Next Contents Previous


An important goal of extragalactic astronomy is to understand the structure of galaxies as encoded, e. g., in the classification schemes of Hubble (1936), de Vaucouleurs (1959) and Sandage (1961, 1975). An updated version of Hubble classification is shown in Fig. 4, and a perpendicular cut through de Vaucouleurs' classification volume at stage Sb is shown in Fig. 5. For each kind of galaxy, we want to understand the present structure and the formation and evolution processes which created that structure.

Figure 4

Figure 4. Revised Hubble classification (Kormendy & Bender 1996) recognizing the difference (the "E – E dichotomy") between boxy, slowly rotating ellipticals that have cores and disky-distorted, more rapidly rotating and coreless ellipticals that are continuous in their properties with S0s. Ellipticals are understood to form via rapid processes; i. e., by major mergers and violent relaxation. These are reviewed in Section 8, where I discuss galaxy formation by hierarchical clustering. S0 and spheroidal galaxies (not included in Hubble types) form – I suggest – at least mostly by environmental secular evolution (Section 7). Spiral galaxies evolve partly by rapid processes (mergers make classical bulges, shown orange in the figure). However, the evolution of many spirals has been dominated by secular processes, including the formation of pseudobulges (also shown in orange, because Hubble classes do not distinguish between classical and pseudo bulges).

Our picture of galaxy formation by hierarchical clustering and our School's subject of secular evolution are both important in creating the structures in Figs. 4 and 5. Ellipticals and classical (elliptical-galaxy-like) bulges of disk galaxies form by hierarchical clustering and major mergers. I review this in Section 8, and Isaac Shlosman discusses it in his lectures. Bulges are usually classical in S0 – Sb galaxies. Secularly built pseudobulges are not distinguished in Hubble classification, but they are already important in some S0s, and they become the usual central component at type Sbc and in later types (Kormendy & Kennicutt 2004). Unlike these permanent components, spiral density waves are temporary features of disks that mostly last only as long as they have driving engines (Toomre 1977b, 1981).

3.1. Inner rings "(r)" and outer rings "(R)"

Figure 4 omits several regular features in galaxy structure that we need to understand. The most important of these are encoded in de Vaucouleurs' (1959) more detailed classification, illustrated here in Fig. 5. Rotate the "tuning fork" diagram in Fig. 4 about the E – S0 axis. In the plane of the page, de Vaucouleurs retains the distinction between unbarred ("SA") and barred ("SB") galaxies. The new, third dimension above and below the page is used for the distinction between galaxies that do or do not contain inner rings. Fig. 5 is a perpendicular cut through this classification volume.

Figure 5

Figure 5. Perpendicular cut at type Sb through the galaxy classification volume as defined by de Vaucouleurs (1959). I am grateful to Ron Buta for this version from the de Vaucouleurs Atlas of Galaxies (Buta et al. 2007).

Figure 6 (top) illustrates the distinction (bottom half of Fig. 5) between SB(s) galaxies in which two global spiral arms begin at the ends of the bar and SB(r) galaxies in which there is a ring of stars and gas surrounding the ends of the bar and the spiral arms start somewhere on this ring. In Figures 5 and 6, there is a prominent, nearly straight dust lane on the rotationally leading side of the bar in SB(s) galaxies. Notwithstanding Fig. 5, there is generally no such dust lane in SB(r) bars (Sandage 1975). Note in NGC 2523 that the inner ring is patchy and blue and contains young stars, like the outer disk but unlike the relatively "red and dead" bar and (pseudo)bulge. This is a common property of inner rings.

Figure 6

Figure 6. (top) Contrast SB(s) and SB(r) structure: (left) SB(s)bc galaxy NGC 1300 has two global spiral arms that begin at the ends of the bar. (right) The SB(r)bc galaxy NGC 2523 has a somewhat more multi-armed but still global spiral pattern with arms that begin from the inner ring, not necessarily at the end of the bar. (bottom) Galaxies with outer rings: (left) the (R)SB(lens)0/a galaxy NGC 1291 and (right) the (R)SA(lens)ab galaxy NGC 4736. The latter is a prototypical oval galaxy. It contains a pseudobulge that is identified by no less than five classification criteria from Section 5.3. NGC 2523 is a Hubble Heritage image; NGC 2523 is from Adam Block (NOAO/AURA/NSF); NGC 1291 and NGC 4736 are from Kormendy & Kennicutt (2004).

Figure 6 (bottom) shows another feature that is recognized by de Vaucouleurs and Sandage although it is not given a separate classification bin in diagrams like Fig. 5. Both barred and unbarred galaxies can contain outer rings "(R)" at approximately twice the radius of the inner disk.

The properties of inner and outer rings are discussed (for example) in Kormendy (1979b, 1981, 1982b), Buta (1990, 1995, 2011, 2012), Buta & Crocker (1991), Buta & Combes (1996) and Buta et al. (2007). I particularly use the following.

  1. Inner rings essentially always encircle the end of the bar.
  2. Outer rings typically have radii ~ 1.2 times the radius of the bar. There is no overlap in the distribution of radii of inner and outer rings.
  3. Both inner and outer rings typically contain H i gas and star formation.
  4. Inner rings typically have intrinsic axial ratios of ~ 0.85 ± 0.10 with the long axis parallel to the bar (Buta 1995).
  5. Outer rings typically have intrinsic axial ratios of ~ 0.87 ± 0.14. In most galaxies, the long axis is perpendicular to the bar (as seen face-on), but a few outer rings are elongated parallel to the bar (Kormendy 1979b; Buta 1995).
  6. Figure 5 contains one mistake. There is no continuity between the inner rings in unbarred and barred galaxies. Instead, SA inner rings are similar to nuclear star-forming rings in SB galaxies (see Fig. 31 in Section 5.1). Nuclear rings are always much smaller than inner rings (Comerón et al. 2010).

All of the above structural properties are natural products of the secular evolution that I review in these lectures.

3.2. Lens Components "(lens)"

Lenses are the final morphological component that we need to recognize observationally and to understand within our picture of galaxy evolution. There is much confusion in the literature about lenses. It is unfortunate that de Vaucouleurs chose to use the name "lenticular" for S0 galaxies. I empathize: a bulge-dominated, edge-on S0 such as NGC 3115 or NGC 5866 is indeed shaped rather like an edge-on glass lens. But this is due to the combination of a thick bulge and a thin disk; it does not happen because these S0s contain lens components. In fact, most unbarred S0s do not contain lens components, whereas some Sa and Sb galaxies – particularly barred ones – do contain lens components. We need to overcome the confusion that results from use of the misleading name "lenticular galaxy". In all of my papers including this one, I avoid use of the name "lenticular galaxy" and always call these objects "S0 galaxies". And I try to identify lens components strictly using the following definition.

A lens component is an elliptical "shelf" in the brightness distribution of a disk galaxy. That is, it has a shallow brightness gradient interior to a sharp outer edge. Typical face-on lenses have axial ratios of ~ 0.85 – 0.9. Lenses occur most commonly in barred galaxies; then the bar almost always fills the lens in one dimension. NGC 1291 in Fig. 6 is an (R)SB(lens)0/a example of this configuration. Lenses can also occur in unbarred galaxies; the prototypical example is the SA(lens)0 galaxy NGC 1553 (Fig. 7). It is important to note that there is no overlap in the intrinsic ellipticity distributions of bars and lenses. Bars always have much smaller (pole-on) axial ratios of ltapprox 1/4. So bars and lenses are clearly different, even though both occur only in disk galaxies. A lens and an inner ring of the same size can occur together (e. g., NGC 3081). The properties of lenses are established in Kormendy (1979b) and in Buta (2011, 2012); Buta et al. (1996, 2007).

Figure 7

Figure 7. (left) Color JHK image of the SA(lens)0 galaxy NGC 1553 from the 2MASS Large Galaxy Atlas (Jarrett et al. 2003; image from NED). The lens is the elliptical shelf in the brightness distribution interior to the outer, exponential disk. It is best seen in the V-band, major-axis brightness profile shown in the right panel (Freeman 1975). The straight line is an exponential fit to the disk profile.

I will argue in Sections that lenses are defunct bars. That is, following Kormendy (1979b, 1981, 1982b), I suggest that bars evolve away into lenses when the secular evolution that they drive increases the central mass concentration so much that the elongated bar orbits can no longer precess together at the same bar pattern speed Ωp.

I use the notation SA(lens) and SB(lens), respectively, for unbarred and barred galaxies with lenses. Buta et al. (2007) uses the letter "el"; I do not do this only because it is difficult to distinguish "el" from the number 1.

It will turn out that lenses, outer rings, inner rings and pseudobulges are like the more familiar classical bulges, bars and disks in the sense that they are distinct components in the structure of galaxies that retain their essential character over many galaxy rotations. They are different from spiral arms, which are density waves in the disk that last – more or less – only as long as the "engine" that drives them. Still, there are many ways in which the above components can be combined into a galaxy. This results in de Vaucouleurs classifications that may be as complicated as (R)SB(r)bc or (R)SB(lens)0/a (NGC 1291 in Fig. 6). This complication makes many people uneasy. However, I emphasize: Every letter in such a complicated galaxy classification now has a physical explanation within our picture of secular galaxy evolution. It is a resounding testament to the educated "eyes" of classifiers such as Allan Sandage and Gerard de Vaucouleurs that they could pick out regularities in galaxy structure that ultimately were understandable in terms of galaxy physics.

3.3. Oval Disks

Fundamental to our understanding of the engines that drive secular evolution is the observation that many unbarred galaxies nevertheless are not globally axisymmetric. The inner disks of these galaxies typically have intrinsic axial ratios of ~ 0.85. Bars are more elongated (typical axial ratio ~ 0.2), but a much smaller fraction (typically ltapprox 1/3) of the disk participates. Observations indicate that oval disks are approximately as effective in driving secular evolution as are bars.

Oval galaxies can be recognized independently by photometric criteria (Kormendy & Norman 1979; Kormendy 1982b) and by kinematic criteria (Bosma 1981a, b). They are illustrated schematically and with observations of prototypical galaxies in Fig. 8. The photometric criteria are:

  1. The disk consists of two nested ovals, each with a shallow brightness gradient interior to a sharp outer edge.
  2. The outer oval has a much lower surface brightness than the inner oval. It can consist of a featureless disk (e. g., in an S0 galaxy) or spiral arms that make a pseudo (≡ not quite closed) outer ring or a complete outer ring.
  3. The nested ovals generally have different axial ratios and position angles, so they must be oval if they are coplanar. But the flatness of edge-on galaxies at these fairly high surface brightnesses shows that these disks are oval.

NGC 4736 and NGC 4151 are outer ring and outer spiral arm versions of oval galaxies (Fig. 8). I use the notation SA(oval) or SB(oval) – exactly analogous to SA(lens) or SB(lens) – for galaxies with oval disks.

Figure 8

Figure 8. Criteria for recognizing unbarred oval galaxies shown schematically (top) and with observations of NGC 4736 and NGC 4151 (adapted from Fig. 9 in Kormendy & Kennicutt 2004). The image of NGC 4736 is from my IIIa-J plate taken with the 1.9 m Palomar Schmidt telescope. The image of NGC 4151 is from The H i velocity contours of NGC 4736 and NGC 4151 are from Bosma et al. (1977a) and from Bosma et al. (1977b), respectively.

The kinematic criteria for recognizing ovals are that the velocity field is symmetric and regular, but:

  1. the kinematic major axis twists with radius;
  2. the optical and kinematic major axes are different; and
  3. the kinematic major and minor axes are not perpendicular.

Here observation (a) is equally consistent with a warp. But (b) and (c) point uniquely to an oval velocity field that is seen at a skew orientation. Given this conclusion, observation (a) then means that the intrinsic orientations of the outer and inner ovals are different. Usually, the outer oval is elongated perpendicular to the inner one, as seen face-on. My favorite poster child for secular evolution, NGC 4736, shows all of these effects. The classical Seyfert 1 galaxy NGC 4151 shows them even more clearly (Fig. 8).

Figure 9

Figure 9. Sequence of outer-ring galaxies that are structurally similar except that a bar embedded in the lens or inner oval structure ranges from strong at upper-left to invisible at lower-right. All images are from

Figure 9 shows four more oval galaxies. I use it to emphasize four points.

  1. Oval galaxies are very recognizable and very common. It is curious that they have not better penetrated the astronomical folklore. Kormendy & Kennicutt (2004) show that they are important engines for secular evolution.

  2. There is a complete structural continuity in outer ring galaxies from (R)SB objects that have strong bars to (R)SA objects that are completely barless.2 NGC 2859 is one of the best examples of an (R)SB(lens)0 galaxy in which the bar fills the lens in one dimension. Another excellent example is NGC 3945 (Kormendy 1979b; Fig. 17 and Fig. 24 here). NGC 3504 and NGC 3368 are two examples among many of galaxies that have well defined oval disks and extremely weak bars. I have typed them as (r,oval) and (s,oval) because there is a bright blue rim of star formation around the outside of the inner oval, exactly as in many (r,lens) galaxies. I am not aware of any measurements of the bar strength that take into account the oval structure. But the bars that are revealed by the 2MASS survey at K' band must be only a few percent of the disk light. NGC 1068 is classified as unbarred by Sandage (1961) and by Buta et al. (2007) there is a hint of a bar in the 2MASS image (Jarrett et al. 2003). NGC 4736 is a purely unbarred, oval galaxy. 2

  3. Albert Bosma has advocated for many years that (lens) and (oval) components are the early-Hubble-type and later-Hubble-type versions of the same kind of structure. For many years, I was unsure about whether or not to believe this. I am now convinced that Albert is exactly correct. The reasons are discussed in Sections Here, I juxtapose (R)SB(lens) and (R)SAB(oval) galaxies in Fig. 9 to emphasize their similarity. I will continue to use this notation throughout this review, but I want to make it clear that I believe that (lens) and (oval) structures are fundamentally the same.

  4. There also is a continuity from complete outer rings (R) such as the one in NGC 2859 through spiral arms that are distorted until they almost closed to form what is called a "pseudoring" (R') as in NGC 3504 through to spiral arms that are less – but similarly – distorted as in NGC 4151 (Fig. 8 here) and NGC 1300 (compare Fig. 6 here with Fig. 13 in Kormendy 1979b). See de Vaucouleurs (1959); de Vaucouleurs et al. (1991: RC3) and Buta (2011, 2012) for further discussion.

All of these continuities prove to have important physical interpretations within our developing picture of disk-galaxy secular evolution. Most of them are reviewed in the following sections. The stellar dynamics of (R)S(r) galaxies are discussed in Lia Athanassoula's (2012) third lecture.

Further examples of the above structures are shown in Ron Buta's (2012) lectures and in the de Vaucouleurs Atlas of Galaxies (Buta et al. 2007).

3.4. Classical versus physical morphology of galaxies

At the start of a new science, it is common to classify the objects under study into "natural groups" (Morgan 1951) whose members share observed characteristics that are judged to be important. The success of the taxonomy depends on how well the natural groups – the classification bins – prove to correlate with physics. The galaxy classification scheme of Hubble (1936) and Sandage (1961) as augmented by de Vaucouleurs (1959) won a Darwinian struggle between taxonomic systems because it succeeded best in ordering galaxies by properties that helped us to understand galaxy formation and evolution. Ron Buta (2012) reviews this subject in his lectures at this School.

It is important – and fundamental to the aims of this paper – to contrast morphological galaxy classification as practiced in the early days of extragalactic research with what becomes necessary as the subject matures. Sandage and Bedke (1994) emphasize the importance early on of not being led astray by preconceptions: "The extreme empiricist claims that no whiff of theory may be allowed into the initial classification procedures, decisions, and actions." Nevertheless, every classifier chooses which features to consider as important and which to view as secondary. Hubble succeeded because he chose properties that became important to our understanding. Sandage recognizes this: "Hubble correctly guessed that the presence or absence of a disk, the openness of the spiral-arm pattern, and the degree of resolution of the arms into stars, would be highly relevant." Hubble based his classification on choices made with future interpretation in mind.

In contrast (Kormendy 2004a): "At the level of detail that we nowadays try to understand, the time has passed when we can make effective progress by defining morphological bins with no guidance from a theory. [Breaking] down the wall between morphology and interpretation successfully has always been a sign of the maturity of a subject. For example, without guidance from a theory, how would one ever conceive of the complicated measurements required to see solar oscillations or use them to study the interior structure of the Sun? In the same way, we need the guidance of a theory to make sense of the bewildering variety of phenomena associated with galaxies and to recognize what is fundamental and what is not. This motivates a `physical morphology' [Kormendy 1979a, b, 1981, 1982b; Kormendy & Kennicutt 2004], not as a replacement for classical morphology – which remains vitally important – but as a step beyond it. Physical morphology is an iteration in detail that is analogous to de Vaucouleurs's iteration beyond the Hubble tuning fork diagram."

One purpose of this review is to bring this process up to date. Our aim is to engineer the best one-to-one correspondence that we currently can between recognized types of galaxies or galactic building blocks and physical processes of galaxy formation.

It is reasonable to expect that an improved understanding of galaxies will show that an initial classification missed some physics. Also, some features of galaxies could not, early on, be observed well enough to be included in the classification. Two substantial additions and one small tweak to classical morphology are needed here. The small tweak is to recognize that nuclear star-forming rings are distinct from and often occur together with inner rings (r) in barred galaxies. This was point (f) in Section 3.1. The first major correction is to recognize and differentiate between real and counterfeit ellipticals (Section 7). The second is to recognize and differentiate between real and counterfeit bulges. The latter are "pseudobulges" (Section 5).

An essential aspect of physical morphology is to treat galaxies as being composed of a small number of building blocks, the distinct components in the mass distribution. The problem that this solves and the benefits that it provides were described as long ago as Kormendy (1979a):

"[Classical morphology] assigns a classification cell to each list of galaxy characteristics. As we look at galaxies more and more closely, the list of observed characteristics becomes alarmingly large. Thus the revised morphological types of de Vaucouleurs already require `about one hundred cells' (de Vaucouleurs 1963) and still do not recognize features such as lenses. It is difficult to attach dynamical significance to the bewildering variety of forms that the system describes."

In physical morphology, "we aim to identify as distinct components groups of stars or gas whose structure, dynamics, and origin can profitably be thought of as distinct from the dynamics and origin of the rest of the galaxy. ... The large number of cells in classical morphology is now thought of as the many ways that components and their secondary behavior [such as spiral structure] can be combined to make a galaxy. The strength of this approach is twofold. It breaks up the complicated problem of galaxy structure into smaller and more manageable pieces to which it is easier to attach dynamical interpretations. Secondly, investigations of correlations and interactions between components are very efficient in suggesting previously unrecognized dynamical processes [emphasis added]. Thus the component approach provides an efficient framework for studies of galaxy dynamics."

Throughout this review, my aim is to tweak the distinctions between galaxy components (and hence their names) until they are both correct descriptions and uniquely tied to physical processes of galaxy formation.

2 The (R)SA(oval)ab galaxy NGC 4736 (Figs. 3, 6, 8, 28, 31, 35, 38) contains a nuclear bar but no main bar. The nuclear bar provides a "connection" to barred galaxies that is relevant in Section 2.8. But it is much too small to affect the overall evolution of the galaxy. NGC 4736 is therefore completely unbarred for the purposes of the present discussion. Similarly, NGC 1068 (Fig. 9) also has a nuclear bar but no main bar. I suggest that the connection between these two nuclear bars and their host oval disks is not an accident – that, in fact, oval disks are the late-type analogs of lenses and thus were once barred (see point 3). The hint is that some nuclear bars can survive even after the main bar has evolved away. Back.

Next Contents Previous