The image of a galaxy can be characterized in an entirely objective and non-controversial way by (1) its total integrated magnitude (usually the Holmberg (1958) magnitude based on the total luminosity inside isophotes of 26.5 and 26.0 mag arcsec−2 in the photographic and photo visual regions, respectively), (2) its integrated colors, U–B, B–V, V–R etc., (3) its isophotal radius r, in which r=(a×b)0.5, and where a and b are the semi-major and semi-minor axes to some specific isophote, (4) its effective radius re, defined as the radius within which half of the total galaxy light is emitted in projection, and (5) for many (but not all) galaxies the disk and bulge scale-lengths. More detailed characterizations, such as those provided by the Hubble, de Vaucouleurs and DDO type, are much more difficult (or perhaps impossible) to carry out in an entirely objective fashion.

Inspection of the images of the SRC Southern Sky Survey on IIIaJ emulsion (van den Bergh 1989b) show that it is just barely possible to recognize grand design spiral galaxies at redshifts of up to about 1×104 km s−1. The images of such objects have diameters of ∼0.5 mm (corresponding to 34') and contain ∼1×103 picture elements. Experience shows that the images of galaxies with redshifts of 1000–2000 km s−1, which typically contain 1×104−1×105 picture elements, can be classified with confidence on plates obtained with the SRC Schmidt telescope.

Selection effects

The disks of galaxies on the Hubble sequence Sa–Sb–Sc have central surface brightnesses that appear to fall in a rather narrow range (Freeman 1970). Zwicky (1957, p. 113) and Disney (1976) were among the first to emphasize the fact that this might be the result of a selection effect which is due to the difficulty in discovering galaxies of very low surface brightness. The fact that dwarf spheroidal galaxies, which are now known to be the most common type of extragalactic objects, were not discovered until the 1930s (Shapley 1939) supports this notion. It was originally thought (e.g. van den Bergh 1959) that all galaxies with a low surface brightness were early or late-type dwarfs. However, radial velocity observations by Fisher & Tully (1975) showed that some galaxies with low surface brightnesses are actually quite large and luminous. This effect is clearly shown in Figure 24 which compares the surface brightnesses of the disks of normal and of low-luminosity galaxies. An interesting feature of Figure 24 (see also Figure 1 of Bothun, Impey & McGaugh (1997)), which is presently not well understood, is that disk galaxies with surface brightnesses that are significantly higher than those of normal spirals do not appear to exist. This is shown most clearly in Courteau (1996b), who finds that there is a rather well-defined upper cut-off at a red central surface brightness of ∼17.5 mag arcsec−2.

The Hubble tuning fork diagram

The Hubble classification system recognizes three form families: ellipticals (E), spirals (S) and irregulars (Ir). The ellipticals are assigned an ellipticity ∈ defined as ∈=10(a –b)/a, in which a and b are the major and minor image diameters, respectively. Classification types for ellipticals range from E0, for objects that appear circular in projection, to E7 for the most highly flattened ellipticals. Spiral galaxies occur in two flavors – normal spirals (S), and barred spirals (SB). Within each of these there are three stages: Early-type galaxies of stage Sa/SBa have large nuclei and tightly coiled (and usually rather smooth) arms, objects in stage Sb/SBb have a more open spiral structure, and smaller central bulges. Late-type galaxies in stage Sc/SBc have small nuclear bulges and exhibit wide-open and rather patchy spiral arms. Finally, irregular galaxies have a patchy structure and exhibit no spiral arms. The original Hubble (1926) classification scheme was modified by Hubble (1936) who introduced a class of lenticular (S0) galaxies to span the chasm between spiral and elliptical galaxies (see Figure 2). The Hubble classification system is described and richly illustrated in The Hubble Atlas of Galaxies (Sandage 1961). Classifications for 1246 bright galaxies are given in A Revised Shapley–Ames Catalog of Bright Galaxies by Sandage & Tammann (1981). The Hubble/Sandage classification system reaches its ultimate form in The Carnegie Atlas of Galaxies (Sandage & Bedke 1994).

Central concentration of light

The galaxy classification system proposed by Morgan (1958,1959a), which is sometimes referred to as the Yerkes system, is a one-dimensional scheme based on central concentration of light. Morgan arranged galaxies in a sequence a–f–g–k, with objects of type a having the weakest central concentration of light and those of type k having the strongest central concentration. The Yerkes system also recognizes the intermediates types af, fg and gk. Abraham et al. (1996b) have shown that it is possible to measure a central concentration index C, determined from measurements of the intensity-weighted second-order moments of a galaxy image, which is closely related to the central concentration classification of the Yerkes system.

Galaxies with Morgan type a tend to have early-type (A) spectra, whereas galaxies of type k mostly exhibit late-type (K) integrated spectra. This linkage between morphology and spectral type shows that the dominant stellar population in centrally concentrated galaxies is old, whereas objects with a low central concentration of light tend to have a strong young population component. Probably this correlation between central concentration and integrated spectral type is largely due to the fact that regions of high gas density will usually collapse at earlier times than is the case for low density regions. This is so because the collapse time-scale τ∝(Gρ)−½. As we shall see in Chapter 11 a puzzling exception is provided by the dwarf spheroidal galaxies, most of which are dominated by an old stellar population, even though they are of low density.

Luminosity effects on morphology

From a study of galaxies in the Virgo cluster Holmberg (1958, p. 69) noted that giant galaxies have a higher surface brightness than dwarfs. When the prints of the Palomar Sky Survey first became available in the late 1950s it was immediately obvious that this large, and very uniform, database of galaxy images enabled one to segregate lowluminosity dwarfs from the much more numerous galaxies of average luminosity. Based on inspection of the prints of the Palomar Sky Survey, van den Bergh (1959, 1966) was able to compile catalogs of 243 DDO dwarf galaxies north of δ= −27°. The entries in these catalogs showed that the distribution of such dwarfs on the sky is broadly similar to that of nearby giant galaxies. This conclusion was confirmed for Virgo dwarfs by Reaves (1956, 1967). Furthermore, observations in the Local Group clearly show that dwarfs cluster around giants. Van den Bergh's data also showed that (1) the fraction of all galaxies classified as irregular increases dramatically with decreasing luminosity, and (2) the fraction of all spirals that are barred is much lower among giants than it is among dwarfs. In a subsequent study van den Bergh (1960a,b,c) was able to show that both the surface brightnesses of spiral galaxies and their morphologies are functions of luminosity. Supergiant spirals were found to have long and well-developed ‘grand design’ spiral arms, whereas low-luminosity spirals tend to have poorly developed ‘scraggily’ spiral arms.

Cores and power-law profiles

Lauer et al. (1995) have used the Planetary Camera of the HST to image the central regions of 57 early-type galaxies. They found that the radial surface brightness profiles of most of these fall into two distinct classes: (1) galaxies that have cores, and (2) galaxies that exhibit power-law profiles that continue down to radii near the resolution limit. Of the galaxies observed by Lauer et al. 15 have cores and 30 exhibit power-law profiles. Among the galaxies that have been classified as having either cores or power-law profiles 21 are contained in A Revised Shapley–Ames Catalog of Bright Galaxies (Sandage & Tammann 1981). Since the statistics of objects in this catalog are better understood than those of the entire sample, only the nine galaxies with cores and the 12 having power-law profiles that are in the Shapley–Ames Catalog will be considered below. The most striking feature of these data (which has already been commented on by Lauer et al. and others) is that the galaxies with cores tend to be more luminous than those with power-law profiles. For the Shapley–Ames sub-sample a Kolmogorov–Smirnov test rejects the hypothesis that the galaxies with bulges were drawn from the same luminosity distribution as those having power-law profiles at the 97% confidence level.

Most spiral galaxies are, to a good approximation, oblate spheroids that can be arranged on the sequence Sa–Sb–Sc. However, a significant minority exhibit bar-like structures and may be placed on the SBa–SBb–SBc tine of Hubble's ‘tuning fork’ diagram. Objects, with less pronounced bars, can be arranged on the intermediate sequence S(B)a–S(B)b–S(B)c between normal and barred spirals. Some disks appear to be globally oval (Kormendy 1982, p. 135). Such oval disks are of interest because, like bars, they represent non-axisymmetric distortions of the gravitational potential. An excellent review on the dynamics of barred galaxies has been given by Sellwood & Wilkinson (1993). Infrared imaging shows that some galaxies contain small IR bars (Frogel, Quillen & Pogge 1996). When such small inner bars are found in galaxies with large outer bars there is no correlation between the position angles of the large and small bars.

Bars are important to the dynamical evolution of galaxies (Kormendy 1982, 1993) because they can (a) lose angular momentum to cold disks and dark halos, and (b) gain angular momentum from rapidly rotating bulges. Furthermore, transfer of significant amounts of gas by bars to the nuclear regions of galaxies will increase the central galactic mass concentration, which in turn will make the galactic disk less prone to the development of bar-like distortions. In other words stellar bars can self-destruct (or transform themselves into lenses) by transporting too much gas to the nuclear regions of their parent galaxies.