* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download The age of elliptical galaxies and bulges in a merger model The age
Survey
Document related concepts
Main sequence wikipedia , lookup
Standard solar model wikipedia , lookup
Leibniz Institute for Astrophysics Potsdam wikipedia , lookup
Planetary nebula wikipedia , lookup
Stellar evolution wikipedia , lookup
Outer space wikipedia , lookup
Weakly-interacting massive particles wikipedia , lookup
Astrophysical X-ray source wikipedia , lookup
Non-standard cosmology wikipedia , lookup
Chronology of the universe wikipedia , lookup
Dark matter wikipedia , lookup
Gravitational lens wikipedia , lookup
Weak gravitational lensing wikipedia , lookup
Cosmic distance ladder wikipedia , lookup
Star formation wikipedia , lookup
Transcript
1996MNRAS.281..487K Mon. Not. R. Astron. Soc. 281, 487-492 (1996) The age of elliptical galaxies and bulges in a merger model Guinevere Kauffmann Max-Planck-Institut fUr Extraterrestrische Physik, D-85740 Garching bei Munchen, Germany Accepted 1996 January 29. Received 1995 September 20 ABSTRACT The tightness of the observed colour-magnitude relation for elliptical galaxies has often been cited as an argument against a model in which ellipticals form by the merging of spiral discs. A common view is that merging would mix together stars of disparate ages and produce a large scatter in colour. Here we use semi~analytic models of galaxy formation in a fiat cold dark matter (CDM) universe to derive the distribution of the mean ages and colours of the stars in cluster elliptical galaxies formed by mergers. It is seen that most of the stars in cluster ellipticals form at relatively high redshift (z> 1.9) and that the predicted scatter in the colourmagnitude relation falls within observational bounds. We conclude that the apparent 'homogeneity' in the properties of the stellar, populations of cluster ellipticals is not inconsistent with a merger scenario for the origin of these systems. The analysis is then extended to the stellar populations of field ellipticals and the bulges of spiral galaxies. The mean stellar age of a bright field elliptical is on average 4 Gyr less than that of a cluster elliptical. The ages of bulges are found to correlate strongly with the luminosity of the associated disc. The bulges of late-type spirals are predicted to be older than the bulges of early-type spirals. Finally, we derive predictions for the properties of elliptical galaxies in clusters at high redshift. The number of recently merged objects increases with redshift and this causes the scatter in the colour distributions to become progressively larger. However, an rms scatter in rest-frame U - V colour greater than 0.1 is predicted only at redshifts greater than 1. Key words: galaxies: elliptical and lenticular, cD - galaxies: formation - galaxies: fundamental parameters - galaxies: stellar content. 1 INTRODUCTION There have long been two competing views on the formation history of the elliptical galaxies we see today. One is that most of the stars in present-day galactic bulges and ellipticals were produced during a relatively short, early phase of intense star formation at high redshift. The second view is that elliptical galaxies are relative latecomers, having been produced as a result of the merging of disc galaxies drawn together by gravity as their surrounding dark matter haloes coalesced. In recent years, the second view has come to enjoy increasing popularity. Detailed numerical simulations have shown that mergers between two spiral galaxies of comparable mass lead to the production of spheroidal merger remnants with physical characteristics such as density profiles, gravitational radii, mean velocity dispersions and surface brightnesses that are quite comparable to those of observed ellipticals (see Barnes & Hernquist 1992 for a review). In addition, observational evidence has begun to accumulate which indicates that mergers and interactions are rather common in the Universe, particularly at high redshift. Hubble Space Telescope (HST) images indicate that, at redshifts ~ 0.4, 25-30 per cent of all galaxies show disturbed morphologies (Driver et al. 1995). At lower redshifts, almost all of the most luminous lRAS galaxies have turned out to be interacting systems, leading to the conclusion that the merging process may fuel very intense bursts of star formation (see Joseph 1990 for a review). Schweizer & Seitzer (1992) present evidence indicating that the UBV colours of elliptical galaxies become systematically bluer as the amount of fine structure in these galaxies increases. If © 1996 RAS © Royal Astronomical Society • Provided by the NASA Astrophysics Data System 1996MNRAS.281..487K \ 488 G. Kauffmann fine structure is indeed a measure of the dynamical youth of a galaxy, these observations suggest that many present-day ellipticals have undergone a merger-induced starburst in the past. Likewise, analyses of the stellar ages of elliptical galaxies using absorption-line index strengths show that the outer parts of elliptical galaxies are both older and more metal-poor than the nuclei, consistent with the merger/starburst picture (Faber et al. 1995). The fact remains, however, that the stellar populations of present-day elliptical galaxies appear to be rather homogeneous. The colours, mass-to-light ratios and Mg II absorption strengths of ellipticals are correlated only with their central velocity dispersions, and the scatter about these relations appears to be surprisingly small. Bower, Lucey & Ellis (1992) show that the intrinsic scatter about the U - V and V - K colour-magnitude relations for early-type galaxies in the Coma and Virgo clusters is less than 0.05 mag. This leads to the conclusion that elliptical galaxies must have formed the bulk of their stars at redshifts greater than 2, or else the observed homogeneity in the colours would require a very precise synchronization of formation epochs and star formation histories for these galaxies. Likewise, Bender, Burstein & Faber (1993) find a very tight relationship between the strength of the Mg II index at the centre of ellipticals and their central velocity dispersions, (1. Galaxies show a mean scatter of only 0.025 mag in Mg II, versus a full range of nearly 0.35 mag in Mg II. This observed scatter sets an upper limit of 15 per cent on the rms variation of both age and metallicity at fixed (1 for bright ellipticals. The tightness of these limits calls into question the tenability of the merger model, as merging might be expected to mix together stellar populations of quite disparate ages (Renzini 1995). In addition, a starburst accompanying the merging event would increase the amount of light contributed by young stars. However, detailed predictions of the resulting scatter in quantities such as colours and Mg II strengths have not yet been made for any realistic theory of elliptical galaxy formation via merging. In this paper, we present results calculated for a b = 1.5 COM universe using semi-analytic techniques. In this model, elliptical galaxies are produced when two progenitor galaxies of roughly equal mass merge with each other. Spiral galaxies are formed in a two-step process: two galaxies of equal mass must first merge with each other to form the bulge component, then gas from the surrounding dark matter halo cools and settles to form a new disc. It has already been demonstrated in Kauffmann (1995) that this model provides a good fit both to the properties of cluster ellipticals seen today, and to the evolution in the colours of cluster galaxies at higher redshift. Models of this type also naturally result in a higher ratio of elliptical to spiral galaxies in clusters than in the field (Kauffmann, White & Guiderdoni 1993, hereafter KWG). This is because the dark matter haloes of cluster galaxies are assumed to be disrupted and stripped by tidal heating and encounters with other galaxies in the cluster potential. As a result, gas can no longer accrete and form a new disc. The results of the analysis in this paper show that, although elliptical formation is far from a synchronized process in this model, merging in clusters does in fact take place at early enough epochs for the scatter in the colours and mean stellar ages to be in good agreement with observations. We also extend the analysis to the stellar populations of field elliptical galaxies and the bulges of spirals. The models predict that bright elliptical galaxies in low-density environments are formed by much more recent merging events than cluster ellipticals and consequently contain younger stars. The colours of field ellipticals are bluer on average and show more scatter. The models also predict that the ages of spiral bulges should correlate strongly with the luminosity of the associated disc. Since discs are accreted gradually over a Hubble time, late-type spirals must form their bulges at higher redshift than earlier types. It is possible that this correlation may be hidden if substantial inflow from the disc to the bulge is able to take place, for example as a result of bar formation. Finally, we make some predictions for the scatter in the colours of cluster elliptical galaxies at high redshift. Morphological classification of galaxies at z ~ 0.6-0.7 is now possible using HST, and the hope is that the colours of galaxies observed at an epoch when the Universe was a fraction of its present age will place strong constraints on cosmological models. 2 SEMI-ANALYTIC MODELS OF ELLIPTICAL GALAXY FORMATION The semi-analytic models we employ are described in detail in KWG and Kauffmann & White (1993). Application of the model to the evolution of the galaxy population in clusters at high redshift is discussed in Kauffmann (1995). Here we present a brief summary of that paper. An algorithm based on an extension of the PressSchechter theory due to Bower (1991) and Bond et al. (1991) is used to generate Monte Carlo realizations of the merging paths of dark matter haloes from high redshift until the present. This algorithm allows all the progenitors of a present-day object, such as a cluster, to be traced back to arbitrarily early times. Dark matter haloes are modelled as truncated isothermal spheres and it is assumed that, as the halo forms, the gas relaxes to a distribution that exactly parallels that of the dark matter. Gas then cools and condenses on to a central galaxy at the core of each halo. Star formation and feedback processes take place as described in KWG. In practice, star formation in central galaxies takes place at a roughly constant rate of a few solar masses per year, in agreement with the rates derived by Kennicutt (1983) for normal spiral galaxies. At a subsequent redshift, a halo will have merged with a number of others, forming a new halo of larger mass. All gas that has not already cooled is assumed to be shock heated to the virial temperature of this new halo. This hot gas then cools on to the central galaxy of the new halo, which is identified with the central galaxy of its largest progenitor. The central galaxies of the other progenitors become satellite galaxies, which are able to merge with the central galaxy on a dynamical friction time-scale. If a merger takes place between two galaxies of roughly comparable mass, the merger remnant is labelled as an 'elliptical' and all cold gas is transformed instantaneously into stars in a 'starburst'. Note that the infall of new gas on to satellite galaxies is not allowed, and star formation will continue in such objects only until their existing cold gas reservoirs are exhausted. Thus the epoch at which a galaxy is accreted by a larger halo © 1996 RAS, MNRAS 281, 487-492 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System 1996MNRAS.281..487K Ages of elliptical galaxies in a merger model 489 delineates the transition between active star formation in the galaxy and passive evolution of its stellar population. The elliptical merger remnants in clusters hence redden as their stellar populations age. Central galaxy merger remnants in the 'field' are able to accrete new gas in the form of a disc to form a 'spiral' galaxy consisting of both a spheroidal bulge and a disc component. As demonstrated in KWG, this model is able to account for the observed numbers and luminosity distributions of galaxies of different morphologies, both in clusters and in lower-density environments. The spectrophotometric models of Bruzual & Charlot (1993) are used to translate the predictions of the models into observed quantities such as magnitudes and colours, which may be compared directly with the observational data. In the present paper, we focus on the ages and colours of the stars of the merger remnants as predicted by the model. The cosmological initial conditions are a b = 1.5 CDM universe with 0=1 and Ho=50 km S-I Mpc- I . Although this model does not match the COBE measurements of the amplitude of the microwave background fluctuations on degree scales, it has become the de facto standard model used in galaxy formation studies, since it provides a reasonable fit to the observed small-scale clustering properties of galaxies. 3 THE AGES AND COLOURS OF ELLIPTICAL GALAXIES Fig. 1 shows the V luminosity-weighted mean stellar age of the galaxies in a cluster of 1015 Mo' The symbols in the plot represent the morphological type of the galaxy: large filled circles correspond to ellipticals, open squares to SOs, small filled circles to spirals and pinwheels to ellipticals that have been formed by a recent merging event ( < 1.5 Gyr ago). The classification into morphological type is based on the B- 12 f- - • e e 10 f- S. III • • a., • rP • a e e e a a· a ~ 8 r- 0 QO _. III -~ r - 8- a - -24 - .. -22 -18 -20 M(V) • . . . .... .. . ..... . -...' . _••• b 4 r- 10 - 6 - a 6 r- - a - -fa:a "'a e , a .. a' a a .J" 'i:' ... ~ e\~~~ • 8 12 _ (a) ; .~ ·.1 •••.1. e • CI) III ~ band disc-to-bulge ratio of the galaxy, as given in table 3B of Simien & de Vaucouleurs (1986). As can be seen, the stellar populations of spiral galaxies span a very wide range in age, but those of early-type galaxies are confined to ages between 8 and 12.5 Gyr. For our adopted cosmological parameters, this means that the bulk of stars in elliptical and SO galaxies were formed at redshifts exceeding 1.9. The age distribution of early-type galaxies found in haloes of mass 1013 Mo is plotted in Fig. 2(a). These galaxies either occur in small groups, or have no companion of comparable luminosity, and thus can be regarded as 'field' ellipticals. In order to get a sizeable sample of these objects, it was necessary to combine the populations of 100 such low-mass haloes. As can be seen, the field elliptical population splits up into two distinct classes. The faint field ellipticals have stellar ages comparable to those of early-type galaxies in clusters. There is also, however, a population of bright field ellipticals that have much younger stellar populations (mean ages between 5 and 8 Gyr). These bright galaxies are all central galaxies and thus are still in the process of accreting cold gas. They are also mostly recent mergers. The fact that we see these galaxies as ellipticals is simply because they have not yet had the time to accrete a new disc following their last big merging event. Fig. 2(b) shows the age distribution of early-type galaxies in massive groups (1014 Mo)' There is some tendency for the most luminous group members to be recent mergers, but the distribution as a whole is rather similar to that of the cluster ellipticals in Fig. 1. 'i:' III ~ - 12 ':.. (b) - 10 - - o ; 8- - a a 2 -24 J I -22 -20 -18 M(V) Figure 1. The V-luminosity weighted mean stellar ages of the galaxies in a cluster of mass 1015 Mo' Large filled circles represent ellipticals, open squares are SOs, small filled circles are spirals and three-pronged pinwheels are elliptical galaxies that were formed by a major merger in the past 1.5 Gyr. - 6 -24 -22 -20 M(V) -18 Figure 2. The mean ages of the early-type galaxies in (a) haloes of mass lO13Mo and (b) haloes of mass 1014 M o ' Symbols are as in Fig. 1. © 1996 RAS, MNRAS 281, 487-492 © Royal Astronomical Society e Provided by the NASA Astrophysics Data System 1996MNRAS.281..487K G. Kauffmann 490 1 (a) 0.8 Elliptical • .. ~ 0.6 r:: 0.4 II e • III 0. 0.2 • • • ..... ..r 0.6 • r:: : • :., ..... 0 0 l) III 0. • 0.2 5 age (Gyear) ... I... . o 0 o· a a. a .1tI a 'll 0 _ 0.,,1Pg • o 0 r9 00 0 1 - • ~ • • Da D a a - I- -22 -20 -18 M(V) • =-=I - Fig. 3 is a histogram of the fraction of the total V-band light of a cluster elliptical or spiral galaxy contribution by stars that form at a given redshift. The solid line represents the average contribution, while the filled squares and circles represent the maximum and minimum contributions respectively. The difference between the average star formation history of a cluster elliptical and that of a spiral is illustrated clearly in this diagram. A large fraction of the Vband light of spirals comes from stars that formed in the past 5 Gyr, whereas on average this contribution is negligible in ellipticals. It should be noted, however, that a small subset of elliptical galaxies does exhibit substantial star formation at late epochs. The dotted line in the top panels in Fig. 3 illustrates the epoch at which elliptical galaxies typically underwent their last major merging event. As can be seen, most ellipticals were formed by a merger that occurred between 5 and 10 Gyr ago. It is also apparent from this diagram that a substantial fraction of the stars in an elliptical galaxy may have been formed before the merging event that actually caused the transformation in morphology. Star formation may take place at a constant rate in the two spiral progenitors of an elliptical for several gigayears before they are able to merge. It is thus important to distinguish exactly what is meant by 3.2 - 3 -• - 2.6 - 0 0 Figure 3. The fraction of the total V-band light of cluster elliptical or spiral galaxies contributed by stars that form at a given epoch. The solid line is the average star formation history. The squares and circles represent, respectively, the maximum and minimum contributions to the total light for any galaxy at that epoch. The dotted line is a histogram of the percentage of elliptical galaxies that experienced their last major merger at a given epoch. - s.. 2.8 :s l) 10 3.4 > - • 00 1-• -24 10 • e0.4 r :s .1 ••••" .... (b) III I- ~ 1 Spiral 1.6 - (a) s.. 1.2 l- 5 age (Gyear) 0.8 I- >I 1.4 • 111,""', 00 - 1.8 - (b) 0 .'W •.0 eg.,.. i .. o. o· • - ';1- 1II~.,i 00 0 oOo.••~ro o y _ a Daa -24 0 -22 -20 0 -18 M(V) Figure 4. The distribution in U - Vand V - K colour of early-type galaxies in a cluster of 1015 M 0 • Filled circles are ellipticals and open squares are SOs. 'the formation epoch of ellipticals'. It can be defined either as the epoch when the bulk of the stars were formed or as the epoch when the merging event took place. Finally, in Fig. 4, we have used the evolutionary synthesis models of Bruzual & Charlot (1993) to generate a scatterplot of the U - Vand V - K colours of early-type galaxies in clusters. In U - V, the average colour of an elliptical is 1.37 with an rms standard deviation, (1, of 0.034. In U - K, the mean is 3.01 and (1 is 0.045. The rms is calculated using the 'median absolute difference' technique so that the results may be compared directly with the data of Bower et al. (1992). If SO galaxies are, included, the rms increases to ~0.07. Although the mean colours are roughly comparable to the average colours in the data of Bower et al., there is no correlation between the colour and the magnitude of the galaxy, in contradiction with the observations, which show strong reddening in the stellar populations of elliptical galaxies with increasing luminosity. As shown in Fig. 1, cluster elliptical galaxies of all luminosities have roughly the same mean stellar age, so any trend in colour would have to result from metallicity differences between galaxies of different masses. We have neglected chemical enrichment in this analysis. The rms scatter in the colour of the model ellipticals is well within the upper limit for the intrinsic scatter in the colour of Coma and Virgo ellipticals quoted by Bower et al. (1992). The inclusion of the SO popUlation increases the scatter by a factor of 2. It is interesting that Bower et al. also found that the removal of SO galaxies © 1996 RAS, MNRAS 281, 487-492 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System 1996MNRAS.281..487K Ages of elliptical galaxies in a merger model 491 caused the scatter to decrease in their data, but not by as much as a factor of 2. However, it is clear that, in the model, the scatter quoted for the early-type population will depend on the precise value of the disc-to-bulge ratio at which a galaxy is labelled as being 'early-type'. 4 THE BULGES OF SPIRAL GALAXIES IN THE FIELD As has been discussed, central galaxy merger remnants in the field are able to accrete new gas, form a disc and become a 'spiral' galaxy consisting of both a bulge component and a disc component. Fig. 5 shows the distribution of bright field galaxies as a function of M(bulge) -M(tot), the difference between the B-band magnitude of the bulge and the total Bband magnitude of the galaxy. Early-type galaxies have M (bulge) - M (tot) < 1 (Simien & de Vaucouleurs 1986). In Fig. 6, we plot the V-luminosity weighted mean stellar ages of the bulges as a function ofM(bulge) -M(tot). There is a 0.5 0.4 0.3 CD 0.2 bO .... III ~ o s... 0.1 CD 0. " 2 3 5 4 M(bulge)-M(tot) Figure 5. The distribution of 'field' galaxies as a function of the difference in the B-band magnitude of the bulge and the B-band magnitude of the whole galaxy. -t; s... 12 - ...... ... f- -< 10 r- II) .c ~: 8 f- • ~ 8 .E I • .. ..,..•'. ...• . •••• • •• .. • • ".... ....CDbO '.): •• •• ., ........... '. . .• • t' -.tl:. . · •• .1 • • ~ r:.::I C , ..-. .-.. : • I I 6 f-. > o . • • :. • ,• • • - • • - • I I 1 2 strong correlation between bulge age and bulge-to-disc ratio. As was already noted, early-type galaxies in the field have young bulges, with ages ranging between 5 and 8 Gyr. Late-type galaxies have bulges as old as 12 Gyr. It would be extremely interesting to investigate whether correlations of this type are found for real spiral galaxies. If not, it may be an indication that bulge formation and disc formation do not follow the simple two-step process assumed here. Indeed, there are indications that bar instabilities in discs may fuel nuclear starbursts in many galaxies (see for example papers by Kenney and collaborators). If bar-driven star formation accounts for a substantial fraction of the total mass of stars in the bulge, the correlation shown in Fig. 5 may be completely wrong. 5 ELLIPTICALS IN CLUSTERS AT HIGH RED SHIFT Present-day elliptical galaxies contain the oldest stars in the Universe. The repair of HST permits the direct identification of elliptical and SO galaxies at significant cosmic depth. The hope is, therefore, that studies of the stellar populations of ellipticals at early epochs will place strong constraints on cosmology, in much the same way that studies of the ages of globular clusters have ruled out high values of the Hubble constant. Ellis (1995) has measured the rest-frame U - V colours of elliptical galaxies in the cluster 0016 + 16 at z = 0.54 and finds an rms scatter of 0.07. In Fig. 7, model predictions for the U - V colours of cluster ellipticals are shown ata series of redshifts. Recent mergers are distinguished from 'ordinary' ellipticals by large open circles in the plot. There are two trends that should be noted. First, the 'red envelope' of the distribution shifts bluewards with increasing redshift. The magnitude of this shift is consistent with what one would expect from the passive evolution of a 12-Gyr-old burst, as given by the models of Bruzual & Charlot (1993). Secondly, the scatter in the colour distribution increases with redshift and most of this increase arises from a growing population of galaxies which have merged in the 1.5 Gyr prior to the epoch of observation. Note, however, that the scatter is not dramatically large until the redshifts are in excess of 1. This may be understood as follows. As pointed out by Charlot & Silk (1994), the colours of stars formed in a short burst evolve very rapidly for the first 3-4 Gyr. After that, the evolution slows down and the colours redden only gradually over time. So long as most of the stars in the elliptical population of a cluster are formed 3-4 Gyr prior to the time the cluster is observed, no great conspiracy is required to produce a small scatter in the colours. At redshifts greater than 1, this is no longer possible (assuming Q = 1 and Ho = 50 Ian S-1 Mpc- 1) and the scatter becomes very large. Spectroscopic age indicators may turn out to be more sensitive tests of cosmology. 6 DISCUSSION AND CONCLUSIONS 3 M(bulge)-M(tot) Figure 6. The V-luminosity weighted mean stellar age of the bulge component of a field galaxy is plotted as a function of the magnitude difference between the bulge and the galaxy as a whole. Two major conclusions follow from this analysis. One is that star formation in cluster ellipticals formed by mergers occurs at high enough redshifts for the predicted scatter in the colour-magnitude relation of these systems to fall within observational bounds. That ellipticals in clusters © 1996 RAS, MNRAS 281, 487-492 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System 1996MNRAS.281..487K 492 G.}(au~ann I 1.5 I I (a) z-O.6 >-I - o 1 •• ::> ~ o '0 u - 0.5 I I -22 -20 I I -18 M(B) 1.5 r- . I I z=O.9 - 1 00 f- >I I (b)- .. ~e,cxo o~ •.. o 0 ctJOO ::> elliptical galaxies. What this paper has demonstrated is that this process can work in detail, and leads to a population of cluster ellipticals that matches the observations rather well. The second major conclusion is that bright elliptical galaxies in the field are intrinsically different from cluster ellipticals. These galaxies have undergone recent merging and star formation. It is only because we 'catch them at the right time' that we observe them as elliptical systems. In a few gigayears, they will have grown new discs and turned back into spirals. There are some observational indications that field ellipticals may be younger (De Carvalho & Djorgovski 1992), but these need to be studied in more detail. We also predict that the ages of the bulges of spiral galaxies in the· field should correlate with the luminosities of the associated discs. Galaxies with higher bulge-to-disc ratios should have younger bulges. If this correlation is not found, it may indicate that bulge formation is an ongoing process. Finally, we derive predictions for the colour distributions of elliptical galaxies in clusters at high redshift. The scatter is predicted to increase very substantially at redshifts greater than 1. ~ 0.5 0 0 ACKNOWLEDGMENTS U I -22 I I I -20 -18 M(B) I thank Ralph Bender and Steve Zepf for helpful discussions and Richard Ellis for pointing out the high-redshift data. REFERENCES 1.5 I • I (c) z=1.5 >-I 1 ::> ~ 0.5 .9 o u -22 -20 -18 M(B) Figure 7. Rest-frame U - V colour-magnitude relations for elliptical galaxies in clusters of 10'5 M0 observed at z=0.6, 0.9 and 1.5. Large open circles denote ellipticals that have been formed by a major merging event in the past 1.5 Gyr. form at high redshift should come as no surprise. For a random Gaussian initial density field, it is well-known that the redshift of collapse of density peaks on galaxy scales is boosted by the presence of the surrounding, large-scale overdensity destined to collapse to form the cluster at z = 0 (Bardeen et al. 1986). Galaxies forming in peaks collapsing at z > 2 quickly merge together within groups and their star formation is truncated when their gas is used up. Thus, in the hierarchical clustering picture, galaxies that form at high redshifts are in some sense pre-selected to end up as cluster Bardeen J. M., Bond J. R., Kaiser N., Szalay AS., 1986, ApJ, 304, 15 Barnes J. E., Hernquist L., 1992, ARA&A, 30, 705 Bender R., Burstein D., Faber S. M., 1993, ApJ, 411, 153 Bond J. R., Cole S., Efstathiou G., Kaiser N., 1991, ApJ, 379, 440 Bower R., 1991, MNRAS, 248, 332 Bower R., Lucey J. R., Ellis R. S., 1992, MNRAS, 254, 601 Bruzual G., Charlot S., 1993, ApJ, 405, 538 Charlot S., Silk J., 1994, ApJ, 432, 453 De Carvalho R. R., Djorgovski S., 1992, ApJ, 389, 149 Driver S. P., Windhorst R. A, Ostrander E. J., Keel W. C., Griffiths R. E., Ratnatunga K. u., 1995, ApJ, 449, L23 Ellis R. S., 1996, in Ostriker J. P., Bahcall J. N., eds, Unsolved Problems in Astrophysics. Princeton Univ. Press, Princeton NJ, in press Faber S. M., Trager S. c., Gonzalez J. J., Worthey G., 1995, in Van der Kruit P. G., Gilmore G., eds, Proc. IAU Symp. 164, Stellar Populations. K1uwer, Dordrecht, p.249 Joseph R. D., 1990, in Wielen R., ed., Dynamics and Interactions of Galaxies. Springer, New York, p.132 Kauffmann G., 1995, MNRAS, 274, 161 Kauffmann G., 1996, MNRAS, 281, 673 (this issue) Kauffmann G., White S. D. M., 1993, MNRAS, 261, 921 Kauffmann G., White S. D. M., Guiderdoni B., 1993, MNRAS, 264, 201 (KWG) Kennicutt R. C., 1983, ApJ, 272, 54 Renzini A, 1995, in Van der Kruit P. G., Gilmore G., eds, Proc. IAU Symp. 164, Stellar Populations. K1uwer, Dordrecht, p.325 Schweizer F., Seitzer P., 1992, AJ, 104, 1039 Simien F., de Vaucouleurs G., 1986, ApJ, 302, 564 © 1996 RAS, MNRAS 281, 487-492 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System