Stellar disks are the main structural features of disk galaxies, which we divide into the spiral, S0, and irregular morphological classes. The disks are highly flattened, with approximate circular symmetry. In spiral and irregular galaxies, the disk contains gas as well as stars, and fine structure is common, including the spiral arms which define that class. Our nearest disk galaxy is our own Mily Way, a spiral galaxy; the Sun lies in the plane of the disk, about 8 kpc from its center.

The stars and gas of a galaxy disk follow near-circular orbits around the center, in the plane of the disk. The disk does not rotate rigidly like a turntable, but differentially: stars in the outer regions take longer to complete an orbit than those close to the center. The orbital motion of the stars and gas supports the disk against collapse under gravity. Organized rotation dominates all other motions; random velocities account for very little of the kinetic energy, so that disks are dynamically ‘cold’, or at least ‘cool’. In the Milky Way, disk stars near the Sun move at about 200 km s–1, taking about 250 Myr to complete an orbit, but their random motions are typically only 30 km s–1. Because the disks are cool, they tend to be unstable to forming internal substructures such as bars and spiral arms.

Giant disk galaxies, with luminosities more than about 6x10exp9 L(solar) (our Milky Way is roughly three times brighter), are composite systems. The round inner bulge is much denser than the disk, and ‘hotter’; the stars have large random motions. Within the bulge is a yet denser star cluster, the galactic nucleus; some nuclei contain massive black holes at their centers. Giant galaxies follow a morphological sequence, named for its originator, Edwin Hubble. The S0 galaxies have large central bulges or spheroids, and a smooth stellar disk; spiral arms, gas and star formation are normally absent. Along the sequence from Sa through Sb to Sc, the central bulge becomes smaller, while the prominence of the spiral arms increases, along with the fraction of gas and newly formed stars. Dwarf galaxies are smaller, less luminous, and less massive than the giants. They are also more diffuse, with reduced densities of stars and gas, and they lack the dense round central bulges. At the boundary between dwarfs and giants are the Sd galaxies, with very thin stellar disks, and only rudimentary spiral structure. The trend towards less organized optical structure continues through the Sm Magellanic irregular class to the dwarf irregulars (dIrr), which are the least massive and least luminous, with blue luminosities below 10exp8 L(solar). All of these classes of dwarf disk galaxies are rich in gas, and have relatively large contributions from young stars.

Stellar disks

The Milky Way is the only galaxy where the three-dimensional structure is well known; elsewhere, we see only two-dimensional projections of the galaxies. By observing many galaxies, randomly distributed over all possible viewing angles, we deduce the average three-dimensional structures of the various classes. In disk galaxies that we view nearly face-on, perpendicular to the plane containing the disk, the projected brightness of starlight declines smoothly with radius. At a given wavelength λ, the run of intensity Iλ(r) with radius r from the disk center, is roughly exponential: Iλ (r) ~~ I(0) e–r/hr. The radial scale length hr is the e-folding length for the starlight integrated vertically through the disk. In most spiral galaxies, hr is 1–5 kpc; dwarfs can have hr<1 kpc while in some peculiar giant galaxies hr>10 kpc.

In most disk galaxies, near the center where r<hr, the luminosity rises above the inward extrapolation of the exponential law describing the main disk. We can define the disk central brightness Iλ(0) by extending the disk model to r = 0; the total luminosity of the disk is simply Lλ=2πhr2Iλ(0). Measured values of the central disk brightness are commonly around Iλ*(0)~~;150 L(solar) pc–2 in blue light; this is the Freeman peak. It is now recognized that the Freeman value reflects an approximate upper bound to disk surface brightness. While few galaxy disks have brighter centers, disks exist with lower surface brightness, ranging down to a few percent of the Freeman luminosity density. These are the low-surface-brightness galaxies. One caution: although it is usual to extrapolate the exponential law in this way, it is not clear if the stellar disks really extend into the central galaxy. We see our own galaxy, the Milky Way, as a luminous path across the sky on a dark night; the Sun is near the mid-plane of our disk, so we observe the stellar disk in projection from within. The disks of other galaxies also appear as narrow bands of light when we see them edge-on. If an edge-on disk contains only stars, with no interstellar matter to block light, we can use brightness maps to derive the distribution of stellar density ‘above’ the mid-plane in the z-direction. Stellar disks have complicated vertical structures. In our own Milky Way, the youngest stars are concentrated near the disk mid-plane, with exponential scale heights hz<0.1–0.2 kpc. Older stars form a ‘thick disk’ with a scale height hz~~1 kpc. The vertical scale height of each component of a stellar disk depends on its velocity dispersion perpendicular to the disk; higher speeds yield thicker disks, since the stars can travel further from the midplane. Young stars are born from clouds of gas that are dynamically ‘cold’; they have little vertical kinetic energy, and are confined close to the mid-plane. As they orbit, stars feel the gravitational forces from large lumps of gas and stars that are present in the spiral arms. The repeated tugs ‘heat’ them, increasing their random motions both vertically and in the plane of the disk.

In external galaxies, the vertical light distributions is often nearly exponential at large z, but flattens near the mid-plane. The average vertical scale heights derived from fits to the luminosity profiles of giant disks are smaller than the radial scale lengths, typically by factors of ~~5–10. Some ‘superthin’ Sd galaxies have even larger ratios, while irregular galaxies are typically fluffy: see figure 2. When we observe a galaxy disk edge-on, there is the additional complication that we integrate the light over a range of radii. The projected light distribution along the mid-plane z = 0 is not proportional to the surface brightness Iλ(r) that we would measure if the galaxy was seen face-on. For example, the radial brightness pro-file of an exponential disk that is edge-on will be fit by modified Bessel functions, which have a shallower drop-off with radius than the exponential distribution.

Gas disks

Except in the S0 galaxies, the disk contains gas as well as stars. The interstellar medium of a disk galaxy like the Milky Way is richly varied. It includes phases ranging from molecular cloud complexes with densities of ~10exp2 particles cm–3, to diffuse highly ionized gas with densities of ~10exp–2 cm–3 and temperatures >10exp6 K. The gas is also threaded with energetically significant magnetic fields, and bathed in the cosmic rays trapped by the field. Interstellar matter is perhaps most readily seen as dark clouds or lanes, where dust grains associated with the gas block our view of the stars behind. In the southern and northern Milky Way, dust is apparent as individual dark clouds, such as the ‘coal sacks’, along the plane of the bright disk. In optical pictures of edge-on disk galaxies, dust in the disk causes the characteristic dark band across the mid-plane.

Interstellar dust grains absorb and scatter starlight at wavelengths from the near infrared to soft x-rays. Interstellar dust grains are small— most have diameters between 0.001 µm and 1 µm—and dust opacity rises rapidly with decreasing wavelength. We get our best views of the distribution of stars in spiral galaxies in the infrared spectral region at λ~2 µm, the longest wavelengths where direct stellar radiation is still important. In the thermal infrared region, at wavelengths 5–300 µm, we see radiation from the dust grains themselves, warmed by the starlight that they have absorbed; most of the power is in the far infrared (FIR), at λ>50 µm. Images of galaxies at FIR wavelengths show emission where there are stellar (or other) heating sources, and the dust absorbs their light efficiently. The most spectacular star-forming disk galaxies, the STARBURST GALAXIES such as M82, contain so much dusty gas that it blocks starlight even at infrared wavelengths. There, the distributions of young stars can only be mapped from the FIR emission, and the radio emission of ionized gas near massive stars, or of the young remnants of supernova explosions.

A typical giant spiral galaxy has about 5–10×10exp9 Mo of cool interstellar gas. About half of this is clumped into dense regions, close to the midplane of the disk, where hydrogen is in molecular form (H2). Young stars are born in these cool molecular clouds. Since H2 is a symmetric molecule, it produces no strong emission, and its ultraviolet absorption bands become difficult to observe in dense regions of the interstellar medium. The locations and amount of H2 are usually derived from observations of tracer molecules which do produce radio frequency line emission, such as CO or HCN. Molecular gas is usually concentrated in the inner galaxy, where it forms a very thin disk, with scale height hz<0.1–0.3 kpc. The remainder of the interstellar gas is more diffusely distributed; most of it is neutral (H I) or ionized (H II) hydrogen. The spatial distribution of the H I in disk galaxies is the most easily measured, through the hyper-fine 21 cm emission line. This component is usually quite thin with vertical scale height hz<1 kpc, and a radial distribution that extends further from the center than the stellar disk. The outer regions of most galaxy disks appear to consist mainly of H I gas. Warm ionized gas, within 1–2 kpc of the midplane, is traced from the recombination emission lines of hydrogen. The disk is surrounded by a much hotter diffuse halo of gas at 10exp5–10exp6 K; we detect it in x–ray emission, or by ultraviolet absorption spectroscopy of resonance lines produced by heavier elements within the hot gas. The first unambiguous evidence for a giant halo of hot gas around a nearby spiral galaxy much like our own Milky Way was found by astronomers using NASA’s CHANDRA X-RAY OBSERVATORY. Chandra found a diffuse halo of x–ray gas, radiating at a temperature of almost 3 million degrees, around galaxy NGC 4631, approximately 25 million light years from Earth. The Chandra image reveals a halo of hot gas that extends for approximately 25 000 light years above the disk of the galaxy. One important feature of the x–ray emission from NGC 4631 is that it closely resembles the overall size and shape seen in the radio emission from the galaxy. This indicates that there may be a close connection between the outflows of hot gas, seen in x–rays, and the galaxy’s magnetic field, revealed by radio emission. A Hubble image of NGC 4631 shows filamentary, loop–like structures enclosing enhanced x–ray–emitting gas and emanating from regions of recent star formation in the galaxy’s disk. Together, these data clearly show that the hot gas is heated by clusters of massive stars and is now expanding into the halo of the galaxy.

The interplay between the gaseous and stellar components of disks is extremely complicated. A gaseous disk experiences dissipation, and can radiate away the energy associated with random motions, so that it tends to collapse in the vertical direction. However, energy and momentum are fed into the interstellar medium, by fast stellar winds, ionizing radiation, and supernova explosions from massive stars. These increase the internal energy of the gas, causing it to expand vertically; in extreme cases it can even escape as a galactic wind. Since stars are made from dense cool gas, star formation can be a self-regulating process. If many young stars are made, they can blow the gas away, or heat it so much as to prevent further starbirth.

Spiral arms and bars

The photogenic arms of spiral galaxies are the most striking luminous structures in any galaxy atlas. In a minority of giant galaxies, such as M81 or NGC 1300 (see figure 4), we see a ‘grand design’ spiral: a pair of well-defined arms winds over azimuth angles of more than 90° and extends in radius over much of the optically visible stellar disk. More frequently, the spiral arms are broken up into shorter segments, and more than two arms are present. In ‘flocculent’ disks, the spiral consists of many small arm fragments; there is no organized spiral structure on large scales. Gas seems to be essential to forming spiral arms; the S0 galaxies lack both gas and spiral pattern. Strong spiral arms form most readily in the rapidly rotating massive disks of luminous galaxies. Lower luminosity (and less massive) galaxies have messy spiral patterns, or none. The Sd systems often have ill-defined arms, while spiral arms are totally absent in the low mass dIrr galaxies. The decline in spiral arm intensity and frequency with decreasing galaxy mass suggests that gravity is important in shaping the spiral structure.

To the surprise of astronomers, the galaxy NGC 4622 appears to be rotating in the opposite direction to what they expected. Astronomers are puzzled by the clockwise rotation because of the direction the outer spiral arms are pointing. Most spiral galaxies have arms of gas and stars that trail behind as they turn. But this galaxy has two leading outer arms that point toward the direction of the galaxy’s clockwise rotation. To add to the conundrum, NGC 4622 also has a trailing inner arm that is wrapped around the galaxy in the opposite direction it is rotating. Astronomers suspect that NGC 4622 interacted with another galaxy. Its two outer arms are lopsided, meaning that something disturbed it. The new Hubble image suggests that NGC 4622 consumed a small companion galaxy. The galaxy’s core provides new evidence for a merger between NGC 4622 and a smaller galaxy. This information could be the key to understanding the unusual leading arms.

Spiral arms can be characterized by their pitch angle, the angle at which the arm cuts through a circle around the galaxy center. Arms in Sa galaxies are tightly wound, with small pitch angles, while those in later-type spirals are more open, with pitch angles of 10–30°. Spiral arms generally trail, with the tips of the arms pointing opposite to the direction of galactic rotation. In figure 4, we see dark dust lanes on the concave sides of the spiral arms; these show where the disk gas is compressed as it flows into the arms. Thus the spiral stands out when observed in H I, or tracers of molecular gas such as CO emission. The bright young stars which make the spiral arms so prominent have been formed from this compressed gas.

Massive stars have short lives; those hot enough to ionize the gas around them last less than 5 Myr, and lifetimes are generally <30 Myr. This is only about 10–20% of a galactic rotation period. So the hot stars that we see in the grand-design spiral arms must have been born in or near them; if they were made elsewhere in the disk, they would die before they could be concentrated into the spiral arms. A classic problem is whether spiral arms stimulate star formation, or simply act to organize it. In the former case, we would expect star formation to be more vigorous in galaxies with grand-design spiral patterns than in flocculent galaxies. But observational comparisons show no significant difference in the rates; the arms seem to act more as star formation organizers.

The young stars in spiral arms radiate 10–100 times as much power per unit mass as older stellar populations, in which the most massive stars are no longer present. Even small fractions of young stars can have major observable consequences. These effects are most pronounced in blue and ultraviolet light, where these hot stars produce most of their luminosity. Young massive stars also photoionize the gas from which they were born, producing brilliant ionized nebulae, known as H II regions. The H II regions are excellent tracers of spiral structure, since they are easily observed in their bright atomic emission lines, from hydrogen recombination and collisionally excited lines of common elements such as nitrogen, oxygen, and sulfur.

About half of all disk galaxies show a central linear bar, containing up to a third of the total light. The ratio of the long to the short axis of the bar can be as extreme as 1:5. Gas is not required for a bar: the disks of S0 galaxies are as likely to be barred as are the gas-rich spiral galaxies. In Hubble’s classification, ‘B’ is added to indicate a bar; NGC 1300 is of type SBb. Bars are the offspring of stellar disks; they appear to be vertically thin, like the disk, and not round like the bulge. The main bars of giant galaxies typically extend to radii of a few kilo-parsecs.

Bulges and nuclei

In less luminous Sc and Sd galaxies, and some dwarfs, the approximately exponential disk extends to the center of the galaxy; for these pure disks, the central surface brightness is simply Iλ(0). Low-luminosity galaxies are often asymmetric, with no well-defined center; e.g. the brightest region may not lie at the kinematic center of the system. The Large Magellanic Cloud is a prime example of an off-center disk galaxy; here, the center of rotation is outside the bright stellar bar. The irregular galaxy NGC 55, also shows an off-center concentration of light.

By contrast, most giant disk galaxies are reasonably symmetric. Extra light is also present in the inner few kiloparsecs, above that expected from the main disk. This is the central bulge: a spheroidal stellar system normally consisting of stars with ages >=3 Gyr. Bulges typically contain little gas, except near their centers. Spheroidal bulges are rotating, but the large random stellar motions allow them to be thicker in the z-direction, perpendicular to the disk midplane, than the surrounding stellar disk. The ‘Sombrero’, shown in figure 3, is an outstanding example of a disk embedded in an unusually large oblate stellar bulge. Smaller bulges, like those of the Milky Way or our neighbor Sb galaxy M31, are much more common.

By contrast, some spirals have inner disks. Like bulges, these small disks have much higher surface brightness than the main stellar disk; but they are relatively flat, roughly axisymmetric, and dynamically ‘cooler’ than spheroidal bulges. Inner disks often support star formation and have their own spiral structure.

Images of the centers of disk galaxies taken at high angular resolution reveal a variety of structures. In some galaxies, small nuclear bars are present, often nested within a larger main bar; these have sizes of <1 kpc. The available evidence suggests the nuclear bars are independent entities, that do not rotate at the same rate as their main bars. At or near the very center is the nucleus, a tiny (r<10 pc) object containing the highest density of stars in the galaxy, up to 10exp6 L(solar). pc–3. Sometimes a massive black hole, with MBH>10exp6 Mo., is also present as well. When gas or stars are swallowed by the black hole, huge amounts of gravitational energy can be released, producing a luminous active galactic nucleus (AGN) that can compete with the power output of all of the stars in the galaxy. However, most nuclei are not AGNs, and derive much of their luminosities from the stars they contain. As a result they are relatively faint, with optical luminosities of 10exp6–10exp8 L(solar).

The stars of a galaxy move under gravity; other forces are rarely important. In the disk, stars are widely separated in space, so we have the additional simplification that close collisions between them are rare, and can be safely ignored. Thus their motions can be studied as a response to the galaxy’s mean gravitational field, which must be computed self-consistently from the mass distribution; this is a challenge, since the system has ~10exp11 particles. By contrast, gas responds both to gravity and to hydromagnetic forces. Since dense gas clouds do collide, they can dissipate the energy of random motions. Because of this, cool disk gas rapidly settles onto closed periodic orbits about the galaxy center; in an axisymmetric disk, these are the circular orbits.

Rotation and mass distribution

We can use Doppler-shifted emission lines from the gas to find the mass of a disk galaxy. If we measure the speed V(r) of gas in a circular orbit at radius r, we can use the radial-force equation V2(r)/r = GM(<r)/r2 (which is exact in a spherical system) to make a rough estimate of the mass M(<r) within that radius. In giant disk galaxies, V(r) climbs steeply over the inner kiloparsec or so, to a maximum that is typically at Vmax~150–300 km s–1, and then stays roughly constant. In dwarfs, the rise is more gradual and peak speeds are lower.

We can compare the observed rotation curves of disk galaxies with what we would expect if their mass had been concentrated entirely in the stars and gas. Even when the assumed ratio of mass to light in the disk is adjusted so that gas and luminous stars account for as much of the galaxy’s rotation as possible, we would expect the rotation speed to fall beyond 2–3 times the scale length hr of the exponential stellar disk. Instead, we usually see that V(r) remains approximately constant to the edge of the gas disk, implying that M(<r) continues to rise; the outer parts of the galaxy contain much mass but emit little light. A spherical halo of dark matter is generally invoked to provide the required extra mass. Dark matter usually accounts for between 50% and 90% of a disk galaxy’s total measured mass, with dwarf galaxies generally having a higher proportion than giants.

More luminous galaxies rotate faster on average, which tells us that they are more massive. The peak rotation speed Vmax increases with the galaxy’s luminosity L, roughly as L : Vmax α, with α~4: this is the Tully–Fisher relation. If galaxies contained no dark matter, we might be able to understand the TULLY–FISHER RELATION. But since the speed Vmax is set largely by the unseen material, while the luminosity comes mainly from the stellar disk, the link between them is puzzling. Nevertheless, the Tully–Fisher relation is very useful in finding distances to galaxies and galaxy groups, since we can estimate the true luminosity of a galaxy from its measured rotation speed.

Disk stability

Galaxy disks live on the edge of gravitational instability; galactic bars, and the spiral arms, are the consequences. The mutual gravitational attraction of disk stars and gas clouds tends to pull them together. On the other hand, since the period of an orbit around the galaxy center increases with radius, the disk’s rotation tries to shear any feature into a trailing spiral arm segment. In a ‘cool’ disk, where the random motion of the stars is small enough that it does not take them outside the spiral arms, their gravity can reinforce a spiral pattern by attracting other stars into it, and so help the spiral to grow. But if the disk contains little mass, or its stars have large random motions, then gravity will be insufficient to amplify a spiral, and no strong arms will develop.

Locally, an axisymmetric disk is unstable, and a spiral will grow, if the Toomre ‘Q’ parameter Q = κσR/3.36GΣ<~1. This criterion specifies that the dispersion σR of stellar velocities in the radial direction must be small enough by comparison with the surface density Σ of mass in the disk; κ is the epicyclic frequency with which a star oscillates radially about the nearest circular orbit, and G is the gravitational constant. Computer simulations of a ‘cool’ axisymmetric disk of ‘stars’ attracting each other by gravity, and initially following nearly circular orbits, show that a spiral pattern generally grows if Q<~1. As it does so, the stars develop larger random motions, and Q rises; so we never expect to see a stellar disk with Q<1. Near the Sun, the density in the disk Σ~50 Mo. pc–2, and κ~~35 km s–1 kpc–1, while for stars about as old as the Sun, the velocity dispersion σR ~~30 km s–1. Hence Q~1.4, which is safely greater than unity.

The simulated disks generally grow a two-armed spiral, and often a straight central bar in addition. The ‘stars’ stream through the spiral, slowing down and lingering in its gravitational potential well, and then pass on through it; the spiral is a density wave, a stellar traffic jam. We believe that the same is true in real galaxies. The pattern of the arms turns in the same sense as the disk’s rotation, at some angular rate Ωp, which is set by the angular frequency of rotation Ω(r) and the epicyclic frequency κ(r) in the region where the spiral is strong.

Why do the arms of a spiral galaxy trail, pointing against the direction of galactic rotation? When a trailing spiral is present, gravitational forces cause the inner parts of the disk to exert a torque on the outer disk, which transfers angular momentum outward and allows material at small radii to move inward. The spiral torques decrease the energy of the disk’s rotational motion, transferring it to increase the random speeds of the stars. As the disk ‘heats up’, the spiral eventually disappears. The disk of an S0 galaxy, where no gas is present, would probably behave in the same way: any spiral pattern would be short lived. But in a gas-rich disk, stars freshly born from the cool gas have very small random speeds. Continued addition of these new stars is probably important in prolonging the life of a spiral pattern, or in re-creating it periodically.

The gravitational pull of the spiral arms affects gas even more strongly than the stellar disk, because the random speeds of gas clouds are only 5–10 km s–1, much lower than for the stars. Except immediately around the corotation radius, where the angular speed Ω(r) of a circular orbit is close to Ωp, the linear speed r[Ω(r)–Ωp] with which gas moves into the spiral arm is supersonic. Shocks develop, compressing the gas enormously as it flows into the arm; we see dark dust lanes there. After about 10 Myr, young stars are born in the compressed gas and begin to shine; the Hα emission from gas ionized by these stars is not concentrated on top of the dust lane, but ‘downstream’ of it. Radiation from the hottest of these stars also splits some of the H2 molecules apart into atomic hydrogen, H I.

Like spirals, the figure of a galactic bar is not static, but rotates with some pattern speed Ωp. Unlike spiral arms, a bar truly captures its stars. Within the bar, stars and gas no longer follow near-circular paths, but are trapped near elongated orbits that close on themselves as seen by an observer moving with the bar’s rotation. Disk gas cannot flow in toward the galaxy center unless it can get rid of its angular momentum; the strongly asymmetric gravitational forces of a bar help it to do just that. The closed oval orbits converge on each other near the ends of the bar. As gas on the orbits approaches those regions, shocks form. The gas is compressed, and we see dark dust lanes. In the shock, the gas loses part of its energy of forward motion as heat, so it drops down onto more tightly bound orbits closer to the center of the galaxy. Bars may be important in feeding gas in towards the central galaxy.

Near the Sun, the oldest disk stars are about 10 Gyr old. By examining the number of stars of each mass that are present today, we can deduce that the Milky Way’s disk has been making new stars at a roughly constant rate over the last few gigayears. We live in an Sc (or Sbc) spiral galaxy; the disks of Sd galaxies, and the irregulars, are bluer than ours, implying a larger proportion of newly formed stars. Either their rates of star formation have increased recently, or they simply began their star-forming histories later. The redder colors of Sa and S0 galaxies tell us that star formation in their disks is decreasing; by now, most or all of their gas supply is used up.

Near the Sun, younger stars contain a larger proportion of elements heavier than helium. The disk has formed successive generations of stars, each producing metals to contribute to the already enriched interstellar gas. In all but the lowest-luminosity disk galaxies, we see a color gradient; the centers are redder than the outer parts. This is because the inner parts contain both a larger proportion of old stars, and stars that are richer in heavy elements. The bulges and inner disks have already made many generations of stars, while the outer disks have lagged behind, producing their stars more slowly, and remaining gas-rich to the present day.

Galaxy formation: inside-out or outside-in?

How might a galaxy like our own Milky Way come into being, out of the hydrogen and helium gas resulting from the Big Bang? Some would argue that giant disk galaxies have grown outward from their central parts. The dense inner bulge, with its low specific angular momentum, formed early on, from a denser-than-average lump of gas that collapsed under its own gravity; it rapidly used up all its gas in forming stars. The disk material, which has more angular momentum per unit mass, would have fallen in later, from greater distances. It must have remained gaseous until almost all its random motion had been dissipated, allowing the material to move inwards and settle onto circular orbits in a thin disk. Only then could the oldest disk stars have been born. Perhaps the disks are still being assembled today, as clouds of cool gas fall in, renewing the supply of raw material to make new stars. The dark matter, unable to dissipate energy, could not fall inwards; hence the dark halo is less concentrated to the galaxy center than are the luminous stars and gas.

Other astronomers point out that the stars of the bulge do not seem to predate the oldest disk stars near the Sun. They offer an alternative idea: the disk formed first, and the bulge grew within it. The disk gas would have flowed slowly inwards, perhaps coaxed towards the center by spiral arms or a galactic bar. When the inner part of the disk had reached a sufficient density, it would have become unstable and formed a bar, which then puffed up vertically into a bulge. This model has received some support from computer simulations. Also, in some bulges the gas is seen to follow oval orbits similar to those in galactic bars, as expected if these bulges are in fact thickened bars. The ‘inside-out’ enthusiasts retort that the bulge of a galaxy like the Sombrero is much more luminous than the disk, and so is unlikely to be an outgrowth from it. There may be no single route to making a disk galaxy.