We present the largest spectroscopic follow-up performed in SWIRE ELAIS-N1. We were able to determine redshifts for 289 extragalactic sources. The values of spectroscopic redshifts of the latter have been compared with the estimated values from our photometric redshift code with very good agreement between the two for both galaxies and quasars. Six of the quasars are hyperluminous infrared galaxies all of which are broad line AGN. We have performed emission line diagnostics for 30 sources in order to classify them into star-forming, Seyferts, composite and LINER and compare the results to the predictions from our SED template fitting methods and mid-IR selection methods.

We report on a statistical study of the 51 radio galaxies at the millijansky flux level from the Faint Images of the Radio Sky at Twenty centimeters, including their optical morphologies and structure obtained with the Hubble Space Telescope. Our optical imaging is significantly deeper (~2 mag) than previous studies with the superior angular resolution of space-based imaging. We that find 8/51 (16%) of the radio sources have no optically identifiable counterpart to AB~24 mag. For the remaining 43 sources, only 25 are sufficiently resolved in the HST images to reliably assign a visual classification: 15 (60%) are elliptical galaxies, 2 (8%) are late-type spiral galaxies, 1 (4%) is an S0, 3 (12%) are point-like objects (quasars), and 4 (16%) are merger systems. We find a similar distribution of optical types with measurements of the Sersic index. The optical magnitude distribution of these galaxies peaks at I~20.7+-0.5 AB mag, which is ~3 mag brighter than the depth of our typical HST field and is thus not due to the WFPC2 detection limit. This supports the luminosity-dependent density evolutionary model, where the majority of faint radio galaxies typically have L*-optical luminosities and a median redshift of z~0.8 with a relatively abrupt redshift cut-off at z>~2. We discuss our results in the context of the evolution of elliptical galaxies and active galactic nuclei.

We present the optical-to-submm spectral energy distributions for 33 radio & mid-IR identified submillimetre galaxies discovered via the SHADES 850 micron SCUBA imaging in the Subaru-XMM Deep Field (SXDF). Optical data for the sources comes from the Subaru-XMM Deep Field (SXDF) and mid- and far-IR fluxes from SWIRE. We obtain photometric redshift estimates for our sources using optical and IRAC 3.6 and 4.5 micron fluxes. We then fit spectral energy distribution (SED) templates to the longer wavelength data to determine the nature of the far-IR emission that dominates the bolometric luminosity of these sources. The infrared template fits are also used to resolve ambiguous identifications and cases of redshift aliasing. The redshift distribution obtained broadly matches previous results for submm sources and on the SHADES SXDF field. Our template fitting finds that AGN, present in about 10% of sources, do not contribute significantly to their bolometric luminosity. Dust heating by starbursts, with either Arp220 or M82 type SEDs, appears to be responsible for the luminosity in most sources (23/33 are fitted by Arp220 templates, 2/33 by the warmer M82 templates). 8/33 sources, in contrast, are fit by a cooler cirrus dust template, suggesting that cold dust has a role in some of these highly luminous objects. Three of our sources appear to have multiple identifications or components at the same redshift, but we find no statistical evidence that close associations are common among our SHADES sources. Examination of rest-frame K -band luminosity suggests that 'downsizing' is underway in the submm galaxy population, with lower redshift systems lying in lower mass host galaxies. Of our 33 identifications six are found to be of lower reliability but their exclusion would not significantly alter our conclusions.

We investigate the clustering of galaxies selected in the 3.6 micron band of the Spitzer Wide-area Infrared Extragalactic (SWIRE) legacy survey. The angular two-point correlation function is calculated for eleven samples with flux limits of S_3.6 > 4-400 mujy, over an 8 square degree field. The angular clustering strength is measured at >5-sigma significance at all flux limits, with amplitudes of A=(0.49-29)\times10^{-3} at one degree, for a power-law model, A\theta^{-0.8}. We estimate the redshift distributions of the samples using phenomological models, simulations and photometric redshifts, and so derive the spatial correlation lengths. We compare our results with the GalICS (Galaxies In Cosmological Simulations) models of galaxy evolution and with parameterized models of clustering evolution. The GalICS simulations are consistent with our angular correlation functions, but fail to match the spatial clustering inferred from the phenomological models or the photometric redshifts. We find that the uncertainties in the redshift distributions of our samples dominate the statistical errors in our estimates of the spatial clustering. At low redshifts (median z<0.5) the comoving correlation length is approximately constant, r_0=6.1\pm0.5h^{-1} Mpc, and then decreases with increasing redshift to a value of 2.9\pm0.3h^{-1} Mpc for the faintest sample, for which the median redshift is z=1. We suggest that this trend can be attributed to a decrease in the average galaxy and halo mass in the fainter flux-limited samples, corresponding to changes in the relative numbers of early- and late-type galaxies. However, we cannot rule out strong evolution of the correlation length over 0.5<z<1.

By combining the VVDS with the SWIRE data, we have built the currently largest spectroscopic sample of galaxies selected in the rest-frame near-infrared. These allow us to investigate, for the first time using spectroscopic redshifts, the clustering evolution of galaxies selected from their rest-frame near-infrared luminosity in the redshift range 0.2<z<2.1. Therefore we use the projected two-point correlation function wp(rp) to study the three dimensional clustering properties of galaxies detected at 3.6 and 4.5 microns with IRAC and for which we have spectroscopic redshifts from first epoch VVDS. We find that in the flux limited samples at 3.6 and 4.5 microns, the clustering length does not change from redshift ~2 to the present. In addition, we find that the clustering lengths are systematically higher when galaxy samples are selected from increasingly redder wavelengths. These values are larger than the typical values found for I-band selected galaxies in the same redshift range, but we find that the difference in clustering length between I-band and 3.6/4.5 microns selected samples is decreasing with increasing redshift to become comparable at z~1.5. We interpret this as evidence that galaxies with older stellar populations and galaxies actively forming stars reside in comparably overdense environments at epochs earlier than z~1.5. The increasing difference in clustering length observed between rest-frame UV-optical and infrared selected samples could then be an indication that star formation is gradually shifting to lower density regions as cosmic time increases.

We analyse the first publicly released deep field of the UKIDSS Deep eXtragalactic Survey (DXS) to identify candidate galaxy over-densities at z~1 across ~1 sq. degree in the ELAIS-N1 field. Using I-K, J-K and K-3.6um colours we identify and spectroscopically follow-up five candidate structures with Gemini/GMOS and confirm they are all true over-densities with between five and nineteen members each. Surprisingly, all five structures lie in a narrow redshift range at z=0.89+/-0.01, although they are spread across 30Mpc on the sky. We also find a more distant over-density at z=1.09 in one of the spectroscopic survey regions. These five over-dense regions lying in a narrow redshift range indicate the presence of a supercluster in this field and by comparing with mock cluster catalogs from N-body simulations we discuss the likely properties of this structure. Overall, we show that the properties of this supercluster are similar to the well-studied Shapley and Hercules superclusters at lower redshift.

We present measurements of the spatial clustering of ultraluminous infrared galaxies in two redshift intervals, 1.5<z<2.0 and 2<z<3. Both samples cluster strongly, with r_0=14.40+/-1.99 h^-1 Mpc for the 2<z<3 sample, and r_0=9.40+/-2.24 h^-1 Mpc for the 1.5<z<2.0 sample, making them among the most biased galaxies at these epochs. These clustering amplitudes are consistent with both populations residing in dark matter haloes with masses of ~6x10^13 solar masses. We infer that a minimum dark matter halo mass is an important factor for all forms of luminous, obscured activity in galaxies at z>1. Adopting plausible models for the growth of DM haloes with redshift, then the haloes hosting the 2<z<3 sample will likely host the richest clusters of galaxies at z=0, whereas the haloes hosting the 1.5<z<2.0 sample will likely host poor to rich clusters at z=0.

We have carried out a deep (21.5<R<23.9) systematic survey of galaxies selected from SWIRE using GMOS to target X-ray sources, AGN and galaxies with photometric redshifts greater than 1. We have obtained a total of 198 redshifts, of which 132 have secure redshifts in the range 0.12<z<1.21. This sample, combined with the 111 redshifts targeted with WIYN in the same field and the 1 square degree survey of ELAIS-N1 with ACIS-I, will allow us to calibrate our photometric redshift techniques; characterise the star formation history of the universe between 0.5<z<1.2 where galaxy formation activity peaks; and explore the connection between AGN and star-formation activity.

We present details of the discovery of XLSSJ022303.0-043622, a z=1.2 cluster of galaxies. This cluster was identified from its X-ray properties and selected as a z>1 candidate from its optical/near-IR characteristics in the XMM Large-Scale Structure Survey (XMM-LSS). It is the most distant system discovered in the survey to date. We present ground-based optical and near IR observations of the system carried out as part of the XMM-LSS survey. The cluster has a bolometric X-ray luminosity of 1.1 +/- 0.7 x 10^44 erg/s, fainter than most other known z>1 X-ray selected clusters. In the optical it has a remarkably compact core, with at least a dozen galaxies inside a 125 kpc radius circle centred on the X-ray position. Most of the galaxies within the core, and those spectroscopically confirmed to be cluster members, have stellar masses similar to those of massive cluster galaxies at low redshift. They have colours comparable to those of galaxies in other z>1 clusters, consistent with showing little sign of strong ongoing star formation. The bulk of the star formation within the galaxies appears to have ceased at least 1.5 Gyr before the observed epoch. Our results are consistent with massive cluster galaxies forming at z>1 and passively evolving thereafter. We also show that the system is straightforwardly identified in Spitzer/IRAC 3.6 and 4.5 micron data obtained by the SWIRE survey emphasising the power and utility of joint XMM and Spitzer searches for the most distant clusters.

We present the rest-frame optical and infrared colours of a complete sample of 1114 z<0.3 galaxies from the Spitzer Wide-area InfraRed Extragalactic Legacy Survey (SWIRE) and the Sloan Digital Sky Survey (SDSS). We discuss the optical and infrared colours of our sample and analyse in detail the contribution of dusty star-forming galaxies and AGN to optically selected red sequence galaxies. We propose that the optical (g-r) colour and infrared log(L_{24}/L_{3.6}) colour of galaxies in our sample are determined primarily by a bulge-to-disk ratio. The (g-r) colour is found to be sensitive to the bulge-to-disk ratio for disk-dominated galaxies, whereas the log(L_{24}/L_{3.6}) colour is more sensitive for bulge-dominated systems. We identify ~18% (195 sources) of our sample as having red optical colours and infrared excess. Typically, the infrared luminosities of these galaxies are found to be at the high end of star-forming galaxies with blue optical colours. Using emission line diagnostic diagrams, 78 are found to have an AGN contribution, and 117 are identified as star-forming systems. The red (g-r) colour of the star-forming galaxies could be explained by extinction. However, their high optical luminosities cannot. We conclude that they have a significant bulge component. The number densities of optically red star-forming galaxies are found to correspond to ~13% of the total number density of our sample. In addition, these systems contribute ~13% of the total optical luminosity density, and 28% of the total infrared luminosity density of our SWIRE/SDSS sample. These objects may reduce the need for "dry-mergers".

We present measurements of the spatial clustering of galaxies with stellar masses >10^11Msun, infrared luminosities >10^12 Lsun, and star formation rates >200Msun per year in two redshift intervals; 1.5<z<2.0 and 2<z<3. Both samples cluster very strongly, with spatial correlation lengths of r_0=14.40+/-1.99 h^-1Mpc for the 2<z<3 sample, and r_0=9.40+/-2.24 h^-1Mpc for the 1.5<z<2.0 sample. These clustering amplitudes are consistent with both populations residing in dark matter haloes with masses of ~6x10^13Msun, making them among the most biased galaxies at these epochs. We infer, from this and previous results, that a minimum dark matter halo mass is an important factor for all forms of luminous, obscured activity in galaxies at z>1, both starbursts and AGN. Adopting plausible models for the growth of DM haloes with redshift, then the haloes hosting the 2<z<3 sample will likely host the richest clusters of galaxies at z=0, whereas the haloes hosting the 1.5<z<2.0 sample will likely host poor to rich clusters at z=0. We conclude that ULIRGs at z>1 signpost stellar buildup in galaxies that will reside in clusters at z=0, with ULIRGs at increasing redshifts signposting the buildup of stars in galaxies that will reside in increasingly rich clusters.

We fit a parametric model comprising a mixture of multi-dimensional Gaussian functions to the 3.6 to 8um colour and optical photo-z distribution of galaxy populations in the ELAIS-N1 and Lockman Fields of SWIRE. For 16,698 sources in ELAIS-N1 we find our data are best modelled (in the sense of the Bayesian Information Criterion) by the sum of four Gaussian distributions or modes (C_a, C_b, C_c and C_d). We compare the fit of our empirical model with predictions from existing semi-analytic and phenomological models. We infer that our empirical model provides a better description of the mid-infrared colour distribution of the SWIRE survey than these existing models. This colour distribution test is thus a powerful model discriminator and complementary to comparisons of number counts. We use our model to provide a galaxy classification scheme and explore the nature of the galaxies in the different modes of the model. C_a consists of dusty star-forming systems such as ULIRG's. Low redshift late-type spirals are found in C_b, where PAH emission dominates at 8um. C_c consists of dusty starburst systems at intermediate redshifts. Low redshift early-type spirals and ellipticals dominate C_d. We thus find a greater variety of galaxy types than one can with optical photometry alone. Finally we develop a new technique to identify unusual objects, and find a selection of outliers with very red IRAC colours. These objects are not detected in the optical, but have very strong detections in the mid-infrared. These sources are modelled as dust-enshrouded, strongly obscured AGN, where the high mid-infrared emission may either be attributed to dust heated by the AGN or substantial star-formation. These sources have z_ph ~ 2-4, making them incredibly infrared luminous, with a L_IR ~ 10^(12.6-14.1) L_sun.

We present an analysis of large-scale structure from the Spitzer Wide-area Infrared Extragalactic legacy survey, SWIRE. The two-point angular correlation functions were computed for galaxies detected in the 3.6-micron IRAC band, on angular scales up to a degree. Significant evolution in the clustering amplitude was detected, as the median redshift of the samples increases from z=0.2 to 0.6. The galaxy clustering in the GALICS semi-analytic models was compared with the observed correlation functions and found to disagree with the data at faint flux limits.

We present a deep ISO observation at 6.7 microns of the 53W002 group of galaxies and AGN at z=2.4. This approximately samples the emitted K band. The faint, blue star-forming objects are not detected, as expected from their very blue color across the emitted optical and UV. However, 53W002 itself is detected at the 3-sigma level, with an emitted V-K color appropriate for a population formed starting at z=3.6-7.0 with most likely value z=4.7. This fits with shorter-wavelength data suggesting that the more massive members of this group, which may all host AGN, began star formation earlier in deeper potential wells than the compact Lyman-alpha emission objects. Two foreground galaxies are detected, as well as several stars. One additional 6.7-micron source closely coincides with an optically faint galaxy, potentially at z=2-3. The overall source counts are consistent with other ISO deep fields.

We present the first analysis of large-scale clustering from the Spitzer Wide-area InfraRed Extragalactic legacy survey: SWIRE. We compute the angular correlation function of galaxies selected to have 3.6 micron fluxes brighter than 32 muJy in three fields totaling two square degrees in area. In each field we detect clustering with a high level of significance. The amplitude and slope of the correlation function is consistent between the three fields and is modeled as $w(theta)=A\theta^{1-\gamma}$ with $A=(0.6\pm0.3)\times10^{-3}, \gamma=2.03\pm0.10$. With a fixed slope of $\gamma=1.8$, we obtain an amplitude of $A=(1.7\pm0.1)\times10^{-3}$. Assuming an equivalent depth of K~18.7 mag we find our errors are smaller but our results are consistent with existing clustering measurements in K-band surveys and with stable clustering models. We estimate our median redshift $z\simeq 0.75$ and this allows us to obtain an estimate of the three-dimensional correlation function $\xi(r)$, for which we find $r_0=4.4\pm0.1 h^{-1}$ Mpc.

In a recent paper Lehnert & Bremer have photometrically selected a sample of galaxies at z>4.8 from a single VLT/FORS2 pointing and spectroscopically confirmed half of them to be at 4.8<z<5.8. To study the properties of such galaxies further, we have photometrically selected a similar sample (V(AB)>28, i(AB)<26.3, i(AB)-z(AB)>0) from the HST ACS images of the Chandra Deep Field South. This selection results in a sample of 44 sources from ~150 sq. arcmin. We find that such galaxies are often barely resolved in the ACS images, having half-light radii of 0.1-0.3 arcsec (<2 kpc). They show no difference in spatial clustering from sources selected by i(AB)<26.3, i(AB)-z(AB)>0, which are generally galaxies of lower redshift. However, their distribution over the field is not uniform and their surface density varies considerably over areas comparable to a single 8m or HST pointing. The reliable determination of the surface and volume densities of such galaxies requires a sky area considerably larger than the current ACS imaging of this field. No individual z>5 candidate was detected to a 3-sigma limit of 6 x 10^-17 erg s^-1 cm^-2 at 0.5-5 keV by Chandra (a limiting luminosity of below 2 x 10^43 erg s^-1 at z~5.3). By summing over all positions, we find that the mean source must be undetected at a level at least a factor 4 times fainter than this. This rules out anything other than a weak AGN contribution to the emission from these objects and thus luminous AGN made little contribution to the final stages of re-ionization of the Universe.

Deep spectroscopy of the two mJy radio galaxies LBDS 53W069 and 53W091 has previously shown them to have old (>3 Gyr) stellar populations at z=1.5. Here we present the results of HST observations of these galaxies. 53W069 has an r^1/4 profile in both the F814W & F110W data with an effective radius of 0.30 arcsec (2.7 kpc). The restframe U-B colour gradient requires a stellar population of super-solar (3Z_sun) metallicity that formed on a very short timescale at high redshift (z>5). 53W091 has a regular r^1/4 profile in F110W with an effective radius of 0.32 arcsec (2.9 kpc). The F814W profile is more extended and is consistent with a blue exponential disk that contributes 20% of the flux within r_e. The restframe U-B colour gradient is significantly larger than that observed in field ellipticals at z<1, implying a stellar population of mixed metallicity (1-3Z_sun) that formed in a high-redshift rapid burst. We have compared these two LBDS radio galaxies with the Kormendy relation of ten 3CR radio galaxies at z=0.8. The LBDS galaxies follow the same relation as the more radio-luminous 3CR galaxies, assuming passive evolution of their stellar populations, although they are smaller than the 3CR galaxies whose mean effective radius is 12 kpc. Their sizes and radio luminosities are consistent with scaling relations applied to the 3CR galaxies, in which both radio power and effective radius scale with galaxy mass. Compared with a sample of z=0.4 cluster ellipticals, 53W069 & 53W091 lie well within the scatter of the Kormendy relation. We conclude that the hosts of these mJy radio sources at z=1.5 are passively-evolving elliptical galaxies that will evolve into ordinary L* ellipticals by the present day. (Abridged)

We present high-resolution imaging in the rest-frame optical of the weak radio source LBDS53W091. Previous optical spectroscopy has shown that this object has an evolved stellar population of age >3Gyr at z=1.55, determined from the amplitude of rest-frame UV spectral breaks. We have obtained deep Hubble Space Telescope imaging over 10 orbits with NICMOS camera 2, using the F160W H-band filter (1.6microns) which is a good approximation to the rest-frame R-band. Our observations reveal a radial light profile which is well fit by a de Vaucouleurs' r^{1/4} law, with a scale length of r_e=0.3" (3kpc/h_{50} for Omega_M=0.3 and Omega_{Lambda}=0). The elliptical morphology of the radio galaxy indicates a dynamically-evolved old system, consistent with the spectroscopic results. Some surrounding objects lie on the color:magnitude relation for a cluster at z=1.55, and are likely to be associated. The group of galaxies are somewhat more luminous than the fundamental plane of ellipticals at z=0 subject to (1+z)^4 surface brightness dimming, but are consistent with estimates of the fundamental plane at high redshift subject to passive luminosity evolution from a formation epoch of z>3.

We present near-IR observations of AGN and star-forming objects in the field of the radio galaxy 53W002 at z=2.39. The star-forming objects, of special interest as candidate protogalactic objects, are uniformly very blue in continuum slope, which fits with the strong Lyman alpha emission in indicating metallicity less than half solar. They fall in a range of luminosity and metallicity not populated by local objects. NICMOS spectroscopy detects [O III] emission for two objects and sets interesting limits for the rest. These data are satisfied by model stellar populations which have been forming stars for the last 2-5 million years. We see no evidence for older populations, either in broadband colors or as redder halos around the star-forming regions. These results suggest that the compact star-forming objects we see at z=2.0-2.5 are indeed early stages in the building of galaxies, rather than transient star-forming events in larger systems. They also also permit an alternative scheme, in which these low-mass systems are blowing winds rather than self-enriching, and should fade rapidly with cosmic epoch. For the three prominent AGN at z=2.39, emission-line ratios show metallicities just as high as in nearby systems. The "ionization cones" appear in [O III] and H lines, with continuum reflection in some cases as well. The contrast between the AGN and fainter star-forming objects can be broadly accomodated in a hierarchical formation picture, though there are still important unknowns as to the fate of the star-forming objects.

{Abridged} A combination of spectroscopy and broadband photometric redshifts has been used to find the complete redshift distribution of the Hercules sample of millijansky radio sources. These data have been used to examine the evolution of the radio luminosity function (RLF) and its high-redshift cut-off. New redshifts have been measured for eleven sources, and a further ten upper limits are given. The total number of sources with known redshifts in the sample is now 47 (65%). We calculated broadband photometric redshifts for the remaining one-third of the sample. For the luminosity range probed by the present study (P_1.4 > 10^24.5 W/Hz/sr), we use the V/V_max test to show conclusively that there is a deficit of high-redshift (z > 2-2.5) objects. Comparison with the model RLFs of Dunlop & Peacock (1990) shows that our data can now exclude pure luminosity evolution. Two of the models of DP90, and the RLF deduced by direct binning of the data, both favour a luminosity dependence for the high-redshift cut-off, with lower-luminosity sources (P_1.4 \simeq 10^24 W/Hz/sr) in decline by z \simeq 1-1.5 while higher-luminosity sources (P_1.4 \simeq 10^{25-26} W/Hz/sr) decline in comoving number density beyond z \simeq 2-2.5.

The results are presented of an extensive programme of optical and infrared imaging of radio sources in a complete subsample of the Leiden-Berkeley Deep Survey. The LBDS Hercules sample consists of 72 sources observed at 1.4 GHz, with flux densities S(1.4) >= 1.0 mJy, in a 1.2 deg^2 region of Hercules. This sample is almost completely identified in the g, r, i and K bands, with some additional data available at J and H. The magnitude distributions peak at r ~= 22 mag, K ~= 16 mag and extend down to r ~= 26 mag, K ~= 21 mag. The K-band magnitude distributions for the radio galaxies and quasars are compared with those of other radio surveys. At S(1.4) <~ 1 Jy, the K-band distribution does not change significantly with radio flux density. The sources span a broad range of colours, with several being extremely red (r-K >~ 6). Although small, this is the most optically complete sample of millijansky radio sources available at 1.4 GHz, and is ideally suited to study the evolution of the radio luminosity function out to high redshifts.

We present the discovery of a radio galaxy at a likely redshift of z = 4.424 in one of the flanking fields of the Hubble Deep Field. Radio observations with the VLA and MERLIN centered on the HDF yielded a complete sample of microjansky radio sources, of which about 20% have no optical counterpart to I < 25 mag. In this Letter, we address the possible nature of one of these sources, through deep HST NICMOS images in the F110W (J) and F160W (H) filters. VLA J123642+621331 has a single emission line at 6595-A, which we identify with Lyman-alpha at z = 4.424. We argue that this faint (H = 23.9 mag), compact (r = 0.2 arcsec), red (I - K = 2.0) object is most likely a dusty, star-forming galaxy with an embedded active nucleus.

We present deep wide-field medium-band imaging in redshifted Lyman alpha of fields surrounding HST detections of faint emitters, to characterize their larger-scale environment. The radio galaxy 53W002 was previously found to be part of a rich grouping at z=2.39, including 5 confirmed, compact, star-forming objects. This is now shown to be part of a larger structure traced by bright AGN, all contained within a projected span of 6.8 arcmin (3.2 Mpc). Of the 14 candidate emitters, six have been spectroscopically confirmed as active nuclei in the range z=2.390 +/- 0.008. Various statistical tests give a significance of 95-99% for the reality of this structure on the sky. This grouping is more extended than a relaxed King model, at the 90% confidence level. Either this configuration has yet to decouple fully from the Hubble expansion, or it consists of subgroups which will themselves form a more compact, relaxed structure. The redshift range for measured members is comparable to the Hubble flow across the structure, which may imply that the structure is seen near turnaround. We surveyed two additional fields at z=2.4, and three contiguous fields at z=2.6. Only a single candidate at z=2.4 was found, with six at z=2.6. Comparison with the WFPC2 surveys suggests that the field-to-field contrast is larger for brighter objects. From this survey alone, groupings such as the 53W002 "cluster" must have an area covering fraction <0.04 in this redshift range.

The results of an optical and infrared investigation of a radio sample drawn from the 1.4 GHz Leiden-Berkeley Deep Survey are presented. This is believed to be the most comprehensive sample of radio sources at millijansky flux limits that is currently available. Optical counterparts have been identified for all but four sources in the two Hercules fields, and 80% of them are identified in the near-infrared. Redshifts have been obtained for 49 of the identified sources, and photometric redshifts were computed from the g,r,i,K data for the remaining 20. The general properties of the sample are summarized. The use of this sample in measuring the 1.4 GHz radio luminosity function is discussed. Finally, HST/NICMOS images of two old, red galaxies at z=1.5 are presented which show that both galaxies are dominated by an r^{1/4} profile with a scale-length of 5 kpc.

Faint extragalactic radio sources provide important information about the global history of star formation. Sensitive radio observations of the Hubble Deep Field and other fields have found that sub-mJy radio sources are predominantly associated with star formation activity rather than AGN. Radio observations of star forming galaxies have the advantage of being independent of extinction by dust. We use the FIR-radio correlation to compare the radio and FIR backgrounds, and make several conclusions about the star forming galaxies producing the FIR background. We then use the redshift distribution of faint radio sources to determine the evolution of the radio luminosity function, and thus estimate the star formation density as a function of redshift.

The results of an optical and infrared investigation of a complete sub-sample of the Leiden-Berkeley Deep Survey (flux > 1 mJy at 1.4 GHz) are presented. Optical counterparts have been identified for 69 of the 73 sources in the two Hercules fields, and redshifts obtained for 49 of them. Photometric redshifts are computed from the g,r,i,K data for the remaining 21 sources. This complete sample is compared with the radio luminosity functions (RLFs) of Dunlop and Peacock (1990). The RLF models successfully trace the evolution of the radio sources with redshift, but there is some disagreement between the luminosity-dependence of the models and the data. The observed RLF for the lower luminosity population (log P < 26) shows evidence for a cut-off at lower redshifts (z=0.5-1.5) than for the more powerful objects.

We have discovered a population of extremely red galaxies at $z\simeq 1.5$ which have apparent stellar ages of $\gs 3$ Gyr, based on detailed spectroscopy in the rest-frame ultraviolet. In order for galaxies to have existed at the high collapse redshifts indicated by these ages, there must be a minimum level of power in the density fluctuation spectrum on galaxy scales. This paper compares the required power with that inferred from other high-redshift populations. If the collapse redshifts for the old red galaxies are in the range $z_c\simeq 6$ -- 8, there is general agreement between the various tracers on the required inhomogeneity on 1-Mpc scales. This level of small-scale power requires the Lyman-limit galaxies to be approximately $\nu\simeq 3.0$ fluctuations, implying a very large bias parameter $b\simeq 6$. The high collapse redshifts of the red galaxies as deduced from gravitational collapse provides independent support for the ages estimated from their stellar populations. Such early-forming galaxies are rare, and their contribution to the cosmological stellar density is consistent with an extrapolation to higher redshifts of the star-formation rate measured at $z<5$; there is no evidence for a general era of spheroid formation at extreme redshifts.

At z=2.39, the cluster around 53W002 is one of the most distant groups or clusters of galaxies known to date. At this redshift the 4000A-break falls between the J and H bands, thus our infrared observations are designed to identify cluster members by a red J-H colour. Out of the 42 objects we have detected in the field, we find ten galaxies with J-H>1.0 and K>18.8, redder and fainter than the radio galaxy, and consistent with the presence of a 4000A-break at the cluster redshift. Two of these reddest galaxies have been confirmed spectroscopically. The colours, sizes and location of these infrared-selected galaxies suggest a cluster much more similar to those nearby than revealed by Hubble Space Telescope observations alone.