The first detection of [OIII] emission from high-redshift damped Lyman-alpha galaxies

0.88 MB·English

by  S. J. Weatherley

Mon.Not.R.Astron.Soc.000,1–14(2004) Printed2February2008 (MNLATEXstylefilev2.2)
iii
The first detection of [O ] emission from high–redshift
α ⋆
damped Lyman– galaxies
S. J. Weatherley1†, S. J. Warren1, P. Møller2, S. M. Fall3, J. U. Fynbo4,
S. M. Croom5
5
0 1Astrophysics Group, Blackett Laboratory, Imperial College London, Prince Consort Road, London SW7 2BW, UK
0 2European Southern Observatory, Karl-Schwarzschild-Strasse 2, D-85748 Garching bei Mu¨nchen, Germany
2 3Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD21218, USA
4Institute of Physics and Astronomy, Universityof ˚Aarhus, DK-8000 ˚Aarhus C, Denmark
n
5The Anglo-Australian Observatory, PO Box 296, Epping, NSW 2121, Australia
a
J

9
Accepted 0000January00.Received0000January00;inoriginalform0000January00
1

1
v ABSTRACT
2
2 We present the detection of[Oiii] emissionlines from the galaxiesresponsible for
4 two high–redshift z > 1.75 damped Lyman–α (DLA) absorption lines. We find two
1 sources of [Oiii] emission corresponding to the z = 1.92 DLA absorber towards the
0 quasar Q2206−1958, and we also detect [Oiii] emission from the galaxy responsible
5 for the z = 3.10 DLA absorber towards the quasar 2233.9+1381. These are the first
0
detectionsofrest–frameopticalemissionlinesfromhigh–redshiftDLAgalaxies.Unlike
/
h the Lyαline, the [Oiii]line providesa measureof the systemic velocity ofthe galaxy.
p We compare the [Oiii] redshifts with the velocity profile of the low–ionisation metal
- lines in these two absorbers, with the goal of distinguishing between the model of
o
Prochaskaand Wolfe ofDLA absorbersas large rapidly rotating cold thick discs,and
r
t the standard hierarchical CDM model of structure formation, in which DLAs arise
s
a in protogalactic fragments. We find some discrepancies with the predictions of the
: formermodel.Furthermoretheimageofthe DLAgalaxytowardsQ2206−1958shows
v
acomplexdisturbedmorphology,whichismoreinaccordwiththehierarchicalpicture.
i
X We use the properties of the rest–frame optical emission lines to further explore the
r questionposedbyMølleretal.:arehigh–redshiftDLAgalaxiesLyman–breakgalaxies
a (LBGs) selected by gas cross section? The measured velocity dispersions of the DLA
galaxies are in agreement with this picture, while the data on the [Oiii] luminosities
andthevelocitydifferencesbetweentheLyαand[Oiii]linesareinconclusive,asthere
areinsufficientLBGmeasurementsoverlappinginluminosity.Finally weestimate the
star formation rates in these two DLA galaxies, using a variety of diagnostics, and
include a discussion of the extent to which the [Oiii] line is useful for this purpose.
Key words:
galaxies: kinematics and dynamics – galaxies: formation – galaxies: high redshift –
quasars: absorption lines – quasars: individual(Q2206−1958,2233.9+1381)
1 INTRODUCTION study the cosmic history of star formation (e.g. Pei, Fall,
and Hauser, 1999), and of metal production (Kulkarni and
The damped Lyman–α (DLA) absorption lines detected in
Fall, 2002). Weare engaged in a programme toidentify the
the spectra of quasars identify gas clouds that contain the
galaxies (hereafter‘DLAgalaxies’) responsible fortheDLA
majority of the neutral hydrogen in the Universe. Analysis
lines at high redshifts, z > 1.75 (Warren et al., 2001; here-
of statistical samples of DLA absorbers has been used to
after W01). The principal goal of this programme is to es-
tablish the connection between the population of DLA ab-
⋆ Based on observations made at the European Southern Ob- sorbers, and galaxy populations identified at high redshifts
in deep imaging studies, firstly by comparing the measured
servatory Very Large Telescope, Paranal, Chile (ESO Programs
63.O-0618and65.O-0707) propertiesofthedetectedDLAgalaxies withtheproperties
† Email:[email protected]
2 S. J. Weatherley, et al.
of other galaxy populations (Møller et al., 2002; hereafter provides the transformation between the luminosity distri-
M02), and secondly by measuring the gas cross sections butions of the two populations (see Fynbo et al. 1999 for
of the absorbers, thereby establishing their space density furtherexplanation).
(Fynbo,Møller, & Warren, 1999; Chen & Lanzetta, 2003). In thispaper we report the detection of rest–frame op-
DLAgalaxies aredifficulttoidentifybecausetheytyp- tical emission lines from two DLA galaxies, with near–ir
ically lie at angular separations of order one arcsec from spectroscopy. These are the first detections of this kind2.
thequasarlineofsight,andarethereforeswampedbylight The galaxies are two of the three studied in M02 (for the
from the quasar. We are aware of only eight high–redshift third, all the strong rest–frame optical emission lines lie at
DLA galaxies with published spectroscopic confirmation of highly unfavourable wavelengths for observation). We use
the redshift1. For reference, we have summarised details of these results to extend our comparison of the properties of
thesedetectionsinTable1.Insuccessivecolumnsarelisted DLAgalaxiesandLBGs(ofsimilarabsolutemagnitudeand
(1) the quasar name; (2) the quasar redshift zQSO; (3) the redshift) to include the rest–frame optical line luminosities
DLA redshift zDLA; (4) a flag Y if zQSO ≫zDLA, N other- and widths.
wise (zDLA ≈ zQSO); (5) the column density log10NHI; (6) In the CDM scheme for the formation of structure in
a flag Y if log10NHI >20.3 (the definition of DLA of Wolfe theUniverse,galaxiesgrowhierarchically,andDLAabsorp-
et al. 1986), N otherwise; (7) the metallicity, if an accurate tion lines at z ∼ 2−3 arise when sightlines pass through
Si or Zn value is available; (8) the reference for the metal- protogalactic fragments. ProchaskaandWolfe(1997b) have
licity; (9) the Lyα luminosity; and (10) thereference of the argued for a very different picture, showing that that the
discoverypaper.WetreatallthesesourcesasDLAgalaxies, detailed velocity structure of the low–ionisation metal ab-
but note that different individuals adopt narrower defini- sorption lines in DLA systems is consistent with theexpec-
tions of a DLAabsorber dependenton theflags in columns tation for sightlines passing through large rapidly–rotating
(4)and(6),sothatunderthenarrowestdefinitiononlythree coldthickdiscsofneutralgas.However,itwassubsequently
sourcesareDLAgalaxies.Foradiscussionofthecaseswhere shown by Haehnelt, Steinmetz, and Rauch (1998) that the
zDLA ≈zQSO, see Møller, Warren, and Fynbo (1998). Five absorption–line kinematic data alone do not permit an un-
of the DLA galaxies listed in Table 1 were discovered with ambiguousinterpretation,andareequallywellexplainedby
telescopes of 4m–class, or smaller (the remaining three are thedynamicsofmergingprotogalacticfragments.Ourdetec-
from the current programme). Motivated by the high spa- tions of rest–frame optical emission lines from DLA galax-
tial resolution of HST, and the large light–gathering power iesprovideimportantadditionalcluesfortheinterpretation
of 8m–class telescopes, we are undertaking a survey to de- of the kinematics of these systems, supplementing the in-
tect a significant sample of DLA galaxies (W01). We have formation provided by the absorption–line velocity profiles,
obtained deepHST NICMOSandSTISimages ofthefields and casting new light on this long–standing debate. In the
of16quasars,aimedatthedetectionofcounterpartgalaxies same context we note that additional evidence in favour of
of18z >1.75DLAabsorbersandfiveLyman–limitsystems. the hierarchical picture comes from recent observations of
In theNICMOS images we found 41 candidate DLA galax- the evolution of the sizes of galactic stellar disks (Ferguson
ies brighter than HAB ∼25, within a box of side 7.5arcsec et al., 2004, Bouwens et al., 2004), and comparison against
centred on each quasar (W01). We are using the VLT and predictions of the extent of the baryons in galaxies, as a
Geminitelescopes toobtain confirmatory opticaland near– function of redshift (Fall and Efstathiou, 1980, Mo, Mao,
ir spectra of thesecandidates. and White, 1998). All the same these analyses rely on as-
In M02, we reported preliminary results from this pro- sumptionsabouttheextenttowhichstarsmapthebaryons
gramme. We tested the hypothesis that DLA galaxies are at any redshift. The analysis of the kinematics of DLA ab-
Lyman–breakgalaxies,selectedbygascrosssection,bycom- sorbers complements this approach.
paringseveralemissionproperties(size,colour,etc.)ofthree The layout of the paper is as follows: in Section 2 we
high–redshift DLA galaxies with the emission properties of describetheobservationsandthedatareduction;inSection
Lyman–break galaxies of similar absolute magnitude and 3 we present the measured properties of the detected lines;
redshift. We found no significant differences and concluded in Section 4 we analyse the kinematics of these systems,
that the data are consistent with the hypothesis posed. It andinSection5wecomparethemeasuredpropertiesofthe
must be appreciated, though, that the hypothesis that the detectedgalaxiesagainst thepropertiesofLBGs.Insection
two populations are drawn from the same parent popula- 6 we estimate the star formation rates in these two DLA
tion,withdifferentselectioncriteria,allows forthepossibil- galaxies, usingavarietyofdiagnostics. Section7providesa
ity that the average properties of the population (e.g. their summary of the main results of the paper. Throughout, we
clustering amplitude) will differ. This will be true, for ex- assume a standard, flat ΛCDM cosmology with Ω = 0.7
Λ
ample, for any quantity that depends on luminosity, since and H =70kms−1Mpc−1. For this cosmology an angle of
0
the average luminosities of the two populations differ, due 1 arcsec corresponds to the physical size 8.4 kpc at z = 2,
to the way they are selected. As noted by M02, the key and 7.7 kpcat z=3.
to confirming the hypothesis conclusively is the detection
of more DLA galaxies, which will allow the measurement
of the relation between gas radius and galaxy luminosity,
the Holmberg relation Rgas ∝ Lt. The Holmberg relation
1 as well as three unpublished (Djorgovski, private communica- 2 Thedetection ofHβ and[Oii]reportedbyElstonetal.,1991,
tion) wasnotconfirmed(Lowenthal, privatecommunication).
[Oiii] emission from DLA galaxies 3
Table 1.Summaryofspectroscopicallyconfirmedhigh–redshiftDLAgalaxies
1 2 3 4 5 6 7 8 9 10
Quasar zQSO zDLA zQSO≫zDLA log10(NHI) log10(NHI>20.3) [M/H] M(refs) LyαLum Original
cm−2 cm−2 ×1042 ergs−1 Discovery
PHL1222 1.922 1.9342 N 20.36 Y — — ∼9.3 [12]
PKS0458−02 2.286 2.0395 Y 21.65 Y −1.17,−1.19 Zn,Zn[5],[14] 1.6+−00..62 [9]
PKS0528−250 2.797 2.8110 N 21.35 Y −0.75,−0.76 Si,Zn[7],[5] 5.2±0.4 [10]
PC0953+4749 4.457 3.407 Y 21.2 Y >−2.09 Si[15] ∼1.1 [1]
Q2059−360 3.097 3.0825 N 20.85 Y — — ∼17[a,6] [13]
Q2206−1958 2.559 1.9205 Y 20.65 Y −0.42,−0.39 Si,Zn[14],[5] 6.8±0.8[b] [11]
2233.9+1318 3.298 3.1501 Y 20.00 N −1.04to−0.56 Si[c,8] 5.6±1.0 [2]
DMS2247−0209 4.36 4.097 Y — — — — 0.9±0.2[4] [3]
Notes: [a]assumes atotal flux2.0×10−16 ergs−1cm−2,[b]thisistheluminosityofN-14-1C, [c]lower and upperlimitsprovidedbyref.[8].Allvaluesin
columns2, 3, 5 are taken from thesummary of Warren et al. (2001), except for Q2059−360 (taken from [7]), and for DMS2247−0209 (taken from [3]).
Thefluxesusedtocomputetheluminositiesaretakenfromthediscoverypaper,unlessreferencedotherwise.Referencesareasfollows:[1]Bunkeretal.(2005,
inprep.),[2]Djorgovski et al.(1996), [3]Djorgovski etal.(1998), [4]Djorgovski (privatecommunication),[5]KulkarniandFall(2002), [6]Leibundgutand
Robertson(1999),[7]Luetal.(1996),[8]Luetal.(1998),[9]Møller,Fynbo&Fall(2004),[10]Møller&Warren(1993),[11]Mølleretal.(2002),[12]Møller,
Warren&Fynbo(1998),[13]Pettinietal.(1995),[14]Prochaskaetal.(2003a),[15]Prochaskaetal.(2003b).
2 OBSERVATIONS AND DATA REDUCTION textthemeaningof‘compact’isthattheS/N ofthedetec-
tion, integrated over an aperture, is greater for the smaller
2.1 Observations apertureused,diameter0.45arcsec,thanforthelargeraper-
tureused,diameter0.90arcsec.SuccessivecolumnsinTable
The observations were taken with the Infrared Spectrom- 2 list the date of observation, the total integration time,
eter and Array Camera (ISAAC) instrument on the Eu- thewavelengthrangecoveredbythespectrum,theslitposi-
ropean Southern Observatory’s 8m UT1 telescope at the tionangle,andtheaverageseeing.Theobservationsof2000
Very Large Telescope (ESO–VLT). We used the medium– Aug 2 were affected by cloud, while conditions were clear
resolution (MR) mode, with a 1arcsec slit, to obtain near– for the other two nights. Further details of the target DLA
ir spectra of a number of targets in the NICMOS candi- absorbers,andthecandidateDLAgalaxies, includingaccu-
date list from W01. Results for three candidates are re- rate coordinates, are provided in W01. Notes on the three
ported here. Results for other candidates will be reported candidates follow.
elsewhere. We observed two candidates in the field of the Quasar Q2206−1958, candidates N-14-1C, N-14-2C:
quasar Q2206−1958, in the H band, where the MR mode The spectrum of this quasar shows two high–redshift DLA
provides a wavelength coverage of 0.079µm, with 1024 pix- absorbers, of redshifts z = 1.9205, and z = 2.0762. Our
els, at a resolving power of 2700 for this slit, corresponding NICMOS observations of this field revealed two compact
to 7 pixels. The third target was theknown DLA galaxy in candidateswithimpactparametersb<2.0arcsec.Thenear-
thefieldofthequasar2233.9+1318 (Djorgovskietal.,1996, est candidate to the line of sight to the quasar, N-14-1C,
and seeTable1), which was observedin theKband,where has impact parameter b = 1.13arcsec, and was detected
the MR mode provides a wavelength coverage of 0.122µm, at S/N = 6.5. Nevertheless this source was considered a
at aresolving powerof 2750, corresponding to7pixels.The marginaldetectionsincetheimageislocatedonadiffraction
spatialscaleis0.147arcsecperpixel.Forspectroscopyinthe spike. The second nearest candidate, N-14-2C, has impact
near–ir it is common practice to nod the telescope between parameterb=1.33arcsec,andwasdetectedatS/N =12.1.
twopositions AandB,insequenceABBAetc.First–order AsreportedinM02,wesucceededindetectingLyαemission
skysubtractionisachievedbysubtractingtheaveragespec- from N-14-1C, confirming that it lies at the redshift of the
trum at position B from the average spectrum at position DLA absorber at z = 1.9205. We also observed N-14-2C,
A,andviceversa.Instead,weplacedthegalaxyatsixposi- butfailed tosecurearedshift.ThesourceisredderthanN-
tions along the slit, ABCDEF, each separated by 7arcsec. 14-1C, isunresolvedin theSTISandNICMOSimages, and
Foreachslitposition,first–orderskysubtractionatanypo- we assumed it was unrelated3. In this paper we show that,
sition is achieved by subtracting the mean of the frames at in fact, thetwo sources are at thesame redshift.
alltheotherskypositions.Forobservationsoffainttargets, The STIS image of this field, provided in M02, detects
thisprocedureshould lead toan increase in signal–to–noise both candidates at much higher S/N, by about a factor of
(hereafter S/N) of a factor p2(N−1)/N where N is the five. This image was taken without a filter i.e. in ‘50CCD’
numberof slit–positions; for six slit–positions we expect an mode. Another version of this image is reproduced here in
increaseinS/N of∼1.3comparedtothenormalprocedure. Fig.1,rotatedsothatNisup(andEtotheleft).Thedeeper
The journal of observations is provided in Table 2. STIS image reveals a more complex morphology than the
Columns 1 and 2 list the names of the quasar, and of the NICMOS image. In addition to the two compact sources,
target DLA galaxy candidate. The target names are taken
fromW01,andmaybedecodedfromtheexampleN-14-2C:
N=NICMOS candidate, 14=14th quasar in the list of 16 3 Because of the wider slit used, 1.3arcsec, and the poor see-
quasars targeted with NICMOS, 2=2nd nearest candidate ingconditions, although weak Lyα emissionwas detected inthe
to the line of sight to the quasar, C=compact morphology spectrum of N-14-2C, the flux was consistent with spillage from
(asopposed toD=diffuse).Asdetailed in W01,in thiscon- N-14-1C.
4 S. J. Weatherley, et al.
some diffuse emission surrounds N-14-1C, and extends to
smaller impact parameters.
For the field of the quasar Q2206−1958, the choice of
central wavelength for the two observations attempted to
maximise the chances of successful detections, from consid-
erationofseveralfactors:thewavelengthsandprobablerela-
tivestrengthsofthevariousrest–frameopticalemissionlines
attheredshiftsofthetwoabsorbers;thewavelengthregions
ofstrongatmosphericabsorption;theresultsfrom theopti-
calspectroscopy;thewavelengthcoverageoftheinstrument;
andthepossibilityofothergalaxiesundetectedintheimag-
ing observations, because at very small impact parameter.
Forbothobservationstheslitwascentredonthequasarand
rotatedtothepositionangleofthecandidate.Wechosethe
samewavelengthrangeforbothslitorientations,whichpro-
videdsimultaneouscoverageof[Oiii]500.7atz=1.9205(at
1.462µm), and Hβ and [Oiii]495.9 at z = 2.0762 (at 1.495
and1.525µm).ThepurposeoftheobservationatPA+25.45,
then, was to detect the [Oiii]500.7 line from the confirmed
galaxy N-14-1C, while also allowing the possibility of de-
tectingthegalaxycounterparttothez=2.0762 absorberif
it lies at very small impact parameter. The purpose of the
observation at PA−13.05, was to search for [Oiii] emission
at either redshift from N-14-2C, as well as to search again
forthez=2.0762galaxyatverysmallimpactparameter.It
wouldhavebeenpossibletoobservebothtargetswithasin-
gleslitorientation,offsetfromthequasar;however,weused
two slit orientations as we considered it important to check
foremission atverysmallimpactparameter.Inaddition,in
tryingtodetectfaintemissionlinesfromagalaxywherethe
spectrum overlaps spatially with the quasar spectrum, and
wherethequasarspectrummustbesubtracted(seebelow),
ourexperienceisthatthisworksbestwhentheslitisaligned
along theline joining thequasar and the galaxy.
Quasar 2233.9+1381, candidate N-16-1D: The spec-
trum of this quasar shows a strong Lyα absorption line,
log10(NHI)=20.0, atz =3.1501. BytheDLAdefinitionof
Wolfe et al. (1986), log10(NHI) > 20.3, this is not a DLA
absorber,butaLyman–limitsystem.Wehaveincludeditin
Table1,sincethemeasuredcolumndensityliesonlyalittle
belowthisthreshold.Inthispaperwetreatthecounterpart
galaxy as representative of DLA galaxies. Nevertheless it
Figure1. Sub–sectionoftheSTIS50CCDimageofthefieldto- shouldbeborneinmindthattheionisation stateofthegas
wardQ2206−1958afterquasarpsfsubtraction.Theupperimage in the absorber is dependent on column density (e.g. Vie-
isthesameasthelowerimage,butwithouttheannotation. Nis gas, 1995), and in this respect this absorber lies at one end
upandEtotheleft.ThesmoothregionintheupperLHcorner of the distribution of the absorbers listed in Table 1. The
contains no data in this image, because the image was formed nearest NICMOS candidate N-16-1D, was first detected in
fromasmallregionoftheSTISfield,andthenrotated toorient
anopticalimageandidentifiedasacandidatecounterpartto
itN–S.Thetwocandidates N-14-1CandN-14-2C,aswellasthe
the absorber by Steidel, Pettini, and Hamilton (1995), who
locationofthequasar,areindicatedbyarrows.Theringcentred
called it N1. The galaxy was confirmed as the counterpart,
onthelocationofthequasarmarksaradiusof0.4arcsec.Thepsf
subtraction is considered to be satisfactory outside this radius, on the basis of the detection of Lyα emission at the DLA
and unsatisfactory inside. Details of the psf subtraction proce- absorber redshift, by Djorgovski et al. (1996). We did not
dureusedwillbeprovidedelsewhere(Mølleretal.,inprep.).The detectanycandidatesatsmallerimpactparameterineither
skycoverageofthe1arcsecslitinthetwoorientations isshown. ourNICMOSimage(W01)orourSTISimage(M02)ofthis
Thepixelsizeis0.056arcsec. ThepixelsizeintheoriginalSTIS field.Thewavelength range of ourobservations covered the
images is 0.050arcsec. The image is a combination of four im- threeredshiftedlinesHβ486.1(at2.017µm),[Oiii]495.9(at
ages taken at two different spacecraft roll angles, totaling 5084s 2.058µm),and[Oiii]500.7(at2.078µm).Theimpactparam-
integration. An approximate 5σ depth of this image, for a point
eterofthisgalaxy,2.78arcsecinourNICMOSimage,isthe
source,wellawayfromthequasarcentroidisV50=27.65. largest of all theeight confirmed galaxy counterparts listed
inTable1.Wealignedtheslittocoverboththequasarand
the galaxy, again to check for the possibility that another
[Oiii] emission from DLA galaxies 5
galaxy at the absorber redshift lies at very small impact
parameter.
2.2 Data reduction
For each source, for first–order sky subtraction we followed
the procedure explained in the previous section. If the six–
step–nod was repeated, the frames were reduced in sets of
six. For six frames taken at positions ABCDEF, from each
flat–fieldedframewesubtractedtheaverageoftheotherfive
frames.Second-orderskysubtractionwasachievedbyfitting
a polynomial up each column. The sky–subtracted frames
werethenregisteredtothenearestpixel,spatiallyandspec-
trally,andcombinedusinginverse–varianceweighting.Inte-
ger pixel shifts were used in order to keep the data in each
pixelstatisticallyindepedent.Asacheckofoursix–step–nod
observing procedure, we also reduced the data in the tradi-
tionalway,bycombiningpairsofnodpositions.Thisverified
that the six–step–nod procedure achieved the expected im-
provement in S/N. To establish the uncertainties, for each
final2Dspectrumweformedacorrespondingvarianceframe
intwoways.Inthefirstmethodweregisteredtherawframes
usingthesameoffsetsasforthesky–subtractedframes. We
then summed the counts at each pixel, and computed the
variance as appropriate, from a knowledge of the measured
gain and read noise, and assuming Poisson statistics. The
second method measured the dispersion in the counts up
eachcolumninthefinalcombinedsky–subtractedframe,us-
ing a robust estimator. This provides an accurate measure
of theaverage noise in the sky,but gives noinformation on
Figure2. Two–dimensionalspectraofQ2206−1958bothbefore
the noise for individual pixels. The second method gave re- andafterquasarremoval.Theverticalbandsofhighernoisemark
sults some 20 to 30percent higher for the variance (10 to the wavelengths ofstrong OH skylines.From top to bottom: a)
15percent for the standard deviation) than the Poisson es- finalcombined2DspectrumofobservationofcandidateN-14-1C,
timate; therefore we scaled the Poisson–estimated variance showing the quasar spectrum. b) the same, after subtraction of
framebythiscorrection factortoproducethefinalvariance the quasar spectrum using our SPSF method, and smoothing,
frame. revealing the emission line from the DLA galaxy N-14-1C. c) fi-
The combined frame formed from the registered raw nalcombined2Dspectrum ofobservation ofcandidate N-14-2C,
showingthequasarspectrum.Thegalaxyemissionlineisvisible
frames, referred to above, emulates the registration of the
inthisframe,asanextensionbelowthequasarspectrum.d)the
sky–subtractedframes,andthereforeprovidedthemeansof
same, after subtraction of the quasar spectrum using our SPSF
accuratewavelength calibration,usingtheskyOHemission
method,andsmoothing,moreclearlyrevealingtheemissionline
lines. This frame also provided the means of measuring the
fromtheDLAgalaxyN-14-2C.Thewavelengthscaleisthesame
spectral resolution, from the widths of the sky lines, and forall the frames. The velocity offset between N-14-1C inframe
yielded the valuesquoted above. b) and N-14-2C in frame d) is clearly visible. The emission line
visible in frame d) displays a faint but significant extension to
redder wavelengths and smaller impact parameters. The wave-
length of this extension matches the wavelength of the emission
3 RESULTS linefromN-14-1Cvisibleinb),andisinterpretedaslightleaking
intotheslitfromthissource.Thiswouldbeexpected,asmaybe
3.1 Reduced 2D frames
appreciatedfrominspectionofFig.1.
Quasar Q2206−1958, candidates N-14-1C, N-14-2C: The
final 2D frames of the spectroscopic observations of
Q2206−1958 are shown in Fig. 2. Because of the poorer total wavelength range observed, for both candidates, both
seeing, shorter integration time, and non–photometric con- before and after subtraction of the quasar spectrum. From
ditions,theS/N ofthespectrumofN-14-1C(toptwopanels toptobottom,thepanelsshowa)N-14-1C,raw,b)N-14-1C,
in Fig. 2) is nearly a factor of two lower than for N-14-2C afterquasarsubtraction,andsmoothingtoenhancethecon-
(bottomtwopanels).Anemission linewasdetectedathigh trast of the line, c) N-14-2C, raw, d) N-14-2C, after quasar
significance from both candidates, with centroid below the subtraction and smoothing. Themeasured spatial offsets of
quasarspectrumineachframe.Forbothcandidatestheim- the emission lines from the quasar are provided in Table
pact parameter is small, and the quasar and galaxy spec- 3, together with the offsets measured in the NICMOS and
tra overlap. Therefore we used our SPSF software (Møller, STISimages.Thespectroscopicandimagingoffsetsarecon-
2000) to subtract thequasar spectrum, as described below. sistent, confirming that the emission is associated with the
Fig. 2 provides the 2D spectra over approximately half the targeted candidates.TheSPSFroutineworksbyfirstly de-
6 S. J. Weatherley, et al.
Table 2.Observationlogforspectroscopicobservations atVLT
Quasar DLAcandidate Date Integration wavelengthrange slitPA seeing
(W01) times µm ◦EofN arcsec
Q2206−199 N-14-1C 2000Aug2 3600 1.453−1.532 +25.45 1.0
=6×600
Q2206−199 N-14-2C 2000Jul8 7200 1.453−1.532 −13.05 0.8
=2×6×600
2233.9+1318 N-16-1D 2000Jul11 7200 1.988−2.110 +158.45 0.8
=2×6×2×300
termininganormalisedprofileofthequasar,smoothlyvary-
ingwithwavelength.Thisprofileisthenscaledtothequasar
counts at each wavelength, and subtracted. Regions where
emissionlinesfromthegalaxyaredetectedarethenmasked,
andtheprocedureisiterated.Thisworkedwellherebecause
the galaxy continuum is negligibly faint, and because the
emission is offset from thequasar centroid.
For the candidate N-14-1C the detected line occurs
at the expected wavelength for [Oiii]500.7 from the con-
Figure3. Two-dimensionalspectrumof2233.9+1381,smoothed
firmed galaxy counterpart to the DLA absorber at redshift
to enhance the contrast of the detected emission line. The thick
z = 1.9205. For the candidate N-14-2C the detected line lineis thespectrum ofthe quasar. The[Oiii] emissionlinefrom
occurs at a very similar, but slightly shorter, wavelength, N-16-1D is clearly visible at a wavelength 2.078µm, offset from
confirmingthatN-14-2CispartofthesameDLAgalaxy as thequasarby2.5arcsec
N-14-1C,andnotanunrelatedsourceaswaspreviouslysus-
pected.Afaint,neverthelesssignificant,extensionofthede-
tectedemissionline,towardsredderwavelengthsandsmaller thegalaxy spectrum and the standard–deviation spectrum.
impact parameters, is visible in thebottom panelin Fig. 2. As expected the counts in these S/N spectra have mean
As explained in the next sub-section, we interpret this as zero,andstandarddeviationunity.Fromtheframesformed
fluxfrom N-14-1C leaking into theslit. bysumming theregistered raw frames, we re–extracted the
Quasar 2233.9+1381, candidate N-16-1D: Fig. 3 pro- quasarspectra,whichnowincludesky.Vacuumwavelengths
vides the 2D spectrum of our observation of the quasar of skyemission lines in thesespectra were used tocalibrate
2233.9+1381 over approximately half the total wavelength thequasarandgalaxyspectraontoalinearwavelengthscale.
rangeobserved.Anemssion lineisclearlyvisible,attheex- All measured wavelengths quoted in this paper have been
pected wavelength for [Oiii]500.7 from the confirmed DLA corrected to theheliocentric frame.
galaxy counterparttotheLyman–limit absorberatredshift We used the spectra of bright AII and GIV stars, ob-
z = 3.1501. The frame has been smoothed to enhance the served on the same nights, at similar airmasses to the tar-
contrastoftheline.Neitherthe[Oiii]495.9northeHβ486.1 gets,tofluxcalibrate the1D spectra.Thecalibration curve
linesweredetected.Thedetectedemission lineiswellsepa- was derived by taking the ratio of the observed spectrum
rated from the quasar, so that it was not necessary to sub- ofeachstandard,andablack–bodyspectrumofthecorrect
tract the spectrum of the quasar. The spatial offset mea- brightness. The spectra of Q2206−1958 are affected by at-
sured from the2D spectrum,listed in Table3, issomewhat mosphericabsorptionbands,whicharecorrectedforbythis
smaller than thevalue measured from the NICMOS image, procedure.Neitherofthedetectedemissionlinesisstrongly
at the 3σ significance level, while it is in agreement with affected by absorption. This procedure provides reasonably
the offset measured in the STIS frame. Such a discrepancy accurate calibration provided slit losses for the object and
might beexplained, for example, by variations in mean age standardstararesimilar.Sinceourtargetshaveangularsize
of thestellar populations across thegalaxy. Itis interesting much smaller than the seeing this will be true if the seeing
to note that the rest–frame UV continuum seems to trace conditions were similar, which was true for the two nights
the [Oiii] line emission better than the rest frame optical which were clear. For the data taken on 2000 Aug 2 (N-14-
continuum. 1C), when conditions were not photometric, thecalibration
was derived from the ratio of the uncalibrated spectrum of
the quasar from that night, to the calibrated spectrum of
3.2 Extracted 1D spectra
the same quasar observed on 2000 Jul 8 (N-14-2C). As an
Thequasar1Dspectraandthecorrespondingvariancespec- additional check we compared the calibrated fluxes of the
trawereextractedusingtheIRAFroutineapall,witha14– quasarsagainsttheNICMOSphotometry,findingagreement
pixelapertureandanoptimalprofileweightingscheme.The atthe0.1mag.level.Thischeckprovidesonlyanindication
same apertures, with appropriate vertical shift, were then of the spectrophotometric accuracy, because the NICMOS
appliedtotheextractionofthegalaxyspectra,andthecor- andspectroscopicobservationswereseparatedbytwoyears,
responding variance spectra, from the 2D frames with the and thequasars may havevaried in theinterim.
quasar spectra subtracted. At this stage ww checked the The galaxy 1D spectra are plotted in Figs 4 and 5. In
variance spectra, by forming, for each galaxy, the ratio of eachcaseweplot:i)theflux–calibrated1Dspectrum(solid),
[Oiii] emission from DLA galaxies 7
Table 3.Spectroscopic andphotometricparametersofthethreecandidates
Candidate impactparameter(arcsec) R(spec.) r1/2(M02)
spec. image(NIC) image(STIS) (kpc) (kpc)
N-14-1C 0.97±0.04 1.13±0.07 0.99±0.05 8.2±0.3 4.2
N-14-2C 1.24±0.03 1.33±0.04 1.23±0.05 10.4±0.3 —∗
N-16-1D 2.51±0.04 2.78±0.07 2.51±0.05 19.0±0.3 1.1
∗
consistentwithpointsource
with the error spectrum (dotted), and our min–χ2 fit of a counterpart to the z = 2.0762 DLA absorber lies at small
Gaussian to the detected emission line (dot–dashed), and impact parameter, bysearching thespectrum of thequasar
ii) the ratio of the object and error spectra i.e. the S/N for the Hβ and [Oiii]495.9 lines at this redshift. The
spectrum. The errors are large at the wavelengths of the lines are undetected. We measured fluxes of (0.2±4.4)×
strong OH sky lines. Residuals from sky subtraction can 10−17ergs−1cm−2,and(−2.0±3.5)×10−17ergs−1cm−2for
mean that the significance of a detected line in the plotted theselines, respectively.
spectrummaynotalwaysbeeasilyappreciated.Plottingthe Candidate N-16-1D: The [Oiii]500.7 emission line is
S/N spectrum can make this clearer. visible in the two panels of Fig. 5. In measuring the prop-
Measurements of the three detected emission lines are erties of the line, because the recorded spectrum covers
summarised in Table 4. Col. 1 lists the galaxy name, and both lines, [Oiii]495.9, 500.7, we fit both lines simultane-
col.2providestheredshiftoftheDLAabsorber(themean- ously, with flux ratio 1:3. The [Oiii]500.7 line is detected
ing of ‘ELA’ is explained in the next section). For each de- at S/N =13.5. The [Oiii]495.9 line is not significantly de-
tected emission line we made a min–χ2 fit of a Gaussian tected because it is a factor of three weaker, and lies in a
profile. The variables, then, are the line centre, line width region of higher noise. The Hβ line was not detected. We
(corrected for resolution), and line flux.Col. 3 provides the measured a flux of (1.1±1.6)×10−17 ergs−1cm−2for this
redshift of the detected [Oiii] line; col. 4 lists the flux in line.
the[Oiii]500.7line;andcol.5liststhelineluminosity.Col.
6 provides the intrinsic line width (FWHM), which is con-
verted to a 1D velocity dispersion in Col. 7. In Col. 8 we
list thevelocitydifferencebetweentheDLAabsorption line 4 KINEMATICS
andthe[Oiii]500.7 emission line.Col.9providestheveloc-
ProchaskaandWolfe (1997b, hereafterPW97) haveconjec-
ity difference between any detected Lyα emission and the
turedthatDLAabsorptionlinesariseinlarge,cold,rapidly–
[Oiii]emission.
rotating discs, of circular velocity vc ∼ 225kms−1, from
CandidateN-14-1C:The1Dspectrumofthiscandidate
an analysis of the detailed velocity structure of the low–
is shown in the top two panels of Fig. 4. The [Oiii]500.7
ionisation metal absorption lines. Typically the strongest
line is detected at 6σ, at a redshift z = 1.9220. The line
component of the absorption complex occurs at either the
has FWHM=220kms−1, corresponding to 14 pixels, which
blue or red limit of the velocity spread, resulting in a char-
is greatly oversampled. Therefore in panel b) we plot the
acteristic asymmetric profile with a sharp line edge. They
S/N in pixels binned by a factor of four. In this plot the
were able to explain this ‘edge–leading asymmetry’ (here-
significance of theline is much more apparent.
after ELA) in terms of the line of sight passing through a
Wecheckedfor thepossibility thatthegalaxy counter- thickdisc.Consideragaseousdiscwithaflatrotationcurve,
part to the z = 2.0762 DLA absorber lies at small impact of circular velocity vc, and in which the number density of
parameter, by searching the spectrum of thequasar for the absorbingcloudsdeclinesmonotonically with distancefrom
Hβ and [Oiii]495.9 lines at this redshift. The lines are un- thecentre.Thesimplestcaseisifthegalaxyisviewededge–
detected. Deeper limits were reached at the other position on,i=90degrees.Thenthefullcircularvelocityisrecorded
angle, and are provided below. at the point at which the line of sight passes closest to the
CandidateN-14-2C:The1Dspectrumofthiscandidate centre. This also corresponds to the strongest absorption if
is shown in the lower two panels of Fig. 4. As stated previ- theface–onsurfacedensityisadecliningfunctionofdistance
ously, the [Oiii] line from N-14-2C, detected at S/N = 12 fromthegalaxycentre.Elsewherealongthelineofsightthe
lies at a lower redshift than N-14-1C. Howeverthefaint ex- column density is lower, and only a component of the cir-
tension to the red, visible as a second peak in the 1D spec- cularvelocity isrecorded.This explainstheELAprofilefor
trum,liesat aredshift consistent with theredshift of N-14- theedge–on case. Thelineedgeoccursat eithertheblueor
1C.Thereforetomeasurethelinewefitsimultaneouslytwo the red end of the profile, depending on whether the gas is
Gaussians, with the higher–redshift line fixed at the red- rotating towards or away from the observer, respectively.
shift of N-14-1C. As seen in the plot, this provides a good At smaller inclinations, i< 90degrees, the situation is
fit.Theweakerlineisdetectedat3.6σ.Thereisnoevidence more complicated, because the absorption profile depends
for light from N-14-2C leaking into the spectrum of N-14- also on the density distribution of clouds perpendicular to
1C,butgiventhelowerS/N ofthelatterspectrum,andthe thedisc, and on thepoint at which the line of sight crosses
slightly larger distance of N-14-2C from the slit edge, Fig. thediscmid–plane.Insomesituationstheabsorptionprofile
1, this is not surprising. can be reversed, in the sense that the strongest absorption
We again checked for the possibility that the galaxy can correspond to the minimum component of rotational
8 S. J. Weatherley, et al.
Figure 4. One-dimensional extracted spectra of candidates N-14-1C and N-14-2C. The top two panels show the data for N-14-1C,
and the bottom two panels show the data forN-14-2C. Thetop panel foreach candidate shows the extracted spectrum (solid), the 1σ
noisespectrum(dotted), andthemin–χ2 fittothedata(dot-dashed). ForN-14-1CwefitasingleGaussiantotheemissionline.Forthe
observation of N-14-2C we alsodetected light fromN-14-1C, and we fit two Gaussians, as described inthe text. The bottom panel for
eachcandidateisthesignal-to-noisespectrum.ForN-14-1Cthishasbeenbinnedupbyafactoroffour.
velocity (PW97, fig. 2). Finally, the absorption profile also the CDM hierarchical model. In this case, the line profiles
varieswithimpactparameter,inthesensethatoneobserves arecausedbyamixtureofrandommotions,rotation,infall,
a smaller spread of velocities at larger impact parameter. and mergers. Therefore the ELA test, on its own, cannot
The PW97 picture of DLAs arising in large, cold, distinguish between the two models.
rapidly–rotating thick discs is inconsistent with the CDM AspointedoutbyWarren&Møller(1996),PW97,and
hierarchical model of structure formation, which for these Lu,Sargent,&Barlow (1997, hereafter LSB),measurement
redshiftspredictssmallerdiskswithsmallerrotationspeeds. of thegalaxy systemicvelocity, through thedetection of an
Nevertheless, as shown by Haehnelt et al. (1998), the ab- emission line, provides a vital additional datum for the in-
sorption line data itself may be equally well explained by terpretation of the kinematics and possible distinction be-
[Oiii] emission from DLA galaxies 9
Figure 5.One-dimensional extracted spectraof candidate N-16-1D.The toppanel shows theextracted spectrum (solid),the 1σ noise
spectrum(dotted),andthemin–χ2fitofGaussianemissionlines,fitsimultaneouslytothe[Oiii]495.9,500.7lines(dot-dashed).Theflux
fromthe weaker 495.9line,near 2.059µm, isbelow the detection limitatthat wavelength. The bottom panel shows the signal-to-noise
spectrum.
Table 4.[Oiii]emissionlinedataforourthreeDLAgalaxies
1 2 3 4 5 6 7 8 9
Candidate zDLA(ELA) z([OIII]) Flux×10−17 Lum×1042 FWHM σ ∆v(DLA-[OIII]) ∆v(Lyα-[OIII])
( ergs−1cm−2) (ergs−1) (km s−1) (km s−1) (km s−1) (km s−1)
N-14-1C 1.919991(2) 1 1.9220(2) (7.6±1.3) (2.0±0.3) 220±50 90±20 −210±20 90±70
N-14-2C 1.91972(8) (10.7±0.9) (2.8±0.2) 180±25 75±11 29±9 —
N-16-1D 3.14930(7)2 3.15137(6) (6.78±0.5) (5.9±0.4) 55+−2300 23+−813 −150±6 118±22
1.Component2fromTable5AinProchaskaandWolfe(1997a)2.StrongestFeII160.8line(Lu,Sargent,andBarlow,1997,1998)
tween these two pictures. The first application of this, was Quasar2233.9+1381:TheredshiftofN-16-1D,fromthe
the analysis by LSB of one of our targets, the z = 3.1501 [Oiii]500.7 line, is z=3.15137±0.00006, which we taketo
DLA absorber in the spectrum of the quasar 2233.9+1381. be the systemic redshift of the galaxy. The low–ionisation
LSB used the redshift of the Lyα emission line from N-16- metal absorption lines in this system display edge–leading
1D,measuredbyDjorgovskietal.(1996),asthegalaxysys- asymmetry, towards the blue. We are interested in the red-
temicredshift;however,theLyαline,beingaresonanceline, shiftoftheELAlineedge,whichwetakeasrecordedbythe
issubjecttocomplexradiationtransferprocesses,whichre- strongest FeII160.8 line, at z = 3.1493. This is entered in
sultinanasymmetricemissionlineprofile.Inthepopulation Table 4 as zDLA(ELA). The absorption line is plotted in
ofLyman–breakgalaxies,Pettinietal.(2001)findthat,rel- fig.1 ofLSB, and theredshift is takenfrom Lu,Sargent, &
ativetotherest–frameopticalemissionlines,theLyαemis- Barlow (1998). No error is quoted, and we will adopt a un-
sionlineisredshiftedbytypicallybetween200and1000 km certaintyof±5km s−1,based onan inspection ofthespec-
s−1, while the interstellar absorption lines are blueshifted trum.TheDLA(ELA)line,then,isblueshiftedby150±6km
by similar amounts. They interpret this as the signature of s−1 relative to the systemic velocity, while the Lyα line is
stronggalacticwinds.Therest–frameopticalemissionlines, redshifted by118±22km s−1.
unaffectedbyresonantscattering,providethebestmeasure- InthePW97picture,thesefeatures(blue–sideELAline
ment of the galaxy systemic velocity, while the Lyα line is edge,blueshiftedrelativetothesystemicvelocity,andasub-
unreliable. stantial velocity difference) are the characteristics of a line
Wenowturntotheinterpretationofourresults,inthe of sight that intersects the disc mid–plane close to the ma-
light of this discussion. Relevant quantitiesare summarised joraxis,atsmallimpactparameter.Anexampleisgivenby
inTable4.Wediscussthequasar2233.9+1381first,because profile no. 3, in fig. 14 of PW97. Because the intersection
it is thesimpler case. is close to the maor axis, we may approximate the velocity
10 S. J. Weatherley, et al.
differenceof150±6km s−1 tovcsini.Wenotethatthisim- discrepancies with the predictions of PW97: in the case of
pliesasubstantiallysmallermassthaninferredbyLSB,who the quasar 2233.9+1381, the DLA galaxy systemic velocity
used the velocity difference given by the Lyα line, 270km lieswithintheabsorptionvelocityfield,ratherthanoutside;
s−1, but is nevertheless in line with the proposal of PW97 inthecaseofthequasarQ2206−1958, weareunabletoin-
of a typical circular velocity of vc ∼225km s−1. corporate the velocity of N-14-2C in a consistent way, and
Although the analysis presented so far sits well with this, coupled with the morphology of the system, supports
thePW97 picture, a prediction of their model, not only for thehierarchical picture.Clearly thedetection ofrest–frame
lines of sight such as this, but for all lines of sight, is that optical emission lines from a larger sample of DLAgalaxies
thesystemicvelocityshouldlieoutsidethevelocityrangeof would beextremely useful, to extend thiscomparison.
thelow–ionisation metal absorption–line profiles. This does
not appear to be the case here. LSB argue that the FeII
line is thebest tracer of theneutral gas, and state that the
5 COMPARISON WITH THE PROPERTIES
velocity spread is ∆v∼200km s−1, meaning that the FeII
OF LBGS
absorption profile extends as far as ∼ 50km s−1 redward
of the systemic velocity. This contradicts the prediction of InthissectionweextendtheanalysisofM02,comparingthe
PW97. The velocity difference is significant, although not propertiesofDLAgalaxiesandLyman–breakgalaxies,toin-
large, and the effect would need to be seen in a number of cludepropertiesoftherest–frameopticalemissionlines.The
systems before ruling out themodel. aimistotestthehypothesisthatDLAgalaxiesareLyman–
Quasar Q2206−1958: A detailed analysis of theveloc- break galaxies, selected by gas cross section. As explained
ityprofilesofbothlow–ionisation andhigh–ionisationmetal in M02, and in the Introduction section here, although the
absorption lines in the z = 1.9205 absorber has been pre- average properties of the two populations may differ (be-
sentedbyProchaskaandWolfe(1997a).Thevelocityprofile cause of the different selection criteria), if the hypothesis
of the low–ionisation absorption lines shows several com- is correct, theproperties of individualDLA galaxies should
ponents, with the two strongest separated by 65km s−1. lie within the range of the properties of a comparison sam-
There is a sharp cut–off to the blue of the lower–redshift ple of Lyman–break galaxies of similar absolute magnitude
component (fig. 2 and table 5A in Prochaska and Wolfe, and redshift. This will be true regardless of how the DLA
1997a), and they considered the profile “to be consistent galaxies were selected.
with an edge–leading asymmetry,but notethatits shapeis Thenewobservationsgiveusthreeextraparametersto
notassuggestiveasthemajorityoftheothercases”.Accept- use. These are the luminosity weighted velocity dispersion,
ing this judgment, we select the strongest blue component, asrecordedbythewidthofthe[Oiii]line,the[Oiii]linelu-
z = 1.919991, as the redshift of the leading edge, entered minosity,andthevelocitydifferencebetweentheLyαemis-
aszDLA(ELA)inTable4.Theothercomponentsidentified sion line andthe[Oiii]line. Althoughourmeasurementsof
extend redward over an intervalof 140km s−1. the three sources considered here are for the [Oiii]line, we
WenowhavetochoosebetweenN-14-1C,whichisred- suppose that for the comparison sample of LBGs any rest–
shifted by 210±20km s−1 relative to the ELA redshift, or frame optical emission line is satisfactory for the first and
N-14-2Cwhich is blueshifted by29±9km s−1.Theformer third of these parameters.
is the simplest interpretation within the PW97 picture. As ThecomparisondataaredrawnfromthestudiesofPet-
forN-16-1D,above,thedataarethenconsistentwithanin- tinietal. (2001), and Erbet al. (2003), hereafter P01, E03,
terpretationofalineofsightthatinteresectsthemid–plane which provide the largest samples of LBGs with spectro-
nearthemajor axis,inwhich casewemayapproximatethe scopic detections of rest–frame optical emission lines. P01
velocity differenceof 210±20km s−1 tovcsini. Again this report observations of 19 targets, with mean redshift 3.1,
value is in line with the proposal of PW97 of a typical cir- and E03 report observations of 16 targets, with mean red-
cularvelocityofvc ∼225km s−1.Furthermore,inthiscase shift 2.3. Some of the relevant quantities were not included
the systemic velocity lies outside the velocity range of the inthesepapers,butwerekindlymadeavailabletousbythe
low–ionisation metal absorption–line profiles, in agreement authors.Intotalthesetwosamplesprovide32measurements
with theprediction of PW97. However, if the whole system (orupperlimits)ofthevelocitydispersion,11measurements
isalargedisk,itisthencuriousthatN-14-1Cisfainterthan ofthe[Oiii]lineluminosity,and18measurementsoftheve-
N-14-2C,ifitisthegalaxynucleus.Afurtherdifficultywith locity difference. For consistency with M02, we use the AB
this interpretation is the velocity of N-14-2C. In the rotat- continuum absolute magnitude at a rest–frame wavelength
ing disc hypothesis the only points blueshifted by as much of 150µm,, as themeasure of galaxy luminosity.
as the ELA line are other points along the major axis. But
this would pass from N-14-1C and run close to the quasar
5.1 Velocity dispersion
position.N-14-2Cliesalmostperpendiculartothisline(Fig.
1). Therefore it must be concluded that it is a separate ob- In Fig. 6 we plot velocity dispersion against absolute mag-
ject.Whenweconsideralsothedisturbedappearanceofthe nitude for our three sources, and the 32 LBGs. All values
DLA galaxy in this field, Fig. 1, these results are certainly are intrinsic, i.e. have been corrected for the resolution of
more in agreement with theCDM hierarchical picturethan the observations. Two of the DLA measurements lie within
with thecold large disc hypothesis. thespread of values for theLBGs. The third,N-16-1D, lies
To summarise, the kinematic evidence from these two below the spread of points. Nevertheless, a number of the
fields is in agreement in a number of respects with the pic- LBG values are upper limits.We conclude that there is no
tureof PW97of DLAabsorbersasarising in large rapdily– evidence for a difference between DLA galaxies and LBGs,
rotating cold discs. Nevertheless there are two significant based on this plot.