. 16
( 28)


observations of interstellar molecules tell us
about the cold dense component. Because of the
great distance to the galactic center, many obser-
vations are limited by poor angular resolution.
In the past several years, a major breakthrough
has resulted from VLA and VLBI observations.
The development of sensitive infrared array
detectors has also been very important. Maps of
radio emission and X-ray emission are shown in
Fig. 16.16.
Studies of the ionized gas show a bent arc of
0 5 10
emission perpendicular to the galactic plane. This
structure is about 15 arc min in extent. It shows
Fig 16.15. (Continued) (b) The fourth quadrant, in a
a filamentary structure. This feature is also seen
galactic longitude range 280° to 330°.These clouds form
in the infrared. The emission seems to be a com-
part of what is called the Carina arm. [(a) Kathryn N. Mead;
bination of thermal (free“free) radiation and
(b) Yasuo Fukui, Nobeyama Radio Observatory]
non-thermal (synchrotron) radiation. The pre-
sense of extended X-ray emission (Fig. 16.16b),
suggests temperatures as high as 107 to 108 K. It
galaxies. Therefore, we will leave further expla-
is speculated that these high temperatures result
nation of spiral structure until Chapter 17, when
from a past explosive event. This event may have
we look at other galaxies.
been a number of supernova explosions follow-
ing an unusual wave of star formation. (We will
16.6 The galactic center discuss these “starbursts” more in Chapter 19.) At
the higher end of that range, the gas would not
16.6.1 Distribution of material near be bound to the galaxy.
the center The molecular material in the inner 200 pc is
remarkably hot and dense, compared with giant
Ever since it was realized that the galactic center
molecular clouds elsewhere in the galaxy (Chapter
is someplace else from where we are, astronomers
15). Typical temperatures are 70 K, and densities
have wondered about its nature. Is it simply a geo-
are greater than 104 cm 3. These are conditions
metric location, or is it the site of unusual activ-
found in normal molecular cloud cores, but they
ity? We will see in Chapter 19 that the centers of
are found over the full extent of the clouds. The
many other galaxies are the sites of unusual
internal velocity dispersions are up to 50 km/s,
activity. This makes the search for activity in our
much greater than even in molecular cloud cores.
galactic center a natural focus. When we talk
The amount of molecular material may be as high
about the center region, we are talking about
as 108 M . It has been suggested that the ionized
roughly the inner 500 pc of the galaxy.



Fig 16.16. Maps of the galactic center. (a) VLA map of
radio emission at a wavelength of 6 cm. An arc like structure
is visible. Much of this pattern is viewed as coming from a
tilted ring of ionized gas. (b) Chandra image of X-ray emis-
sion from hot gas hear the galactic center. (c) The blue image
shows the X-ray emission relative to the radio emission,
which is shown in red. [(a) NRAO/AUI/NSF; (b), (c) NASA]

wider than in the normal interstellar medium.
However, a variety of techniques suggest very
high field strengths, of the order of a few milli-
gauss. (Remember, typical interstellar fields are
tens of microgauss.)
The galactic center provides a totally unique
environment for star formation. There is a high
density to help, but it is inhibited by higher tem-
peratures, velocity dispersions, magnetic pressure,
and tidal effects. Observations (such as those
shown in Fig. 16.17) tell us that the central parsec
gas filaments are the boundaries of dense molecu-
is very rich in past and current star formation.
lar clouds, which have been exposed to the intense
The star formation rate in the central region is
ionizing radiation near the center.
estimated to be as high as 1M /yr. The central
If we are to understand the dynamics of this
cluster shows evidence for a new 104 M superim-
unusual environment, we must also know the
posed on an older 106 M cluster. It has been sug-
magnetic field strength (and arrangement). As
gested that the formation of the newest cluster
we discussed in Chapter 14, magnetic fields are
triggered the formation of at least two other rich
generally hard to measure. For the galactic cen-
nearby clusters. (We discussed triggering of star
ter, even the Zeeman effect in HI is hard to
formation in Chapter 15.)
measure because the spectral lines are so much

Fig 16.17. HST near IR images
of the galactic center, showing rich
star clusters. [STScI/NASA]

16.6.2 A massive black hole? racy, in their case to within 0.002 arc sec. When
we have previously talked about using velocities
There has been considerable speculation on the
to measure the mass of a system, in using the vir-
nature of the central object. It has been suggested
ial theorem for example, we used radial velocities
that it might be a few million M black hole. In
as measured from Doppler shifts. We could have
establishing the existence of such a black hole,
used proper motions, but they are generally too
there are two observational steps that must be
small to measure over a period of a few years.
made. First, we must accurately determine the
Velocities near a massive black hole would be large
mass of the object. The best way to do this is by its
gravitational influence on nearby surrounding
stars or gas. Second, we must show that the object
is smaller than the Schwarzschild radius for that
You can get an idea of how difficult this is. By
equation (8.10), the Schwarzschild radius for a
106 M black hole is 3 106 km, which is about
10 7 pc, and would subtend an angle of about
10 6 arc sec.
One interesting approach to this problem was
started in 1995 by Andrea Ghez (Fig. 16.18) of UCLA,
and co-workers, using the Keck 10 m telescope
(described in Chapter 4). Their goal was to meas-
ure the proper motions of stars in the direction of
SgrA*, the source in the center of the galaxy.
Ghez™s group observed in the near infrared
(2.2 m), which provided good angular resolution,
but still had less extinction to the galactic center
than they would have had in the visible. To obtain
the best possible resolution, they took a number
of short exposures, so there was little atmospheric
smudging in each image. In adding the images
together, they shifted the image to remove changes
in atmospheric refraction from image to image.
The final images had a resolution of about 0.5 arc
sec. If the diffraction pattern is very clean (and it
is for their system) then it is possible to measure
Fig 16.18. Andrea Ghez. [Andrea Ghez, UCLA]
the position of each star to much greater accu-

enough to produce measurable proper motions. With three observing runs, a year apart, Ghez™s
Measuring Doppler shift for these stars would be group were able to identify 90 stars with proper
hard, because they required such short exposures motions large enough to measure. The largest
to achieve good angular resolution. Their results proper motion corresponded to a tangential veloc-
are summarized in Fig. 16.19. ity of 1400 km/s. They found that this motion is
organized about the position of SgrA* to within
0.1 arc sec. The projected positions and velocities
0.4 are consistent with Keplerian motion. This
means that essentially all of the mass is closer to
the center than any of the stars measured. The
∆DEC from Sgr A* (arcsec)

best estimate of the mass in the central object
106 M . The observations don™t
is (2.6 0.2)



Velocity Dispersion (km/sec)

1996 400
’0.4 1997

’0.2 ’0.4
0.4 0.2 0
∆RA from Sgr A* (arcsec)

∆DEC from Sgr A* (arcsec)

0.01 0.1
Radius (pc)


Minimum Mass (Mo)

4 2 0
∆RA from Sgr A* (arcsec)
Fig 16.19. (a) Measured positions for three observing
years near the position of SgrA*. (b) Positions of the 90 stars
whose proper motions were measured, with the size of the
symbol scaled to the size of the motion. (c) Projected stellar
velocity dispersion as a function of distance from SgrA*.The
solid line is the value for Keplerian orbits (still outside all of 0.01 0.1
Projected Radius (pc)
the mass). (d) Enclosed mass as a function of projected dis-
tance for 30 stars. (d)

By following these stars for another two years
(through 1999), the group found three stars whose
paths had a measurable curvature about the SgrA*
position (Fig. 16.19e). These stars had projected posi-
tions on the sky 0.005 pc from SgrA*. From this
curvature, they could directly measure the acceler-
ation in the circular orbit. It turns out to be numer-
ically close to the value of the Earth™s acceleration
in its motion about the Sun. The shortest orbital
period is 20 years, so we have the prospect of being
able to watch something make a complete revolu-
tion about the galactic center in our lifetimes.
Acceleration vectors don™t allow for a better mass
estimate, but if you use the mass estimate from the
proper motions, the volume the mass is con-
strained to is decreased by an order of magnitude.
This greatly strengthens the case for that mass to
(e) be included within its Schwarzschild radius.
There is also indirect evidence for explosive
Fig 16.19. (Continued) (e) Curvature in the paths of
activity in the galactic center region. For example,
three stars. [Andrea Ghez, UCLA; (a)“(d) Ghez, A. M. et al.,
there is an armlike feature in our galaxy, some 3
Astrophys. J., 509, 678, 1998, Figs. 3“6; (e) Ghez, A. M. et al.
to 4 kpc from the galactic center, called the “3 kpc
Reprinted by permission from Nature, 407, 307, Fig. 1,
Copyright (2000) Macmillan Publishers Ltd.] arm”, which appears to be expanding at about
50 km/s. Speculation is that this expansion is due
to some explosion in the relatively recent past (see
conclusively show that this mass is contained Problem 16.14). We see other similar features
within RS, but alternative explanations, such as a closer in, suggesting that this activity has taken
very rich cluster, don™t seem very likely. These place on a continuing basis. We will see in
masses are consistent with previous determina- Chapter 19 that the activity in our galactic center
tions using other techniques, but the new meas- is small compared with that in many galaxies.
urements place much tighter constraints on the However, it gives us our best opportunity to study
confinement of the matter, and the coincidence such activity “close up”. For this reason, study of
with SgrA*. the galactic center is a very active field.

Chapter summary
In this chapter we saw how stars and interstellar In looking at the average gas distribution, we
material are arranged in the Milky Way. found that the HI is extended beyond the Sun™s
We saw how the rotation curve tells us about orbit, while the number of molecular clouds
the mass distribution in the galaxy, and how that falls off more sharply with distance from the
rotation curve is determined. Different tech- center of the galaxy. The molecular clouds are
niques are needed for material inside and outside also more tightly concentrated toward the galac-
the Sun™s orbit about the galactic center. We saw tic plane than the HI clouds. This concentration
how the rotation provides evidence for dark mat- toward the plane suggests that the molecular
ter. Once the rotation curve is known, velocities clouds are younger.
of objects can be used to estimate distances. This We discussed evidence for spiral structure in
works better outside the Sun™s orbit than inside, the Milky Way and the difficulties in tracing out
because of the distance ambiguity. spiral arms in our galaxy. We saw that tracers for

spiral structure include HII regions, OB associa- has a mass concentration which may be a few
million M black hole.
tions and molecular clouds.
In looking at the galactic center, we found
that it contains a small, active region. The center

16.1. How do we know that v(R) tells us the total 16.5. Contrast the method of measuring the rota-
mass interior to R? tion curve inside the Sun™s orbit with that
*16.2. Why should the kinematic and dynamical for measuring it outside the Sun™s orbit.
definitions of the local standard of rest give 16.6. Describe the difficulties in studying spiral
the same rest frame? structure in our galaxy.
16.3. Why is it sufficient to measure the rotation 16.7. Compare the methods for using GMCs and
curve only for subcentral points for R R0? HII regions to study spiral structure.
*16.4. Why can™t we use transverse velocities to 16.8. Why is it easier to study spiral structure in the
measure the rotation curve of our galaxy? outer part of our galaxy than in the inner part?

16.1. Show that the radial velocity equation of sight. Show that the component of their
(16.11), derived for the first quadrant of the velocity along the line of sight is the same.
galaxy, holds for the second quadrant. 16.9. Show that the radial velocity of a point in the
*16.2. For material observed at a radial velocity vr fourth quadrant is the negative of that of the
along a line of sight at galactic longitude , corresponding point in the first quadrant.
find an expression for the separation between 16.10. If we can only measure kinematic distances
the near and far points producing that vr. for points with vr 10 km/s, what range of R
*16.3. Calculate (R) and v(R) for the following den- can we study?
sity models: (a) all the mass M at the center 45 , we observe vr
16.11. For 30 km/s.
What are R and d?
of the galaxy; (b) a constant density, adding
up to a mass M(R0) at the Sun™s orbit and no 16.12. If v(R) 220 km/s for R0 R 2R0, find an
expression for vr as a function of ( , R).
mass beyond.
16.13. When the constants R0 and v0 changed from
16.4. Suppose the rotation curve of the Milky Way
is flat out to 2R0. What mass does that imply 10.0 kpc and 250 km/s to 8.5 kpc and
out to that distance? 220 km/s, respectively, by what factor did
16.5. Convert the HI and H2 masses given in the the mass inside the Sun™s orbit change?
chapter into: (a) average volume densities 16.14. What is the Schwarzschild radius for a 5
106 M black hole? How does this compare
and (b) average surface densities for the
regions R R0 and R0 R 2R0 (assuming with the sizes of the structures found in the
that all of the mass outside R0 is between R0 galactic center?
and 2R0). *16.15. The mass distribution for material near the
*16.6. For material outside the Sun™s orbit, derive galactic center can be determined from stud-
an expression to convert observed vr and d ies of the rotation curve close to the center.
into (R) and R. When the rotation velocity is plotted vs. log R
16.7. (a) Estimate the age of the 3 kpc arm from for the inner 10 pc, the result is approximately
a straight line. v(R) is 200 km/s at r 2 pc
its radius and expansion speed. (b) Is this an
and 70 km/s at R 10 pc. (a) Use this data to
upper or lower limit to its age? Explain.
find M(R) for 2 pc R 10 pc. (b) Assuming
16.8. Draw a diagram showing two points with
the same radial velocity along the same line a spherical distribution, find (r).

Computer problems

16.1. Plot a graph of Doppler shift as a function of dis-
tance along a line of sight for 45 and for
135 .
16.2. Plot a graph of vmax vs. for ranging from 0 to
90 degrees.
16.3. Complete the following table by giving kinematic
R and d for the indicated and vr. (In case of a dis-
tance ambiguity, give both values.)


30 30
30 30
60 30
60 30
120 40
240 40
330 30
330 30
Part V
The universe at large
To this point we have been studying the stellar life cycle and how stars
and other material are arranged in the Milky Way Galaxy.We will now
turn to studies on a much larger scale.We will ¬rst look at other galaxies,
and see that some of them tell us more about our own galaxy, which is so
hard to observe.When we talk about how the universe is put together,
each galaxy has only as much importance as a single molecule of oxygen
has in describing the gas in your room.
As we go to larger scales, we will look at how galaxies are distributed
on the sky, and how they move relative to one another.We will also see
how the problem of dark matter becomes more important as we go to
larger and larger scales.
As we go to larger scales, increasing the number of galaxies that we
observe, we also ¬nd a variety of interesting phenomena associated with
galaxies. In Chapter 19 we will discuss aspects of galactic activity, particu-
larly as evidenced by radio galaxies and quasars.
In Chapter 20 and 21 we will turn to cosmology, the study of the uni-
verse on the largest scales.This also includes the past and future evolution
of the universe. It is in the study of the past that we encounter one of the
most fascinating aspects of modern astrophysics research, the merging of
physics on the smallest (elementary particles) and largest (structure of the
universe) scales.
Chapter 17

Normal galaxies

Furthermore, he found galaxies that do not have
Our study of the Milky Way has been aided greatly
a spiral structure. Hubble classified the galaxies
by studies of other galaxies. However, for a long
he studied according to their basic appearance. It
time it wasn™t clear that the spiral nebulae we see
was originally thought that the different types of
in the sky are really other galaxies. From their
galaxies represented different stages of galactic
appearance, it might just be assumed that these
evolution. (Similarly, some astronomers thought
nebulae are small nearby objects, just as HII
that different spectral type stars along the main
regions are part of our galaxy.
sequence were evolutionary states of the same
The issues were crystallized in 1920 in a debate
star.) We now know that this is not the case.
between Harlow Shapley and Heber D. Curtis. Curtis
However, Hubble™s classification scheme, depicted
argued that spiral nebulae were really other galax-
in Fig. 17.1, is still quite useful.
ies. His argument was based on some erroneous
assumptions. First, he confused novae in our galaxy
17.1.1 Elliptical galaxies
with supernovae in other galaxies. Shapley
Elliptical galaxies have, as their name suggests,
thought the spiral nebulae were part of our own
simple elliptical appearances. Some examples of
galaxy, partly based on an erroneous report of a
ellipticals are shown in Fig. 17.2. The ellipticals are
measurable proper motion for some nebulae.
classified according to their degree of eccentric-
The issue was settled in 1924 by the observa-
ity. The ones that look spherical (zero eccentricity)
tional astronomer Edwin Hubble (after whom the
are called E0, and the most eccentric are called E7.
Space Telescope is named). Hubble studied Cepheids
The most common type of elliptical galaxies
in three spiral nebulae (including the Andromeda
are called dwarf ellipticals, since they are also the
Galaxy), and clearly established their distance as
smallest. Their sizes are typically a few kilopar-
being large compared with the size of the Milky
secs and their masses are a few million solar
Way. There is some problem with Hubble™s analysis,
masses. More spectacular are the giant ellipticals,
involving type I vs. type II Cepheids. However, even
with extents up to 100 kpc and masses of about
this factor of 2 error in the distance was not enough
1012M , with some with masses up to a factor of
to alter the basic conclusion that spiral nebulae are
ten higher.
not part of our own galaxy. Following this work,
The gas content of ellipticals is low. Studies of
Hubble made a number of pioneering studies of
HI, using the 21 cm line emission, as well as IRAS
other galaxies, essentially opening up the field of
observations of weak emission from their dust,
extragalactic astronomy.
suggest that the mass of the interstellar medium
may be up to about 1% of the mass of the stars
17.1 Types of galaxies that we see. The low gas content rules out the pos-
sibility that ellipticals eventually flatten to form
spirals. The continuing process of star formation in
In his studies Hubble realized immediately that
a galaxy depletes its supply of interstellar matter,
not all spiral galaxies have the same appearance.

metal abundances are not low. Giant ellipticals
have metal abundances that are quite high, about
twice the solar value.
Some ellipticals are rotating very slowly. They
have a higher ratio of random velocities to rota-
Sa Sb Sc
tional velocities than do spirals. We think that
their slow rotation means that they could
collapse without much flattening. Remember,
E0 E4 E7
when we discussed collapsing interstellar clouds,

Barred Spirals
Fig 17.1. Hubble classi¬cation of galaxies. Ellipticals range
from E0 (round) to E7 (the most oblate).The regular spirals
are divided according to the relative size of the nucleus and
the disk, and the tightness of the spiral arms.The Sa have the
largest nuclei and the most open arms.The barred spirals,
SB, follow the same classi¬cation as the normal spirals. S0
galaxies have nuclei and small disks but no spiral arms.

so if spirals are merely evolved ellipticals, we have
no way of understanding why spirals have more
gas and dust than ellipticals.
Ellipticals generally contain an evolved stellar
population, with no O or B stars. However, their


Fig 17.2. Elliptical galaxies. (a) M87, in Virgo, which is a
giant elliptical, type E0.The fuzzy patches visible near the
edge of the galaxy are globular star clusters.The inset shows
a blow-up of the center. (b) M49, in Virgo, type E1. It is about
15 Mpc away, and about 15 kpc across. (c) M32 in
Andromeda, which is a dwarf elliptical companion to the
Andromeda Galaxy (M31, Fig. 17.3b) and is type E2. It is only
800 pc across. [NOAO/AURA/NSF]

we found that rapid rotation retards collapse to galaxies that are so far away that we cannot
perpendicular to the axis of rotation, resulting in distinguish individual stars of HII regions. As sug-
gested by Sydney Van den Bergh, astronomers have
the formation of a disk.
taken to adding a luminosity class to the spiral
We can use photometry to study the bright-
ness distribution in ellipticals. Since we see the
galaxy projected as a two-dimensional object on
the sky, it is convenient to speak of the luminos-
ity per unit surface area L(r), where r is the pro-
jected distance from the center of the elliptical.
Studies show that the light from most ellipticals
can be described well by a simple relationship
(known as de Vaucouleur™s law):

L10 2 e 1r r0 2 1 4
L1r2 (17.1)

In this expression L(0) and r0 are constants. The
values of L(0) are found not to vary very much,
with a typical value of about 2 105 L /pc2. The
values of r0, however, show a very large spread.

17.1.2 Spiral galaxies
Spirals make up about two-thirds of all bright
galaxies. They are subdivided into classes Sa, Sb
and Sc. The two important features of the classi-
fication are (1) the openness or tightness of the
winding of the spiral pattern, and (2) the relative
importance of the central bulge and the disk of
the galaxy. Sa galaxies have the largest bulges
and the most tightly wound arms. Sc galaxies
have the smallest bulges and the most open
arms. We think that the Milky Way is between Sb
and Sc. Different types of spirals are shown in
Fig. 17.3.
Some spirals have a bright bar running
through their center, out to the point where the
arms appear to start. These are called barred spi-
rals. Some examples are shown in Fig. 17.4. The
barred spirals are also subclassified into SBa, SBb
and SBc, according to the same criteria as Sa, Sb
and Sc. In general, the spiral pattern in barred
spirals is quite well defined.
When the spiral pattern is well defined, we
call the galaxy a grand design spiral. They have a
continuous pattern running throughout the
galaxy. Other spirals have a less organized appear- Fig 17.3. Various types of spiral galaxies. (a) NGC 7217, an
ance. These are called flocculent spirals. A compari- Sa galaxy. (b) The Andromeda Galaxy (M31), type Sb, one of
son is shown in Fig. 17.5. our nearest neighbors. It is at a distance of about 700 kpc, and
is more than 20 kpc across. Notice the two companions,
It has been realized that spirals of the same
including M32.We think that it is very similar to our own Milky
type can have different luminosities. This point
is important in trying to determine the distances




Fig 17.3. (Continued) (c) NGC 4622 in Centaurus, at a
distance of 60 Mpc. It is type Sab. (d) M101, in Ursa Major, at a
distance of 5 Mpc, type Sc. (e) A section of M101. Note the
spiral arms of collections of bright patches, probably HII
regions and OB associations, with dust lanes superimposed.
(f) M33, in Triangulum, is type Sc. (g) NGC 1232 (Sc).

Fig 17.3. (Continued) (h) M83
(the Southern Pinwheel) in Hydra,
is 3 Mpc away, and is type Sc.
(i) M104, (Sab) the Sombrero
Galaxy, looks like a Mexican hat
viewed almost edge-on.This is an
edge-on spiral with a prominent
bulge.The dark lane is the collec-
tive extinction of dusty molecular
clouds in the disk. (j) The edge-on
spiral NGC 4565, in Coma
Berenices, is type Sb. [(a) Image
from the OSU Bright Spiral Galaxy
Survey; (b)“(f), (h), (j), (k) NOAO;
(g), (i) ESO]



classification. This is done by adding a I through magnitude of a galaxy is known, and its apparent
V following the Hubble classification, with I magnitude is observed, its distance can be
being the brightest (just as for stars). Efforts are determined.
still underway to find other properties of spirals An important feature of spirals is the obvious
that correlate with luminosity class. In this way, presence of an interstellar medium gas and
the luminosity of a galaxy can be determined dust. Even when a spiral is seen edge-on, we can
without needing to know its distance. (Similarly, tell that it is a spiral by the presence of a lane of
the luminosity class of a star can be determined obscuring dust in the disk of the galaxy. The
from the shapes of certain spectral lines, allow- light from spirals contains an important contri-
ing us to know the absolute magnitude of a star bution from a relatively small number of young
without knowing its distance.) Once the absolute blue stars, suggesting that star formation is still



Fig 17.5. Grand design and ¬‚occulent spirals. (a) A grand
design galaxy, M81, type Sb. (b) A ¬‚occulent spiral, M94, also a
type Sb. [NOAO/AURA/NSF]

taking place in spirals. Where galaxies are found
in a cluster, to be discussed in Chapter 18,
approximately 80% of the galaxies are ellipticals.
Outside of clusters, approximately 80% are spi-
rals. Typical radii for the luminous part of the
Fig 17.4. Various types of barred spirals. (a) NGC 1530, in disk in spirals are about 10 to 30 kpc. Stellar
Cameleopardalis, is type SBb. (b) NGC 1365, in Fornax, type masses of the galaxies we can see range from 107
SBc. [(a) NOAO/AURA/NSF; (b) ESO]
to 1011 M .

Fig 17.6. HST images of six
spiral galaxies, showing regions of
star formation.These are false
color images made through three
separate ¬lters.The red repre-
sents the Paschen line from H at
1.87 m. Blue shows near IR
emission (1.4 to 1.8 m ). Green
is a mixture of the two.

17.1.3 Other types of galaxies
Fig. 17.6. shows the distribution of tracers of
young stars in a selection of spirals. In studying There is an additional type of galaxy that has cer-
general trends, we can get a good idea of the dis- tain features in common with spirals, but does
tribution of young stars as we go farther out in not show spiral arms. This type is called S0 (˜S-
the disks of spiral galaxies. The luminosity of the zero™) (Fig. 17.7). The bulge in an S0 is almost as
disk falls off sharply with r, the distance from the large as the rest of the disk, giving the galaxy an
center. We can approximately fit the observed almost spherical appearance. Some S0 galaxies
falloff with an exponential expression. That is, if also contain gas and dust, suggesting that they
L0 is the luminosity at the center, the L(r), the belong in the spiral classification. However, most
luminosity at radius r, is given by
L0 e(
L(r) (17.2)

In this expression, D is called the luminosity
scale length and gives a measure of the character-
istic radius of the galaxy as seen in visible light.
Typical values of D are about 5 kpc. This means
that the luminosity of the disk of a spiral falls to
1/e of its peak value at r 5 kpc.

Table 17.1. Properties of spirals and
Property Spirals Ellipticals

Gas yes some
Dust yes some
Young stars yes none
Shape ¬‚at round
Stellar motions circular rotation random
Color blue red Fig 17.7. M102, type S0. [NOAO/AURA/NSF]

(a) Fig 17.9. This galaxy, M82, is the scene of very unusual activ-
ity.At ¬rst it was thought that this galaxy was exploding.
However, it just seems to have undergone a rapid wave of star
formation. Galaxies like this are known as starburst galaxies, and
we will discuss them more in Chapter 19. [NOAO/AURA/NSF]

reported their existence in 1943. These are spiral
galaxies with a bright small nucleus. The spectra
show broad emission lines, an indication of a
very hot or energetic gas. Seyferts make up about
2 to 5% of all spiral galaxies. An example of a simi-
lar phenomenon in ellipticals is found in N galaxies
(b) (where the N stands for ˜nucleus™). There is also a
class of galaxies that give off very strong radio
Fig 17.8. The Magellanic Clouds. (a) The Large Magellanic
emission, radio galaxies. All of these active galaxies
Cloud (LMC) is 50 kpc from us. (b) The Small Magellanic
will be discussed in Chapter 19.
Cloud (SMC), which is 65 kpc from us. [NOAO/AURA/NSF]

17.2 Star formation in galaxies
S0 galaxies have no detectable gas. The role of S0
galaxies is still not well understood.
Some galaxies have no regular pattern in their
An important opportunity provided by other
appearance. These are called irregular galaxies. The
galaxies is the opportunity to study star forma-
Magellanic Clouds, shown in Fig. 17.8, are irregular
tion in a variety of environments. We can see how
companions to our own galaxy. Irregulars make up a
various factors, such as the type of galaxy, its
few percent of all galaxies. We distinguish between
metallicity, and the interstellar radiation field
two types of irregulars: Irr I galaxies are resolved
within the galaxy, affect star formation. We dis-
into stars and nebulae; Irr II galaxies just have a
cussed the basic idea of star formation in the
general amorphous appearance. Lenticular galaxies
interstellar medium in Chapters 14“16. We would
have an irregular elongated structure. Ring galaxies
like to apply the ideas we developed when study-
have prominent bright rings around their centers.
ing star formation in our galaxy to help us under-
Peculiar galaxies have a general overall pattern,
stand other galaxies. In turn, our understanding
but also have some irregular structure indicative
of other galaxies will help our analysis of our
of unusual activity in the galaxy. An example is
galaxy. We can ask a number of questions about
shown in Fig. 17.9.
star formation in galaxies.
There are also types of galaxies that are char-
(1) Does star formation take place in molecular
acterized by a very bright nucleus. Seyfert galaxies
clouds, as it does in the Milky Way? If so, are
are named after their discoverer Carl Seyfert, who

the properties of those clouds similar to the
ones we find in our galaxy?
(2) What is the large-scale distribution of star-
forming material within the galaxy?
(3) What is the distribution of newly formed
(4) Does the mix of stellar masses (the initial
mass function) appear to vary from galaxy to
galaxy, or within galaxies?
(5) How does the star formation rate vary from
galaxy to galaxy and within galaxies?
To study molecular material, we still need to
observe trace constituents, such as CO, using mil- (a)
limeter telescopes. For more distant galaxies,
angular resolution with single dishes is a limit-
ing factor, but millimeter interferometers are
becoming more powerful, and give much more
useful resolutions. To study hot cores and proto-
stars, near infrared observations are useful, and
angular resolution is not a serious problem. To
study recent sites of massive star formation, we
look for HII regions, either by their H emission
(through H filters), or by observing their radio
continuum emission, at centimeter wavelengths.
The latter requires interferometers for good
angular resolution.
17.2.1 Star formation in the Large
Magellanic Cloud
The Magellanic Clouds (Fig. 17.8) are the closest 30DOR-CENTER
galaxies in which we can study star formation. At
the 50 kpc distance of the LMC, a 1 parsec extent, (b)
subtends an angle of 4 arc sec. So the 20 arc sec res-
Fig 17.10. HII regions in the LMC. (a) Optical image of the
olution of typical millimeter telescopes corre-
Tarantula Nebula, 30 Dorado. (b) Image of the LMC, taken with
sponds to a size of 5 pc. This is adequate for study-
an H ¬lter, shows the locations of the HII regions.The two
ing giant molecular clouds with extents of tens of rectangular boxes show the regions of detailed study in Fig.
parsecs (though ultimately arrays like ALMA, shown 17.11. [(a) STScI/NASA; (b) Monica Rubio, University of Chile]
in Fig. 4.32, will be needed to study small clouds
and cloud cores). The Swedish-ESO-Submillimetre
times referred to as the 30 Dor complex. An H
Telescope (SEST, Fig. 4.28b), placed in Chile, has been
image of the whole LMC is shown in Fig. 17.10(b),
particularly well suited to study the Magellanic
showing HII regions all over the LMC. We will look
in a little detail at the large dark cloud (visible in
As Fig. 17.10 shows, the LMC is the site of many
Fig. 17.8b) that runs south from 30 Dor, and an iso-
HII regions. As we discussed in Chapter 15, this
lated HII region in the northeast corner of the
implies the existence of recent massive (O and B)
image, known as N11. Both of these regions are
star formation. Fig. 17.10(a) shows an optical image
indicated by the rectangular boxes on Fig. 17.10(b)
of the brightest HII region in the LMC. Because of
Maps of the CO emission from these regions
its appearance, it is called the Tarantula Nebula.
of detailed study are shown in Fig. 17.11, both by
The exciting star is 30 Dorado, and this is some-

Fig 17.11. CO images (contours) of star forming regions in the LMC, with
observations done on the SEST, on La Silla, Chile.These are superimposed on FIR
images from IRAS (the gray scale). (a) The 30 Dor Complex. Notice the complex
-70 10' of giant molecular clouds. More detailed maps show two dozen GMCs in this
complex, with properties very much like Milky Way GMCs. FIR peaks all have
associated strong CO emission, suggesting there are dense cores there.There are
also CO peaks away from FIR emission. Perhaps these are clouds that are not as
far along the star formation process. (b) The N11 region. In this more open ring,
the individual clouds are more easily seen. HIRES refers to a type of image
DEC (J2000)

-70 20' processing, which enhances the angular resolution [author].
100 pc

(determined from the virial theorem) of a few
times 105M . These are very much like giant
molecular clouds, and this whole long dark cloud
-70 30'
is like a GMC complex in the Milky Way. The FIR
image shows a number of embedded regions
where dust is being heated by ongoing or recent
star formation, just as for massive star forming
regions in the Milky Way. It appears that the
Tarantula Nebula is at the northern end of this
05h41m 05h40m complex, and is in a region where there are more
RA (J2000) young stars, but less molecular clouds. It has
(a) been suggested that this is a site of sequential
star formation, also similar to situations found in
the Milky Way.
-66 10'
The N11 region, Fig. 17.11(b), has a more open
appearance, so it is easier to see the structure.
Here we see a ring of clouds (with an extension to
the northwest). These clouds also look like Milky
DEC (J2000)

-66 20'
Way GMCs, and they also have masses of a few
times 105 M . All of these clouds have internal
velocity dispersions comparable to Milky Way
100 pc

GMCs. This grouping has a similar appearance to
-66 30'
the Orion region (partly shown in Fig. 15.4). So, it
appears that star forming regions in the LMC are
very similar to those in the Milky Way, despite the
-66 40' CO many differences between the LMC and the Milky
04h58m 04h56m
17.2.2 Star formation in spiral galaxies
RA (J2000)
When we look at the various images of spirals
(Fig. 17.3), we see that the spiral arms are traced
themselves, and superimposed in the FIR images. by strings of bright HII regions. This suggests that
We first look at the 30 Dor complex (Fig. 17.11a). the spiral arms are sites of enhanced massive star
The CO emission shows a complex of molecular formation. We would like to apply the ideas we
clouds that extend over part of an arc for about developed when studying star formation in our
600 pc. A more detailed picture shows that this is galaxy to help us understand other spiral galax-
composed of some two dozen clouds, each with ies. In turn, our understanding of other galaxies
an extent of tens of parsecs, and each with a mass will help our analysis of our galaxy. We can ask a

Fig 17.12. 21 cm map of the Andromeda Galaxy, M31,
M31 is the nearest spiral, so it provides the
made using the Westerbork interferometer in the
best opportunity for studying the interstellar
Netherlands. [Elias Brinks, Sterre wacht, Leiden University]
medium in detail. At 700 kpc distance, 1 arc sec
number of questions about star formation in spri- corresponds to a linear size of 3 pc, so a 100 pc
ral galaxies. long giant molecular cloud would subtend an
angle of 30 arc sec. This corresponds to the reso-
(1) What is the large-scale distribution of star form-
lution of typical single-dish millimeter tele-
ing material? How does it vary in the disk with
scopes. To study the large-scale distribution of
distance from the center? How does it vary with
molecular material, we could use single-dish
distance from the central plane of the disk?
observations, but to study individual clouds we
(2) Is the interstellar medium concentrated into
have to use millimeter arrays. To study the HI, sin-
the spiral arms?
gle-dish observations at 21 cm do not give suffi-
(3) How do the sizes of molecular clouds compare
cient resolution, so we must use arrays.
with those in our galaxy? Are the physical
Fig. 17.12. shows an interferometer map of the
conditions within the star forming clouds the
large-scale distribution of HI in M31. The large-
scale molecular distribution is shown in a single-
In studying the interstellar medium of our dish CO map in Fig. 17.13. One of the problems in
galaxy, or any other, radio observations play an studying the spiral structure in M31 is that it is
important role. Except for the nearest galaxies, tilted so it is hard to trace accurately the spiral
single-dish radio observations do not provide arms as they would appear if we were looking
much spatial detail. However, with the extensive from overhead. The large-scale distribution of
use of interferometers we have now obtained very star forming regions in M31 is shown by the FIR
detailed maps of many galaxies. Continuum image in Fig. 17.14.
observations can be used to study the positions of The single-dish CO observations of spirals
HII regions and young supernova remnants, both reveal a sharp falloff in brightness with radius,
signs of relatively recent star formation. Studies similar to that of the visible light. The falloff in
of spiral structure have been limited by poor CO emission may indicate the true gas distribu-
angular resolution for single-dish studies. tion. However, it may be due to the fact that the
However, sufficient resolution is available to gas cools, and therefore radiates less strongly
study nearby galaxies. where there are fewer stars to heat it. There may


Y (deg)



-1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
X (deg)

0.2 0.4 0.6 0.8 1.0 1.2 1.4
Wco (K km/s)
Fig 17.13. Large-scale distribution of CO emission in the
galaxy M31.This map was made using a 1 m diameter tele- within a galaxy and from galaxy to galaxy. Within
scope at the Center for Astrophysics (Harvard). [Thomas
a galaxy, the general trend is to have the molecu-
Dame, CFA/Dame,T. et al., Astrophys. J., 418, 730, 1993, Fig. 1(a)]
lar hydrogen abundance fall off faster with radius
than the atomic hydrogen abundance. We find
also be significant amounts of molecular gas in some galaxies in which the molecular hydrogen
the outer parts of spiral galaxies. makes up over half of the interstellar medium, and
In studies of several spirals it is found that the others in which it seems to be less than 10%, as
relative amounts of molecular and atomic hydro- determined from a deficiency of CO emission. In
gen vary significantly. These variations occur both galaxies that seem to be deficient in molecular
hydrogen, we still don™t have observations with
sufficient resolution to tell us whether this is
because they contain fewer molecular clouds than
the other galaxies, or whether the clouds are less
dense. Also, even galaxies that appear to be defi-
cient in molecular clouds have O stars. This tells
us that we still do not fully understand the con-
nection between molecular clouds and massive
star formation.

17.3 Explanations of spiral

It is actually quite surprising that we see any spi-
ral structure in galaxies. The differential rotation
of a galaxy should smear out any pattern on a
time scale comparable to the orbital period. For
the Sun, the orbit period is about 200 million
years. We think that the age of the galaxy is about
Fig 17.14. IRAS map of M31. [NASA]
10 billion years. This means that any initial spiral

A scenario for temporary arm formation is
shown in Fig. 17.15. For one reason or another, star
formation starts in one region. It may even spread
via some of the triggering mechanisms discussed
in Chapter 15. The region with new stars is then
stretched out by the differential rotation into a
piece of a spiral pattern. This may explain why we
see pieces of spiral features but no complete arms
in our galaxy. In this scenario each ˜arm™ lives for
(a) (b)
a short time, and new ones are always forming.
If spiral patterns persist for many revolutions
of a galaxy, then it may be that the pattern and
the matter itself are moving at different speeds.
At first this may seem strange, but we can use an
analogy to see how it might work (Fig. 17.16).
Suppose we have cars moving along a two-lane
highway, and we are looking from above, in a traf-
fic helicopter. A truck breaks down in the right
lane, causing a traffic jam. If we look from above,
(c) (d)
we see cars backed up for some distance behind
Fig 17.15. Scenario for temporary spiral structure. (a) The
the truck. The density of cars is higher for this
dark area is the center of the galaxy, and the arrows indicate
region. Far behind the truck, the cars are still
the orbital speeds of material at different distances from the
moving at their normal speed, and after the cars
center. (b) A large-scale burst of star formation takes place.
squeeze past the truck, they will resume their
(c) The differential rotation stretches the stars out, producing
normal speed. If we come back a few minutes
part of a spiral arm. (d) After a few rotations the arm is
stretched out so much that we can no longer detect its later, we will see the same pattern of cars.
presence. However, the specific cars involved in the buildup
will be different. The cars that we saw originally
pattern would have had ample time to smear out. will be far down the road. In this case, the cars are
Therefore, spiral arms must be temporary, or moving along, but the pattern stays in the same
there must be a way of perpetuating them. place because the truck stays in the same place.

Fig 17.16. Car analogy to density wave.
(a) A truck (shaded) is broken down in the
right lane. Far in front and behind the truck,
14 10 2
6 5
cars have the normal speed and low density.
11 9 1
13 7
Just behind the truck, the density of cars
goes up, and their speed goes down. (b) As
time goes on, cars slowly squeeze by the
truck.The basic pattern is retained. However,
as the numbers on the cars show, different
17 15 14 7 5
12 10 8
cars are stuck behind the truck than in the
16 6
earlier frame.We therefore have a density
concentration along the highway while the
individual cars are not permanently attached
to the concentration. (c) This concentration
can even move. Instead of being stuck,
19 8
15 13 11
suppose the truck is moving slowly.The
12 7
16 10
20 14
pattern moves along with the truck, while
the individual cars move at higher speeds.

Now suppose the truck is moving at a slow
speed. Again, there will be a buildup behind the
truck as cars squeeze into one lane to move past
the truck. As in the case of the stationary truck,
we see the cars moving at their normal speed.
However, now the pattern moves. The speed of
the pattern is not related to the speed of the
cars “ it is determined by the speed of the truck. Stars
HI Cloud
The truck is responsible for the pattern. The cars
with or without
simply respond to the presence of the truck. This HII
Small Molecular
is the type of situation in which the pattern (the Clouds
traffic jam) can move at one speed, and the mat-
Fig 17.17. Scenario for molecular clouds tracing spiral
ter (the cars) at another.
arms. If the gas is circulating faster than the spiral pattern, as
There is a theory that the same type of situa-
with the density wave picture, the gas can overtake the spiral
tion can occur in spiral galaxies. Since the matter arms from behind. Before reaching the arms, the gas is in the
moves at a different speed from a density form of low density HI cloud, possibly containing some small
buildup, the theory is called the density wave molecular clouds.The entry into the arm slows the HI
theory. In a galaxy, the dynamics is controlled by clouds down and compresses them. It may also gather
the halo, which contains most of the mass. The together small molecular clouds. In any event, giant molecular
clouds form.These clouds give birth to O and B stars (as
bright spiral arms contain a small fraction of the
well as to all the other types).The radiation from the O and
mass of the galaxy, and represent material which
B stars disrupts the clouds.
is orbiting at its normal speed, but responding to
the gravitational effects of the asymmetric distri-
bution of the stars in the halo. The mathemati- As the front of the cloud enters the arm, it
cian C. C. Lin has shown that once a spiral pattern slows down. The only way for the back of the
is established in a galaxy, it can sustain itself for cloud to know that this has happened is for a
a long time in this type of wave. Eventually, the pressure wave (sound wave) to travel from the
wave will die out and a new one must be front of the cloud to the back. However, the speed
generated. of the cloud is greatly in excess of the speed of
In the density wave picture, the visible arms sound within the cloud (a few kilometers per sec-
are a result of a gathering of interstellar matter. ond). The back of the cloud doesn™t receive the
When high enough densities are reached, star message to slow down until it has almost over-
formation may take place. One scenario for this is taken the front. The cloud has been compressed.
illustrated in Fig. 17.17. A large HI cloud, or a (This may also gather small molecular clouds into
group of small clouds, approaches an arm at a giant molecular clouds.) At this point, we see the
speed of about 100 km/s relative to the arm. (In cloud as part of a dust lane, explaining why dust
this case, the arm may be moving at 100 km/s, lanes appear at the back of spiral arms. Since the
and the matter overtaking it at 200 km/s.) The cloud has been compressed, star formation is ini-
arm acts like a gravitational potential well, caus- tiated. Massive (as well as low mass) star forma-
ing material to take more time to traverse the tion takes place. The bright stars don™t live very
arm than a similar distance between arms. The long, so we see them only over a small range,
matter entering an arm will leave its circular forming the bright chains that mark the fronts of
path, and have some motion along the arm, the arms.
before finally emerging. It should be noted that, The massive stars also have the effect of driv-
even if the density waves don™t cause strong ing the clouds apart. This can be through the
visible arms, they alter the orbits, resulting in effect of stellar winds, expanding HII regions, and
noncircular motions. Some of the results of criti- supernova explosions. The clouds dissipate. The
cal calculations of density waves are shown in material again resembles that which originally
Fig. 17.18. entered the back of the arm. It remains this way

Fig 17.18. The results of
theoretical calculations
demonstrating the ability of galaxies
to amplify small perturbations into
a well de¬ned spiral pattern. [Alar
Toomre, MIT]




until it overtakes the next arm. According to this
picture, giant molecular clouds should be seen
almost exclusively in spiral arms. It is a good
observational test of the theory, but, as we have
seen, the limited angular resolution of single
dishes makes these observations difficult.
We have already seen that spirals seem to fall
into two categories: grand design and flocculent.
It may be that the underlying cause of the two
types of spiral structure is different. One observa-
tion to support this is that grand design spirals
seem to occur in galaxies with internal bars or
with nearby neighbors. It is suspected that
the tidal interactions between the galaxy and the
neighbors set up the spiral density wave in the
mass distribution of the galaxy. As the gas
streams through the density wave, it is com-
pressed into giant molecular clouds, which give
birth to stars and then dissipate, as shown in Fig.
17.17. In flocculent spirals, the spiral structure
may be a series of temporary patterns, in which
the results of local bursts of star formation are
drawn into a spiral by the differential rotation of
the galaxy.
The density wave theory seems to be best
applied to grand design galaxies. We will briefly
Fig 17.19. Optical image of the Whirlpool Galaxy, M51,
look at one that has been studied in detail, M51
9.6 Mpc distant [NOAO/AURA/NSF].
(the ˜Whirlpool™). This is shown in Fig. 17.19. In

Fig 17.20. Tracers of the spiral
pattern in M51. ˜Visible™ is R band
taken at Mt Palomar. ˜IR™ is K
band, taken at Kitt Peak. ˜Doppler™
is H velocities, taken at Mt
Palomar. ˜Gas Flow™ are the resid-
ual velocities from ˜Doppler™ with
the average rotation taken out.
This gives the streaming motions.
˜Molecular Gas™ is CO emission
from BIMA. ˜Atomic Gas™ is HI
emission with the VLA. ˜Ionized
Gas™ is H from Kitt Peak.
˜Combined Gas™ is HI in blue, H
in green and CO in red, to com-
pare the locations of these com-
ponents. [Stuart Vogel, University
of Maryland; H1 emission, Arnold
Rots, CFA]

that the molecular gas dominates the inner disk
M51 we can see clearly the location of the dust
and the atomic gas dominates the outer part of
lanes with respect to the bright arms. The lanes
the disk.
are on the inside edges of the bright arms.
The face-on appearance of M51 means that we
Remember, in the density wave picture, the inter-
can easily see the relative placement of features.
stellar gas is swept up, forming giant molecular
However, we have very little velocity information,
clouds. These giant molecular clouds are visible
since all of the galactic rotation is perpendicular
as the dust patches. Eventually, these clouds give
to our line of sight. We see that the largest
birth to O stars, which illuminate HII regions.
Doppler shifts are in edge-on spirals, but we can-
The O stars and HII regions trace out bright arms.
not make out any spatial structure from their
The density wave picture therefore makes a spe-
edge-on appearance. Studies of more inclined
cific prediction of the relative positions of the
spirals, such as M81, shown in Fig. 17.5(a), have
dust and bright arms. Radio continuum maps of
been useful in testing velocity shift predictions of
M51 show that the synchrotron emission is
the density wave theory.
strongest in the direction of the dust lanes. At
first we might expect the synchrotron emission
to be strongest where there are the most super-
17.4 Dark matter in galaxies
novae, on the bright side of the arm. However, the
compression of the interstellar medium on the
When we look at a galaxy in visible light we obvi-
dark dust side of the arms produces relatively
ously see the most luminous objects. However,
strong magnetic fields, and the synchrotron
some of the mass may not be luminous. It could
emission becomes stronger when the magnetic
be there but hard to detect. The only sure way to
field strengthens.
trace out the total mass, whether it is bright or
Fig. 17.20. shows the results of a detailed
dark, is to study its gravitational effects. In a
study of the spiral structure made by Stuart
galaxy, the easiest way to study the gravitational
Vogel (University of Maryland). One of the fea-
forces is to measure the rotation curve. We have
tures that we see is that the CO emission comes
already discussed the rotation curve in our galaxy
from the inside of the arms, consistent with the
in Chapter 16.
predictions of density wave theory. Also note

that v(r) stays roughly constant out as far as we
see luminous material. This immediately tells us
that the mass doesn™t fall off as fast as the lumi-
nosity. The masses that are found are as high as
2 1012M . In many galaxies, no edge has yet
been found. The rotation curves are still flat out
to radii where the interstellar medium can no
longer be detected, even using 21 cm observa-
tions, which show material farther out than the
H emission.
Where can this matter reside? One possibility
is that it is part of the disk. However, theoretical
models show that such a large mass would gather
the disk into a bar. The disks that we see would

Fig 17.21. (a) Vera Rubin.


. 16
( 28)