Table Of ContentThe Astrophysical Journal,???:??–??,2010
PreprinttypesetusingLATEXstyleemulateapjv.03/07/07
THE ENVIRONMENTAL DEPENDENCE OF THE EVOLVING S0 FRACTION*
Dennis W. Just1, Dennis Zaritsky1, David J. Sand1,2,3, Vandana Desai4,5, Gregory Rudnick6
Accepted 2010 Jan 12
ABSTRACT
WereinvestigatethedramaticriseintheS0fraction,f ,withinclusterssincez ∼0.5. Inparticular,
S0
wefocusontheroleofthe globalgalaxyenvironmentonf bycompiling,eitherfromourownobser-
0 vations or the literature, robust line-of-sight velocity dispSe0rsions, σ′s, for a sample of galaxy groups
1
and clusters at 0.1<z <0.8 that have uniformly determined, published morphologicalfractions. We
0
find that the trend of f with redshift is twice as strong for σ < 750 km s−1 groups/poor clusters
2 S0
than for higher-σ, rich clusters. From this result, we infer that over this redshift range galaxy-galaxy
n interactions, which are more effective in lower-σ environments, are more responsible for transforming
a spiral galaxies into S0’s than galaxy-environment processes, which are more effective in higher-σ en-
J
vironments. The rapid, recent growth of the S0 population in groups and poor clusters implies that
2 large numbers of progenitors exist in low-σ systems at modest redshifts (∼0.5), where morphologies
1
and internal kinematics are within the measurement range of current technology.
Subject headings: Galaxies: Clusters: General — Galaxies: Groups: General — Galaxies: evolution
]
O
C 1. INTRODUCTION hosts the S0’s. Processes that are expected to operate
. bestinlower-σenvironments,wherethelowerrelativeve-
h The fraction of galaxies morphologically classified as
locities between galaxiesallow them to interact more ef-
p S0 (f ) increases by a factor of ∼ 3 in galaxy groups
S0
- and clusters over the past ∼ 5 Gyr, at the expense of fectively,include mergersandgalaxy-galaxyinteractions
o (Toomre & Toomre 1972; Icke 1985; Lavery & Henry
the spiral fraction (Dressler et al. 1997). This evolu-
r 1988; Byrd & Valtonen 1990; Mihos 2004). Those ex-
t tion has generally been interpreted as the result of the
s pected to work best in higher-σ environments,either di-
transformationof spirals into S0’s within dense environ-
a
rectly because of the high velocities, the deeper poten-
[ ments (Dressler et al. (1997); Fasano et al. (2000), here-
after F00; Smith et al. (2005); Postman et al. (2005); tial implied by the high velocities, or the higher den-
1 sity intracluster medium, include ram pressure strip-
Poggiantiet al. (2006); Desai et al. (2007), hereafter
v ping(Gunn & Gott1972;Abadi et al.1999;Quilis et al.
D07), although the physical mechanism remains un-
8 2000), strangulation (Larson et al. 1980; Bekki et al.
determined. As highlighted by Dressler (1980), the
0 2002), and harassment (Richstone 1976; Moore et al.
relationship between morphologies and environment
0 1998).
can help distinguish between hypothesized formation
2
To investigate the dependence of f on environment,
. mechanisms for S0’s. As practiced, this effort in- S0
1 we return to published morphological samples. We use
volves tracing galaxy populations as a function of
0 published visual morphological classifications as the in-
environment (Dressler 1980; Postman & Geller 1984;
0 dicator of galaxy type. Quantities related to f , such
Zabludoff & Mulchaey 1998; Helsdon & Ponman 2003), S0
1 as B/T and color distributions, have also been used to
increasingly at higher redshifts (Dressler et al. 1997;
: investigatesuchquestions,butmorphologiesprovidead-
v Kautsch et al. 2008; Wilman et al. 2009). Those studies
i in turn haveproduced the evidence for significantevolu- ditional,complementaryinformation. Infact,variousre-
X cent studies are suggesting that morphologicalevolution
tionoftheS0fraction(Dressler et al.1997),buthavenot
r examinedwhethertherateofevolutionitselfdependson is somewhat decoupled from the evolution of the stel-
a lar population (Poggianti et al. 2006; Tran et al. 2009).
environment.
Morphologies are available across a significant range of
WefocusontherelationshipbetweenS0evolutionand
redshifts and velocity dispersions, and significant effort
the velocity dispersion (σ) of the group or cluster that
hasbeenexpendedinputtingtheseonacommonfooting
*ObservationsreportedherewereobtainedattheMMTObser- across redshift (F00; D07). We compile an internally-
vatory,ajointfacilityoftheUniversityofArizonaandtheSmith- consistent set of velocity dispersions, recalculating the
sonian Institution; and based on data collected at the Magellan velocitydispersionusingeitherpreviouslypublishedindi-
Telescope,whichisoperatedbytheCarnegieObservatories. vidualgalaxyredshiftsorredshiftsfromourownobserva-
1StewardObservatory,UniversityofArizona,933NorthCherry
tions,toprovideameasureofenvironment. Again,alter-
Avenue,Tucson,AZ85721,USA
2Harvard Center for Astrophysics and Las Cumbres Observa- native measurements of environment exist, for example
toryGlobalTelescopeNetworkFellow X-rayluminositiescouldhavebeenused. However,X-ray
3Harvard-Smithsonian Center for Astrophysics, 60 Garden
measurements, particularly for low-mass, high-redshift
Street,Cambridge,MA02138,USA
4Division of Physics, Mathematics and Astronomy, California environments, are scarce and velocity dispersions pro-
InstituteofTechnology, Pasadena, CA91125,USA vide the most uniform and extensive data. Studies us-
5Spitzer Science Center, California Institute of Technology, ing different measures of either galaxy type or environ-
Pasadena,CA91125,USA
ment are mixed. For example X-ray luminosities corre-
6TheUniversityofKansas,DepartmentofPhysicsandAstron-
omy, Malott room 1082, 1251 Wescoe Hall Drive, Lawrence, KS late with B/T at z ∼ 0 (Balogh et al. 2002) and with
66045,USA early-type fraction at z > 1 (Postman et al. 2005), but
2 JUST ET AL.
velocity dispersions correlate only weakly with the frac- directly compare their results to F00, who present mor-
tionofredgalaxieswithinthevirialradius(Balogh et al. phological fractions for non-uniform apertures that cor-
2004). Apparently conflicting results such as these high- respondtoaperturesofradiispanningfrom∼500to700
lighttheimportanceofusingconsistentmeasurementsof kpc,D07usedtheclassiccosmologytomeasuremorpho-
both galaxy type and environment across redshift when logicalfractionswithinfixed600kpcradiusaperturesfor
investigating evolution. the EDisCS clusters. This selection of a fixed physical
In§2,wedescribethetwosampleswechosetouse,the aperture attempts to best match, on average, the F00
spectroscopic measurements we acquired in an attempt measurements,which are for a range of apertures. How-
to obtain velocity dispersions to complete the sample, ever, D07 demonstrated that a choice of aperture that
andthe calculationofa consistentset of velocity disper- scales with R (0.6R ) results in f values that are
200 200 S0
sionmeasurements. In§3,wepresentourresults,discuss in all cases within the uncertainty estimates. Lastly, re-
their implications in §4, and summarize in §5. When garding the magnitude limit, D07 classify galaxies down
computing the aperture size used for calculating the ve- to a fixed absolute magnitude across the redshift range,
locity dispersion, we assume H = 70 km s−1 Mpc−1, chosentomatchtheF00classificationprocedure,assum-
0
Ω = 0.3, and Ω = 0.7 (hereafter, the “Lambda cos- ingthe rest-framecolorsandI-bandmagnitude ofanel-
m Λ
mology”). However, for the aperture size within which liptical galaxy (details provided in D07). Applying the
galaxiesare included in the calculation of morphological incorrectcosmology(i.e. classicratherthanLambdacos-
fractions, H =50 km s−1 Mpc−1, Ω =1, and Ω =0 mology) results in differential magnitude limits across
0 m Λ
(hereafter, the “classic cosmology”)is assumed. the redshift range from 0.2 to 0.8 of a few tenths of a
magnitude, comparable to the uncertainties in the ob-
2. DATA servedmagnitudesthemselvesandthereforenotexpected
to have a noticeable effect.
2.1. Sample A sample of z ∼1 clusters with morphologicalclassifi-
Morphological fractions can depend sensitively on the cations and redshift measurements from Postman et al.
aperturewithinwhichclustermembersareclassifiedand (2005) also appear in D07. However, those morphologi-
on the absolute magnitude to which the classification is calfractions were notexplicitly matched to those of F00
done. As such, it can be quite difficult, and potentially (i.e.,bytakingstepstominimizesystematicdifferencesin
misleading, to use classifications from disparate sources. classification,suchasthosestatedabove)and,therefore,
D07 presented their own classification of a set of galax- we exclude these clusters to avoid any possible confu-
ies and combined these with a set from the literature sion in the interpretation of our results. Including these
for which they were able to closely match the classifica- clusters does not alter our main results.
tion procedure, the aperture used, and the magnitude
limit. Specifically, the sample presented in D07 con- 2.2. New and Archival Redshifts
sists of 23 galaxyclusters at z ∼0.1–0.5drawnfrom the
Of the 33 galaxy clusters and groups from the com-
F00 sample and 10 clusters at z ∼ 0.5–0.8 drawn from
bined sample of F00 and EDisCS, seven (∼ 20%; all
EDisCS.TheF00sampleinturnconsistsofnineclusters
from F00) do not have previously published velocity
at0.1<z<0.3addedbytheauthorsthemselves,fiveclus-
∼ ∼ dispersion measurements. All of these clusters are at
tersat0.15<z<0.3thateither appearedinCouch et al.
∼ ∼ z < 0.25, where less than half of the clusters have ve-
(1998) or were classified in a manner consistent with
locity dispersion measurements. This important part
that study, and nine clusters at 0.3<z<0.5 from the
∼ ∼ of parameter space drives much of the f -z trend ob-
MORPHSstudy(Dressler et al.1997;Smail et al.1997), S0
served in F00. Although several of these clusters have
all of which were classified in a consistent manner. D07
enough individual galaxy redshifts available in the lit-
usedtheF00procedurewhenclassifyinggalaxiestomin-
erature with which to calculate a reliable velocity dis-
imize systematic differences betweenthe two samples;in
persion (>10, see Beers et al. 1990), we still targeted
particular, the five authors who did the morphological ∼
them for observation because a higher number of red-
classification also reclassified the highest redshift clus-
shifts allows us to calculate a more robust velocity dis-
tersofF00(from0.3<z <0.5),followingthe samepro-
persion. Wetargetedsixclusters(Abell951,Abell1643,
cedure as the original authors (Smail et al. 1997), and
Abell 1878, Abell 1952, Abell 2192, and Abell 2658)
found good agreement.
using Hectospec (Fabricant et al. 2005) on the MMT
Errors on the morphological fractions for those from
between 2007 November to 2008 April. We observed
the ESO Distant Cluster Survey (EDisCS; White et al.
each cluster for a total of 30–60 minutes and mea-
2005) were computed using the method of Gehrels
sured redshifts using the iraf task rvsao. We used
(1986). The situation is somewhat more complicated
HSRED (e.g., §3.2 of Papovichet al. 2006) for the Hec-
for the F00 morphological fractions. We calculate the
tospec data reduction. We also targeted four clusters
uncertainties using the Gehrels method, but some of the
(Abell 1878, Abell 3330, Cl0054−27,and Cl0413−6559)
necessaryinformation,suchasthevariouscorrectionand
using the Inamori-Magellan Areal Camera and Spectro-
completenessfactors,arenotavailableandweinferthem
graph (IMACS; Bigelow et al. 1998) on the Magellan
indirectly from the data provided by F00. To test the
Baade telescope during two observation runs in 2008
sensitivity of our results to the uncertainties, we also do
June and 2008 September. IMACS data was reduced
all the analysis described subsequently using the quoted
using the COSMOS package8, following standard reduc-
uncertaintiesinF00,whichwerenotcalculatedusingthe
Gehrels method. None of the results (including the sta-
8 The Carnegie Observatories System for Multiobject Spec-
tistical significances quoted) change sufficiently between
troscopy was created by A. Oemler, K. Clardy, D. Kelson, and
the two approaches to alter any of our conclusions. To G.Walth. Seehttp://www.ociw.edu/Code/cosmos.
S0 EVOLUTION VS ENVIRONMENT 3
TABLE 1
Log of Observations
TotalRedshifts ClusterRedshifts
Cluster Date Telescope Instrument Measured Measured Notes
Abell951 2007Nov MMT Hectospec 23 19 ···
Abell2658 2007Oct MMT Hectospec 146 41 ···
Abell1952 2008Mar MMT Hectospec 131 46 ···
Abell2192 2008Mar MMT Hectospec 100 13 ···
Abell1643 2008Mar MMT Hectospec ··· ··· Lostduetoweather.
Abell1878 2008Apr MMT Hectospec ··· ··· Lostduetoweather.
2008Jun Magellan IMACS 25 18 ···
Cl0054−27 2008Jun Magellan IMACS ··· ··· Lostduetoweather.
Abell3330 2008Sep Magellan IMACS ··· ··· Lostduetoweather.
Cl0413−6559 2008Sep Magellan IMACS ··· ··· Lostduetoweather.
tion procedures. Based on our comparison of 15 objects sions,sothatallmeasurementsforthevelocitydispersion
for which previous redshift measurements exist, we cal- are calculated using the same method. We now describe
culate that our velocity measurement uncertainty is 86 our procedure for evaluating the velocity dispersion, in-
km sec−1. This is a conservative estimate in that we as- cluding our iterativeprocedure to define anaperture. In
signtheentiredifferencebetweenourmeasurementsand theend,wefindthatthevelocitydispersionshaveonlya
the published ones to ourselves. slightdependenceontheapertureaslongastheaperture
Alogoftheobservationsofthe clustersispresentedin is a significant fraction of the virial radius.
Table1. ThetargetgalaxiesareselectedfromtheNASA Starting with both the literature and newly-measured
ExtragalacticDatabase(NED)andsothereisnouniform redshifts, we include only those galaxies within 3 Mpc
selectioncriteria. We prioritizewhatappeartobe early- of the cluster center in our initial estimate of the veloc-
type galaxies and use whatever other information is in ity dispersion, although we do not always have spectro-
NED to maximize our return on cluster members, but scopic redshifts out to that radius. The cluster center
giventhe heterogeneityof the sourcematerialthe target is as defined in the previous studies and remains un-
sample is ill-defined. Furthermore, as with all multiob- changed through our procedure. Because of the small
ject spectroscopy, the effective selection is complicated number of spectroscopic members in most of these clus-
by fiber/slit allocation algorithms and then by the in- ters and the nature of the iterative procedure, we use
trinsic spectrum of an object. In detail, such biases can the initial center, which is often defined either by X-ray
lead to differences in measured velocity dispersions due contours, brightest cluster galaxy, or weak lensing con-
to differences in the dispersions of different morphogical tours rather than from the galaxy population centroid.
types within a cluster (cf. Zabludoff & Franx 1993), but FollowingHalliday et al.(2004), wealsoapply aredshift
hereweusethevelocitydispersionsonlyasaroughrank- cutof∆z =0.015abouttheredshiftofthecluster. Only
ing of environment and are not interested in differences redshifts from the literature with quoted errors <0.01
∼
at the ∼10% level. Both of these spectrographsprovide are included; a difference in redshift of 0.01 corresponds
large (> 24 arcmin) wide fields-of-view, so the galax- to 3000 km s−1, which is much larger than the veloc-
ies sample the dynamics well beyond the cluster core. ity dispersion itself for even our richest clusters. We use
These observations provide enough redshifts for all but the biweight statistic of Beers et al. (1990) to calculate
one cluster (Abell 1643 from the F00 sample, which was the value of σ, which gives robust velocity dispersion
observed during poor weather) to measure the velocity measurements with as few as ∼ 10 galaxy redshift mea-
dispersions for nearly the full sample (32/33 clusters). surements. The velocity dispersions are corrected to be
The other clusters lost due to weather had enough red- rest-frame velocity dispersions. Regarding our choice of
shifts to reliably measure the velocity dispersion. In the initial aperture, we find that varying it within the range
analyses that follow, only these 32 clusters are included. ∼ 1.5–3 Mpc affects the velocity dispersion by <10%
∼
In all, we present new redshift measurements for five for all our clusters, most often <5%. In fact, the ve-
∼
clusters (four fromHectospec observationsandone from locity dispersion calculated within any aperture varying
IMACS observations). Although this is a small number from∼1.5–3Mpc (when notimplementing our iterative
of clusters relative to the entire sample, they lie in the aperture scheme outlined below) changes by <15% for
∼
regionofparameterspaceresponsibleformuchoftheS0 all our clusters except Abell 951 and Abell 2658, whose
evolution(i.e.,low-z,high-f ). Inadditiontothesenew velocity dispersions change by ∼50%within that range.
S0
redshift measurements, we took advantage of the large After calculating the velocity dispersion, 3σ outliers are
numberofpreviously-measuredredshifts availableinthe rejectedandtheprocessiterateduntilnooutliersremain
literature. These redshifts came from various studies, (see §5.2 of Halliday et al. 2004). This value of σ is then
andweusedNEDtosearchforandselectthedata. This used to calculate an estimated virial radius, R , using
200
provides improved velocity dispersion measurements for Equation (5) of Finn, Zaritsky, & McCarthy (2004):
many of the clusters.
σ
R =1.73 [Ω +Ω (1+z)3]−1/2 h−1 Mpc.
200 1000km s−1 Λ 0 100
2.3. Velocity Dispersion Measurements
(1)
Wecalculatevelocitydispersionsfortheentiresample, AnewcutisappliedatR ,andtheprocessiteratedun-
200
includingthosewithpreviouslymeasuredvelocitydisper- til convergence. Sometimes R is greater than 3 Mpc,
200
4 JUST ET AL.
TABLE 2
MainProperties of the Sample
Name z σ R200 Niter Nmem fE fS0 fS fE+S0 Sample
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
A3330 0.091 732+−28327 1.73 ··· 9 0.307+−00..008790 0.501+−00..008834 0.193+−00..008555 0.807+−00..005865 1
A389 0.116 662+−117350 1.55 3 40 0.353+−00..009848 0.629+−00..009897 0.019+−00..007104 0.981+−00..001740 1
A951* 0.143 537+−16268 1.24 4 23 0.313+−00..102975 0.649+−00..019289 0.038+−00..009361 0.962+−00..003916 1
A2218 0.171 1520+−17142 3.45 1 98 0.437+−00..009825 0.240+−00..009607 0.324+−00..008835 0.677+−00..008853 1
A1689 0.181 1876+−9781 4.24 1 206 0.363+−00..006531 0.363+−00..006531 0.274+−00..005498 0.726+−00..004589 1
A2658* 0.185 498+−9598 1.12 ··· 15 0.491+−00..112512 0.410+−00..115129 0.099+−00..103602 0.901+−00..016320 1
A2192* 0.187 635+−113192 1.43 ··· 16 0.287+−00..008756 0.511+−00..007979 0.202+−00..009554 0.798+−00..005945 1
A1643 0.198 ··· ··· ··· ··· 0.242+−00..007703 0.476+−00..007950 0.282+−00..007755 0.718+−00..007755 1
A1878* 0.222a 828+−218305 1.83 1 13 0.364+−00..100863 0.282+−00..017004 0.354+−00..101763 0.646+−00..017136 1
A2111*b 0.229 1129+−18201 2.49 2 80 0.465+−00..006667 0.336+−00..006643 0.200+−00..006447 0.800+−00..004674 1
A1952* 0.248 718+−229039 1.57 1 18 0.413+−00..007788 0.380+−00..007821 0.207+−00..008522 0.793+−00..005822 1
AC118 0.308 1748+−91939 3.69 1 83 0.246+−00..006513 0.527+−00..006644 0.227+−00..006429 0.773+−00..004692 1
AC103 0.311 965+−18312 2.03 1 55 0.301+−00..007781 0.313+−00..008664 0.386+−00..007851 0.614+−00..008715 1
AC114 0.315 1889+−8714 3.98 1 196 0.223+−00..004591 0.318+−00..006510 0.459+−00..006508 0.541+−00..005680 1
Cl1446+2619 0.370 1397+−228178 2.85 2 20 0.338+−00..008720 0.248+−00..007648 0.415+−00..008762 0.585+−00..007826 1
Cl0024+1652 0.391 764+−4500 1.54 2 235 0.348+−00..008746 0.227+−00..007750 0.426+−00..008825 0.574+−00..008852 1
Cl0939+4713 0.405 1331+−91609 2.65 1 72 0.250+−00..009658 0.257+−00..009770 0.493+−00..100806 0.507+−00..018060 1
Cl0303+1706 0.418 769+−19240 1.52 2 56 0.227+−00..008742 0.126+−00..007554 0.647+−00..008858 0.353+−00..008885 1
3C295 0.461 1907+−124025 3.69 1 32 0.463+−00..019031 0.197+−00..009657 0.341+−00..100806 0.659+−00..018060 1
Cl0412−6559 0.510 626+−211709 1.17 1 19 0.347+−00..008899 0.090+−00..006543 0.564+−00..018005 0.437+−00..100850 1
Cl1601+42 0.539 749+−9776 1.38 1 55 0.509+−00..006648 0.165+−00..006412 0.326+−00..006588 0.674+−00..005688 1
Cl0016+16 0.545 1307+−111123 2.41 2 99 0.502+−00..007860 0.208+−00..007565 0.291+−00..007649 0.709+−00..006794 1
Cl0054−27 0.560 700+−228544 1.28 2 17 0.310+−00..008777 0.246+−00..008743 0.444+−00..008952 0.556+−00..009825 1
Cl1138−1133 0.480 746+−9769 1.43 1 49 0.305+−00..116240 0.095+−00..101834 0.600+−00..114652 0.400+−00..116425 2
Cl1232−1250 0.541 1171+−17505 2.16 1 54 0.350+−00..004400 0.170+−00..003300 0.470+−00..004400 0.530+−00..004400 2
Cl1037−1243 0.578 344+−7634 0.58 1 16 0.281+−00..112446 0.000+−00..100090 0.625+−00..113586 0.281+−00..112446 2
Cl1227−1138 0.636 584+−9730 0.64 ··· 22 0.290+−00..116356 0.146+−00..105975 0.394+−00..116674 0.436+−00..116602 2
Cl1054−1146 0.697 603+−117400 1.01 2 33 0.245+−00..007619 0.000+−00..003060 0.755+−00..006791 0.245+−00..007619 2
Cl1103−1245b 0.703 235+−28063 0.39 ··· 9 0.250+−00..102800 0.000+−00..007000 0.750+−00..018200 0.250+−00..102800 2
Cl1040−1155 0.704 535+−8791 0.89 2 15 0.377+−00..113166 0.066+−00..009538 0.419+−00..114115 0.444+−00..114213 2
Cl1054−1245 0.750 570+−114013 0.93 2 22 0.300+−00..100970 0.267+−00..100847 0.433+−00..110082 0.567+−00..110028 2
Cl1354−1230 0.762 732+−24383 1.18 1 21 0.170+−00..007500 0.290+−00..007600 0.550+−00..008700 0.450+−00..007800 2
Cl1216−1201 0.794 1066+−8824 1.69 1 67 0.490+−00..003200 0.220+−00..002200 0.270+−00..002200 0.710+−00..002200 2
Note. —(1)ClusterName. Anasterisk(*)denotesaclusterwithnewdata;(2)Redshift;(3)VelocityDispersioninunitsofkms−1;
(4) Virial Radius in units of Mpc; (5) Number of iterations until convergence, see §2.3; (6) Number of redshifts ultimately used in
calculatingthevalue in Column3; (7) Fractionof Ellipticalgalaxies; (8) Fractionof S0galaxies; (9) Fractionof Spiralgalaxies; (10)
FractionofElliptical+S0galaxies;(11)Sample,1-Fasanoetal. (2000),2-EDiscS
a This redshiftis differentthan that which appearsin F00, who use z = 0.254. The origin of the discrepancycan be tracedback to
Sandage,Kristian, & Westphal(1976), wheretwo potentialredshiftsfortheclusterare givenat z =0.222and z =0.254. Thelower
valuewasassumedtobeforeground,sothelattervaluewasadoptedinlaterstudies. However,withournewlymeasuredredshiftsof18
galaxiesneartheclusterpositionthatarewithin±0.015ofthelower valueandonly2thatarewithin±0.015ofthehighervalue,we
adoptz=0.222astheclusterredshift.
b Whilenonewredshiftshavebeenmeasuredforthiscluster,it’svelocitydispersionhasnotbeenpublishedasfarastheauthorsknow,
andispresentedforthefirsttimehere.
resulting in moreredshifts being included in the later it- sion obtained with that aperture. We estimate the 1σ
erations. The main properties of our 32 cluster sample, errors by selecting random subsamples of the data from
including these new velocity dispersion measurements, which to evaluate the velocity dispersion.
appears in Table 2. The values for R , the number For three of the clusters, Abell 1952, Cl0024+1652
200
of iterations until convergence, N , and the number (both part of the F00 subsample), and Cl1037−1243
iter
of redshifts used in the final iteration, N , appear (partofthe EDisCSsubsample), there is clear9 evidence
mem
in Columns 4, 5, and 6, of Table 2, respectively. For of substructure in their phase-space plots. We remove
five of the clusters, Abell 3330, Abell 2658, Abell 2192,
Cl1103−1245b, and Cl1227−1138, this process of iter- 9ForCl1037−1243,thesubstructureonlybecomesobviousafter
ation removes galaxy redshifts until there are too few vtheleoficirtsitesitoefra≈tio−n.15T0w0oangdala−x2ie0s00lockamteds−21′′raeplaatritveontoththeeskcyluhstaevre.
(<10) to reliably calculate a velocity dispersion. For
∼ Due to the relatively few galaxies in the cluster (16), these two
these systems,the velocity dispersionis calculated using galaxieschange thevelocity dispersionfrom≈300to650kms−1
afixed3Mpccut,andtheR thatappearsinTable2is when they are included (such that they are then not excluded in
200
calculated from Equation (1) using the velocity disper- the 3σ clipping). Inspection of the histogram leads us to believe
the former value is more accurate, although adopting the latter
S0 EVOLUTION VS ENVIRONMENT 5
by hand the galaxiesbelonging to these subgroups when
calculating the velocity dispersion for the three clusters.
Asidefromthisstep,thevelocitydispersioniscalculated
using the same procedure outlined above.
We present velocity histogramsfor the clusters in Fig-
ure 1 (placed at the end of the paper). The bin size
is set to one-third the velocity dispersion, and the red-
shifts plotted are those that remain after the various
cuts/iterations in the calculation (see above). Over-
plotted on each panel is a Gaussian with the mea-
sured velocity dispersion, normalized to the area of
the histogram. Our newly calculated velocity disper-
sions are in good agreement with those previously mea-
sured for the EDisCS clusters (Halliday et al. 2004;
Milvang-Jensen et al. 2008), but tend to give larger val-
ues for some of the σ > 1000 km s−1 F00 clusters (see Fig. 2.— S0 Fraction (fS0) plotted against redshift. Triangles
representF00systems,whilecirclesrepresentEDisCSsystems.
D07, and references therein). This discrepancy is not
duetoaperture-sizeeffects,butmorelikelyfromthe dif-
ferent methods employed in calculating the velocity dis- as we have noted earlier some of the most interesting
persion. Althoughthe velocitydispersionwascalculated clusters were missing such measurements.
using the Lambda cosmology, while the morphological In Figure 3, we present f plotted against velocity
S0
fractions were calculated within an aperture defined by dispersion for our sample. Although the f -z trend in
S0
theclassiccosmology,wefindthatthevalueofσ isfairly Figure 2 appearsmuch stronger than any trend between
insensitive to aperture size (see above). f and σ in Figure 3, we quantify which is the more
S0
Lastly, we address the impact of observational uncer- dominantwithapartialcorrelationanalysis. Thepartial
tainties on our measured velocity dispersions. As men- correlationcoefficient ρ,
tioned previously, comparison of our redshift measure-
r −r r
ments with those in the literature suggestsa single mea- ρ= A,B A,C B,C , (2)
surementuncertaintyof86kms−1. Thisislikelytobe a (1−r2 )(1−r2 )
q A,C B,C
significantoverestimateformostsystems,butweusethis
value to estimate the impact on the dispersions. If we is useful for disentangling the interdependence between
simply addrandomvelocities usinga Gaussianwith this three variables (A, B, and C), where one wants to ac-
σ toanintrinsicGaussianofwidthcommensuratetothe count for the influence of the third variable (C) on the
line-of-sight velocity distribution of a specific group and correlation of the first two. It is normalized to +1 for
cluster, we find that even in for our lowest velocity dis- a perfect correlation, 0 for no correlation, and −1 for a
persion system (Cl1102-1245b) the observational errors perfectanticorrelationbetweenAandBafteraccounting
inflate the dispersion by less than 20 km s−1. This un- for C. However, the distribution of ρ does not approxi-
certainty is in allcases significantly less than the quoted mate a normal distribution, so we follow the work of
errors on the velocity dispersion and does not affect our Kendall & Stuart(1977)inusingastatisticZ ,where
B,C
results. B is the dependent variable and C is the controlledvari-
able. Z is defined as
B,C
3. RESULTS
1 (1+ρ)
We explore the environmental dependence (charac- Z = ln . (3)
terized by velocity dispersion) of the apparent evolu- B,C 2 (1−ρ)
tion of f with redshift (Figure 2). Our sample spans
S0 with a variance σ2 =1/(N−2), where N is the number
a range of dispersions from that typical of groups (∼ Z
ofdatapoints. Themorepositive(negative)thevalueof
200−500kms−1)topoorclusters(∼500−750kms−1)to
Z the stronger the correlation (anticorrelation). We
richclusters (>750km s−1). Although there is no strict B,C
∼ treatz andσ astheindependentandcontrolledvariable,
ruleforwhatvelocitydispersionconstitutesagroupver-
and then vice-versa. We find a stronger correlation for
sus a cluster, in what follows we use the above conven-
f with redshift than with σ, with Z =−0.91±0.18
tion. S0 z,σ
and Z =−0.02±0.18.
σ,z
3.1. Analysis of the Full Sample
3.2. Analysis of Groups vs. Clusters
We begin by determining whether a relationship be-
The results of the previous correlationanalysis do not
tween f and environment (velocity dispersion) exists
S0
necessarily imply that environment (velocity dispersion)
across the full redshift range. Due to the selection cri-
plays no role. From Figure 3, it is apparent that there
teria for the F00 sample (clusters were selected based
is a wide spread in f below ∼750 km s−1 and a much
on being “well-studied”), it is possible that some unap- S0
narrower spread above. We therefore split the sample
preciated selection bias manifests itself as a correlation
intoahigh-σ binandalow-σ binatthisvaluetoinvesti-
between f and z. Figure 6 of D07 shows a weak trend
S0
gatetheeffectofenvironmentonthef -z relation. This
between f and σ, although they were limited to the S0
S0
choice divides the sample into nearly equal parts as well
subsetofF00clusterswithdispersionmeasurementsand
asintosamplesthataremoretypicalofgroups/poorclus-
valuedoesnotsignificantlychangeourresults. ters(σ<∼750km s−1)andrichclusters(σ>∼750km s−1).
6 JUST ET AL.
Some of the clusters have suspiciously high velocity dis-
persions (σ>1500 km s−1) and are presumably unre-
∼
laxed systems (e.g., A1689). Nevertheless, given our
gross binning scheme they are still likely to be systems
with σ>750 km s−1 and placed in the appropriate ve-
∼
locity dispersion bin. Selecting a boundary anywhere
up to 1050 km s−1 or down to 650 km s−1 (after which
the number of clusters in the low-σ bin drops sharply)
leavesthe results thatfollowqualitativelyunchanged,as
does removing the clusters with σ>1500 km s−1 from
∼
the analysis.
InFigure4,weshowf plottedagainstredshiftinthe
S0
high-σ and low-σ bins. While the f -z trend is evident
S0
inthegroups/poorclusters,thecorrelationappearstobe
muchweaker,ifpresentatall,inthe richclusters. Using
uncertainty-weighted least-squares fitting, we find that Fig. 3.—S0fraction(fS0)plottedagainstgalaxyclustervelocity
dispersion(σ). Thereisnosimplecorrelationbetweenthesequan-
the slope for the groups/poor clusters, −0.75±0.14, is
tities but clearly a divergence of fS0 at low σ. The clusters with
steeper than the slope for the rich clusters, −0.18±0.09 z < 0.3 (squares) are entirely non-MORPHS clusters from F00,
(a 3.4σ difference in slope). For the high-σ clusters, one theclusters from0.3<z<0.5(stars) aremostlyMORPHSclus-
ters from F00, and the clusters with z >0.5 (crosses) are mostly
may worry that there is only one data point at z > 0.6,
EDisCSclustersfromD07.
whichhasananomalouslysmallerrorof±0.02andthere-
fore stronglyinfluences the slope. To explorethe impact
ofthisoneclusteronthefit,wehaveassigneditanuncer- must be affected by selection effects and methodology,
tainty equal to the scatter in f for the high-σ clusters, may be a result of unappreciated biases. The ability
S0
±0.07. With this larger uncertainty estimate the new to distinguish between S0’s and ellipticals at higher red-
slope is −0.38±0.13, resulting in only a 1.9σ difference shifts, or other problems associated with morphological
in slope between the low- and high-σ clusters. To bol- classification, could in principle result in spurious corre-
ster the case for the flat relationship among the massive lations. With this specific issue in mind, we investigate
clusters,wecomparethemorphologicalfractionstothose the relationships of various morphologicalfractions with
from Postman et al. (2005). Although we arguedin §2.1 redshift and velocity dispersion. We have already ar-
against using these clusters for our statistical analyses, gued against a redshift-dependent classificationproblem
they support our finding that the relationship with red- in E’s vs S0’s (see above). What if there is ananalogous
shift is nearly flat for massive clusters (Figure 4). We problem with environment? For example, if ellipticals
concludethat the difference inbehaviorbetweenthe low are more common in the more massive environments to
and high-σ clusters is not the result of the one high-z the limits of our classification,and if a constant fraction
EDisCS cluster. Lastly, the two lowest-σ clusters in the of those are misclassifiedas S0’s, then f would appear
S0
EDisCSsamplehavef =0andarepotentiallyveryun- higher in more massive environments. For Figure 5 we
S0
usual, although excluding them from this analysis does also conclude that there is no discernible difference in
not alter the results. the f as a function of environment over the range of
E
The clusters driving most of the trend in the environments explored here.
groups/poor clusters are the high-f systems at low z. We now remove the ellipticals from consideration and
S0
Among those at z <0.3,there is anapparentdichotomy consider a plot similar to Figure 4 in which we replace
between those with a dense concentration of ellipticals the ordinate, f , with N /(N +N ), where N and
S0 S0 S S0 S0
toward the cluster center and those less centrally con- N are the numbers of S0’s and spirals in each cluster,
S
centrated, in the sense that the latter have higher f respectively (Figure 6). The dichotomy in the rate of
S0
(F00). Therefore,itisalsopossiblethatS0evolutionde- evolution between low-σ groups/poor clusters and high-
pends further on an environmental property marked by σ rich clusters remains, with slopes of −1.19±0.24 and
thedistributionofclusterellipticals. Evenso,thereisan −0.07± 0.17, respectively (a 3.8σ difference in slope).
increaseinf sincez ∼ 0.5(F00)whenconsideringthe The difference between the morphological fractions of
S0
high-andlow-ellipticalconcentrationsystemsseparately. the two environments at low redshifts indicates that the
InFigure5,weshowtheellipticalfraction(f )plotted morphologicaldistinction between spirals and S0’s is re-
E
againstredshiftforthe entiresample,the low-σ subsam- flecting a true underlying difference between the two en-
ple, and the high-σ subsample. In all three cases, there vironments. The difference in evolutionary trends does
is no significant trend of f with redshift. This argues not, unfortunately, necessarily imply that the trends are
E
againstamisclassificationbetweenS0’sandellipticalsas unaffected by misclassification; if the two environments
the origin of the S0 evolution. havedifferent intrinsic fractions of spiralsand S0’s,then
redshift-dependentmisclassificationcouldaffecteachen-
4. DISCUSSION
vironment differently.
As we have described, previous studies have found Giventheresultsdescribedsofar,weinterpret(asoth-
a factor of ∼ 3 increase in f between z ∼ 0.5 and ers before have, e.g. Dressler et al. 1997; Fasano et al.
S0
z ∼ 0, with a corresponding decrease in the spiral frac- 2000; Smith et al. 2005; Poggiantiet al. 2006) that the
tion and a constant elliptical fraction (Dressler et al. evolvingS0fractionrepresentsthetransformationofspi-
1997; Fasano et al. 2000). Some authors (e.g., Andreon rals into S0’s. The difference here is that the S0 evo-
1998) have noted that the trends, which at some level lution (over these redshifts) is taking place primarily in
S0 EVOLUTION VS ENVIRONMENT 7
Fig. 4.—S0Fractionplottedagainstredshiftinalow-mass,σ<750kms−1 bin(left)andahigh-mass,σ>750kms−1 bin(right)for
theF00andEDisCSclusters(trianglesandcircles,respectively);thisbinningroughlysplitsthesampleintogroups/poorclustersandrich
clusters,respectively. Thetrend isclear inthegroups/poor clusters sample(withaslopeof−0.75±0.14), buthardlyevident inthe rich
clusters (with a slope of −0.18±0.09), consistent with the idea that morphological transformation is taking place in group/poor cluster
environmentsoverthisredshiftrange. ThesubsetofclustersfromPostmanetal.(2005)withvelocitydispersionmeasurementsareplotted
asopendiamonds;theseclustersarenotusedinthefitsforreasonsgivenin§2.1andareshownforillustrativepurposesonly.
groups/poorclusterswithσ<750kms−1(Figure4),sug- distinctly, Christlein & Zabludoff (2004) find that S0’s
∼
gesting that this is the location of S0 formation. This differ from normal spirals due to a higher bulge lu-
result then supports the hypothesis that direct galaxy minosity rather than fainter disks, and interpret this
interactions, i.e. mergers and/or close tidal encoun- as requiring bulge growth during S0 formation. They
ters, are the dominant mechanisms in converting spi- conclude that such formation mechanisms as strangula-
rals into S0’s over the redshift interval examined. The tionandrampressurestrippingare thereforedisfavored.
value of σ where galaxy-galaxy processes dominate and Hinz et al. (2003) arguethatthe largescatter they mea-
where galaxy-environment process dominate is not the- sure in the local S0 Tully-Fisher relationsupport forma-
oretically well constrained. Although we choose a cutoff tionmechanismsthatkinematicallydisturbthe galaxies,
at750kms−1 to divide thesample intoequalparts,and i.e. interactions. The unique aspect of our observations
expect mergers and/or tidal interactions to dominate in is that we establish both the redshift and the environ-
the low-σ subsample, the division into two subsamples ment at which this formation is occurring. Thereby, we
onlycrudely reflectsa distinctionofenvironmentswhere identifytheexactplacetofocusfurtherinvestigationand
different physical effects may dominate. However, the perhaps distinguish the progenitors. Fortunately, this
existence of high f systems with low velocity disper- evolution happens at redshifts that are relatively easily
S0
sions demonstrates that neither the nature or nurture of accessed with current technology.
massive environments is necessary to the formation of Although S0 evolution is seen primarily in the low-
S0’s. σ clusters and the values of f reach between 0.5 and
S0
The conclusion that groups are the site of S0 forma- 0.6 at z ∼ 0, the rate of S0 formation must reverse it-
tion,andthereforethatmergers/interactionsarethefor- self at some low value of the velocity dispersion so as
mation mechanism, has been arrived at in various ways. not to overpopulate the field with S0’s (the local field
Wilman et al. (2009) find a high f already in place f ∼ 0.10; Sandage & Tammann 1987). Determining
S0 S0
in z ∼ 0.5 groups. Poggiantiet al. (2009) find more- this transitional value of the velocity dispersion would
pronouncedS0 evolutioninclusters with σ<800kms−1 further aid our understanding of the environmental pro-
∼
by comparing a z ∼ 0 sample to a high-z sample, al- cesses at work. For example, one might find that this
thoughtheirinclusionofthesameEDisCSclustersmeans velocity dispersion corresponds to that of environments
the resultsarenotentirelyindependentfromours. More where the probability of interactions in a Hubble time
8 JUST ET AL.
Fig. 5.—Ellipticalfraction(fE)plottedagainstredshiftforthe Fig. 6.—NS0/(NS+NS0),whereNS0 andNS arethenumbers
fullsample(top),thelow-mass,σ<750kms−1 bin(middle),and ofS0’sandspirals,respectively,plottedagainstredshiftinthelow-
the high-mass, σ > 750 km s−1 bin (bottom). Symbols are the mass,σ<750kms−1bin(left)andthehigh-mass,σ>750kms−1
same as in Figure 2. Neither the full sample nor the subsamples bin(right). Thedashedlineshowsbest-fittrends,withsignificantly
showasignificanttrendinellipticalfractionwithredshift. differentslopesof−1.19±0.24and−0.07±0.17(a3.8σdifference
inslope)intheleftandrightpanels,respectively. Symbolsarethe
sameasinFigure2.
becomeunlikely(e.g. slightlymoremassivethantheLo-
cal Group). Our lowest-z clusters extend down to ∼500
kms−1,whilethez ∼0clustersofPoggiantiet al.(2009) Wechsler et al. (2002) models, and fS0,gr is the S0 frac-
probe down to ∼ 400 km s−1, setting an upper limit on tion for low-z groups, for which we adopt a conserva-
where the trend must reverse (our two lowest velocity tive value of 0.4. From our best-fit trend in the high-σ
dispersion systems, both with σ <400 km s−1, but high panel of Figure 4, the S0 fraction for a massive cluster
redshifts, have fS0 ∼ 0, perhaps suggesting where this at z = 0.5, fS0,z=0.5, is 0.25. Using Equation (4) gives
turnover occurs). fS0,z=0 ≈0.3,consistentwithourbest-fittrendatz =0.
So far, we have not accounted for the effects of the hi- Therefore, this simple model suggests that the trend of
erarchicalgrowthof groups and clusters on the question increasing fS0 with z in the high-σ clusters could be ac-
of S0 evolution. Groups and clusters grow over time, countedforsolelybytheaccretionofS0-richgroups. Re-
accreting galaxies from the field and/or groups, so that gardlessoftheactualaccretionhistory,weconcludethat
systemsatz ∼0.8withaparticularvalueofσdonotcor- the accretionofatleastsomeS0-richgroupswillexplain
respondtothoseofthesameσ atz =0. Ithasgenerally part of the increase in fS0 in clusters.
been assumed, due to the expectation that S0’s would The results presented here (and elsewhere) that S0
be rarer in low density environments, that any accre- galaxies are forming at relatively low redshifts (z <0.5)
tion these systems experience would be S0-poor, hence and in low-σ groups, implies that we should be able
the need to transform some fraction of these galaxies to identify and study both the progenitor class and
into S0’s. From Figure 4, we now know that this is not the galaxies undergoing this transition. Post-starburst
the case, at least for z < 0.3. In fact, at low z it ap- galaxies are commonly suspected to be late-time exam-
pears that high-z clusters could increase their f over plesofthelatter(Dressler et al.1985;Couch & Sharples
S0
time by accreting these smaller systems without requir- 1987; Yang et al. 2004, 2006). If so, this transformation
ing any morphological transformation mechanism. How affectsboththemorphologyandstellarpopulationofthe
muchoftheobservedf -z trendinthe high-σ richclus- galaxy and we expect based on our results that 1) S0’s
S0
ters could simply be due to the accretion of smaller, S0- inrich clusters atz =0 will have mostly oldstellar pop-
rich groups/poor clusters similar to those in our low-σ ulations (>∼7 Gyr) because most of their S0 population
subsample? has been in place since z ∼0.8 and 2) the S0’s in low-σ,
Toestimatetheincreaseinthenumberofclustergalax- z =0clusterswillhaveamixofyoungandoldstars,with
ies with redshift, we note that the mass of rich clus- roughly 50% of the S0’s having a significant fraction of
ters at z ∼ 0.5 typically increases ∼ 40% by z = 0 their stars that are younger than ∼3 Gyr old (evidence
(Wechsler et al. 2002), and assume that this increase in for some relatively young S0 galaxies in the field now
mass corresponds to the same relative increase in the exists; Moran et al. 2007; Kannappan et al. 2009).
number of cluster galaxies. We also assume that the
mass accretion comes in the form of our low-σ groups. 5. CONCLUSION
To the degree that field galaxies, with their lower fS0, By compiling a large set of clusters with both
accountforthe accretedmass thenthis modelwillbe an internally-consistent morphological classifications and
overestimate of the effect. The final S0 fraction fS0,z=0 uniform velocity dispersions, σ, we examined the rate
in this simple model is ofchangeinthe S0fraction, f ,with redshiftasa func-
S0
f +ηf tionofenvironment. We showthatforourentiresample
S0,z=0.5 S0,gr
fS0,z=0 = 1+η , (4) fS0 is primarily correlated with redshift and not signifi-
cantly correlated with velocity dispersion. However, the
wheref istheS0fractionoftheclusteratz =0.5, evolution of f with redshift is much stronger among
S0,z=0.5 S0
η is the fractional increase in number of cluster galax- σ < 750 km s−1 galaxy groups/poor clusters than in
ies from z = 0.5 to z = 0, i.e. η = 0.4 based on the higher-σ rich clusters. We interpret this result to mean
S0 EVOLUTION VS ENVIRONMENT 9
that direct processes like galaxy mergers, which are ex- mation. We cannot measure the evolution of f as a
S0
pectedtodominateinlower-σenvironments,arethepri- function of local density from our data due to the small
marymechanismsformorphologicaltransformationover number of spectroscopic members per system, but both
the redshift range explored, 0<z<0.8. largercluster/groupsamples andmore redshifts per sys-
∼
Furtherstudieswouldbenefitfromalargersamplesize, tem would enable such a study.
in particular having f and σ measurements for both
S0
groups/poor clusters and rich clusters with comparable
numbers across a similar range in redshift. This study We thank the anonymous referee for insightful com-
highlights the importance of having velocity dispersion ments that improvedthe contentandpresentationofhis
measurementsinevolutionarystudies,sothatonecanac- paper. DJ thanks Michael Cooper for useful conversa-
countforanyenvironmentaldependenceoftheevolution tionsandDanielChristleinforhelpingwiththeMagellan
itself. In particular, we emphasize that more complete observations. DZacknowledgesfinancialsupportforthis
samplesofenvironmentsareneededandthatlargenum- work from NASA LTSA award NNG05GE82G, NASA
bersofredshiftspersystemarenecessarytoconvincingly XMMgrantsNNX06AG39A,NNX06AE41G,andNASA
measurevelocitydispersionsoflow-masssystems. Lastly, Spitzergrant1344985. DZ alsothanksthe NYUPhysics
asemphasizedby Dressler(1980) andPostman & Geller Department and CCPP for their hospitality during his
(1984), local density may be a critical factor in S0 for- visit.
REFERENCES
Abadi,M.G.,Moore,B.,&Bower,R.G.1999,MNRAS,308,947 Lavery,R.J.,&Henry,J.P.1988,ApJ,330,596
Andreon,S.1998,ApJ,501,533 Mihos,J.C.2004,inClustersofGalaxies:ProbesofCosmological
Balogh, M. L., Baldry, I. K., Nichol, R., Miller,C., Bower, R., & StructureandGalaxyEvolution,ed.J.S.Mulchaey,A.Dressler,
Glazebrook, K.2004,ApJ,615,L101 &A.Oemler,277–+
Balogh,M.L.,etal.2002, ApJ,566,123 Milvang-Jensen,B.,etal.2008, A&A,482,419
Beers,T.C.,Flynn,K.,&Gebhardt, K.1990,AJ,100,32 Moore,B.,Lake, G.,&Katz,N.1998,ApJ,495,139
Bekki,K.,Couch,W.J.,&Shioya,Y.2002,ApJ,577,651 Moran,S.M.,Ellis,R.S.,Treu,T.,Smith,G. P.,Rich,R.M.,&
Bigelow, B. C., Dressler, A. M., Shectman, S. A., & Epps, Smail,I.2007, ApJ,671,1503
H. W. 1998, in Presented at the Society of Photo-Optical Papovich,C.,etal.2006,AJ,132,231
Instrumentation Engineers (SPIE) Conference, Vol. 3355, Pello,R. etal.2009,˚a,inpress
Society of Photo-Optical Instrumentation Engineers (SPIE) Poggianti, B. M., Smail, I., Dressler, A., Couch, W. J., Barger,
ConferenceSeries,ed.S.D’Odorico,225–231 A.J.,Butcher,H.,Ellis,R.S.,&Oemler,A.J.1999,ApJ,518,
Byrd,G.,&Valtonen, M.1990,ApJ,350,89 576
Christlein,D.,&Zabludoff, A.I.2004,ApJ,616,192 Poggianti,B.M.,etal.2006,ApJ,642,188
Couch, W. J., Barger, A. J., Smail, I., Ellis, R. S., & Sharples, —.2009, ApJ,697,L137
R.M.1998,ApJ,497,188 Postman,M.,&Geller,M.J.1984, ApJ,281,95
Couch,W.J.,&Sharples,R.M.1987,MNRAS,229,423 Postman,M.,etal.2005, ApJ,623,721
Desai,V.,etal.2007,ApJ,660,1151 Quilis,V.,Moore,B.,&Bower,R.2000,Science,288,1617
Dressler,A.1980,ApJ,236,351 Richstone,D.O.1976,ApJ,204,642
Dressler,A.,Gunn, J.E.,&Schneider,D.P.1985, ApJ,294,70 Sandage, A.,Kristian,J.,&Westphal, J.A.1976,ApJ,205,688
Dressler,A.,etal.1997,ApJ,490,577 Sandage, A., & Tammann, G. A. 1987, A revised Shapley-Ames
Fabricant,D.,etal.2005,PASP,117,1411 Catalogofbrightgalaxies
Fasano, G., Poggianti, B. M., Couch, W. J., Bettoni, D., Smail, I., Dressler, A., Couch, W. J., Ellis, R. S., Oemler, A. J.,
Kjærgaard,P.,&Moles,M.2000,ApJ,542,673 Butcher,H.,&Sharples,R.M.1997, ApJS,110,213
Finn, R. A., Balogh, M. L., Zaritsky, D., Miller,C. J., & Nichol, Smith, G. P., Treu, T., Ellis,R. S., Moran, S. M., & Dressler,A.
R.C.2008,ApJ,679,279 2005,ApJ,620,78
Finn,R.A.,Zaritsky,D.,&McCarthy,Jr.,D.W.2004,ApJ,604, Toomre,A.,&Toomre,J.1972,ApJ,178,623
141 Tran,K.-V.H.,etal.2009,ApJ,705,809
Finn,R.A.,etal.2009,inprep Wechsler,R.H.,Bullock,J.S.,Primack,J.R.,Kravtsov,A.V.,&
Gehrels,N.1986,ApJ,303,336 Dekel,A.2002,ApJ,568,52
Gunn,J.E.,&Gott, J.R.I.1972,ApJ,176,1 White,S.D.M.,etal.2005,A&A,444,365
Halliday,C.,etal.2004,A&A,427,397 Wilman,D.J.,Oemler,A.,Mulchaey,J.S.,McGee,S.L.,Balogh,
Helsdon,S.F.,&Ponman,T.J.2003,MNRAS,339,L29 M.L.,&Bower,R.G.2009,ApJ,692,298
Hinz,J.L.,Rieke,G.H.,&Caldwell,N.2003,AJ,126,2622 Yang, Y., Tremonti, C. A., Zabludoff, A. I., & Zaritsky, D. 2006,
Holden,B.P.,etal.2009,ApJ,693,617 ApJ,646,L33
Icke,V.1985, A&A,144,115 Yang, Y., Zabludoff, A. I., Zaritsky, D., Lauer, T. R., & Mihos,
Kannappan, S.J.,Guie,J.M.,&Baker,A.J.2009,AJ,138,579 J.C.2004,ApJ,607,258
Kautsch, S. J., Gonzalez, A. H., Soto, C. A., Tran, K.-V. H., Yang,Y.,Zabludoff,A.I.,Zaritsky,D.,&Mihos,J.C.2008,ApJ,
Zaritsky,D.,&Moustakas,J.2008,ApJ,688,L5 688,945
Kawata,D.,&Mulchaey,J.S.2008,ApJ,672,L103 Zabludoff,A.I.andFranx,M.1993,ApJ,106,1314
Kendall,M.,&Stuart,A.1977,Theadvancedtheoryofstatistics. Zabludoff,A.I.,&Mulchaey,J.S.1998,ApJ,498,L5+
Vol.1:Distributiontheory
Larson, R. B., Tinsley, B. M., & Caldwell, C. N. 1980, ApJ, 237,
692
10 JUST ET AL.
Fig. 1.— Rest-frame velocity histograms. The bin size is one-third the velocity dispersion, and the velocities plotted are those that
remain after the various cuts/iterations in the calculation of σ. Overplotted on each panel is a Gaussian normalized to the area of the
histogram.
