2.2. Classical Galaxy Morphology

Excellent reviews of the history and current status of galaxy classification systems are given by de Vaucouleurs (1959a), Sandage (1961) and especially Sandage (1975). The present discussion is restricted to a sketch of the basic Hubble-Sandage-de Vaucouleurs classification, including refinements having to do with structure but not those dealing primarily with stellar content. The purposes and strengths of the classical morphology are well described in Sandage (1975). It has a secure position in galaxy studies because it correctly identifies sets of different characteristics without attempting to attach interpretations. In these lectures it is useful to adopt a different (and more hazardous) approach, which in no way undermines the classical approach, but which is efficient in leading to new physical results. I will emphasize the physical processes which underly the classification, anticipating some results from later sections to provide feedback for the refinement of the classification. In particular, I emphasize the limitations of classical morphology, because these are the driving force for further progress. These limitations motivate the discussion of section 2.4 on physical morphology.

Figure 1 illustrates the basic classification scheme. Hubble's (1936) original tuning-fork diagram recognized the fundamental distinction between elliptical galaxies, which do not contain disks, and spiral galaxies, in which a disk becomes progressively more important along the sequence Sa-Sc. The other classification parameters along the spiral sequence are the tightness of the winding of the arms (Fig. 1; see also Fig. 7 of Kennicutt 1981) and the degree of resolution into young stars and H II regions, which increases toward the right. Subsequent work has shown that these criteria are to some extent physically related. For example, a more massive bulge results in more tightly wound arms (Lin and Shu 1964; Roberts, Roberts and Shu 1975). However, the visible structure at late types is dominated by the distribution of young stars, and the factors which control the rate of gas conversion into stars of various masses are very poorly understood. Elliptical galaxies were interpreted until the mid-1970s as a sequence of in creasing rotation, which was plausibly connected to the spiral sequence. Illingworth (1977) and others have subsequently shown that flattening correlates poorly with rotation (section 4.2.6). It is probably still useful to think of the horizontal sequence E-Sc as one of the increasing importance of dissipation versus ongoing star formation during the galaxy collapse stage. However, the situation is greatly complicated by the possible rearranging of morphological types through galaxy mergers and through other interactions (e.g., section 3.3.7). The point of these remarks is to emphasize how much more complicated the real world may be than a one- or two-dimensional morphological sequence. Also, inspection of photographs may not be sufficient to recognize all physical differences. There may well exist a sequence of ellipticals (with increasing specific angular momentum) which attaches naturally onto the spiral sequence. However, since ellipticals can be flattened by velocity anisotropies as well as by rotation, the position of a galaxy in the physical sequence is not easily deduced from its apparent shape. An added complication is the fact that the apparent flattening contains projection effects.

Figure 1. The basic Hubble-Sandage-de Vaucouleurs classification scheme. Hubble's (1936) well known tuning-fork diagram (upper panel) was a two-dimensional classification. The lower panel shows the three-dimensional classification volume envisaged by de Vaucouleurs (1959a). There is a continuous sequence of classes (E, S0, Sa-m, Im) horizontally, families (SA, SAB, SB) vertically and varieties [(r), (rs), (s)] perpendicular to the page. For classification purposes this continuum is somewhat arbitrarily divided into discrete cells (lower right).

Another limitation of Hubble types is illustrated in Figure 2. This shows, separately for types Sb and Sc. galaxies of widely differing spiral structure. The top two panels show spirals with non-global spiral filaments. The second panels show galaxies with two coherent spiral arms; these are the "textbook" spirals embodied in the folklore and sketched in Figure 1. The third panels show galaxies which also have global spiral structure, but which have two sets of arms imbedded in disks of different surface brightnesses. The galaxies illustrated are chosen to be unbarred in blue photographs and to have similar luminosities. Although many of the above distinctions in spiral structure have been recognized for years and are included in descriptions in the Hubble Atlas, none are embodied in the morphological notation. They will be discussed further in section 2.5.

Figure 2. Varieties of spiral structure for galaxies of types Sb (left) and Sc (right). Flocculent spirals (top), global-pattern spirals (middle) and oval galaxies (bottom) are discussed in sections 2.5.3, 2.5.3 and 2.5.1, respectively. Absolute magnitudes are: NGC 2841, -21.5; NGC 4571, -20.2; M81, -20.8; NGC 5364, -21.5; NGC 210, -21.9; NGC 1566, -21.5 ( H0 = 50km s-1 Mpc-1). Sources of photographs: NGC 2841, Mt. Wilson 2.5 m telescope; NGC 4571, M81 and NGC 5364, Palomar 1.2 m Schmidt telescope; NGC 210, Palomar Observatory Sky Survey blue plate; NGC 1566, ES0/SRC J Survey film. The upper four plates are taken with J emulsions and Wr 2c filters.

The second dimension in the Hubble system is the distinction between "normal" and barred spirals. Hubble recognized that the arms of the tuning fork are the extremes of a continuum of intermediate cases. However, the notation gives only minimal recognition to transition types; objects are as much as possible assigned pure types in the Hubble Atlas (Sandage 1961) and in the Revised Shapley-Ames Catalog (RSA; Sandage and Tammann 1981). Since the S-SB distinction poorly measures the strength of the bar (just as the Sa-Sc sequence imperfectly measures the strength of the spheroid), there is considerable heterogeneity of physical properties at any Hubble type. Depending on the application, the large size of the classification bins can be an advantage or a disadvantage. A second remark concerns elliptical galaxies, which are now believed to be triaxial objects (section 4). Figure 1 immediately suggests the question of whether the elliptical sequence should be thought of as having width in the same (S versus SB) sense as the spiral sequence. This question is taken up in section 5 on barred galaxies. Preliminary indications are that the two types of triaxiality are rather different, and that the degree of triaxiality of ellipticals represents the width of the E sequence in the horizontal, "dissipation" direction of the tuning-fork diagram.

The Hubble Atlas sets up a definitive catalog of morphological types using photographs of prototypical galaxies taken with the large-scale Mt. Wilson and Palomar telescopes. Apart from refinements, the main improvement is the addition of the S0 class of galaxies with smooth disks of mostly old stars and no spiral structure. (This class was "more or less hypothetical" in 1936, see Fig. 1). The class of Magellanic irregulars is also attached at the end of the spiral sequence. Further improvements and a more detailed discussion of features such as lenses and luminosity classes are emerging from an ongoing survey (Dressler and Sandage 1978; Sandage and Brucato 1979, 1982; Sandage and Tammann 1981). This benefits especially from the large scale (10.8" mm^-1) of the Las Campanas 2.5 m telescope. Some of the details are discussed below.

The Revised Morphological Types of de Vaucouleurs (1959a, see also 1963, de Vaucouleurs and de Vaucouleurs 1964; de Vaucouleurs, de Vaucouleurs and Corwin 1976; hereafter RC2) does not change the Hubble-Sandage types, but rather adds detail by giving explicit recognition to a number of additional features. As illustrated in Figure 1, the classification of barred (SB) and unbarred (SA) galaxies is made symmetric. Transition types SAB are no longer identified less frequently than pure types (de Vaucouleurs 1959a, 1963). Unbarred galaxies are termed "ordinary" rather than "normal", because they are not more normal than SB galaxies (they are actually in the minority). A larger addition to Hubble types is the introduction of a third dimension in the classification, to distinguish between (s) galaxies, which have spiral arms beginning at the center, in an amorphous bulge, or at the ends of a bar, and (r) galaxies, whose arms begin tangent to an inner ring (Fig. 40). As indicated in Figure 1, the A-B and (r)-(s) distinctions are largest at intermediate types. According to de Vaucouleurs (1959a), all positions in the three-dimensional continuum depicted in Figure 1 represent real galaxies. This should be viewed with caution. The discussion of section 5.4 will suggest that the features identified as inner rings (r) are different in SA and SB galaxies. Also, lens components are denoted (r) in the RC2 (section 5). Note that a third dimension is introduced for the (r) - (s) distinction, but a physically similar although operationally distinct form of outer ring (section 5.4) is recognized only by adding the symbol (R) at the beginning of the classification. Thus, the Revised Morphological Types (Fig. 1), although recognizing additional important features, are still incomplete and physically inhomogeneous, and also do not make clear the relative importance of the features described.

Another refinement of the de Vaucouleurs types is the subdivision of the Hubble-Sandage class Sc into Sc, Scd, Sd, Sdm and Sm. These subdivisions recognize a sequence of increasing resolution into patches of star formation at the expense of coherent spiral structure. Evidence discussed in section 2.3 shows that this is a sequence of decreasing luminosity and mass.

The increasingly chaotic structure along the Sc-d-m sequence is an indication that the regularity of structure correlates with luminosity. This was first exploited by van den Bergh (1960a, b, c; 1966) to construct an important luminosity classification system. van den Bergh showed that high-luminosity Sb-c galaxies have long, well developed spiral arms of relatively high surface brightness (luminosity class I). Less luminous galaxies (classes II, III, ...) have progressively less well developed spiral structure: the arms are shorter and more patchy, and ultimately are impossible to recognize (class V). Galaxies of adjacent luminosity classes differ on average by ~ 1 mag in luminosity (van den Bergh 1960a, b, c, 1982). van den Bergh emphasized further that Hubble types are defined by supergiant galaxies, and describe progressively fainter galaxies less and less well. As luminosity decreases, the distinction between types decreases and ultimately almost disappears (van den Bergh 1976b, 1977). Since there are no dwarf Sb galaxies (the lowest recognized luminosity class is III), low-luminosity galaxies are ultimately distinguished only as relatively non-descript dwarf ellipticals (which are largely gasless and old) and dwarf irregulars (which contain gas and form stars, but which lack regular structure). This important point deserves emphasis and enlargement. It is my impression that the sophistication of essentially all galaxy structure correlates with luminosity. Only the most luminous galaxies produce features such as well-defined rings. At lower luminosities, spiral structure begins to disappear (van den Bergh's luminosity classes). Disks and bars are the most robust structures, and disappear at the lowest luminosities (e.g., van den Bergh 1966). I suspect that a viable luminosity classification could be constructed using virtually any kind of regular structure.

The physical basis for luminosity classification is discussed by Roberts, Roberts and Shu (1975, see also Iye and Kodaira 1976), who suggest that more massive galaxies have stronger density-wave shocks, increased gas compression and more vigorous (and therefore more visibly regular) star formation outlining the arms.

Recent work casts doubt on the accuracy to which morphology can be used to estimate luminosities. A luminosity classification of 1246 galaxies in the Shapley-Ames (1932) catalog has been published by Sandage and Tammann (1981) and analyzed in that paper and in Tammann, Yahil and Sandage (1979). These studies show that there is a spread of more than 3 mag in the absolute magnitudes of galaxies of a given luminosity class (see Fig. 1 of Tammann, Yahil and Sandage 1979 and the illustrations on pp. 124-127 of the RSA, but contrast van den Bergh 1982). Tammann, Yahil and Sandage conclude that factors other than luminosity influence the classification. It is possible that the luminosity discrimination would be improved by taking account of the many physical factors now known to influence spiral structure (section 2.5.3).

Another proposed revision of Hubble types involves the interpretation of S0 galaxies. van den Bergh (1976a) notes that the sequence S0-c is not very precisely a sequence of increasing flattening. For example, there exist S0 galaxies such as NGC 4762 which are very flat and which have small bulges. These differ from spirals mainly in their lack of gas and star formation. van den Bergh suggests that S0 galaxies form a sequence S0a-S0b-S0c paralleling the sequence Sa-Sb-Sc and distinguished from it by a lack of gas. He classifies spirals with "anemic" structure in an intermediate sequence Aa-Ab-Ac, which is inferred to be intermediate in H I content. Finally, he notes that S0 and anemic galaxies are more common in rich clusters than in the field, and suggests that environmental effects such as H I sweeping by an intercluster gas might cause spirals to evolve into anemics and ultimately into S0s. This work has added fuel to the controversy (see White 1982) between "stripping" theories, which suggest that environmental effects produce S0s from spirals, and "intrinsic formation" theories, which claim that the difference between S0s and spirals was built in during galaxy formation.

A final morphological type which is relevant here involves non-isolated galaxies. Morgan (1958, 1959) has developed a classification system based largely on the importance of the bulge. Since the main aim is to describe galaxy populations, only one feature of the classification is crucial here. Morgan introduces the term cD for supergiant elliptical-like galaxies with very extensive outer halos, which are up to several Mpc in diameter if H₀ = 50 km s^-1 Mpc^-1 (Matthews, Morgan and Schmidt 1964; Morgan and Lesh 1965). Generally these are the brightest galaxies in rich clusters, but cD-like galaxies have been reported in poor clusters (Morgan, Kayser and White 1975; Albert, White and Morgan 1977). The possible production of cD halos through environmental effects is discussed in section 3.3.6.