemissionlines (Misawaetal. 2003). Time Gangulyetal. 2003). Systematicmonitoringofindividual intervalsbetweenobservationsaregiveninthequasarrest mini-BALs using high-resolution spectra taken on several frame throughout the paper, unless otherwise noted. We epochs canpotentially provideveryimportantconstraints useacosmologywithH =75kms−1Mpc−1,Ω =0.3,and 0 m onthepropertiesoftheabsorbers. However,nosuchcam- Ω =0.7. Λ paign has been attempted so far, to our knowledge. The opticallybrightquasarHS1603+3820(z =2.542, em 2. observations and data reduction B=15.9), first discovered in the Hamburg/CfA Bright QuasarSurvey(Hagenetal. 1995;Dobrzyckietal. 1996), We observed HS 1603+3820 eight times over a period is known to have a large number (13) of CIV doublets at of 4.2 years in the observed frame, from March 2002 to 1.965< z < 2.554 (Dobrzycki, Engels, & Hagen 1999). May 2006. We will call the dates of these observations abs Using high-resolution spectra (R = 45,000) taken with epoch 1 through epoch 8. Six of the observations were the Subaru telescope, Misawa et al. (2003, 2005) clas- obtainedwith Subaru+HDS, using a 0′.′8 or 1′.′0 slit width sified all C IV doublets into 8 C IV systems, and found (R = 45,000 or 36,000), and adopting 2×1 pixel binning that only a mini-BAL system at z ∼ 2.43 (shift veloc- along the slit. The red grating, with a central wavelength abs ity7, vshift = 8,300–10,600 km s−1) shows both partial of 4900 ˚A, was used to cover as many C IV systems as coverage and time variability in an interval of 0.36 years possible,exceptforthefirstandthelastobservationsthat in the quasar rest frame. Based on these results, Mis- usedcentralwavelengthsof6450˚Aand5700˚A.Theother awa et al. (2005) placed constraints (i) on the electron two spectra were taken with the HET+MRS. The MRS density (n ≥ 3.2 × 104 cm−3) and the absorber’s dis- features a pair of fibers: one for the target and another e tance from the quasar (r ≤ 6 kpc), if a change in the for the sky, which enables us to perform sky subtraction ionization state causes the variability, or (ii) on the time effectively. We used a 1′.′5 fiber to get R=9,200 spectra. scale for the absorber to cross the continuum source and A setup with a central wavelength of 7,000 ˚A is required the distance from the continuum source, if gas motion for the queue observations with MRS+HET. across the background UV source causes the variability. We reduced both the Subaru and HET data in a stan- dard manner with the IRAF software8. We used a Th-Ar 7 Theshiftvelocityisdefined aspositiveforabsorptionlinesthatareblueshiftedfromthequasar. 8 IRAFisdistributedbytheNationalOpticalAstronomyObservatories,whichareoperatedbytheAssociationofUniversitiesforResearchin Astronomy,Inc.,undercooperativeagreementwiththeNationalScienceFoundation. 9 The broad emission lines probably do not vary between observations, because bright quasars like HS 1603+3820 do not show significant variabilityinmagnitudeonatime-scaleasshortas∼1year(e.g,Giveonetal. 1999, Hawkins2001,Kaspietal. 2007). Dramatically Variable Mini-BAL in HS 1603+3820 3 spectrumforwavelengthcalibration. Wedirectlyfittedthe at z = 2.5114. This system was probably a false de- abs continuum, which also includes substantial contributions tection(Dobrzycki2005;privatecommunication). Among frombroademissionlines9,withathird-ordercubicspline the 9 CIV systems, only the system at zabs =2.42–2.45is function. Aroundheavilyabsorbedregions,inwhichdirect classified as a mini-BAL system, because its line width is continuum fitting is difficult, we used different techniques larger than 500 km s−1. for the Subaruand HET spectra. For the Subaru spectra, WecarriedoutmodelfitstoallCIVsystemsinthesame we adopted the interpolation technique introduced in Mi- mannerasinourpreviouspaper(Misawaetal. 2005),ex- sawa et al. (2003). Since the echelle blaze profile in HDS cept for the mini-BAL. Misawa et al. (2005) performed shows time variability during a night, we cannot directly a manual Voigt profile fit to the mini-BAL, because self- usetheinstrumentalblazefunctionasacontinuumprofile. blending (blending of the blue and red members of the In this scheme, we obtain the continuum shape by inter- doublet) occurs in this system. In its original form, the polating between the two echelle orders adjacent to the minfit code was not able to handle such self-blended re- orderofinterest10,afterweightingflux levelsbasedonthe gions. The code was originally written for narrowabsorp- flat-frame spectrum [see eqn. (1) of Misawa et al. 2003]. tion lines, thus it fitted the models to the blue and red We then divide the quasar spectrum by this interpolated members of doublets separately. This procedure overpro- continuum shape to obtain the normalized spectrum. We duced the residual intensity at the self-blending regions have verified the validity of this technique by applying it because it added contributions from two components. To to a stellar spectrum; the continuum error is always less avoidthis problem,werevisedthecodesoastofitmodels than 3.3%. For HET spectra, we used a blaze profile con- to the blue and red members of doublets simultaneously, structedfromtheflatfieldspectrumasacontinuummodel i.e.,bymultiplyingthecontributionsfromthetwodoublet after multiplying it by a scaling factor to adjust its count members. In this paper, we fit the observed mini-BALs level to that of the quasar spectrum (the scaling factor is automatically using this improved minfit. a ratio of counts in the quasar spectrum to those in a flat As in Misawa et al. (2005), we fitted only the first spectrum for the same unabsorbed region). (bluest) two C IV kinematic components in the system, An observation log is given in Table 1, in which we list atλ =5284–5318˚A, because the other componentsare obs the observation epoch, date of observation, relative time soheavilyblendedwitheachotherthatwecannotseparate in the quasar rest frame, instrument, wavelength cover- them, even with the improved minfit (i.e., minfit gives age, total exposure time, spectral resolving power, and multiplesolutions). Atfirst,weremovethecontamination signal-to-noise ratio (S/N) per pixel (after rebinning to from Si II λ1527 in System B, at λ = 5303–5318 ˚A. The pixel scales of 0.03 ˚A and 0.15 ˚A for Subaru and HET SiII λ1527line in the systemdid not show any detectable data, respectively). The S/N is evaluated around 5370 ˚A, variability between the first two observations. The C IV close to the C IV mini-BAL. In Figure 1, we show a nor- doublets for System B show neither partial coverage nor malized spectrum, after combining 6 spectra taken with time variability, and are consistent with it being an inter- Subaru+HDS in the region redward of the Lyα emission vening system. Therefore, the Si II λ1527 line is also not line of the quasar (λ>4300 ˚A). likely to vary. Thus, we removed the contamination from 3. fitting procedure Si II λ1527 by dividing the observed spectra by a model profilesynthesizedwiththe line parametersofSiIIλ1527, We used the line-fitting software package minfit determined from an average of the first two observations (Churchill 1997; Churchill et al. 2003), with which we presented in Misawa et al. (2005). can fit absorption profiles with four free parameters: red- After removing the narrow Si II λ1527 components, we shift (z), column density (log N), Doppler parameter (b), applied the improved minfit to the spectrum using two and coverage fraction (Cf). Here, the coverage fraction is C IV components, one narrow and one broad, as in Mis- the fraction of the background continuum source and the awa et al. (2005). During the fitting process, we initially broad-emission line region that is covered by the intrin- forcedthesamecoveragefractionforthenarrowandbroad sic absorber. The coverage fraction can be systematically components, as in Misawa et al. (2005). In that paper we evaluated in an unbiased manner by considering the op- showed that, during the first two epochs of observation, tical depth ratio of resonant, rest-frame UV doublets of thissolutionproducedanequallygoodfitassolutionsthat Lithium-like species (e.g., CIV, NV, and SiIV), namely, allowedthese coveragefractions to differ. We also consid- (R −1)2 ered the possibility of differing coverage fractions for the r Cf = 1+R −2R , (1) narrow and broad components in later epochs of observa- b r tion. where R and R are the residual fluxes of the blue and b r Figure 2 shows 6 spectra taken with Subaru+HDS, for red members of the doublets in the continuum normal- which we find good model fits. The fits themselves are ized spectrum (Hamann et al. 1997b; Barlow & Sargent shown in Figure 3. We cannot fit spectra obtained with 1997; Crenshaw et al. 1999). A C value less than unity f HET+MRS in the same manner, because of their lower signifies that a portion of the background source is not resolutionandS/N.Therefore,ontheHET+MRSspectra, occulted by the absorber. This, in turn, means the dou- weoverplottedmodelprofilessynthesizedwithparameters blet is probably produced by an intrinsic absorber (e.g., that are linearly interpolated between the two adjacent Wampler et al. 1995; Barlow & Sargent 1997). We found epochs observed with Subaru. These models provide rea- 9 C IV systems at zabs = 1.88–2.55, as identified in Do- sonably good fits to the HET+MRS profiles. The fitting brzyckiet al. (1999),but we did not detect a CIV system 10Iftheorderofinterestism,weuseordersm−1andm+1asreferenceorders. However,inourdatathoseadjacentordersarealsosomewhat affected bybroadabsorptionfeatures. Therefore,weusedordersm−3andm+3asreferenceorders. 4 Misawa et al. parameters for the narrow and broad C IV components, identified because of heavy contamination from the Lyα at different epochs, are summarized in Table 2. Columns forest. SiIV was not detected, even in our final, high-S/N (1)–(3) give the epoch number, observation date and the spectrum. Figure 5 presents graphically the evolution of relative time in the quasar rest frame. Columns (4)–(8) the fit parameters of the C IV mini-BAL, including (from list,forthenarrowcomponent,theshiftvelocity,CIVcol- top to bottom) the total rest frame equivalent width at umn density, C IV Doppler parameter, coverage fraction, λ = 5288–5346 ˚A, the C IV column density, the Doppler and observed-frame equivalent width evaluated from the parameter, the coverage fraction, and the shift velocity. syntheticmodelspectrum. Columns(9)–(13)arethesame We discuss each in detail below. ascolumns(4)–(8),butforthebroadcomponent. Wecon- Thetotalrestframeequivalentwidthincreasedbyabout firmed that the fit parameters for the first two epochs are afactoroftwooverarestframetimeintervalof0.36years, consistentwithin1σerrorswiththosepresentedinMisawa and then decreased over an interval of 0.6 years. During et al. (2005), based on a manual fitting procedure. this entire period, the general appearance of the line pro- We also applied minfit to all metal lines in the other file did not change significantly. Based on the results of NAL systems. Since, as described later, we did not see just the first two epochs, Misawa et al. (2005) specu- anyremarkablevariabilityintheseNALsystems,wecom- lated that this mini-BAL may soon evolve into a BAL, bined all 6 Subaru spectra to produce a final spectrum with more than 10% of the flux absorbed over at least a with higher S/N (∼96 per pixel) before applying minfit 2,000kms−1range. Thisideaisnotconfirmedbythesub- to the NALs11. The best fit parameters are summarized sequentepochs,since the mini-BAL was foundto weaken. inTable 3. Column(1)isanidentificationnumber forthe We also monitored the equivalent widths of the narrow line. Columns (2) and (3) give the observed wavelength and broad components, separately. At first, both compo- and redshift. Column (4) gives the shift velocity from the nents have almost the same equivalent width. But after quasar systemic redshift. Columns (5) and (6) give the epoch3,theequivalentwidthofthebroadcomponentwas column density and the Doppler parameter with 1σ er- much higher than that of narrow component. rors. Column(7)reportsthe coveragefractionwith its 1σ The variability of the column density in the mini-BAL error(whichincludes the errordue tocontinuumlevelun- is complex. This is especially true for the broad com- certainty as described below). During a fitting trial, min- ponent, which showed a large increase in column density fit sometimes gives unphysical coveragefractions such as within 0.13 years, from epoch 3 to epoch 7, as shown in Cf < 0 or Cf ≥ 1. When this happens, the other fit pa- Figure 5. This increase in column density was concurrent rameterssuchasthecolumndensityandDopplerparame- with the increase of equivalent width, as shownin the top ter havenophysicalmeaning. These unphysicalfitresults window of the figure. One possible explanation for this could be causedby errorsin the continuumfit. Misawaet dramaticincreaseisachangeofthe ionizationstateofthe al. (2005)foundthatthederivedcoveragefractionscanbe gas (Hamann et al. 1997b; Narayananet al. 2004),which significantly affected by continuum level errors, especially we discuss further in §5. for very weak components whose Cf values are close to 1. An interesting trend is seen in Doppler parameter vari- Therefore, if minfit gives unphysical Cf values for some ability. Fromepoch3to epoch5 (anintervalof0.09years components, we carry out the fit again for the entire line, in the quasar rest frame), the Doppler parameter of the assuming Cf =1 for those components, following Misawa broad component increased suddenly from ∼ 350 km s−1 et al. (2007). In this case, we do not evaluate the error to ∼ 860 km s−1, and then it decreased gradually. This in Cf. The best fit models are overlaidwith the observed abrupt jump is correlated with the sudden increase of spectra in Figure 4(a–i). the column density in the broad component, as described Therearealsoothererrorsourcesforthefitparameters, above. Onthe other hand, the narrowcomponentdid not such as blending with other lines, and convolution of the show a significant change in Doppler parameter. true spectrum with the instrumental line spread function The coverage fraction (assuming it is the same for the (LSF). Both of these are negligible if (1) the doublet is narrowandbroadcomponents)showsavariabilitysimilar strong enough, i.e., log (N/cm−2) ≥ 14.0 in the case of to that of the total equivalent width. After increasing to C IV doublets, (2) the normalized residual flux is smaller 0.45,thecoveragefractiondecreasedto0.24. Inspectionof than ∼0.5,and (3) the line profile is much wider than the the separate contributions of the broad and narrow com- instrumentalLSF(Misawaetal. 2005). Mostcomponents ponents to the total equivalent width shows that the cov- in our sample are broad enough compared with the LSF erage fraction evolves similarly to the equivalent width of (correspondingtob∼4km s−1),andnotseverelyblended the narrow component. If the change of coverage fraction with each other, as described in section 4.2. is caused by the absorber’s motion across the background flux source, the absorber size is constrained to be smaller 4. variability of fit parameters thanthebackgroundsource(abouthalfofthebackground 4.1. The Mini-BAL System sourcebecauseofthe maximumC valueof∼0.5). Atthe f same time, the column density increased, suggesting an We previously reported that the CIV mini-BAL of Sys- inhomogeneous absorber. A more detailed discussion will tem A shows trough dilution (Misawa et al. 2002), and be presented in §5. variability within 0.36 years (Misawa et al. 2005). With The column density and equivalent width of the broad theadditionalspectraofthispaper,wefollowthevariabil- component increased over the same period (epoch 3 to ity of this system for 1.2 years in the quasar rest frame, epoch8)asitscoveragefractiondecreased. However,since sampling it at eight epochs. N V and Lyα could not be 11 For system G, we combined only spectra in epochs 3, 7, and 8, because the other spectra are affected by a data defect near the blue/red CCDgap. Dramatically Variable Mini-BAL in HS 1603+3820 5 we assumed that the broad and narrow components had fractions of two components, 11 and 14, deviate the same coverage fractions, we must assess whether this from unity by more than 1σ, but both are heav- assumption could lead to the observed trends. Specifi- ily blended with other components and their C f cally, we consider the effect ofassuming a constantC for values are not very reliable. All components in the f the broad component (adopting Cf = 0.36 for epoch 3 N V NAL are consistent with full coverage. Only through epoch 8, and allowing the narrow component Cf the epoch 8 spectrum covered the Si IV doublet of to change as before). This model, which also producedan this system. For the Si IV doublet, only compo- adequatefittothedata,isillustratedwiththedottedlines nent 7 suggests partial coverage, but the model fit in the 2nd, 3rd, and 4th panels from the top of Figure 5. isnotgoodaroundthe leftwingofthis component. For this case, the column density of the broadcomponent We concludethat this systemdoes notshowstrong doesnotincreaseasrapidly,butitstilldoesincrease. The evidence for partial coverage, and is probably an same applies for the equivalent width. Thus our conclu- interveningsystem, most likely a galaxybecause of sions regarding the evolution of the column density and the strength of the low-ionization absorption lines. equivalent width are not qualitatively changed by our as- sumption about the C value of the broad component. System C —TheCIVNALinthissystemwithasimple f absorptionprofile(seeFigure4c)isfittedwithonly We have looked for correlated changes in all of the one component, as in Misawa et al. (2005). Both aboveparametersofthemini-BALprofilebyplottingthem fits are consistent with full coverage. We also did against each other. In Figures 6 and 7, respectively, we not find any evidence for time variability. Except show the relevant plots for the narrow and broad com- for its small shift velocity, ∼ 430 km s−1, there is ponents of the mini-BAL. Other than a tentative trend no evidence to suggestthat this system is intrinsic. between C and b in Figure 6 and between C and N f f CIV in Figure 7, we see no convincing correlation. This lack System D — Misawa et al. (2005) fitted this system’s of correlated variability between model parameters is an C IV and Si IV profiles with 4 and 2 components, important clue, which we take into consideration in our respectively, and did not find any partial coverage. later discussion. However, we found an additional C IV component The observed shift velocity represents the ejection ve- at v = −160 km s−1 from the system center locity of the absorber from the central engine projected shift that shows partial coverage, and we have no rea- alongthe lineofsight. While thelinecenterofthenarrow son to suspect systematic error. The line profiles componentremainsnearlyconstant,thebroadcomponent shifts by about 300 km s−1 (v ∼ 10,100 km s−1 → are shown in Figure 4d. This component was not shift studied by Misawa et al. (2005) because of a data 10,400 km s−1) from epoch 1 to epoch 7. If this veloc- defect. We also now find that component 1 in the ity shift is really due to the radiative pressure from the SiIVNALpossiblyshowspartialcoverage,although continuum source, the acceleration of the gas can be cal- it was consistent with full coverage in the shorter culated to be ∼ 0.01 m s−2. However, the measurement exposure presented by Misawa et al. (2005). We of the center of the broad component is highly uncertain, also note that it is surprising that the C IV λ1548 because the observed velocity shift is much smaller than is black at the position of this Si IV component if the total line width because of self-blending of the blue the partial coverageis true. However,the fact that and red members of the CIV doublet. Moreover,the cen- partial coverage is apparent for both a Si IV and terofthebroadcomponentmovedinbothdirections(i.e., a C IV component suggests that this system may bluewardandredward)duringourobservations. Although be an intrinsic system. The fact that this system it is still not clear if there is a truly systematic increase is redshifted from the quasar systemic redshift also in this parameter, we could regard the above value as an supports this idea. Although this system is located upper limit on the acceleration of the absorber. near the top of the C IV emission line, there is no residualflux at the position of the CIV doublet for 4.2. NAL Systems components4and5. Thesecomponentsmustthere- Inadditiontothemini-BALsystem,wedetected8NAL fore cover both the continuum source and broad systems in the HS 1603+3820 spectra. For these NALs, emission line region. We do not see any variability we do not detect significant variability. Model fit results over the course of our observations. to the combined spectrum of the 6 Subaru observations System E — Misawa et al. (2005) fitted the C IV NAL (with a total exposure time of 8.9 hours and a S/N per withonlytwocomponents,andoneofthemshowed 0.03 ˚A pixel of about 91 at λ ∼ 5370 ˚A) are consistent evidence of partial coverage. However, our higher- with the results based on only the epoch 2 spectrum (a S/Nspectrum(showninFigure4e)requires5com- 1.7 hour exposure with S/N ∼ 49 per pixel; Misawa et ponents to fit the C IV NAL, and three of them al. 2005), with only a few small exceptions as described are consistent with full coverage. Although com- below. Inthissectionwediscussindividualsystems,other ponents 3 and 5 deviate from full coverage, they than system A (the mini-BAL). The profiles of the lines are heavily blended with the other components, so of these systems are shown on a common velocity scale in this result is quite uncertain. The shift velocity of Figure 4(a–i). thesystemisalsoextremelylarge,∼60,000kms−1. There is no compelling reason to believe that this System B — We fit this strong C IV NAL, shown in system is intrinsic to the quasar. Figure 4a,b,with15components,althoughMisawa et al. (2005) used 19 components. The coverage System F —Ourbest-fitting modelis almostconsistent 6 Misawa et al. with the previous one, having the same number of appear to be driven largely by changes in the coverage components at almost the same positions (see Fig- fraction,as shownin Figure 5. Other physicalparameters ure 4f). All components are consistent with full (N andb)arevaryingaswell,buttheredoesnotseem CIV coverage except for component 1 that is blended to be an obvious correlation between changes of different with the strongcomponent2. This systemalso has profile parameters, at least in the bluest troughs of the a very large shift velocity, ∼ 50,000 km s−1. We mini-BAL (see Figures 6 and 7). consider it probable that this is an intervening sys- 5.1. Transverse Motion of a Homogeneous Absorber tem. In Misawa et al. (2005) we favored an interpretation in System G — Because the C IV NAL in this system was which motion of the (homogeneous) absorber across the positionedatthe edge ofthe blue CCDof the HDS line (or cylinder) of sight causes the observed variability (around which bad pixels are common) in some of (see also Hamann et al. 1997a). The assumed scenario is our observing setups, the absorption profiles are depicted in the cartoon of Figure 2a of Hamann & Sabra slightlyorseverelydifferentinsomespectra. There- (2004). With time, an increasing fraction of the cylinder fore, we combined only spectra from epochs 3, 7, of sight was covered by the absorber, leading to changes and 8 (these are not affected by the data defect) inthe coveragefraction. Inthis contextandbasedontwo to produce a final spectrum for this system (Fig- snapshots of the spectrum, we derived a transverse veloc- ure 4g). We fit the C IV NAL with 6 components, ity of 8,000 km s−1 and a constraint of r < 0.2 pc on the and all of them are consistent with full coverage. distance of the absorber from the continuum source (see Component5wasnotfittedbyMisawaetal. (2005) details in Misawa et al. 2005). because of the unphysicalline ratio of the blue and That model was motivated largely by the observation red component, due to line blending. This system that the coverage fraction was the only quantity that is probably an intervening system. changed substantially between the first two observations. The subsequent observationsshowthat scenarioto be too System H — Two components are necessary to fit this simple,namelythecolumndensityoftheabsorberchanges C IV NAL (see Figure 4h). The shallower absorp- with time as well, suggesting that the absorber is inho- tion feature, centered on component 2, has an un- mogeneous and its internal structure plays an important physical ratio of its blue and red members. Proba- role in the variability of the mini-BAL profile. Thus, a blythebluememberiscontaminatedbyotherlines. more plausible scenario is the one depicted in Figure 2b Only component 1 is useful, and it yields C = 1. f of Hamann & Sabra (2004) and in Figure 5 of de Kool, In this system, we againdo not see any evidence of Korista, & Arav (2002), where the absorbing medium is intrinsic properties. inhomogeneous, with clumps spanning a range of column System I — This system was detected only in the densities scattered over the cylinder of sight. epoch 8 spectrum, because it was not covered in There is a remaining difficulty with this scenario, how- the other epochs. We used 7 components to fit the ever. Thefactthatallthetroughsinthemini-BALvaryin CIV NAL (see Figure 4i). Component 4 may show concert would require a coincidence between the motions partialcoverage,butits C valuediffers fromunity of the individual clumps. More specifically, the projected f by only slightly more than 1σ. The case for partial areas of all the clumps within the cylinder of sight should coverage for component 6 is somewhat more com- vary in the same way (i.e., rise and fall with time in the pelling, since the minfit fitting procedure yielded same manner for all clumps). We consider such a coordi- a value C = 0.87± 0.02. Component 4 is not nated sequence of variations rather unlikely, therefore, we f blended with other components, and component 6 regard this interpretation as implausible. We note that if is only weakly blended. Although the S/N for this future observationsshowthis patterntobe common,then spectrum is not very high in the relevant region, this interpretation can be conclusively rejected. this system may be another candidate for an in- 5.2. Internal Motions in the Absorbing Gas trinsic system. At first glance, compression of the absorber might be 5. discussion: the origin of time variability abletoproduceatrendsimilartothatofFigure7b,where the coverage fraction decreases as the column density in- In view of the observational constraints derived above, creases in one of the kinematic components of the mini- we evaluate here different scenarios for the origin of the BAL.However,oncloserexamination,suchacompression variability of the CIV mini-BAL. A fundamental assump- would require highly supersonic motions within the ab- tion underlying our discussion is that the kinematic com- sorber. Assuming the background continuum source size ponents making up system A represent parcels of absorb- is 0.02 pc (this is the UV-emitting region of the accretion ing gas at different positions along the line of sight. In disk; see Misawa et al. 2005), a change in the coverage effect, we assume that the absorberis embedded in an ac- fractionfrom 0.36to 0.27 within 0.13 years(epochs 5 and celerating outflow from the quasar,thus the blueshift of a 7) implies a speed of order 104 km s−1. In comparison, kinematic component increases with its distance from the quasar continuum source. the speed of sound in a ∼ 104 K gas is only 10 km s−1. A very important observationalclue from the data pre- Thus we consider such a scenario untenable. sentedaboveisthefactthatallthekinematiccomponents 5.3. Scattering of Continuum Photons Around the ofthemini-BALsystemvaryinconcert(i.e.,thedepthsof Absorber all troughs increase or decrease together). These changes Dramatically Variable Mini-BAL in HS 1603+3820 7 The observed partial coverage signature can, in prin- tive recombinationcoefficients of5.3×10−12 cm3 s−1 and ciple, be explained if continuum photons are re-directed 2.8×10−12 cm3 s−1 (from Arnaud & Rothenflugh 1986; by a scattering medium towards the observer, when they assuming a nominal gas temperature of 20,000 K, follow- were initially traveling in a different direction. Thus, the ingHamann1997). Thusweobtainalimitontheelectron absorber can be homogeneous and its projected size need density of n > 1×105 cm−3, using the methodology of e not be smaller than the cylinder of sight towards the con- Narayanan et al. (2004) and a bolometric luminosity of tinuum source. The dilution of the troughs by photons L =2.5×1048 erg s−1 (Misawaetal. 2005). Toconvert bol scattered around the absorber can change with time as thislimittoalimitonthedistanceoftheabsorberfromthe the absorber moves relative to the scattering medium or continuum source, r, we must assume a value of the ion- as the conditions in the scattering medium change. In izationparameter. IftheCIVcolumndensityfluctuations the context of this scenario, the apparent coverage frac- reflectconversionsofCIII↔CIVinalower-ionizationab- tion can vary according to the behavior of the scattering sorber,thenthecalculationsofHamann(1997)indicatean medium. Moreover,the variationsofthecoveragefraction ionization parameter of U ∼0.002 and lead to r <8 kpc. need not be tied in any way to variations of other param- If,ontheotherhand,theCIVcolumndensityfluctuations etersofthe absorber,suchasthe CIVcolumndensity,the reflectconversionsofCV↔CIVinahigher-ionizationab- b-parameter,or the shift velocity. This idea can be tested sorber,thenthecalculationsofHamann(1997)indicatean observationallythroughspectropolarimetry. Inparticular, ionization parameter of U ∼ 0.06 and lead to r < 1 kpc. we wouldexpect the fractionalpolarizationinthe absorp- Inthiscontextthesimultaneousvariabilityofallthemini- tionmini-BALtroughstobehigherthaninthecontinuum BAL troughs implies that these troughs should represent (see the analogous test in BAL quasars by Brotherton et gas parcels of similar densities for the following reason. al. 1997,Ogleetal. 1999,andLamy&Hutsem´ekers2000, Since all absorbing gas parcels must be within the cylin- forexample). Furthermore,wewouldalsoexpectthisfrac- derofsighttothecontinuumsource,thetimelagbetween tional polarizationto increaseas the coveragefractionde- changes in the continuum intensity at the source and our creases. observationofachangeinthemini-BALprofileisthesum of the light traveltimes fromthe continuum source to the 5.4. Change of Ionization State of the Absorbing Gas absorber and from the absorber to the observer, plus the recombination time of the gas. The sum of light travel Changes in the level of the ionizing continuum can also times is the sameregardlessofthe locationofagasparcel bring about absorption-line variability since they change along the cylinder of sight, therefore, a difference in time the relativeionic populations in the absorber(in this par- lagcanonlyresultfromdifferentrecombinationtimes,i.e., ticular case, the relative populations of C III, C IV and different gas densities. We can then conclude that the C V). In principle, a monotonic change in the continuum density does not vary significantly across the line profile, intensity can produce a non-monotonic change in the ab- whichimpliesinturnthattheshapeofthemini-BALpro- sorption line strength. For example, an absorber initially file is determined primarily by the combinationof column at a low ionization state that sees a rising ionizing flux density (i.e., physical thickness) and coverage fraction as will first respond with an increase in the strength of its a function of velocity. C IV absorption lines (as C III → C IV) and then with a Fluctuations in the ionization state of the absorber are decrease (as C IV → C V). The variability time scale of anattractiveexplanationofthe observedvariationsinthe the absorptionlinesprovidesanupper limitontheioniza- strengthofthemini-BALinHS1603+3820. However,the tion or recombinationtime of the absorbinggas,hence its mini-BAL appears to vary much faster than any expected electrondensity(see the detaileddiscussioninHamannet variationsintheUVcontinuumofsuchaluminousquasar al. 1997b and Narayanan et al. 2004). The density limit (see for example, Giveon et al. 1999, Hawkins 2001, and can be obtained from n ≥ (α t )−1, where α is a re- e r var r Kaspietal. 2007). Moreover,Fleming&Kennefick(2006) combinationcoefficientandt isavariabilitytime scale. var foundno variabilityofthis particularquasarin 10daysin Thisisasimplifiedpictureinwhichthegasisinionization the observed frame. 12 Therefore we propose that the equilibrium and CIV is the dominant ionic species of C. variations of the ionizing continuum seen by the absorber Since the C IV ionic column densities have fluctuated arecausedbyascreenofvariableopticaldepthbetweenit up and down more than once over the course of our mon- andthe (relatively steady)continuum source. This screen itoring observations, the ionizing continuum seen by the could be the ionized, clumpy, inner part of the outflow, absorber must be fluctuating too. Moreover, the contin- i.e., the shielding or hitchhiking gas invoked in the out- uum intensitymustbe fluctuating byatleasta factorof3 flow models of Murrayet al. (1995),andarisingnaturally onrest-frametimescalesof6monthsorlesstoproducethe in the more detailed calculations of Proga et al. (2000). observedchangesintheCIVioniccolumndensity(seethe Thetwo-dimensional,axisymmetricmodelsofProgaetal. calculations of Hamann 1997). Taking this interpretation (2000)also show that disk winds can be acceleratedup to at face value (but see discussion below), we can use the 15,000 km s−1, and that they are unsteady and generate shortest variability time scale that we have observed (16 dense knots periodically. The time variability seen in the days in the quasar rest frame, epochs 5 → 7; see Fig. 5) mini-BALsystemofHS1603+3820mayberelatedtothis to constrain the density of the gas. We consider both an mechanism. The conditions in the screen may be similar increase and a decrease in the ionization state of the gas tothoseof“warm”absorbers,observedinthe X-rayspec- as a cause for the observed change in the C IV column tra of many Seyfert galaxies,where it often co-exists with density (i.e., C III → C IV or C V → C IV), with respec- 12 Unfortunately, our spectra cannot provide information on variations of the continuum level because we cannot calibrate their flux scale. Thisisaconsequence ofthefactthattheblazefunctionoftheHDSonSubaruchanges withtimeoverthecourseofanight. 8 Misawa et al. a UV absorber (e.g., Crenshaw et al. 1999). We note, systemsin the quasarspectrum. The partialcover- by way of example, that models for the warm absorbers age signature and time variability of the mini-BAL in NGC 5548, IRAS 1339+2438, and NGC 3783 require show it to be intrinsic to the quasar. Two of the multiple ionizationphasesforanadequatefittotheX-ray NALsystems(DandI)mayalsobeintrinsic,based spectrum (Kaastra et al. 2000, Sako et al. 2001, Blustin upon our partial coverage analysis. et al. 2002), one of which may fulfill the requirements for the inner screen. 2. Weareabletofitmodelsonlytothebluestportion Iftheopticaldepthofthe screenatwavelengthsaround of the CIV mini-BAL profile where self-blending is the C IV edge is close to unity, then small fluctuations not severe. All fit parameters (i.e., column density, in the column density can cause large changes in the Doppler parameter,coveragefraction,and shift ve- shapeandintensityofthe transmittedionizingcontinuum locity) as well as the total equivalent width of the (e.g., Fig. 6 of Murray et al. 1995, or Fig. 1 of Hamann system vary significantly with time, even on short 1995),leadingtotheobservedchangesinthemini-BALof time scales. However, the profile parameters do HS1603+3820. Thismaybeanalogoustothebest-studied not appear to change in concert with each other, warm absorber, in the Seyfert galaxy NGC 3783, where withoneexception: theequivalentwidthsofallthe fluctuations in the ionization structure of the absorbing troughs in this system vary together and approxi- gas have been observed over the course of 6 months (e.g., matelyfollowthevariationsofthecoveragefraction Netzeretal. 2003). Thereisalsoevidencethattheopacity determined from the model fits. of the warm absorber in NGC 3783 fluctuates over time 3. We have examined a number of ways of explain- intervals as short as a month, in response to variations in the X-ray continuum by a factor of 2 (Krongold et al. ing the above variations of the CIV mini-BAL and we have found two viable possibilities. The first 2005). Within this context we can also understand why possibilityisthattheobservedpartialcoveragesig- thevariationsinthecoveragefractionofthemini-BALgas nature is the result of continuum photons scatter- are unrelated to variations of its column density. Because ing around the absorber and into our cylinder of both media can be clumpy and since differential rotation sight. The observed changes in mini-BAL equiv- of the wind causes the inner screen to move faster than alent widths are thus produced by variations in the mini-BAL gas across the cylinder of sight, the inten- thescatteredcontinuumthatdilutestheabsorption sity of the continuum transmitted throughthe screen and troughs. In the second possibility, the illumination receivedbythemini-BALgaswouldvarywithtime. More- of the UV absorber fluctuates on short time scales over,thecoveragefractionneednotbecorrelatedwiththe by a factor of up to 3. We suggest that these fluc- intensity of the transmitted continuum. tuations are caused by a screen of variable optical We have considered the possibility that one of the ob- depth between the mini-BAL gas and the contin- served NAL systems at smaller blueshifts relative to the uum source. This screen might be identified with quasarcouldrepresentthe inner screen. However,none of the shielding gas invokedor predicted in some out- the three candidate C IV systems, B, C, and D, can sat- flowmodels. Moreover,itcouldbeanalogoustothe isfy the requirements for the screen. The very fact that “warm”absorbersobservedin the X-ray spectra of they have CIV absorptionlines suggeststhat their ioniza- Seyfertgalaxiesandsomequasars. Thispicturecan tionstateistoolow,andmoreover,systemsBandDhave also explain the variations in the coveragefraction, even lower-ionizationlines. In addition, none of the three which appear to be unrelated to the ionic column systems are variable, which is necessary in order for the density changes. screen to perform the desired function. More promising waysoffindingthescreensignatureareX-rayobservations in search of the classic warm absorber feature, O VII and O VIII edges. Thus, we will be able to determine which This workwassupportedby NASA grantNAG5-10817. of the Lyα absorption lines seen at shorter wavelengths Wearegratefultothe staffoftheSubarutelescope,which might be associated with the screen material. is operated by the National Astronomical Observatory of Japan. We wouldalsolike tothank ChristopherChurchill 6. summary and conclusions for providing us with the minfit software package. ME We have been monitoring the absorption lines in the acknowledges partial support from the Theoretical As- spectrum of the quasar HS 1603+3820 since 2002, us- trophysics Visitors’ Fund at Northwestern University and ing the Subaru and Hobby-Eberly Telescopes. We have thanks the members of the theoretical astrophysics group obtained six Subaru+HDS and two HET+MRS spectra for their warm hospitality. The Hobby-Eberly Telescope spanning an interval of approximately 1.2 years in the (HET) is a joint project of the University of Texas at quasar rest frame and probing rest-frame time scales as Austin, the Pennsylvania State University, Stanford Uni- short as 16 days. We have determined the physical pa- versity, Ludwig-Maximillians-Universit¨at Mu¨nchen, and rameters of the 9 absorption-line systems by fitting Voigt Georg-August-Universit¨atG¨ottingen. The HET is named modelstothelineprofiles. Ourmainresultsareasfollows: inhonorofitsprincipalbenefactors,WilliamP.Hobbyand Robert E. Eberly. We also thank the anonymous referee 1. We detect one CIV mini-BAL and eight CIV NAL for helpful comments and suggestions. Dramatically Variable Mini-BAL in HS 1603+3820 9 REFERENCES Arav, N., Korista, K.T., Barlow, T.A., & Begelman, M.C., 1995, Kaspi,S.,Brandt,W.N.,Maoz,D.,Netzer,H.,Schneider,D.P.,and Nature,376,576 Shemmer,O.,2007, ApJ,inpress(astro-ph/0612722) Arnaud,M.,&Rothenflugh, R.,1985,A&AS,60,425 Kaastra,J.S.,Mewe,R.,Liedahl,D.A.,Komossa,S.,&Brinkman, Bachev, R.,Strigachev,A.,&Semkov, E.,2005, MNRAS,358,774 A.C.2000, A&A,354,L83 Barlow,T.A.,&Sargent,W.L.W.,1997,AJ,113,136 Krongold,Y.,Nicastro,F.,Brickhouse, N.,Elvis,M.,&Mathur,S. Barlow,T.A.,1993,Ph.D.thesis,UniversityofCalifornia,SanDiego, 2005, ApJ,622,842 (1994, PASP,106,548) deKool,M.Korista,K.T.,&Arav,N.2002,ApJ,580,54 Barlow, T.A., Junkkarinen, V.T., Burbidge, E.M., Weymann, R.J., Lamy,H.&Hutsem´ekers,D.2000,A&A,356,L9 Morris,S.L.,&Korista,K.T.,1992, ApJ,397,81 Lundgren, B.F., Wilhite, B.C, Brunner, R.J, Hall, P.B., Schneider, Becker, R.H., Gregg, M.D., Hook, I.M., McMahon, R.G., White, D.P., York, D.G., Vanden Berk, D.E., & Brinkmann, J., 2006, R.L.,&Helfand,D.J.,1997, ApJ,479,L93 astro-ph/0610656 Blustin, A. J., Branduardi-Raymont, G., Behar, E., Kaastra, J. S., Ma,F.,2002,MNRAS,335,99L Kahn, S. M., Page, M. J. Sako, M., & Steenbrugge, K. C. 2002, Misawa,T.,Charlton, J.C.,Eracleous,M.,Ganguly, R.,Tytler,D., A&A,392,453 Kirkman,D.,Suzuki,N.,&Lubin,D.,2007, ApJ,submitted Brotherton, M. S., Tran, H. D., van Breugel, W., Dey, A., & Misawa,T.,Eracleous,M.,Charlton,J.C.,&Tajitsu,A.,2005,ApJ, Antonucci, R.1997,ApJ,487,L113 629,115 Chiang,J.&Murray,N.,1996,ApJ,466,704 Misawa, T., Yamada, T., Takada-Hidai, M., Wang, Y., Kashikawa, Churchill,C.W.,Vogt,S.S.,&Charlton,J.C.,2003,AJ,125,98 N.,Iye,M.,&Tanaka,I.,2003,AJ,125,1336 Churchill,C.W.,Schneider,D.P.,Schmidt,M.,&Gunn,J.E.,1999, Misawa,T.,Tytler, D., Iye, M.,Storrie-Lombardi,L.J., Suzuki, N., AJ,117,2573 &Wolfe,A.M.,2002,AJ,123,1863 Churchill,C.W.,1997,Ph.D.thesis,Univ.California,SantaCruz Murray, N., Chiang, J., Grossman, S.A., & Voit, G.M., 1995, ApJ, Crenshaw, D.M., Kraemer, S.B., Boggess, A., Maran, S.P., 451,498 Mushotzky, R.F.,&Wu,C.-C.,1999,ApJ,516,750 Narayanan, D., Hamann, F., Barlow, T., Burbidge, E.M., Cohen, Dobrzycki, A., Engels, D., & Hagen, H.-J., 1999, A&A, 349, L29 R.D.,Junkkaribeb, V.,&Lyons,R.,2004, ApJ,601,715 (D99) Netzer,H.etal.,2003,ApJ,599,933 Dobrzycki, A., Engels, D., Hagen, H.-J., Elvis, M., Huchra, J., & Noguchi, K., Aoki, W., Kawanomoto, S., Ando, H., Honda, S., Reimers,D.,1996,BAAS,188.0602 Izumiura, H., Kambe, E., Okita, K., Sadakane, K., Sato, B., Elvis,M.,2000,ApJ,545,63 Tajitsu, A., Takada-Hidai, M., Tanaka, W., Watanabe, E., and Everett, J.E.,ApJ,2005,631,689 Yoshida,M.,2002,PASJ,54,6 Fleming,B.T.,&Kennefick, J.,2006,BAAS,141,01 Ogle, P. M.,Cohen, M.H.,Miller,J. S.,Tran, H. D.,Goodrich, R. Foltz, C.B.,Weymann, R.J., Morris,S.L., & Turnshek, D.A.,1987, W.,Martel,A.R.1999, ApJS,125,1 ApJ,317,450 Petitjean,P.&Srianand,R.,1999,A&A,345,73 Ganguly,R.,Eracleous,M.,Charlton,J.C.,&Churchill,C.W.,1999, Proga,D.,Stone,J.M.,&Kallman,T.R.,2000,ApJ,543,686 AJ,117,2594(G99) Richards,G.T.,2001, ApJ,133,53 Ganguly, R., Bond, N.A., Charlton, J.C., Eracleous, M., Brandt, Sako,M.etal.2001,A&A,365,L168 W.N.,&Churchill,C.W.,2001, ApJ,549,133(G01) Sargent, W.L.W., Boksenberg, A.,& Steidel, C.C.,1988, ApJS, 68, Ganguly, R., Masiero, J., Charlton, J.C., & Sembach, K.R., 2003, 539 ApJ,598,922 Scannapieco, E.&Oh,S.P.2004,ApJ,608,62 Giveon, U., Maoz, D., Kaspi, S., Netzer, H., & Smith, P. S. 1999, Smith,L.J.&Penston,M.V.,1988,MNRAS,235,551 MNRAS,306,673 Springel,V.,DiMatteo, T.,&Hernquist,L.2005,ApJ,620,L79 Granato, G.L., De Zotti, G. Silva, L., Bressan, A., & Danese, L., Srianand,R.&Petitjean,P.,2000,A&A,357,414 2004,ApJ,600,580 Steidel,C.C.,1990,ApJS,72,1 Hagen, H.-J., Groote, D., Engels, D., & Reimers, D., 1995, A&AS, Turnshek,D.A.,Grillmair,C.J.,Foltz,C.B.,&Weymann,R.J.,1988, 111,195 ApJ,325,651 Hamann,F.1997, ApJS,109,279 Vestergaard,M.,2002, ApJ,571,733 Hamann,F.,Barlow,T.A.,&Junkkarinen, V.,1997a, ApJ,478,87 Vilkoviskij,E.Y.&Irwin,M.J.,2001,MNRAS,321,4 Hamann, F., Barlow, T.A., Junkkarinen, V., & Burbidge, E.M., Weymann,R.J.,Morris,S.L.,Foltz,C.B.,&Hewett,P.C.,1991,ApJ, 1997b,ApJ,478,80 373,23 Hamann,F.,&Sabra,B.,2004,2004,ASPC,311,203 Wise,J.H.,Eracleous,M.,Charlton,J.C.,&Ganguly,R.,2004,ApJ, Hawkins,M.R.S.,2001, ApJ,553,97 613,20 Horner,S.D.,Engel,L.&Ramsey,L.W.,1998,SPIEConf.3355,375 Yuan, Q., Green, R.F., Brotherton, M., Tripp, T.M., Kaiser, M.E., &Kriss,G.A.,2002,ApJ,575,687 10 Misawa et al. Table 1 Log of Monitoring Observations Timea WavelengthCoverage Exposure Resolving Epoch Date (years) Instrument (˚A) (s) Power S/Nb (1) (2) (3) (4) (5) (6) (7) (8) 1c 2002Mar23 0.00 Subaru+HDS 5080–6420 2700 45,000 46.5 2c 2003Jul7 0.36 Subaru+HDS 3520–4850,4930–6260 6000 45,000 48.7 3 2005Feb26 0.83 Subaru+HDS 3520–4855,4925–6275 7100 36,000 43.0 4 2005May10 0.89 HET+MRS 4500–6200 1800 9,200 17.4 5 2005Jun29 0.92 Subaru+HDS 3530–4855,4925–6280 3600 45,000 31.1 6 2005Aug3,8 0.95 HET+MRS 4500–6170 4000 9,200 28.5 7 2005Aug19 0.96 Subaru+HDS 3520–4850,4920–6235 3600 36,000 43.6 8 2006May31–Jun1 1.18 Subaru+HDS 4320–5630,5740–7050 9000 45,000 52.5 aTherelativetimesincethefirstobservation,inthequasarrestframe. bS/Nperbinaround5370˚Aafterrebinning. Thefinalbinsizeis0.03˚AfortheSubaru+HDSspectraand0.15˚Aforthe HET+MRSspectra. cThesespectraweoriginallypresentedanddiscussedbyMisawaetal. (2003,2005).