Astronomy & Astrophysics The ESO Nearby Abell Cluster Survey

18 Pages·2002·0.43 MB·English

A&A 387, 8{25 (2002) Astronomy
DOI: 10.1051/0004-6361:20020340 &
(cid:13)c ESO 2002 Astrophysics
The ESO Nearby Abell Cluster Survey?;??
XI. Segregation of cluster galaxies and subclustering
A. Biviano1, P. Katgert2, T. Thomas2, and C. Adami3
1 INAF,Osservatorio Astronomico di Trieste, Italy
2 Sterrewacht Leiden, The Netherlands
3 Laboratoire d’AstrophysiquedeMarseille, France
Received 22 October 2001 / Accepted 30 January 2002
Abstract.Westudyluminosityandmorphologysegregationofclustergalaxiesinanensembleclusterbuiltfrom59
rich,nearbygalaxyclustersobservedintheESONearbyClusterSurvey(ENACS).Theensembleclustercontains
3056 member galaxies with positions, velocities and magnitudes; 96% of these also have galaxy types. From
positionsandvelocitiesweidentifygalaxieswithinsubstructures,viz.asmembersofgroupsthataresigni(cid:12)cantly
colder than their parent cluster, or whose average velocity di(cid:11)ers signi(cid:12)cantly from themean.
We compare distributions of projected clustercentric distance R and relative line-of-sight velocity v, of galaxy
subsamples drawn from the ensemble cluster, to study various kinds of segregation, the signi(cid:12)cance of which is
obtained froma 2-dimensional Kolmogorov-Smirnov test. We(cid:12)ndthat luminosity segregation isevident onlyfor
theellipticals that are outside (i.e. not in) substructuresand which are brighter than M =−22:0(cid:6)0:1. This is
R
mainly duetothe brightest cluster membersat rest at thecentre of the cluster potential.
We con(cid:12)rm the well-known segregation of early- and late-type galaxies. For the galaxies with M > −22:0 of
R
all types (E, S0, S and emission-line galaxies, or ELG, for short), we (cid:12)nd that those within substructures have
(R;v)-distributionsthat di(cid:11)erfromthoseofthegalaxies thatarenot insubstructures.Theearly andlatespirals
(Sa{Sb and Sbc{Ir respectively) that are not in substructures also appear to have di(cid:11)erent (R;v)-distributions.
Forthesereasonswehavestudiedthesegregation propertiesof10galaxysubsamples:viz.E,S0,S ,S andELG,
e l
both within and outside substructures.
Amongthe5samplesofgalaxiesthatarenotinsubstructures, atleast3ensemblescanandmustbedistinguished;
these are: [E+S0], S , and [S+ELG]. The [E+S0] ensemble is most centrally concentrated and has a fairly low
e l
velocitydispersionthathardlyvarieswithradius.The[S+ELG]ensembleisleastconcentratedandhasthehighest
l
velocity dispersion, which increases signi(cid:12)cantly towards the centre. The class of the S galaxies is intermediate
e
tothetwoensembles.Itsvelocitydispersionisverysimilartothatofthe[E+S0]galaxiesintheouterregionsbut
increases towards thecentre.
Thegalaxieswithinsubstructures donotallhaveidentical(R;v)-distributions;weneedtodistinguishatleasttwo
ensembles,becausetheS0and[S+ELG]galaxies havedi(cid:11)erentdistributionsinR aswellasin v.The[S+ELG]
l l
galaxies are less centrally concentrated and, in the inner region, their velocity dispersion is higher than that of
theS0 galaxies. Ourdata allow theother 3 galaxy classes to be combined with these two classes in 4 ways.
We discuss briefly how our data provide observational constraints for several processes inside clusters, like the
destruction of substructures,the destruction of late spirals and the transformation of early spirals intoS0s.
Key words.galaxies: clusters: general { galaxies: elliptical and lenticular, cD { galaxies: evolution {
galaxies: kinematics and dynamics { cosmology: observations
1. Introduction Dressler(1980)werethe(cid:12)rsttoquantifythesedi(cid:11)erences.
Dressler (1980) showed that the di(cid:11)erent distributions
It has been known for a long time that in clusters,
arise mainly from the so-called morphology-density rela-
galaxies of di(cid:11)erent classes have di(cid:11)erent projected dis-
tion (MDR): i.e., the relative fractions of ellipticals, S0s
tributions. Oemler (1974), Melnick & Sargent (1977) and
and spirals correlate very well with local surface density.
Hence, the composition of the galaxy population changes
Send o(cid:11)print requests to: A.Biviano,
with distance from the cluster centre.
e-mail: [email protected]
? Based onobservationscollected at theEuropeanSouthern Postman & Geller (1984) derived an MDR over 6
Observatory (La Silla, Chile). decades of local space density from the CfA redshift
?? http://www.astrsp-mrs.fr/www/enacs.html survey, and in the Pisces-Perseus supercluster { and in
A.Biviano et al.: ENACS XI: Segregation and subclustering. XI. 9
particular in its long (cid:12)lament { Giovanelli et al. (1986) explained the observed degrees of luminosity segregation
found a clear MDR. In this supercluster, even early and by their merging models.
late spirals have di(cid:11)erent distributions, and this was also Adami et al. (1998a) studied a sample of about 2000
found for spirals in groups of galaxies (Giuricin et al. galaxies in 40 nearby Abell clusters and con(cid:12)rmed that
1988). The MDR was also studied in several individual the overall velocity dispersion depends on galaxy type,
nearbyclusters(e.g.Andreon1994,1996;Caon&Einasto and increases along the Hubble sequence. The velocity
1995) and in general redshift surveys (e.g. Santiago & dispersion pro(cid:12)les for the various galaxy types indicate
Strauss 1992). that the spirals may not yet be fully virialized, and may
In spite of the wealth of observational data, it is still be mostly on radial, infalling orbits. The spirals may
still not totally clear how the MDR arises. In clusters, thus haveproperties similar to the galaxieswith emission
galaxy encounters must play a ro^le, so that gas-rich disk lines (ELG), which were studied in A576 by Mohr et al.
galaxies cannot survive in the dense cores of clusters. (1996), and by Biviano et al. (1997, hereafter Paper III)
Contraryto Dressler (1980), Whitmore& Gilmore(1991) in the clusters observedin the ESO NearbyAbell Cluster
andWhitmoreetal.(1993)foundthatmorphologicalfrac- Survey (ENACS). Biviano et al. concluded that the ELG
tion correlates as tightly with clustercentric distance as probably havea signi(cid:12)cantvelocityanisotropy.De Theije
with projected density. An explanation for the MDR in a &Katgert(1999,hereafterPaperVI)distinguishedearly-
colddarkmatter-dominateduniversewasgivenbyEvrard and late-type galaxies in the ENACS from their spectra,
et al. (1990). andconcludedthattheevidenceforradialorbitswasonly
signi(cid:12)cant for the ELG.
The study of the MDR was extended towards higher
Recently,Thomas(2002,hereafterPaperVIII)derived
redshifts,e.g.byDressleretal.(1997),Couchetal.(1998)
morphologiesforcloseto2300ENACSgalaxiesfromCCD
and Fasanoetal.(2000), andwaslinkedto themoregen-
imaging.Byaddingmorphologiesfromtheliterature,and
eral question of the evolution of galaxies in environments
spectral types from the ENACS spectra, this provides es-
of di(cid:11)erent densities (e.g. by Menanteau et al. 1999).
sentially complete type information for the galaxies in a
Dressler et al. (1997) found that the regular, centrally
sampleof59ENACSclusters.Thisdatasetallowsavastly
concentrated clusters at a redshift of about 0.5 show a
improvedanalysisofthedistributionandkinematicsofthe
strongMDR,asdothelow-redshiftclusters.However,the
less concentrated and irregular clusters at z (cid:25) 0.5 do not variousclasses of galaxiesin clusters,which we presentin
thispaper.Inasubsequentpaper(Katgertetal.2002)we
showa clearMDR, unliketheir low-redshiftcounterparts.
derive the mass pro(cid:12)le of the ENACS clusters.
Dressler et al. also noted that the fraction of S0s ap-
In Sect. 2 we summarize the data that we used. In
pears to decrease quite strongly with increasing redshift
Sect.3 wediscussthe method by whichwestudythevar-
(by as much as a factor of 3 from z = 0 to z (cid:25) 0:5),
ious types of segregation. In Sect. 4 we discuss the e(cid:11)ect
and Fasano et al. (2000) studied this e(cid:11)ect in clusters
of substructure and in Sects. 5 and 6 we discuss the evi-
at redshifts between 0.1 and 0.25. The reality of this de-
dence for luminosity and morphology segregation,as well
creasewasquestionedbyAndreon(1998)whoarguedthat
as the minimum number of galaxy ensembles that must
it is not trivial to establish a reliable elliptical/S0-ratio.
be distinguished. In Sect. 7 we discuss the nature of the
Also, the elliptical/S0-ratio may not be very meaningful
morphological segregations and in Sect. 8 we discuss the
(even if it can be established accurately) because the dif-
implications of our results for ideas about cluster galaxy
ferencesbetweenellipticalsandS0smaynotbemajor(e.g.
evolution. In Sect. 9 we present a summary and the main
J(cid:28)rgensen & Franx 1994).
conclusions.
Morphological segregation in position is often accom-
paniedbymorphologicalsegregationinvelocityspace,i.e.
2. The data
galaxies of di(cid:11)erent types have di(cid:11)erent velocity disper-
sions, or velocity dispersion pro(cid:12)les (e.g. Tammann 1972; We use data from the ENACS (see Katgert et al. 1996 {
Moss & Dickens 1977; Sodr(cid:19)e et al. 1989; Biviano et al. Paper I { and Katgert et al. 1998 { Paper V). We have
1992). The e(cid:11)ect is sometimes reported as a correlation imposedaredshiftlimitof z <0:1andweappliedalower
betweenkinematicsandcolours(e.g.Colless&Dunn1996; limitof20to thenumber ofmember galaxies;this de(cid:12)nes
Carlberg et al. 1997a). a sample of 67 clusters that is essentially volume-limited
Luminosity segregation was detected by Rood & (Mazure et al. 1996 { Paper II). Clusters were de(cid:12)ned in
Turnrose(1968),Capelatoetal.(1981),Yepesetal.(1991) redshift space, from the distribution of the line-of-sight
and Kashikawa et al. (1998). Luminosity segregation was velocities.Weusedadensity-dependentgap(Adamietal.
detected both as a segregation in clustercentric distance, 1998b) rather than a (cid:12)xed gap (as was used in Paper I)
and as a kinematical segregation, viz. the most luminous to accommodate di(cid:11)erent total numbers of galaxies. The
galaxies have the smallest velocity dispersion (see e.g. membership of the ENACS clusters with at least 20 red-
Rood et al. 1972). Yet, kinematical segregation appears shifts hardly changes when we use a variable instead of a
to occur mostly (Biviano et al. 1992), if not exclusively (cid:12)xed gap.
(Stein 1997), for ellipticals, and much less { if at all { for Interlopers (non-members, seen in projection onto
the other galaxy types. Fusco-Femiano & Menci (1998) the cluster) were eliminated with the interloper removal
10 A.Biviano et al.: ENACS XI: Segregation and subclustering. XI.
procedure devised by den Hartog & Katgert (1996). For
thesystemswithatleast45galaxieswithredshifts(N (cid:21)
z
45)wecalculatedan\interim"masspro(cid:12)le.This predicts
the maximum line-of-sight velocity at the projected posi-
tionofeachgalaxyfromwhichwedetermineifthegalaxy
canbewithintheturn-aroundradius.Thisprocedurewas
repeated until it converged. For clusters with N <(cid:24) 45,
z
such a procedure generally does not work; therefore we
usedforthemtheseparationbetweenmembers andinter-
lopers as de(cid:12)ned in a statistical manner by the N (cid:21) 45
z
clusters (for details, see Katgert et al. 2002).
Our magnitudes are R-band, and the absolute magni-
tudes, M , were derived for H = 100 km s−1 Mpc−1,
R 0
with K-corrections according to Sandage (1973) and cor-
rections for galactic absorption according to Burstein &
Heiles(1982).Forthegalaxiesinoursample,thesecorrec-
tionsarequitesmall,viz.(cid:24)0.1mag.Informationongalaxy Fig.1.Thenumberofclusterscontributingtotheanalysis,as
a function of R=r , for the various galaxy classes. The full-
typecomesfromvarioussources:eitherfromaCCDimage 200
drawn line refers to all 59 clusters, the dashed line to the 17
(mostly from Paper VIII), or from the ENACS spectrum
clusters with only spectral galaxy types (i.e. no pure E-type
(using a Principal Component Analysis in combination
nor S ), and the dot-dashed line to the 42 clusters with CCD
with anArti(cid:12)cial Neural Network,see PaperVI). A com- e
imaging (i.e. with all galaxy types).
parison of the various type estimates and a discussion of
their robustness is given in Paper VIII. Several hundred
galaxies have one or more emission lines in their ENACS
spectrum (we refer to those as ELG, see Paper III). Most
de(cid:12)nes a sample of 59 ENACS clusters with z < 0:1; all
of these ELG have narrow lines due to warm gas.
clusters have 20 or more ENACS member galaxies, for
The galaxies with type information were assigned to
at least 80% of which a galaxy type is known. The total
the following classes: ellipticals (E), S0, spirals (S), and
numberofgalaxiesinthe59clustersis3056,andfor2948
the intermediate classes, E/S0, and S0/S. Whenever pos-
(96%) of those a galaxy type is known. In the sample of
sible,wealsodistinguishedbetweenearly(S )andlate(S)
e l
59 clusters, there are 429 ELG, i.e. galaxies with one or
spirals,i.e.,spiralswithtypeearlierthanorasearlyasSb,
more emission lines in the spectrum. Information about
and later than Sb, respectively. The classes E, S0, S and
the 59 clusters is given in Appendix A.
ELG are \pure" and \exclusive": i.e. the E, S0 and S do
not containELG;as a matter offact weignoredthe ELG Combination of data in clusters of various sizes and
in Es and S0s. The class E/S0 is not used separately, but masses into an ensemble cluster requires that projected
it is included in the class of early-type galaxies, together distances and relative velocities (or rather, their line-
with E and S0, when these two classes are linked in one of-sight components) are properly scaled. Projected dis-
sample (see Sect. 6). The S0/S class was never used. In tances R were scaled with r200, the radius within which
clusters where galaxy types could only be estimated from theaveragedensity is 200times the criticaldensity ofthe
spectra, the pure E class does not occur and the \earli- universe and which is very close to the virial radius. We
est" galaxy class is E/S0 (see also Paper VIII). Similarly, assumed, like Carlberg et al. (1997b), that M(< r) / r,
earlyandlatespiralgalaxiescanonlybeclassi(cid:12)edonCCD sothatr200 followsfromtheglobalvalueofthedispersion
images. However, late spirals can be recognized from the of the lipne-of-sight component of the velocities, (cid:27)p; viz.
spectrum alone (see Paper VIII). r200 (cid:25) 3(cid:27)p=(10H(z)), with H(z) the Hubble parame-
We considered including galaxies and clusters with ter at redshift z. The line-of-sight components (v−v) of
non-ENACS data, but segregation can only be studied the relativevelocities of the galaxies were scaled with the
usefully for data with a su(cid:14)ciently uniform completeness global velocity dispersions (cid:27)p of the parent cluster. Note
limit in apparent magnitude. For literature data this re- that (cid:27)p was calculated for all galaxies together, irrespec-
quirement often is not met, so literature data were not tive of type, so that the relative velocities of the di(cid:11)erent
used to enlarge the ENACS galaxy samples; we only used types of galaxies are all normalized by the same overall
galaxy types from the literature if there was no ENACS velocity dispersion. Forthe clustersinTableA.1the aver-
galaxy type. Redshifts from the literature were only used age value of (cid:27)p is about 700 km s−1, so that the average
in the identi(cid:12)cation of interlopers in the ENACS galaxy value of r200 is about 1:2 h−1 Mpc.
samples. In Fig. 1 we show the number of clusters contribut-
The analysis of the distribution and kinematics of the inginthevariousR=r -intervals(solidline).Alsoshown
200
various galaxy classes can only be done for clusters with arethenumbersfortheclusterswithonlyspectralgalaxy
galaxytypesforasu(cid:14)cientlyhighfractionofthegalaxies, types(i.e.thatdonothaveE-andS -types),andclusters
e
and we required this fraction to be at least 0.80. This with CCD-imaging (that have all types).
A.Biviano et al.: ENACS XI: Segregation and subclustering. XI. 11
3. Detecting segregation sampled out to di(cid:11)erent apertures. Selection of di(cid:11)erent
morphological types sometimes results in the selection of
Segregation, i.e., the fact that the various galaxy pop-
a subsample of the original 59 clusters (Fig. 1). Di(cid:11)erent
ulations have di(cid:11)erent phase-space distributions, may
cluster subsamples have di(cid:11)erent degrees of incomplete-
show up either in the projected distribution, or in the
ness at a given radius. We estimate this radial bias by
kinematics, or in both. In order to use our data opti-
using an approach similar to that adopted by Merri(cid:12)eld
mally, we searchedfor the various types of segregationby
& Kent (1989).
comparing (R;v)-distributions in an unbinned way, using
Whencomparingtwo(R;v)-distributions onemustei-
the combined evidence from projected positions and rel-
ther ensure that these biases are identical, or take the
ative velocities. Luminosity and morphology segregation
di(cid:11)erences in the biases into account before making the
are generally presented through compression of (R;v)-
comparison. Since galaxy subsamples can have di(cid:11)erent
distributions, i.e., through projection onto the R- or the
radialdistributions, the fact that clusters have been sam-
v-axis. Luminosity and morphology segregation are thus
pled out to di(cid:11)erent radii may be a complicating factor.
oftenreferredtoas\kinematicalsegregation"withrespect
ForallKS2Dcomparisonspresentedinthispaper,thera-
to magnitude or morphology. However, it is important to
dialbiasesinthetwo(R;v)-distributionswerefoundtobe
consider radial and kinematical segregation together, be-
either identical, or so similar that a straight comparison
cause they may not be independent.
was justi(cid:12)ed.
We use an ensemble cluster constructed from the 59 Thenumberofclustersthatcontributetotheensemble
ENACS clusters listed in Table A.1. From this ensemble clusters of the various galaxy types, at various projected
cluster weselect variousgalaxysubsamples.Combination radial distances, is shown in Fig. 1. In order to ensure
of the 59 clusters is necessary to have the best \sig- that an ensemble cluster is su(cid:14)ciently representative(i.e.
nal/noise". However, the implicit assumption is that the is built from a su(cid:14)cient number of clusters) we have al-
distributions of the various galaxy type and magnitude ways compared (R;v)-distributions over the radial range
subsamples are su(cid:14)ciently similar in the individual clus- 0 (cid:20) R=r (cid:20) 1:5. This means that the ensemble cluster
200
ters, or in di(cid:11)erent classes of clusters, so that their com- includes at least 13 out of 59 clusters (or 9 out of 45, for
bination is meaningful. This is not guaranteed, as cosmic E and for S ).
e
variance is not negligible. It is therefore possible, if not The actual comparisonof two (R;v)-distributions was
likely, that no real cluster is described satisfactorily by donewiththe2-DversionoftheKolmogorov-Smirnovtest
the ensemble cluster; however, the ensemble cluster gives (KS2D, for short), as described by Peacock (1983) and
the best picture that we have of an average rich nearby Fasano & Franceschini (1987). The Kolmogorov-Smirnov
cluster. (KS)testisrelativelyconservative:ifitindicatesthattwo
For the estimate of the projected clustercentric dis- distributions have a high probability of not being drawn
tanceR,itisimportantthatthedeterminationofthecen- from the same parent population, other tests (e.g. Rank-
treoftheparentclusterisasunbiasedaspossible.Wehave Sum tests or Sign Tests) indicate the same. However, the
takenspecialcarethatthecentresofallclustersaredeter- KS test does not always support di(cid:11)erences indicated by
minedwithsimilarmethods,andwithsu(cid:14)cientaccuracy. other tests. Therefore, even if the number of galaxies in
For the calculation of the centre position we followed the one or both of the samples is relatively small, a small
procedure described in Paper III: in order of decreasing probability for the samples to be drawn from the same
preference we used the X-raycentre, the brightest cluster population in general is trustworthy. However, if the dif-
member in the core of the cluster, the peak in the galaxy ference is not signi(cid:12)cant, this does not prove that the two
surface density (if necessary luminosity-weighted) or the samples are drawn from the same parent population, be-
biweight average(e.g., Beers et al. 1990) of all galaxy po- causerealdi(cid:11)erencescanbemadeundetectablebylimited
sitions to derivethe centralposition.Theestimatedaccu- statistics.
racy that can be obtained in this way is 50{60 kpc (see We checkedthe performanceof the KS2Dtest by ran-
also Adami et al. 1998c). The positions of the adopted domlyassigninghalf of the galaxiesin eachcluster to one
cluster centres are given in Table A.1 in Appendix A. of two ensemble clusters, each comprising half the total
The advantage of comparing (R;v)-distributions is number of galaxies in the ensemble cluster built from the
that structure in the (R;v)-distribution is not diluted in 59 clusters. According to the KS2D-test the probability
projection onto either the R- or v-axis. However, as a re- that the two subsets aredrawnfrom the same parentdis-
sult of the generally non-circular shapes of the apertures tributionis83%.Thisisfullyconsistentwiththefactthat
in which the ENACS redshift surveys of the clusters were they were drawn from the same parent distribution and
done(see,e.g.,Fig.10inPaperI),thedistributioninpro- thattherearenodi(cid:11)erencesbetweenthemotherthansta-
jected radial distance is always biased. We estimate this tistical fluctuations.
radial bias by assuming circular symmetry for the clus-
ter galaxy distributions, knowing the positions (and size)
4. The e(cid:11)ect of substructure
of the Optopus plates that were used to sample the clus-
ters(seeFig.10inPaperI).Another sourceofradialbias The combination of data for many clusters into an en-
arisesfromthefactthatwestackclusterswhichhavebeen semble cluster is unavoidable, especially because galaxy
12 A.Biviano et al.: ENACS XI: Segregation and subclustering. XI.
subsamples in individual clusters are too small for seg- Table1.Theevidenceforsubstructureintheclusterswithat
regation studies. However, combining clusters in an en- least 45 members with ENACS redshifts.
semble cluster inevitably reduces the relative amplitude
ACO v ENACS P (cid:27)
of substructure. Because substructure may play an im- 3K (cid:1) p
kms−1 z type kms−1
portantro^leintheformationandevolutionofclusters,we
119 12997 102 87 0.917 720
have estimated the e(cid:11)ect of substructure on the kinemat-
168 13201 76 71 0.305 518
ics and distribution of the various galaxy classes, in two
514 21374 82 74 0.684 875
ways.First,wecomparedclusterswithandwithoutglobal 548 12400 108 108 0.005 710
substructure, and secondly we comparedgalaxiesthat re- 548 12638 120 116 0.000 824
side in and outside local substructures in their respective 978 16648 56 52 0.081 497
clusters. 2734 18217 77 77 0.010 579
2819 22285 49 44 0.778 409
3094 20027 66 64 0.000 654
4.1. Clusters with and without substructure 3112 22417 67 60 0.211 954
3122 19171 89 88 0.000 782
Because substructure may take di(cid:11)erent forms, di(cid:11)erent
3128 17931 152 152 0.000 765
(cid:12)lters are required (and have been devised) for its detec-
3158 17698 105 102 0.771 1006
tion. A general problem for the detection is that obser-
3223 17970 66 65 0.204 597
vations only provide the 2+1-D projected version of the 3341 11364 63 63 0.569 561
3+3-D phase-space, which reduces the detectability. We 3354 17589 56 56 0.008 367
have used a slightly modi(cid:12)ed version of the test devised 3558 14571 73 73 0.063 1035
by Dressler & Schectman (1988). This test is sensitive to 3562 14633 105 105 0.000 903
spatiallycompact subsystems that either haveanaverage 3651 17863 78 78 0.019 662
velocity that di(cid:11)ers from the cluster mean, or have a ve- 3667 16620 103 102 0.151 1037
3806 22825 84 83 0.600 808
locity dispersionthat di(cid:11)ers fromthe globalone,or both.
3822 22606 84 68 0.079 971
For each galaxy we selected the n neighbours that
loc 3825 22373 59 57 0.106 699
apre closest in projection, where nloc was taken to be
N (see,e.g.,Bird1994)withN thetotalnumber
mem mem
of cluster members with redshifts. For these n neigh-
loc
theprobabilityP thattheobservedvalueisduetonoise,
bours we calculated the averagevelocity, v and the ve- (cid:1)
loc
and thus not indicative of real substructure. Thus, a low
locity dispersion (cid:27) . From these parameters, we calcu-
loc
valueofP indicatesahighprobabilityofsigni(cid:12)cantsub-
lated for each galaxy a quantity (cid:14), designed to indicate (cid:1)
structure.
groups (of n members) that are\colder"than the clus-
loc There are 9 clusters with P (cid:20) 0:05, i.e. with signif-
ter and/or have an average velocity that di(cid:11)ers from the (cid:1)
icant global substructure, and these contain 851 galax-
global cluster mean.
ies. The other 14 clusters, containing 1069 galaxies, have
The parameter (cid:14) was calculated as follows:
v P > 0:05, i.e. are without signi(cid:12)cant global substruc-
u (cid:1)
(cid:14) = 1 ut nloc(cid:14)v2 + h q (cid:14)(cid:27)2 i (1) ture. The average number of galaxies in the substructure
(cid:27)p(R) [tnloc−1]2 1− (n −1)=(cid:31)+ 2 clusters is higher than it is in the non-substructure clus-
loc nloc−1 ters (95 against 76). This is a reminder that some of the
with(cid:14) =jv −v j,and(cid:14) =max((cid:27) −(cid:27) ;0),where non-substructureclustersmayhavesubstructurethatwas
v loc glob (cid:27) p loc
the Student-t and (cid:31)2 distributions are used to calculate notdetecteddueto limitedstatistics.Thereisnorelation
theuncertaintyinthevelocityandvelocity-dispersiondif- between the presence of substructure and global velocity
ferences, respectively. To suppress noise, we (cid:12)nally calcu- dispersion:the9substructureclustershaveanaverageve-
lated(cid:14) foreachgalaxyastheaverageofthe(cid:14)-valuesofits locity dispersion of 694(cid:6)52 km s−1, for the 14 clusters
n −1neighbours.Thelargerthevalueof(cid:14)thelargerthe without substructure this is 763(cid:6)59 km s−1.
loc
probabilitythatthegalaxy(cid:12)ndsitselfinamovingand/or A KS2D comparison of the (R;v)-distributions of
cold subgroup within its cluster. the total galaxy populations in substructure and non-
As shown in Paper III, at least 40{50 galaxies are substructure clusters shows that the two samples have a
neededforadecisionaboutwhetheraclustercontainssig- probability of <0.1% to have been drawn from the same
ni(cid:12)cant substructure or not. In Table 1 we list the results parent sample. This forces us to analyze the segregation
forthe23clusterswithatleast45members.Foreachclus- properties of galaxies within and outside substructures
teraglobal(cid:1)parameterwascalculatedasthesumofthe separately.
individual(cid:14)’s ofallgalaxies.Theobservedvalueof(cid:1)was
compared with the 1000 (cid:1) values obtained in 1000 ran-
4.2. Galaxies in and outside substructures
dom azimuthal scramblings of the galaxypositions of the
same cluster. In the scramblings the incomplete azimuth Insteadofsummingallindividualvaluesof(cid:14)togetamea-
coverage was taken into account. The fraction of scram- sure of the amount of substructure in the cluster as a
blings with a value of (cid:1) larger than the observed one is whole (as was done in Sect. 4.1), one can also use the
A.Biviano et al.: ENACS XI: Segregation and subclustering. XI. 13
Thelowerlimitin(cid:14) fortheselectionofgalaxieswithin
substructures is less obvious. Using (cid:14) > 1:8, the com-
pleteness and reliability of the substructure-sample (of
752galaxies)arebothabout65%.I.e.,oneinthreeofthe
galaxies with (cid:14) > 1:8 is not in substructures. Increasing
thelowerlimitin (cid:14) toreducethecontaminationbygalax-
iesoutsidesubstructuresalsoreducesthenumberofgalax-
ies in substructures available for the tests. For compar-
isons of (R;v)-distributions involving samples of galaxies
in substructures, we therefore always de(cid:12)ned 4 parallel
samples, with (cid:14) > 1.8, 2.0, 2.2 and 2.4 to vary the bal-
ance between contamination and statistical weight. If the
results for those 4 samples areidentical, contamination is
notimportant;otherwisethe4resultsmustbeinterpreted.
We made a KS2D comparison of the (R;v)-
Fig.2. The distribution of the substructure parameter (cid:14) for distributions of the total galaxy populations within and
the 3056 galaxies in the 59 clusters. The solid line represents outside substructures, using (cid:14) > 1:8 for the substructure
the observations, the dashed line the distribution for the az- sample. The probability that the (R;v)-distributions of
imuthallyscrambledclusters(normalizedtotheobservednum- the two samples are drawn from the same parent distri-
ber with (cid:14) < 1), and the dash-dotted line gives the di(cid:11)erence bution is very small, viz. again<0.1%.
between thetwo.
4.3. Characteristics of the substructures
individual (cid:14)-values to select galaxies in signi(cid:12)cant local Thede(cid:12)nitionofsubstructuresthatweusedwasdesigned
substructures in their cluster. In other words: whereas in toselectcoldand/ormovinggroups.However,becausethe
Sect. 4.1 all galaxies in a cluster were made to follow the membership of a group is (cid:12)xed to be n , the properties
loc
classi(cid:12)cation of their cluster, one may also consider all ofthegroupsarenotnecessarilyconstant.E.g.,the\size"
galaxies in signi(cid:12)cant local substructures, independent of of a group depends on n and on the surface density of
loc
the classi(cid:12)cation of the parent cluster. Even in clusters galaxies, which in turn depends on the redshift sampling
withoutsigni(cid:12)cantglobalsubstructure,somegalaxiesmay oftheclusterN ,sothatthegroupsindi(cid:11)erentclusters
mem
be inlocalsubstructures.Similarly,inclusterswithsignif- may havedi(cid:11)erent sizes. This disadpvantageis outweighed
icant substructure, not all galaxies are in local substruc- by the fact that by setting n = N one maximizes
loc mem
tures. the sensitivity to signi(cid:12)cant substructures while reducing
In Fig. 2 we show the distribution of (cid:14) for all 3056 thesensitivitytoPoissonnoise(e.g.Silverman1986).The
galaxies in the 59 clusters in our sample. In order to use latter is important, asin our clusters nloc is never as high
(cid:14) to select galaxies in and outside cold and/or moving asthe optimumvalue of(cid:25)25,derivedbyKnebe & Mu¨ller
groups, we must determine the value of (cid:14) that optimally (2000) from an analysis of simulated clusters.
separates them. To that end, we azimuthally scrambled However, even within a cluster the \size" of the se-
the galaxy distributions, taking into account the incom- lected groups is not constant, but varies with distance
plete azimuth coverage due to the generally non-circular from the centre because the surface density increases no-
shapes of the apertures within which the ENACS spec- ticeablytowardsthecentre.Thise(cid:11)ectisclearlyvisiblein
troscopy was done. The resulting distribution of (cid:14) (the ourdata: theharmonicmeanradius of then −1 neigh-
loc
dashed line in Fig. 2) was normalized to produce the ob- bours increases with projected distance from the cluster
served numbers of galaxies with (cid:14) <1:0. centre.Onecouldavoidthisbiasbychoosinga(cid:12)xedphys-
This normalization was chosen because for (cid:14) <1:0 no icalscalefortheselectedsubclusters.However,ifthescale
signi(cid:12)cant contribution of galaxies in substructures is ex- ischosenlargeenoughforareasonablenumberofgalaxies
pected. This is borne out by the fact that the observed to be selected at large radii, substructures in the central
and scrambled (cid:14)-distribution have essentially the same region would be averagedout.
shape for (cid:14) < 1:0. The di(cid:11)erence between the observed Using the (cid:14)-values of the individual galaxies, we have
andscrambled(cid:14)-distributionsgivesthedistributionofthe attempted to identify \subclusters" as follows. First, we
galaxies that presumably are in substructures (the dash- selected all galaxies in each cluster with a value of (cid:14) >
dotted line in Fig. 2). The galaxies that are not in sub- (cid:14) . Then we calculated the harmonic mean projected
lim
structurescanbeselectedquitesatisfactorilybyrequiring radius and the velocity dispersion of the group of n
loc
(cid:14) < 1:8. Both the completeness and reliability of that nearest neighbours around each of these galaxies. These
sample of 2304 galaxies are very close to 90%. We will groups were subsequently merged if both their projected
therefore de(cid:12)ne all (sub-)samples of galaxies not in sub- distance was less than the sum of their harmonic mean
structures with an upper limit in (cid:14) of 1.8. radii,andthedi(cid:11)erenceoftheiraveragevelocitieswasless
14 A.Biviano et al.: ENACS XI: Segregation and subclustering. XI.
than the mean of their velocitydispersions. The resulting
\subclusters" consist of all galaxies with (cid:14) > (cid:14) in the
lim
groups from which they were built.
For (cid:14) = 2:0 we (cid:12)nd 62 subclusters in the 59 clus-
lim
ters, i.e. on averageone per cluster. The mean number of
galaxies with (cid:14) > 2:0 in a subcluster is 8.6, and individ-
ual numbers go up to about 60 (in the very rich cluster
A3128). Note that 16 clusters do not have a subcluster,
while some subclusters are found in clusters that do not
showsigni(cid:12)cantevidenceforsubstructurewiththe testof
Dressler&Schectman.Notealsothat,whiletheharmonic
mean radius of the selected \subclusters" is observed to
increasewithclustercentricdistance,theirvelocitydisper-
sionstaysremarkablyconstantat(cid:27) (cid:24)400−500kms−1.
loc
The(cid:12)lterthatweusedwillnothavedetectedallgalax-
ies that belong to substructures. Yet, as shown by our
azimuthal scramblings, a large fraction of those that are
selected, do belong to subclusters (where this fractionob-
viously increases with increasing lower limit in (cid:14)). The
galaxies selected to be in subclusters do not form a com-
pletesample,butwetreatthemasadistinctclass,ifonly
because their (R;v)-distribution is likely to be influenced
by the fact that they are dynamically linked in subclus-
ters.Thisisactuallycon(cid:12)rmedbytheresultsoftheKS2D
tests described in Sect. 4.2.
From histograms like those in Fig. 2 in several radial
intervals, we conclude that the (cid:14)-distribution of galax-
ies in substructures, correctedfor accidental substructure
through azimuthal scrambling (the dashed-dotted line in
Fig.2),doesnotvarysigni(cid:12)cantlywithradius.Therefore,
we felt justi(cid:12)ed to apply radius-independent (cid:14)-cuts for
galaxies inside and outside of substructures.
With our normalization of the (cid:14)-distribution of the
Fig.3. Top: the distribution of the galaxies with (cid:14) > 2:2 in
azimuthally scrambled clusters (see Fig. 2), the average
the 59 clusters, as a function of projected radius R. The solid
fraction of galaxies in substructures, corrected for acci-
line represents the observed distribution, the dashed line the
dental substructure, is 0.22. However, this fraction ap-
distribution for the azimuthally scrambled clusters, while the
pearstodependonthecentralconcentrationofthegalaxy
dash-dottedlinegivesthedi(cid:11)erencebetweenthetwo.Bottom:
distribution. We quantify the latter by a concentration the fraction of galaxies in substructures, with (cid:14) > 2:2, as a
index, calculated as the ratio of the number of galaxies function of projected radius R.
within 0.25r , and between0.25 and0.50 r . This in-
200 200
dex is not a(cid:11)ected by azimuthal incompleteness (Sect. 3)
becausethatisnegligiblewithin0.50r .The23clusters dash-dottedlineisthedi(cid:11)erencebetweenthetwo.Thelat-
200
with high concentration index contain 1527 galaxies, the terisareliableestimateoftheradialdistributionofgalax-
36 low-concentration clusters contain 1529 galaxies. The ies in substructures. In the lower panel of Fig. 3 we show
corrected fractions of galaxies in substructures in the two the fraction of galaxies in substructures, with (cid:14) > 2:2,
clustersamplesare0:17(cid:6)0:01and0:27(cid:6)0:01respectively. with respect to the total number of galaxies. The most
In other words:in clusters with low central concentration remarkablefeatureinFig.3is theverystrongandabrupt
the fraction of galaxies in substructures is signi(cid:12)cantly decrease for R (cid:20) 0:3r of the number of galaxies really
200
higher than in clusters with high central concentration. insubstructures(thedash-dottedlineintheupperpanel).
Such a correlation is also seen in numerical cosmological One might wonder to what extent the strong de-
simulations (Thomas et al. 2001). crease of the number of galaxies in substructures within
The number of galaxies in substructures also appears R (cid:25) 0:3r could, at least partially, be caused by a
200
to decrease markedly towards the centre. This is shown selection e(cid:11)ect. As we discussed above, the \size" of
in the upper panel of Fig. 3 which gives the radial distri- the substructures selected by our \(cid:12)lter" is smaller in
bution for the galaxies with (cid:14) > 2:2 (similar results are the centre than it is at the periphery. Therefore, we
obtained for other values of (cid:14) ). The solid line is the are insensitive to substructures with a large linear scale
lim
observed distribution, the distribution in the azimuthally in the centre, and to small-scale substructures in the
scrambled clusters is given by the dashed line, while the periphery. Whereas small-scale substructures might exist
A.Biviano et al.: ENACS XI: Segregation and subclustering. XI. 15
Fig.4. The relations between absolute magnitude and projected radial distance (left) and normalized relative velocity (right)
for ellipticals outside substructures.
in the periphery, it is unlikely that in the central region combinationofmanyclustersmaydilute realLSin(some
large-scale substructures, could have survived, if they ex- of) the individual clusters.
isted. As was shown by Gonza(cid:19)lez-Casado et al. (1994),
We searched for LS with many KS2D tests in which
the less massive subclusters are tidally disrupted in one
we compared two (R;v)-distributions of the same class
cluster crossing, while the more massive clumps migrate
of galaxies, like ELG or S0 etc., which di(cid:11)er only
nosub sub
towards the centre through dynamical friction and disap-
in the range of absolute magnitude. In other words: the
pearassubstructure.Therefore,webelievethatthestrong
parent sample of ELG or S0 etc. was split in ab-
nosub sub
decrease of the number of galaxies in substructures to-
solute magnitude at M , the value of which we varied.
cut
wardsthecentreisreal.Thee(cid:11)ectisreminiscentoftheal-
Forthegalaxiesinsubstructureswedidthetestsnotonly
mosttotalabsenceofbinarygalaxiesintheinnerregionof
forseveralvaluesofM butalsoforthefourlowerlimits
rich clusters (R<(cid:24)0:4h−1 Mpc), discussed by den Hartog cut
in (cid:14) that we discussed in Sect. 4.2. All those tests show
(1997).
only one robust case of LS: namely for the ellipticals out-
side substructures. As elsewhere in this paper, di(cid:11)erences
are considered real only if there is less than 5% probabil-
5. Luminosity segregation
itythat two (R;v)-distributions aredrawnfrom the same
parent distribution. For the ellipticals outside substruc-
Evidence for luminositysegregation(LS) is generallypre-
tures,weconsistentlygetprobabilities ofless than5%for
sentedasadependenceonmagnitudeofthedistributionof
all M ’s in the range −22:5 to −21:0.
intergalaxydistances,i.e.,oftheangularcorrelationfunc- cut
tionofthegalaxies.TheglobalcharacterofLSisthatthe InFig.4we showtherelationsbetweenabsolutemag-
brightest galaxies have a more central distribution than nitudeandprojectedradialdistance(left),andnormalized
the other galaxies. As an extreme example, the brightest relativevelocity(right),fortheensembleclusterofellipti-
galaxies (frequently cDs) arefound verycloseto the clus- cals with (cid:14) <1:8. The brightest ellipticals have velocities
tercentre.However,notallclustersthathavebeenstudied closeto the systemic velocity andare foundmostly in the
for LS do show evidence for it. very centres of their parent clusters. The fact that the
KS2D tests give a signal for a range of M must be due
A robust detection of LS requires a large number of cut
to \cross-talk":the segregationof the brightest ellipticals
member galaxies. In several cases, (cid:12)eld galaxies are in-
is so strong that it shows up even if fainter ellipticals are
cluded in the analysis as no redshifts are available, and
included. To estimate the value of M that optimally
thosearethenasourceofnoise(seee.g.Yepesetal.1991; cut
separates the bright galaxies that show LS from the faint
Kashikawaetal.1998).IntheENACS,(cid:12)eldgalaxieswere
ones which do not, we have proceeded as follows.
eliminated quite well, but the number of member galax-
ies in most of the clusters in Table A.1 is not su(cid:14)cient The faint ellipticals with −20:6 < M < −20:0 were
to study LS in individual clusters. The ensemble of all taken as a reference sample, presumably una(cid:11)ected by
59 clusters does have su(cid:14)cient statistical weight, but the LS. We compared the (R;v)-distribution of this reference
16 A.Biviano et al.: ENACS XI: Segregation and subclustering. XI.
Table2.ThenumberofgalaxieswithR=r (cid:20)1:5andM > As explained in Sect. 4.2 we de(cid:12)ned, for each of the 5
200 R
−22:0 in each of thesamples used. classes of galaxies within substructures, 4 samples with
lowerlimitsin(cid:14) of1.8,2.0,2.2and2.4,respectively.From
gal outside within substructures
theresultsofeachsetof4parallelsamples,wegaugedthe
type substr. (cid:14)>1:8 (cid:14)>2:0 (cid:14)>2:2 (cid:14)>2:4
\diluting" e(cid:11)ect (by contamination of galaxies with a (cid:14)-
E 200 60 41 30 15
valueabovethelimit,butnotinsubstructures)onrealdif-
S0 795 261 176 114 70
ferences (see also Fig. 2).At the sametime, weestimated
S 183 63 41 28 21
e
theinfluenceoftheoppositee(cid:11)ect,viz.aspurious\di(cid:11)er-
S 113 25 19 16 12
l
ence" due to contamination. In many cases the results of
ELG 236 88 73 56 44
E/S0 165 72 51 36 19 the 4 samples of galaxies in substructures with di(cid:11)erent
S 119 33 25 18 13 lowerlimits in (cid:14) areconsistent.There are12 comparisons
generic
whichdo notshowtotalagreementbetweenthe4 parallel
sample with that of the brighter ellipticals with [M − tests. These comparisons are discussed in more detail in
cut
0:25;M +0:25],for values of M between −23:25and Appendix B, where we give the 4 results as well as our
cut cut
−21:55. The faintest M for which the two samples are interpretation.
cut
di(cid:11)erent is −22:0, and we estimate that the uncertainty Withthe10classes,wedidall45possiblecomparisons
in this value is at least 0.1. In the following we adopt (whichrequireatotalof150KS2Dtests,duetothe4lower
M = −22:0 as the absolute magnitude above which LS limitsto(cid:14)usedfortheclassesofgalaxiesinsubstructures).
R
occurs for ellipticals. Although the KS2D tests do not Of these 45 comparisons,22 show a signi(cid:12)cant di(cid:11)erence.
show signi(cid:12)cant evidence for LS of galaxies other than Westressagainthattheverdict\signi(cid:12)cantdi(cid:11)erence"in-
the brightellipticals,wedecided to excludegalaxiesof all dicates that the probability that the two galaxy samples
types with M < −22:0 in the analysis of morphological weredrawnfrom the sameparent sample is less than 5%.
R
segregation, to avoid possible cross-talk of low-level LS It should be appreciated that comparisons for which no
into morphologicalsegregation. believable di(cid:11)erence was found are \undecided". In other
We have investigated the relation between the bright- words:inthosecasesit is not proventhatthegalaxysam-
est ellipticals and the 1st- and 2nd-ranked galaxies in the ples have identical (R;v)-distributions, because our data
clustersasfollows.Comparisonofthe(R;v)−distributions do only indicate that they are not signi(cid:12)cantly di(cid:11)erent.
of these classes shows that 1st-rankedgalaxies (or bright-
The results are as follows:
estclustergalaxies,BCGs)andthebrightestellipticalsare
not signi(cid:12)cantly di(cid:11)erent. On the contrary, those of the
2nd-ranked galaxies and the brightest ellipticals are, and { Of the 10 comparisons between classes of galaxies not
this is also the case for the 1st- and 2nd-ranked galaxies. in substructures, 6 show a di(cid:11)erence: viz. E{S,
l
It is noteworthy that of the brightest ellipticals 15% are E{ELG, S0{S , S0{S, S0{ELG and S {S.
e l e l
neither 1st-ranked nor 2nd-ranked galaxies. As a matter { Of the 10 comparisons between classes of galaxies in
of fact, the average type of 1st-ranked galaxies is inter- substructures, 2 show a di(cid:11)erence: viz. S0{S and
l
mediate between E and S0, while that of the 2nd-ranked S0{ELG
galaxies is S0. { Of the 25 \mixed" comparisons between classes of
galaxies in and outside substructures, 14 show a dif-
ference. The latter are not very surprising in view of
6. Morphology segregation
the result discussed in Sect. 4.2. Instead, the compar-
6.1. Segregation results isons for which no di(cid:11)erence was found may be more
informative in this case; these are: E {S ,
sub l;nosub
We investigated the evidence for morphological segrega-
E {ELG , S0 {S , S0 {S ,
sub nosub sub e;nosub sub l;nosub
tion by means of a large number of KS2D comparisons.
S0 {ELG , S {E , S {S0 ,
sub nosub e;sub nosub e;sub nosub
Becauseitturnedoutthatthe(R;v)-distributionsofearly
S {S , S {S , S {ELG and
e;sub e;nosub e;sub l;nosub e;sub nosub
and late spirals (S and S) that are not in substructures
e l S {S .
l;sub l;nosub
have less than 4% probability of being drawn from the
In view of the sample sizes (see Table 2) the latter 6
same parent distribution, we did not consider them to-
results may well be due to limited statistics.
gether. For the spirals within substructures there is no
evidence that we must consider S and S; however, for
e l
consistency, we also treated them separately. We there- As mentioned in Sect. 2, all projected distances R were
foremadeKS2Dcomparisonsofthe(R;v)-distributionsof expressed in r , which was derived from the velocity
200
the following 10 galaxy classes: E , S0 , S , dispersion (cid:27) on the assumption of a mass pro(cid:12)le with
nosub nosub e;nosub p
S , ELG , E , S0 , S , S and ELG . M(< r) / r. We redid all KS2D comparisons with pro-
l;nosub nosub sub sub e;sub l;sub sub
The number of galaxies with R=r200 (cid:20)1:5 in each galaxy jected distances scaled with an alternative value of rp200,
class is shown in Table 2. calculated for an assumed mass pro(cid:12)le M(< r) / r.
The5 classesofgalaxiesoutside substructureswereall The results areessentially identical, and lead to the same
de(cid:12)ned with a (cid:12)xed upper limit in(cid:14) of1.8 (seeSect. 4.2). de(cid:12)nition of galaxy ensembles, as we describe now.
A.Biviano et al.: ENACS XI: Segregation and subclustering. XI. 17
6.2. The minimum number of galaxy ensembles S ,weaddthegenericspirals(withinsubstructures)to
l;sub
the two ensembles which contain both S and S .
e;sub l;sub
BasedonthesegregationresultsdiscussedinSect.6.1,we
While we were able to identify a unique 3-ensemble
havetriedto(cid:12)ndtheminimumnumberofgalaxysamples
con(cid:12)guration for the galaxies outside substructures, our
that must be distinguished. The KS2D tests tell us which
data do not uniquely de(cid:12)ne the two ensembles into
galaxy classes cannot be combined (i.e. those that have a
which the galaxies within substructures are segregated.
probabilityof less than 5% of being drawn fromthe same
Nevertheless, the 10 galaxy classes that we started with
parent population), and which classes can in principle be
can be reduced to 5 ensembles. However, the choice be-
combined(thosewithaprobabilityofmorethan5%...),
tween the 4 possible 5-ensemble con(cid:12)gurations cannot be
but they do not tell us which ones must be combined.
made with our data.
Wewill(cid:12)rstconsidertheclassesofgalaxieswithinand
Sofar,wehaveconsideredensemblesthatconsistonly
outsidesubstructuresseparately.Thisismotivatedbythe
ofgalaxieseitherinoroutsidesubstructures.Yet,ourdata
factthattheindividualgalaxyclasses,viz.ellipticals,S0s,
allow combinations of galaxies in and outside substruc-
all spirals (including the generic spirals) and ELG in and
tures in a single ensemble. If such combinations are not
outside substructures have di(cid:11)erent (R;v)-distributions.
consideredunacceptableforphysicalreasons,onecancon-
Note that this is true for all 4 substructure samples (i.e.
struct ensemble con(cid:12)gurations with 4 ensembles instead
for di(cid:11)erent values of (cid:14) ). It is true that our data do
min of 5. The reason is that, even though galaxies of a given
not indicate that S and S have di(cid:11)erent (R;v)
e;nosub e;sub class have di(cid:11)erent (R;v)-distributions within and out-
distributions,andsimilarlyforS andS ,butthis
l;nosub l;sub side substructures (except maybe the S and S), this is
may be due to limited statistics. e l
not true in general. In other words: the KS2D tests al-
Forthegalaxiesthatarenotinsubstructures,thedata
lowe.g.S0 andELG tobecombined.However,no
sub nosub
do not allowless than threegalaxyensembles.This is be-
con(cid:12)gurations with 3 ensembles are possible. This is be-
cause the S ’s and S’s cannot be combined, while at the
e l cause, according to our data, the [E+S0] cannot be
nosub
same time neither of these canbe combinedwith the S0s.
combined with any of the 8 substructure ensembles. The
According to the other \segregation rules" in Sect. 6.1,
factthatthe[E+S0] canbecombinedwiththeS
nosub e;sub
the Es and ELG can be combined in three di(cid:11)erent ways
which are in one of the two substructure ensembles does
with S0s, S ’s and S’s. E.g., the Es can be combined
e l not a(cid:11)ect that conclusion.
with S0s as well as S ’s, while the ELG can be combined
e If one tries to construct mixed ensembles, simply by
with S ’s and S’s, as long as this does not imply com-
e l combiningthe3ensemblesoutsidesubstructureswiththe
bining them with Es or S0s. Two of these three ensemble
4 sets of 2 ensembles within substructures, one can con-
con(cid:12)gurations are not meaningful, because in them there
struct6con(cid:12)gurationsof4ensembles.However,thereisno
aretwoensembles thatarenotsigni(cid:12)cantlydi(cid:11)erent (i.e.,
goodreasonwhy one could then not \open" the 3 ensem-
whichhavemorethan5%probabilityto havebeendrawn
bles outside substructures and the two within substruc-
from the same parent population). That leaves a unique
tures, and the number of possible 4-ensemble con(cid:12)gura-
ensemblecon(cid:12)gurationforthegalaxiesthatarenotinsub-
tions then certainly becomes larger than six. However,
structures, viz.:
we do not consider it very useful to explore all those
possibilities.
{ [E +S0 ], S , [S +ELG ].
nosub nosub e;nosub l;nosub nosub
We notein passingthat the fact that the Es and S0s out- 7. The nature of the morphological segregations
side substructures togetherconstitute anensemble means
Having studied, through KS2D tests in which we com-
that the intermediate galaxy class, E/S0 can be in-
nosub
pare (R;v)-distributions, which morphological segrega-
cluded in that ensemble.
tions are indicated by our data, it remains to charac-
For the galaxies within substructures, the \segrega-
terize the nature of the various segregations. In Fig. 5
tion rules" require a minimum of two ensembles. This is
we show the (R;v)-distributions of the following 8 galaxy
remarkable, because the de(cid:12)nition of substructures and
classes: outside substructures, the E with M < −22,
the assignment of galaxies to substructures was done to- nosub
the [E+E/S0+S0] , the S , the [S+ELG] ,
tally independent of galaxy type. Applying the \segrega- nosub e;nosub l nosub
andinsubstructurestheE ,theS0 ,theS andthe
tionrules"forthegalaxiesinsubstructures,weobtainfour sub sub e;sub
[S+ELG] .AsexplainedinSect.5,thelast7classesdo
possible two-ensemble con(cid:12)gurations, viz.: l sub
not containgalaxies with M <−22.In order to minimize
{ S0 , [E +S +S +ELG ] the e(cid:11)ects of contamination in the substructure samples,
sub sub e;sub l;sub sub
{ [S0 +E ],[S +S +ELG ] while still retaining a reasonable statistics, we applied a
sub sub e;sub l;sub sub
{ [S0 +S ], [E +S +ELG ] lower limit in (cid:14) of 2.2 (see also Sect. 4.2). Because Es
sub e;sub sub l;sub sub
{ [S0 +E +S ], [S +ELG ]. and S0s outside substructures have (R;v)-distributions
sub sub e;sub l;sub sub
that are not signi(cid:12)cantly di(cid:11)erent, we added the class
In each of these con(cid:12)gurations, the two ensembles always E/S0 to that sample.
nosub
have signi(cid:12)cantly di(cid:11)erent (R;v)-distributions. Because Figure5visuallyillustratesthesegregationresultsdis-
there is no evidence that we need to separate S and cussedinSects.5and6.Thebrightellipticalsclearlyhave
e;sub