Annu. Rev. Astron. Astrophys. 1994. 32: 115-52
Copyright © 1994 by . All rights reserved

Next Contents


One of the remarkable aspects of galaxies is that they can be classified into relatively few categories. The well-ordered sequence of galaxy types appears to offer a clue to possible formation and evolutionary processes. It is thus not surprising that morphology is so frequently an underlying theme in the study of galaxies, and serves as the principal subject of this review. Excellent discussions of recent classification systems are given by Buta (1992a, b). Other reviews with historical references are given by de Vaucouleurs (1959) and Sandage (1975).

Hubble (1926, 1936) introduced an early scheme to categorize galaxies; its concepts are still in use. In its simplest form, three basic types are recognized: ellipticals, spirals, and irregulars. Most modern schemes try to employ multiple classification criteria. There are two systems in common use today, both similar in application and notation, and both derived from Hubble's original classification scheme. One is the Hubble system as detailed by Sandage (1961) (Sandage and Tammann 1987, Sandage and Bedke 1993). The other system, developed by de Vaucouleurs (1959), adds more descriptive details to the notation and also extends the spiral sequence beyond Sc. Because the application of this system to over 20,000 galaxies in the Third Reference Catalog of Bright Galaxies (RC3, de Vaucoulers et al. 1991) has given it wide usage, it is adopted here. A few percent of all galaxies are unclassifiable. Many of these have unusual morphology because they are interacting systems. For the current purpose of looking for trends among average galaxies, we exclude these peculiar objects from our discussion.

Although the criteria for a type assignment are well recognized, the process is in reality subjective. Rather, we seek to replace qualitative measures with quantitative ones and ultimately to uncover the physics underlying galactic structure. As (we hope) will become evident, various trends do exist, but regardless of the parameters, the dispersion within each type is always large, much more so than errors of measurement. One of these trends, that of color with type, has been long recognized (Hubble 1936). Others, e.g., the H2 content, are only now being evaluated.

Most recently, numerous authors have attempted to classify galaxies using multivariate analysis of available quantitative measures (Whitmore 1984; Watanabe et al. 1985). Such quantitative studies show the existence of two principal categories of galaxy parameters: those that measure the absolute scale (size, luminosity, mass) and those that describe more its form (morphology). Because he undertook his classification scheme at a time when distance estimates were available for only a handful of galaxies, Hubble could not discriminate the scale dimension: that bright galaxies are bigger and more massive than faint ones of the same morphology.

With notable exceptions and for obvious reasons the study of galaxies is directed primarily to those listed in catalogs, e.g., the New General Catalog (NGC), the Shapley-Ames Survey and its update (RSA), the Uppsala General Catalogue (UGC), and the various editions of the Reference Catalogues (RC). They are all flux or diameter limited. They all suffer from Malmquist bias and they are all deficient in low surface brightness systems. They list the brighter galaxies though generally not the most or least luminous systems. And they mostly contain nearby galaxies, i.e., z << 0.1. These are systems for which we have the most data, and it is such ``catalog galaxies'' that we discuss here. For some of the data considered here we can also construct an approximately volume-limited sample and properties so derived are also noted. They generally give smaller values (the Malmquist bias operating) but show trends with type similar to those for the biased samples.

The ability to measure properties is not equal for all types. Thus HI is rare in elliptical systems, and we lack any meaningful HI parameters for these systems. Similarly, total mass estimates are poorly determined for E's and when available are generally based on different approaches than that used for spirals. Comparison of total mass estimates between E's and spirals are accordingly uncertain and are not made. Other properties such as CO and its derivative, H2 content, or X-ray luminosity are available only for relatively small samples and suffer accordingly. For X-rays, the best that can be done at present is to contrast data for ellipticals with that for the overall spiral category; we note interesting differences.

The most obvious omission is the lack of any distinction between regular and barred spirals. Here we have been guided by Holmberg's (1958) remarks in his classic paper on the photometry of galaxies. He notes:

. . . that the majority of spiral nebulae exhibit a more or less pronounced bar; the bar may not always be recognizable on the blue plate, but is usually visible on the photovisual exposure. It seems quite possible that a bar is a structural detail common to all, or most, spiral nebulae and that the observed differences are of a quantitative rather than qualitative nature.

This is strikingly illustrated for M51 by Zwicky (1957, Figure 41) by means of a composite of yellow and blue sensitive images. In this composite M51, a classic ``regular'' spiral shows a small but pronounced bar in its central region.

Some caveats must be emphasized. There are many type-dependent trends to be found in the literature, based on widely ranging sample sizes and of various levels of confidence. We are unable to discuss all or even a significant fraction of these. Since our focus is on type-dependencies, other and possibly related issues are treated only briefly. We apologize at the onset if your favorite relationship is omitted and hope that the references will guide the reader to more extensive discussions on these topics.

Next Contents