Deep imaging of Q2112+059: A bright host galaxy but no DLA absorber

6 Pages·0.57 MB·English

by  J. U. Fynbo

A&Amanuscriptno. ASTRONOMY
(willbeinsertedbyhandlater)
AND
Yourthesauruscodesare: ASTROPHYSICS
11(11.01.2;11.05.1;11.17.1;11.17.4;Q2112+059)
Deep imaging of Q2112+059: A bright host galaxy but no DLA
absorber ⋆,⋆⋆
J.U.Fynbo1,P.Møller1,andB.Thomsen2
1 EuropeanSouthernObservatory,Karl-Schwarzschild-Straße2,D-85748,GarchingbyMu¨nchen,Germany
2 InstituteofPhysicsandAstronomy,A˚rhusUniversity,DK-8000A˚rhusC
1
0 Received;accepted
0
2

n Abstract. In a ongoingprogrammeaimedatstudyinggalaxy Lyα absorbers or DLAs for short (Møller and Warren 1993,
a counterpartsof Damped Lyα Absorbers(DLAs) we have ob- 1998; Fynbo et al. 1999, 2000a,b). DLAs cause Lyα absorp-
J
tained high resolution deep I-band imaging data of the field tion lines with damping wings corresponding to HI column
4 around the z = 0.457 BAL QSO Q2112+059. In the liter- densities larger than 2 × 1020cm−2. This very large column
em
ature this QSO is listed to have a candidate DLA at z = densityabsorptionoccursinregionsofselfshielding,coolgas,
1 abs
v 0.2039 along the line of sight. After subtraction of the QSO i.e. where one would expect stars to form. Furthermore, they
4 PointSpreadFunction(PSF)wedetectagalaxycentredonthe contain nearlyall the neutralgas in the universe(Wolfe et al.
6 position of Q2112+059. To help answer whether this galaxy 1995). Hence DLAs at high redshift are prime candidates for
0
is the DLA or the QSO host galaxy we retrieved a GHRS beingtheprogenitorsofpresentdaygalaxies.
1
spectrumofQ2112+059fromtheHST-archive.Thisspectrum At redshifts z < 1.65 DLAs are both rarer and observa-
0
1 shows that there is no Lyα absorption line at zabs = 0.2039. tionallymoredifficulttodetectsinceLyαisnotshiftedabove
0 Thisfactincombinationwiththeperfectalignmentonthesky atmosphericcut–offandhencecannotbeobservedwithground
/ ofthegalaxyandQ2112+059leadustotheconclusionthatthe basedtelescopes(Lanzettaetal.1995;Jannuzietal.1998;Rao
h
p galaxymustbethehostgalaxyofQ2112+059. & Turnshek 2000). Le Brun et al. (1997) reported candidate
- ThehostgalaxyofQ2112+059isbright(Mobs = −23.6), galaxy counterparts of 7 such low redshift absorbers that are
o I
andhas a radialprofilewell fitted by a modifiedHubble+de either confirmed DLAs or strong MgII/FeII absorbers. They
r
t VaucouleursprofilewithR = 0.5kpcandR = 3.6kpc.Our found that those candidate galaxy counterparts cover a wide
s c e
a resultsarewellinlinewiththeconclusionofearlierworkdone rangeinmorphologicaltypes.
v: atlowerredshifts,thatbrightlowredshiftQSOspreferentially The zem = 0.457 BAL QSO Q2112+059 (2000.0 coordi-
i resideinluminous,ellipticalgalaxies.ThehostofQ2112+059 natesRA 211452.6,DEC +060741)isradioquiet,andhas
X
is however, despite it’s brightness, very compact when com- optical magnitudes B = 15.52 and V = 15.77. Lanzetta et al.
ar paredtoearlytypegalaxiesatlowerredshifts. (1995) reported the detection of a candidate DLA at zabs =
0.2039intheIUEspectrumofQ2112+059.InSeptember1996
Keywords:quasars–absorptionlines,quasars–Q2112+059, weperformedanimagingstudyofQ2112+059inanattemptto
galaxies–active,galaxies–elliptical,lenticular,cD detect the optical counterpart of this DLA. The imaging and
data reduction is described in Sect. 2, in Sect. 3 we present
spectroscopicdatafromtheHubbleDataArchiveandinSect.
4and5wediscussourresults.
1. Introduction
For easy comparison to earlier work on this subject, we
We have during the last several years been undertaking pro- shallhereassumeaHubbleconstantof50kms−1 Mpc−1and
grammes aimed at detecting optical counterparts of the most Ω=1.0.
HI rich class of QSO absorption line systems, the Damped
Sendoffprintrequeststo:J.U.Fynbo 2. Imaging
⋆ BasedonobservationsmadewiththeNordicOpticalTelescope,
operatedontheislandofLaPalmajointlybyDenmark,Finland,Ice- 2.1. ObservationsandDataReduction
land,Norway,andSweden,intheSpanishObservatoriodelRoquede
The observationswereperformedin 1996withHiRAC (High
losMuchachosoftheInstitutodeAstrofisicadeCanarias.
⋆⋆ BasedonobservationswiththeNASA/ESAHubbleSpaceTele- Resolution Adaptive Camera) on the 2.56 m Nordic Optical
Telescope underphotometricand excellentseeing conditions.
scope,obtainedfromthedataArchiveattheSpaceTelescopeScience
Institute,whichisoperatedbytheAssociationofUniversitiesforRe- The CCD used was a 1k×1k back-side illuminated thinned
searchinAstronomy,Inc.underNASAcontractNo.NAS5-26555. Tektronix with a pixel scale of 0.176 arcsec. The data were
Correspondenceto:[email protected] allobtainedintheBesselIfilter.Thejournalofobservationsis
2 J.U.Fynboetal.:DeepimagingofQ2112+059:AbrighthostgalaxybutnoDLAabsorber
Fig.1. Combined frame containing 973×1023 pixels
(171×180arcsec2)surroundingtheQSOQ2112+059(marked
byanarrow).Northisupandeasttotheleft.Alsomarkedare
the field stars A–E which we use to check the stability of the
PSFinSect.2.2.
given in Table 1. A total integration time of 3.25 hours was
obtained. Also observed were 7 Landolt sequences (Landolt
1992).
The39individualframeswerebiassubtracted,flat-fielded
andsubsequentlycoaddedusingtheoptimalcombinationcode
describedinMøllerandWarren(1993).
InFig.1weshowthefinalcombinedimageof973×1023
pixels (171×180 arcsec2) surrounding Q2112+059 which is
marked with an arrow. Using the Landolt sequences we Fig.2.–a)Theradialprofilesof5brightfieldstarsandofthe
obtained the calibration from the instrumental system to QSO. The profiles were determined as the azimuthal average
Cousins I. The uncertaintyin the zero–pointis 0.02 mag. For ona9by9sub–pixelgrid.Theprofileshavebeennormalised
Q2112+059wefindI=15.16±0.02,wheretheerrorisdomi- to the same valuein the centre(illustrated in the insertwhere
natedbythezero–pointerror.TheskybrightnesswasµI(sky) weplotthecentralpartoftheµI profilesvs.log(R).TheQSO
= 19.0 mag arcsec−2 and the sky noise in the final combined clearly has excess extended flux compared to the stellar pro-
image 26.8 mag arcsec−2. The full–width–at–half–maximum files.b)Residualsaftersubtractionoftheweightedmeanstel-
(fwhm)ofpointsourcesinthecombinedimageis0.69arcsec. larprofiledeterminedfromthestarsA–E.Theresidualsforthe
5starsare0towithinthenoise,butthereisaclearlydetected
residualforQ2112+059.
Table1.JournalofNOTobservations
Date SeeingFWHM Numberof Exp.Time
(arcsec) Frames (sec)
1996Sep15 0.6-0.8 7 2100
1996Sep17 0.5-0.7 10 3000 There is no indication of variability of Q2112+059 with
1996Sep18 0.6-0.8 13 3900
amplitudeabove0.01magontime-scalesfromtensofminutes
1996Sep20 0.5-0.7 4 1200
todays.
1996Sep21 0.8-0.9 5 1500
J.U.Fynboetal.:DeepimagingofQ2112+059:AbrighthostgalaxybutnoDLAabsorber 3
2.2. PSFqualityandstabilitytest
In order to search for faint sources near the line of sight to
thebrightQ2112+059agoodPSF-subtractionismandatory.To
testthestabilityofthePSFweselected5brightstarsdistributed
over the full field. These stars are marked A–E in Fig. 1. We
then determined the radial profiles of the field stars by aver-
aging over the azimuth angle on a 9 by 9 sub–pixel grid. In
Fig. 2a we have plotted the radial profiles determined in this
way.AlsoplottedistheradialprofileofQ2112+059.Allpro-
fileswerenormalisedtothesamelevelinthecentre(shownin
detail in the inserted logarithmicplot in Fig. 2a). The surface
brightness(µI) zero–pointin Fig. 2awas chosento be thatof
theQ2112+059profile.Itisseenthat,i)the5brightstarshave
radialprofiles(fulldrawnlines)thatareidenticaltowithinthe
thickness of the lines out to about 2.8 arcsec, and ii) the ra-
dialprofileofQ2112+059(shortdash)showsexcessfluxover
thatofthefieldstars.ToquantifytheexcessweplotinFig.2b
theresidualsurfacebrightnessaftersubtractionoftheweighted
mean stellar profile determinedfrom the field stars A – E. At
radial distances ranging from 1.4 to 3.6 arcsec from the cen-
troid of the QSO, the excess surface brightness is 0.08–0.14
mag.InthesamerangethescatterofthestellarPSFsisabouta Fig.3. Upper left : 100×100 pixels (17.6×17.6 arcsec2) sur-
factor10–20smaller.Atradialdistanceslargerthan3.6arcsec
rounding Q2112+059. Upper right : Same as left frame, but
theprofilesaredominatedbyskynoise.
after subtraction of the 2D PSF. Lower panels : For compari-
We endthissectionconcludingthatthePSFisverystable sonweshowthefieldstarmarkedCinFig.1beforeandafter
acrossthefield andthatthe profileofQ2112+059hassignifi- PSF–subtraction.Theimageshaveallbeensmoothedbya3×3
cantexcessfluxoverthatofaPSF. boxcarfilter.Thecontourlevelsare-4,4,6,8,12,20,38,...in
termsofsigmaoftheskyintheunfilteredimage,withtheneg-
ativecontourbeingdotted.
2.3. 2DPSF–subtraction
Toexaminethenatureoftheexcessfluxweproceededwitha
(0.07arcsec).The isophotesofthe residualshownoclear ev-
twodimensionalPSFsubtractionusingDAOPHOTII(Stetson
idence for ellipticity, rather they are consistent with e = 0.
1987,1999).Forthedeterminationofthe2DPSFwechoseto
There are two additionalobjects that are both consistentwith
usestar A,whichistheclosesttoQ2112+059,butasdemon-
being point sources 5 and 7 arcsec east of S, and patches of
strated above our results are not dependent on this specific
faint extendedfuzz both east and west of the object S. In the
choiceofPSFstar.WhenwesubtractedthePSFoftheQSOus-
lower panels of Fig. 3 we show for comparison the field star
ingthescalingofthePSFautomaticallychosenbyDAOPHOT
C (0.35 mag fainter than Q2112+059)before and after PSF–
II this produced a residual consisting of an extended positive
subtraction.
residual surrounding a negative “hole” in which the PSF had
beenover–subtracted.Thisisthe typicalresidualonewill see
3. SpectroscopyfromtheHDA
whenapointsourceandanextendedobjectaresuperimposed
on roughlythe same position(see e.g. Fig. 2b where the PSF Our original aim with the observations was to detect the op-
subtracted QSO profile is 0 in the centre but 0.13 mag 3 arc- tical counterpartofthe DLA, butthe perfectalignmentofthe
sec out).To obtaina image ofthe underlyingextendedobject galaxySwiththeQSOreportedabovesuggeststhatweprob-
withoutthe“hole”fromthePSFsubtraction,wemustimpose ablyhavedetectedtheQSOhostgalaxyratherthanthegalaxy
someadditionalcondition.Areasonableassumptionisthatthe causing the reportedcandidateDLA absorption.To helpclar-
extendedobjectis“flat”acrossthecentroidoftheQSO,sowe ifythiswesearchedtheHubbleDataArchive(HDA)forspec-
shallusethisastheextraconstrainttobreakthedegeneracy. troscopic data of Q2112+059.In this section we describe the
Fig.3showstheresultofthisfinaltwodimensionalPSF– resultsofthissearch.
subtraction. In the upper left panel is shown 100×100 pixels
(17.6×17.6arcsec2)aroundQ2112+059.Theupperrightpanel
3.1. RetrievalofthespectraandDataReductions
showsthesameareaafterthefinalsubtractionofthePSF.The
residualimageintheupperrightframeshowsafairlylargeex- We foundtwo data sets suitable for this study. GoddardHigh
tendedobject(markedS),presumablyagalaxy.Thecentroidof ResolutionSpectrograph(GHRS)spectraovertheregionfrom
this objectcoincideswith thatof the QSO to within 0.4 pixel 1330 A˚ to 1610 A˚ (grism G140L) was obtained in 1995 (PI
4 J.U.Fynboetal.:DeepimagingofQ2112+059:AbrighthostgalaxybutnoDLAabsorber
Fig.4.The80A˚ regionaroundLyαatz=0.2039ofthenormal-
ized and binned spectrum of Q2112+059. Also shown is the
1σ error per 0.72A˚ bin (lower curve). The dotted line shows
a damped Lyα absorption line with a HI column density of
2×1020cm−2ataredshiftofz=0.2039.Thepresenceofsucha
lineaexcludedatthe29σlevel.
Lanzetta) and in 1992 spectra of the region from 1610 A˚ to
3250A˚ wasobtainedwiththeFaintObjectSpectrograph(FOS)
(PI Bahcall). Only the FOS-spectra has been published (Jan-
nuzietal.1998).
We retrieved the calibrated GHRS spectrum from the
archive. Hence, the data reduction was limited to extracting
the1dimensionalspectrumand1σ errorspectrafromthecal-
ibrated science data files and using the calibrated wavelength
solutionfiles.
3.2. Results Fig.5. Upper panel: The crosses show the surface brightness
for the host galaxy determined by azimuthal averaging on a
Fig. 5 showsthe spectrumofQ2112+059coveringthe region 9 by 9 sub–pixel grid plotted against R1/4 (in arcsec). Also
from1425A˚ to1505A˚.Alsoshownisthe1σerrorper0.72A˚ plotted(fulldrawnlines)are 6exponentialprofileswithscale
bin(lowercurve). lengths0.1–0.6arcsecconvolvedwiththePSFandazimuthally
ThepresenceofadampedLyαabsorptionlineintheLyα averagedinthesamewayasthedata.Asseennoexponential
forestofQ2112+059atz = 0.2039isexcludedatthe29σ modelprovidesagoodfittothedata.Lowerpanel:Sameasthe
abs
confidencelevel. The FOS spectrum shows absorptiondue to upperpanelexceptherewehaveoverplottedtheradialprofile
gas in the Milky Way and intervening metal line systems at of a modelconsistingof an outerR1/4 profile anda modified
redshifts z=0.370 and z=0.418 as well as metal line systems Hubble profile in the core (see text for details). As seen this
associatedwithQ2112+059,butnoabsorptionlinesatthered- modelmatchthedatawell.
shiftz=0.2039(Jannuziprivatecommunication;Turnsheketal.
inprep.).
4. Natureofthehostgalaxy
AsdiscussedbyRao&Turnshek(2000)only2of14can-
didate DLA systems from the IUE survey of Lanzetta et al. As there clearly is no DLA absorber in the sightline towards
(1995)havebeenconfirmed.Q2112+059addsonetothelistof Q2112+059,andasthecentroidoftheextendedobjectwehave
non-confirmedcandidates. detected is closely aligned with the position of the QSO (to
within0.07arcsec),weconcludethatwehavedetectedthehost
galaxyofQ2112+059.We nowreturntoadiscussionofwhat
J.U.Fynboetal.:DeepimagingofQ2112+059:AbrighthostgalaxybutnoDLAabsorber 5
can be learned about the nature of this galaxy from our deep Table 2. Size, absolute magnitude and redshift of the host of
imaging. Q2112+059,andofRadioGalaxies(RG),RadioLoadQuasars
An attempt to fit a 2D galaxy model to the 2D PSF sub- (RLQ)andRadioQuietQuasars(RQQ)intheworkbyMcLure
tractedimage shownin Fig 3 wasunsuccessfulas the signal– etal.(1999).
to–noiseoftheimagewastoopoortoprovidemeaningfulcon-
¯ ¯
straints.However,thealignmentoftheQSOwiththecentroid object Re,rms Lgal/LQSO Mgal,rms z
kpc
of its host, and the fact that there is no evidence in the 2D Q2112+059 3.6 0.035 −23.6 0.457
residualimage(Fig.3)foranyelongationoftheextendedob- RG 14.9,5.4 25–3000 −23.65,0.50 0.174–0.244
RLQ 11.2,4.5 0.17–1.3 −23.89,0.29 0.173–0.258
ject, suggest that we could significantly lower the number of RQQ 9.8,4.8 0.06–5 −23.27,0.52 0.115–0.239
fitting parameters, and at the same time increase the signal–
to–noise of the data to fit, if we returned to the 1D regime.
Again,asinthe2Dcase,weareleftwiththedegenerateprob-
largeand at radiilargerthan 3.5arcsec where the uncertainty
lemofdecidinghowtoscalethePSFwesubtractfromthedata.
in themeasurementofthesky–levelbecomesdominant(sky–
Galaxy cores are knownoften to have centralspikes (see e.g.
levelerrorsareshownasshortdashcurvesinFig.5).Atradii
Møller etal. 1995)which couldbe dueeither to centralpoint
between0.5and3.5arcsectheprofileisverywelldetermined.
sources(AGN),tocentraldensitycuspsor(whichisthemost
likely)toacombinationofboth(foradiscussionseeStiavelli Fig.5ashowsthatitisnotpossibletoobtainanacceptable
etal.1993).InthecaseofQ2112+059thereclearlyisacentral fitwithanexponentialdiscprofileforanyscalelength.Incon-
AGN(theQSO),whichisthepointsourcewewishtoseparate trast,asseeninFig.5b,amodelwithanR1/4outerprofilewith
out.Therefore,asinMølleretal.(1995)webreakthedegener- aneffectiveradiusofRe = 0.53arcsec(3.6kpc)andamodi-
acybyimposingtheconditionthatthecentralprofilemustbe fiedHubbleprofilewithacoreradiusofRc =0.07arcsec(0.5
isothermalinthesensethatthegalaxycoreisfittedbyamod- kpc)inthecoreprovidesanexcellentfit.
ifiedHubbleprofile.Thisconditioncloselycorrespondstothe
requirementwe used for the 2D PSF subtraction,namelythat
5. SummaryandDiscussion
theresidualmustbe“flat”or“smooth”inthecentre.Withthis
assumptionweeffectivelyassignthemaximumpossiblefluxto Based on an IUE spectrum of Q2112+059 Lanzetta et al.
the centralpoint–source,and thereforeobtaina lower limitto (1995)reportedacandidatedampedLyαlineatz=0.2039.Sub-
thefluxofstellarorigin.However,thecentraldensitycuspsof sequentHSTspectroscopy(Fig.4)hasnotconfirmedthis.This
localearly–typegalaxiesdonotcontainanysignificantfraction result is important in its own right as DLAs are very rare at
ofthetotalstellarlight(1–10%ofthefluxmeasuredinsidethe z <1.InthesampleofDLAspresentedinWolfeetal.(1995)
galaxycore,Mølleretal.1995),sotheresultinguncertaintyon thereare10DLAsatz<1.0.Ofthe10only4areatz<0.5,and
thefluxfromthetotalstellarcomponentisinsignificant. oneofthosewasthecandidateDLAtowardsQ2112+059.
Forthe1DPSFweagainusedtheweightedmeanofthe5 Having ruled out the presence of an intervening DLA at
field stars(allfield stars were givenweightscorrespondingto z = 0.2039, we find that we are forced to interpret the ellip-
their flux, except star E which was given a lower weight be- tical galaxy we detect under the PSF of Q2112+059 as the
cause of the nearby saturated star which caused variable sky host galaxy of the QSO. The host galaxyhas an I-bandmag-
across the profile). In Fig. 5a and Fig. 5b we plot the host nitude of 18.8 (total magnitude determined from the model
surfacebrightnesspersquarearcsec(µ )againstR1/4.Super- fit), and an effective radius of 3.6 kpc. I=18.8 correspondsto
I
imposed on the data are profiles corresponding to a range of Mobs = −23.6 (no k-correction) in the assumed cosmology
I
galaxymodels.InFig.5awehavesuperimposedsixexponen- (Sect.1).
tialdiscprofiles McLure et al. (1999) find that both radio galaxies (RGs),
radioloudQSOs(RLQs)andbrightradioquietQSOs(RQQs)
I(R)=I0×exp(−R/Rs)
preferentiallyarehostedbymassive,brightellipticalgalaxies.
withscalelengthsR (0.1–0.6arcsec),andinFig.5btheradial Our detection of a bright elliptical galaxy hosting the radio
s
profileofamodelwithanR1/4outerprofile quiet Q2112+059is consistent with this result. In Table 2 we
I(R)=I0×exp(−7.67×((R/Re)0.25−1)), cthoamtpthaerehoousrt greasluaxltystooftQho2s1e12o+f0M59cLisurienetthealb.r(i1g9h9t9e)nadnodffitnhde
andamodifiedHubbleprofile absolutemagnitudedistribution,buthasasmallereffectivera-
I(R)∝(1+(R/R )2)−1, diusthananyRQQhostgalaxyintheirsample.
c Schneideretal.(1983)studiedthebrightestclustergalax-
in the core. The models shown were produced on a 2D sub– ies in 83 Abell clusters (with redshifts in the same range as
pixelgridandconvolvedwiththeobserved2DPSF. Thefinal theMcLureetal.samples).Theyfoundatightcorrelationbe-
1Dmodelprofileswerethendeterminedasazimuthalaverages tween the core radii and effective radii. The host galaxy of
usingthesameprocedureasforthedata. Q2112+059fallsnicelyonthiscorrelation,butissmallerthan
It is seen that the radial profile of the host is noisy in the anyofthebrightestclustergalaxiesoftheSchneideretal.sam-
central0.5arcsecwheretheerrorsfromthePSFsubtractionare ple.
6 J.U.Fynboetal.:DeepimagingofQ2112+059:AbrighthostgalaxybutnoDLAabsorber
Compared to the work on local early type galaxies by
Kormendy (1977), the effective radius of the host galaxy of
Q2112+059fallsintheoverlappingregionbetween“compact”
and“normal”earlytypegalaxies.
Finally,sincethehostgalaxywehavedetectedisaspheroid
we can estimate the mass of the black hole using the correla-
tionsderivedbyMagorrianetal.(1998)assumingthatthecor-
relationfoundatz=0isalsovalidatz=0.457.Wederiveablack
holemassofabout9×109M⊙.
Acknowledgments
We thank B. Januzzi for providing us with results on metal
absorption lines in the FOS spectrum of Q2112+0519 prior
to their publication.We are gratefulto our refereeDr. Villar–
Martinforsuggestionsandcommentswhichhelpedclarifying
our manuscript on several important points and significantly
improvedthepresentationofourresults.
References
BertinE,ArnoutsS,1996,A&AS117,393B
FynboJ.U,MøllerP.,WarrenS.J.,1999,MNRAS305,849
FynboJ.U,MøllerP.,ThomsenB.,2000,In:‘AstrophysicsResearch
and Science Education’, edt. C. Impey, (Notre Dame Press), p.
237
FynboJ.U,ThomsenB.,MøllerP.,2000,A&A353,457
JannuziB.T.,BahcallJ.N.,BergeronJ.,etal.,1998,ApJS118,1
KormendyJ.,1977,ApJ218,333
LandoltA.U.,1992,AJ104,340
LanzettaK.,WolfeA.M.andTurnshekD.A.,1995,ApJ440,435
LeBrunV.,BergeronJ.,Boisse´P.,DeharvengJ.M.,1997,A&A321,
733
MagorrianJ.,TremaineS.,RichstoneD.,etal.,1998,AJ115,2285
McLure R.J., KuKula M.J., Dunlop J.S.,et al., 1999, MNRAS 308,
377
MøllerP.,Stiavelli,M.,Zeilinger,W.W.,1995,MNRAS276,979
MøllerP.,WarrenS.,1993,A&A270,43
MøllerP.,WarrenS.,1998,MNRAS299,661
RaoS.M.,TurnshekD.A.,ApJinpress(astro-ph/9909164)
SchneiderD.P.,GunnJ.E.,HoesselJ.G.,1983,ApJ268,476
StetsonP.,1987,PASP99,191
StetsonP.,1999,”User’sManualforDAOPHOTII”
Stiavelli,M.,Møller,P.,Zeilinger,W.W.,1993,A&A277,421
WolfeA.M.,LanzettaK.M.,FoltzC.B.,ChaffeeF.H.,1995,ApJ454,
698