Why do you want to classify galaxies? «You ask what is the use of classification, arrangement, systematization. I answer you; order and simplification are the first steps toward the mastery of a subject – the actual enemy is the unknown». This sentence from Thomas Mann’s The Magic Mountain, is quoted by Allan Sandage (1926-2010) in The Classification of Galaxies: Early History and Ongoing Developments (ARAA = Annual Review of Astronomy and Astrophysics, vol. 43, 581, 2005). There Sandage urges that «that the classification be made on the basis of morphology alone, not on the basis of supposed physics that some wish to be introduced to “explain” the classification».
Appearance is the first but not the only key to built a classification. Other characteristics or physical properties may be effectively used.
We start with morphology, the oldest and the simplest approach. Morphological categorization is a valuable tool to spot possible lines of attack in exploring a completely new territory (first step of any new natural science). It is also precious for establishing a common synthetic language (a jargon), provided that the number of classes is kept small (otherwise classification turns into a plain description). Taxonomy shall not be confused with the naturalistic approach of medieval science [what is it? why?].
The very first classification of nebulae is that by the German astronomer Max Wolf (1863-1932), pioneer of astrophotography. It was published in 1908, when galaxies had not yet been understood as island universes. Wolf however thought that «there are no two nebulae in the sky that are similar»: indeed a bad starting point for a synthetic taxonomy.
The most famous and widely utilized morphological classification of galaxies is that devised by Edwin Powell Hubble in 1926, just after he had discovered galaxies to be “island universes”. We still use this classification extensively, and not only for semantic purposes, as the Hubble sequence catches a basic property of galaxies.
Said in short, Hubble classes define a sequence of increasing specific angular momentum (momentum per unit mass [give units]). Since, as we shall see, momentum is grossly responsible of the efficiency of the star formation [why?], the Hubble sequence is also a line of increasing gas-to-stars ratio.
Hubble published the first version of his scheme, the famous “tuning fork” diagram, as early as 1926 (Ap.J., 64, 321), in a row with its discovery of the extragalactic nature of a few nearby nebulae. The scheme was then reviewed in the book The realm of the nebulae where the S0 class was introduced as a “junction” between E and S galaxies. Just after Hubble’s death, the scheme was frozen by Allan Rex Sandage in the Hubble Atlas of Galaxies (Carnegie Institution of Washington, 1961).
Given the importance of Hubble’s classification, we shall spend some time on it. The other non-physical classifications reported at the end of this lecture will be exposed in short, mainly to catch names and symbols that remain in to-date extragalactic jargon.
In his original scheme (1926) Hubble subdivided most galaxies into two classes according to their smooth rather than complex appearance:
Each class was broken up according to morphological characteristics as the amount of flattening for ellipticals and the appearance of bulge and arms for spirals.
Galaxies which did not fit into these categories were designed as irregulars.
Later Hubble introduced the S0 class (lenticulars) to create a morphological transition between ellipticals and spirals: disk galaxies with no arms.
Hubble had photographs (taken with blue emulsions [why this is important to remember?]) of a few tens of bright nearby, basically regular galaxies. A richer data-base would have been much more difficult to synthesize in just a few classes.
Galaxies resembling a given Hubble type, but not fitting all the characteristics or showing asymmetries and irregularities, are qualified as peculiar (pec).
Following Hubble, E and S0 galaxies are often referred to as early types, and spirals as late, in a sequence with no temporal meaning. It rather describes an increasing morphological complexity. Similarly early spiral means Sa-Sba, late spiral Sc-SBc.
The universality of Hubble’s system is demonstrated by the fact that, out of 338 “strange” galaxies of Halton Arp’s Atlas of peculiar galaxies (1966), only 13% escaped a Hubble class.
Galaxies on LHS of this diagram are designated early types, those on the RHS late types. Spiral sub-types can also be generically specified as early (Sa-SBa) and late (Sc-SBc). These expressions have no temporal (evolutionary) meaning (Credit: SDSS).
A sample of Hubble types from modern color pictures for which credit is due to CFHT + Coelum, NASA (HST), ESA.
Elliptical galaxies = E In his 1926 paper Hubble wrote that elliptical «give images ranging from circular through flattening ellipses to a limiting lenticular figure in which the ratio of the axes is about 1 to 3 or 4. They show no evidence of resolution and the only claim to structure is that the luminosity fades smoothly from bright nuclei to indefinite edges. Diameters are functions of the nuclear brightness and the exposure time.»
Classification of an elliptical galaxy is straightforward, since there is little structure present. The isophotes (lines of equal surface brightness) are roughly elliptical.
Types from E0 to E7 (to E6 if E7 = S0s) are recognized, where the integer following the symbol E is the rounded value of 10x(1-b/a), with b/a = apparent (projected) axis ratio. Specifically, an E0 is an elliptical galaxy that appears circular [Is it necessarily a sphere seen in projection? Is the Hubble sub-type for ellipticals a physical or a descriptive parameter?].
No ellipticals flatter than E7 have been found, possibly because near this shape there is a stability limit. Note that even the apparent ellipticity is not completely well-defined, as many Es have changes in ellipticity with radius or isophotal twists. Most are likely triaxial systems.
NGC 4874 in the Coma cluster: a giant E galaxy surrounded by smaller Es and S0s. Ellipticals prefer dense environments. Credit: SDSS.
The Perseus Cluster, dominated by ellipticals. Note the different luminosities and sizes [and masses?] of these equally distant objects. [Why stars are so different in color and galaxies not?]. Credit: SDSS.
Spiral galaxies = S Spirals are divided by Hubble into the two families of ordinary (S) and barred (SB), each with stages from Sa-SBa (early spiral types: large bulge, tightly wound and fairly smooth arms) to Sc-SBc (late spiral types: small bulge, loosely would and very textured arms, show resolution in stellar cluster and HII region). Arms in some spirals are designed more by the dust lanes than by the light of stars. The sub-classes a, b, or c were originally set on the basis of the pitch angle of the spiral arms (see graphical definition). The classification criteria recognize the following types:
Sa: large central bulge, smooth, tightly wrapped spiral structure; example: M104, the Sombrero galaxy (possibly S0-Sa);
Sb: less noticeable bulge, looser spiral structure; example: M31, the Andromeda galaxy;
Sc: weak or no bulge, open spiral structure, very knotty appearance; example: M33, the Triangulum galaxy.
Parallel to the normal spirals are the barred spirals;
example: NGC 1365, an SBc.
Transition stages between two consecutive stages may be assigned; e.g. SBab means an intermediate stage between SBa and SBb.
In a later version of the Hubble classification (but actually introduced by de Vaucouleurs), the stage “d” following “c” was added to account for newly observed galaxies with particularly small bulges and loose structures.
The French mathematician Henry Poincaré, at the conclusion of the preface to his book Hypthéses Cosmogoniques, states: «One fact that strikes everyone is the spiral shape of some nebulae; it is encountered much too often for us to believe that it is due to chance. It is easy to understand how incomplete any theory of cosmogony which ignores this fact must be. None of the theories accounts for it satisfactorily, and the explanation I myself once gave, in a kind of toy theory, is no better than the others. Consequently, we come up against a big question mark.»
Sandage (1975) reports an interesting comment on Hubble’s classification made by an influential British astronomer, John Henry Reynolds. Soon after publication of Hubble’s 1926 paper, Reynolds (1927) criticized the system because of a supposed inadequate number of classification bins. Impressed by the enormous variety of galaxy structures, Reynolds wrote: «The problem I have always found in attempting a general classification of spiral nebulae is that one meets case after case where a special class is required for the individual object. Spectral classification of stars is a simple and straightforward matter compared with this.» Reynolds was wrong.
A sample of disk galaxies seen on edge. Note the different sizes of bulges and the occurrence of equatorial dust lanes. Credit: SDSS.
Irregular galaxies = I The fortune of Hubble’s classification depends on two key properties. In a simple format the tuning fork hosts the majority of bright galaxies (of the kind of those which were at Hubble’s disposal). Moreover, as we shall see later, the purely morphological classes correlate well with observables such as bulge/disk ratio, light distribution, colors, kinematics, gas content, star-forming properties, ISM chemical composition. At Hubble’s time there were however some galaxies with no obvious regular structure (either disk-like or ellipsoidal), which could not be classified either E or S types. Hubble named them Irregular.
Actually, the class of Irregulars is far richer than Hubble thought. It includes a large fraction of dwarfs (see those of the Local Group), the interacting systems as M51, some accreting giants (as NGC 1275 in Perseus), and a remarkable portion of active galaxies.
Sandage not only explicitly recognized transition S0 galaxies, but found subtypes depending on the amount of dust in the disk plane (and strength of the disk) denoted S0-, S0, S0+ as in the Hubble Atlas. S0 systems pose some interesting problems of recognition when not seen close to edge-on, since they are tough to tell from ellipticals with extended envelopes.
S0 galaxies = S0 What are S0 galaxies? (Ap.J., 601, L37, 2004) van den Bergh writes: «Hubble (1936) introduced the classification type S0 to bridge the gap between objects of types E7 and Sa. Galaxies that are morphologically intermediate between elliptical and spiral have also been referred to as “lenticulars” by de Vaucouleurs (1959). Not unexpectedly S0 galaxies, on average, appear more flattened than E galaxies … The physical difference between elliptical and lenticular galaxies is that S0 galaxies contain an old disk, whereas elliptical galaxies do not. It often becomes difficult to unambiguously distinguish between these two classes of objects when either (1) only a small fraction of the light originates in the disk or (2) the disk is viewed almost pole-on.» See also Sandage, Galaxies and the Universe, 9, Chicago. Univ. Press, 839, and Spitzer and Baade, Ap.J., 113, 413, 1951.
S0s are then the flat galaxies with little or no structure in their disks. Their existence might be (wrongly) interpreted as a hint that the tuning fork is an evolutionary diagram.
In the Hubble Atlas of Galaxies, Sandage (1961) defined sub-types of S0s: S0-, S0, S0+ depending on the increasing strength of the disk and amount of dust in the disk plane.
Arranged from Sandage & Tammann, The Revised Shapley-Ames Catalogue (RSA), Carnegie Inst. of Washington, 1981.
Starting with the seminal papers for volume 53 of the Handbuch der Physik (1959), de Vaucouleurs (1918-1995) added significant innovations to Hubble’s scheme, aimed at increasing the resolution with new morphological features and at establishing the continuity between adjacent types.
This second goal was achieved by replacing the one-dimensional double sequence with the notion of classification volume (see figure). New stages are added for spirals and irregulars: Sc-Scd-Sd-Sdm-Sm-Im (where “m” stays for Magellanic, the LMC being the prototype in view of the discovery by de Vaucouleurs of its very weak spiral arms).
The continuity between S type, now called SA, and SB types, is secured by adding the mixed types SAB and SBA (where the first of the letter A and B designates the dominant type).
Details of this classification are given in the introduction to the Reference Catalogue of Bright Galaxies (RC1), that de Vaucouleurs and his wife Antoinette (1921-1987) published in 1964.
The classification recognizes if the inner structure is ring like or purely spiral, with mixed types indicated as (r), (rs), (sr), (s), and if there is an outer ring (R). There is another class of irregulars, that de Vaucouleurs named Irr II, grouping objects with amorphous texture of the luminous form.
The prototype of the class is M82. Sandage warns not to place in this class those galaxies clearly made peculiar by tidal interaction. The classification scheme of de Vaucouleurs is indeed complex, and it has been gradually abandoned.
The figure presents a cross section of de Vaucouleurs’ classification volume showing the transition structures for the same S type.
Left: the late end of the classification sequence. Right: various interacting galaxies. From Sandage, in Galaxies and the Universe, 1975.
One feature of de Vaucouleurs’ scheme has remained. In order to correlate morphology with observables and with astrophysical parameters, de Vaucouleurs has introduce a numerical type T, an integer ranging from -6 to +10. The first 3 entries are reserved to ellipticals with or without envelopes (-6 is for compact ellipticals, cE, placed at the beginning of the sequence for reasons of continuity in the correlation of T with astrophysical parameters), -3 to -1 is for S0, 0 for S0/a, on through 5 for Sc and 10 for Im.
The association of de Vaucouleurs’ T index with the morphological classes in Hubble’s and de Vaucouleurs’ classification schemes is given at RHS.
In a series of papers from 1960 to 1966, the Dutch astronomer Sidney van den Bergh, then at the David Dunlap Observatory in Canada (DDO), suggested that the intrinsic luminosity of (chiefly late-type) spiral galaxies could be deduced by the appearance of their spiral arms.
Following the luminosity-class categorization for stars, he defined five types of spirals, from the Type I (supergiant) with well-developed global spiral structure, to the Type V (dwarfs) with worn-out ill-defined spiral arms.
The roman number of the type follows the Hubble class: e.g. ScI, for supergiant spiral of class “c”. While this classification is no longer in use, it has provided some new terms to extragalactic jargon (e.g. ScI) and the starting point for fundamental discoveries about the motions of nearby universe. The basic criteria of this DDO-classification are outlined in the table below.
van den Bergh has also introduced a new class of “anemic spirals”: objects with a low rate of star formation and a low surface brightness that he placed parallel to the Hubble sequences of S and SB, in line with the sequence of S0s. Anemic galaxies are indeed present in very rich cluster, and seem therefore connected to strong environmental effects.
For his classification of galaxies, named “of Yerkes” after his Observatory, William W. Morgan (1906-1994) considered a sort of bulge-to-disk ratio rather than the occurrence and strength of the spiral arms, as Hubble did. In the founding paper of 1958 he introduced three parameters. The first is a “form factor”.
To the classical Hubble types, S = spiral, B = barred, E = elliptical, I = irregular, Morgan added four “form families”:
The second parameter is an inclination class, from 1 for round objects to 7 for the flattest ones, mimicking Hubble’s ellipticities. The third is a spectroscopic type based on Morgan and Mayall’s correlation between form and stellar population in galaxies spectra. This is used to set up a form classification indicative of the general kind of stellar population encountered in the majority of galaxies classified. Note that the spectral class is deduced by morphology.
Classifications as Yerkes, based on light concentration, are receiving a renewed interest in studies of high-redshift galaxies.
Morphological classification is not the only source of jargon. Many galaxy types have been introduced by grouping objects according to other properties than just the shape as perceived, typically, in one optical band. For instance, some galaxies look quite different if seen in the UV rather than in the IR [why so?].
The morphological differences increase dramatically in the radio on the one side and in the X-ray domain in the other. In fact, some galaxies are powerful radio and/or X-ray emitters, other are not. This is enough to create categories based on an extended concept of color excess, such as Blue Galaxies, Radio Sources, or X-ray Objects. Other classes may be generated by consideration on the total luminosity (Ultra-bright Galaxies or Dwarfs), on the average surface brightness (Low Surface Brightness = LSB), on the degree of concentration (Compact galaxies) or resolution (Amorphous).
Galaxies may be also classified according to:
Some of these categorizations do not pertain to intrinsic (genetic) properties but rather to occasional/transient phenomena. In addition to their Hubble calls, each galaxy may appear in more than one of the above classifications.
Normal galaxies are built up by stars, gas, and dust and their spectra are the integral (along the line of sight) of the spectra of these components. Spectral lines are broadened by the random motions of the stars (gaseous nebulae usually participate to ordered motions, which do not contribute to the broadening of the spectra).
Active galaxies have spectra with emission lines brighter and broader than normal galaxies. They are otherwise normal objects, made peculiar by a violent release of energy from their nuclei (orders of magnitude higher than that of all the stars of the Milky Way, but produced in a region of the size of the Solar System). The wealth of phenomena produced by a complex central structure powered by a supermassive black hole is currently described by a unified model, where the various classes of Active Galactic Nuclei (AGN) correspond to different viewpoints of the same morphology.
Carl K. Seyfert (1911-1960) has been one of the first astronomers to realize that some galaxies were characterized by peculiar compact and bright nuclei. In 1943, while a PhD student, he selected six galaxies from the images collected at the Mt. Wilson Observatory which had in common broad emission lines, in order to analyze their spectra: NGC1068, NGC1275, NGC3516, NGC4051, NGC4151 e NGC7469. Five of these galaxies are spirals, except NGC1275 which is a peculiar irregular galaxy. Since them the term “Seyfert galaxy” is used to name a whole class of active galaxies, i.e. galaxies whose emission cannot be explained only by the combined emission of their stars. Seyfert galaxies were originally noted as having unusually bright, compact (starlike) nuclei.
Left: HST imaging (Oxygen) of the Seyfert galaxy NGC 4151. Right: spectrum (in false colors) across the nucleus showing the two very complex and broad lines of [OIII]. Adapted from NASA-HST.
In 1974, owing to improvements in spectroscopic techniques, Ed Khachikian e Dan Weedman re-classify Seyfert galaxies in two classes:
In 1963 the Armenian astronomer Benjamin Egishe Markarian (1931-1985) published the results of the study of 14 peculiar galaxies showing evidence of non-thermal emission, i.e. emission not directly linked to the stellar light [What is thermal emission? What is non-thermal emission?]. The sample included all galaxies analyzed by Seyfert. Markarian started a survey (with the 1m Schmidt telescope of the Byurakan Oservatory in Armenia), selecting a sample of objects with a blue-UV excess in their spectral continuum, which resulted in the publication of a catalog containing approximately 1,500 galaxies: the so called “Markarian galaxies”. Actually this is not a class of AGNs, but a sample of low-redshift objects with a peculiar blue-UV spectrum. It includes 200 Seyfert galaxies, and hundreds of starburst, blue compact, and HII galaxies.
Quasars were identified for the first time (around the 1950s) as strong and extended radio sources. Since their optical counterparts were point-like, as observed from ground telescopes of the time, they were labeled QSRSs (Quasi-Stellar Radio Sources), or Quasars. The optical emission seen in Quasars is variable on timescales of months (much less in some sources, and faster at shorter wavelengths) suggesting that the emitting region is small, at most across.
The RHS shows the Quasar 3C 273: the radio astronomers of the Jodrell Bank used the lunar occultation method to measure the position of this source that they believed to be extragalactic. Based again on the Palomar photographic plates the optical counterpart looked like a 13° magnitude star, surrounded by a faint nebulosity. The spectrum of 3C273 revealed hydrogen lines corresponding to a redshift of (i.e. a relative speed of 16% of . Thus this source is located about 685 Mpc away (H0 = 70 km/s/Mpc). From this distance we can estimate an absolute magnitude of the object of , equivalent to
Ground telescope and HST images of QSO 1229+204. The HST resolution reveals the presence of the host galaxy with a spiral structure. Credit: NASA-HST.
Blazars represent a class of extremely “violent” objects: their emission extends to gamma-ray energies and they vary on very short timescales. Many of the brighter blazars were first identified, not as powerful distant galaxies, but as irregular variable stars in our own galaxy. These blazars, like genuine irregular variable stars, changed in brightness on periods of days or years, but with no pattern. The early development of radio astronomy had shown that there are numerous bright radio sources in the sky.
By the end of the 1950s the resolution of radio telescopes was sufficient to be able to identify specific radio sources with optical counterparts, leading to the discovery of quasars. Blazars were highly represented among these early quasars, and indeed the first redshift was found for 3C 273, a highly variable quasar which is also a blazar. In 1968 a similar connection between the “variable star” BL Lacertae and a powerful radio source VRO 42.22.01 was made. BL Lacertae shows many of the characteristics of quasars, but the optical spectrum was devoid of the spectral lines used to determine redshift. Faint indications of an underlying galaxy – proof that BL Lacertae was not a star – was found in 1974.
The extragalactic nature of BL Lacertae was not a surprise. In 1972 a few variable optical and radio sources were grouped together and proposed as a new class of galaxy: BL Lacertae-type objects. This terminology was soon shortened to BL Lacertae object, BL Lac object, or simply BL Lac.
All known BL Lacs are associated with core dominated radio sources.
The name “blazar” was originally coined in 1978 by astronomer Edward Spiegel to denote the combination of OVV quasars and BL Lac objects. An optically violent variable quasar (OVV quasar) is a type of highly variable quasar. It is a subtype of blazar that consists of a few rare, bright radio galaxies, whose visible light output can change by 50% in a day. They are similar in appearance to BL Lacs but generally have a stronger broad emission line, and tend to have higher red shift components.
Radio galaxies and their relatives, radio-loud quasars and blazars, are types of active galaxy that are very luminous at radio wavelengths (up to 1039W between 10 MHz and 100 MHz). The radio emission is due to the synchrotron process. The observed structure in radio emission is determined by the interaction between twin jets and the external medium, modified by the effects of relativistic beaming. The host galaxies are almost exclusively large elliptical galaxies. Radio-loud active galaxies are interesting not only in themselves, but also because they can be detected at large distances, making them valuable tools for observational cosmology.
In 1974, radio sources were divided into two classes, now known as Fanaroff and Riley Class I (FRI), and Class II (FRII) by the name of the authors. The distinction was originally made based on the morphology of the large-scale radio emission (the type was determined by the distance between the brightest points in the radio emission): FRI sources were brightest towards the centre, while FRII sources were brightest at the edges. B.L. Fanaroff and J.N. Riley observed that there was a reasonably sharp divide in luminosity between the two classes: FRIs were low-luminosity, FRIIs were high luminosity.
The figure shows the galaxy NGC 5128 in Centaurus from a Palomar Sky Survey picture with superimposed IR emission (red) and the contours of the radio emission at 20 cm.
A low-ionization nuclear emission-line region (Liner) is a type of galactic nucleus that is defined by its spectral line emission. The spectra typically include line emission from weakly ionized or neutral atoms, such as O, O+, N+, and S+. Conversely, the spectral line emission from strongly ionized atoms, such as O++, Ne++, and He+, is relatively weak. The class of galactic nuclei was first identified by Timothy Heckman in the third of a series of papers on the spectra of galactic nuclei that were published in 1980. Liner galaxies are very common; approximately one-third of all nearby galaxies (galaxies within approximately 20 to 40 Mpc) may be classified as Liner galaxies. Approximately 75% of Liner galaxies are either elliptical galaxies, lenticular galaxies, or S0/a-Sab galaxies (spiral galaxies with large bulges and tightly-wound spiral arms). Liners are found less frequently in Sb-Scd galaxies (spiral galaxies with small bulges and loosely-wound spiral arms), and they are very rare in nearby irregular galaxies.
We have presented different types of astronomical sources discovered for peculiarities in their spectral, temporal behavior, and/or morphology. It can be show (see High Energy Astrophysics course) that they are all part of the same class of astronomical objects called AGN (Active Galactic Nuclei):
Morphological classification is clearly subjective, if not arbitrary. The availability of large databases of galaxies with ample sets of well measured parameters (e.g. luminosity, mean surface brightness, colors, boxiness/diskyness, concentration, gradients, masses, subcomponents, spectral features, nuclear activity, crowding degree of the environment, etc.) allows (and requires) automated classification procedures. Techniques have been devised to this purpose which go from the estimators of the significance of a given parameter by a multivariate analysis to the exploration of the parameter space searching for clusters of objects; cf. A. Baillard et al., Project EFIGI: Automatic Classification of Galaxies, in Astronomical Data Analysis Software and Systems, XV ASP Conference Series, vol. 351, 2005.
Average size (top) and luminosity (bottom) of morphological galaxy types (adapted from Roberts and Hanes, 1994).
8. Spiral arms
10. Scale relations
14. Galaxy dynamics
19. Galaxy clusters
Elmegreen, D. M., and Elmegreen, B. G., Classification based on spiral arm appearance, Ap.J., 314, 3, 1987.
Sandage, A., Bedke, J., The Carnegie Atlas of Galaxies, Carnegie Institution of Washington Public. 638, 1994.
Nakamura, O., et al., Visual Hubble classification of 1500 SDSS galaxies, A.J., 125, 1682, 2003.
Ball, N., et al, Classification of SDSS galaxies using neural nets, MNRAS, 348, 1038, 2004.