Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.arcetri.astro.it/science/k20/papers/bzk.ps Äàòà èçìåíåíèÿ: Tue Feb 17 13:50:01 2004 Äàòà èíäåêñèðîâàíèÿ: Fri Feb 28 06:56:48 2014 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: active galaxy

arXiv:astro­ph/0308456
v1
26
Aug
2003
TO APPEAR ON ASTROPHYSICAL JOURNAL LETTERS, SPECIAL ISSUE FOR GOODS
Preprint typeset using L A T E X style emulateapj v. 3/25/03
NEAR­IR BRIGHT GALAXIES AT z ' 2. ENTERING THE SPHEROID FORMATION EPOCH ? 1
E. DADDI 2 , A. CIMATTI 3 , A. RENZINI 2 , J. VERNET 3 , C. CONSELICE 4 , L. POZZETTI 5 , M. MIGNOLI 5 , P. TOZZI 6 , T. BROADHURST 7 , S.
DI SEREGO ALIGHIERI 3 , A. FONTANA 8 , M. NONINO 6 , P. ROSATI 2 , G. ZAMORANI 5
Submitted 30 April 2003; Accepted 17 June 2003
ABSTRACT
Spectroscopic redshifts have been measured for 9 K­band luminous galaxies at 1:7 < z < 2:3, selected with
Ks < 20 in the K20 survey region of the Great Observatories Origins Deep Survey area. Star formation rates
(SFRs) of  100--500 M yr -1 are derived when dust extinction is taken into account. The fitting of their
multi­color spectral energy distributions indicates stellar masses M >
 10 11 M for most of the galaxies. Their
rest­frame UV morphology is highly irregular, suggesting that merging­driven starbursts are going on in these
galaxies. Morphologies tend to be more compact in the near­IR, a hint for the possible presence of older stellar
populations. Such galaxies are strongly clustered, with 7 out of 9 belonging to redshift spikes, which indicates
a correlation length r 0  9--17 h -1 Mpc (1  range). Current semianalytical models of galaxy formation appear
to underpredict by a large factor ( >
 30) the number density of such a population of massive and powerful
starburst galaxies at z  2. The high masses and SFRs together with the strong clustering suggest that at z  2
we may have started to explore the major formation epoch of massive early­type galaxies.
Subject headings: galaxies: evolution --- galaxies: formation --- galaxies: high­redshift --- galaxies: starbursts
1. INTRODUCTION
The remarkable success of the Cold Dark Matter (CDM)
scenario to account for the cosmic microwave background
power spectrum (Bennett et al. 2003), leaves understand­
ing galaxy formation and evolution as one of the most com­
pelling, unresolved issues of present cosmology. Semiana­
lytical renditions of CDM hierarchical paradigm have so far
favored a slow growth with time, with a major fraction of the
mass assembly taking place at z <
 1 (e.g., Baugh et al. 2002;
Somerville, Primack, & Faber 2001), with virtually all mas­
sive galaxies disappearing by z  1:5. Recent results from
the K20 project (Cimatti et al. 2002a,b,c; Daddi et al. 2002;
Pozzetti et al. 2003) appear to be at variance with these expec­
tations. The K20 project consists of a spectroscopic survey of
 500 Ks < 20 objects selected over 52 arcmin 2 , and has re­
vealed a sizable high redshift tail in the galaxy redshift distri­
bution, where  30 galaxies ( 6% of the total sample) were
found at z > 1:7. Semianalytical CDM models would have
predicted no galaxy at all at such high redshifts in the whole
sample (see Fig. 4 in Cimatti et al. 2002c). The redshift
distribution could be reproduced with pure luminosity evolu­
tion (PLE) models, although not for all realizations (see also
Somerville et al. 2003). However, for only a few among the
z >
 1:5 galaxies was a spectroscopic redshift available, while
for all other such galaxies only the photometric redshifts
could be obtained. In order to put on firmer grounds such
major result of the K20 project (and to understand the nature
of these high­z galaxies) we have conducted new VLT spec­
troscopic observations of galaxies with either photometric or
1 Based on observations collected at the European Southern Observatory,
Chile (ESO programs 70.A­0140, 168.A­0485), and with the NASA/ESA
Hubble Space Telescope, obtained at the Space Telescope Science Institute,
which is operated by AURA Inc, under NASA contract NAS 5­26555.
2 ESO, Karl­Schwarzschild­Str. 2, D­85748 Garching, Germany
3 Osservatorio Astrofisico di Arcetri, L.go E. Fermi 5, Firenze, Italy
4 Caltech, Mail code 105­24, Pasadena (CA) 91125
5 Osservatorio Astronomico di Bologna, via Ranzani 1, Bologna, Italy
6 Osservatorio Astronomico di Trieste, via Tiepolo 11, Trieste, Italy
7 Racah Institute for Physics, The Hebrew University, Jerusalem, Israel
8 Oss. Astron. di Roma, via Dell'Osservatorio 2, Monteporzio, Italy
(uncertain) spectroscopic redshift above z  1:7. This let­
ter reports the results of the new spectroscopic observations,
and combines them with the optical HST+ACS and infrared
VLT+ISAAC imaging made available by the Great Observa­
tories Origins Deep Survey (GOODS) project (Giavalisco et
al. 2003). We assume a Salpeter
IMF;

;
M = 0:7; 0:3, and
h = H 0 [km s -1 Mpc -1 ]=100 = 0:7.
2. THE SPECTROSCOPIC SAMPLE
A sample of 20 galaxies with photometric redshifts
z phot >
 1:7 were selected among the 41 galaxies without spec­
troscopic redshift identification in the 32 arcmin 2 K20 field
that is included in the GOODS­South area. The photometric
redshifts where improved over an earlier estimate (Cimatti et
al. 2002b) by including the ultra­deep JHKs photometry from
the GOODS VLT+ISAAC imaging. Within this sample, 10
objects with most secure z phot have been observed in Novem­
ber 2002 with VLT+FORS2, integrating for 7.2ks with 0 00 :6
seeing, and using the 300V grism covering the range 3600--
8000 å with 13 å resolution for a 1 00 slit.
The spectra were reduced and calibrated in a standard way
(Cimatti et al. 2003b) and co­added to already existing spec­
tra, when available. Redshifts were measured for 7 galax­
ies from absorption features in their blue continua identified
as UV metal lines (Fig. 1). For the z > 2 galaxies Ly
in absorption and the onset of Ly forest are also detected.
One galaxy (ID#5) shows HeII1640 and CIII]1909 emis­
sion lines. Hints for those emission lines are found also for
object ID#9, for which redshift is less secure because of the
faint and noisy spectrum. Together with 2 previously iden­
tified galaxies (ID#1 and ID#2), a sample of 9 galaxies with
spectroscopic redshift z spec > 1:7 is now available among the
304 Ks < 20 galaxies in the K20/GOODS­South area (Ta­
ble 1). Correspondingly, the fraction of Ks < 20 galaxies
with z spec > 1:7 is 3:0 +2:8
-1:4 %, for a surface density of 0:28 +0:26
-0:13
arcmin -2 , and a comoving density of 4:6 +4:3
-2:2 10 -4 h 3 Mpc -3
(the range 1:7 < z < 2:25 is used, hereinafter, for volume
calculations). The uncertainties are poissonian at the 95%
c.l. Accounting for cosmic variance due to clustering (Sect.

2 E. Daddi et al.
FIG. 1.--- The panels show the spectra of, from top to bottom: (1)
ID#7 at z = 2:227; (2) ID#5, the X­ray and radio source at z = 2:223;
(3) the average spectrum of 8 z > 1:7 K­band luminous galaxies (ID#2
has only a near­IR spectrum showing H emission) (4) the average spec­
trum of  250 LBGs with Ly in absorption (Shapley et al. 2003). The
principal metal lines observed in the UV for starburst galaxies are marked
for reference, namely the ISM lines SiII1260;1304;1527;1534;, OI1303,
CII1334:5, CIV1548;1551, SiIV1394;1403, FeII1608, AlII1671,
AlIII1855;1863; as well as the photospheric absorption lines (shown with
an asterisk) FeIII1926, SV1502, SiIII1294, SiII1485,CIII1427.
3) would significantly alter only the upper bounds (see e.g.
Eq. 8 in Daddi et al. 2000). These densities would increase
by a factor of two by including the remaining objects with
z phot >
 1:7. The good agreement between z phot and z spec in
the poorly tested region 1:7 < z < 2:3 (Table 1) suggests a
fair fraction of the latter to be genuine z  2 galaxies. Ob­
taining z spec for these is difficult due to their redder colors and
thus fainter optical magnitudes with respect to galaxies with
measured z spec .
3. PROPERTIES OF K­BAND LUMINOUS GALAXIES AT z  2
Using the wealth of subsidiary data available on the
GOODS­South area we investigate in this section the nature
of these K­band luminous galaxies with z spec  2.
-- Star formation rates. The rest­frame UV spectra of these
galaxies indicate that they are actively star­forming. They are
very similar to the template Lyman Break Galaxies (LBG)
spectrum (Shapley et al. 2003), and often show SiIII1294
absorption due to OB stars (Fig. 1). Ly emission is never
detected, as for 25--50% of LBGs (Shapley et al 2003), a hint
for significant dust extinction. The UV fluxes at 2800å corre­
spond to star formation rates (SFRs) in the range of  10--40
M yr -1 before extinction correction (Kennicutt 1998). The
extinction at 1600å was estimated from the UV spectral slope
 (Meurer et al. 1999) derived from the multicolor photom­
etry. The SFRs derived in this way (adopting the Calzetti
et al. 2000 extinction law) are very high, with a median of
 400 M yr -1 . As an alternative estimate hyperz was used
(Bolzonella et al. 2000) to fit Bruzual & Charlot model spec­
tral energy distributions (SED) to the observed ones from the
available deep VLT BVRIzJHKs photometry, assuming con­
stant SFR (CSF), and solar metallicity. The best fitting models
require reddening in the range E(B-V )  0:3--0.6 and intense
SFRs up to  500 M yr -1 , with a median of 150 M yr -1 .
Using a SMC extinction law would lower both the SFRs and
E(B-V ) estimates. Interestingly, ID#5 (which has the high­
est extinction­corrected SFR) is a faint (soft) X­ray source
(XID563 in the Chandra Deep Field South catalog, Giacconi
et al. 2002). If completely due to star formation, its X­ray
luminosity L rest
2-10keV  2:710 42 erg s -1 cm -2 corresponds to
SFR  500 M yr -1 (Nandra et al. 2002). The object is also a
faint 1.4 Ghz radio source with a flux density of 10313 Jy
(Kellermann et al. 2003, private comunication), fully consis­
tent with the tight X­ray--radio luminosities correlation shown
by Ranalli et al. (2003) for actively star­forming galaxies. We
cannot definitively rule out the presence of an obscured AGN,
but the lack of AGN signatures in its UV spectrum, showing
instead a strong SiIII1294 photospheric absorption line, the
non detection in the Chandra hard band and the low X­ray
to optical flux ratio (log( f 0:5-2kev = f R )  -1:3) indicate this is
a vigorous star­forming galaxy. The stacked X­ray fluxes of
the undetected sources give a 2 detection corresponding to
< SFR > 100 M yr -1 ( = 2:1 is assumed following Brusa
et al. 2002). In general the limits on the X­ray luminosities
(L X <
 10 42 erg s -1 cm -2 ) and the low X­ray to optical flux
ratios (log( f 0:5-2kev = f R ) <
 -1:5) imply that our sample con­
tains virtually no AGN. The high SFRs qualify these galaxies
as starbursts, and allow to build up the equivalent of a local
M   10 11 M galaxy in 0.2--1 Gyr.
-- Stellar masses. Although the SEDs are reddened in the
rest­frame UV, they appear even redder toward the near­IR,
where they show a steep flux increase, starting in the F850LP
band or beyond, which is suggestive of relatively old stellar
populations. The CSF models discussed above imply M stars =
0:3 - 5:5  10 11 M , and luminosity­weighted ages of about
250--1700 Myr. In this case, the near­IR bump is reproduced
by a prominent Balmer break. The CSF models are likely
to underestimate the masses of the galaxies, as older stars,
with higher mass to light ratios, may well be present and yet
their light would be outshined by the younger ones. In order
to estimate reliable upper limits, the minimal contribution to
the K­band light by the ongoing starburst is determined by
fitting a very young ( <
 10 Myr) reddened component to the
SEDs between the B and I bands. This component accounts
for 30--50% of the K­band light. Assuming the remaining K­
band light is due to a maximally old 3 Gyr stellar­population
component, the resulting masses are typically a factor of 2--5
higher than estimated from CSF models, similarly to what is
found for LBGs (Papovich et al. 2001).
-- Clustering. Significant redshift pairing is observed
among z  2 galaxies, a clear indication of strong cluster­
ing. Monte Carlo simulations are used to constrain the corre­
lation length r 0 from the short scale pairing, assuming a slope
= 1:8 for the correlation function (Daddi et al. 2002). A flat
selection function between z = 1:7 and z = 2:25 is used. Seven
independent pairs within 5 h -1 comoving Mpc are found in
our sample of 9 galaxies, implying r 0 > 7 h -1 Mpc comoving
(95% c.l.), and a most likely range 9--17 h -1 Mpc (68% c.l.).

Near­IR bright galaxies at z ' 2 3
FIG. 2.--- ACS (top and center row for F435W and F850LP respectively, epochs 1+2+3) and VLT+ISAAC (bottom row for Ks, seeing 0.5 '' ) imaging for the
galaxies with spectroscopic identification. The images are 5 00 on a side. Redshift measurement for each galaxy is given in the bottom panels.
TABLE 1. K­BAND LUMINOUS STARBURSTS AT z > 1:7 IN THE K20/GOODS AREA
ID z spec z phot Ks J­Ks R­Ks  SFRUV SFR E(B­V) t M r hl C A S
Vega Vega Vega M /yr M /yr Gyr 10 10 M 00 kpc
(1) (2) (3) (4) (4) (4) (4) (5) (5) (5) (5)
1 1.727 1.74 19.99 1.44 3.15 ­0.7 26 93 0.3 0.36 3.4 0.6 4.8 2.2 0.21 0.9
2 1.729 2.54 19.07 1.99 4.54 0.7 13 490 0.6 0.25 11 0.8 7.1 1.8 0.30 1.4
3 1.901 1.65 19.68 1.57 3.42 ­0.9 27 155 0.3 0.52 7.9 0.6 5.4 1.9 0.41 0.6
4 2.060 1.78 19.31 1.83 4.15 ­0.2 35 487 0.5 0.50 25 0.7 5.9 2.0 0.34 1.3
5 2.223 2.40 18.72 2.15 4.15 0.0 38 540 0.4 1.0 55 0.4 3.5 3.2 0.25 0.3
6 2.226 2.24 19.94 1.69 3.30 ­0.4 18 88 0.3 0.71 6.0 0.7 5.5 2.6 0.33 0.5
7 2.227 2.11 19.45 1.80 3.46 ­1.2 31 178 0.3 0.72 13 0.7 5.5 2.2 0.41 1.3
8 2.228 2.43 19.74 2.02 3.66 ­0.4 26 121 0.3 1.4 17 0.9 7.8 2.1 0.25 0.9
9 2.25 2.29 19.94 1.94 3.65 ­0.7 30 153 0.3 1.7 26 1.1 9.4 2.7 0.29 1.2
NOTE. --- (1) The redshift for ID#9 is less secure. (2) UV spectral slope. Typical errors are 0:1. (3) SFR derived from the 2800å luminosity without extinction correction. (4)
Star formation rate, extinction, luminosity weighted stellar age and stellar mass derived from SED fitting of CSF models with reddening. (5) Quantities measured in the F850LP band,
typical errors are
C;
A;
S = 0:15;0:15;0:05.
-- Morphology. In the HST+ACS and VLT+ISAAC images
taken for the GOODS project all galaxies show a rather ir­
regular light distribution (Fig. 2), with bright knots and low
surface brightness regions, often split into separated compo­
nents. We measured the CAS parameters (Conselice 2003;
Bershady et al. 2000; and references therein), finding rela­
tively high clumpiness (S), high asymmetry (A) and very low
concentration (C) (see Table 1). As S is known to correlate
with the SFRs, the large S values are consistent with the high
SFRs estimated above. The A values of most galaxies are
consistent within the errors with the limit of A > 0:35, typical
of galaxies undergoing merging or that experienced merging
in the last Gyr (Conselice 2003). The low C values are also
typical of local merging­driven starbursts, or ultra luminous
infrared galaxies (ULIRGs). There is a trend for increasing
C from the rest frame far­UV (F435W band) to the optical
(K­band, resolution effects having been taken into account),
implying a morphological K­correction. Also this is typical
of starburst galaxies (see e.g. Dey et al. 1999; Smail et al.
2003), and may indicate the presence of an older bulge/disk
component (e.g. LabbÈ et al. 2003b) or a higher reddening in
the central regions. All the galaxies appear rather extended,
allowing to host the high estimated SFRs. The average half
light radius is r hl = 0 00 :7 in the F850LP band, about  6 kpc.
4. RELATING TO OTHER z >
 2 GALAXY POPULATIONS
We now compare the properties of these K­band lumi­
nous galaxies to those of other relevant populations at z >
 2,
namely: LBGs at z  3, very red z > 2 galaxies, and SCUBA
sources. Compared to LBGs (e.g. Giavalisco et al. 2002),
these near­IR bright starbursts at z  2 have, on average, larger
sizes, higher masses and SFRs, and stronger clustering. De­
spite their spectral similarity, these galaxies are not just a spe­
cial subsample of LBGs. Indeed, it appears that they have
redder UV continuum than the reddest LBGs template (Fig.
1). In fact, most objects in Table 1 have  > -1, while most
LBGs have UV slopes between  = -2 to -1, and virtually
none has  > -0:5 (Adelberger & Steidel 2000). Hence, the
two populations appear only partially overlapping.
These z  2 starbursts are red in the near­IR, with J -
K >
 1:7, and their clustering is consistent with that of much
fainter Ks < 24 galaxies at z > 2 with J - K > 1:7 colors
(Daddi et al. 2003). In fact, van Dokkum et al. (2003)
found significant redshift pairing among 5 galaxies at z > 2
selected with J - K > 2:3 (Franx et al. 2003; LabbÕ et al.
2003a). Nevertheless, the two samples show different proper­
ties, as strong Ly emissions, regular morphologies, and AGN
signatures are common among van Dokkum et al. objects.
Our 9 spectroscopically confirmed galaxies have J -K < 2:3
and very clumpy and asymmetric morphologies. We conclude
that there is a large variety of properties among K­band bright
galaxies at z > 2, that we are just starting to explore.
Given the estimated SFRs, redshift range and peculiar mor­
phology, some of our galaxies are potential SCUBA sources
(Chapman 2003a,b). Red UV SEDs with  > -0:5 are in­
deed common among SCUBA sources (e.g. Dey et al. 1999,
Chapman et al. 2002), which also appear to have large clus­
tering (e.g. Webb et al. 2003). Nevertheless, SCUBA sources

4 E. Daddi et al.
have much lower spatial density and are often much fainter
than Ks = 20, while some of our sources may have too low
SFRs to be submm­bright. Hence, also in this case the two
populations are likely to overlap only partially.
5. DISCUSSION
These K­band luminous starbursts provide a substantial
contribution to the cosmic SFR density (SFRD) at z  2:
adding up the SFRs from SED modeling we derive SFRD 
0:04 M yr -1 Mpc -3 from the 9 spectroscopically confirmed
galaxies alone. This estimate is certainly affected by in­
completeness, and yet it already represents  30--60% of the
SFRD within the range 1:5 < z < 3 (see e.g. the compila­
tion by Nandra et al. 2002). These galaxies are also among
the most massive systems detected at z  2. Six objects in our
sample have M stars > 10 11 M (conservative estimates), result­
ing in a number density  10 -4 Mpc -3 and a mass density of
 210 7 M Mpc -3 , both  10 % of the corresponding local
value (Cole et al. 2001). Integrating over the mass function
predicted by the Baugh et al. (2002) model at z = 1:92, one
expects on average only 0.2 galaxies with M stars > 10 11 M
within the explored volume, and even less if the semianalyt­
ical mass function was properly normalized at z = 0. There­
fore, these semianalytical models underestimate the number
of massive galaxies at z  2 by about a factor of 30, and pos­
sibly much more given the incompletenesses of our spectro­
scopic sample. The assembly of massive galaxies apparently
took place at a significantly larger redshift (earlier epoch) than
predicted by the models (see also Genzel et al. 2003). On the
other hand, these z  2 galaxies are too actively star­forming
and irregular to be consistent with PLE models with high red­
shift of formation. The agreement between the observed red­
shift distribution at z > 1:7 and the PLE model described in
Cimatti et al. (2002c) is therefore likely to be just chance.
These K­band luminous starbursts are very strongly
clustered, suggesting they are hosted in very massive and
biased environments, which itself argues for these objects
being quite massive. At z < 2 the only known sources with
r 0 >
 7 h -1 Mpc are old, passively evolving EROs (Daddi et
al. 2000, 2001) and local massive ellipticals (Norberg et al.
2002). These z  2 galaxies are therefore likely to evolve
into such classes of objects. If star­formation ends rapidly,
it would take them >
 1 Gyr to develop very red optical
to near­IR colors and to morphologically relax to regular
bulge­dominated galaxies. This scenario, with massive
spheroids still forming at z  2--3, would be quite in good
agreement with some properties of z  1 old EROs, including
their number counts (Daddi, Cimatti & Renzini 2000), hints
for residual star­formation present in their UV rest­frame
(McCarthy et al. 2001), and with the inferred formation
redshifts (2:4  0:3) of their stellar populations (Cimatti et
al. 2002a). At the same time, it would predict a paucity of
passive EROs at, say, z > 1:3--1.5. Finally, we notice that a
major shift seems to happen for the clustering properties of
star­forming galaxies from z  1, where they have very low
clustering (see e.g. Daddi et al. 2002), to z  2, where they
have a very large one. The straightforward interpretation is
that while at z <
 1 star formation is mostly confined to low­
mass galaxies, at z  2 we are starting to see the major build
up phase of massive early­type galaxies. It remains to be
determined whether this z  2 activity represents the peak or
the low­z tail of the massive spheroid formation epoch. With
the existing technology we should soon be able to answer this
question, mapping massive galaxy assembly as a function of
both redshift and large scale structure environment.
We are in debt to K. Kellermann for measuring the radio
flux density of ID#5. We thank J. Bergeron, M. Brusa, R.
Fosbury, M. Franx and V. Mainieri for discussions, C. Steidel
for providing the LBG composite spectra, and the referee, J.
Primack, for useful comments.
REFERENCES
Adelberger K. L., Steidel C. C., 2000, ApJ, 544, 218
Baugh C.M., Benson A.J., Cole S., et al., 2002, in `The Mass of Galaxies at
Low and High Redshift', eds. R. Bender, A. Renzini (astro­ph/0203051)
Bennett C.L., et al., 2003, submitted to ApJ (astro­ph/0302207)
Bershady M.A., Jangren J.A., Conselice C.J. 2000, AJ, 119, 2645
Brusa M., Comastri A., Daddi E., et al., 2002, ApJ, 581, L89
Bolzonella M., Miralles J.­M., PellÑ R., 2000, A&A, 363, 476
Calzetti D., Armus L., Bohlin R. C., et al., 2000, ApJ, 533, 682
Cimatti A., Daddi E., Mignoli M., et al., 2002a, A&A 381, L68
Cimatti A., Mignoli M., Daddi E., et al., 2002b, A&A 392, 395
Cimatti A., Pozzetti L., Mignoli M., et al., 2002c, A&A 391, L1
Chapman S. C., Shapley A., Steidel C., Windhorst R., 2002, ApJ, 572, L1
Chapman S. C., Blain A. W., Ivison R. J., Smail I., 2003a, Nature, 422, 695
Chapman S. C., et al., 2003b, conference proceedings (astro­ph/0304236)
Cole S., Norberg P., Baugh C. M., et al., 2001, MNRAS, 326, 255
Conselice C.J., 2003, ApJS, 147, 1
Daddi E., Cimatti A., Pozzetti L., et al., 2000, A&A 361, 535
Daddi E., Cimatti A., Renzini A., 2000, A&A 362, L45
Daddi E., Broadhurst T. J., Zamorani G., et al., 2001, A&A, 376, 825
Daddi E., Cimatti A., Broadhurst T. J., et al., 2002, A&A 384, L1
Daddi E., RÆttgering H., LabbÈ I, et al., 2003, ApJ, 588, 50
Dey A., Graham J. R., Ivison R. J., et al., 1999, ApJ, 519, 610
Franx M., LabbÈ I., Rudnick G., et al., 2003, ApJ, 587, L79
Genzel R., Baker A. J., Tacconi L. J., et al., 2003, ApJ, 584, 633
Giacconi R., Zirm A., Wang J., et al., 2002, ApJS, 139, 369
Giavalisco M., 2002, ARA&A, 40, 579
Giavalisco M., et al., 2003, this volume
Kellermann K.I., Fomalont E.B., Rosati P., Shaver P., 2003, in preparation
Kennicutt R. C., 1998, ARA&A, 36, 189
LabbÈ I., Rudnick G., Franx M., et al., 2003, ApJ, 591, L??
LabbÈ I., Franx M., Rudnick G., et al., 2003, AJ, 125, 1107
McCarthy P. J., Carlberg R. G., Chen H.­W., et al., 2001, ApJ, 560, L131
Meurer G. R., Heckman T. M., Calzetti D., 1999, ApJ, 521, 64
Nandra K., Mushotzky R. F., Arnaud K., et al., 2002, ApJ, 576, 625
Norberg P., Baugh C. M., Hawkins E., et al., 2002, MNRAS, 332, 827
Papovich C., Dickinson M., Ferguson H. C., 2001, ApJ, 559, 620
Pozzetti L., Cimatti A., Zamorani G., et al., 2002, A&A, 402, 837
Ranalli P., Comastri A., Setti G., 2003, A&A, 399, 39
Somerville R. S., Primack J. R., Faber S. M., 2001, MNRAS, 320, 504
Somerville R. S., Moustakas L. A., Mobasher B., et al., 2003, this volume
Shapley A., Steidel C.C., Pettini M., Adelberger K.L., 2003, ApJ, 588, 65
Smail I., Chapman S.C., Ivison R.J., et al., 2003 (astro­ph/0303128)
van Dokkum P. G., et al., 2003, ApJ, 587, L83
Webb T. M., Eales S. A., Lilly S. J., et al., 2003, ApJ, 587, 41