Astron. Astrophys. 319, 747-756 (1997)

5. Classifying by means of structural components

In view of the drawbacks of the traditional method, especially when applied to the specific task of classifying galaxies in clusters, other methods have been introduced. Most of them have the advantage of giving a stable estimate of the Hubble type, in particular when prototypes or training sets are kept fixed, but they ultimately infer the Hubble type from the galaxy properties by using the fact that the Hubble type is correlated to the galaxy properties (colors, bulge on disk ratio, spectral shape, etc.). The observed properties of galaxies in classes defined in this way are strongly influenced by the constraints used to define the classes; for example, if colors are used for classifying galaxies, blue ellipticals will be missing whether they exist or not in the Universe. We are thus not in favor of adopting such methods.

We propose using an alternative classification method based on the criteria presented in Michard & Marchal (1993) and adopted with minor changes in Andreon et al. (1996). This method relies on the detection of structural components, rather than on the resemblance to standards.

The morphological classification is done as follows. A number of radial profiles (of surface brightness along the major and minor axes, of ellipticity, of the position angle, of isophote deviations from the perfect elliptical shape) are computed and then visually inspected for the detection of signatures typical of the morphological components described below. When a component (a disk, a bar, an envelope, etc.) is detected, we take note of it.

The segregation between the E and S0 types is based upon the examination of the surface brightness profile along the major axis of the projected galaxy, plotted on an r scale. The presence of a disk gives a characteristic bump above the linear profile which characterizes pure spheroids. This photometric signature of S0s should not be confused with other often observed deviations from the de Vaucouleurs law, both in giant ellipticals with an envelope enhanced above the extrapolated r line, and in minor objects with evidence for a cut-off below this line (Schombert 1986). This signature may be absent in some face-on S0s, which are then classified as Es; there is no way of avoiding this type of misclassification, as shown by the photometric similarity between the face-on NGC 3379, prototype of Es, and the edge-on NGC 3115, prototype of S0s (Capaccioli et al. 1991). Round galaxies with elliptical isophotes and exponential profiles are classified as S0s. In this spirit, these criteria are identical to those used in classical morphology, when the observer looks for subtle changes of gradient in a galaxian image.

The E galaxies are further subclassified into 'disky', or diE, 'boxy', or boE, and 'undefined', or unE. This is done mainly from the radial profile and sign of Carter's coefficient ¹. Roundish galaxies where the values fluctuate around zero are classified unE. To this class also belong Es with inner disky and outer boxy isophotes (or vice versa), and lemon-shaped Es. Note however that non zero are also produced by the presence of dust and shells, which also give a signature in higher-order coefficients and in the two-dimensional image.

The S0s are subclassified into the SA0, SAB0 and SB0 families, based upon the radial profile of the position angle (PA) of the isophotal major axis. For a galaxy observed with a sufficient resolution, the PA of the bulge and disk are nearly the same, while the PA of the bar stays in a limited range (except in cases of unfavorable projection). At the distance of nearby clusters ( km s^-1), it often happens that the contrast between bulge and bar is washed out by the seeing.

The segregation between S0 (or E) and S is based on the presence of spiral arms or very irregular or asymmetric isophotes. Because of this choice, the spiral class defined in such a way may include (a small number of) galaxies showing strong signs of interaction (tidal arms, double nuclei, non concentric isophotes, etc.). On the one hand, such a misclassification could be viewed as a contamination of the 'pure' spiral class; on the other hand, interacting galaxies often show important star formation. But in no case should we introduce another class, that of interacting galaxies, because this goes one step beyond that of classification, since we then interpret peculiarities in terms of gravitational interaction.

Spiral galaxies are not classified in more subclasses than S and Irr for two reasons. First of all, we lack an objective and unique method to distinguish stages in the S class. As noted by Sandage (1961), Hubble's (1936) description of spirals does not allow a unique classification of some spirals (e.g. NGC 4941) and some criteria have to be relaxed (Sandage, 1961, p. 13). Furthermore, splitting the S into the traditional stages does not separate galaxies of different physical types, since each stage is a blend of galaxies showing a range in properties, sometimes as large as the one found among the stages (see, e.g., Gavazzi & Trinchieri 1989, Roberts & Haines 1994, and Staveley-Smith & Davies 1988).

Finally, we stress that the structural classification scheme does not (implicitly or explicitly) use any other structural parameter, different from the ones above, to classify the galaxies; in particular, it does not make use of the bulge to disk ratio to discriminate between lenticulars and spirals.

In summary, the galaxy morphological type is nothing else than the obvious composition of morphological components detected following the rules outlined above. For example, a galaxy having (almost perfect) elliptical isophotes with a bump in surface brightness along the major axis and a characteristic twist of the position angle, is noted as having a photometric disk, a bar and no spiral arms, and therefore is classified as SB0. For this galaxy, whether it is boxy or disky is not a matter of concern, since boxiness or diskiness have no effect on its morphological type.

The Hubble sequence therefore looks like this:

The criteria of structural morphology are as close as possible to those of morphologists. The main difference is that structural components are measured, and galaxies are classified according the presence (or absence) of these structural components without exceptions and that galaxies are not classified according to the resemblance to standards. This alternative approach has several advantages.

© European Southern Observatory (ESO) 1997

Online publication: July 3, 1998
helpdesk.link@springer.de