1. Galaxy Classification In 1924, Edwin Hubble divided galaxies into different “classes” based on their appearance. Why begin here? Hubble classification serves as the basic language of the field. The morphological sequence reflects a fundamental physical and evolutionary sequence, which offers important clues to galactic structure, formation and evolution.

2. Hubble Tuning Fork diagram (Hubble 1936) Ellipticals Lenticular (S0) Spiral and Barred Spiral Irregular

3. Spiral Galaxies Disk + spiral arms + bulge (usually) Subtype a b c defined by 3 criteria: Bulge/disk luminosity ratio Sa: B/D>1 Sc: B/D<0.2 Spiral pitch angle Sa: tightly wound arms Sc: loosely wound arms Degree of resolution into knots, HII regions, etc.

4. Barred Spiral Galaxies Contain a linear feature of nearly uniform brightness centered on nucleus Subclasses follow those of spirals with subtypes a b and c

5. Elliptical Galaxies Smooth structure and symmetric, elliptical contours Subtype E0 - E7 defined by flattening En where n = 10(a-b)/a where a and b are the projected major and minor axes (doesn’t tell what the 3-D shape is)

6. Lenticulars or S0 Galaxies Smooth, central brightness concentration (bulge similar to E) surrounded by a large region of less steeply declining brightness (similar to a disk) No spiral arm structure Originally thought to be transition objects between Sa and E but typical S0 is 1-2 mags fainter than typical Sa, E (van den Bergh 1998)

7. Irregular Galaxies No morphological symmetry Lots of young, blue stars and interstellar material Smaller than most spirals and elliptical galaxies Two major subtypes: Irr I: spiral-like but without defined arms, show bright knots with O,B stars Irr II: asymmetrical with dust lanes and gas filaments (e.g. M82) - explosive M82-Irr II NGC 4485-Irr II Irr I

8. General trends within Hubble sequence E Sc: Decreasing Bulge/Disk Decreasing stellar age Increasing fractional gas content Increasing ongoing star formation Limitations of the Hubble Classification Scheme 1.Only includes massive galaxies (doesn’t include dwarf spheroidals, dwarf irregulars, blue compact dwarfs) 2.Three different parameters for classifying spirals is unsatisfactory because the parameters are not perfectly correlated. 3.Bars are not all-or-nothing. There is a continuum of bar strengths.

9. de Vaucouleurs’ Revised Hubble Classification System (de Vaucouleurs 1958, Handbuch der Phys. 53, 275) (de Vaucouleurs 2 1964, Reference Catalog of Bright Galaxies) Basic idea: retain Hubble system, but add lots of optional bells and whistles Mixed types: E/S0, Sab, Sbc Mixed barred/normal: SA (unbarred), SB (barred), SAB (in between) Inner rings:S(s) (arms out of ring), S(r) (arms in ring), S(rs) Outer rings:(R) S Extended spiral, irr types:Sm (between spiral and Irr), Im (magellanic), Sd (extreme Sc), Sdm (between Sd and Im) “t-types” scaleAdded in later editions of the Reference Catalog (de Vaucouleurs 2, Corwin 1976) E0  S0  Sa  Sb  Sc  Im -5 -1 1 3 5 10 (t-type)

10. No Bar Bar Spiral shaped Ring shaped Cross section of diagram E E+ S0- S0 S0+SaSbScSdSmIm Schematic Diagram of Revised Hubble Classification Limitations: E --- Im is not a linear sequence of one parameter Rings and bars are not independent Does not take into consideration mass or other important parameters. All based on optical surface brightness morphology.

11. Luminosity Classification or “DDO System” van den Bergh (1960) - In spirals and irregular galaxies, some properties correlate with galaxy mass rather than type. For spirals, the key parameter is arm development (i.e. arm length, continuity and width relative to size) Sc I - long, well-developed arms Sc III - short, stubby arms Sc IV - dwarf, spiral galaxy -faint hint of spiral structure Van den Berg speculates that all disk galaxies are born as gas-rich spirals and gradually evolve to anemic and finally S0’s. who was at David Dunlop Observatory in Ontario, Canada - hence the “DDO” Revised DDO - van den Bergh (1976): Placed disk galaxies into 3 parallel classes based on luminosity: Gas-rich, anemics and lenticulars Anemics have weak and diffuse spiral arms and low level of ongoing SF Parameters which change systematically from Lenticular to Gas-rich Mean stellar age Gas fraction Recent SF

12. Yerkes System (Morgan 1958) Utilizes fact that there is a strong correlation between the nuclear light concentration (how big the bulge is) and its integrated spectrum. Type is based on this one parameter - integrated spectral type. E, S0K-type spectrum SF-K stars dominate IrrA stars dominate Nomenclature: g S 2 Spectral type (dominant stars) Hubble type flattening (I.e. bulge/disk)E - elliptical 10(a-b) a, af, f, fg, g, gk, kD - S0 a S - spiral B - barred I - Irregular R - rotationally symmetric but no S or E structure

13. Kennicutt (1992) Galaxies shown in order of increasing Hubble type from top to bottom.

14. A couple of galaxy classes not addressed in these classification systems…. Leo I dSph Dwarf Spheroidals - dSph - overall star density in these systems is so low that on images with even the largest telescopes, they appear as mere clusterings of faint stars. The Sculptor system (Shapley 1938) was the first dSph galaxies to be discovered. dSph are the low-luminosity counterparts of the brighter dEs. Dwarf Ellipticals - dE - much less luminous than the standard elliptical galaxy. While giant Ellipticals are a few Mpc across and contain 1 trillion stars, dE can be as small as 1 kpc across and contain 1 million stars. NGC 205

15. Morphological Distributions The range and frequency of different morphological types is sensitive to the sample studied (more detailed discussion of this later…). Some key results: The Local Group is the only sample that includes a significant number of very faint galaxies. Of the 35 galaxies now considered members of the Local Group, only the 3 brightest (M31, MW and M33) are spirals, the remainder are equally divided between irregular and dwarf elliptical /spheroidal galaxies (Hodge 1995). Magnitude-limited sample of galaxies outside of clusters (in the “field”) are biased towards late-type (Sc) spirals. A typical field sample might consist of 80% S galaxies, 10% S0 galaxies, and 10% E galaxies. Within rich clusters, the population of bright galaxies is dominated by early-type systems (Dressler 1980). An intermediate density cluster will have 40% S galaxies, 40% S0 galaxies, and 20% E galaxies. A high density cluster will have 10% S, 50% S0, and 40% E.

16. Automated Classification Visual classification is inherently time consuming and different observers are unlikely to agree in ambiguous cases. This motivates the development of algorithms to automatically and impartially classify galaxy images - very important for large surveys like 2MASS and SDSS. Abraham et al. (1994, 1996): Concentration parameter C - fraction of light within ellipsoidal radius 0.3 x outer isophotal radius (1.5  above sky level). Asymmetry parameter A - fraction of light in features not symmetric wrt a 180 degree rotation Naim, Ratnatunga & Griffiths (1997) use 4 parameters: blobbiness, asymmetry, filling factor and elongation. Naim et al. (1995) used artificial neural nets to classify galaxies into the numerical T types. Achieved uncertainty of +/- 1.8 in T which is comparable to the dispersion between observers. For distant galaxies (greater than z=0.5), classification is difficult because of small angular size and apparent faintness of galaxies. HST “field” galaxies (z~1) classified by 2 humans (Ellis and van den Bergh) and A and C parameters of Abraham. For faint galaxies (I>21mag), C parameter alone is fairly good. For brighter galaxies, C is degenerate between E and S0.

17. Abraham et al. (1996)

18. The Gini Coefficient and M20 parameter Used in economics to measure distribution of wealth in population G = relative distribution of flux in galaxy’s pixels (Abraham et al. 2003) G=0 for completely egalitarian society (uniform surf brightness) G=1 for absolute monarchy (all flux in single pixel) 2 x area= G constant galaxy M20 = 2nd order moment of the brightest 20% of the galaxy - measures concentration

19. M20 G E/S0/Sa Sc/Sd/Irr Sb/Sbc Mergers more light in fewer pix more uniform surface brightness centrally concentrated less concentrated Lotz et al. 2005

20. ...but we haven’t seen the end of visual classification! No matter how good the automated classifications become, the human eye is still better at determining patterns than neural networks (e.g. detecting spiral structure, smoothness) Galaxy Zoo is a project to employ volunteers to classify galaxies imaged in the Sloan Digital Sky Survey About 250,000 people have participated in this project to visually classify about 400,000 galaxies. Each galaxy receives over 20 classifications and the results are used together to determine the true classification. Some results? go to http://www.galaxyzoo.org/