Bohmvitense e Introduction To Stellar Astrophysics Volume 1

Introduction to stellar astrophysics
Volume 1
Basic stellar observations and data
Introduction to Stellar Astrophysics:

Volume 1
ISBN 0 521 34402 6 (hardback)
ISBN 0 521 34869 2 (paperback)
Volume 2
Stellar atmospheres
Volume 3
Stellar structure and evolution
Introduction to
stellar astrophysics
Volume 1
Erika Bohm-Vitense
University of Washington
The right of the

University of Cambridge
to print and sell
all manner of books
was granted by
Henry VIII in 1534.
The University has printed
and published continuously
since 1584.
CAMBRIDGE UNIVERSITY PRESS

Cambridge
New York
Port Chester
Melbourne
Sydney
Published by the Press Syndicate of the University of Cambridge

The Pitt Building, Trumpington Street, Cambridge CB2 1RP
40 West 20th Street, New York, NY 10011, USA
10 Stamford Road, Oakleigh, Melbourne 3166, Australia
Cambridge University Press 1989
First published 1989
British Library cataloguing in publication data
Bohm-Vitense, Erika, 1923Introduction to stellar astrophysics.
Vol. 1: Basic stellar observations and data
1. Stars
I. Title
523.8
Library of Congress cataloguing in publication data
Bohm-Vitense, E.
Introduction to stellar astrophysics/Erika Bohm-Vitense.
p.
cm.
Includes index.
Contents: v. 1. Basic stellar observations and data.
ISBN 0-521-34402-6 (v. 1). ISBN 0-521-34869-2 (pbk.: v. 1)
1. Stars. 2. Astrophysics. I. Title.
QB801.B64 1989
523.8dcl9 88-20310 CIP
ISBN 0 521 34402 6 hardback
ISBN 0 521 34869 2 paperback
Transferred to digital printing 2003
TM
Contents
Preface
ix
1
1.1
1.2
1.3
Positions of stars
The coordinate system
Direction of the Earth's rotation axis
Visibility of the sky
1
1
3
4
Proper motions of stars
3
3.1
3.2
Distances of nearby stars

The distance of the sun
Trigonometric parallaxes of stars
9
9
11
4
4.1
4.2
4.3
13
13
14
4.4
4.5
4.6
The brightnesses of the stars

The apparent magnitudes
The colors of the stars
Correction for the absorption of radiation in the Earth's
atmosphere
The black body
The solar radiation
The absolute magnitudes of stars
16
21
26
28
5
5.1
5.2
5.3
5.4
5.5
Color magnitude diagrams

Photometric parallaxes for
Photometric parallaxes for
31
31
33
36
38
39
6
6.1
6.2
6.3
The luminosities of the stars

Fluxes, luminosities and angular radii
The luminosity of the sun
Luminosities of stars and bolometric magnitudes
of nearby stars
for open clusters
for globular clusters
star clusters
single stars
41
41
43
44
vi
Contents
7
7.1
7.2
7.3
7.4
Angular radii of stars

The problem
The Michelson interferometer
The Hanbury Brown interferometer
Angular diameters from lunar occultations
49
49
50
53
58
8
8.1
8.2
8.3
Effective temperatures of stars

General discussion
The solar surface flux and effective temperature
63
63
64
65
9
9.1
9.2
9.3
9.4
9.5
9.6
Masses and radii of stars

General discussion of binaries
The Doppler effect
Radial velocities and masses of binaries
Determination of orbital parameters for binaries
Data for stellar masses, radii, and effective temperatures
The mass-luminosity relation
67
67
70
71
75
84
85
10
10.1
10.2
10.3
Spectral classification
The spectral sequence
Luminosity classification
White dwarf spectra
89
89
95
97
11
11.1
11.2
11.3
Understanding stellar spectra

The solar spectrum
Line identification
Understanding the spectral sequence
99
99
100
100
12
Population II stars
107
13
Stellar rotation
111
14
14.1
14.2
Stellar magnetic
General discussion
The Zeeman effect
15
15.1
15.2
15.3
15.4
Stars with peculiar spectra

General discussion
Peculiar A stars, or magnetic stars
Diffusion in stars
Metallic line stars, or Am stars
fields
117
117
117
127
127
127
136
140
Contents
vii
15.5
15.6
Barium stars
T Tauri stars
144
147
16
16.1
16.2
16.3
16.4
Pulsating stars
The different types of pulsating stars
The d Cephei stars
The RR Lyrae stars
The period-luminosity relation
153
153
156
159
161
17
17.1
17.2
Explosive stars
Supernovae
Novae
165
165
183
18
18.1
18.2
18.3
18.4
Our sun
Introduction
The surface of the sun
The outer layers of the sun
The active sun
193
193
193
197
204
19
19.1
19.2
19.3
Interstellar absorption
Introduction
The interstellar dust
The interstellar gas
213
213
213
227
Appendixes
A. Problems
B. Some important astronomical quantities
231
231
237
References
239
Index
243
Preface
The topic of this volume is stellar astronomy or more accurately stellar

astrophysics. We call it astrophysics because all our knowledge about stars is
based on the application of the laws of physics to the stars. We want to find
out how big the stars are, how much mass they have, what material they are
made of, how hot they are, how they evolve in time, and how they are
distributed in space. The last question does not strictly belong to thefieldof
stellar astrophysics but knowledge of stellar structure and evolution will
provide a means by which to determine their distances. There are also
important correlations, for instance, between the location and motion of the
stars in our Galaxy and their physical properties.
In Volume 1, we shall be concerned mainly with finding out about the
global properties of stars, such as brightnesses, colors, masses and radii.
Brightnesses and colors can be measured directly for all stars, for masses and
radii we have to study binaries. Parallax measurements can give us distances
to nearby stars. We shall first discuss the majority of stars which we call
normal stars. In the latter parts of this volume we shall also look at stars
which seem to be different, the so-called 'peculiar' stars.
How can we get information, for instance, about the physical properties of
the stars such as their temperatures, pressures, and chemical compositions?
For most of the stars we have to get this information from that tiny little bit
of light which we can receive. Only the sun gives us a lot of light to study, but
even for this nearby star we get little more than just the light. For all the other
stars it is only this tiny little dot of light that has to tell us all about the
properties of the star from which it comes. It will be the topic of Volume 2 to
study how we can do this.
In Volume 3 we will use theoretical considerations and the known laws of
physics to derive the properties of stellar interiors, and discuss stellar
evolution, as well as the origin of the chemical elements. We shall see which
observations can tell us something about the structure of the insides of the
stars.
IX
Preface
The critical reader might wonder whether we are justified in applying the
laws of physics, as we know them from our experiments on Earth, to the
stars. How do we know whether the same laws apply to the stars? We really
do not know. We can, however, try to understand the stellar observations,
assuming that the same laws of physics hold. As long as we can do that
and get sensible results, and if we can predict successfully the results of
further observations we can feel that we are on the right track. Our space
travel within the solar system so far has obeyed the laws of physics as we
know them. Once we come to a point where we can see that the observations
clearly contradict our earthly laws of physics we will have to make
corrections. So far this has not been necessary, except that some refinements
have been made. These refinements are, however, believed or proven to hold
also in experiments on Earth, except that some effects are too small to be
measurable in laboratory experiments. Our laboratories are too small to
measure some effects such as the bending of a light beam in a gravitational
field as predicted by the theory of general relativity. To measure these effects
we need the largest laboratory which we can get, the universe.
The three volumes were written for students in their junior and senior
years. They should be understandable by the educated layman with some
basic knowledge of physics and mathematics.
I apologize for not giving all the references to the authors who have
contributed to our present day knowledge of stars as described here. They
would fill a large volume by themselves. I quote at the end only a few
textbooks which also describe some of the observational results discussed
here, and I list a number of reference books and tables which also give basic
data about stars. I do list all the references from which figures and tables were
used. Frequently these references are also textbooks or review articles which
will lead the reader to a more detailed discussion than is possible in this short
volume.
I am very much indebted to Dr G. Wallerstein for detecting several errors
and weaknesses in the manuscript, and to Dr R. Schommer for a critical
reading of the whole manuscript and for many helpful comments.
I am also very grateful for a JILA fellowship which permitted me to write
most of this book. I acknowledge especially the help of the JILA secretaries
with the typing of the manuscript.
1
Positions of stars
1.1
The coordinate system
If we want to study stars, the first thing to look at might be their

positions in the sky. This in itself does not tell us much about the nature of
the stars, but it is very helpful when we want to find a particular star or a
group of stars in the sky. We have to have a reference point with respect to
which we can describe the position of the star in which we are interested. We
all know or have at least heard about the constellations of stars which in
earlier times were extremely helpful in describing the positions of stars with
respect to a given star in a particular constellation. We still name the
brightest stars according to the constellations in which they are found, but
we like to have a more general description of the positions. When looking at
the sky we can measure the positions only as projected against the sphere of
the sky, i.e., against a two-dimensional surface. We can therefore describe the
positions of the stars by two quantities. Since the surface against which we
measure the positions is a sphere, we use spherical polar coordinates. Since
our telescopes are fixed on the Earth, we use a coordinate system which is
fixed with respect to the Earth. The Earth is rotating, but we do not want to
have a rotating coordinate system, which would cause many problems. We
keep the coordinate system fixed in space. The equatorial plane of our
spherical polar coordinate system is identical with the equatorial plane of the
Earth, which means that the equatorial plane is perpendicular to the
rotation axis of the Earth. Unfortunately, the direction of the rotation axis of
the Earth is notfixedin space, but due to the gravitational forces of the sun
and the moon on the Earth, the Earth's axis of rotation is precessing, i.e.,
describes approximately a cone around an axis fixed in the Earth. This
causes our reference plane also to precess, which means that the coordinates
of the stars are changing in time because the axes of the coordinate system
are changing in time, not because the stars are changing their positions. Of
course, the stars are also moving in space but that leads to much smaller
1 Positions of stars
changes in the coordinates than does the precession of the Earth's rotation
axis.
The coordinates which the astronomers use are the right ascension a and
the declination 8. The right ascension a corresponds to the longitude, which
we use on the Earth's surface to describe the position of a particular place,
and the declination 5 corresponds to the latitude which we use on the surface
of the Earth, see Fig. 1.1. As we know from the Earth, we still have to define
the meridian which we call longitude zero. On Earth this is defined as the
meridian which goes through Greenwich. On the celestial sphere we also
have to define a meridian through a given point as being longitude or right
ascension zero. We could define the position of a given star as right ascension
zero, but then that star might turn out to move in space and then that
coordinate system would move with this arbitrarily chosen star. We could
choose the position of a very distant object, for instance, the position of a
quasar. Even a large space motion of such a distant object would not change
its position measurably. At the time when the coordinate system was defined
the quasars were not known and the distances of other astronomical objects
were not known either. The zero point for the right ascension was therefore
defined by the direction of a line, namely the line given by the intersection of
two planes, the equatorial plane of the Earth and the orbital plane of the
Earth around the sun, the ecliptic, see Fig. 1.2. As the orientation of the
equatorial plane changes with time because of the precession of the Earth's
rotation axis, the direction of the line of intersection of the ecliptic and the
equatorial plane also changes with time which means the zero point for the
right ascension also changes with time. So the coordinates for all the stars
change with time in a way which can be computed from the known motion
of the Earth's rotation axis. The right ascension is measured in hours,
Celestial pole
Equatorial plane
Fig. 1.1. The equatorial plane of the Earth defines the plane for the celestial
polar coordinate system, which describes the positions of the stars by right
ascension a and the declination 3.
1.2 Direction of the Earth's rotation axis
minutes, and seconds. 24 hours correspond to 360 degrees. The right ascension gives the siderial time when the star has its greatest altitude above the
horizon. The declination 3 is measured in degrees, 90 < 3 < + 90.
There are catalogues which give the coordinates of the stars for a given
year, we then have to calculate the corrections to the coordinates for the
time, when we want to observe the object. The equations for computing these
corrections can be found in Smart's textbook on spherical astronomy (1977).
Tables for the corrections are given by Allen (1982).
Catalogues with positions of stars for the year 1855 are, for example, the
'Bonner Durchmusterung' (BD), and for 1900 the 'Henry Draper' Catalogue
(HD). Stellar positions for the year 1950 are given in the catalogue of the
Smithsonian Astrophysical Observatory (S.A.O.).
1.2
Direction of the Earth's rotation axis
From the previous discussion it is apparent that we have to know

how the position, or better, the direction of the Earth's rotation axis changes
in time. How can we determine this direction? The best way is to take a long
exposure photograph of the sky with a fixed orientation of the telescope,
preferably close to the direction of the north polar star. Because of the
Earth's rotation, which will cause the telescope to change its orientation in
space, the stellar positions will apparently move in circles around the
direction of the rotation axis during the course of a day, see Fig. 1.3.
Repeated observations of this kind permit a determination of the changing
direction of the Earth's rotation axis.
Equatorial plane
Fig. 1.2. The direction of the intersection between the equatorial plane and the
plane of the ecliptic defines the meridian for the zero point of the right
ascension. When the orientation of the equatorial plane changes, the position of
the zero point meridian changes.
1 Positions of stars
1.3
Visibility of the sky
From every point on the surface of the Earth we can see only a
fraction of the sky which is determined by our horizon and by the rotation of
the Earth, as illustrated by Fig. 1.4. Suppose the observer is at point P on the
Earth early in the morning. The plane of his horizon is indicated by the solid
line. He can observe everything which is above his horizon. Of course, he will
only see the stars if the sun is not shining on his side of the Earth. Twelve
hours later the observer will be at point P' because of the Earth's rotation
Fig. 1.3. A long-exposure photograph with a fixed position of the telescope

pointing towards the North Pole. The positions of the stars describe circles in
the sky with the centers of the circles showing the direction of the rotation axis
of the Earth. The lengths of the circle segments seen in the photograph are
determined by the duration of the exposure. A 12-hour exposure would
give a half-circle. (From Abell 1982.)
1.3 Visibility of the sky
around the axis a>. The plane of his horizon is now indicated by the dashed
line. He can see only what is above this plane. The whole cone, which is cut
out by the rotating plane of the horizon is excluded from his view. Only
observers at the equator have a chance of seeing the whole celestial sphere
during one day, however, they will only be able to see all of the stars during
the course of one year, because the sun always illuminates about half of the
sky.
from point P
Fig. 1.4. From any point P on the surface of the Earth, a cone is excluded from
observation, except for points on the equator.
Proper motions of stars
In the previous chapter we have seen that the coordinates of the stars
change with time because the coordinate system, which is defined by the
rotation axis of the Earth, changes with time. The coordinates of the stars
may also change because the stars themselves move in space. Only the
motions perpendicular to the direction of the line of sight will actually give a
change in the coordinates, see Fig. 2.1. Motions along the line of sight will
change the distance but not the coordinates. The motions perpendicular to
the line of sight are called proper motions, because they give coordinate
changes which are due to the star's proper and not to the Earth's rotation.
Velocities along the line of sight are called radial velocities because they go in
the direction of the radius of a sphere around the observer. Proper motions
are measured by the changes in their right ascension and declination, which
are angles. The proper motions are therefore measured in sec or arcsec per
year, while the radial velocities are measured by means of the Doppler
shift, see Section 9.2, which gives the velocities in km s " i . It would be difficult
to give proper motions in km/s because the relation between proper motions
in arcsec and in k m s " 1 depends on the distance of the star: A given velocity
vr (Radial velocity)
. v (True velocity)
vp (Proper motion)
Observer
Fig. 2.1. Only motions in the direction perpendicular to the line of sight change
the position of the star in the sky, i.e., the point of the projection of the star
against the background sphere. Radial velocities do not change the coordinates
of the star, only its distance.
2 Proper motions of stars
of a star in a direction perpendicular to the line of sight will lead to a

relatively larger change in position, i.e., to a relatively large proper motion, if
the star is nearby and only to a very small change in position if the star is far
away, see Fig. 2.2. In fact, proper motion studies are often used to find
nearby stars.
>
Fig. 2.2. For a given velocity vp perpendicular to the line of sight the proper
motion in arcsec (//) is larger for a nearby star than for a more distant star (f
Distances of nearby stars
3.1
The distance of the sun
The distances to nearby objects on Earth are determined by

measuring how often a stick of standard length, for instance, a meter stick,
fits in between the two objects whose distance we want to measure. For
larger distances, this very often does not work. For instance, in a mountain
area it would be impossible to measure the distances of two mountain tops in
this way.
Our eyes make rough distance determinations without using a meter stick.
Our eyes actually use the so-called method of triangulation. For triangulation we observe a given object from two different points whose distance we
know, for instance, by measuring with a meter stick.
From two observing points A and B the observed object C will appear
projected against the background at different positions E and Z), see
Figure 3.1. For a nearby object there will be a large angle y between the
projection points, for an object further away the angle will be smaller, see
Fig. 3.2. The relation between the angle y measured from the two points and
the distance to the object is given by
a/(2dl
(3.1)
where a is the distance between the two observing points, A and B, and d is
the distance to the object, C or C, from the center of the two observing
points,
For large distances we can set sin y = y if y is measured in radians. (3.1) can
then be replaced by
y = a/d
(3.2)
from which d can be determined if a and y have been measured. For us our
two eyes serve as the two observing points, a is then the distance between the
eyes.
10
3 Distances of nearby stars
From (3.2) it is obvious that we can measure larger distances if the baseline
a is large, because there is a limit to the size of the angle y which we can still
measure. On Earth there is a limit to the length of the baseline a, this limit is
determined by the diameter of the Earth. It turns out that this baseline is not
large enough to measure even the distance to the sun accurately. We can,
however, measure distances to nearby asteroids this way and then use
Kepler's third law to determine the distance to the sun. We can now also
determine the distance to Venus by radar measurements and then again use
Kepler's third law to derive the distance to the sun. The method works as
fellows: Kepler's third law states that the squares of the orbital periods of
the planets are proportional to the third powers of the semi-major axis b of
their orbits around the sun, or
p2/b3 = const. = A.
(3.3)
For two planets, for instance, the Earth and Venus (or an asteroid) this tells
us that
P(Venus)2/P(Earth)2 =fc(Venus)3/fc(Earth)
(3.3a)
wherefc(Venus)and b(Earth) are the semi-major orbital axis of Venus and

Earth. P(Earth) is one year and P(Venus) is the orbital period of Venus,
which is 224.7 days. Equation (3.3a) is one equation for the two semi-major
axes of Venus and Earth. If we had one more equation we could determine
Fig. 3.1. From two observing points A and B, distance a apart, the object C is
projected against the background at different points D and E. The angle y at
which the object appears from the observing points A and B is given by (3.1);
see text.
Fig. 3.2. For a distant object C the angle y' at which the object appears
from the observing points A and B is smaller than for a nearby object C.
3.2 Trigonometric parallaxes of stars
11
both semi-major axes. The second equation is provided by a measurement of

the distance between Venus and Earth at their nearest approach when it can
best be measured, see Fig. 3.3. In order to demonstrate the principle, we
approximate both orbits by circles (the ellipticities are actually quite small).
From Fig. 3.3 we see that then the distance Venus-Earth d is
d = b(Venus) - b(Earth).
(3.4)
If we measure d, then (3.4) provides the second equation needed to determine

both b (Venus) and b (Earth). Of course,ft(Earth) is the distance Earth-sun,
which we want to determine, and which is generally called one astronomical
unit or abbreviated 1 au. The distance Venus-sun is only 0.72 au. At closest
approach the distance Venus-Earth is therefore only 0.28 au, see Fig. 3.3.
This distance is small enough to be measured by radar. Since the radar signal
travels with the speed of light c, it takes the radar signal a time t to travel to
Venus and back which is given by
t = 2d/c.
(3.5)
The time t can be measured and d can be determined from (3.5). Using (3.3a)
and (3.4) we then derive the distance Earth-sun to be
l a u = 1 . 4 9 x 10 13 cm.
Taking into account the ellipticities of the orbits will complicate the
mathematics but does not change the principle.
3.2
Trigonometric parallaxes of stars
Once we know the orbital diameter of the Earth we can use this
length as the baseline, a, for further triangulation. If we make one
observation on day 1 and the second observation half a year later, we have
changed our position in space by the orbital diameter of the Earth. It is not
necessary to make the observations from the two observing points at the
same time.
From Fig. 3.4 we infer that for a star at the ecliptic pole we will see a change
Earth
Venus
Sun
0.72 au
1 au
Fig. 3.3. The smallest distance Earth-Venus is only 0.28 times the distance
Earth-sun.
12
3 Distances of nearby stars
of position during half a year which is given by the parallax angle 2n = 2 au/d
or
n = 1 au/d,
where d is the distance sun-star. Knowing the size of one au and measuring n
will give us the distance to the star, provided the angle n is large enough to be
measured. Since the distances of the stars are so large it is convenient to
introduce a new unit of length to measure stellar distances. This unit is called
the parsec (pc) and is the distance at which a star would be if it had a parallax
angle n equal to one arcsec. From the geometry in Fig. 3.3 we derive that
such a star would have to be at a distance d equal to 3.08 x 1018 cm. The light
would need about three years to travel this distance. Actually, there are no
stars which have a parallax angle as large as 1 arcsec. The closest star,
Proxima Centauri, has a parallax angle of 0.76 arcsec. We can only measure
parallax angles larger than about 0.05 arcsec with reasonable accuracy,
which means we can determine distances out to about 20 pc with fairly good
accuracy; beyond that trigonometric parallaxes cannot be of much help. As
our measuring instruments improve we may be able to measure parallaxes
down to 0.02 arcsec. We are lucky that we find about 200 stars within 20 pc
distance, bright enough to be the basis for the distance determinations. There
are several thousand very faint stars within 20 pc distance.
In Fig. 3.4 we have demonstrated only the case for a star at the ecliptic pole
which during the course of one year describes a circle projected against the
background sky due to the orbit of the Earth around the sun. If the stars are
not near the ecliptic pole they will, of course, also show parallax motions
against the background sky but their 'parallax orbits' will not be circles but
ellipses. For the major axes of these ellipses, the same equation holds as for
the 'orbital' radii of the stars at the ecliptic poles.
Fig. 3.4. The parallax angle n at which a star is seen when measured at sixmonth intervals is a measure of its distance. The further away the star, the
smaller is n.
The brightnesses of the stars
4.1
The apparent magnitudes
The brightness of a star is a quantity which can be measured easily, at

least in a qualitative way, by comparing the brightnesses of different stars.
Even with the naked eye we can see whether star A is brighter than star B or
vice versa. Ancient astronomers called the brightest stars first magnitude
stars, the next fainter ones were second magnitude stars and so on. These
magnitudes were determined by the sensitivity of the eyes because all
observations were made by the naked eye. The sensitivity of the eye is
logarithmic to enable us to see a large range of light intensity. The magnitude
scale is therefore a logarithmic one. A given difference in brightness
corresponds to a given factor in the amount of energy received.
When astronomers learned to make quantitative intensity measurements,
they did not want to part with the well-established magnitude scale. They
therefore put the magnitude scale on a quantitative basis. Since it is a
logarithmic scale, as we saw, they found the old scale could best be
represented by defining
Amv=-2.5-Alog/v,
(4.1)
where Amv is the difference is magnitude as seen by the eye, i.e., by visual
observation, and Jv is the intensity received with an instrument which has a
color sensitivity similar to our eye. For two stars A and B, we then derive
my(A) ~ mw(B) = - 2.5 [log Iy(A) - log JV(B)].
(4.2a)
The minus sign on the right-hand side takes care of the fact that the
magnitudes become larger when the stars become fainter. Remember the
stars of first magnitude are brighter than those of second magnitude!
If we actually want to determine the magnitude of star 5, we have to know
the magnitude of star A:
mv(B) = my(A) + 2.5 [log Iy(A) - log JV(B)].
(4.2b)
13
14
4 The brightnesses of the stars
The magnitude scale was originally defined by the north polar sequence of
stars. It is now defined by a number of stars measured accurately by Johnson
and Morgan in 1953. For all practical purposes, we can say that the
magnitude scale is defined by assigning the magnitude 0 to the star a Lyrae,
also called Vega. So if star A is Vega, then my(A) = 0 and
mv(B) = 2.5 [log /v(Vega) - log Iy(B)l
(4.2c)
(Actually mv(Vega) = 0.02 0.01, but we do not here worry about this small
difference.) In practice, we compare the brightness of all stars with that of
Vega. If a star is fainter than Vega, then mv > 0, if a star is brighter than Vega,
then it has mv < 0. There are some stars brighter than Vega, for instance,
Sirius, these stars then have negative magnitudes. Sirius has mv = 1.6.
We have always indicated these magnitudes with a lower case m because
they refer to the brightnesses as we see them. They are called apparent
brightnesses or apparent magnitudes. These apparent brightnesses have to be
distinguished from the intrinsic brightnesses, called absolute brightnesses or
absolute magnitudes which are designated by a capital M (see Section 4.6).
4.2
The colors of the stars
We have emphasized that the apparent brightnesses or magnitudes

discussed above refer to visual observations. They compare the brightnesses
as seen with the eye. They are called the visual magnitudes.
Quantitative measurements are now made either with photographic plates
or more frequently with photoelectric instruments. These measuring devices
have a different sensitivity from our eyes. Originally photographic plates
were mainly sensitive to blue light. With photographic plates we compared
the brightness of the stars in the blue wavelength band. A star, which in the
visual has the same brightness as Vega and therefore has mv = 0, may
actually be brighter than Vega in the blue, if it is a blue star, so its blue
magnitude will be mB < 0. A given star will generally have different
magnitudes for different wavelength bands. If it is more blue than Vega, its
blue magnitude will be smaller than its visual magnitude, a star that has
relatively more energy in the red as compared to Vega will have a larger blue
magnitude than visual magnitude. The difference in magnitude of a given
star for different wavelength bands tells us something about the color of the
star. For mB > mv the star is fainter in the blue, i.e., the star looks more red
than Vega. For Vega all apparent magnitudes are zero by definition, see (4.2c)
which holds actually for all wavelength bands. (This does not mean that Vega
has the same brightness at all wavelength bands.)
4.2 The colors of the stars
15
The most widely used system of apparent magnitudes is the so-called UBV
system which measures apparent magnitudes in the ultraviolet, U, referring
to ~3600 A, in the blue, B, referring to ~4300 A, and in the visual, V,
referring to ~ 5500 A. The sensitivity functions of the measuring instruments are shown in Fig. 4.1. There are many different magnitude systems in
use, almost as many as there are astronomers who measure magnitudes.
They all serve their special purposes. The UBV system has the largest
number of measurements. Each band uses a large fraction of the spectrum
(1000 A) and can therefore be used to measure magnitudes of rather faint
stars. In the following discussion we will therefore use only this system.
The difference in the ultraviolet magnitudes and blue magnitudes for a
given star is abbreviated by
U - B = mu-mB,
(4.3)
B V = mB mv.
(4.4)
and
According to our discussion above, B V > 0 means the star is more red
than Vega. B V < 0 means the star is more blue than Vega; it has relatively
more energy in the blue. U B > 0 means also more energy at the longer
wavelenghts, where longer' now means the blue wavelength band as
compared to the ultraviolet, and U B < 0 means the star has relatively
more energy in the ultraviolet than Vega.
Since in the colors we use V instead of mv many astronomers now
7
7000 A
Fig. 4.1. The sensitivity function Sx for the U, B, V magnitude scales, i.e., the
relative intensities that would be measured through the U, B, V filters, for
constant (wavelength independent) intensities Ik, are shown. (From Unsold,
1982.)
16

Table 4.1. The relation between
B V, absolute magnitude M v and
U B colors for main sequence
stars
Main
sequence
U-B
Main
sequence
-4.0
-1.6
-0.4
+ 0.6
+ 1.4
+ 2.1
+ 2.7
+ 3.3
+ 4.0
+ 4.6
+ 5.2
+ 5.8
+ 6.3
+ 6.7
+ 7.1
+ 7.5
+ 8.0
-1.08
-0.71
-0.32
0.00
+ 0.10
+ 0.11
+ 0.07
+ 0.01
+ 0.03
+ 0.13
+ 0.26
+ 0.43
+ 0.63
+ 0.81
+ 0.96
+ 1.10
+ 1.22
Mv
B-V
-0.30
-0.20
-0.10
0.0
+ 0.10
+ 0.20
+ 0.30
+ 0.40
+ 0.50
+ 0.60
+ 0.70
+ 0.80
+ 0.90
+ 1.00
+ 1.10
+ 1.20
+ 1.30
abbreviate mv by V. So V means the same as m v .

In Table 4.1 we list the relation between the U B and B V colors for
'normal' (so-called main sequence) stars. These are, however not the colors as
we measure them directly. Before we can say anything about the true colors
of the stars we must correct the measured intensities for the absorption in the
Earth's atmosphere.
4.3
Correction for the absorption of radiation in the Earth's

atmosphere
When we measure the radiative energy arriving at the telescope we

measure the radiation after it has passed through the Earth's atmosphere,
which means after the Earth's atmosphere has absorbed a certain fraction of
the light. This fraction is not always the same becasue we observe different
stars at different positions in the sky, and even the same star is observed at
different positions in the sky depending on the time of the year and the time
of the night. Depending on the zenith distance z of the star the path length
of the light beam of the star through the atmosphere is different, see Fig. 4.2,
17
4.3 Radiation absorption in the Earth's atmosphere
and that means that different amounts of light are absorbed during the
passage through the atmosphere. If we want to know the amount of
radiation arriving above the Earth's atmosphere - this is the only quantity
which truly tells something about the stars - then we have to correct for the
absorption in our atmosphere, also called the atmospheric extinction.
Due to the extinction in the Earth's atmosphere, the light is reduced by a
certain amount which is proportional to the intensity /A of the beam at
wavelength X. (The more photons there are passing through the atmosphere
the larger is the chance that one of them will hit an atom and will be absorbed
by the atom.) The chances of absorption are also larger if the path lengths
through the atmosphere are larger. It is also large if the atoms are of a kind
which want to absorb especially light of the wavelength X considered. The
properties of atoms concerning the absorption of light of a given wavelength
X are described by the absorption coefficient KX per cm, which may be very
strongly dependent on the wavelength X. We then find that the intensity
change d/A along the path element ds is
(4.5)
Dividing by JA and remembering that d/A//A = d(ln JA) we find

d(ln/A) = -Kkds=
-dr A .
(4.6)
Here we have defined the so-called optical depth TA by

rs0
drA = KX ds
and
TA(S0)
(4.7)
Kxds.
J o
Equation (4.6) can be integrated on both sides between 0 and 5 and yields
= ln/ A(s)-ln/ A (O) = -
d r A = - T A ( S ) , (4.8)
Star
Zenith
Top of the atmosphere
Surface of the Earth
Fig. 4.2. Stellar light is entering the atmosphere at an angle z with respect to
the zenith direction (exactly overhead). Light is absorbed along the path s. The
longer 5 the more light is absorbed.
18
where xx is the variable optical depth along the path s.

Taking the exponential on both sides gives
/A(s) = JA(0)e T^(s).
(4.9)
The optical depth along the path of the light xXs depends on the zenith
distance z, as can be seen in Fig. 4.2. We see that cos z = t/s = dt/ds, or
ds =
= sec z dt.
(4.10)
Pft
(4.11)
cos z
It then follows that
- fsP
Jo
Jo
where xXt is the optical depth measured perpendicularly through the

atmosphere. We can then write equation (4.9) in the form
JA(s,z) = J A ( 0 ) e - ,
(4.9a)
where xXt is now independent of z.

xXt is called the optical depth of the atmosphere at wavelength X and is
usually written at xx.
In order derive from /A(s, z) the intensity above the Earth's atmosphere we
have to know TA, which may depend strongly on the wavelength X.
We can in principle determine xx from two measurements of /A(s, z) to give
us two equations for the two unknowns /A(0) and xx. Let / A1 = / A (z x , s) be the
intensity measured at zenith distance zx and IX2 be the intensity measured at
zenith distance z 2 . Then, according to equation (4.9a)
ln/ A 1 - l n / A 2 = - i ^ s e c z ! - s e c z 2 )
and
ln/A1-ln/A2
(secz2 secz x )
As there are always measuring errors, it is safer to make many measurements and plot the result in a diagram, as shown in Fig. 4.3. The best fitting
straight line gives the relation In Ix = In Ixo sec Z-TA. (Here we have written
Ixo for /A(0)). The gradient of the line is determined by TA, and In Ixo can be
read off on the In Ix axis from the extrapolation of the best fitting line to
sec z = 0. (Never mind that sec z = 0 does not actually exist; this is just
a convenient way to read off ln/ A0 ).
Actually, Fig. 4.2 is a simplification of the situation, since the Earth's
surface is not plane parallel and, in addition, the beam of light is bent due to
4.3 Radiation absorption in the Earth's atmosphere
19
refraction in the atmosphere. These effects give a relation ixjxkt # sec z. The
actual ratio is called the air mass. For sec z < 2 the difference is in the third
decimal place and is in most cases negligible.
It is important to note that the above derivation applies only to each given
wavelength X with a given xx. It cannot be applied to a broad wavelength
band with varying KX. Fig. 4.4. and 4.5 show the overall gross variations of
"n h
In 7xo-
2
sec z
Fig. 4.3. The logarithms of the measured intensities JA are plotted as a function
of sec z. The best fitting straight line through these points can be determined.
The intersection of this line with the In lx axis determines the logarithm of the
intensity above the Earth's atmosphere, In / A0 .
1.0-
7
8
MA) 10"3
Fig. 4.4. The optical depth TA of the Earth's atmosphere for the continuous
absorption is shown as a function of wavelength X, according to Allen (1968).
20
the absorption in the Earth's atmosphere with wavelength. Within the broad
molecular bands seen in Fig. 4.5, which are due mainly to the absorption by
the water molecules in the atmosphere, we actually have very many narrow
lines. The actual variation of KX with X is much stronger than shown. If we use
the method described above to determine the average TA for such wavelength
regions we may get very wrong values for JA0.
According to (4.9a), the correction factor for the measured intensity is
esecz'T\ Suppose we have measured star 1 at zenith distance zx and measured
/ A0 (l)-e- secziT \ For Star 2 we measured /A(2) = / A0 (2)-e secz2T \ We
now determine
l (secz 1 -secz 2 )-T A
Ix(2) e'
(4.13)
if f A1 = fA2 = TA. If we have determined a wrong fA the error is minimized if z1

is chosen as close to z 2 as possible. It is zero if zx= z 2 . The error is also
smaller if we choose small zl and z 2 because then also sec zx and sec z 2 will be
small and the difference remains small, even if
zi^z2.
If we measure stars with very different energy distributions then fA1 7^ T A 2 .
In this case the error does not cancel out, even if zt = z 2 , but the error is still
minimized for z x = z 2 and small secz.
ENERGY CURVE FOR BLACK BO0Y AT 6 0 0 0 K

SOLAR ENERGY CURVE OUTSIDE ATMOSPHERE
AT SEA LEVEL
.1
.2 ,Z\A
.5 .6 .7
BIOLOGICAL EFFECT
1.2
1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.6 2.9 3JO 3.1 3.2
AM
Fig. 4.5. The measured energy distribution of the sun, which means the amount
of radiation which we receive from the sun here on the surface of the Earth, is
plotted as a function of wavelength of the radiation. The black areas indicate
the amount of radiation taken out by the molecular bands of water vapor and
of oxygen. Also shown is the energy distribution as measured above the Earth's
atmosphere. For comparison we also show the energy distribution of a black
body with a temperature of 6000 K, which matches the solar energy distribution
rather well. (From Pettit 1951.)
4.4 The black body

4.4
The black body
4.4.1
Definition of a black body
21
If different stars have different colors, the colors must tell us

something about the nature of the stars and their differences. In order to find
out about the possible meaning, we have to compare the colors of the stars
with colors of known objects on Earth. We do not learn anything if we
compare the stars with just any colored object on Earth, we need some sort of
ideal source of light whose properties depend only on one parameter. Such a
source is the so-called 'black body', whose radiation depends only on the
temperature.
What is a black body? We would call an object black if it looks black in
bright daylight, this means that the object does not send out any light rays
that enter our eyes. The light that our eyes receive from objects which do not
look black in bright daylight is usually not light which these objects emit of
their own but is solar light that is reflected by these objects. Solar light
appears yellow to us. Most objects on Earth do not look yellow even though
we see only solar light reflected by them. The reason for the difference in
color is the fact that any object absorbs part of the solar light, and only the
part which is not absorbed can be reflected. If mainly blue wavelengths were
absorbed, then the object will appear more red than the sun, if mainly red
wavelengths were absorbed then the object will appear more blue. If all
wavelengths are strongly absorbed, then no light is left to be reflected into
our eyes and the object will appear black. We therefore define a black body as
an object that absorbs all light which falls on it. This does not mean that a
black body always has to look black. A black body may generate radiation
by itself, then it may actually be quite bright even though it absorbs all the
light reaching it from other light sources. Think about a hot plate on an
electric stove. If it is not turned on, it looks almost black because it absorbs
nearly all the light which falls on it, but when you turn it on, it becomes hot
and now produces its own light. It then shines brightly even if you turn off all
the outside lights. The hot plate is still nearly a black body because it still
absorbs nearly all the light falling on it but it certainly does not look black
any more. When you turn the hot plate on 'LOW' you will see it glow in a
dark red color, when you turn it on higher you will find that it becomes
brighter and more yellowish; if it becomes still hotter, it will shine still
brighter and more bluish. In this example, we see already that the color
of such a near black body tells us something about the temperature of the
object.
How can we realize an ideal black body, which means an object that
22
absorbs all the light falling on it? The best way to do this is to make a small
hole in a box, see Fig. 4.6. A light beam falling into the hole will hit a wall in
the box and will be reflected, part of it may be absorbed by the wall. The
reflected part will hit another wall and will be partly absorbed and partly
reflected again and so on. Chances that even a small fraction of the light
might finally find its way back out through the hole are extremely small and
will be smaller the smaller the hole is. The hole is indeed a black body.
Actually the windows in a building are approximations to this kind of a
black body; when you look at the windows from the outside, they appear
black, except if you put a light source inside this black body, such as a
burning electric bulb, then the black body radiates on its own merits.
If we insulate such a black body very well from its surroundings and leave
it to itself for a long time, then the inside of the box, shown in Fig. 4.6,
achieves an equilibrium, which means that nothing in the box will change
anymore in time or in space. We say the box has reached thermodynamic
equilibrium. In physics we only call a box with a hole a black body if it has
achieved thermodynamic equilibrium. This is included in the definition of a
'black body.'
4.4.2
Radiation of a black body
For a black body we can measure the radiation that comes out of the
hole, it is radiation emitted by the black body itself, either by gas in it or by
the inside walls of the box. For the radiation of such a black body in
thermodynamic equilibrium we can measure that the radiation coming out
of the hole is independent of the material of the box and independent of the
gas in the box. No matter which kind of box we take, after thermodynamic
equilibrium is achieved the radiation is always the same if the temperature of
the box is the same, but the radiation changes if the temperature of the black
Fig. 4.6. A black body can best be realized by a small hole in a box. Light
falling into the hole in the box will be reflected back and forth and will not find
its way out of the hole before being absorbed at the walls. For a black body, as
defined in physics, we have to insulate the box very well from its surroundings
and leave it to itself for a long time so that thermodynamic equilibrium can be
achieved in the box.
4.4 The black body
23
body changes. The change is similar to the changes seen for the hot plate with
increasing temperature. For higher temperatures, more radiation is emitted
and the radiation looks more blue. From such measurements of the
radiation of a black body we find that the total radiative energy E emitted
per cm 2 and per s in all directions by a black body of temperature T increases
with the fourth power of the temperature, i.e.
E = cjT* = nF.
(4.14)
where a is the so-called Stefan-Boltzmann constant and is a = 5.67 x

1 0 " 5 e r g c m " 2 s " 1 K" 4 , F is called the flux. Equation (4.14) is the StefanBoltzmann law.
4.4.3
Spectral energy distribution
If we measure the amount of radiation per wavelength interval of

1 cm coming out of the hole per s perpendicular to the area of the hole and
into a cone with a solid angle Aco = 1, and if we divide this amount by the
area of the hole, then this energy is called the intensity Ix (see Fig. 4.7)
In Fig. 4.8 we show the wavelength dependence of the intensities emitted
by black bodies of different temperatures.
Of course, we cannot make black bodies with temperatures of 6000 K in
the laboratory because any box of any kind of material would melt at such
temperatures, but we can derive the law for the intensity distribution from
measurements for lower temperatures and then use this law to calculate what
the energy distribution for higher temperatures will look like. You will recall
that it was the observed law for the radiation of a black body which led Max
Planck to the discovery of the light quanta.
For a black body the intensity lx = Bx is given by
O7
h = ks ehcixkT_l
=BX = Planck function.
(4.15)
Here, c = velocity of light, k = Boltzmann constant, and h = Planck constant.
Aco
Fig. 4.7. The energy emitted from a surface area A of 1 cm 2 per s per
wavelength band AA = 1, perpendicular to A and into a cone with opening
Aco = 1, is called the intensity Ik. For a black body the intensity Ik = Bx =
Planck function, given by (4.15).
24
We certainly do not measure the intensity with a wavelength band of 1 cm

because the intensity would vary by very large factors over such a wide band.
What we mean is that we measure the intensity in a small wavelength band
Al and then divide the measured energy by the wavelength interval AX. The
amount of energy received will of course increase if the bandwidth is
increased.
Keeping in mind the energy distributions of black bodies and the
increasingly more blue color for higher temperatures we may suspect that
the blue stars have higher temperatures than the red stars.
4.4.4
Wieris displacement law
The intensity distribution of a black body as given by (4.15) has a

very important property. As we see in Fig. 4.8, the maximum shifts to shorter
wavelengths for higher temperatures. From (4.15) it can be easily derived
that the wavelength /lmax, for which the maximum intensity is measured, (i.e.
the wavelength >lmax for which dIJdX = 0) is inversely proportional to the
temperature T, or in other words,
Amax* T= const. = 0.2897 cm-degree
(4.16)
if the wavelength Amax is measured in cm.

Equation (4.16) offers an easy way to determine the temperature of a black
body without having to measure the whole energy distribution, we only have
1000
5000
10 000
15 000
MA)
Fig. 4.8. The intensity Ik = Bk is shown as a function of the wavelength X for
black bodies with different temperatures T, as indicated on the curves.
4 A The black body
25
to determine the wavelength at which the energy distribution has its

maximum. If the stars were to radiate like black bodies we could easily
determine their temperatures in this way.
4.4.5
The colors of black bodies
Because black bodies become more blue with increasing temperature we suspect that the colors of the stars tell us something about the
temperatures of the stars. We expect the blue stars to have higher
temperatures than the red stars because for black bodies the radiative energy
shifts to shorter wavelengths for higher temperatures. In order to check how
well the colors of the stars agree with those of black bodies we have to
determine the B V and the U B colors for black bodies with different
temperatures. This means we have to determine the U, B, V magnitudes of
the black bodies which in turn requires us to compare the brightness of the
black bodies in the U,B and V bands with the brightness of Vega.
Unfortunately, this is not very easy. In order to have the same transmission
functions for the star and the black body we have to observe the star and the
black body with the same instrument, namely with the telescope. The
telescope gives only sharp images of light sources which are very far away.
We therefore have to put the black body at a large distance which makes it
rather faint.
Even more of a problem is the correction for extinction in the Earth's
atmosphere. The beam of light from Vega passes through the whole
atmosphere, all layers of the atmosphere with their different absorption
coefficients contribute to the extinction. For the black body, which we may
put at the next mountain top, only the lower layers of the atmosphere
contribute to the extinction. We would have to know the height dependence
of the atmospheric absorption coefficient in order to make the correct
extinction correction for the black body. While from the physical point of
view it would be more logical to compare the stellar brightnesses directly
with those of black bodies to begin with, this causes so many difficulties that
it is intrinsically much more accurate to compare stars with stars. The final
comparison with black bodies does not then influence the intrinsic
comparison of stars among themselves.
When we do the best we can for the determination of the colors of black
bodies, we find the B V and U B colors for black bodies as shown in the
two color diagram, Fig. 4.9. Table 4.2 shows the actual numbers. Every point
on the black body curve in Fig. 4.9 corresponds to a black body with a given
temperature, as indicated for a few values of the temperature. Remember
26
stars with positive values of B V are more red than Vega. B V therefore
increases for lower temperature. The same holds for U B.
In the same diagram we have also entered values measured for nearby
stars. They also seem to follow one curve, which is, however, different from
the one for black bodies, but still all stars with a given color B V have
nearly the same U B color. Both colors are determined by one parameter,
which we still suspect to be the temperature. Before we can, however,
establish a relation between the colors of a star and its temperature we have
to understand the difference between the stellar energy distributions and
those of black bodies, and especially why the stars have so much less energy
in the U band than the black bodies. This we will discuss in volume 2, when
we will discuss the radiation and the structure of stellar atmospheres.
4.5
The solar radiation
The brightest star is of course our sun. We can measure the amount
of radiative energy received from the sun and correct it for the absorption in
16 000 K
-1.0
U-B
8000 K
6000 K
4000 K
Superglants
1.0
-0.3
0.5
B-V
1.0
GO
Fig. 4.9. The U, B, V colors for 'normal', main sequence stars are shown in a
two-color diagram. For increasing B V colors the U B colors generally also
increase, except in the range 0.1 < B - V < 0.5. Also shown are the colors for
supergaints, see chapter 10.2, for which the U B colors always increase with
increasing B V. We have also plotted the U B versus B V colors for
black bodies. For a given B - V the stars have less radiation in the ultraviolet
than black bodies.
4.5 The solar radiation
27
Table 4.2 The B-V

and U B colors of
black bodies with
different temperatures
according to Lamia, 1982
B-V
U - B ( + 0.03)
Black body
-0.30
-0.20
-0.10
0.00
+ 0.10
+ 0.20
+ 0.30
+ 0.40
+ 0.50
+ 0.60
+ 0.70
+ 0.80
+ 0.90
+ 1.00
+ 1.10
+ 1.20
+ 1.30
-1.22
-1.11
-LOO
-0.88
-0.76
-0.66
-0.54
-0.44
-0.33
-0.22
-0.10
+ 0.02
+ 0.14
+ 0.25
+ 0.37
+ 0.49
+ 0.61
the Earth's atmosphere in exactly the way described in the previous section.
We can now also measure the total solar radiation directly, outside the
Earth's atmosphere from satellites. The satellites called 'Space Lab' and the
'Solar Maximum Mission' (the first satellite to be repaired while in orbit)
have made very careful measurements of the solar radiation. What is found is
an energy distribution, also shown in Fig. 4.5. The continuum energy
distribution, which means the energy distribution between the spectral lines,
looks very similar to that of a black body at 6000 K. The energy has a
maximum around 5000 A, the wavelengths at which our eyes are most
sensitive, luckily enough also in the wavelength region in which the Earth's
atmosphere is rather transparent.
How much energy do we actually get from the sun altogether? The total
radiation received from the sun per s per cm2, perpendicular to the beam of
light above the Earth's atmosphere, is called the solar constant S. It is
measured to be S = 1.368 x 10 6 ergcm" 2s~ 1 .
Since there is so much talk about solar energy as the energy source for our
28
industrialized life, let us stop for a moment and realize how much that
actually is.
Since 1 square foot has about 103 cm 2 and since 107 erg sec" * is 1 watt we
can also write S ^ 0.14 x 103 watt per ft2 or S 0.14 kW per ft2.
If your room has a roof area of about 200 ft2 and the sun shines on average
5 hours per day, you would collect on your roof energy of the order of
140 kW hours each day. This is not really correct since the sun is not shining
perpendicularly on your roof and there is some absorption in the atmosphere, but it gives the correct order of magnitude, it may be a factor of 2 or 3
less. This would still be plenty to cover your personal needs. Certainly for a
large industrial plant the sun shining on their roof would not be sufficient,
but our main problem still is that we do not yet know how to convert the
solar ergs per s into kW of electrical energy efficiently. Currently we lose a
factor of between 10 and 100 in the conversion.
4.6
The absolute magnitudes of stars
The apparent magnitudes do not really tell us anything about the

intrinsic brightness of the stars which to the astrophysicist is much more
important than the apparent brightness. A star may be intrinsically very
bright and just appears very faint because it is so distant. If we can determine
the distance we can correct for this effect. We can then in our minds put all
the stars at the same distance and compare their intrinsic brightnesses. The
magnitudes which the stars would have if they were at a distance oflOpc are
called the absolute magnitudes. They are designated by a capital M. Again we
have absolute magnitudes M v , M B , Mv for the different wavelength bands.
Vega is at a distance of 8.4 pc, i.e., is closer than 10 pc. If Vega was at a
distance of 10 pc it would be fainter. Its absolute magnitude is therefore
larger than its apparent magnitude: M v (Vega)>0. It comes out to be
My(Vega) = 0.5. We have to calculate the magnitude which Vega would
have if it were at the distance of 10 pc and compare this with the magnitude
which it actually has, namely mv = 0.
In order to determine the absolute magnitude we have to know the
distance d of the star
because Icc
M v = mv(10pc)
My-mv=
- 2.5 Dog J(10pc)-log/(<*)]
(417)
4.6 The absolute magnitudes of stars
29
= -2.5[21ogd-21oglOpc]
= -51ogd[pc] + 5
(4.18)
mv M v is called the distance modulus of the star because it is determined by

the distance of the star.
For stars with parallax angles n > 0".05 we can measure n fairly accurately.
We can therefore determine their distances from their parallaxes. For these
stars we can also determine their absolute magnitudes M u ? M B , M v .
What are the stellar colors in absolute magnitudes?
From (4.18) we see
My = my- 5-logd[pc] + 5
M B = mB - 5 log d [pc] + 5
(4.19)
(4.20)
and
M B - M v = B - V = mB - m v .
(4.21)
This means we can always determine the colors from the apparent
magnitudes. The absolute colors are the same. See, however, Chapter 19 for
interstellar reddening.
5.1
Color magnitude diagrams of nearby stars
For nearby stars, say within 20 pc, we can determine the distances
from trigonometric parallaxes. From the apparent magnitudes and the
distances we can calculate the absolute magnitudes, i.e., the magnitudes
which the stars would have if they were at a distance of 10 pc. This means that
for absolute magnitudes we compare the brightness the star would have if it
were at a distance of 1 Ope with the actual brightness of Vega at its actual
distance, i.e. with its apparent brightness. It turns out to be quite instructive
to plot the absolute magnitudes of the stars as a function of their B V
colors. In Fig. 5.1 we do this for the nearby stars. While we might have
expected that stars with a given color could have quite different absolute
magnitudes, it turns out that this is generally not the case. Most of the stars
with a given B V color have the same absolute magnitude. Most of the
stars fall along one line in the color magnitude diagram. This line is called the
main sequence. The intrinsic brightnesses and the colors of these stars are
obviously determined by just one parameter, since they follow a onedimensional sequence. It turns out, as we shall see in Volume 3, that this one
parameter is the mass of the star.
While most of the stars follow the main sequence we see a few which are
outsiders. For a given value of the B V color most of these stars have a
larger intrinsic brightness than the main sequence stars, which means a
smaller absolute magnitude. Because brighter stars have smaller magnitudes
the smallest magnitudes are always plotted at the top in the color magnitude
diagrams. On the other hand, the B V colors increase towards the right in
the color magnitude diagrams. If, as we suspect, and as is actually the case,
the B V colors increase for lower temperatures because the black bodies
become more red for lower temperatures, then the temperatures decrease
toward the right-hand side of the diagram.
Let us compare two stars with a given B V, which means, as we suspect,
31
32
5 Color magnitude diagrams
with nearly the same temperature, but with different brightnesses. For a
given temperature the radiation of a black body per cm2 increases with T 4 as
we saw in the previous chapter. For a given temperature the total radiation
of a black body per cm2 is always the same. If the B V colors of the stars are
a measure of the temperatures then stars of a given B V should radiate
approximately the same amount of energy per cm2. If the stars were to
radiate like black bodies their total radiation, called their luminosity L
should be
L = 4nR2oT\
(5.1)
where R is the radius of the star and therefore 4nR2 is the surface area of the
star. If two stars with a given B V radiate vastly different amounts of
energy as shown by their different absolute magnitudes, then the brighter
star must have the larger surface area, which means it must be larger. The
Fig. 5.1. The color magnitude diagram for nearby stars is shown according to
Johnson and Morgan (1953). For better comparison with the following color
magnitude diagrams we have added an eye-fitted average curve for the main
sequence stars.
5.2 Color magnitude diagrams for open clusters

stars which are brighter than the main sequence stars are therefore called
giants. In contradistinction to the giants the main sequence stars are called
dwarfs. In astronomy, we have either giants or dwarfs but nothing in
between (except perhaps subgiants; see Section 5.3). When we now look at
the stars which are fainter than main sequence stars, we know that these stars
must be smaller than the main sequence stars if they have the same B V but
less radiation. Since the name 'dwarf has been used already for the main
sequence stars we are in trouble, we do not know what to call them. Most of
these faint small stars have rather bluish or white colors. They are therefore
called white dwarfs. Not all of them are actually white, some of them are
rather blue and a few are also red. So now we have blue (or hot) 'white dwarfs'
and we have fairly red (or cool) 'white dwarfs', which sounds confusing. But,
as always in astronomy, there is a long tradition and it would be even more
confusing to change the names now.
5.2
Color magnitude diagrams for open clusters
Since we see this one-dimensional main sequence for the nearby stars
we might wonder whether this is also true for the stars at larger distances.
But how can we determine their absolute magnitudes? Fortunately, we do
not have to know the distances of the stars which we plot in a color
magnitude diagram as long as we know that they are all at the same distance.
We could have plotted the same kind of color magnitude diagram for nearby
stars, putting them in our minds all at a distance of 20 pc, then they would all
have been fainter by a factor of 4, which means their magnitudes would all be
larger by Am =1.5. The whole diagram would be shifted down by
1.5 magnitudes but otherwise remain the same. If we have a group of stars
that are at the same distance, no matter what that distance is, we can still
compare their intrinsic magnitudes. The unknown distance only introduces
an upward or downward shift of all the magnitudes dependent on the
distance, but this shift is the same for all the stars at the same distance. There
are groups of stars visible in the sky which we can see are at the same
distance, these are the star clusters. The best known one which is easily seen
by the naked eye are the Pleiades. A photograph of this cluster is seen in
Fig. 5.2(a). There are so many stars close together that it cannot be an
accident. Most of these stars must actually belong together and form a star
cluster. They must then be all at the same distance. The differences of the
apparent magnitudes of these stars must then also be the differences of their
absolute magnitudes. In Fig. 5.3 we show the color magnitude diagram with
the absolute magnitudes for the Pleiades. Most of the stars again lie along
33
Fig. 5.2. Photographs of the well-known star clusters, the Pleiades (a), and the
Hyades {b\ in the constellation of Taurus. (From Burnham 1978.)
Pleiades
Hyades
I,
at as
B-v
(a)
02
OB
as
12
i*
B-v
(b)
Fig. 5.3. The color, absolute magnitude diagrams for the Pleiades (a) and
Hyades (b) star clusters are shown according to Arp (1958). A distance modulus
m v - My = 5.3 was used for the Pleiades. For the Hyades individual distance
moduli were used, averaging m v - M v = 3.08. Distance moduli larger by 0.2
magnitudes would be considered more appropriate now.
For better comparison of the different main sequences we have added some
eye-fitted average curves.
35
one sequence, probably the same main sequence which we see in Fig. 5.1, but
the sequence for the Pleiades has to be shifted upwards to smaller
magnitudes by m v M v ~ 5.3. In Fig. 5.2(b) we reproduce a photograph of
another well-known cluster, the Hyades, and in Fig. 5.3(b) we show the color
magnitude diagram for this cluster. The difference between the mv for a given
B V in the Hyades and the M v for the nearby stars is m v M v = 3.3.
There is, however, a distinction between the different color magnitude
diagrams with respect to the bluest stars seen. In the Pleiades we can see stars
which are much more blue than for instance in the Hyades. If we shift the
lower parts of the main sequences on top of each other, as seen in Fig. 5.4,
then the blue part of the main sequence for the Pleiades extends to smaller
magnitudes or to intrinsically brighter stars.
1
\'
\
-
2-
3-
K
Pleiades
^
\
5_
field stars
V
\
7 --
8-
9L
-0.2
0.5
1.0
1.5
B-V
Fig. 5.4. For the nearby fields stars and for the Pleiades and Hyades we have
plotted the eye-fitted average main sequences from Figs. 5.1 and 5.3 together in
one color magnitude diagram. A betterfitof the three sequences is obtained if
the distance moduli of the Hyades and Pleiades are increased by 0.2
magnitudes, as compared to the distance moduli adopted for Figs. 5.1 and 5.3.
At the high luminosity end the main sequences diverge because the average
ages of the field stars, the Hyades and the Pleiades are different. For further
discussion see Volume 3.
36
If we had superimposed color magnitude diagrams of a large number of

open clusters, we would have detected another group of stars which are even
brighter than the giants. These very bright and therefore very large stars are
called supergiants.
5.3
Color magnitude diagrams for globular clusters
So far we have shown only clusters with rather bright, blue stars. We
can easily distinguish the different stars in these clusters. They are called
open clusters. There are also clusters which look very different, these are the
so-called globular clusters. In Fig. 5.5 we reproduce photographs of two of
these globular clusters called M92 and M3. (They have the numbers 92 and 3
in the Messier Catologue of nebulous objects.) The globular clusters
contain 10000-1000000 stars. In the centre they cannot all be resolved.
Only in the outer parts of the clusters can we resolve single stars. When the
colors and magnitudes of these stars were measured it came as a big surprise
that the color magnitude diagrams for these clusters looked very different
from the ones for the open clusters. In Fig. 5.6 we show the color magnitude
diagrams for the globular clusters M92 and M3. You can hardly find a
sequence which looks similar to the main sequence. The main branches seen
are two nearly horizontal sequences of different magnitude and a nearly
Fig. 5.5. Photographs of the two globular clusters M3 and M92 are shown.
From Sandage and Walter (1966) and from Clayton (1983). Photographs from
Mt. Wilson and Palomar Observatories.
37
vertical branch of red stars. On the blue side of the diagram we see, however,
a short stub of a sequence which has the same inclination as the main
sequence for the open clusters, but no bright blue stars are seen in these
clusters. The color magnitude diagrams for other globular clusters all look
!
.. :
;
* 1
[..
I
1
"r.
S
* '
i/s
:3:':K:
* \:
'
'
M92
!
*i
V*
t
"**
1 ."'.
ye
13
-Off
-4
-02
-.2
' . ' .
02
Of
06
08
Color index P-V
1.0
1.6
1.8
Fig. 5.6. The color apparent magnitude diagram for the stars in the globular
cluster M92 is shown (top), according to Arp (1958). For the magnitude range
between the dashed lines a smaller field of the cluster was used than for the
brighter stars. P stands for photographic.
In the lower diagram the color apparent magnitude diagram for the stars in
the globular cluster M3 is shown, according to Arp (1958).
38
very similar to these except that the upper horizontal sequence always looks
different and sometimes consists only of a blue sequence or a short stub on
the red side. The upper horizontal sequence is called the horizontal branch
(HB). The lower horizontal sequence is called the subgiant branch and the
nearly vertical sequence is called the red giant branch. The short stub of a
sequence of faint stars in the diagram is indeed what is left over of the main
sequence. In Fig. 5.7 we show a schematic color magnitude diagram of a
globular cluster, in which the names for the different branches are given.
It is the main topic of Volume 3 to explain and understand the differences
between the color magnitude diagrams for open and globular clusters. It is
the big success of stellar evolution theory that we are actually able to
understand most of the details of these diagrams.
5.4
Photometric parallaxes for star clusters
In Section 5.2 we have seen that the main sequences in the different
color magnitude diagrams of open clusters can be shifted on top of each
other by a vertical shift, i.e., a shift by Amv only.* This is to be expected if
indeed all the main sequences agree but the clusters are at different distances.
The necessary vertical shift Amv = m v M v is then actually the distance
modulus for each cluster. This distance modulus determines the distance
-V
-3
36
-2
-I
' brand
i
i
IRLyrat
gap
heirizonia
0
I
^ /
vh
3
if
S
6
-az
0.2
OM
as
as
B-v
i.o
1.2
is
Fig. 5.7. A schematic color magnitude diagram for globular clusters is shown.
The names of the different branches are indicated. (From Arp 1958.)
* Except for a color correction which compensates for reddening by interstellar absorption
(see Chapter 19).
5.5 Photometric parallaxes for single stars
39
of the cluster, which can be derived from (4.21). For the Hyades with
m v My = 3.27, we derive a distance of 45 pc, for the Pleiades with
mv My = 5.5 we find a distance of 130 pc. By fitting the stub of the main
sequence for globular clusters with the main sequence for nearby stars we
can also determine the distances to the globular clusters if we can still see
the faint stars of the left over piece of the main sequence. For very distant
clusters, these stars are too faint to be observable except with the largest
telescopes and modern receivers. These new receivers, the CCDs (Charge
Coupled Devices) now permit us to observe the main sequences of globular
clusters to much fainter magnitudes than shown in Fig. 5.6. They look
indeed very similar to what we expected.
5.5
Photometric parallaxes for single stars
For main sequence stars we can also use this method to determine
the distance to a given star. Since for main sequence stars the absolute
magnitude M v is a unique function of the B V color, which we can read off
from Fig. 5.1, we only need to measure the B V color and the apparent
magnitude mv of the stars. For the measured B V color we read off the M v
from Fig. 5.1.f The values of M v for main sequence stars for different values
of B V are also given in Table 4.1. mv M v is the distance modulus from
which the distance can be obtained again according to (4.18). These
parallaxes are called photometric parallaxes.
Unfortunately, there are severe problems with this method of determining
parallaxes for single stars;
1. The star could be a giant or supergiant or perhaps a white dwarf. We
will see in Chapter 10 how we can distinguish these different types of stars by
their spectra.
2. The color of the star could be altered by interstellar absorption as we
will discuss in Chapter 19, where we will also see how we can determine this
color change.
t Except for a color correction which compensates for reddening by interstellar absorption
(see Chapter 19).
The luminosities of the stars
6.1
Fluxes, luminosities and angular radii
What is of interest to the astrophysicist even more than the amount

of energy which is emitted by the star in a given wavelength band, for
instance in the visual, is the total amount of energy emitted by a star. The
radiation of the stars is emitted in all directions. We can therefore say that the
same amount of energy goes through every cm2 of a sphere of radius d
around the star, see Fig. 6.1. If we are at a distance d from the star, then we
can measure the amount of energy arriving here above the Earth's
atmosphere per cm2. We denote by e this energy arriving per s per cm2,
e = nf, where / is called the flux*. The energy arriving per cm2 and
per s per cm wavelength band at the wavelength X is called nfx. Obviously,
the total flux
nf= f
J
nf.dl,
integrated over all wavelengths.
Fig. 6.1. The amount of energy leaving the star per s will pass through a
sphere of radius d some time later. The time difference is given by the travel
time between the star and the observer at distance d. The amount of energy
passing through the sphere of radius d per s is the same as the amount of
energy leaving the sphere of radius R per s, namely the star.
* Most observers write e = f and include the factor n in the definition of/. We prefer the definition
given above for theoretical reasons which will become apparent in Volume 2.
41
42
6 The luminosities of the stars
The energy nf which we receive does, of course, depend on the type of star
which we observe and on its distance. At any given moment the whole star
emits a certain amount of energy which is given by the amount emitted
per cm 2 on the stellar surface, which we call TTF, multiplied by the number of
cm2 on the stellar surface, which is given by
total surface area = 47LR2,
(6.1)
where R is the radius of the star. The total amount of energy emitted by the
star per s is then given by
E(total)/s - nF-4nR2 = L = luminosity.
(6.2)
The energy emitted by the star per second is called the luminosity L of the
star. After a time t = (d R)/c this same amount of energy will pass through
the sphere at distance d. If there is no energy absorbed on the way, for
instance by interstellar material, then the amount of energy passing through
the sphere at distance d must be the same as was emitted by the star t seconds
earlier. This leads to the equation
if the luminosity of the star does not change in time. Otherwise (6.3)
describes the luminosity at the time t = (d R)/c before we observe the flux
nf. (As usually dRwe can replace d R by d). From (6.3) we then derive
= nf-(d/R)2
(6.4)
We can determine the luminosity L by measuring the amount of energy

arriving from the star above the Earth's atmosphere if we know the distance.
We can determine the amount of energy leaving the stellar surface per cm 2
and per s if we measure nf and if we determine the ratio R/d9 which is the
angular radius of the star (see Fig. 6.2), if we measure the angular radius in
radians. It is important to notice that it is often easier to determine nF the
surface flux on the star than it is to determine the luminosity, because for
Fig. 6.2. The ratio R/d is the angular radius of the star as measured by the
observer at distance d. A pencil piece of length s at a distance / from the
observer covers the diameter of the sun. The ratio s/l can easily be measured,
and equals 2R/d.
6.2 The luminosity of the sun
43
stars not too distant it is easier to measure the angular diameter than it is to
measure their distance.
6.2
The luminosity of the sun
The sun is close enough for us to see its disk and not just an
unresolved dot of light as we see with our eyes for the other stars. For the sun
everyone of us can therefore easily measure the angular diameter. An easy
way to do so is to take a pencil in your hand, stretch out your arm and see
what length s of the pencil will cover up the diameter of the sun (see
Fig. 6.2). As can be seen from Fig. 6.2 the ratio of s to the length of your
arm / gives you the angular diameter of the sun, which turns out to be
about D/d 1/100 (actually 0.0093). It is a very fortunate accident that the
angular diameter of the moon is almost exactly the same, otherwise we
would never be able to see total solar eclipses.
Knowing that the flux nf from the sun equals the measured value for
the solar constant S = 1.38 x 10 6 ergcm~ 2 s" 1 = TC/(sun) we find with K/d =
D/(2d) = 1/2 the solar angular diameter that
7iF = S - 2 0 0 2 ^ 6 - 1 0 1 0 e r g c m - 2 s - 1
(6.5)
for the sun.

For the determination of the solar luminosity we have to know the
distance to the sun, which, as we saw, is much more difficult to determine
than the angular diameter, but which we know is l a u = 1.49 x 10 13 cm.
Inserting this into (6.3) we obtain for the sun
L(sun) = 3.96 x 10 33 erg per s.
(6.6)
This is a big number which really does not tell us a lot. It means more to us
when we look at the energy generation of the sun. Knowing that the sun loses
per s the energy L and seeing that it does not change with time, we must
conclude that the sun must replenish the energy loss by some sort of energy
generation. We can say that the energy generation per s equals L = 3.96 x
10 33 erg. We understand this number better if we express L in kW, 1 erg =
10" 1 0 kW, so L(sun) = 4 x 10 23 kW. From the orbits of the planets
around the sun and from Kepler's third law we can determine the mass of the
sun and find (see Section 9.5) that M o = M(sun) = 1.98 x 10 33 g. (If we
measure the luminosity of the sun in erg s~x and the mass in g the ratio of
L/M = 2. This can be remembered easily.) If we now calculate how much
energy the sun produces per g of material we are surprised how little this is,
namely 4 kW for each 2 x 10 10 g of material, or 2 kW for each 10 000 tons of
44
material. If we were to imitate the solar energy generation to cover our

energy needs it would not do us much good, we need to do it much more
efficiently. The sun has such an enormous mass that it can afford to be very
inefficient and still be a very large energy source.
6.3
Luminosities of stars and bolometric magnitudes
If we measure the amount of energy which we receive from a star

per cm per s above the Earth's atmosphere we can determine the total
amount of energy emitted by the star at all wavelengths if we also know the
distance to the star. Actually we can only determine the luminosity for those
wavelengths for which we receive energy above the Earth's atmosphere.
Before we had rockets and satellites we could actually measure only the
luminosity for those wavelengths which could still penetrate the Earth's
atmosphere. But even now with satellites we do not get radiation from all
wavelengths for most of the stars, because the interstellar material between
us and the stars absorbs the light for wavelengths shorter than 912 A for all
the distant stars and absorbs a large fraction of that light even for the nearby
stars. If stars emit a large fraction of their energy at such short wavelengths
we cannot measure their luminosities directly, we have to use theoretical
extrapolations to estimate the amount of energy emitted in those wavelength
bands. We will see later (in Volume 2) how this can be done.
The luminosities or brightnesses of the stars integrated over all wavelengths are again measured in magnitudes, the so-called bolometric magnitudes. We can again distinguish apparent and absolute bolometric magnitudes, though generally only absolute bolometric magnitudes are in use. The
difference between visual magnitudes mv and bolometric magnitudes mbol is
called the bolometric correction BC. We then have
2
mhol = mY-BC
or Mbol = MY-BC.
(6.7)
Different authors disagree about the sign on the right-hand side of (6.7),
which leads to different signs for the bolometric correction. This is of course
only a matter of definition. Since the total radiation at all wavelengths is
more than the radiation in a limited wavelength band, which means the
bolometric brightness is larger than the visual, we expect that the bolometric
magnitudes are smaller than the visual magnitudes. If we take the definition
(6.7) with the sign, then the BC are generally positive, though unfortunately
not always for the following reasons. While all the apparent magnitudes for
any wavelength band for Vega are 0 this is not the case for the apparent
bolometric magnitude. For the bolometric magnitudes astronomers agreed
6.3 Luminosities and bolometric magnitudes
45
on a different normalization. As we said the bolometric magnitudes measure

the radiation for all wavelengths including the infrared and the ultraviolet
region. For stars which have a large fraction of their radiation in those
wavelength regions the bolometric corrections must be large. From the
energy distribution of the black body radiation we know that hot objects
emit a large fraction of their radiation at very short wavelengths while cool
objects emit a large fraction in the infrared. We therefore expect the
bolometric corrections to be large for very hot and very cool stars. There
must be a minimum for intermediate temperatures. In Fig. 6.3 we have
plotted the bolometric corrections for main sequence stars as a function of
Fig. 6.3. The bolometric corrections, BC, are shown for main sequence stars
with different B - V colors (solid line). These bolometric corrections have a
minimum at B - V = 0.3. For main sequence stars the minimum bolometric
correction was set equal to zero by definition. This now requires a change of
sign of the BC for some supergiant stars as shown by the supergiant curve in
this diagram (dashed curve).
46

Table 6.1. Effective temperatures T*ff and bolometric
corrections BC for main sequence stars and supergiants
M bol = My - BC.
B-V
Main
sequence
BC
Main
sequence
Te(f
Super
giants
-0.25
-0.23
-0.20
-0.15
-1.10
-0.05
0.00
+ 0.10
+ 0.2
+ 0.3
+ 0.4
+ 0.5
+ 0.6
+ 0.7
+ 0.8
+ 0.9
+ 1.0
+ 1.2
24500
21000
17700
14000
11800
10500
9480
8530
7910
7450
6800
6310
5910
5540
5330
5090
4840
4350
2.30
2.15
1.80
1.20
0.61
0.33
0.15
0.04
0.00
0.00
0.00
0.03
0.07
0.12
0.19
0.28
0.40
0.75
26000
23500
19100
14500
12700
11000
9800
8500
7440
6800
6370
6020
5800
5460
5200
4980
4770
4400
BC
Super
giants
2.20
2.05
1.72
1.12
0.53
0.14
-0.01
-0.09
-0.10
-0.10
-0.09
-0.07
-0.03
+ 0.03
+ 0.10
+ 0.19
+ 0.30
+ 0.59
1. Effective temperatures will be discussed in Chapter 8
the B V colors. They have a minimum for B V = 0.3. Astronomers

decided a long time ago to set the minimum bolometric correction to zero by
definition. For main sequence stars with B V = 0.3 we therefore have by
definition mbol = m v and M bol = M v .
Vega has B V = 0 so for Vega mbol is not equal to m v which means it is
not equal to zero. For stars like Vega the bolometric correction BC = +0.15.
The reason for this irrational definition of the BC was the desire to have
bolometric corrections as small as possible without getting negative values.
It now turns out, however, that we find negative bolometric corrections for
the supergiants. So the original reason for the choice of the zero point for the
bolometric corrections no longer holds. It seems to be much better to define
the zero point for the mbol consistently with the zero point for all other
apparent magnitudes, but again astronomers do not like to make changes.
With this definition of the zero point the bolometric correction for the sun
becomes BC(sun) = 0.07, and the absolute bolometric magnitude of the sun
6.3 Luminosities and bolometric magnitudes
47
becomes Mbol(sun) = 4.75. It seems easiest to remember this number and

derive all other absolute bolometric magnitudes from the relation
Mbol(star) - Mbol(sun) = - 2.5Bolometric corrections derived from measurements and theoretical extrapolations are given in Table 6.1.
Angular radii of stars
7.1
The problem
For the study of the structure of the outer layers of a star and also for
the overall structure of the star it is very important to determine the surface
flux of the radiation for the stars, which we call F. We saw in the previous
section that the amount of energy leaving 1 cm 2 of the stellar surface per s,
nF, can be obtained from the amount of energy reaching 1 cm 2 per s above
the Earth's atmosphere, provided we know the angular radius of the star.
Basically there are four methods in use to determine angular radii: the
Michelson interferometer, the intensity interferometer, or Hanbury Brown
interferometer, and lunar occultations. Speckle interferometry has also been
used.
The basic difficulty with measuring stellar diameters is the fact that the
angular diameters are so small that the blurring of the stellar image, which
occurs when the light passes through the atmosphere, completely wipes out
the size of the image. This blurring of the image is called the 'seeing' and is
due to the turbulence elements in the air, which have slightly different
temperatures and densities and therefore slightly different refractive indices.
The light beams passing through different turbulence elements at different
times therefore do not arrive exactly parallel at the telescopes but seem to
come from slightly different directions and will be focused by the telescope at
slightly different points. Since during the time of exposure of a photographic
plate different turbulence elements pass by the telescope the images due to
the different turbulence elements are all registered by the photographic plate.
The sum of all these images then gives a rather large image of the star ranging
from image sizes of about 1/2 arcsec for excellent seeing to 20 or 30 arcsec for
very bad seeing. This means that even for the best seeing conditions stellar
images of less than 1/2 arcsec cannot be resolved with ground-based
telescopes which have to observe through the Earth's atmosphere. If we
could observe for very short instances, when all the light beams from the star
49
50
7 Angular radii of stars
reaching the telescope pass through the same turbulence elements, we would
be able to resolve much smaller images, but we would not receive enough
light in such a short time. The interferometers circumvent this problem by
comparing at any instance two beams of light which arrive at the telescopes
simultaneously and have therefore in general passed through the same
turbulence elements. The beams may change their directions slightly
in time but they do so together, and that does no harm.
When the Space Telescope is in orbit we will be able to resolve images of
0.02 arcsec. Also the astrometric satellite Hipparchos will be able to resolve
images of such small separation.
7.2
The Michelson interferometer
The Michelson interferometer was the first instrument in use to

determine angular diameters of stars. Fig. 7.1 shows how the Michelson
interferometer works. Two parallel light beams from the star enter the
telescope lens at points A and , where we have to provide two holes in a
screen for the light to enter. At the focal point of the lens the beams form an
image at the point Pl9 where we see the image of the star. Since at points A
and B the light enters through small holes there is light scattered at the edges
of these holes. This scattered light is dispersed in all directions and appears
on the focal plane all around point P i . Scattered light comes from both holes,
but the light coming from point A arrives at point P with a different phase
from the light coming from point B because of the different path lengths. If at
a point P the path length difference is Xj2 the phases differ by n and the
electric and magnetic vectors cancel, there will be no light at the point P. On
the other hand, if the path length difference is X the phase difference is 2% and
the electric and magnetic vectors add up, we see another, secondary light
maximum at P = P2- Another light maximum occurs at point P = P3 (not
plotted) for which the path length difference is 2/1. From Fig. 7.1 we see that
the path length difference As = yD, where D is the distance between the
entrance holes A and B. For light maxima we require As = 2, for light
minima As = A/2. The distance / between points P x and P2 is given by / = yd
where d is the focal length.
For light maxima we therefore require
(7.1)
for light minima
l/2
(7.2)
7.2 The Michelson interferometer
51
Let us now look at an image of a star 2 which is a small angular distance 8

from star 1. The primary image of star 2 appears in the focal plane at a
distance 8 from the primary image of star 1. The image of star 2 also shows
secondary maxima just the same as those from star 1. If the angular distance
Star 2 Star 1
Al
As
Star 1 interference pattern

Star 2 interference pattern '
Fig. 7.1. A schematic plot of the light paths in the Michelson interferometer is
shown. Light from star 1 enters through the entrance slits at points A and B.
The primary image of star 1 is formed at point P1 in the focal plane. The light
scattered at the entrance slits forms an interference pattern around Pt.
Secondary maxima occur around points P2, P3 etc., where the path length
differences for light from A and B are multiples of the wavelength X of the light.
Light minima occur halfway between these points. Light from a star 2, which is
at an angular distance S from star 1 forms a similar interference pattern around
its primary image S^ If the light maxima from star 2 fit into the light minima
of star 1 the interference pattern in the focal plane becomes invisible if both
stars are equally bright. In this case (7.3) holds, from which S can be
determined.
52
d of star 1 from star 2 happens to be

(7.3)
then the primary image of star 2 will appear at the place of minimum light for
star 1 and the subsequent light maxima for star 2 all fall on the subsequent
light minima for star 1. This means that the interference pattern of star 1 will
be wiped out by the interference pattern of star 2, provided both stars are
equally bright. If this happens we know that the angular distance between
star 1 and star 2 is given by (7.3). We have then actually measured the
angular distance d of star 1 from star 2. Given a distance d between two
neighboring stars we can adjust the distance D between the two entrance
holes in such a way that the interference patterns for the two stars becomes
invisible, we then know that (7.3) must hold. We can measure the distance D
for which this happens and then calculate 3 from (7.3).
How can this help us to measure angular radii of stars? In our minds we
can subdivide the star into two half stars (see Fig. 7.2), half 1 and half 2. The
two half stars are now separated by approximately the angular radius of the
star 9/2 = /^/distance. Knowing the brightness distribution on the stellar
surface we can determine the distance of the 'centres of gravity' for the two
halves more accurately. We can then make the same kind of observation as
described above for the two stars. When the interference pattern, which is
now observed for just one star at the focal plane, becomes invisible we know
that 3 = ymin = X/2D. We can again adjust D so that this happens. Instead of
taking one very big lens with a screen in front with two openings, it is easier
to take two separate lenses at a large distance apart, and then vary the
positions of the two lenses until the interference pattern becomes
invisible.
Due to the seeing effects the interference patterns of both stellar images will
move back and forth on the focal plane, but they always move together, so
the light maxima for star 2 always fall on the light minima for star 1, even
Fig. 7.2. In our minds we can divide the disk of a star into two halves, 1 and 2.
The two halves are separated by approximately the angular radius of the star,
which can thus be measured with the Michelson interferometer.
73 The Hanbury Brown interferometer
53
Table 7.1 Angular diameters measured with the Michelson

interferometer
Star
a Bootis
a Tauri
a Orionis
P Pegasi
a Herculis
o Ceti
a Scorpii
Spectral1
type
Distance D
between lenses
(meters)
Angular diameter
KOIII
K5III
MOIII
M5III
M8III
M7III
MOIII
7.3
7.3
3.0
6.7
4.9
3.0
3.7
0.020
0.020
0.047 var.
0.021
0.030
0.047 var.
0.040
<%>
1. The spectral types will be discussed in Chapter 10.

2. This is the angular diameter which the star would have to have if its disk had
uniform brightness. All stars are expected to be dimmer at the limb than in the
center, as is observed for the sun (see Volume 2). The true angular diameter is
therefore probably slightly larger.
though they are at different places on the screen. As long as the light beams
from both stars go through the same turbulence elements the seeing does not
disturb the observation. But this condition sets a limit to the distance D
between the lenses, if D is too large then the light beams going through the
two lenses pass through different turbulence elements and have additional
phase differences which disturb the interference pattern. This upper limit on
D sets a lower limit on the distance 6 and on the angular radius which can be
measured. Only angular radii of very large stars can be measured in this way.
In Table 7.1 we give some values for angular radii of stars which have been
measured in this way.
7.3
The Hanbury Brown interferometer
The Hanbury Brown interferometer works on a very different

principle. It also uses two large mirrors but interferences are measured for
each mirror alone and then the interferences for these mirrors are compared
(see Fig. 7.3). If the light falling on these two mirrors is coherent, then there is
a good correlation between the interferences measured for the two
telescopes. How can there be interferences in each of the separate mirrors?
Observations are made in a narrow wavelength band and the light waves of
slightly different wavelengths show interferences, which are also called beat
phenomena. The amplitude of a wave with a given frequency itself has a wave
54
pattern. In Fig. 7.4 we show the beat phenomenon on a wave pattern

resulting from the superposition of two waves of slightly different frequencies. The frequency of the amplitude modulation is given by the difference
in the frequencies of the two interfering waves:
Writing
*-icot
we find quite generally,

(D2
COS COx t + COS (D2t = 2 COS ^ r
-t'COS
t,
(7.4)
which describes a wave with a frequency (co1 + a>2)/2 with an amplitude

which is modulated with the frequency (a>1 co2)/2.
Different wave amplitudes of a light wave mean different intensities. Due
to this beat phenomenon light fluctuations are observed in each telescope
over very short time scales. With the Hanbury Brown interferometer
intensities therefore have to be measured at very short time intervals in order
REFLECTOR
Fig. 7.3. The general layout of the Hanbury Brown interferometer is shown.
(From Hanbury Brown, 1974.)
Fig. 7.4. The beat phenomenon resulting from the superposition of two waves
of nearly equal frequencies is shown. The amplitude of the resulting wave is
itself a wave.
7.3 The Hanbury Brown interferometer
55
to measure these intensity fluctuations. In order to get measurable energies

over such short time scales large mirrors have to be used and also bright stars
are needed for the measurements. As we shall see soon, the stars must
not be too large in order to be measurable with this type of instrument,
otherwise the mirrors would have to be too small and could not gather
enough light.
In order to understand how these intensity fluctuations can be used to
measure angular distances between nearby stars or to measure angular radii
of stars we have to give a clear definition of what is meant by coherent light.
In the strict definition light is coherent or two photons are coherent if they
belong to the same quantum cell in phase space. The phase space element is
given by
d (phase space) = dx dy dz dpx dpy dpz
x9 y, z are the space coordinates and px, py and pz are the momentum
coordinates in the directions of x, y and z. For coherent light the two
photons being coherent have to be in a phase space element d (phase space)
which is only one quantum cell, which means it cannot be larger than /z3,
where h is the Planck constant. The intensity fluctuations measured in the
two mirrors are correlated as long as the light falling on the two mirrors is
coherent. From the schematic representation of the Hanbury Brown
interferometer in Fig. 7.5 we see the following relations. We put the
coordinate system in such a way that the x coordinate is in the direction from
Fig. 7.5. A schematic plot of the geometry of the Hanbury Brown

interferometer is shown: d is the distance between the two mirrors, D the
distance between two points near the limb of the star. 6 is the angular
diameter of the star, and r the distance to the star, y is the angular distance of
the mirrors as seen from the star. The orientation of our coordinate system is
also shown.
56
the star to the interferometer. The y and z directions are in the plane
perpendicular to this direction. The photons reaching the mirrors at any
given time come from any points Px and P2 on the star, which means they
can, in the y and z directions, have distances of the order of the stellar
diameter D. This means Ay = Az = D. If they arrive at the mirrors at the same
time they must have had a distance Ax equal to the path length A, which
means Ax = A. From Fig. 7.5, we see that A = Dy and y = d/r9 where d is the
distance between the two mirrors and r the distance of the star.
We now look at the differences in the momenta px, py, pz of the photons
reaching the mirrors. The main direction of propagation is the x direction,
therefore px = hv/c, where v is the frequency of the light observed. If a
wavelength band of width Av is used, then the spread in px is Apx = hAv/c. The
photons reaching mirrors A and B must have a difference in Apy = pxy where
y = d/r. The same holds for Apz.
We thus find for the volume element in phase space populated by the
photons which reach the two mirrors at any given time
AxAyAzApxAPyApz = AD2(yY
= AVAP
(7.5)
If we observe the photons in the two mirrors to be coherent they must all
belong to the same quantum cell, which means we must conclude that
AFAP < h3
(7.6)
A02^72^3
(7.6a)
According to (7.5) this means
With A = Dy = Dd/r we find that for coherence we must have with X = c/v
and after dividing by h3
^l
(7.6b)
or, taking the third root we find
z>Vsr -'V^
(7J)
where we have made use of the relation that the angular diameter = D/r.
For larger distances d than given by (7.7) the photons reaching the two
mirrors are not coherent any more because the py and pz components of their
momenta differ too much. Starting at small distances d between the mirrors
7.3 The Hanbury Brown interferometer
57
and slowly increasing d we expect first to measure a good correlation

between the intensity fluctuations in the two mirrors. Once we reach the
distance d for which the equal sign holds in (7.7) we expect the correlation to
decrease and finally go to zero for large distances of the mirrors: If A = Ax
becomes too large the photons reaching the two mirrors are not coherent
any more. For which A does this happen? From (7.7) we see that for the equal
sign we find
I ,.
(7.8)
is
p Crucis
o
82
z 0
20
40
60
80
100
Baseline (in m)
2 3
8
120 *140
160
Gruis
|
2 1
0
20
40
60
80
Baseline (in m)
100
120
Lyrae
a 1
0
20
40 60
Baseline (in m)
Fig. 7.6. The correlation between the intensity fluctuations of the two mirrors
of the Hanbury Brown interferometer is shown as a function of distance
between the mirrors. For large angular diameters, (see a Lyrae) the correlation
breaks down for very small distances. For small angular diameters of the stars
(for instance, /? Crucis) the correlation keeps up for much larger distances.
(From Hanbury Brown 1974.)
58
with Av/v = 10 ~2 and ^/(v/Av) = 4.6 we find that the coherence stops for
A5A.
From (7.7) we see that large values of can only be measured with small
distances d of the mirrors, but the smallest distance of the mirrors is limited
by the size of the mirrors. For the measurement of large 0 we would need
very small mirrors which would, however, gather too little light. So very
large angular diameters can not be measured with the Hanbury Brown
interferometer, while small angular diameters can be measured well as long as
the star is still bright enough to measure very short time (10 " 9 s) fluctuations.
For small X small angular diameters can be measured especially well. The
Hanbury Brown interferometer works best for bright stars with small
angular diameters, which means for hot luminous main sequence stars. It is
therefore complementary to the Michelson interferometer with which only
large angular diameters can be measured, which are mainly observed for
cool supergiants.
In Fig. 7.6 we show for some stars the measured correlations between the
fluctuations in the two mirrors as a function of distance between the mirrors.
For large distances the correlation drops. The smaller the angular diameter,
the smaller the distance for which the correlation decreases.
In Table 7.2 we give the angular diameters of stars as measured by
Hanbury Brown.
7.4
Angular diameters from lunar occultations
For stars near the ecliptic, which means near the plane of the Earth's
orbit, angular diameters of stars can also be measured by means of lunar
occultations when the dark side of the moon covers up the star.
Since the moon orbits the Earth in 28 days it apparently goes around the
sky once a month, it can eclipse stars which happen to be behind its orbit as
seen by us. Because the moon's orbit is inclined with respect to the Earth's
orbital plane by about 5 and because the Earth orbits the sun, different stars
can be occulted at different times. Even though only stars in a small angular
range about the ecliptic can be studied in this way a fairly large number of
stellar angular diameters have been determined by lunar occultations. Of
course, we always have to wait for a time when the stellar eclipse happens
naturally and hope that at that particular time the weather is also
cooperative.
In principle, we just have to measure the time between the start of the
decline of the stellar light until the light is completely gone. The moon moves
360 in one month, or roughly in 3 x 106 s. This means that in 1 s of time it
59
7.4 Angular diameters from lunar occupations

Table 7.2 Angular diameters, 9, measured with the Hanbury Brown
interferometer
Star
number
Star
name
472
1713
1790
1903
1948
2004
2294
2326
2421
2491
2618
2693
2827
2943
3165
3207
3685
3982
4534
4662
4853
5056
5132
5953
6175
6556
6879
7001
7557
7790
8425
8728
a Eri
1.
2.
3.
4.
pOri
y Ori
Ori
COri
jcOri
j?CMa
a Car
y Gem
aCMa
CMa
<5CMa
rj CMa
aCMi
CPup
y2 Vel
Car
a Leo
Leo
y Crv
Cru
a Vir
s Cen
S Sco
COph
aOph
Sgr
a Lyr
a Aql
a Pav
aGru
aPsA
Angular diameter3
in 10~3s of arc
2
Type
B3(Vp)
B8(Ia)
B2 (III)
B0(Ia)
O9.5(Ib)
B0.5 (la)
Bl(II-III)
F0 (Ib-II)
A0(IV)
A1(V)
B2 (II)
F8 (la)
B5 (la)
F5 (IV-V)
O5(f)
WC8 + O9
Al(IV)
B7(V)
A3(V)
B8(III)
B0.5(III)
Bl(IV)
Bl(III)
B0.5(IV)
O9.5(V)
A5(III)
A0(V)
A0(V)
A7(IV,V)
B2.5(V)
B7(IV)
A3(V)
1.92 + 0.07
2.55 0.05
0.72 0.04
0.69 0.04
0.48 0.04
0.45 0.03
0.52 0.03
6.6 0.8
1.39 0.09
5.89 + 0.16
0.80 + 0.05
3.60 0.50
0.75 + 0.06
5.50 + 0.17
0.42 0.03
0.44 0.05
1.59 0.07
1.37 0.06
1.33 0.10
0.75 0.06
0.722 0.023
0.87 + 0.04
0.48 0.03
0.46 0.04
0.51 0.05
1.63 0.13
1.44 0.06
3.24 0.07
2.98 0.14
0.80 0.05
1.02 0.07
2.10 0.14
Temperature4
[reff<7]/K
13 700 600
11 500 700
20 800 1300
24 500 2000
26 100 2200
30 400 2000
25 300 1500
7500 250
9600 500
10 250 150
20 800 1300
14 200 1300
6500 200
30 700 2500
29 000 3000
9500 350
12 700 800
9050 450
13 100 1200
27 900 1200
22 400 1000
26 000 1800
8150 400
9650 400
9250 350
8250 250
17 100 1400
14 800 1200
9200 500
Bright star catalog number (Hoffleit and Jaschek, 1982).

Spectral type and luminosity class (in brackets), to be discussed in Chapter 10.
True angular diameter allowing for the effects of limb-darkening.
Effective temperatures will be discussed in Chapter 8.
60
moves about 0.5 arcsec. If the star has an angular diameter of 10 arcsec,
then we have to have a time resolution of the order of 10" 3 s or better.
During such short time intervals we must be able to measure the radiation
flux of the star. This again means that the stars have to be bright and we have
to use large telescopes in order to gather enough light in 10~ 3s.
In reality the stellar light does not disappear monotonicly, but the light
rays from the star are diffracted at the moon's limb (Fresnel diffraction). The
superposition or the 'interference' of the different diffracted beams, originating on different parts of the star, results in an interference pattern (see
Fig. 7.7, where two such calculated interference patterns are reproduced).
The interference pattern depends on the angular diameter of the star. The
measured pattern has to be compared with the ones calculated for different
angular diameters. The one which fits the observation best tells us the
POINT SOURCE
UNFORM DISK 0-006*
INTENSITY
20
1
s
1
t
10
1
20
1
20
1
25
SO
METRES
1
30 X 10 ~ * StC Of
1
40
I
SO
I
60
40
30
1
19
1
30
MILLISECONDS
Fig. 7.7. The calculated interference pattern during lunar occultations is shown
for a point source. The three scales of the abscissae represent the size of the
pattern on the Earth in meters, the angle subtended by this pattern at the
moon in seconds of arc, and the variation of light with time at a fixed point in
milliseconds. (From Hanbury Brown 1968.)
7.4 Angular diameters from lunar occultations
61
^v \\
60
100
40
EO
-40
0 - 2 0
-60
M E T E R S
180
150
120
60
30
-30
-60
-90
M E T E R S
Fig. 7.8. The measured interference patterns are shown (asterisks) for the lunar
occultation of Y Tau (top) and for U Ori (bottom). Also shown are the
theoretical patterns for an angular diameter of 0 = 7.98 milli-arcsec (center
curve for Y Tau), which is considered to give the best fit for Y Tau, and the
theoretical curves for 0 = 8.98 milli-arcsec (top graph) and for 0 = 6.98 milliarcsec (bottom curve for Y Tau). In the plot for U Orionis the theoretical curve
for 0 = 15.45 milli-arcsec is shown, which is considered to be the best fit curve
(upper curve), and also the one for a point source (lower curve). (From
Ridgway, Wells, & Joyce 1977.)
62
angular diameter of the star. In Fig. 7.8 we reproduce some of the observed
patterns and the best fit calculated ones.
In Table 7.3 we give some of the stellar angular diameters determined from
lunar occultation measurements.
Table 7.3 Some angular diameters measured from lunar occultations
according to Ridgeway, Wells and Joyce (1977)
HR No. 2 Star
2286
867
1977
5301
4902
7150
9004
3980
7900
1.
2.
3.
4.
Sp. type3
\x Gem M3III
RZ Ari M6III
Y T a u C5II
M2III
\jj Vir
M3III
\2 Sgr K1III
TXPsc C5II
31 Leo K4III
v Cap M2III
B-V
mv
10 3 arcsec
n
arcsec
R/RO
TM
1.64
1.47
3.03
1.72
1.60
1.18
2.60
1.45
1.66
2.88
5.91
6.95
4.91
4.79
3.51
5.04
4.37
5.10
13.65
10.18
8.58
4.41
5.86
3.80
9.31
3.55
4.72
0.020
0.014
?
0.012
0.021
0.011
- 0.004
?
0.019
73.2
78.0
?
39.4
29.9
37.1
?
?
26.6
3650
3160
?
4040 4
3530
4210
?
4000
3440
Considering the limb darkening of the disk.

Number in the Bright Star catalogue, Hoffleit and Jaschek (1982).
From the Bright Star catalogue
Assuming BC = 2.0.
8
8.1
General discussion
As we have seen in Chapter 6 (6.4), we can determine the surface flux

nF per cm 2 and per s if we measure the flux nf arriving above the Earth
per cm 2 and per s and if we know the angular diameter or angular radius R/d
for the star. With 9 = R/d we have
nF = nf'l/62.
(8.1)
The surface flux tells us how much radiation is emitted from 1 cm 2 per s by
the star. In order to get another estimate for the temperature of a star we can
compare this surface flux of the star with the flux emitted by a black body of
temperature T which is, according to (4.5),
nF (black body) = a T 4 .
(8.2)
If the star did radiate like a black body we could use this equation to
determine its temperature. We know, however, that the stars are not black
bodies because they are not well insulated from the surroundings, they
actually lose a lot of energy all the time to the surroundings, fortunately for
us, because otherwise we would not be able to see them and we would have
no knowledge of their existence.
So we know that the stars are not black bodies but nevertheless we can
compare their energy loss per cm 2 with that of a black body and say that if
they did radiate like black bodies then their temperatures would have to be
given by
a T 4 = %F = a T 4 f .
(8.3)
The temperature defined in this way is called the effective temperature of
the star. It is another way of describing the surface flux of a star. All it means
is that the energy loss of the star per cm 2 and per s is the same as that of a
black body with T= Teff.
63
64
8 Effective temperatures of stars
If we measure the black body radiation coming out of the hole in the box, that means the
black body is permitted to lose energy by radiation it is then, of course, in the strict sense, not a
black body any more because it is no longer insulated from the surroundings. We have to
make sure that the hole in the box through which the radiation escapes is very small, such that
the energy loss remains very small in comparison with the energy content of the black body.
This consideration also gives us hope that, may be, the stars are not so different from a black
body after all because their energy loss, large as it may be, may still be small compared to their
energy content.
Equation (8.3) is the definition of the effective temperature of a star. In

Volume 2 we will see that this Teff is indeed the temperature in the surface
layer of the star from which the radiation which we receive is mainly emitted.
We can consider the effective temperature as an average temperature of the
surface layers of the star.
In Chapter 7 we have discussed how angular radii can be measured for
several groups of stars. For these stars we can directly determine Teff if we
know nf. The determination of nf requires, however, a knowledge of the
stellar radiation at all wavelengths including those which cannot penetrate
the Earth's atmosphere.
8.2
The solar surface flux and effective temperature
The sun fortunately does not emit much radiation in the infrared
and ultraviolet wavelengths which cannot penetrate the Earth's atmosphere.
We can now even measure the total radiation of the sun from satellites above
the Earth's atmosphere and find
7c/(sun) = S = 1 . 3 8 x 106 e r g o n " 2 s" 1 .
(8.4)
With the known angular diameter of the sun (see Section 6.2), namely
= 0.0093 radians, we calculate
7iF(sun) = 6.3 x 10 1 0 ergcm" 2 s" 1 .
(8.5)
With TcF(sun) = a T*u (sun) we calculate

Teff(sun) = 5780K
(8.6)
It is very helpful to remember this value as a reference.

We remember that we estimated the solar temperature to be about 6000 K
from the wavelength Amax for which we observe the maximum of the solar
energy distribution, using Wien's displacement law, (4.7). The close agreement of this value with Tcff for the sun gives us hope that these values do
actually tell us something about the temperature of the outer layers of the
sun, and that the solar radiation is actually not so different from that of a
65
8.3 Effective temperatures of stars
black body. In Volume 2 we will specify what we mean by 'not so different.'

8.3
For stars more blue than the sun, which we expect to be hotter than
the sun, we expect more radiation at the short wavelengths, i.e. in the
ultraviolet. With the astronomical satellites we can now measure radiation
for all wavelengths. In principle we could therefore determine the/ A for all
wavelengths. There is however one basic difficulty for which there is no
straightforward solution: the interstellar material which is between us and
the stars has a very large absorption coefficient for light with wavelengths
shorter than 912 A (see Chapter 17). We do not get any radiation in these
wavelengths for stars further away than say about 50 pc. Also for wavelengths longer than 912 A there is already rather strong absorption for stars
with distances larger than about 200 pc. For hot stars we can therefore only
determine nf accurately for stars which are closer than about 20 pc. There
are no very blue stars that close. We therefore do not have accurate
knowledge of Teff for very hot stars, though we have some good estimates
Table 8.1 Adopted values of Teff for stars with different BV colors and
different luminosity classes (The uncertainty is at least 5%)
B-V
V1
III2
I3
-0.32
-0.31
-0.30
-0.29
-0.28
-0.27
-0.26
-0.24
-0.22
-0.20
-0.16
-0.12
-0.08
-0.04
0.00
+ 0.04
+ 0.08
37000
33000
30500
28700
27200
25700
24500
22000
19800
17700
14500
12500
11100
10200
9480
9080
8700
37000
33000
30500
28700
27200
25700
24500
22000
19800
17700
14500
12500
11100
10200
9480
9080
8700
35000
34700
34400
33800
32500
26000
23500
21200
19100
15700
13500
11900
10700
9600
9000
8500
1. V stands; for main sequence stars

2. Ill stands for giants
3. I standsfor supergiants
B-V
III
0.12
0.16
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
1.10
1.20
1.30
1.40
1.50
8380
8150
7910
7300
6750
6310
5910
5540
5330
5090
4840
4600
4350
4100
3850
(3500)
8370
8130
7890
7270
6710
6270
5860
5480
5260
5020
4780
4580
4300
4060
3900
3780
8000
7700
7440
6800
6370
6020
5740
5460
5220
4980
4770
4570
4410
4260
4140
4020
66
8 Effective temperatures of stars
from theoretical extrapolations from the observed part of the energy

distribution. The most primitive method of extrapolation would be to
determine an approximate Teff and assume that the star would indeed
radiate like a black body with T = Teff. The Planck function then tells us
how much radiation there should be in the short wavelengths, for which we
cannot observe the flux. We have to add this amount to the observed energy
and determine a better value for Teff. This does not give very good
corrections, because the hot stars do not radiate like black bodies in the short
wavelength region. In Volume 2 we will learn how to do it better.
In Table 7.2 we have given the Teff for those stars whose angular
diameters have been determined by the Hanbury Brown interferometer and
in Table 7.3 we collected the data measured with lunar occultations. In
Table 8.1 we give the values of Teff, which we consider to be the best ones, as
a function of the B V colors for main sequence stars (V), for giants (III) and
for supergiants (I).
These values refer to stars in our neighborhood, which we call population I
stars. We will see later (Chapter 12) that there are other stars which do not fit
into this population. For those we find another relation between colors and
effective temperatures.
Masses and radii of stars
9.1
General discussion of binaries
For the stars for which angular diameters have been measured we
can, of course, determine the radii by multiplying the angular radius by the
distance to the star, provided we know the distance. For stars further away
than about 20 pc we cannot measure trigonometric parallaxes accurately, we
can therefore determine distances only in indirect ways. Star stream
parallaxes (see for instance Becker, 1950) have been used to determine the
distance to the Hyades cluster, but it turned out that the photometric
parallaxes were more accurate. Fortunately, we can also determine radii for
stars in special binary systems, namely in eclipsing binaries. For these
binaries we can also determine the masses of the stars. We shall therefore
devote this section to the discussion of binaries in general, and the next
sections to the discussion of the special types of binaries for which stellar
radii can be determined, namely eclipsing binaries, and for those binaries for
which stellar masses can also be determined, again the eclipsing binaries and
also the visual binaries.
We can determine stellar masses for binaries using Kepler's third law. Let
us briefly discuss the mechanics of a binary system.
From the solar system we are accustomed to the case that one body, the
sun, has a much larger mass than the other bodies, the planets. For binaries
with two stars we have to take into account that both bodies have elliptical
(or perhaps circular) orbits around their center of gravity S (see Fig. 9.1).
From the definition of the center of gravity we know that
Mlrl = M2r2
(9.1)
where M x and M 2 are the masses and rx and r2 are the distances of the stars
from the center of gravity. In a binary system there must be equilibrium
between gravitational and centrifugal forces which means
GMiM2
- ^
,,
= M1w2lrl
n4
= M2co22r2
<c\^\
(9.2)
67
68
9 Masses and radii of stars
and where co1 and co2 are the angular velocities, rx and r2 are the orbital radii
of the two stars, and G is the gravitational constant. For the two stars to
remain in phase, which means for the center of gravity to remain at constant
velocity, we must have co1 = co2 = co.
For (9.2) to be true, we must then also have M1r1= M2r2 which means
that both stars must orbit the center of gravity.
For circular orbits co can be replaced by
(o = 2n/P
(9.3)
where P is the orbital period. We then derive

GMlM2
division by M1r14n2
..
4TC2
yields
Jkwh'-FUsing (9.1) we find Ml = (M2'r2)/r1

(M1+M2)
(95)
and
= M2 M - + 1 I or M 1 + M 2 = i - ^ ( r 2 + r 1 ).
(9.6)
This yields
M2
M t + M2
rt
r2 + r t
(9.7)
Inserting (9.7) into (9.5) yields the general form of Kepler's third law
(9.8)
If distances are measured in astronomical units and times in years and
masses in units of solar masses, the factor 4n2/G becomes unity.
Star 2 (f)^
L^r*
Si
*/]) Star 1
Fig. 9.1. In a binary system the two stars, Star 1 and Star 2 orbit their center
of gravity, S.
9.1 General discussion of binaries
69
While we have derived (9.8) only for circular orbits (otherwise rur2 and co
would be functions of time) it can be shown that this law also holds for
elliptical orbits, if r t and r2 are replaced by the semi-major axes a1 and a2
of the orbits around the center of gravity.
In order to determine the masses, we then have to determine a1 and a 2 , or
at least the sum. If we look at visual binaries, i.e., binaries for which we can
observe both stars, we will generally only be able to observe the orbit of one
star relative to the other. For circular orbits of the two stars, the relative
orbits of each star around the other will also be a circle with radius rt + r 2 .
Making a plot of the relative positions of one star around the other for
elliptical orbits one can see that each star will also go around the other in an
ellipse and the second star will be in the focal point of the true elliptical
orbit.* The semi-major axis of the orbit of one star around the other is
ax + a2. This can be seen from Fig. 9.2 as follows.
The major axis of the relative orbit is
b1 + b2 + d2 + d1 = bi+d1 + b2 + d2 = (ax + a2)2 = la
(9.9)
which is the sum of the major axes of the two orbits around the center of
gravity. We have written ax and a2 for the semi-major axes of the stellar
orbits around the center of gravity, and a for the semi-major axis of the
relative orbit of one star around the other.
What we actually observe for a visual binary system is unfortunately not
Fig. 9.2. The major axis of the relative orbit of M x around M 2 is the sum of
the major axes of the orbit of each star around the center of gravity, S. t stands
for time and P for the period, t0 is the time of maximum distance apart of the
two stars.
* What we see is, of course, generally not the true orbit but is the projection of the true orbit against
the background sky. We will outline below how the true orbit can be determined from the projected
orbit and the measured velocities.
70
the major axis itself but the angular major axis, namely 2a/distance. In order
to convert this into the actual major axis we have to determine the distance
or find another way of conversion. As we know good parallaxes can only be
determined for nearby stars. Fortunately the Doppler effect is independent
of distance and provides a geometrical scale.
9.2
The Doppler effect
The Doppler effect is very important for many branches of

astronomical research because it permits us to measure the component of the
velocity along the line of sight, the so-called radial velocity (see Fig. 9.3). For
the stars, we cannot of course measure their velocity by measuring the
distance which they travel and then dividing by the time it takes them to do
so, as we do for objects here on Earth. Remember that we can only observe
the light from the star, from which we have to derive everything. Fortunately
the frequency of the light that we see changes when the light source moves
towards us or away from us. This shift in frequency or wavelength is called
the Doppler effect. Fortunately the stellar spectra have very narrow
wavelength bands in which they have very little light intensity, the so-called
absorption lines, which mark distinct wavelengths whose Doppler shift can
be measured. If the light source moves towards the observer the wavelength
becomes shorter by A/I where
=\
Ao
(9-10)
here Xo is the so-called rest wavelength of the line (i.e., the wavelength which
the line has, if the light source and the observer have the same velocity). vT is
the component of the velocity in the direction of the observer, the line of sight
or the radial velocity (see Fig. 9.3).
If the light source moves away from us, the observers, the wavelength
becomes longer by the same relative amount. It is this wavelength shift which
permits us to measure the radial velocities of the stars in cm/s.
-Observer
Fig. 9.3. The component of the velocity in the direction of the observer, the
radial velocity, vr, leads to a shift in wavelength of the light, given by (9.10).
9.3 Radial velocities and masses of binaries
9.3
71
Radial velocities and masses of binaries
In order to measure the radial velocities of the stars we have to

observe the energy distributions of the stars in very narrow wavelength
bands, i.e., we have to take so-called high ressolution spectra, such that we
can see and measure the exact wavelengths of the spectral lines. If our line of
sight happens to be in the orbital plane of the binary system, consisting of
stars A and B orbiting in circular orbits around their center of gravity S (see
Fig. 9.4), then at the time of largest angular distance of the two stars, star B
will move towards us while star A moves away from us. If stars A and B have
the same mass, as assumed in Fig. 9.4, then the velocities will be of the same
magnitude but in opposite directions. At this time, the stars have the largest
radial velocity component. While the speed of the two stars in their circular
orbits remains the same, the component in the line of sight decreases when
they move on and is zero when they reach points P and P 2 , where their
motions are perpendicular to the line of sight. At this time, the spectral lines,
also called Fraunhofer lines, appear at the wavelength Xo corresponding to
the velocity of the center of gravity. The radial velocity components then
change directions for both stars and the Doppler shift changes sign. The lines
of both stars are now shifted in the opposite direction. Altogether the lines of
each star show a periodic shift whose amount changes sinusoidally in time if
the orbits are circular.
In Fig. 9.5 we show the spectra of a binary system, in which the stars are
too close to be observed separately. We can only obtain spectra of both stars
together, which show the spectral lines of both stars. We see a time series of
spectra, proceeding through one orbital period from the top spectra to the
Observer
Fig. 9.4. At points A and B the spectra show Doppler shifts of the lines
corresponding to the orbital velocities. At the positions P1 and P2 the
radial velocities are zero.
1945
UT.
28
506
29
4:05
30
411
July
500
4:08
5:28
530
10
4:35
12
4MI
15
3:03
Fig. 9.5. From top to bottom a time sequence of the spectra of the binary star system Mizar is shown. Both stars
are too close together to separate their spectra. The line systems of both stars are seen on each spectrum. The
line positions of both stars shift sinusoidally due to the Doppler shift which results from the orbital motions. The
wavelengths of the spectral lines of stars 1 and 2 vary in antiphase. (From Binnendijk 1960.)
9.3 Radial velocities and masses of binaries
73
bottom spectra. The approximately sinusoidal change in separation between

the lines of the two stars can be seen.
Radial velocity curves as measured for the two stars in the binary system
Phoenicis are reproduced in Fig. 9.6B.
Stars which cannot be observed separately but whose binary nature can be
seen from the periodic Doppler shifts of their spectral lines are called
spectroscopic binaries.
For binaries, for which both spectra can be obtained, and for circular orbits
with period P we must have
and v? =
2nr7
(9.11)
200 -
-200 -
Phase
Fig. 9.6. Light and velocity curves for the eclipsing binary f Phoenicis are
shown as a function of orbital phase. During the times when one star is behind
the other we see a reduced light intensity. This happens twice during one
period, once if star 1 is in front of star 2, and once if star 2 is in front of star 1.
If star 2 has the larger surface brightness, i.e., if star 2 emits more light per cm 2 ,
then less light will be received when star 2 is eclipsed during the first eclipse
than will be received when star 1 is eclipsed, i.e., during the second eclipse. The
first light minimum is then deeper than the second one. (From Paczinski 1985.)
74
and
(9.12)
The periods must of course be the same for both stars. vx and v2 can be
measured at maximum angular distance of the two stars. With the measured
period, rx and r2 and the sum r1 + r2 can be determined in cm, which then
also gives the sum of the masses according to (9.8). From (9.12) the mass ratio
can be obtained. With the mass sum and the mass ratio known it is a simple
mathematical problem to determine both masses.
Unfortunately, most stars do not have circular orbits and our line of sight
usually is not in the orbital plane.
If the stars are not in circular orbits, their radial velocity curves are in
general not sinusoidal, the shape depends on the degree of eccentricity and
on the orientation of the major axis of the orbit with respect to the line of
sight. It also depends on the tilt of the orbital plane with respect to the line of
sight, which is described by the angle i between the normal n on the orbital
plane and the line of sight.
If the inclination angle i between the normal n on the orbital plane and the
line of sight is not 90, which means the line of sight is not in the orbital plane,
then all measured radial velocities only measure the component of the
velocity projected onto the line of sight. This means all velocities are reduced
by the factor sin i. Even for circular orbits the radial velocity measured at the
largest distance is then only Fsin L (see Fig. 9.7). Then we cannot determine
the masses without knowing sin i. For i = 90 the stars appear to move in a
straight line. Only for sin i 0 are we able to see the true shape of the orbit,
but then we cannot measure the velocities. For angles i between 0 and 90
orbits appear as ellipses even if they should be circular. In order to be able to
derive semi-major axes and masses we therefore have to determine the
,v
Observer
\/sin /
Fig. 9.7. If the normal n to the orbital plane is inclined with respect to the line
of sight by the angle i, then the radial velocity vr is V sin i if V is the true
orbital velocity.
9 A Determination of orbital parameters for binaries
75
orbital parameters like ellipticities, orientation of the ellipse and the

inclination angle i of the orbital plane.
9.4
Determination of orbital parameters for binaries
9.4.1
General discussion
Before we discuss in detail the procedure for determining the orbital

parameters for binaries let us see how the radial velocity curves change for
different orbits. The orbit is determined if we know the ellipticity, the
orientation of the ellipse and the inclination angle i between the line of sight
and the normal n on the orbital plane.
The ellipticity e is defined by
e2 = \ - ^
(9.13)
where a and b are the semi-major and semi-minor axes respectively. The
orientation of the ellipse is described by the angle co, which is the angle
between the minor axis of the ellipse and the projection of the line of sight
onto the orbital plane.
In Fig. 9.8 we show radial velocity curves to be observed for a circular
orbit, e = 0.0(a), and for e = 0.5 for different orientations of the ellipse. For
different angles i all the velocities would be multiplied by the factor sin i but
the shape would remain the same. We cannot therefore determine the angle i
from the "shape of the velocity curve. The ellipticity can however be
determined from the shape of the radial velocity curve. If e i=- 0 the curve
is not sinusoidal because the stars have larger velocities when they are closer
together. In Fig. 9.8 (b) we see the high peaks of the radial velocity when the
star moves through point b, the perihelion of its orbit. When it moves
towards us at point d it is far away from the other star and therefore has a
much lower velocity. We measure a broad, flat valley for the negative
velocities. A very asymmetric velocity curve is obtained if co = 45 as seen in
Fig. 9.8 (c), but we still see larger positive velocities than negative ones. If
co = 90 (Fig. 9.8 (d)) the position of the ellipse is symmetric with respect to the
line of sight. The largest positive and negative velocities have therefore equal
absolute values, but it takes the star a much longer time to get from the
largest negative velocity to the largest positive velocity, because the distance
is larger and the speed is smaller when going from d to b than when going
from b to d. From Fig. 9.8 we can easily see that we can determine e and co
from the deviations of the radial velocity curve from a sinusoidal curve.
76
For the details of the mathematical procedure see for instance Binnendijk
1960.
9.4.2
Determination of i for visual binaries
From the study of the radial velocities we have no means of

determining the inclination angle i between the line of sight and the normal n
to the orbital plane. Since the shape of the radial velocity curve is not
changed by a change in i we have no way to determine i if we have only radial
velocity curves. If we are however dealing with a visual binary for which we
can measure the shape of the orbits we can determine i. We know that for
binaries the second star must be in the focal point of the orbit for thefirststar.
Suppose the stars have circular orbits which appear as ellipses because i is
not 0. For a circular orbit the secondary star will therefore appear in the
(0
(d)
Fig. 9.8. Plotted are radial velocity curves to be seen for stellar orbits with zero
ellipticities (a) and with an ellipticity e = 0.5 for co = 0 (b\ and for the same
ellipticity but co = 45 (c), and also for co = 90 (d). e / 0 generally introduces a
difference between maximum positive and negative velocities (see (c) and (d))
except in the case when co 90. In this case the time difference between points
b and d on the one side and between points d and b on the other side (d) gives
a measure of the ellipticity. (From Becker 1950.)
9.4 Determination of orbital parameters for binaries
11
middle of the major axis and not in the focal point (see Fig. 9.9). We know
immediately that the true orbit must be a circle. From the apparent ratio of
the minor to the major axis we can in this special case, determine how large i
is. If the second star appears off center but not in the focal point we also know
that i is not 90, but that the true orbit is an ellipse. We can figure out how
much we have to tilt the observed projected ellipse in order to fit the second
star onto the focal point of the orbital ellipse.
For the procedure and details of the orbit determinations see for instance
Binnendijk 1960.
9.4.3
Distance to visual binaries
In Section 9.3 we saw that we can determine the orbital radii in cm if

we can determine the true orbital velocity by measuring the radial velocities,
determining the orbital parameters, and determining sin i. If we know the
true orbital velocities then the orbital radii and the velocities are related as
given by (9.11) for the case of a circular orbit. From the two Equations (9.11)
we can determine r1 and r2 and thereby ri + r2. Observing the orbit of one
star around the other permits us to measure the angular semi-major axis
( r i + ri)/d = a, where d is the distance to the binary system. Once we
have determined r1 and r2 in cm we can also find d in cm (or in pc) from
For visual binaries we can therefore determine the distance even if we

cannot measure trigonometric parallaxes.*
Line of sight
Fig. 9.9. For a circular orbit of star A around star B whose normal n is
inclined at an angle i with respect to the line of sight, the apparent orbit
projected against the background sky is an ellipse with the second star B in the
middle of the major axis, instead of in the focal point F.
* In practice there are only a few systems which are close enough for the two stars to be seen separately
while the periods are not too long to be determined in a (wo)man's lifetime and for which the radial
velocities are large enough to be accurately measurable.
78
9.4.4
Eclipsing binaries
a. Mass determination. Another group of binaries for which i can be

determined are the eclipsing binaries. For these stars the line of sight is
almost in the orbital plane such that at some time the secondary star is in
front of the primary (see Fig. 9.10). Half a period later the primary star will be
in front of the secondary. If along the line of sight one star gets in front of the
other, we can be fairly certain that sin i has to be close to 1 otherwise the
'front' star would remain above or below the 'back' star, except if the stars are
very close (see Fig. 9.11). By far the most binaries are separated far enough
that eclipses will only occur if i > 75 which means sin i ^ 0.96 and
sin3 i ^ 0.89. This means that if any sign of an eclipse is seen sin3 i cannot
differ by more than 10% from 1. Assuming sin 3 i = 1 will at most cause a 10%
Observer
Fig. 9.10. If along the line of sight one star is in front of the other star system
the total amount of light of the system is reduced. We see the eclipse of the
star.
Observer
Orbital plane
Fig. 9.11. If the line of sight is inclined too much to the orbital plane then star
M2 is not in front of star Mt along the line of sight. The angle i has to be very
close to 90 for an eclipse.
79
error in the masses (see equation 9.26). With refined technique we can even
determine sin i for partial eclipses. We can usually assume sin i = 1 for
eclipsing binaries.
Such eclipses can be recognized by the decreasing amount of light received
during the phases of occultation. In Fig. 9.6 we showed the light curve for the
eclipsing binary f Phoenicis. In Fig. 9.12 we reproduce schematic light
curves for binaries with different ratios of the size of their radii and with
different surface brightnesses. In Fig. 9.13 we relate the geometrical situation
to the observed light curve as a function of time. From Fig. 9.14 we can see,
that for a circular orbit of star A around star B the ratio of the total length of
the eclipse time te to the length of the period P is given by
(9.15)
2nr
where RA and RB are the radii of stars A and B respectively, and r is the
orbital radius of star A around star B. Equation (9.15) assumes that star B is
much more massive than star A such that the velocity of star B can be
neglected. It also assumes r RA and RB.
If tJP 1 the stellar radii must be much smaller than the orbital radius
and the stars cannot be very close. Eclipses must then mean that i 90 and
<9O
90*
Fig. 9.12. Schematic light curves are shown for different ratios of stellar radii
and different surface brightnesses of the eclipsing stars. Curves (a) and (b) show
the light curves for eclipsing binaries for which both stars have the same radius
and the same surface brightnesses. Curves (c) and (d) apply to binaries with
different radii and surface brightnesses. From Becker (1950).
80
sin i 1. We can get even more accurate information about i if we study the
shape of the light curve in detail. The decrease of the light intensity will be
less steep if the secondary star does not pass over the equator of the other star
(see Fig. 9.13(b)).
For such eclipsing binaries we then also know the inclination angle i. We
can therefore determine the masses of both components as described in
Sections 9.4.1 and 9.4.4.
b. Radius determination. Eclipsing binaries are also very useful for
radius determinations, especially if we see a total eclipse.
Fig. 9.12 shows light curves which are obtained for stars with different
radii and different surface brightnesses. Since these light curves look very
different we can use the light curves to determine the properties of the stars.
For simplicity we again assume circular orbits. Generally orbits are elliptic
(a)
(b)
- time
Fig. 9.13. For a total eclipse (a) the light curve shows a flat bottom (except if
the radii are equal), while for a partial eclipse (b) the light curve does not have
a flat bottom.
81
but the method does not change, only the mathematics gets more
complicated.
If radial velocities for the stars are measured we can determine the radii of
both stars by means of (9.15). The maximum radial velocity of star A is
K
r(max) ~" ^'u A/1
\y'LKJ)
We also know that for a total eclipse the duration for the flat part of the light
curve tt, the time during which star A is totally in front of star B, is related to
the length of the period P by
(a)
(b)
Observer
(0
Fig. 9.14. Figure (a) shows the positions of the eclipsing star at first, second,
third, and fourth contact as seen from an observer above the plane of the
paper. Figure {b) shows the geometry of the eclipse for an observer whose line
of sight is in the orbital plane (which is the plane of the paper), not drawn to
scale. For a large distance between stars A and B the path of star B can be
approximated by a straight line. Figure (c) shows the light curve as a function
of time for the duration of the eclipse. During the time te star A travels a
distance de = 2RA + 2RB. During the time tt star A travels the distance
dt=2RB-2RA.
82
tJP = (2RB-2RA)/2nrA.
(9.17)
if again the mass of star B is much larger than the mass of star A.
Equation (9.16) determines the orbital radius rA. From (9.15) and (9.17) we
obtain
AR
{t
*-tt)/P
2^A-
(9
*18)
and
AR
<'
+ t)/
>
<919>
These equations determine the radii of both stars.

If stars A and B have comparable masses we have to consider the motions
of both stars. The eclipse times will then be shorter, because the relative
velocities of the two stars will be larger, namely VA + VB, with VA = 2nrA/P
and VB = 2nrB/P.
If the stars have elliptical orbits we have to know in which part of the
ellipse the occultation occurs, such that we can relate the maximum
measured radial velocity to the relative velocity at the time of the eclipse.
The mathematics becomes more involved but the principle of determining
the radii of both stars remains the same.
c. Surface brightness of eclipsing binaries. From the light curves of
eclipsing binaries we can also determine the ratio of the surface brightnesses
of the stars.
From Fig. 9.14 we see that maximum light is given by
/max = (nRl-FB + nR2A FA)const
(9.20)
where FA and FB are the amounts of radiation emitted per cm 2 of the stellar
surface of star A and star B respectively.
The constant is determined by the distance and the sensitivity of the
receiving instrument as well as the transmission of the Earth's atmosphere.
During the first eclipse the minimum intensity Jx is given by (see Fig. 9.15)
h = L(KRB - KR2A)'FB + nR2A-FA]consl.
(9.21)
During the second eclipse the minimum intensity I2 is given by

I2 = {nR2B FB)const.
(9.22)
/max-/2 = (^^)const.
(9.23)
/ - / ! =(^-F B )const.
(9.24)
From this we derive

and
83
Knowing RB and RA from our previous discussion we can then determine FA and FB from (9.23) and (9.24), except that we have to know the
constant, which means we have to know the distance, the transmission of
the Earth's atmosphere, the sensitivity of the instrument, etc. So in general
we can only determine ratios namely
*2
(9.25)
If we know the flux for one star, we can determine the flux of the other star.
Since the radiative fluxes FA and FB determine the effective temperatures of
the stars we can also determine the ratio of the effective temperatures of the
two components of an eclipsing binary system.
9.4.5
Spectroscopic binaries
Most binaries are spectroscopic binaries, which means we can see

their binary nature from the periodic line shifts due to the Doppler effect.
Very often we see only lines from one star because the companion is too faint
to be recognized in the combined spectrum. If we see both spectra we can
determine the mass ratio of the two stars according to (9.2). It does not cause
any problems that we can only measure v1 sin i and v2 sin i, without knowing
what sin i is. In the ratio the sin i factor cancels out. But if we want to
determine the mass sum we are in trouble. We know that rt = v1P/(2n) and
similarly r2 = v2P/{2n). Since we only know v1 sin i and v2 sin i we also can
2nd eclipse
1st eclipse
Star B
starA
~rf)]
in front \ \ ^
Star A
in back
Fig. 9.15. During the first eclipse a fraction of the surface of star B is occulted.
During the second eclipse star A is completely covered up. The depths of the
light minima depend on the surface fluxes of star A and star B.
84

Table 9.1 Masses and radii for some visual binaries according to D. Popper (1980)
Star
aCMaA
aCMi
Her A
HerB
a Cen
y Vir
r\ Cas A
i; C a s B
{ Boo
M/MQ
*P
A1V
F5IV-V
GO IV
KOV
G2V
FOV
GOV
MOV
G8V
n"
Wo
1.68
2.06
2.24
0.79
1.27
1.35
0.98
0.59
0.77
2.20
1.77
1.25
0.70
1.14
1.08
0.91
0.56
0.90
0.377
0.287
0.104
0.104
0.743
0.094
0.172
0.172
0.148
only determine ^sini and r2sinf. According to (9.8) we therefore can

determine only
IXM . AJI \ 3
3
(Mi + M2)sin i = ^
(r1-\-r2)3sin3i4n2
-.
(9.26)
The mass sum remains uncertain by the factor sin3 i, and therefore both
masses remain uncertain by the same factor.
9.5
Data for stellar masses, radii, and effective temperatures
In Table 9.1 we list stellar masses as determined from the study of

visual binaries according to D. Popper (1980). In the same table we also list
stellar radii determined for the same binary systems.
In Table 9.2 we have given average values of masses, radii and effective
temperatures for main sequence stars of different B V colors, corrected for
interstellar extinction (see Chapter 19).
In Fig. 9.16 we have plotted a color magnitude diagram with the position
of the average main sequence drawn in. On this average main sequence we
have also given values for the mass, radius, and Teff for stars at that
particular position on the main sequence. Also given are spectral types,
which will be discussed in the next chapter.
At the top of the main sequence we find the most massive stars, which are
also the largest ones and the hottest ones. It becomes obvious that the main
sequence is indeed a one parameter sequence: If we know the B V color for
a star we also know its absolute magnitude, its mass and radius and its
9.6 The mass-luminosity relation
85
Table 9.2 Masses and radii for main sequence stars according to Schmidt-Kaler
(1982) and Popper (1980)
Spectral1
Type
B-V
R/Ro
M/Mo
O8
BO
B2
B4
B6
B8
AO
A5
FO
F5
GO
G5
KO
K5
MO
M5
-0.32
-0.30
-0.24
-0.18
-0.15
-0.11
-0.01
+ 0.15
+ 0.30
+ 0.44
+ 0.58
+ 0.68
+ 0.81
+ 1.15
+ 1.40
+ 1.62
10
7.5
5.4
4.2
3.4
2.9
2.4
1.6
1.4
1.3
1.1
0.95
0.85
0.72
0.60
0.27
23
16
9
6
4.5
3.8
2.8
2.0
1.6
1.3
1.1
0.93
0.82
0.65
0.65
0.20
1. Spectral types will be discussed in Chapter 10.
effective temperature. In Volume 3 we will try to explain why most of the

stars form a one-dimensional sequence.
9.6
The mass-luminosity relation
As we saw in the previous section there is a relation between mass

and the absolute visual magnitude for the stars on the main sequence, the
more massive the star is, the brighter it is in the visual. We also saw in
Section 6.3 that for stars hotter than 8000 K the bolometric corrections
increase and therefore the bolometric luminosities also increase for increasing mass of the stars. For stars cooler than 7000 K the bolometric corrections
increase for stars with smaller visual brightness, but we still find decreasing
bolometric brightnesses for decreasing masses of the stars. On the main
sequence we therefore find a simple relation between mass and luminosity:
the larger the mass the larger the luminosity. This may not be surprising
since we have just emphasized that the main sequence is indeed a onedimensional sequence and therefore all the stellar parameters are related in a
unique way. What is surprising, however, is the fact that the mass-
86
luminosity relation also holds approximately for the giants and supergiants.
Wefindseveral of them in binaries for which we can determine the masses. In
Fig. 9.17 we show the mass-luminosity relation as seen for all welldetermined masses in binaries. The exceptions are the red giants which for a
given luminosity have generally smaller masses. In Volume 3 we will explain
this result which at first sight appears very surprising.
supergiants Ja
- 5
\ \ B 0 ; 30500K-, 7R-, 16 M
\
^ vAO; 9500K-, 2.4R; 2.8M J
- "_
4vF0; 7500K; I.5R-, I.6M

+v0i 6300 K; MR; I.05M
Sun; 5 8 0 0 ( 0 ^ x 0 ; 5350 K;0.85RT
MV
0.8 Mo
M0;3850Kj 0.6R o 0 . 5 7 1 ^
10
M5; 3000 K; 0.27 Roi 0.22 M \

15
0.5
1.0
1.5
B-V
Fig. 9.16. The solid line outlines the main sequence in the color magnitude
diagram. At the different points of the main sequence the masses, radii, and
effective temperatures of the stars are given.
The dashed line shows the position of the giant sequence, and the dotted line
the approximate positions of the brightest supergiants, called luminosity
class la.
9.6 The mass-luminosity relation
87
-1
log M/MQ
Fig. 9.17. The relation between stellar masses and their luminosities is shown
for binaries with well determined masses, according to Popper (1980). The
different symbols refer to different kinds of binaries. The open circles refer to O
type binaries, the filled circles to O, B pairs, the x to visual binaries, and the
triangles to giants. The open squares refer to resolved spectroscopic binaries.
10
Spectral classification
10.1
The spectral sequence
So far we have talked only about global properties of the stars and
about their brightnesses in broad wavelength bands. We get, of course, more
information when we reduce the widths of the wavelength bands in which we
study the radiative energy emitted by the stars. If we reduce the bandwidth to
the order of a few A or even a fraction of an A, and cover all wavelengths,
we talk of stellar spectra. If we compare spectra of different stars, we see that
there are many different kinds of spectra. Most of them can be ordered in a
continuous sequence of spectra, the so-called spectral sequence. In Fig. 10.1
we show the sequence as it is used now, and as it was established finally
by Morgan, Keenan and Kellman (1943).
If we plot the energy distribution in a spectrum as a function of
wavelength, we get plots as shown in Figs 10.2. Basically, we see a
continuous energy distribution (Fig. 10.2(a)), but there are many wavelengths
for which the energy is reduced by varying amounts, the so-called spectral
lines, see Fig. 10.2(fc). If we look at these wavelengths in a spectrum, they look
dark because there is little energy at these wavelengths. Such dark lines are
called absorption lines. There are also some spectra which have bright lines,
i.e., for which there is more energy at these wavelengths. Such lines are called
emission lines. Most stellar spectra show absorption lines. This is quite
different from the situation which is known from the chemistry laboratory,
where emission line spectra are used to identify chemical elements. Kirchhoff
and Bunsen discovered in 1859 that an absorption line spectrum is seen if in
the laboratory we place a cold gas in front of a hot light source (see Fig. 10.3).
The fact that we see absorption lines in the stellar spectra thus tells us that
the gas at the stellar surface is cooler than the gas in the somewhat deeper
layers (see Fig. 10.4). In Volume 2 we will understand better when emission
lines and when absorption lines are formed.
The spectral classification is done according to the lines and the line
89
90
10 Spectral classification
?G Lac
OS
r Sco
SO
y Ori
87
OH
82
83
85
0 Pep
B8
or A y
8$
1
I
HR
mo
AD I
cttyr
AD V
1 III
36
Cas
S 222
ffr GfifA
S S389
Fig. 10.1. We show the sequence of stellar spectra. At the top of the figure the
O type spectrum of the star 10 Lac is reproduced. At the bottom we see an N
type spectrum as observed for the star S 3389. All the stars for which the
spectra shown have been observed are indicated on the left side. For a
description of the spectral types see the text. Shown are negatives. The white
lines are wavelengths in which no or very little energy is received, i.e. they are
absorption lines.
91
10.1 The spectral sequence
strengths seen in the spectra. These lines are also called the Fraunhofer lines
after the optician whofirstdiscovered them in the solar spectrum. Originally
when astronomers did not then understand the origin and the meaning of
these spectral lines, they realized that there were a number of strong lines
which were strong in almost all spectra, so they put the spectra in a sequence
according to the strengths of these lines. We now know that these lines are
due to absorption by the hydrogen atoms in the stars. The early astronomers
called the spectra with the strongest hydrogen lines, spectral type A, those
with somewhat weaker hydrogen lines were called spectral type B, and went
through the alphabet for weaker and weaker hydrogen lines, down to
spectral types M, N, O, R, S. It turned out, however, that within this sequence
the line strengths of the other lines varied quite irregularly, also the B - V
colors varied irregularly along the sequence. It was much better to put the
spectra into a sequence which also took into account the changes of the other
Baschek+Oke 1965
x This study
4.5
29Psc
5
0.009*6
0.025
0.015
0.010=5
3.6
3.7
a8
Fig. 10.2(a). Continuous energy distributions for A stars are shown as a _

function of wavelength for the wavelength interval 3000 A < X < 9000 A. S are
the standard deviations for the measurements. The energy distributions of A
stars are strongly wavelength dependent, especially around 3800 A where the
Balmer discontinuity is seen. (From Bohm-Vitense & Johnson 1977.)
92
lines. After the reordering had been done, it appeared that the new spectral
sequence was also a sequence according to the colors of the stars. As usual, in
the meantime the astronomers had become accustomed to the old spectral
types and did not want to change the spectral type designation for their stars.
HS Sri
He+CaECaE
HTJ
H6
Hi
HH
r~
HX HAL
L_J_
9000
I
3700
Fig. 10.2(b). The energy distributions in two spectra of the variable star 3
Cephei are shown as a function of wavelength. The upper spectrum refers to
the phase of maximum light and maximum temperature when the star has a
spectral type F5Ib. The hydrogen Balmer line series is clearly visible. The lower
spectrum refers to the phase of minimum light and minimum temperature,
when the star has a spectral type G2Ib. At this time the star is so cool that the
hydrogen lines are barely visible for this low dispersion. The Ca+ lines are very
strong and many other lines of metallic ions and atoms are seen. (From Ledoux
& Walraven 1958.)
Spectrum
Na
Cold
X = 5889
5896 A
Fig. 10.3. Cold Na vapor in front of a light source emitting a continuous

spectrum absorbs light at the wavelengths of 5889 and 5896 A, giving dark
absorption lines in the spectrum. A hot light source with Na vapor in it would
show emission lines, as seen in the Bunsen burner.
93
So now we still have most of the original spectral types, but they do not
appear in alphabetical order in the new sequence. The A stars with the
strongest hydrogen lines are now in the middle of the sequence. The sequence
now goes O, B, A, F, G, K, M. At the cool end, the classification becomes
more complicated. In addition to the M stars, we also have N stars and S
stars, depending on which molecular bands are seen.
The O, B, and A types are called the early types, just because A and B are
'early' in the alphabet. The term 'early' has nothing to do with the age of the
stars. Correspondingly, the G, K, M, N, and S stars are called late type stars
because these letters are later in the alphabet.
The sequence now starts with spectral type O, which indicates that these
stars have rather weak hydrogen lines, but they are the bluest stars. In the
spectra of these stars we can recognize lines which we know from laboratory
experiments to be due to the helim atom and the helium ion, i.e., a helium
atom which has lost one electron. The O stars are the only stars in whose
spectra the lines of ionized helium can be seen. It takes a lot of energy, 24
electron volts (ev), to separate the electron from the helium atom. Only in
very hot stars do the particles have enough energy to ionize the helium atom.
The next spectral type in the new sequence is type B. As the letter indicates,
these spectra show rather strong hydrogen lines but they also show lines of
the helium atom, thereby indicating that they should be placed close to the O
stars. Most other lines in the spectra of the B stars are very weak.
The next type in the sequence are the A stars with the strongest hydrogen
lines. They do not show helium lines, or at least only extremely weak ones
which can only be seen if very high spectral resolution is used, which means if
the energy distribution is measured in very narrow wavelength bands. The A
Observer
Radiation
Stellar surface
Cold gas
MMM!//,
Stellar interior
Fig. 10.4. The stellar radiation comes from a hot interior and passes through
the cooler surface layers. The spectrum which we observe therefore shows
absorption lines.
94
stars also begin to show a number of other spectral lines which, as we know
from laboratory experiments, are due to heavier elements like iron (Fe),
chromium (Cr), and many other elements. The lines seen are due mainly to
the ions of these elements, which means again that these atoms have lost one
of their electrons. It takes only about 8 eV to ionize these elements.
For the F stars, which follow the A stars, these lines from the ions of the
heavier elements become stronger, and we also start to see lines which are
due to the neutral particles, the atoms.
Astronomers are usually somewhat sloppy and call all the elements whose
atoms are heavier than helium the 'metals.' It is clear that, for instance,
carbon or oxygen are not metals, but in the astronomer's jargon these
elements are often included when the word 'metal' is used. We can then say
that for F stars the metallic lines become stronger. This is even more true for
the G stars in whose spectra some metallic lines and the hydrogen lines are of
comparable strengths. The solar spectrum is of type G. For stars still later in
the sequence, the hydrogen lines become still weaker and the metallic lines
stronger.
In the K stars we see mainly lines of neutral atoms as the lines of the
metallic ions decrease in strength. For the M stars, another type of spectrum
becomes visible. We see groups of many lines very close together. These types
of spectra are formed by molecules. For stars with M type spectra, also
called the M stars, we know that molecules must be present. Indeed, we know
that some of these molecular bands, as the line groups are called, are due to
molecules CN, CH, and TiO, i.e. cyanogen, carbon hydride and titanium
oxide.
Other stars with molecular bands are the R, N, and S stars. They differ
from the M star spectra by the kind and strengths of the molecular bands.
The S stars do not show TiO bands but instead show ZrO bands. The N and
R stars are now usually called C stars, where the C stands for carbon. They
show strong bands of various molecules containing a carbon atom, while no
molecules with oxygen are seen. These are apparently stars which have a
larger than average carbon abundance, carbon is more abundant than
oxygen. Under those circumstances the oxygen is all bound in the CO
molecule, whose bands are in the ultraviolet where they are not observable
for cool stars, because there is so little flux at these short wavelengths.
The C stars are again subdivided into classes Cl, C2, C3, etc with
increasing numbers for increasing strength of the carbon features in their
spectra.
Dividing all the different spectra into just these seven or ten main types is,
of course, only a very coarse division which turned out to be much too
95
coarse for modern astronomical research. Each spectral type is, therefore, now
subdivided into ten subclasses, in principle, though some of the subclasses
are not really in use. The subclasses have numbers. We now have AO, Al,
A2,... A9. A9 is then followed by FO, etc. For some spectral classes, even this
subdivision did not turn out to be fine enough, so we also have a spectral
class B0.5, which means the spectrum is between BO and Bl. We have to get
accustomed to the fact that astronomy is a very old science and that
everything is built on tradition even though for modern astrophysics many
procedures do not make much sense anymore. It would be very difficult to
give up old traditions because the old observations and catalogues still have
to be used and are very important for many studies.
For the new spectral sequence, the hydrogen lines now have their
maximum somewhere in the middle of the sequence and decrease on both
sides as shown schematically in Fig. 10.5.
10.2
Luminosity classification
We started out trying to find spectral criteria which would tell us

whether a given star is a main sequence star or whether it might be a giant,
also called luminosity class III star, or a supergiant, also called luminosity
class I star, or perhaps a white dwarf.
For stars in clusters and for very nearby stars, we know from the color
magnitude diagram which are main sequence stars and which are giants or
supergiants. We can therefore look at the spectra of these stars and see in
which way the spectra of the giants and supergiants differ from those of main
sequence stars. For blue stars it is obvious that the hydrogen lines become
more narrow for luminous stars (see Fig. 10.6). For stars of later spectral
Hydrogen
line
strength
Fig. 10.5. We show schematically the dependence of the hydrogen line strength
on the spectral type. When going from the O type stars to later spectral types
the hydrogen line strength first increases until we get to the spectral type AO,
for later spectral types it decreases.
96
types this is no longer the case, but other criteria can be found, as
demonstrated in Fig. 10.7 in which we show the luminosity criteria for the
spectral type F8. Generally the metallic lines increase in strength for
supergiants while the hydrogen line strengths do not change for higher
luminosity stars of spectral types G or later. The spectral types are
determined mainly by ratios of line strengths for metallic lines rather than by
the line strengths themselves, because these do change with luminosity.
Once we have learned how to recognize giants and supergiants by means of
their spectra, we know whether a given star is a main sequence star or not
and whether we can determine its distance by means of spectroscopic
parallaxes.
Giants and supergiants have a much larger spread in intrinsic brightnesses. Generally we cannot determine distances of supergiants by means of
spectroscopic parallaxes, only rough estimates are possible.
In order to obtain a somewhat better luminosity classification the
supergiants are now subdivided into classes la and Ib and sometimes also
lab. The la supergiants are the brightest ones, then follow the lab, and the Ib
are the faintest supergiants.
The giants are also subdivided into the bright giants, or luminosity class II
17C M
B5lo
67OpK
B5Ib
B5fll
B5V
I
II
Fig. 10.6. A series of B5 star spectra for stars with different luminosities is
shown. The hydrogen lines become narrower and apparently fainter for the
more luminous stars. (Luminosity class I means supergiants, II means bright
giants, III are giants, IV are subgiants, and luminosity class V are the main
sequence stars.) (From Morgan, Abt & Tapscott 1978.)
10.3 White dwarf spectra
97
stars, the 'normal' giants with luminosity class III, and the subgiants with
luminosity class IV. The luminosity classification is generally uncertain by at
least one subclass.
The population II main sequence stars, which lie below the standard main
sequence in the color magnitude diagram are often called luminosity class
VI objects.
10.3
White dwarf spectra
So far we have only discussed how we can recognize giants and

supergiants from their spectra. We have not yet discussed how white dwarfs
may be recognized. In Fig. 10.8 we reproduce white dwarf spectra of different
spectral types. It is obvious that the spectral lines are extremely broad; in
fact, some are so broad that they can hardly be recognized. For most white
dwarfs, only hydrogen lines and sometimes the lines of He can be seen. White
dwarfs with He lines are called DB stars, while those with hydrogen lines are
called DA white dwarfs because, generally, no lines other than hydrogen
HD
2 7806
Fig. 10.7. The spectra of F8 stars with different luminosities are shown. The
supergiant spectrum (luminosity classes la, Ib are at the top), the giant
spectrum (luminosity class III) is in the center, the subgiant (luminosity class
IV) is below the giant, and the main sequence star spectrum (luminosity class
V) is at the bottom. The metallic line strength increases for higher luminosities,
especially the strength of the SrII line at 4078 A. The G band at 4300 A is
broad for giants and main sequence stars. (For Morgan, Abt & Tapscott 1978.)
98
lines are seen. There are very few white dwarfs for which some of the
strongest metallic lines can be recognized, like, for instance, the lines of
ionized calcium at wavelengths of 3933 A and 3968 A. Some white dwarfs'
spectra show no lines at all, and they are called DC spectra or, the stars,
DC white dwarfs (C from continuum).
HZ21
EG&6
Comparison
Fig. 10.8. Spectra of white dwarf stars are shown. In the top DO1 and DB4
spectra only helium lines are seen, for instance at 4713 A and 4471 A. For the
DA star spectra the hydrogen lines at 4101 (H^), 4340 (Hy), and 4861 (HJ are
strong and very broad. In the bottom spectra of R640 and van Maanen 2 the
hydrogen lines cannot be recognized. The Ca+ lines at 3933 and 3968 A are
very broad and strong. (From Greenstein 1986.)
11
Understanding stellar spectra
11.1
The solar spectrum
In Fig. 6.7 we have shown the overall energy distribution of the sun.
We have seen that the solar energy distribution has its maximum at a
wavelength of about 5000 A, which, according to Wien's displacement law,
would mean that the temperature of the solar surface has to be about 5800 K
if the sun radiates like a black body (see (4.7)), however, we do not yet know
if the sun does, indeed, radiate like a black body. (We saw, however, in
Chapter 8, that the effective temperature of the sun is also close to 5800 K). In
this section we want to look at the solar line spectrum. In Fig. 11.1 we
reproduce a small part of a 'positive' spectrum of the sun, which means we see
dark spectral lines at those wavelengths where little or no energy is received.
The strongest lines are the lines from Ca + . This notation means that the lines
are due to a calcium atom which has lost one electron and is therefore
positively charged by one elementary charge. We call this a calcium ion. The
lines are also called Ca II lines, the strongest of which are visible as broad
bands at 3933 and 3968 A. The hydrogen lines seen at 4340, 4861, and
6562 A are much weaker. Strong lines of magnesium and sodium are also
seen as well as many lines of iron, chromium, titanium, silicon, manganese,
and other elements.
Fig. 11.1. A small part of the solar line spectrum is shown. Many dark
absorption lines, the so-called Fraunhofer lines, are seen. The wavelengths of
these lines agree with the wavelengths of spectral lines seen in laboratory
spectra of different elements. By comparison with laboratory spectra we can see
which elements are present in the sun.
99
100
11 Understanding stellar spectra
11.2
Line identification
How do we know which atoms or ions cause these spectral lines? We

have to study in the laboratory the spectra of light sources which we know
contain atoms and ions of a certain element, say, iron. We compare spectra of
such a light source with the lines which are seen in stellar spectra or in the
solar spectrum. If we see the same lines which we see in the laboratory iron
spectrum we know that these lines in the stellar spectrum are caused by iron,
at least if we know that no other elements are present. If other elements are
also present, we have to study another light source which contains only
atoms of that other element in order to identify the lines which do not belong
to iron. It has taken a long period of laboratory studies to disentangle all the
stellar spectra and to identify which lines are due to which element. We
still cannot identify all spectral lines, either in the laboratory or in the solar
spectrum. But for the majority of lines we know which element causes them.
There are large tables available, for instance the tables by Mrs Moore
Sitterly, which list the wavelengths of the lines and give their identification in
terms of which atom or ion is responsible for a given line. Sometimes several
different elements form a line at nearly the same wavelength. One then has to
study in more detail which other lines can be seen. If, for instance, an iron line
is a possibility, we have to check whether other strong iron lines are also
there; if not, then the identification as an iron line is rather unlikely. In some
sense, the identification of spectral lines in a stellar spectrum is like a puzzle:
all the pieces have to fit together.
The presence of spectral lines of a given element in the solar or in a stellar
spectrum tells us that a particular element is present on the star, but does the
line strength tell us how much of that particular element is present? For
instance, does the high strength of the Ca + lines in the solar spectrum tell us
that the sun consists mainly of calcium? Or that the A stars are mainly made
of hydrogen? In order to answer this question, we have to discuss in detail the
origin of the spectral lines and the factors which determine their strengths. A
full discussion of the stellar spectrum analysis must be left until Volume 2.
Here we can give only the main results.
11.3
Understanding the spectral sequence
11.3.1 General discussion

In Chapter 10 we saw that most of the stellar spectra can be classified
in a two-dimensional scheme, the one dimension given by the spectral
sequence and the other by the luminosity classes. We suggested earlier that
11.3 Understanding the spectral sequence
101
the spectral sequence is indeed a temperature sequence, while the luminosity

classification describes mainly the size of the star (actually the gravitational
acceleration in the photosphere). The temperature sequence was suggested
because the spectral types are strongly correlated with the colors of the stars.
On the other hand, we also saw that different lines and also lines of different
elements are seen in different spectral types. Does this suggest that stars with
different temperatures have different chemical compositions? A detailed
analysis of the stellar spectra shows that this is not the case. Almost all
stars in our surroundings have the same chemical composition. How, then,
can we understand the changing line strengths of the lines of different
elements for stars with different temperatures?
11.3.2 The changing strengths of the hydrogen lines for different
spectral types
We discussed in Chpater 10 the fact that the hydrogen lines show

maximum strengths for stars of spectral types A0, and decrease their
strengths for earlier spectral types (B and O stars) as well as for later spectral
types (F, G, K, and M stars). The reason for the decrease of the hydrogen line
strengths for the later spectral types is the decreasing temperature for the
later spectral types. Only hydrogen atoms in the second quantum level of the
hydrogen atom can contribute to the hydrogen line absorption in the visual
spectral region. The hydrogen atoms have to be excited into this level and it
takes 10 eV to bring the atoms into this excited level. The excitation can occur
due to collisions with electrons which need to have high energies for this to
happen and there are not many electrons with energies high enough if the
temperature is too low. Excitation can also occur due to the absorption of a
photon with a wavelength of 1216 A. If you remember the wavelength
dependence of the Planck function you will realise that for low temperatures
there are very few photons with such short wavelengths. For decreasing
temperatures we therefore find a decreasing number of hydrogen atoms in
the excited, second quantum level which can absorb in the hydrogen Balmer
lines which we observe in the visual spectral region.
The reason for the decreasing strengths of the Balmer lines at higher
temperatures is more difficult to explain briefly. People tend to say that this
is due to the ionization of hydrogen which also reduces the number of
hydrogen atoms in the second quantum level, because many hydrogen
atoms lose their electron altogether. While this is true it is not the real reason
for the decreasing line strength. What really counts is the ratio of the number
of atoms absorbing in the hydrogen lines to the number of atoms absorbing
102
in the underlying continuum, which for the hot stars are the number of
hydrogen atoms in the third quantum level of hydrogen. This number also
decreases with increasing temperature in the stellar surface layers, but with
increasing temperature their number decreases more slowly than those in the
second quantum level, therefore the ratio of those in the second quantum
level to those in the third level decreases and this reduces the strength of the
Balmer lines.
11.3.3 The strengths of the helium lines
The helium lines which can be observed in the visual spectral region
need even higher excitation energies than the hydrogen lines. They can only
be excited by collisions with electrons which have at least about 20 eV
energy or by photons which have wavelengths as short as about 500 A.
Very high temperatures are required for such high energies of the electrons
or to produce a measurable number of photons at such short wavelengths.
Temperatures higher than 10 000 K are required to make the helium line
visible. For low temperature stars we can never see the helium lines in the
absorption spectra of the stars. You may wonder why then the element got
the name 'Helium' which means the element only seen in the sun? The sun
obviously is a cool star, too cool for helium lines to be seen in the
photospheric absorption spectrum. The helium lines were observed in the
solar spectrum during solar eclipses when the outer layers of the sun become
visible. These are not usually seen because the solar disk spectrum normally
seen is so much more luminous than the spectrum which becomes visible
during solar eclipses.
The spectrum seen during an eclipse is an emission line spectrum emitted
by the high layers above those layers which we normally see. In these high
layers, called the solar chromosphere and corona and the transition layer
between the chromosphere and corona, the temperature increases outwards
and reaches values high enough for the higher energy levels in the helium
atom to be excited and then emit light in the helium lines which can be
observed. (There is actually also one very weak infrared line of helium which
is seen in the normal solar spectrum, but which is also caused by the helium
atoms in the high chromospheric layers.) The absence of helium lines in cool
stars makes it very difficult to determine helium abundances in cool stars.
For the lines of the helium ions to become visible we need still higher
energies first to ionize the helium, which requires about 24 eV as compared
to the 13.6 eV required for the ionization of hydrogen, and in addition we
need to excite an energy level in the helium ion from which the ions can
103
absorb the visual lines. These lines are therefore seen only for temperatures
higher than about 30 000 K.
11.3.4 The strengths of the metallic lines
The lines of the heavier elements are generally expected to be weaker

than those of hydrogen because these elements turn out to be very rare in
stars. For each 1000 hydrogen atoms wefindone atom which is heavier than
helium. For high temperatures, when the helium and hydrogen atoms are
excited to the energies necessary for the line absorption, the 'metallic' lines
are therefore generally much weaker than the helium and hydrogen lines, at
least in the visual spectral region; only at lower temperatures, when the
hydrogen atoms are not excited to the second quantum level in large enough
numbers do the lines from the heavier elements become comparable in
strength. For very low temperatures they may even become stronger. For the
K stars we see mainly the lines of the neutral atoms because the energies of
the free electrons and the energies of the available photons are not high
enough to remove one electron from these heavier elements. For somewhat
higher temperatures the photons and electrons become more energetic and
in those elements from which the electron can be removed easily they can
then be stripped from the atoms; they become ionized and we start to see the
lines from the ions of iron, titanium, chromium etc. for the G and F stars. The
lines of the atoms become weaker because their number decreases.
For A stars, the temperatures become high enough so that now even the
ions can again be ionized; they can lose an additional electron and become
doubly ionized. The ionization can proceed even further for the B stars and
the still hotter O stars. We may then also see lines of Si or Fe which are doubly
ionized, the so-called Fe III or Si III lines. Most of the lines of these ions with
higher degrees of ionization can only be seen in the ultraviolet, but a few lines
are also in the visual spectral region and can be studied to determine the
abundance of these elements in the hot stars. In a spectrum analysis of all
these different types of stars, the changes of the continuous absorption and
the changes in the ionization and excitation have to be taken into account in
order to determine the element abundances.
If we do this, we find that nearly all the stars in our neighborhood and
actually in our stellar system, our galaxy, have nearly the same chemical
composition. We even find that the relative abundances of the heavy
elements agree rather well with the relative abundances found originally on
earth, especially if we take into account that some of the elements which can
form volatile substances have been lost earlier from the Earth's atmosphere.
104
11.3.5 The molecular bands
In Chapter 10 we discussed the fact that for very red, presumably

very cool stars, the M, N, S stars, we also see strong molecular bands which
disappear for higher temperature stars.
The difference between atomic lines and molecular bands is due to the
large number of energy levels which molecules have. For each excited state of
an electron within the molecule, there are an infinite number of rotation and
oscillation energy states. The energy differences between the rotational
energies are very small. Since spectral lines can originate due to transitions
between all these many very close rotational energy states, these lines have
only very small wavelength differences. They lie very close in the spectrum,
forming a band which can be resolved into separate lines only if very high
resolution spectra are taken. On low resolution spectra, the many close lines
appear as broad bands.
Molecules form when two atoms or a molecule and an atom collide. They
may also be destroyed by photon absorption or in collisions, in the same way
as atoms are ionized when they collide with an electron or absorb a photon.
As for atoms, higher temperatures lead to more frequent destruction of
molecules and the equilibrium shifts from a large number of molecules to
larger numbers of atoms. The binding energy of most molecules like H 2 O,
TiO, ZrO, CN, etc., is of the order of 4-6 eV. Molecules can, therefore, be
destroyed at rather low temperatures. For stars with temperatures higher
than about 5000 K, molecular bands become very weak. For the sun, they
can only be observed with high resolution, when weak bands can still be
detected. In the sun, molecules are only present in the coolest outer layers,
and even then they are present only in small numbers. The presence of
absence of molecules in the stellar spectra tells us mainly about the
temperature of the star, though higher pressures favor the formation of
molecules. For most stars the strengths of the molecular bands are quite
consistant with the element abundances also observed for the sun.
In Table 11.1 we list the element abundances determined for different stars
and compare them with element abundances determined for chondritic
meteorites. We see that the relative abundances of the heavy elements
determined for the different objects agree rather well. The average abundances determined for the Earth depend on the assumptions made about the
interior of the Earth for which we can not make a direct chemical analysis.
The best we can tell is that the heavy element abundances also agree with
those found in stars if we take into consideration that volatile atoms and
molecules have been lost.
105
Table 11.1 Logarithmic Element abundances for the most abundant

elements determined for stars in our neighbourhood and in
the carbonatious chondrites (meteorites).
Population I
Atomic
Number Element
Sun2
G2V
aLyr
A0V
1
2
8
6
7
10
26
14
12
16
13
28
20
11
24
17
15
25
22
27
19
12.0
(11.0)
8.8
8.5
8.0
(7.9)
7.6
7.5
7.4
7.2
6.4
6.3
6.3
6.3
5.9
5.6
5.5
5.4
5.1
5.1
5.0
12.0
(11.4)
9.3
?
8.8
?
7.1
8.2
7.7
?
5.7
7.0
6.3
7.3
5.6
?
?
5.3
4.1
Hydrogen (H)
Helium (He)
Oxygen (O)
Carbon (C)
Nitrogen (N)
Neon (Ne)
Iron (Fe)
Silicon (Si)
Magnesium (Mg)
Sulfur (S)
Aluminum (Al)
Nickel (Ni)
Calcium (ca)
Sodium (Na)
Chromium (Cr)
Chlorine (Cl)
Phosphorus (P)
Manganese (Mn)
Titanium (Ti)
Cobalt (Co)
Potassium (K)
Population II
Meteorites,
carbonaceous
chondrites
7.6
7.6
7.6
7.2
6.5
6.2
6.4
6.2
5.6
4.8
5.5
9
9
5.1
HD
140283
HD
19445
12.0
?
?
6.4
?
?
5.2 0.3
5.1
4.9
12.0
?
?
?
?
3.7
4.2
4.0
3.5
3.6
?
?
2.8
2.8
2.7
9
5.7 0.2
6.0
6.5?
?
4.5
4.7
4.8
?
3.8
?
?
3.7
3.4
?
?
1. Normalized to log N(H) = 12.00

2. These abundances are considered to be 'Cosmic Abundances.1 '
It is found that the vast majority of stars consist mainly of hydrogen with
about 10% + 5% of helium mixed in, if we count the number of atoms or ions.
Since each helium nucleus is four times as heavy as a proton, the abundance
of He by weight, Y, can be computed from
4-ab(He)
_ 4-0.1 _0.4
01
4-ab(He) + 1 ab(H) " 0.4 + 0.9 ~ 1.3 "

The next most abundant elements are C, N, O, and Ne. These four
elements together make up about 1/10 of 1% of the total number of particles.
Since their atomic weight is, on the average, about 15, we find that their
fraction by mass, called Z, is about
Z = 0.02 0.01
106
The next most abundant elements are Si, Mg, Fe, and Al, though these
elements are more rare than C, N, O, and Ne by another factor of about 10.
We realize that the heavy elements which make up our Earth constitute only
a very small fraction of the gas in the cosmos.
Generally the relative abundances of the heavy elements with respect to
one another are very similar for all the stars and even for our Earth, though
for cool, luminous stars we mayfindsomewhat different relative abundances
especially of C, N, and O.
There are, however, a few stars for which the abundances of the heavy
elements as compared to hydrogen are very different from those seen in
the sun. The overall abundance of the elements heavier than He may be
lower by two or even almost three powers of 10. These are the so-called
population II stars which we will discuss in the next chapter.
12
Population II stars
There are a few stars in our neighborhood whose spectra show a different
chemical composition for their photospheres. These stars were previously
known as subdwarfs. The reason for this name was their position in the color
magnitude diagram: they appeared below the main sequence, which means
they either are too faint for their color or they are too blue for their
brightness. A spectrum analysis showed that the latter is the case. It turned
out that, for these stars, the relative abundances of the heavy elements with
respect to one another are quite similar to the ones observed for the sun, but
the overall abundances of the heavy elements with respect to hydrogen and
helium are considerably reduced by up to a factor of 500, though most of
them have much smaller abundance reductions. In these metal-poor stars,
the metallic lines are much weaker than for normal stars of the same
temperature. Since spectral lines are generally stronger in the blue, and
especially in the ultraviolet, than in the red, the lines take more energy out of
the ultraviolet and blue spectral region than out of the red. If the lines are
weakened in the metal-poor stars, more energy is restored to the ultraviolet
and blue spectral region than in the red and the stars therefore look bluer,
especially in the ultraviolet. They show an ultraviolet excess which can, in
fact, be used to determine their metal deficiencies. In Fig. 12.1 we show the
two-color diagram for the stars in the globular cluster M 92 (object No. 92 in
the Messier catalogue of nebulous objects in the sky), and compare it with
the two-color diagram of the Hyades star cluster in which the stars have
normal or perhaps even slightly higher than normal metal abundances. The
cool stars (which have the strong metallic lines) in M 92 have colors which
are considerably bluer in the U B colors than the Hyades stars. The M 92
stars have an 'ultraviolet excess' because they are metal-poor.
For the very metal-poor stars (reduction factors 50-100), the relative
abundances of the heavy elements with respect to each other appear to be
nearly the same as in the sun. For heavy element reduction factors between
10 and 50, it turns out, however, that different elements have different
107
108
12 Population II stars
reduction factors. Apparently, the lighter elements are less reduced than the
iron group elements. The differences in the reduction factors, and what this
means with respect to the chemical evolution of the stars and the Galaxy, are
the subject of many modern studies. A final answer cannot yet be given.
It is interesting to note that these metal-poor so-called subdwarfs also
differ from the normal stars in other respects. They all show high velocities
relative to the sun. Within the galactic plane, the sun and all the stars
in our neighborhood orbit the galactic center. We can also say they
rotate around the galactic center the way the planets orbit around the sun.
Stars in our neighborhood must orbit with nearly the same speed and
direction of velocity as we do, at least if they have nearly circular orbits as the
sun has. If stars in our neighborhood have very different velocities, it means
that they must have different galactic orbits: they must have very elliptical
orbits and/or their orbital planes must be inclined with respect to the orbital
plane of the sun, which is identical with the plane of our Galaxy. The high
relative velocity of the subdwarfs shows that their orbits are different from
the orbits of the 'normal' stars in our neighborhood. They have elliptical
orbits which are also inclined with respect to the galactic plane. Indeed, it
appears that the overall abundances of the heavy elements in these stars are
correlated with the maximum distance z above the galactic plane which these
stars can reach (see Fig. 12.2). These very metal-poor stars appear to belong
to a stellar population which does not fit into the galactic plane; rather, they
belong to a stellar population which originated above the galactic plane.
Once they were formed, they fell towards the galactic center. Their angular
-0.8
M 92
-0.4
0
QA
o V < 17.0 (Giants HB)
0.8
19>V>17 (Main Seq.)
1.2
-0.2
0.2
(M
0.6
B-V
0.8
1.0
Fig. 12.1. The two-color diagram for the stars in the globular cluster M92 is
shown and is compared with the two-color diagram of the Hyades stars. The
stars in M92 have more radiation in the U band, which means they have an
ultraviolet excess because they are 'metal'-poor.
109
momentum prevented them from falling into the galactic center, but made
them orbit around it in an elliptical orbit, with the orbital plane inclined with
respect to the galactic plane (see Fig. 12.2). This population of stars, which
does not fit into the galactic plane but, instead, belongs to the space above
and around it, the so-called galactic halo, are called population II stars in
contradistinction to the population I stars which we called the 'normal'
stars - those stars with orbits in the galactic plane which also must have
been formed within the galactic plane.
In Section 5.3, we discussed the color magnitude diagrams of stars in
clusters, and we saw that globular clusters (that is, clusters with a large
number of stars and a rather high density of stars), show different color
magnitude diagrams from those of the open clusters which have a much
lower number of stars. It turns out that the open clusters belong to the
population of the galactic plane: their orbits lie in the galactic plane. The
globular clusters, on the other hand, belong to the halo population, or the
population II. Like the subdwarfs, they also show reduced abundances of
the heavy elements. As for the subdwarfs, the most metal-poor globular
clusters belong to the largest distances from the galactic plane.
Since it appears that stars form in regions of high densities of interstellar
gas and dust, and since we do not see much gas and dust now in the galactic
halo, it appears that the population II stars must have formed at a time when
there was more gas and dust in the halo than now. This means they must
have formed at a time when the galaxy had not yet contracted into a disk. We
are, therefore, led to suspect that the population II stars are older than the
population I stars, which were formed in the galactic plane. As we just
saw, the population II stars are deficient in heavy elements; they must
have been formed from gas and dust which was poor in heavy elements.
Population I stars, which apparently were formed later, were formed from
material which contained a much higher proportion of heavy elements. This
suggests that, with time, the interstellar medium from which the stars formed
Orbit of Population II star
Galaxy
Fig. 12.2. Population II stars have orbits which do not fit into the galactic
plane. They generally have elliptical orbits which reach high above the galactic
plane into the galactic halo.
110
was enriched in heavy elements. We expect to find a correlation between the

age of the stars and the overall abundances of the heavy elements. We will
learn in Volume 3 how we can determine the ages of clusters and stars. Here,
we can only state that we have not been able to demonstrate such a
correlation, but we also do notfinda contradiction. The oldest stars all seem
to have about the same age regardless of their chemical composition, at least
to within the errors of the age determination. It appears that the enrichment
in heavy elements took place in a timespan which is short in comparison with
the age of the oldest stars.
13
Stellar rotation
We can measure rotational velocities of stars from the widths of their

spectral lines. For a rotating star, one side of the star moves away from us
while the other side moves toward us, at least if the line of sight does not
coincide with the rotational axis (see Fig. 13.1). The superposition of all the
spectra originating from the different parts of the star gives a broadened
spectral line (see Fig. 13.2). The width of the broadened line tells us the
rotational velocity of the star except for the factor sin z, where i is the angle
between the line of sight and the rotational axis.
The effects of rotation on the appearance of the spectral lines is illustrated
in Fig. 13.3.
It turns out that the radial velocity, i.e., the component of the rotational
velocity in the direction of the line of sight, is constant along stripes across
the stellar surface, which are parallel to the direction of the rotational axis
projected against the background sky, as we will now show. Suppose the
rotational axis is perpendicular to the line of sight. At a given latitude </> on
the star, the rotational speed is given by (see Fig. 13.4)
^rot = iw(equator) cos </>.
(13.1)
If X is the geographical longitude on the star counted from the central

meridian, then the line of sight component of the rotational velocity, i.e., the
radial velocity vr seen for each point on the stellar surface, is given by (see
Fig. 13.1(b))
t;rot = vrot sin X = vrot(equator) x cos </> x sin X.
(13.2)
The radial velocity vr is therefore constant for cos (/> sin X = const.
If we describe the sphere by x, y9 z coordinates in a rectangular coordinate
system (see Fig. 13.5), we find, where r is the radius of the sphere,
z = r sin <j> )
x = psinX [
(133)
y = p cos X J
111
112
13 Stellar rotation
Vrot
(a)
(b)
Fig. 13.1. For a rotating star whose rotation axis is inclined with respect to the
line of sight, a distant observer will see one side of the star move away from
him, while the other side moves towards him.
Fig. (a) shows a view of the star with the line of sight in the equatorial plane,
Fig. (b) a view from an observer whose line of sight is parallel to the rotation
axis. An observer whose line of sight is in the equatorial plane sees one side
move towards him and the other side move away from him. If the observer
looks along the rotation axis no Doppler shift is seen.
Stellar line
= Sum of all
lines
\J
(a)
(b)
Fig. 13.2. The spectral lines of a non-rotating star (a) look generally sharp. For
a rotating star, the lines of the opposite sides of the star are shifted in opposite
directions, while the center of the star shows the unshifted spectrum (b). Since
we can not spatially resolve the star, we can only observe the sum of the
spectra originating on different parts of the stellar surface. This means we see a
broadened line.
113
13 Stellar rotation
(a)
(b)
(c)
4467
4495
Fig. 13.3. The effects of rotational broadening on the spectral lines are shown
for three stars with different rotational velocities. (From Babcock 1960.)
P<0=/'COS(|)G)
= ve C O S <|>
Fig. 13.4. The rotational velocity at the stellar latitude 0 is given by vrot = p-w
where p = r c o s 0, ve = vrot(equator)
p = r cos <t>
J \
z = r sin <J>
x = pcos X
y =psin X
(a)
Fig. 13.5. The relations between the rectangular coordinates x, y, z and the
spherical polar coordinates r, 0, X are shown, (a) View from the equatorial
plane, (b) View from the pole.
114
13 Stellar rotation
Here, p is the radius of the cross section through the sphere at height z. From
Fig. 13.4 we see that
p = rcos</>.
With this we find
x = rcos(/)sin>l and y = rcos</)cosX.
(13.4)
We saw above that vT = constant for cos (j> sin X constant. From (13.4) we
see that this is the case for x = constant on the sphere, which means along
cross sections through the sphere at constant distance from the z-y plane.
These are stripes on the surface of the star parallel to the projected
direction of the rotation axis. In order to calculate the line profile for a
rotationally broadened line we have to add up the contributions from these
stripes, as shown in Fig. 13.2.
For (13.1)(13.4) we have assumed that the line of sight is perpendicular to
the rotation axis. If this is not the case, all the vr are multiplied by the factor
sin i where i is the angle between the rotational axis and the line of sight.
Generally, we have no way to determine this angle i; we can, therefore, only
measure vT sin i. In the tables giving rotational velocities of stars, we always
find this value vr sini, which gives us only a lower limit to the actual
rotational velocity of a given star.
Assuming a random distribution for the orientation of the rotation axes
(probability of i proportional to sin i) we can, however, in a statistical sense
correct for the sin i factor and determine the true distribution of equatorial
rotational velocities.
In Fig. 13.6, we show the distribution of average rotational velocities
vrot (equator) for field stars as a function of the spectral types. Rather high
rotational velocities are found for O, B, and A stars. For late A and early
F stars, the rotational velocities drop almost to zero.
In Fig. 13.7, we show the distribution of rotational velocities for stars in
different clusters in comparison with the distribution for field stars. While we
see a similar picture for all the clusters except for the Pleiades, the drop in
rotational velocities is more or less pronounced in the different clusters. In
Fig. 13.7, we have actually ordered the clusters according to age. (We will see
in Volume 3 how the ages can be determined.) In the old field stars and in
the Coma Berenices cluster, the drop in rotational velocities at B V = 0.3 is
well pronounced. In the somewhat younger clusters, Praesepe and the
Hyades, this drop at B V = 0.3 is less obvious. The rotational velocities
remain high also for stars with B V > 0.4. One clearly sees that for stars
with B V ^ 0.3, a braking mechanism is at work which is even more effective
200
Ve km/s
100
AO
FO
Spectral type
Fig. 13.6. The dependence of the mean equatorial rotational velocities on the
spectral types is shown for main sequence stars, for giants, and for supergiants,
according to Schmidt-Kaler (1982). V: main sequence stars, III: giants, la:
supergiants.
300
(a)
200
_
100
300
20
(c)
FIELD STARS
(b)
COMA BERENICES
~*b
PRAESEPE
(d)
HYADES
(f)
a PER
100
0
300 -
(e)
-y
PLEIADES
200
100 _. \
O.I
0.2 0.3 0.4 0.5
O.I 0.2 0.3 0.4 0.5 0.6
B-V
Fig. 13.7. The rotational velocities of main sequence stars are shown as a
function of B V for field stars, i.e., stars which are not in clusters, and for
stars in different Galactic clusters. Where necessary, reddening corrections were
applied (see Chapter 19). The clusters are ordered according to their ages (see
Volume 3). The Coma Berenices, the Hyades and Praesepe clusters are the
oldest, the a Perseus and the Pleiades cluster are the youngest. In the Pleiades
cluster the stars all still have their large rotational velocities. (From BohmVitense & Canterna 1974.)
116
13 Stellar rotation
for stars with B V ^ 0.4. While these subtle differences are not yet
understood, astronomers generally believe that this braking mechanism is
related to stellar winds, which in turn are related to the presence of stellar
coronae which can be observed for the cooler stars. We will come back to this
discussion in Volume 2.
14
Stellar magnetic fields
14.1
General discussion
We are all accustomed to the fact that our Earth has a magnetic field
whose shape comes rather close to that of a dipolefield,with the magnetic
axis not being very different from the rotational axis of the Earth. This raises
the suspicion that the magnetism is due or at least related to the rotation of
the Earth. We may therefore wonder whether stars which have much higher
rotational velocities than the Earth might have much stronger magnetic
fields than the Earth. It seems very interesting to check. The question arises:
how can we measure magnetic fields on stars? We clearly cannot take a
magnetometer to the star's surface. All we can get from the star is its light.
Fortunately, nature has provided an effect of a magnetic field on the light
which can be used to measure magnetic fields of stars. This is the Zeeman
effect, named after its discoverer.
14.2
The Zeeman effect
Zeeman discovered that for a laboratory light source which emits an

emission line spectrum in a laboratory magnetic field, the spectral lines
generally split into several components. If you observe the light in an
arbitrary direction with respect to the direction of the magnetic field, you
will, in the simplest case, see a line split into three components, the socalled
Lorentz triplet. The central component remains at the original wavelength
Ao, which the line had without a magneticfield.The other two components
are displaced symmetrically to both sides of the central component. The
displacement in frequency, Av, is given by
Av= + - x - ^ - = 4.66x l ( T 5 x # , with v = |
(14.1)
or, expressed in AA, we find with

117
118
14 Stellar magnetic fields
Av
dvl v /
v"
\ /
AX= P x 4.7 x l O " 5 x i f
(14.1(a))
where H is the magnetic field strength to be measured in Gauss, e is the

electronic charge to be measured in electromagnetic units, and m is the
electronic mass.
The relative strengths of the three components change if we observe in
different directions. If we observe in a direction parallel to the magnetic field
lines, we see what is called the longitudinal Zeeman effect. In this case, we see
only the two displaced components: the intensity of the central component
has decreased to zero (see Fig. 14.1). The light in the two displaced
components turns out to be circularly polarized in opposite directions. The
direction of the circular polarization for each component depends on the
direction of the magnetic field. If, in Fig. 14.1, the magnetic field around the
light source were to point in the opposite direction, the directions of the
circular polarization for the two lines would have to be inverted.
If we were to observe the light of the light source in the direction
perpendicular to the direction of the magnetic field, we would observe the socalled transverse Zeeman effect, which means we would see all three
components of the Zeeman triplet, the central component being twice as
strong as each of the displaced components. Now all three components turn
o o
Light
H
AA
source
Observer
(b)
Observer
(a)
Fig. 14.1. The splitting of a spectral line due to the Zeeman effect is shown for
the case where we observe the light source in the direction parallel to the
magnetic field, i.e., we observe the longitudinal Zeeman effect (a), and for the
case when we observe perpendicular to the magnetic field, i.e., when we observe
the transversal Zeeman effect (b). The directions of circular and linear
polarizations of the emitted light are indicated by the arrows.
14.2 The Zeeman effect
119
out to be linearly polarized. The two shifted components are polarized in the
direction perpendicular to the magnetic field, while the central component is
polarized parallel to the magnetic field (see Fig. 14.1, the lower left plot).
Since the splitting of the lines increases with increasing magnetic field, a
measurement of the line splitting gives a measurement of the magnetic field
strength. Hale, in 1908, was the first to make use of this Zeeman effect to
measure the magnetic fields in sunspots. In Fig. 14.2 we show a picture of a
solar sunspot, taken in the light of the hydrogen Ha line. This picture
reminds us very much of the magnetic lines of force outlined by iron filings
on a piece of paper on top of a small magnet in the laboratory (see Fig. 14.3).
This made Hale suspicious, and he actually measured magnetic fields of
several thousand Gauss in the sunspots.
From (14.1 (a)) we can calculate how large the splitting of a spectral line at
about k0 = 4300 A will be. For a field of 1000 Gauss, we find AA//1 = 2 x
10 ~6. Due to rotation, the spectral line of a star has a width of about AXD =
Ivjc, where c is the velocity of light. For i?rsini = 100 km s" 1 , we find a
width of AX = 1.4 A. This means that the widths of the spectral lines due to
rotation are much larger than the splitting of the lines due to the Zeeman
effect. We will not be able to measure the splitting of the line. Only in cases of
a small vr sin i and in the case of a very strong magnetic field, can we hope to
see the splitting of the lines directly. In Fig. 14.4 we show the spectrum of HD
215441 (star No. 215441 in the Henry Draper catalogue of stars), as taken by
Babcock 1960. In this spectrum the three Zeeman components of the Cr line
Fig. 14.2. High spatial resolution pictures of sunspots are seen, taken in the
light of the H a line of hydrogen. We have the impression that we actually see
the outline of magnetic field lines of force emerging from the sunspots. (From
Bray & Loughhead 1974.)
120
at 4254 A can actually be seen. The amount of splitting indicates a

magnetic field of 34 000 Gauss for this star.
For stars with magnetic fields of the order of 1000 Gauss, as are observed
in sunspots, and for rotational velocities of the order of 50 km s ~ 1 , the
Fig. 14.3. The magnetic lines of force of a laboratory magnet as outlined by

iron filings on a piece of paper on top of the magnet (top picture), and on the
side of a magnet (bottom picture).
121
different components cannot be resolved. We observe only slightly broadened lines (see Fig. 14.5). We clearly have the best chance to separate the
components if we observe the longitudinal Zeeman effect only, which
eliminates the central component. However, we cannot arrange our line of
sight. We therefore have to build an instrument which manages to separate
the components with circular polarization. Babcock, at Mount Wilson
observatory, was the first to design such an instrument for the observation of
I9A
f
\1
Cr n 4558
Fig. 14.4. We reproduce a small section of the spectrum of the star No. 215441
in the Henry Draper catalogue. This star has a magnetic field of 34 000 Gauss
and a rotational velocity of only Skms" 1 . The lines are narrow enough and the
magnetic field strong enough so the Zeeman triplet of the Cr line at 4558 A
can be seen clearly. The laboratory (emission line) spectra seen above and
below the stellar spectrum are needed for the line identification. (From Babcock
1960.)
Line without H
Line with H
Fig. 14.5. For broad spectral lines and small magnetic fields, no splitting, but
only a broadening of the line can be seen due to a magnetic field. The light on
the two sides of the line is circularly polarized in opposite directions.
122
stellar magnetic fields. His instrument is shown in Fig. 14.6. The essential
parts are the calcite crystal and the A/4 mica. The calcite crystal has the
property of having different diffraction angles for linearly polarized light in
perpendicular directions. The 1/4 mica makes linearly polarized light out of
circularly polarized light. For opposite directions of the circularly polarized
light the directions of the linearly polarized light obtained are perpendicular
to each other. A generally unpolarized beam of light from the star, which we
can regard as consisting of two beams of circularily polarized light, being
F/30 beam from
100 inch mirror
Decker
A/4 mica
'/'
'/'
'/' 4/ ///
4?
Glass
/ -
~EH
Calcite
Plane of slit
v y/"^"f""/\ v
Slit of spectrograph
Fig. 14.6. The essential parts of Babcock's magnetograph are shown. The A/4
mica plate transforms circularly polarized light into linearly polarized light:
opposite directions of circular polarization lead to linear polarization in
perpendicular directions. The two coinciding light beams from the star with
perpendicular directions of linear polarization suffer different diffraction in the
calcite crystal. The beams appear side by side when they exit the crystal. They
then enter the spectrograph slit separated along the direction of the slit and
form two spectra side by side on the photographic plate, where the spectra of
opposite directions of polarization are then seen side by side. (From Babcock
1960.)
123
polarized in opposite directions, enters the A/4 mica plate. On the other side
of the mica plate the beam now consists of two linearily polarized beams
which have perpendicular directions of polarization. They now pass through
the calcite crystal in which one of the beams passes straight through, while
the other beam is diffracted. The beams with different polarization therefore
exit the calcite crystal at different places. The calcite crystal thickness is such
that the two beams exit side by side. The beams then enter the spectrograph
slit. In the spectrograph they are each dispersed into a spectrum. The two
spectra from the two beams now appear side by side. Since the one beam
contained light from the one direction of polarization a given spectral line is
shifted, say, to longer wavelengths if the star has a magnetic field. In the other
spectrum the light with opposite circular polarization is seen. In this
spectrum we see a line which is then shifted to shorter wave-lengths if the star
has a magnetic field. In those two side by side spectra the line therefore
appears at slightly different wavelengths, if the star has a magnetic field. The
shift corresponds to the separation of the two shifted lines from the Lorentz
triplet.
For different lines the amount of splitting in a given magnetic field is
different, depending on the structure of the energy levels to which the line
transition belongs. For some lines the splitting is larger than is given by
(14.1), and for some it is smaller. Some lines do not split at all. The sensitivity
of the different lines to the Zeeman effect can be measured in the laboratory,
and we can check whether the measured splittings for the different lines show
the same ratios as in the laboratory. In this way we can make sure that we are
actually seeing the effect of a magnetic field on the star.
The line shifts for the two spectra are measurable even if the lines are
broader than the shift, as can be seen from Fig. 14.7 and also in Fig. 14.8
where we show such pairs of spectra as observed by Babcock. Even with this
r>
-*> A
Fig. 14.7. Sets of two spectra of opposite circular polarization are obtained
with a magnetograph.
124
clever instrument, we are limited to fairly strong magnetic fields for our
measurements. Fields of less than 100 Gauss cannot be measured even today,
though there are still more sophisticated instruments now in use. Modern
instruments make use of the polarization created by the magnetic field line
shifts in the hydrogen lines. The principle is the same as described here. The
shifts of the lines in the polarized beams causes a slight polarization in the
wings of the hydrogen lines which has opposite signs in the long wavelength
ans short wavelength wings. A difference in polarization for the two wings
indicates a stellar magnetic field.
As we mentioned previously, the splitting of different metallic lines due to
the Zeeman effect is different. For each line, the splitting can be measured in
the laboratory for different field strengths, and the ratio of the splitting for
different lines for a given field strength can be determined. As a further check
on our interpretation of the splitting as the result of the Zeeman effect, we can
determine in the stellar spectra the ratio of the splitting for the different lines
seen in the stellar spectra and check whether they agree with the laboratory
values. Such checks were done by Babcock and he found good agreement.
It turned out that for most stars for which magnetic fields could be
measured (only a small fraction of all stars), these fields vary with time in a
periodic way. In Fig. 14.9 we have plotted the fields measured by Babcock
for stars with measurable magnetic fields as a function of their spectral types.
For the stars with variable magnetic fields, the maximum field strength was
plotted regardless of whether we see N or S polarity. From Babcock's
measurements it appeared that the maximum field strengths occur for early
A stars. No measurable magnetic fields were found for cool stars. We know
that the sun has a weak magnetic field which is also periodically variable. We
can measure this field of a few Gauss on the sun only because we can measure
single spots on the solar surface where the lines are not rotationally
4205
EuH
Fig. 14.8. Spectra of three magnetic stars are reproduced, taken with the
Babcock magnetograph, showing the spectra of light with opposite directions of
circular polarization side by side. The shift of the spectral lines due to the
Zeeman effect can be seen especially well for the star 53 Camelopedalis (f).
(From Babcock 1960.)
125
broadened. In addition, we can use very high resolution spectra of the solar
light because the sun is so bright. If another star like the sun had a magnetic
field comparable in strength to the solar field, we would never be able to
measure it. Modern measurements in the infrared do however reveal
magnetic fields of the order of about 100 Gauss for some cool stars. For hot
stars, which means for O and early B stars, strong magnetic fields were not
found by Babcock, because the metallic lines are so weak and there are
hardly any stars which have vT sin i small enough to permit the measurement
of the Zeeman splitting for the metallic lines. The B stars do have, however,
strong hydrogen lines. The polarization measurements in the hydrogen line
wings have revealed strong magnetic fields for some peculiar B stars, the socalled weak and the so-called strong helium line stars, whose helium line
strengths do not match the strengths expected for their colors. The weak
helium line stars are mainly late B type stars, while the strong helium line
stars are early B stars. It appears that the peculiar late B type stars have in the
average magnetic fields comparable to the peculiar A stars (Borra et al.
1983), while for the peculiar early B type stars even stronger fields are found
(see Bohlender et al. 1987). We do not yet know whether there might be
peculiar O stars which have even stronger magnetic fields.
6000 -
5000 -
4000 -
B8
9A0
F0 1
Spectral type
Fig. 14.9. The maximum measured field strengths for stars with magnetic fields
(only a small fraction of stars of any spectral types have magnetic fields) are
plotted as a function of their spectral types according to the data given by
Babcock (1960). The largest fields were measured for early A or late B type
stars. Even larger fields are now found for peculiar early B stars with unusually
strong helium lines.
15
Stars with peculiar spectra
15.1
General discussion
In the previous sections we have discussed stars which are generally

considered to be normal stars, which means their spectra fit into the twodimensional classification scheme according to spectral type and luminosity.
True, the weak-lined stars, or population II stars, do not fit into that scheme,
but generally their peculiarity can be understood by the change of just one
parameter, the ratio of the metal abundance to the hydrogen abundance,
though recently it has been found that this may not always be the case. More
than one parameter may actually be necessary to describe the abundances
of the heavy elements. The population II stars are still generally considered
to be 'normal' stars because we believe that all their peculiarities can
apparently be traced back to different chemical abundances. For the stars we
are going to discuss in this chapter, this does not seem to be the case. There
are, of course, a large number of different kinds of peculiar stars, but we are
not able to discuss all of them in the framework of this introduction to stellar
astrophysics. We shall only discuss the most frequent kinds of peculiar stars
and those which are of special interest in the framework of understanding
stellar structure and evolution.
15.2
Peculiar A stars, or magnetic stars
15.2.1 The observations

In the previous section we saw that some stars with very strong
magnetic fields are found among the early A stars. These stars were
previously known as peculiar A stars or Ap stars because of their peculiar
spectra. They show unusually strong lines of Y, Si, Sr, Cr, and Eu, as well as
other rare earths. These peculiarities are clearly recognizable on low
resolution spectra which are used for classification. In Fig. 15.1 we compare
spectra of peculiar A stars with those of normal A stars. On higher resolution
127
128
15 Stars with peculiar spectra
spectra it becomes obvious that the lines of nearly all the rare earth elements
are stronger than in normal spectra. Spectrum analysis indicates abundances which are increased by factors of up to 1000 for the rare earth
elements. Astronomers found it hard to believe that the rare earth elements,
especially, should be enhanced by such large factors in these stars. They
Fig. 15.1. A comparison of peculiar A star spectra and spectra of normal A

stars is shown. The peculiar A stars show much stronger lines of silicon
(k = 4128-31 A), strontium {X = 4077 and 4215 A), and the rare earths.
(From Morgan et al 1978.)
75.2 Peculiar A stars, or magnetic stars
129
looked for all possible explanations by means of peculiar structures of the

photospheres which might explain the peculiar line intensities, but no
explanation was found.
A peculiar structure of the photospheres was actually suggested by the fact
that the line intensities in these stellar spectra are periodically variable in
time. The chemical abundances can hardly be expected to be variable in time.
If, however, different parts of the stellar surface have different photospheric
structures, perhaps due to the influence of magnetic fields, then because of
stellar rotation, we may see different parts of the surface at different times.
This may explain the variations; that is, if different parts of the stellar surface
have different chemical compositions, we expect, of course, similar variations
in their spectra. The period of line intensity variations must then be the
period of rotation. There was only one difficulty, namely, that some stars
showed different periods of variation for different lines. Some lines showed
periods which were shorter by a factor of two. In Fig. 15.2, we show the line
intensity variations for the lines in the peculiar A star 56 Arietis. We see that
the period for the variation of the line strengths for the helium lines is half as
long as the period for the Fe and Si lines.
In 1950, when Babcock started to measure stellar fields, he found that all
the A stars with magnetic fields were peculiar A stars, and all the Ap stars
+120
km./sec
%
Maximum
0
-0
Sin 128-31
f|oooo
SiJL,e/c.
o_ -
Maximum
o o
o O
"
H
OP
o 08
JO
OS
0-8
Phase
06
Phase
0-8
Fig. 15.2. For the peculiar A star 56 Arietis the line strengths of different lines
are shown as a function of phase. For some lines the period appears to be half
as long as for other lines, for instance, for those of helium. (From Deutsch
1958.)
02
+10
km/sec
0
0-
OS
0-2
OS
Pel
-10
-20
HO
0-2
Fen
.. I
-13
..Crn
EuE
:
Tin
\+20
\
0
0
-20
CrE .
s-20
-Eun
-Gdn
20
Can
Gdn
*
-(0
HO
+20
-23
+0
. .
0-2
Pen
-20
* . '
3$
i-fO
0
HO
-20
-to
0-6
nn
Pel ' '.
Srn
-/J
+20
-0
Can
-10
+/0
Man
i
-20
Mgn
-fO
3
Phase
0
-20
Phase
Fig. 15.3. For the peculiar A star a2 CVn the line strengths, radial velocities, and the magnetic fields are shown as a function of
phase. If a given line strength increases, its radial velocity is negative. For decreasing line strength the radial velocity is positive.
(From Ledoux & Walraven 1958.)
131
whose vT sin i was small enough, and whose lines, therefore, were narrow
enough for the Zeeman effect to be measurable, had strong magnetic fields.
We can, therefore, conclude that all Ap stars have strong magnetic fields. As
we said earlier, the magnetic fields for many of these stars are also
periodically variable, and it was found that the periods agree with the
spectroscopic periods, i.e., with the longer periods of line strength variation.
Radial velocities measured from different lines in the stellar spectra also
turned out to be periodically variable, but different kinds of variation are
observed for different lines. For phases of increasing line strength the radial
velocity is generally negative, which means the material is moving towards
us. For decreasing line strength the material is generally moving away from
us. In Fig. 15.3 we show the observed variations of the line strengths, the
radial velocities, and the magnetic field for the star a2CVn in the correct
phase relations. The observed line intensities and radial velocities are very
nearly what we would expect for an oblique rotator (see next section) if lines
of different elements have different line strengths on different parts of the
stellar surface.
15.2.2 The oblique rotator hypothesis
If we assume that some lines are especially strong on one part of the
stellar surface, then the rotational velocity of that spot is directed toward us
when rotation brings it around the stellar limb into the visible hemisphere
(see Fig. 15.4). When the spot disappears on the opposite limb, the radial
Observer
Observer
Fig. 15.4. If rotation brings a spot with especially large line strength into the
visible hemisphere, i.e., when the lines become stronger, the radial velocity of
that line is directed toward us, which means it is negative. If the spot
disappears on the other limb and the line becomes weaker, the radial velocity is
directed away from us, which means positive.
132
velocity is directed away from us, which means it is positive. With this
oblique rotator hypothesis, we can also understand why some line intensities
have half the period of others. These lines originate in two spots on the star
and, therefore, have two maxima. We call this model the oblique rotator
model because rotation leads to changes in the magnetic field strength only if
the magnetic axis is inclined with respect to the rotation axis (see Fig. 15.5). If
we assume that the magnetic field of the star is similar to a dipole field, then
we can find a reversal of the polarity of the magnetic field, as is measured for
several of the Ap stars: we sometimes look at one pole and sometimes at the
other. If the magnetic axis and the rotation axis were parallel, we would
always have the same aspect of the magnetic field, at least if the magnetic field
was symmetric around the magnetic axis.
Another question is whether the magnetic field is, indeed, similar to a
dipole field. Stibbs calculated that, for a dipole field and the oblique rotator
model, we should always observe sinusoidal variations of the magnetic field.
Babcock's measurements did not show sinusoidal field variations, but new
measurements made with the hydrogen line polarization method suggest
that the variations may, indeed, be sinusoidal and that the measurements of
metallic line Zeeman splitting may lead to somewhat erroneous field
variations because of the changing line strengths on different parts of the
stellar surface which have different magnetic field strengths. The shape of the
magnetic field is, at present, not very well known. One fact, however, seems
to be well established: the polar strengths of the field are not equal at both
poles for many of the stars. It appears that the field consists of at least two
parts: one nearly dipole field and one nearly quadrupole field, as illustrated
in Fig. 15.6. The sum of the two shows different field strengths at both poles
because at the one pole, the dipole and quadrupole fields add up, but at the
Observer
Fig. 15.5. Only if the rotation axis and the magnetic axis are inclined with
respect to each other does the rotation lead to changing aspects of the magnetic
field, and thereby to changing values of the measured magnetic field strength.
133
other pole they have opposite sign and the difference of thefieldstrengths is
observed. With such a field, symmetric around the magnetic axis and an
inclined rotation axis, all the observed variations can be understood quite
well. We conclude that the lines which show half the period lengths of the rest
of the lines originate at both polar regions, while the other lines must be less
strong at the poles and stronger in the equatorial belt, where 'equatorial'
means 'with respect to the magnetic axis.'
The oblique rotator hypothesis can be checked in yet another way. Stars
with short periods of rotation must generally have broad lines unless we
happen to see them nearly pole on (here we mean the rotational poles). Those
with long rotation periods must have narrow lines. If, indeed, the periods for
the line strength variations are the rotation periods, then stars with short
period variations must generally have broad lines, while those with longer
periods should have narrow lines. This test wasfirstdone by Deutsch (1952)
at Mt. Wilson, and the correct relation was found. While this does not
necessarily prove that the oblique rotator hypothesis is correct, it does prove
that this hypothesis is a viable possibility to explain all the observed
variations.
How else can we understand the periodic variations of the magnetic field
with periods of the order of days to weeks? Geologists are talking about
reversals of the Earth's magnetic field, though not within weeks, but
nevertheless it seems to have happened over very long time intervals.
Fig. 15.6. The magnetic field strengths measured by Babcock at the two poles
of a magnetic star are not equal. We then have to assume that the actual field
is the sum of a dipole and a weaker quadrupole whose axis is aligned with that
of a dipole. At the one pole the field strengths add up, while at the other they
are of opposite polarity and, therefore, only the difference in field strength is
observed. The two poles then are observed to have different field strengths.
134
We also know of reversals of the general solar magnetic field which happen
about every 11 years, not within a week either, but slowly and irregularly
over several years. Could that also happen with the magnetic stars?
In order to answer this question, we have to remember that the general
magnetic field of the Earth, and also of the sun, is of the order of one Gauss.
We also know that matter and magnetic fields can only move together, but
which controls the motion, the matter or the magnetic field? We can judge
that when we compare the magnetic energy of the field, which is a measure of
the resistance of the field to any changes, with the kinetic energy of the gas. It
is easiest to compare the magnetic pressure PM with the gas pressure P g .
In a homogeneous magnetic field, the magnetic pressure is
where H is the magnetic field strength.

For a field H of one Gauss, the magnetic pressure, and therefore the
magnetic forces, are very small. In the solar atmosphere, the gas pressure P g
is on the order of 105 dyncm" 2 . The kinetic energy of the matter is much
larger than the magnetic energy. The material controls the motion and the
changes in the magnetic field, at least at the surface. For a field of 1000 Gauss,
the situation changes. We now find PM = 10 6 /8TI ~ 105. Now, magnetic
energy and kinetic energy are comparable. In sunspots, the magnetic field
determines the structure. In A stars, the gas pressure in the atmosphere is
only 10 3 10 4 dyncm" 2 (the absorption coefficient is so large that we only
see the top layers where the gas pressure is low). For a field of several 1000
Gauss, the magnetic energy is much larger than the kinetic energy. The
motion of the material is not able to control the structure of the magnetic
field at least not at the stellar surface.
Nevertheless, let us consider one possibility of varying magnetic field
strength which has been discussed in the literature. In order for the magnetic
field to increase in strength, the material has to squeeze the field lines
together (see Fig. 15.7). On a magnetic star, one can imagine that the field
strength increases when the equatorial material moves toward the poles,
which would give an increase in the field strength near the poles and a
decrease near the equator (see Fig. 15.8). However, we see from Fig. 15.8
that, while we can get a change in the magnetic field in this way, we can never
get a reversal of polarity, as is observed. While this was first discussed as one
possibility to explain the magnetic field variations, most astronomers are
now convinced that this mechanism does not work.
The other possibility is, then, an oblique rotator (see Fig. 15.5). If the axis of
135
rotation is inclined close to 90 and the star has a dipole field, the observer
sees a sinusoidal variation of the magnetic field with changing polarity. The
period of variation is the period of rotation. If some spectral lines are
stronger at the magnetic poles, we expect two maxima of line strengths
during one magnetic period.
But why are some lines stronger at the poles and some stronger at the
equator? In the beginning, it was discussed whether nuclear reactions could
take place in the polar regions, but the observed abundance ratios do not
agree with those expected from nuclear reactions. There are also difficulties
in accelerating particles to energies high enough for nuclear reactions: In
order to obtain high velocities the densities have to be low enough so that
collisions do not occur, but at such low densities, nuclear reactions do not
occur either. There is also the peculiar observation that the enhancement of
line strengths depends on the effective temperature of the stars. For the
hotter stars, we see strong Si lines; the cooler stars have strong Eu, Sr, and Cr
lines. It would be peculiar if the nuclear reactions were dependent on the
effective temperature of the star rather than on the magneticfieldstrength of
the star. Diffusion is considered to be the only possible explanation for the
Fig. 15.7. The magnetic field strengths in stellar material can be increased if the
material contracts.
Fig. 15.8. If the equatorial material of the star were to move toward the poles,
the magnetic field strength at the poles would increase. It would decrease at the
equator. Figure (a) shows the original star and its field and (b) the shape and
the field of the star after the material has moved from the equator to the poles.
136
peculiar abundances in the photospheres of the peculiar A stars, though

there are still some problems if one wants to explain the observed
abundances in detail.
15.3
Diffusion in stars
Generally, we know diffusion as a slow process of moving a

substance to another place by molecular motion. Light particles can diffuse
through a membrane into a mixture with heavier particles, or a gas of one
kind can mix with a large volume of a different gas by diffusion which
actually means by thermal motion. When we think about a star we usually
think about a well-mixed gas ball. In the case of the Ap stars we do not
actually think about mixing but rather about demixing or sedimentation. In
nature that will never happen without an external force. In stars, the only
forces appear to be gas pressure and gravitation. Under the influence of
gravitation, we find a stratification of the gas pressure. On a high mountain,
the gas pressure is much lower than at sea level. We can calculate how the gas
pressure changes with height in the atmosphere if we consider the
equilibrium between pressure and gravitation. At each level the weight of the
overlying material has to be balanced by the gas pressure P g , (see Fig. 15.9).
At the height h this means
P (h) = mass {h)-g
(15.2)
where mass (h) is the mass of the material above the height /z, and g is the
gravitational acceleration. At the height h + Ah we must have accordingly
PJh + Ah) = mass (h + Ah)-g
(15.3)
1 cm 2
PAh + Ah)
m = p-Ah
Fig. 15.9. The pressure at the height h has to balance the weight of the
overlying material, which means it has to balance the weight of the mass
column of 1 cm 2 cross-section above the height h, which is m(h). The pressure at
the height h + Ah must balance the mass column above h + Ah, which is
m(h 4- Ah). The difference in weight of the two mass columns is given by the
weight of the mass in the volume element of 1 cm 2 cross-section and height Ah.
This mass is given by p-Ah.
15.3 Diffusion in stars
137
where Ah is the height of the volume element under consideration. The mass
(h) is larger than the mass (h + Ah) by the mass contained in the volume
element. If this volume element has a cross-section of 1 cm2 and if the density
of the material is p then the mass in this volume element is given by
mass (h) mass (h + Ah) = p-Ah
(15.4)
We can now express the gas pressure at the height h + Ah by means of the
first terms of a Taylor expansion around the height h and obtain
Pg(h + Ah) = Pg(h) + ^
Ah = mass(/z + Ah)-g
(15.5)
and in the same way we find

mass(/z + Ah) = mass(/z) H-rrAh = mass(/i) 4- pAh
(15.4a)
Subtracting (15.2) from (15.5), making use of (15.4a) we derive

or -JT= -p-g= - ^ V # >
(15.6)
if we replace p = (Pgin/RgT), where // = molecular weight of the gas,

Rg = gas constant = 8.315 x 107 [cgs].
Equation (15.6) tells us that P g decreases with increasing height, and
decreases faster if ja is larger. We can see this even better if we integrate (15.6).
Division by P g yields
Pgdh
dh
RJ'
If the right side is a constant, which means for an isothermal atmosphere,

we can easily integrate both sides of this equation over dh between the
limits h = h0 and h = h and find
UQ lh
= --^-'h\
J\. 1
and
Ch d l n P
, since
ho
"I*
gdff = lnPg
dfi
(15.8)
<=h-hn.
If we insert these limiting values, we find

In PJh)-\nPJho)= --^F(h-h0)
=lnp^.
(15.9)
138
Taking the exponential yields

,
or
(15.10)
where we have abbreviated (RgT/fig) by //, which is called the scale height. If
(h ho) = H, then the exponent in the exponential is 1. The pressure
decreases by a factor e over one scale height. The scale height H is smaller if \i
is larger. Heavy particles form an atmosphere of smaller height than
lightweight particles (see Fig. 15.10).
It turns out that in a mixture, each group of particles tries to set up its own
equilibrium. If you make an emulsion of heavier particles in water, the heavy
particles will sink. If you have a mixture of air with N 2 (/i = 28), CO 2 (n = 44),
and H2O(pi= 18), then H 2 O will become more abundant in the higher
atmospheric layers, provided there are no atmospheric currents or winds
which keep the atmosphere well-mixed.
The stellar material consists of//, He, and heavier particles, and very light
electrons. If there were no mixing, the electrons would like to float on top.
This would, however, generate a charge separation, creating an electric field
which prevents further charge separation (see Fig. 15.11). The electrons and
ions are kept together by the electric forces. But there is no force which
inhibits the falling of the still heavier ions with respect to the protons. The
helium atoms and ions may well fall with respect to the hydrogen atoms and
ions, but the density in stars below the surface is so large that there are very
many collisions which prevent the particles from falling or diffusing
downward fast. It would take more than the lifetime of the stars to produce
any measurable separation over the radius of the star, even for the still
heavier ions. This was already discussed by Eddington (1926).
/
<>
G
o
0
o
O
>- ^
'A\:;\'::.
/. . .
oo oo /
0
(a)
(b)
Fig. 15.10. In an atmosphere with heavy particles (a) the pressure decreases
faster with height than in an atmosphere with light particles. In an atmosphere
consisting of heavy and light particles (b\ the heavy particles stay close to the
bottom, the light ones extend to greater heights.
15.3 Diffusion in stars
139
To see deficiencies at the surface does not, however, take as long as

diffusion to the center. The layers which we see are only a few hundred km
thick. Diffusion can go rather fast in these layers where 'fast' means about 105
years, provided there is no mixing mechanism. If measured line widths
indicate turbulent motions, then diffusion cannot work.
Also, we are talking about Ap stars, which show an enrichment, not a
depletion, of heavy elements like the rare earths and Si, Cr, and Eu at the
surface. Why are these elements enriched?
There is another diffusion process which opposes the gravitational
settling. This process is called radiative diffusion, and the force causing this
diffusion is radiation pressure.
In a star, energy is transported outwards from the inside by radiation. This
means that there is more radiation coming from below than from above
(Fig. 15.12), otherwise there would be no net flux to the surface and to the
outside (see Fig. 15.13(a)). The atoms in the atmosphere absorb and re-emit
the photons. In each absorption process, the photon transmits some
momentum to the absorbing atom or ion. The photon will generally be reemitted in another direction (see Fig. 15.13(b)). In fact, there is no preferred
direction for the re-emission. Therefore, in the re-emission the net effect of all
the pushes is zero. But this is not the case for the absorption because there is
more radiation coming from below than from above, so there is a net push to
the surface. The atoms and ions which absorb and re-emit most, receive the
biggest push. The rare earth elements have a very complicated energy-level
Fig. 15.11. In an atmosphere with free electrons and protons, the electrons
would like to diffuse to the top. This would give a charge separation, yielding
an electric field, which prevents further separation.
Top of
atmosphere
Fig. 15.12. In order to transport radiative energy from the inside of the star to
the surface, more radiation has to flow in the outward than in the inward
direction, / " i s the downward going intensity, and J + the outgoing intensity.
140
diagram; they have, therefore, very many lines in the spectral region where
there is a large flux coming out of the star. Ca + has very few lines because it
has only one electron in its outer shell. It has more lines in the far ultraviolet
but there are no photons to absorb. Therefore, the Ca + can sink while the
rare earths do not.
Also, He and He + have only a few lines because they have only one or two
electrons and will, therefore, tend to sink. Helium appears to be depleted in
the peculiar A stars. The radiative diffusion can, indeed, be different in stars
with different effective temperatures because the degree of ionization
changes and the radiation field changes with Teff.
A remaining problem is the convection zone below the surface of the Ap
stars which is not important otherwise, but which could keep the top layers
well-mixed. Probably the magnetic field is strong enough to prevent these
motions. In fact, this is supposed to be the reason why diffusion works in
magnetic stars but not in non-magnetic A stars.
Diffusion of charged particles is difficult in a direction perpendicular to the
magnetic field lines, but easy along the direction of the field lines. We may,
therefore, expect a larger enrichment of the upward diffusing elements in
both polar regions. This explains the different element abundances on
different parts of the magnetic stars.
15.4
Metallic line stars, or Am stars
The most abundant peculiar stars are called metallic line stars.
About one-third of all the late A or early F stars belong to this group. As their
name says, the lines of most of the heavy elements are too strong in
comparison with their hydrogen line strength. On the other hand, the
usually very strong lines of the calcium ion Ca + , the lines at 3933 and
Push = hv/c
Push =
Photon
momentum =
hv/c
(a)
(b)
Fig. 15.13. When the atom or ion absorbs a photon from below, it receives a
push to the surface (a). When the photon is re-emitted in another direction, the
atom or ion receives a push in a direction different from that during the
absorption (b).
15.4 Metallic line stars, or Am stars
141
3968 A, the H and K lines, are too weak in these stars (see Fig. 15.14, where
we compare a spectrum of an Am star with those of normal A and F stars). In
Fig. 15.15 we show schematically for 'normal' stars the dependence of the
hydrogen line strengths on the spectral types, the general dependence of the
metallic line strength on the spectral types, and also the dependence of the H
and K line strengths on the spectral types. While the strengths of the metallic
lines in general increase for cooler stars, the H and K line strengths decrease
20CVn
F3 m
63Tou
kA2, mF3I
HR1279
F3Y
Fig. 15.14. The spectrum of the Am star, 63 Tau, is compared with spectra of
normal A and F stars. In the metallic line star the metallic lines look like those
of an F star, but the hydrogen lines and the calcium lines look like those of A
stars. (From Morgan etal. 1978.)
Line
strength
Metallic
lines
Hydrogen
lines
lines
Fig. 15.15. The dependence of the hydrogen, calcium and other 'metallic' line
strengths on spectral types for 'normal' stars is shown schematically. For the
spectral type indicated by the H lines in an Am star the metallic lines are too
strong, or for spectral type indicated by the metallic lines the hydrogen lines
are too strong. The Ca+ K lines are too weak for any of the other spectral
types.
142
for lower temperatures because of the decreasing degree of ionization. In the

cooler stars, the lines of the neutral Ca become stronger. If in the metallic line
stars the Ca + lines, also called the Ca II lines, are too weak, then the spectral
type derived from the strengths of these lines will be too early, and we would
think the stars are rather hot. On the other hand, if we look at the Fe lines,
which appear too strong, we would think the stars are cooler than indicated
by the hydrogen lines. If we determine their spectral type according to the
metallic line strengths, they would be called late A or early F stars. However,
if we classify them according to the Ca I I H and K lines at 3933 and 3968 A,
they would be classified as early A stars, i.e., AO to A3. The strengths of the
Hydrogen lines would indicate a spectral type, say, around A5, somewhere
intermediate between the metallic line and the Ca II line spectral types.
Which spectral type tells the correct temperature?
We discussed previously that the line strength is determined by the ratio
/c(line)//c(cont). For A stars, the continuum absorption is due to hydrogen; the
hydrogen line absorption is also due to hydrogen. It is very difficult to
envisage any mechanism which could reduce the hydrogen line strengths
except a lower density, which would require the stars to be giants. But many
of the Am stars are members of clusters: the Hyades cluster, for instance.
Their brightness shows that they are not giants even though the ionization
equilibrium of iron indicates a somewhat reduced pressure in the atmospheres of these stars.
The simplest way to explain the peculiar line strengths of the metallic lines
would be to say that these stars have a higher abundance of the metals, but
they must then have a lower abundance of Ca. It turns out that the scandium
lines are also weak, as are sometimes, the carbon lines and those of some
other elements. The lines of the rare earth elements are rather strong,
compared to the Fe lines. We then must ask: what can cause these peculiar
element abundances?
15.4.1 The binary nature of Am stars
These Am stars have some other peculiarities. When a careful study
of their radial velocities was made, it turned out that they all have
periodically variable radial velocities, which means they are all binaries, and
all rather close binaries, with periods of the order of days to weeks. (For
'normal' stars, about 30% of all are binaries). What does that have to do with
their peculiar abundances? How short can the periods of binaries be without
the stars touching each other? For the stars not to touch, the distance d must
15.4 Metallic line stars, or Am stars
143
be larger than ~ 2R*. For Am stars, R ~ 2Re. This means

d>4Ro
A
1 au
200 so
If we measure all quantities in solar units, we know that Kepler's third law
reads
3
where p is the period, and M x and M 2 are the masses of the two stars. With
M1 + M2 ~ 3M O and d = 1/50 au, we find for the period in years,
L
^^ \50/
3 r^ 3
125 '"^ 4^
VCclIo
or P ^ 1.5 x 10" 3 years, or roughly P ^ 1 day.

The Am stars have periods longer than that, but not much longer. They
must have strong tidal interaction, but are not contact binaries.
The Am stars are also peculiar in another respect, as we will discuss in the
next section.
15.4.2 Rotational velocities of Am stars
As we saw in Chapter 12 we can measure the rotational velocities of
stars from the widths of their spectral lines. Rotational velocities for many
stars have been determined in that way. In Figs 13.5 and 13.6 we have shown
the dependence of the average rotational velocities on the spectral types. The
rotational velocities decrease rapidly for stars cooler than spectral type F5. It
is believed that this decrease of rotational velocity is due to the development
of a stronger stellar wind, like the solar wind, which causes the braking of
rotation.
The Am stars occur in the region of spectral classes where the rotational
velocities are generally still high, but the Am stars have much smaller
rotational velocities than the normal stars.
What does all this have to do with the stronger metallic lines and with the
apparent over-abundance of heavy elements?
Most astronomers believe that the peculiar abundances observed in the
Am stars are also due to diffusion processes even though there are still some
severe problems with that explanation: if Ca is depleted due to the sinking of
Ca + in the layer where calcium is mainly ionized once, why is not Mg also
144
depleted? Its atomic structure is very similar to Ca, yet the Mg abundance is
normal in the Am stars.
The other question is, of course, why should diffusion work in Am stars
and not in normal stars? Why are all the Am stars in binaries and why are
they slow rotators? The reason is again that diffusion can only work if there
are no mixing processes which prevent it. Normal stars of the same
temperature as the Am stars would still be rapid rotators. Since the studies of
Eddington, it is well known that rapidly rotating stars cannot be in a static
equilibrium, but that slow currents, the so-called Eddington currents,
develop which carry material upward in the polar regions and downward in
the equatorial region. These currents can keep the star well-mixed.
Therefore, diffusion can only work for slowly rotating stars. What makes the
Am stars rotate slowly? Their binary nature - they are all members of
rather close binaries, for which tidal interaction is very important. Tidal
interaction slows down rotation until the rotational period equals the
orbital period.* For an orbital period of one week, this would mean an
equatorial rotational velocity of about 15km/s for the Am stars, which
would be considered a slow rotation. The binary nature then causes the
slow rotation which in turn may permit diffusion to work.
15.5
Barium stars
There is another group of stars with peculiar element abundances,

called barium stars. These have gained interest during recent years even
though they are relatively rare stars. They are certainly not made out of
barium, but show unusually strong lines of barium and strontium (see Fig.
15.16). In Table 15.1, we list the element abundances as determined for some
of these stars. They show a general increase of abundances for elements
which are made by the so-called slow neutron capture process. This means
that the atomic nuclei of these elements were originally formed by adding
neutrons to the atomic nuclei in a process called neutron capture. Neutron
rich atomic nuclei become unstable to the so-called ft decay, which means
they like to emit an electron thereby transforming a neutron into a proton.
Very neutron rich atomic nuclei cannot bind any more neutrons, they must
wait until one of the captured neutrons has emitted an electron before it can
capture another neutron. Due to the decay the number of surplus neutrons is
reduced and an additional neutron can be bound. The transformation of a
neutron into a proton can occur for different numbers of surplus neutrons,
* Our moon rotates once in 27 days, which means during its orbital period. We therefore always see
the same side of the moon.
145
15.5 Barium stars
depending on the available number of neutrons to be captured. If neutrons

are added very rapidly in a very neutron rich environment, then there is no
time for the less neutron rich nucleus to decay before the next neutron is
added. We call this the 'fast' neutron capture process. If fewer neutrons are
around then neutron captures are less frequent and the less neutron rich
nucleus has time to emit an electron before capturing the next neutron. This
is called the 'slow' neutron capture process. The resulting atomic nuclei
which are formed by these different neutron capture processes are different.
From the observed abundances we can thus see which of the processes has
taken place. For the barium stars we see the products of the slow neutron
/? SCT
t CAP
HD 211594
HR 2 392
HH 5 0 5 8
HR
774
v OPH
Bo I
Sri
Fig. 15.16. The spectra of five barium stars are compared with the spectra of
the 'normal' stars /? Scuti and v Ophiuchi. In the Ba stars the Ball line at
4554 A is much stronger than in the normal stars and also the SrII line at
4078 A is stronger. (From Bidelman & Keenan 1951.)
Table 15.1 Abundances1 of slow neutron capture

elements in two barium stars and in the sun.
Element
logN
Solar
HR774
HD 44896
A2
Sr
Y
Zr
Ba
La
Eu
Gd
H
2.94
2.23
2.59
2.20
1.22
0.55
1.09
12.0
4.22
3.34
3.59
3.28
2.30
1.32
2.04
12.0
3.79
3.44
3.45
3.07
2.27
1.07
1.80
12.0
1.06
1.16
0.97
0.97
1.06
0.65
0.83
0.0
1. The numbers given are the logarithms of the number of atoms

plus ions in the amount of material which contains 1012
hydrogen atoms and ions.
2. A = log N o - logN (Ba star).
146
capture process. This generally means that between successive neutron

captures there was a timespan of roughly 104 years. There must have been a
weak neutron source present for at least that length of time in order to make
the slow neutron capture process work. Such slow neutron capture processes
are believed to be possible only in the interior of stars which are in very
advanced states of their evolution. At that time they are supposed to be very
bright red giants, as we shall see in Volume 3. The barium stars are not that
bright, though they are red giants. It has been a puzzle for many years how
these stars could have manufactured these slow neutron capture elements
and, in addition, if they did so, how they could have mixed the processed
material to the surface where we can observe them. There is presently no
deep mixing.
The problem appeared in a different light when McClure, Fletcher, and
Nemec (1983) discovered that all the barium stars showed radial velocity
variations indicative of their binary nature. From the radial velocity
variations of some of the barium stars, they concluded that the masses of the
companions were of the order of one solar mass. Since the companion
spectra could not be detected, they speculated that the companions might be
white dwarfs. Shortly thereafter, white dwarf companions of the brightest
barium stars, Cap and Cyg, were detected by means of the ultraviolet
observations with the IUE (International Ultraviolet Explorer) satellite.
Cyg is a so-called mild barium star which shows only a small enhancement of
the barium line intensity. An additional white dwarf companion was
detected for the mild barium star } Cet. For other nearby barium stars, the
presence of a white dwarf companion appears probable because emission
lines indicative of a rather hot gas are seen. No main sequence companions
with masses comparable to the masses of the barium stars, about two solar
masses, could be detected, though they should be visible in the ultraviolet if
present. This supports the suspicion that all the barium star companions are,
indeed, white dwarfs. If the companions were mainly normal stars, the
chances offindingstars of almost the same mass as the barium stars would be
rather large since statistics show that about half of all binaries have two
components with about equal masses, to within 50%. Since we do not find
companions which are normal stars we conclude that all barium star
companions are white dwarfs.
As we shall see in Volume 3, white dwarfs are believed to be thefinaldeath
stages of stellar evolution. They supposedly form when a star blows off its
outer envelop at the state of stellar evolution when it is a very luminous red
giant. The exact cause of this mass loss is presently still unknown, but we
know that it must happen. The reasons will be discussed in our stellar
evolution discussions.
15.6 TTauri stars
147
If, indeed, all barium stars have white dwarf companions, then this must be
a necessary condition for the formation of these stars. The presence of the
white dwarf must be essential to the peculiar element abundances. It seems
likely that the slow neutron capture process did actually take place in the
interior of the progenitor star of the present white dwarf, which must have
been a very luminous red giant before it lost its envelope and became a white
dwarf. The barium star probably accreted a major part of the envelope of its
companion when it was blown off. We probably see the material manufactured in the interior of the white dwarf progenitor displayed now on the
surface of the present barium star. There are still many studies needed before
we can be sure that this is actually what happened, but these observations
open the exciting possibility that we can actually study, on the surface of the
barium star, what happened in the interior of the white dwarf progenitor.
15.6
T Tauri stars
The so-called T Tauri stars are named after their prototype. The T in
front of the constellation indicates that the stars are variable, they are in fact
irregularly variable. This type of star is defined by the peculiar spectra which
show strong emission lines of hydrogen and of the Ca + , H and K lines at
3933 A and 3968 A (see Fig. 15.17). They frequently also show emission
!!
i ii M i
B IIIIIJUJ
M I ' M l i 1 U>
mniiiiiii^n tn
m i ti nt M it if
i l l It I I MM
\[\ rawiiittil in
II it HI1JM Ml
I
I
t It II
I iIII
I II It
I!
IE iniltlfl
Mi
in ii iiitiiii in Mi i
utt
III IHMI! Ill

IN JNfHtJin
Ml SMMilJil
iiiiiHinniiiii
mil mini i
III
nil
"HI
\\\\
ii
j
ii
f
ii
ii
IIII
mi
III!
IN:
in
lint n t i fi if
t
[in
hill II It
mil ii n
IfIlllH ill
ijiiiiii HI
mini! Mi
-v
t
f
ill SI i l t f f t l Itf
II II riMl HI Ml
iiiiiin ii ii itn
i:
in
! II
i i n m i <ii
! _ J M I!
HI !i it n
i l l ! II
E
Ml
i t i n
II! i
ill
;
111 i i
niiitiii
nmi 11
fi u IIIMIi m
IIIJ
i i int
K H
i! ii mm m
HS
4233
Hy
4563
H$
Fig. 15.17. Negative T Tauri spectra (the narrow horizontal lines) are shown for
different T Tauri stars. The strong emission lines of the Ca+ ion, the H and K
lines, are seen (dark blobs on the left). On both sides of the stellar spectra are
laboratory spectra, such that wavelengths can be measured for line
identification. (From Joy 1960.)
148
lines of other elements like iron or sulfur. The particular lines of these latter
elements are not usually seen in normal stellar spectra but are lines which are
generally seen in objects of very low density like gaseous nebulae. They are
the so-called forbidden lines indicating low density material around these
stars.
As judged from the absorption line spectra also seen in these T Tauri stars
they are of spectral types F to M which means they have effective
temperatures between 7000 and 3500 K. As discovered by ultraviolet
obserations by means of the IUE satellite these stars show very strong
chromospheric and transition layer emission lines much stronger than those
of the sun (see Chapter 18).
These T Tauri stars are especially interesting because they appear to be
very young stars as demonstrated by the fact that they appear only in regions
of the sky where we also see hot O and B stars, which we know cannot live
longer than about 106 years. The T Tauri stars always appear in dense
interstellar gas clouds which we think are the birth places for stars. If such a
gas cloud is large and dense enough the gravitational force can pull it
together to form a star. The association of these stars with young O and B
main sequence stars whose absolute magnitudes we know and for which we
can therefore determine the distances (see Chapter 5) makes it possible for us
to also get approximate distances to the T Tauri stars, though the absorption
by the gas in the cloud surrounding the star introduces some uncertainty,
(see Chapter 19). It turns out that the T Tauri stars are much more luminous
than main sequence stars. In Fig. 15.18 we show the position of the T Tauri
stars in the color magnitude diagram. The stars are still much larger than
main sequence stars. From their association with the young O and B stars we
know they must be very young. They must still be contracting finally to
become main sequence stars.
From the study of these young, still contracting T Tauri stars we can get
interesting information about our sun at an evolutionary stage when it was
still contracting and when the solar system, including the planets and the
earth was forming.
A more detailed discussion of star formation will be given in Volume 3.
What is of interest to us here are the more or less violent phenomena which
seem to occur in connection with the birth of a star. For many T Tauri stars
we see for instance P Cygni line profiles (see Chapter 17, Fig. 17.9)
indicating that material is flowing out from these stars, while for a newly
forming star we would naively have expected that matter is falling inward. It
appears that newly formed stars develop very thick chromospheres, coronae
and stellar winds, though X-ray emission from the coronae is not observed.
15.6 TTauri stars
149
We do not know whether the coronae do not become very hot because of the
large energy loss in the strong winds or whether the material in the dense gas
cloud in which the star is embedded absorbs all the X-rays.
The T Tauri stars also show collimated, high velocity outflows with
velocities of 300-400 km/s, so-called jets (see Fig. 15.19) going in opposite
directions, so-called bipolar outflows, which in their extension often show
Herbig-Haro objects, named after the astronomers who first discovered and
discussed them. These Herbig-Haro objects change their appearance in the
course of a few years, (see Fig. 15.20). It was therefore at some time suggested
that they were actually newly formed stars and that we could witness directly
the birth of new stars. The masses of these objects are, however, much too
small. The amazing thing about the Herbig-Haro objects is their fast motion
away from the central source, always believed to be a T Tauri star. The
Fig. 15.18. The positions of the T Tauri stars in the color magnitude diagram
are shown. They all lie above the main sequence, which is the boundary with
the numbers on the left hand side. They must, therefore, be larger than main
sequence stars. The solid lines are evolutionary tracks (see Volume 3) for stars
of different masses as given by the numbers (in units of solar masses). (From
Cohen & Kuhi 1979.)
40
Aa
(a)
20
(arc-sec)
Fig. 15.19. A photograph of the Herbig-Haro objects HH 1 and HH2 is shown

on the left {a). (This is again a negative, the dark spots are bright.) Their proper
motions (corresponding roughly to 250km/s) point radially away from a source
which is about halfway in between HH 1 and HH 2. It can only be seen in the
infrared because all the other light is absorbed by the interstellar dust (see
Chapter 19) around it. In the right-hand picture (b) the directions of proper
motions are shown. The lengths of the arrows are proportional to the measured
velocities. (From Schwartz 1983. Originals from Herbig & Jones 1981.)
t
1947
1954
Fig. 15.20. Changes in the appearance of the Herbig-Haro object no. 2 are
seen. In 1947 only three major blobs are seen; in 1954 five can be recognized.
Two new ones have appeared in the course of seven years. (From Herbig 1969.)
15.6 TTauri stars
151
kinetic energy emitted in these clumps of matter corresponds to the energy

emitted by the sun during a whole year. It is not at all clear what the energy
source is for these jets.
If the original gas cloud from which the star formed, had some angular
momentum, we might expect a disk to form in a star forming collapse. We
might then understand why a stellar wind can only flow out in directions
perpendicular to the plane of the disk and form a jet. It is still surprising that
the jet keeps so narrowly focussed, the opening is about 10 degrees, over
distances of the order of 1 pc, and the energy source for the jet is still a puzzle.
The spectra of the Herbig-Haro objects can be explained by a spectrum of
a shock, forming either when a high density, high velocity gas ball, a so-called
interstellar bullet, hits a less dense gas cloud at rest, or forming when a high
velocity, low density jet hits a clump of matter.
16
Pulsating stars
16.1
The different types of pulsating stars
The pulsating stars (not to be confused with pulsars, which are

rotating not pulsating), are recognized by their periodic changes in
brightness accompanied by periodic variations of their radial velocities. We
distinguish several classes of pulsating stars. The most frequent type are the d
Cephei stars, which have periods of light variation of a few days to a few
weeks. Another very common type is the RR Lyrae star, named after the
first-discovered variable of this kind. The RR Lyrae stars vary with periods
of the order of half a day. Other types of short period variables are the 5 Scuti
stars, which are population I variables like the longer period S Cephei stars.
The 3 Cephei stars are generally of spectral types F and G, and are
supergiants, while the S Scuti stars are A stars of luminosity class IV to V.
The amplitudes of the 3 Cephei variables are of the order of one magnitude,
i.e., the light output changes by a factor of 2 or 3, while the light variations of
the 8 Scuti stars are so small that they are hard to detect.
The RR Lyrae stars are frequently found in globular clusters; therefore,
they belong to population II. The amplitudes of their light variations are also
of the order of one magnitude, and can therefore be recognized fairly easily.
The radial velocity amplitudes for the 3 Cephei stars are several tens of
kilometers per second. Those of RR Lyrae stars are of comparable values,
though somewhat smaller. In Fig. 16.1 we reproduce the light and velocity
curves of the prototype, namely 8 Cephei.
There are also population II star analogues to the 8 Cephei stars, called the
W Vir stars, again named after the first star discovered of this type.* These
*We should mention here, perhaps, that variable stars are named after the constellation in
which they are seen. The first one discovered in the constellation of Virgo would be called A
Vir, the second one B Vir. After all the letters have been used, one would continue with AA Vir,
AB Vir, etc. Historically it developed though that the first ones discovered were actually
designated with letters occurring late in the alphabet.
153
154
16 Pulsating stars
stars frequently show very broad light maxima. In Fig. 16.2 we show the light
curve of the population II Cepheid, W Vir.
Another group of intrinsically variable stars are the long period variables
with periods of the order of several years. They are very luminous red giants
and are of spectral type M. Their light variations are not quite as regular as
those of the other variables discussed so far.
Even less regular are the light variations of the so-called RV Tau stars
which, in the HR diagram, are placed between the Cepheids and the long
period variables. In Fig. 16.3 we show the positions of all the different kinds
of intrinsically variable stars in the HR diagram.
The most frequent types of variables are found in a narrow band cutting
diagonally through the HR diagram along the so-called Cepheid instability
strip. All stars inside this band in the HR diagram are pulsating stars. In
Volume 3 we will explain why.
These pulsating stars actually reveal many properties of the stellar
"1
-2.8
-2.6
-2.4
%v
-2.2
1
0.0
0.5
1.0
-10 km/s
-20 -
Fig. 16.1. The light and velocity curves for the pulsating star S Cephei are
shown as a function of phase in the pulsational cycle. For this star the period
of variation is 5.4 days. (From Cox 1980.)
155
16.1 The different types of pulsating stars
interiors to us. They serve to check our ideas about the interior structure of
the stars and about stellar evolution. They are, therefore, very important to
the theory of stellar structure and have been studied extensively.
So far we have mentioned only stars which pulsate in a spherically
symmetric way, the surface moving in and out. There are other types of
variables which pulsate in a non-spherically symmetric way, i.e., the polar
regions may be moving inwards while the equatorial regions move
outwards. More complicated patterns are also possible. Spherically symmetric pulsations may be superimposed. The most prominent stars of this
kind may be the /? Cephei stars. They are B type stars and are not positioned
in the Cephei instability strip. The reasons for their pulsational instability is
presently unknown.
There is also a group of white dwarfs, the ZZ Ceti stars, which are
pulsationally unstable and show a very complicated pattern of pulsation
consisting of the superposition of pulsations of many frequencies. They are
very helpful for studying the interior structure and chemical composition of
these white dwarfs.
TR
A.SVSO
9.8 :
10.0
102
10.6
P-V
10.8
y
Vm/sec
-to
/
i
-50
-60 -70
*>
on
SU
-90
-WO
0.7
0.8 09 0.0 OJ 0.2 03 0.<t 0.5 0.6 0.7 0.8 03

phase
Fig. 16.2. The light curve of the population II Cepheid W Virginis is shown. In
comparison with the light curve of S Cephei it shows a very flat top. Such flattopped light curves can be well distinguished from those of population I
Cepheids, but not all population II Cepheids have flat-topped light curves. The
velocity curve is shown at the bottom. The open circles indicate velocities from
emission lines. For phases between 0.9 and 0.6 it appears that two absorption line
components are seen. (From Ledoux & Walraven 1958.)
156
16 Pulsating stars
In the following sections, we will describe the major types of pulsating

stars, the 6 Cephei stars and the RR Lyrae stars, in somewhat more detail.
16.2
The 8 Cephei stars
Because of their periodic light and radial velocity variations,

astronomers first thought that the d Cephei stars were eclipsing binaries. In
Fig. 16.4 we show again the light and velocity variations observed for d
Cephei to be compared with those observed for an eclipsing binary (see
Fig. 9.6). The non-sinusoidal velocity curve for S Cephei would suggest a
very elliptical orbit (see Fig. 9.8). The second eclipse would have to be
invisible. The stars would have to be very close companions because there
is no constant light phase between eclipses. Also, the periods are rather
short, only a few days.
When astronomers learned how to determine the luminosities of the stars
-6
/RED SEMI/ REGULARS
-4
-2
0
Mv,
DWARF
CEPHEID
10
12
BO
(27,000)
AO
(10,400)
FO
(7200)
GO
(6000)
KO
(5120)
MO
(3750)
SPECTRAL TYPE
Fig. 16.3. The positions of the different kinds of pulsating stars in the color
magnitude diagram are shown. The main groups of pulsating stars fall along a
diagonal sequence, the so-called Cepheid instability strip. The long period Mira
variables, the irregularly variable RV Tauri stars and, surprisingly, also the /?
Cephei stars, fall essentially along a constant luminosity strip close to the top
of the diagram. The pulsation mechanisms for all these latter stars are not
yet understood. (Adapted from Cox 1980).
16.2 The 5 Cephei stars
157
from their spectral features, it became obvious that the d Cephei stars are
supergiants. This means that the stellar radii are actually larger than were
the orbital radii calculated for the stars when interpreting them as binaries. It
then became clear that these stars could not be eclipsing binaries, but had to
be intrinsically variable stars.
The variable radial velocities indicate that the surface is moving outwards
when the light increases and it is falling inwards when we see minimum light.
We can directly measure the change in radius AR(t) by calculating
-r- = v and AR(t) =
vdt.
to
In fact, we can also calculate the radius itself if we compare two phases of
-0.2
0.2
0M
0.6
Phase
0.8
1.0
1.2
Fig. 16.4. The observed variations of the apparent visual magnitudes, the
effective temperatures, the spectral types, and the radial velocities for 3 Cephei
are shown as functions of phase. At the bottom the changes in the radius are
shown, which can be obtained by integrating over the pulsational radial
velocities (the stellar velocity has to be subtracted from the observed velocities).
It can be seen that the radius is nearly the same for maximum and minimum
brightness. (From W. Becker 1950.)
158
16 Pulsating stars
equal temperature. The difference in bolometric magnitude is given by

Am bol = - 2 . 5 - l o g T 7 ^ = m bol (f 1 )-m bol (t 2 )
(16.1)
and
J?2(t \
R(t\
(16.2)
The Ambol can be measured; in fact, Ambol = Amv since, for equal temperature,
the bolometric correction is nearly the same. The change with gravity is
minute. We can then calculate the ratio R(t1)/R(t2) from (16.2). We can also
calculate
= R(t2)-R{t1)=
[v(t)dt.
(16.3)
Equations (16.2) and (16.3) are two equations for the determination of R(ti)
and R(t2). These stars change their radii by 5 to 20% during one cycle, i.e.,
during days or weeks, and their radii are about 50-100 solar radii. The
velocities are of the order of 30 k m s " 1 . If the period is 10 days (~ 106s)
and if they move outwards during half that time with an average velocity
of 15kms" 1 , we can estimate AJR to be
A i ? - 1 . 5 x 106-5 x 1 0 5 c m ~ 7 . 5 x 1 0 1 1 c m ~ 1 0 / l o
For a radius of 50 RQ this means a 20% change. The stars are really blown up
like a balloon.
The method of determining the radii of pulsating stars was developed by
Baade and Wesselink and is, therefore, called the Baade-Wesselink method.
Using the B V colors, measured as a function of phase in the pulsational
cycle, we can determine the effective temperature also as a function of
phase.* In Fig. 16.4 we reproduce the results obtained for S Cepheii, as given
by W. Becker. We find that the star has the highest Teff during maximum
light and the lowest Teff during minimum light. This means that the observed
light variations are mainly due to temperature changes; in fact, the stellar
radii during minimum and maximum light are nearly identical. Maximum
outward radial velocity also occurs during the maximum temperature phase.
The outward velocity then slowly decreases and the matter falls back onto
the star while it cools off. Any theory of stellar pulsations has to explain these
phase relations. If the stars were to pulsate in an adiabatic way, we would
expect the highest temperature for maximum contraction, which means for
* The phase is defined as the fraction of the period which has elapsed since maximum light.
159
163 The RR Lyrae stars
the smallest radius and the lowest temperature for the largest radius. This is
not what is observed.
The lengths of the periods for stars of given luminosity, the shape of the
light and velocity curves and the phase relations all depend on the internal
structure of the star, as we shall see in Volume 3. By means of these pulsating
stars we can therefore check our understanding of the interior structure and
the evolution of stars.
16.3
The RR Lyrae stars
Where the Cepheid instability strip cuts through the horizontal

branch of the population II HR diagram, we find the RR Lyrae stars. This
A
771
no
_
:
/ I
MS
/so
IV.0
-
b)
MS
/so
MS
c)
u''
&v
/SO
phase
Fig. 16.5. The light curves for the three Bailey types of RR Lyrae stars are
shown. Types a and b only differ in amplitude. The c RR Lyraes show a
nearly sinusoidal light variation. (From Ledoux & Walraven 1958.)
160
16 Pulsating stars
group of pulsating stars is found in globular clusters as well as in the field.

They have periods of around 1/3-1 day.
People distinguish three types of RR Lyrae stars: Bailey types a, b, and c,
called so after the astronomer who first made this distinction. Light curves
for the three types are shown in Fig. 16.5. While the type c stars have almost
sinusoidal light curves, the types a and b have very non-sinusoidal curves.
The main difference between types a and b is the amplitude, and there is
really no basic difference between these two types except for the smaller
amplitude and the somewhat longer periods for the type b stars. But there is a
very basic difference between types a and c.
In Fig. 16.6 we show the relation between amplitude and period for RR
Lyrae stars. Again we see that there is a smooth transition between type a
and type b RR Lyraes, while those of type c are clearly distinct. The type c
have the shorter periods. They have all periods less than 1/2 day, while the a
and b types all have periods longer than 1/2 day. It turns out that the type a
and b RR Lyrae pulsate in their fundamental mode, i.e., with the longest
possible period, while the type c RR Lyraes pulsate in the first overtone.
There are also RR Lyrae stars which have variable amplitudes in light
variation, like the star RR Lyrae itself. We can understand this as a beat
phenomenon of several periods.
ampl.
I * Type a
Typeb
Typec
ft
0.3
0.5
06
0.7
0.8
Period days
Fig. 16.6. We show the relation between amplitude and period for the three
types of RR Lyrae stars. The c types have the shortest periods and the lowest
amplitudes. They are clearly distinct from the a and b types. (From Ledoux &
Walraven 1958.)
16.4 The period-luminosity relation

16.4
161
The period-luminosity relation
The S Cephei stars are also very important for a very different reason,
namely their period-luminosity relation. For the Cepheids found in the
nearest external galaxy, the Large Magellanic Cloud, Ms Levitt discovered
that the brightest Cepheids had the longest periods. When she plotted the
luminosities of the Cepheids as a function of their periods, she found a rather
tight correlation. For the Cepheids within the Large Magellanic Cloud we
know that they must all be at nearly the same distance because the diameter
of this galaxy is much smaller than the distance from us. This periodluminosity relation found for the variables in the Large Magellanic Cloud
therefore showed that it is intrinsic to the Cepheids and can be used for other
Cepheids as well. Once it is calibrated, which means once we know the
absolute magnitude for one Cepheid of a given period, we can, in fact, use this
period-luminosity relation to determine the distance to any Cepheid with a
measured period. It is usually not very difficult to determine the periods of the
light variations, and the absolute magnitudes can then be read off from this
period-luminosity relation. The apparent visual magnitude can be measured and the distance modulus can be determined, which tells us the distance
for instance to the Large Magellanic Cloud or to the Andromeda galaxy.
The calibration of the period-luminosity relation can be achieved by
means of Cepheids which are members of open clusters in our Galaxy. The
distances to such clusters can be obtained by means of spectroscopic
parallaxes, or by main sequence fitting, as discussed in Section 5.4.
The period-luminosity relation for the Cepheids has been extremely
important for the determination of the distances to the nearest extra-galactic
systems in our local group of galaxies.
In Fig. 16.7 we show the period-luminosity relation for population I
Cepheids, as determined by Sandage and Tammann (1969), Schmidt (1984),
and by us (1986)*. There is still an uncertainty of about 0.5 magnitudes in the
absolute calibration, at least for the longer period Cepheids, which are the
most important ones, because they are the brightest which can be observed
best in extragalactic star systems. Discussions about the accurate cluster
distances are still going on. The problem is the uncertainty in the
trigonometric parallaxes for the nearby stars upon which all our distance
determinations are based. It is very difficult to measure high precision
trigonometric parallaxes if the parallax angles are less than 0.05 arcsec!
How can we understand the existence of the period-luminosity relation?
* There is actually a B V dependent term in the period-luminosity relation. We have used the
average relation between (B V)o and the period to eliminate this term.
162
16 Pulsating stars
Intuitively, it seems plausible that it should take the more luminous stars,
which are the largest stars, longer to expand and contract than it takes the
less luminous, smaller ones. Actually, the periods of the pulsating stars are
determined by a resonance phenomenon. We are essentially observing a
standing wave. The phenomenon can be compared to a rope fastened at one
end to a wall. If you move the free end of the rope, you can generate a limited
number of standing waves with very distinct frequencies, the eigenfrequencies. The wave which you generate by moving the free end proceeds along
the rope until it is reflected at thefixedend. We observe standing waves if at
,Sch.
-5
B.-V.
-3
using (B-V) o = 0.45 log P + 0.34

<B>0-<V>0 = 0.37 log P + 0.37
_2
-1
0.5
1.0
Log period
1.5
Fig. 16.7. The relation between absolute magnitudes and periods (periodluminosity relation) is shown for population I Cepheids according to different
recent studies. S.T. stands for Sandage, Tammann 1969. B.-V. stands for BohmVitense 1986, who adopts the Sandage, Tammann relation, except for a
decrease in intrinsic brightness by 0.2 magnitudes. Sch. stands for Schmidt
1985, who redetermined the distances to clusters with Cepheids.
16.4 The period-luminosity relation
163
any given point the reflected wave has the same phase as the oncoming wave.
We can observe a low frequency, the fundamental mode, and we can generate
some higher frequency waves, the overtones. The star acts in very much the
same way as this rope. The center of the star represents the fixed end of the
rope because the high-density center cannot move. The surface of the star is
the free end of the rope which, due to some excitation mechanism, is moving
back and forth. Only if the reflected wave has the same phase as the inward
moving wave, can the star generate a standing wave or a radial pulsation.
Clearly, the travel time from the outside in and back is very important for the
determination of these resonance waves. The travel time through the star
depends on the density and temperature stratification in the star. This is why
the frequencies of pulsation can tell us so much about the interior structure
of the star.
In a qualitative way, the fundamental period P can be estimated from the
condition that the period P is given by the travel time of a pressure wave to
the centre of the star and back, i.e.,
(16.4)
where C s is the velocity of sound. With
C
where
s=J(y'j)
7 =^-5/3
(16.5)
for a monatomic gas. Cp and Cv are the specific heats at constant pressure
and constant volume, respectively. At any place in the star the pressure must
balance the weight of the overlying material. We derive for the center.
n
- r, -
- GM
Pg p-R-g P'R'^r
_ G'M
= PR-
<+rr^
( 16 - 6 )
where G is the gravitational constant and M the mass of the star, p is the
average density and p-R is the mass in a column of 1 cm 2 cross-section above
the center. This means
5 /"
We then find for the period
or
GM
164
16 Pulsating stars
or
9
// 1 \
"2
(16.9)
or
P = const. p " 1 / 2 .
(16.10)
This qualitative estimate gives only the correct dependence on density.

The constant has to be determined by more accurate calculations.
Equation (16.9) explains the observed period-luminosity relation. The
larger the luminosity, i.e., the radius, the smaller the density, and the longer
the period.
17
Explosive stars
17.1
Supernovae
17.1.1 Classification of novae and supernovae
The ancient astronomers had already noted that sometimes new

stars became visible in the sky and after some time disappeared again. In the
Middle Ages the astronomers called these stars novae, which is the Latin
word for new stars. Some of these new stars were exceedingly bright, and
were later called supernovae. Three of these supernovae were observed in
historic times: Tycho de Brahe's supernova, which occurred in the year 1572,
Kepler's supernova which became very bright in the year 1604, and a
supernova which was observed by Chinese astronomers in the year 1054. At
the location of the Chinese supernova we now see the Crab nebula in the
constellation of Taurus. The nebula got its name from its appearance which
reminds us of a crab. The Crab nebula still expands with velocities of about
1400 km s ~ S showing that a truly gigantic explosion must have occurred 900
years ago.
What are these novae and supernovae? How often do they occur? What
kinds of objects are their progenitors? What leads to such gigantic
explosions? What distinguishes novae and supernovae? Are all supernovae
similar events or do we have to distinguish different kinds of novae or
supernovae? These are questions for which we would like tofindthe answers.
Both novae and supernovae are objects which suddenly increase their light
output by many orders of magnitude. They are usually not visible before the
outburst, so they appear as new stars. Since we do not see them before the
outburst, we really do not know directly what types of objects the
progenitors are, except for the very unusual supernova which was discovered
in the Large Magellanic Cloud in February 1987, for which the progenitor
has been observed before and was classified as a B3 Ib supergiant. This
supernova was very unusual, not only because it was intrinsically much
165
166
17 Explosive stars
fainter than other supernovae observed so far, but also because the time
dependence of its light output was very different from that of other known
supernovae. Other supernovae therefore probabhy have very different
progenitors.
In Fig. 17.1 we reproduce a plot by W. Zwicky (1965), showing the
frequency of new stars with different maximum brightnesses. We see a
continuous distribution which is determined by both the intrinsic brightnesses of the novae and their distances. If we want to know the distribution of
the intrinsic brightnesses we have to compare objects at nearly equal
distances, such as, for instance, the novae or supernovae in the Andromeda
galaxy or all the ones in the Magellanic Clouds. Supernovae are not very
frequent events, so we cannot observe supernovae in only one galaxy, but we
have to combine observations from many galaxies at approximately the
same distances. In Fig. 17.2 the apparent magnitudes of the observed new
stars in nearby extragalactic systems have been plotted. In this graph two
groups can clearly be distinguished; the brighter group being the supernovae
and the fainter one of the novae.* Novae are less bright and more frequent.
They can best be observed in nearby galaxies and in our own Galaxy. The
one supernova observed in the Andromeda nebula, also called M31, had a
maximum apparent magnitude mv between 7 and 8, while the supernova
2 -
-3
-2
-\
visual magnitude
Fig. 17.1. The frequency of 'new' stars in our Galaxy with a given apparent
magnitude is shown for the last 700 years, according to Zwicky (1965).
*On average the supernovae plotted here are further away then the novae shown, The difference in
intrinsic brightness is therefore still larger than indicated by Fig. 17.2.
17.1 Supernovae
167
1987A observed in the Large Magellanic Cloud reached a maximum

apparent brightness of 2.9 magnitudes. It clearly appeared much brighter
than any other extragalactic supernova observed so far.
With a distance of 700kpc = 7 x 105pc the absolute magnitude at
maximum light for the Andromeda supernova was (neglecting interstellar
absorption)
Mv = mv - 5 x 4.85 = mv - 24.3 = 7.3 - 24.3 = - 17.
For the Large Magellanic Cloud 1987A supernova we find with a distance
modulus of mv Mv = 18.3 an absolute magnitude at maximum visual light
of
Mv = 2.9-18.3 = -15.4
This has to be compared with the absolute magnitude Mv = 8 for the
brightest 'normal' star in the galaxy, or with the solar absolute magnitude of
Mv = + 4.8. The supernova is 23 magnitudes brighter, or as bright as 109
suns, in the visual. Actually, the average Mv for supernovae appears to be
Mv~-18l.
According to the records of the Chinese astronomers, the Taurus
supernova reached a maximum apparent visual magnitude of about 5,
which means it was visible during daytime. For Tycho's supernova the
maximum mv was about 4 and for Kepler's about 3. These magnitudes
7
4
Novas in nearby galaxies.

Within the past hundred
yeors
Supernovae
1
i
9
10
II
12
13
apparent visual magnitude
Fig. 17.2. The frequency of 'new' stars for nearby extragalactic systems
with a given apparent visual magnitude is shown, according to
Zwicky (1965).
168
17 Explosive stars
are, of course, influenced by different distances and by absorption due to the

interstellar gas and dust which is especially large for Kepler's supernova. The
brightness of a supernova decreases exponentially with time. After one or
two years it becomes invisible.
Altogether, several hundred supernovae have been observed. Unfortunately, none has occurred in our own Galaxy lately. The maximum
brightnesses of the supernovae observed in external galaxies are about two
magnitudes fainter than the whole galaxy. For the smaller, fainter galaxies
the supernova can be just as bright as the whole galaxy. In Fig. 17.3 we show
a photograph of a supernova in an external galaxy, IC 4281.
According to Zwicky, who has observed most of the supernovae, there are
at least five different types as judged by their light curves. Most of the
observed supernovae belong, however, to two types, called type I and type II.
They can be distinguished by the shapes of their light curves as well as by
their spectra. The observed light curves of some supernovae of type I are
shown in Fig. 17.4. A typical type I supernova declines by 15-19 magnitudes
during thefirsttwo years after the outburst, or by about 8 magnitudes in one
year.
Figure 17.5 shows some light curves of type II supernovae. They generally
look much more irregular than those of type I, although some do not look
Fig. 17.3. Two pictures of the faint galaxy IC 4281 are shown, one before the
outburst of the supernova and one after the occurrence of the supernova.
(From Zwicky 1965.)
17.1 Supernovae
169
m
8.0
"iIr
rsilO.0
GO
IC4I82
OI2.0
ao
14.0
15.0 o
NGC 4636
m
I4Q
ID
OI6.0 ^
o
17.0
m
-rg
NGC 6946 No.2
l2
NGCU62I
16.0
18.0
m
12.0
J4.0
GI6.0
NGC 1003
O
Z
18.0
>/.
20.0
I
80
240
160
320
480
Fig. 17.4. Observed light curves are shown for type I supernovae. The names
of supernovae are given. (From Unsold 1982.)
4.0
\
1
15.0
16.0
NGC 4725 ~
^ - .
- ^
I4?O
,C 4273
15.0
16.0
I3?O
15.0
17.0
19.0
r \
\
\\
-(
NGC 5907
-
NGC 4559
\
1
Fig. 17.5. Light curves for type II supernovae. (From Zwicky 1965.)
170
17 Explosive stars
too different. When looking at such light curves, we have to be careful with
the interpretation.
The spectra of the type I supernovae are still not understood. They show
broad bands (see Fig. 17.6) and it is hard to decide whether the light regions
are emission lines or whether the dark bands are absorption bands in a
bright continuum. We probably have both absorption and emission lines. It
seems that in type I supernovae we see broad absorption lines of heavy
elements while in type II supernovae hydrogen absorption lines are
identified. The new classification distinguishes between types la and Ib. In
type Ib supernovae Hell lines are seen. If the widths of the lines are
interpreted as being due to Doppler shifts, then they indicate velocities of up
to 20 000 km s" 1, or - 15 000 miles s"1 . The material would fly halfway
around the whole Earth in one second. It appears that we are watching
gigantic explosions.
A few days after maximum light the overall spectral energy distributions
resemble that of a supergiant with temperatures around 10 000 K at
maximum and cooling down to 5000 or 6000 K after two or three weeks.
In Fig. 17.7 we show the spectrum of a type II supernova. Broad lines of
S, Hy, and US are seen. The line widths are also large, corresponding to
IJ
i
i
O CD
Jh> - O*
O
00
OD
00
en
iff*
31
Fig. 17.6. We show a spectrum of a supernova of type I. It is difficult to decide

whether we see broad absorption bands in a continuous spectrum, or whether
we see broad emission bands on a much lower continuum. A comparison
spectra with hydrogen and helium emission lines are also shown on both sides
of the supernova spectrum. (From Zwicky 1965.)
171
17.1 Supernovae
velocities of up to about 10 000 km s \ The velocities seem to decrease as

time goes on. Whether this means an actual slowing down of the motion or
whether it means that we see successively deeper layers with decreasing
velocities is not quite clear.
The Large Magellanic Cloud supernova also showed hydrogen and
helium lines. The hydrogen lines were in the beginning shifted towards the
blue, corresponding to Doppler shifts for 20 000 km s"1, but then slowing
down to velocities of about 10000kms" 1 and remaining at the same
velocity for several weeks.
In the first days this supernova had an overall energy distribution
resembling that of a hot star but deep broad absorption troughs seemed to
Fig. 17.7 (a). The spectrum of a type II supernova is seen. The strong
absorption lines can be identified with lines of hydrogen or helium, seen in the
laboratory comparison spectra on the side of the stellar spectrum. (From
Zwicky 1965.)
s ||lf;
Supernova in NGC 7 3 3 l g
_J_J
,|
[*
n i
Fig. 17.7(b). The intensities lx measured in the spectrum of a type II supernova

are shown as a function of wavelength L The broad Ha emission line at
6562.7 A can well be recognized, as well as the emission for the He I line at
4471.5 A.
172
17 Explosive stars
be superimposed. After about two weeks the overall energy distribution

suggested temperatures around 6000 K.
There seems to have been one other supernova which was also observed
shortly before maximum light, when its overall energy distribution suggested
a temperature around 20 000 K. At maximum light the temperature
appeared to have dropped to 10 000 K, but no lines were seen.
It looks as if the supernovae become very hot briefly and then start
expanding while they cool at the surface, probably due to the expansion.
After the maximum the supernovae appear to be rather cool, therefore
recent studies have attempted to identify the broad, deep absorption bands
with lines of Fe II, Ca II, Na I, Mg II, etc., all lines which are normally seen in
rather low temperature stars with temperatures between 5000 and 10 000 K.
There seems to have been some success in explaining the spectra.
After 10 days the radius of the ejected shell is about 10 9 cms" 1 10 6 s =
1015 cm = 60 AU, which is more than the distance from the sun to Pluto. If
the explosion started from a star of, say, 10 /? o , then the radius has increased
by a factor of 1200.
In Figs 17.8 and 17.9 we show the geometry of the expansion. In the
beginning a rather large fraction of the material (section d), in our drawing
about 10%, is in front of the star and can absorb light. The corresponding
fraction c in the back is covered up by the star and is invisible. The material
on the sides of the star (sections a and b) contributes to an emission line. This
material is moving transversely to the line of sight and therefore the emission
Blue shifted
absorption
from d
Emission from
sections a and b
Shell A
Shell B
Fig. 17.8. In the beginning of the supernova expansion the material in the outer
shell (shell A) is rather dense. There is much material in section d in front of the
star to absorb light from the deeper layers, causing a blueshifted absorption
line. Sections a and b cannot contribute to the absorption line; they can only
emit light, giving rise to an emission line. These sections move on average
transverse to the line of sight; the emission line is therefore only broadened but
not shifted. Redshifted emission lines are emitted from section c but are
obscured by the star. The expected absorption-emission line profile is shown
on the right side of the figure. These types of line profiles were first observed
for the star P Cygni (which is not a supernova) and are therefore called P
Cygni lines. In Fig. 17.10 we show a part of the spectrum of this star.
17.1 Supernovae
173
lines have essentially no Doppler shift. The expected line profile is shown
qualitatively in Fig. 17.8 on the right. Shell B below shell A contributes only
to the background against which A is seen in absorption.
Sometime later we have the situation shown in Fig. 17.9. Shell A has
expanded so far that only a small fraction d is in front of the star and can
absorb light from the star. The density has decreased by a large fraction.
Shell A contributes little to the absorption and emission. It is now
transparent and shell B becomes visible. Shell B has a smaller velocity that A,
so the absorption lines caused by shell B show a smaller Doppler shift and
the spectrum shows a smaller shift in absorption and emission, although the
expansion did not actually slow down. This only shows how careful one has
to be when interpreting changes in the velocity shifts of emission and
absorption lines.
In Fig. 17.10 we show the spectrum of the star P Cygni, which is not a
supernova, but which also has an expanding shell, but the velocities are
much smaller for this star. We see the typical combination of absorption and
emission lines.
Fig. 17.11 demonstrates that we also have to be very careful when we try to
interpret the change of light output by a supernova. The supernova may be
Spectrum from shell A
Absorption from d
x
Emission from a and b

continued
Shell >a
Shell B
Spectral line from shell B
Fig. 17.9. The geometry at a later stage of the supernova expansion is shown.
Shell A has greatly expanded and becomes transparent because of its low
density. There is now much less material in section d in front of the star, which
can cause a blueshifted absorption line. The next deeper shell B now gives the
main contribution to the emission and absorption. Shell B has lower velocities
than shell A. The expected line profiles due to shells A and B are shown on the
right-hand side. The observed line profile is a superposition of these profiles.
174
17 Explosive stars
in the neighborhood of dust clouds. When the supernova explodes light will
be emitted in all directions. At first, we only see the light which is emitted into
our direction (see Fig. 17.11). Later on we may also see light which is
reflected from a neighboring cloud. We see then a so-called light echo.
17.1.2 Estimation of age and distance of supernova remnants and mass
loss of supernovae
1 u1
11 1
4)
X
ooo
0
4116
4121
1=1
4089
M
Z XX
4026
M
*>
X
Hel 4009
3995
3956
3965
ao
3927
If the ejected mass expands so fast, we should be able to see the shell
growing. Suppose the supernova is at a distance of 2000 pc, the probable
distance of the Crab nebula. Then a shell of 1 AU diameter would have an
angular diameter of 1/2000 arcsec. (At a distance of 1 pc, the diameter of
1 AU has an angular diameter of 1 arcsec, by the definition of the pc.) With an
Ed
m T
i/> X
1
PCyg
Cas
Fig. 17.10. A small wavelength interval of the spectrum of the Be star (Be
means B star with emission lines) P Cygni is reproduced, showing the
emission-absorption line profiles seen for an expanding gas shell around a star.
For comparison the spectrum of the star K Cas is also shown. (From Morgan,
Abt & Tapscott 1978).
Observer
Fig. 17.11. The change of the light output of a supernova after maximum light
is not only determined by the light output of the supernova itself, but may also
be influenced by light reflected from nearby dust clouds, the so-called light
echo.
17.1 Supernovae
175
expansion velocity of lOOOOkm/s = 10 9 cm/s, as observed shortly after

maximum, the shell would have a radius of 3 x 10 16 cm, or 2000 AU after
one year if the velocity remained constant. This corresponds to 1 arcsec if it is
at a distance of 2000 pc. Actually, the expansion velocity of the Crab is
observed to decrease. The present expansion velocity of the Crab nebula, for
instance, is about 1400km s~ \ or ~ 5 x 10 15 cm per year. After 1000 years
the nebula will then have a radius of 5 x 10 18 cm, or 1.7 pc, or 3 x 105 AU. At
a distance of 2000 pc, this corresponds to ~ 150 arcsec. The actual radius is
about 180 arcsec, showing that the velocity has decreased during the 950
years and was, on average, ~ 2000km s" 1 if the Crab is at a distance of
2000 pc. The average velocity could hardly have been less than the present
1400 km s" 1 , so the Crab nebula must be at a distance d^ 1400 pc.
We can use this method in a more refined way to determine the distance to
the Crab nebula and thereby the maximum absolute magnitude of the Crab
supernova. The present expansion of the nebula is 0".222 arcsec per year.
The measured radial velocity, i.e., perpendicular to the direction in which
we measure the expansion, is 1450 km s" 1 . If the velocity in the two perpendicular directions is the same, then the Crab must be at a distance of
1380 pc. At this distance the supernova would not have been very bright
intrinsically, only M v = 15.7, if there were to be no interstellar absorption, or with 1.5 magnitude interstellar absorption, as seems reasonable (see
Chapter 19) an absolute magnitude of M v = 17.2 is obtained.
The expansion may, however, not be spherically symmetric. If the
expansion velocity perpendicular to the line of sight is 1.25 times larger than
the radial velocity in the line of sight, then the distance becomes 1730 pc. At
such a distance it is reasonable to assume that the interstellar absorption
dims the supernova by about 2 magnitudes. Taking this into account, we
find a maximum absolute magnitude for the Crab supernova of M v =
18.2, which would be close to the maximum brightness observed for other
extragalactic supernovae of type I.
From the reconstructed light curves it seems very likely that the Chinese
nova and Kepler's and Tycho's novae were all supernovae of type I. There
seems to be a spread in maximum M v for type I supernovae. The brightest
ones are called type la. Since all Galactic ones appear to have been
somewhat fainter they may have been of type Ib, or they may have been more
obscured or further away than we think.
How much mass is ejected in such a supernova explosion? From the light
emitted by the Crab nebula now, an estimate of the luminous mass can be
made which gives a range 0.35 M o < M < 3.5 M 0 , which really does not tell
us too much, except that it is not a negligible fraction of the mass of the
176
17 Explosive stars
supernova. Part of the luminous mass in the nebula was swept up from the
interstellar medium. If the density was 1 hydrogen atom per cm3 then the
accreted mass was ~0.2M o . Most of the mass is probably from the
supernova outburst.
17.1.3 The origin of supernovae
Unfortunately nobody has ever observed a presupernova until

recently when the supernova in the Large Magellanic Cloud exploded. In
this case the star at the position of the supernova was previously observed
and has been classified by Sanduliac as a B3 Ib supergiant. This supernova
increased its brightness by about 10 magnitudes. For other supernovae we
can only make rough estimates. Since the brightest normal stars have M v =
8, a supernova must increase its brightness by at least 10 magnitudes,
probably much more. If the Crab presupernova had M v = 8 and at
maximum it had mv = 5, then the presupernova would have had mv =
5 + 11= +6. A trained astronomer would just barely have been able to
see it with the naked eye, but it might have escaped observation. It was not
recorded, probably it was fainter and therefore not visible. The other galactic
supernovae were further away and more obscured by interstellar dust.
Galileo had a telescope but he obviously had not seen the presupernova
before the sudden brightening. Whether he could have seen it, we do not
know. He did observe Kepler's supernova after it had become bright.
Three supernovae have been observed in nearby galaxies. From a
comparison with the brightest stars in those galaxies, an absolute visual
magnitude of Mv = 18 + 1 was derived.
We can get some information about the mass of the presupernovae if we
check with which kinds of stars they are associated. We saw earlier that old
stars are seen in globular clusters and they all have masses less than 1 solar
mass. Elliptical galaxies supposedly have mainly such old stars. On the other
hand, young massive stars are found mainly in the spiral arms of galaxies, of
course together with less massive stars. If presupernovae are massive stars,
we might expect to see them associated with young star clusters. Zwicky at
the California Institute of Technology was the first one to start a systematic
search for supernovae. He did some statistics with 111 supernovae, observed
until 1965. He found that about 26% of all types I supernova occurred in
elliptical galaxies, i.e., among old stars. Other type I supernovae were found
in spirals and irregular galaxies in about the proportions of the number of
existing galaxies. In the spirals, the supernovae are found mainly in the
outer parts. Since a rather large number of type I supernovae occur in
17.1 Supernovae
177
elliptical galaxies, those must belong to the old population. The progenitors
are probably low mass stars.
Type II supernovae have not been seen in elliptical galaxies, it seems, but
even that does not seem to be quite established. The classification is
sometimes difficult; a few supernovae seen in ellipticals could not be
classified very well because not enough observations existed. So it is not
quite clear at the moment, but it appears that type II supernovae are not
found associated with ellipticals, suggesting that they belong to a young
stellar population and probably have massive stars as progenitors. In
external spiral galaxies type II supernovae often appear to be associated
with regions containing the massive O and B stars. If they indeed have 520 M o stars as progenitors, (the LMC supernova progenitor appears to have
had 15 M o ) there should have been several supernovae explosions in the
Pleiades and in the Hyades star clusters. Should we be able to see the
remnants? How old are the oldest remnants that we actually see? Some
spherical nebulae are supposed to be supernova remnants like the Cygnus
loop, the Veil nebula, the Vela remnant, or the Gum nebula (see Figs 17.12
and 17.13).
Fig. 17.12. The Crab nebula (in the constellation of Taurus) is shown. Near the
center of the nebula there are two stars close together. The lower right star of
these two shows a featureless blue continuous spectrum. It is the pulsar. (From
Abell 1982.)
178
17 Explosive stars
If type II supernovae have massive stars as progenitors, should we not

expect to see supernovae or at least supernovae remnants in young galactic
clusters with many massive stars?
From the observed radius and the radial velocities, the ages of the remnants
can be estimated, as discussed above. They are of the order of 10 4 -10 5 years,
and none is older than 105 years. If there were perhaps five massive stars in
the Pleiades which became supernovae during the lifetime of 108 years, there
was one every 107 years. The chance of seeing a remnant now are thus 1 in
100, which is consistent with the fact that we do not see any. There are not
enough young clusters around for us to have a good chance of seeing one in a
cluster. So that does not tell us much.
There are, however, young, very populous clusters in the LMC. Perhaps
we have a chance to see a supernova go off in one of them. It would be very
interesting to study the effects on the other cluster stars. The observed LMC
supernova was not in a cluster.
1.7.1.4 Energetics of supernova explosions
How much energy is involved in a supernova explosion? How can
this energy be supplied?
The most exciting thing about the supernovae is the enormous energy
Fig. 17.13. The filamentary nebula in Cygnus is shown. It is believed to be the

remnant of an old supernova. (From Abell 1982.)
17.1 Supernovae
179
output. If its absolute visual magnitude is M v ~ 19, as compared to the

solar absolute visual magnitude of M v ~ + 5, we see that the visual light
output of the supernova is about 10 96 times as much as that of the sun.
Considering that the supernova has T ^ 10 000 K and has a bolometric
correction say of about 0.4 magnitudes, we find that the supernova at
maximum has a luminosity which corresponds to at least 1010 suns.
This is about a factor of 10 less than the luminosity of the whole galaxy. We
saw above that a supernova at maximum is only two magnitudes fainter than
a bright galaxy.
The luminosity at maximum is then 10 43 -10 44 ergs~ 1 , which means the
total visible energy output is 1049 -1050 erg. It is generally assumed that the
energy output in the form of neutrinos is even higher, and the neutrinos
observed after the outburst of the LMC supernova seem to confirm this.
If a star were to convert all its hydrogen into helium by nuclear fusion, then
0.7% of its mass would be converted into energy and the total amount of
nuclear energy liberated would be (see Volume 3)
where M is the hydrogen mass of the star. For the sun this would be
In an outburst, the supernova releases in 10 days almost as much energy as

the whole nuclear energy available to a star with 1 M o . But it is our present
understanding that the supernova explosion happens at the end of the stellar
evolution and therefore most of the nuclear energy has been used up
already. There must be another energy source.
The gravitational energy released by a star when it collapses from a dust
cloud into a star with radius R is
f
h
*~ R '
which for the sun comes out to be ~ 1048 erg, which again is not enough for a
supernova.
If, however, the sun were to collapse to a white dwarf with a radius 100
times smaller than that of the sun, then the gravitational energy release
would be 1050 erg. We would be already coming close to the energy release of
a supernova. If the sun were to collapse to a neutron star with R < 70 km =
7 x 106 cm, the total gravitational energy release would be ^ 1052 erg. The
gravitational collapse of a normal star into a neutron star could provide
enough and even more energy than observed for a supernova explosion. In
180
17 Explosive stars
fact this seems to be the only known energy source which can provide
enough energy.
17.1.5 The Crab pulsar
Does a presupernova collapse into a neutron star?
The only way to answer this is to look at supernova remnants like the
Crab. In the center of the Crab nebula we see two stars (see Fig. 17.12). The
one star is a normal F star. The other one emits a blue continuum with no
spectral features. It has long been suspected to be the remnant of the
supernova outburst. The real confirmation came only when it was discovered that it is a pulsar. This means it was observed to have pulsed light
emission, with a period of 1/30 of a second. These pulses were first detected at
radio wavelengths, but they were later also confirmed in the visual.
In Fig. 17.14 we show a series of pictures of the Crab nebula taken at
different phases of the pulse cycle. The first picture was apparently taken
shortly after the off phase of the pulses, the third and fourth and the ninth
and tenth ones were taken during the on phases. The star with the F star
spectrum is always visible, but the blue star with the featureless spectrum
comes and goes, showing that the light is turned on and off 30 times a second.
Because the period for these pulses is shorter than the resolution time of our
eyes, which is 1/16 of a second, these light flashes remained undetected until
the radio astronomers developed receivers with very high time resolution
which enabled them to detect such short pulses of radio waves.
Fig. 17.14. A time sequence of photographs of the central pair of stars in the
Crab nebula is shown as a function of phase in the pulse cycle. The time
differences are less than 1/100 of a second. (Repeated exposures is successive
cycles are necessary to get enough light.) While the F star remains constant in
light, the pulsar turns on and off. (From Abell 1982. Photographs from Kitt
Peak National Observatory.)
17.1 Supernovae
181
Between the main pulses there is always a smaller interpulse. Each pulse is
subdivided into many small pulses with very short timescales. The main
pulses occur extremely regularly. The period is very well defined, with no
irregularities. The only phenomenon known to be so regular is rotation. But
this means that this star must rotate 30 times a second. For such rapid
rotation we must ask ourselves, how can the star stay together and not be
torn apart by centrifugal forces? If the star does not disintegrate, the
gravitational force has to overcome the centrifugal forces at the surface of the
star, i.e.,
v2
GM
_
or
-R^-RT
GM
2nR
u
~^r
where t? = - p -
n _
(17.1)
R<
P is the rotation period and v the equatorial rotational velocity. Inserting

the expression for v9 we find
D3
GMP2
7X 1(T8-2X
<-*?-"
10 33 cm
40(30?
in21
3 X 1
Cm
__
(17 2)
'
or R ^ 1.5 x 107 for a 1 M o star or, in other words, the radius of the star must
be less than 150 km.
The shortest possible period is obtained if we insert the highest possible
velocity, namely v = c where c = velocity of light, into the stability criterion
(17.1). We then find
GM 7 x l ( T 8 - 2 x l 0 3 3
5
-^- =
=1.5xl0 5 cm
9 x 1Q2o
(17.3)
and
In principle, much shorter periods than observed for the Crab pulsar would
be possible. However, such objects would have extremely high densities. If
we assume the Crab pulsar would have a radius of ~ 100 km = 107 cm, then
the density for a 1 M o would be
Z
111
A ISS.
A A 1 O
If you pack nuclear matter densely, you find densities around 1014 g cm 3.
Remember that for white dwarfs we found densities around 1 ton per cm3.
For the Crab pulsar we find, if its mass is about 1 solar mass, that the density
is at least 1 million tons per cm3. There is only one kind of star known, which
can have such high densities, namely the so-called neutron stars. Because of
182
17 Explosive stars
the extremely high densities in these stars the electrons and protons are
squeezed together to form neutrons. The stars therefore consist mainly of
neutrons and have a radius of about 15 km.
How does the pulsar get such high rotational velocity?
It turns out that this is actually no problem. If a star collapses, conserving
angular momentum for each unit volume, i.e.,
pf2-a> = const,
(17.5)
then
1
Shrinking r by a factor
1000 000 km
100 km
1.5 Re
100 km
increases co by a factor of 108 or the period P = l/co would decrease by a factor

of 108. If the star originally had a rotational period of 1/30 x 108 s = 3x
106 s or 1 month like the sun, then there is no problem.
Consider, for instance, a B star with vrot ~ 200 km s" \ which corresponds
to a rotational period of
2 x 106km-27r_1.26x 107km ,
~
1A5
nn
vrnt
" 200kms" 1 = 6x 105s
(17.6)
or P~6 days, or even a star with i;rot = 20kms \ which would have a
rotation period of 60 days, they could easily have a rotational period as
short as the Crab pulsar if they were to collapse into a neutron star.
For the Crab supernova, it then seems very likely that it gained its energy
from the gravitational energy released in the collapse of a major fraction of
its mass into a neutron star now visible as the Crab pulsar.
17.1.6 Other supernova remnants
The question then remains, do all supernovae explosions leave a

neutron star? In order to answer this, we have to know how a neutron star
can be discovered. In the Vela remnant, another pulsar is found. No pulsar
has been seen in the Tycho and Kepler remnants, but this does not necessarily
mean that there is no neutron star left.
The Crab pulsar emits light for only ~ 10% of its period. The only way to
explain this is by assuming that the pulsar emits strongly beamed light (see
Fig. 17.15). If the beam is narrow, the observer will be 'hit' by the beam only
in exceptional situations. For the Crab, we actually see a pulse and an
17.2 Novae
183
interpulse. We must be close to situation of Fig. 17.15(ft). The Crab pulsar is

the only pulsar seen to pulse in visual light. The radio beam is probably
wider, so it can be more easily observed.
For the remnants in which we do not see a pulsar, the geometry may be
like case (c) where we cannot see the beam. There may be neutron stars in all
supernova remnants: we have no way of knowing. We are just lucky that the
Crab pulsar beam points at us.
It is also possible that only certain types of supernovae leave behind a
neutron star. Not knowing what the masses of presupernovae are, we must
ask for which kind of star we might expect such an implosion-explosion
process: which means implosion in the center leading to the formation of the
neutron star, and explosion in the outer layers which leads to the explosion
and the strong light increase.
In Volume 3 we will see that in the late stages of stellar evolution we may
actually expect a situation when the interior of a massive star will become
unstable leading to a collapse which will free enough gravitational energy to
cause an explosion of the outer layers. We believe that such stars are the
origin of type II supernovae.
17.2
Novae
17.2.1 Observations of novae

We shall now talk about the group of novae which is the less
luminous but the more frequent. These are the so-called classical novae.
They are much more frequent and have therefore been studied in much
more detail. They reach maximum absolute visual magnitudes up to 9,
Observer
Observer
Observer
(a)
(c)
Fig. 17.15. If the pulsar emits strongly beamed light the observer may either see
one light pulse per rotation period, case (a) or two pulses per rotation period,
case (b) or she/he may not see a pulse at all, case (c), depending on the
orientation of the light beam with respect to the rotation axis and on the
direction of the line of sight.
184
17 Explosive stars
i.e., about 10 magnitudes less than supernovae. Several of the novae have
been observed before they erupted. Some of them have recurrent eruptions.
During eruption they change their brightness by more than 9 magnitudes in
a few days, they then decline to pre-eruption magnitude, sometimes within
one or two years, like the supernovae, though the light curves and decline
times are very different for different novae. There are fast, slow and moderate
novae. They show a pre-maximum spectrum like O, B or A stars, but are
definitely not main sequence stars, and are very subluminous. Their
distances can again be determined from the radial velocities and the
expansion of the nebula resulting from the explosion (see Fig. 17.16) in the
same way as we discussed for the Crab nebula.
With this distance, we can determine the absolute magnitude of the
prenova and see that it was a subluminous star. It is not faint enough for the
light to be only due to a white dwarf. There is an additional light source.
Two observations were very important for the understanding of the nova
phenomenon: (a) the nova DQ Herculis was found to be an eclipsing binary;
subsequently more postnovae were found to be binaries, (b) several novae
show small amplitude light variations with periods around 70 s.
Observations (a) suggest that all novae are close binaries. Observations (b)
Fig. 17.16. The expanding shell of Nova Herculis is seen on this picture taken
in 1972. The nova exploded in 1934. (From Abell 1982.)
17.2 Novae
185
indicate that one of the components must be a white dwarf, because only a
white dwarf is small and dense enough to have oscillation periods of minutes.
Normal stars would have longer periods, as was discussed earlier.
Since the post- and presupernovae are blue objects, brighter than a white
dwarf, there must be an additional subluminous blue light source, which we
think is a disk of hot material. In addition, there must be the second star,
which has been seen in only one case, where it is a K2 IV star which means
slightly more luminous than a main sequence star. In all other cases it is
obviously too faint to be seen. It probably is a lower main sequence star.
17.2.2 Close binaries
Before we can go any further with our discussion of the origin of the
nova explosions, we have to talk a little about close binaries. As we know,
binaries exert gravitational forces on each other. If we replace both stars in
our minds by point sources, then gravitational force and centrifugal force
balance if they are in Kepler orbits, as we have discussed previously. But this
is not the whole story, as we can see in Fig. 17.17.
Real stars are not point masses but extended objects. On star 2, with mass
M 2 for instance, point A is closer to star 1 with mass Mx than point 0 or point
B. While at point 0 gravitational and centrifugal forces balance, this is not
true of point A. At point A the gravitational force Fg working on 1 g is
while the centrifugal force is the same for all points on M x . (During one orbit,
points 0, B and A all describe a circle with the same radius d2 if M 2 is not
rotating.)
Star 2
Star 1
Fig. 17.17. For two stars in Kepler orbits gravitational and centrifugal forces
balance for point masses at distance d = dl+ d2. For the matter at point A
which is closer to star 1 the gravitational force is, however, larger than the
centrifugal force; at point B the centrifugal force is larger than the gravitational
force exerted by Mt. Point A is pulled toward M1 while point B is pushed
away.
186
17 Explosive stars
For a Kepler orbit, the centrifugal and gravitational forces balance at

point 0, which means that if we write the equation for the forces working on
1 g of material we find
we can then calculate, for point A, the difference between gravitational and
centrifugal force AF(A), namely
or
according to (17.8). For R2 (di + d2) we find from the Taylor expansion
with
and
that
(17,2)
and the difference in (17.11) becomes
This means we have a surplus gravitational pull on point A. At point B the

situation is similar, though with the opposite sign. The gravitational force is
now smaller by the same amount. At point B the gravitational force is less
than the centrifugal force. The net effect is a surplus gravitational force at
point A, pulling out point A from star 2, and a surplus centrifugal force
pulling out point B from star 2. In other words, the gravitational pull with
which point A and point B are attracted by star 2 is reduced, which makes the
star more extended in the direction connecting the two stars. This is what we
call the tidal bulges (see Fig. 17.18).
If the stars rotate once during one orbital period (as our moon does), then
the tidal bulges always stay at the same place on the stars. However, if the
stars have different rotational and orbital periods, then the bulges have to
17.2 Novae
187
move around the stars during rotational and orbital periods. This takes
some energy which has to come from the rotational energy. In fact, the bulges
do not move quite fast enough; they lag behind because of friction. The
deformed star then exhibits a torque on the other star in the sense that it tries
to equalize the rotational and orbital periods. In the long run, rotational and
orbital periods will become the same, but it takes time.
It can also be shown that the tidal forces will generally make the orbits
circular. For close binaries which have been around for some time, we can
therefore expect circular orbits and synchronized orbital and rotational
periods. Both stars will then orbit their center of gravity with constant
angular velocity co.
Let us now ask the question, what will happen to a small third body, i.e., a
dust grain or one asteroid or an atom P, which is also in the system.
Let us write down the equation for the forces acting on the mass point P.
We describe the force F by means of the potential by setting
F = gradiA
(17.14)
which defines \j/. The surfaces for which \j/ = constant are called the
equipotential surfaces. Equation (17.14) shows that along these surfaces
there is no force component, the forces are always perpendicular to the
equipotential surfaces. For the point P the gravitational potential with
respect to each mass M is \j/g = GM/r, where r is the distance to the mass M.
The sum of the gravitational potentials for the point P is then
^
(.7.15,
where r1 and r2 are the distances of P from Mx and M 2 respectively (see

Fig. 17.19).
In a rotating system we also have to consider the centrifugal force. Its
potential is
<Ac=+^-r 2 ,
Star 2
(17.16)
Star 1
Fig. 17.18. Due to the geometrical extent of the stars there is an imbalance of
gravitational and centrifugal forces at the 'inside' and at the 'outside' surfaces,
leading to tidal bulges on the stars at the places where the surface material of
the stars is closest and furthest apart.
188
17 Explosive stars
where r is the distance from the center of gravity S, around which the whole
system is orbiting. The total potential is therefore
+^ .
(17.17)
The condition \\i = constant describes the equipotential surfaces. There are
no force components in the equipotential surfaces. In the case of one spherical
star when we consider only gravitational forces the equipotential surfaces
are spheres around the center of the star. For a plane parallel atmosphere the
horizontal planes are equipotential surfaces. There are no forces in the
horizontal direction. The equipotential surfaces tell us in which direction the
force acts. In the case of the two masses Mt and M 2 working on the mass
point P and in the presence of the centrifugal force the equipotential surfaces
are rather complicated. In the neighborhood of each star they are almost
spheres around each star, because the other star is far away and of little
influence. The further away we go from each star the more distorted are the
equipotential surfaces, as can be seen in Fig. 17.20. For one equipotential
surface we find a crossing point between the two masses. This singular point
is called the Lagrange point Lv In the coordinate system of the orbiting
binaries the forces on a point P in this location are zero, which means a mass
point could reside in this location in theory for ever. If, however, a small
disturbance brings it closer to the mass M x it will fall toward M 1? if a small
disturbance brings it closer two M 2 it will fall towards M 2 . A mass point at
Lx would be in a very unstable equilibrium. The equipotential surface which
goes through Lx is called the Roche lobe. We show in Figure 17.20 a plot of
Fig. 17.19. In a close binary system with circular orbits around the center of
gravity S, the gravitational potential at point P is given by the sum of the
gravitational potentials for each star determined by the masses of the stars and
the distances rx and r2 from these stars. The potential for the centrifugal forces
is determined by the distance r from the center of gravity around which the
whole system is rotating.
77.2 Novae
189
the Roche lobe for the case that the mass ratio of Mx to M 2 is 2 to 3.
This Roche lobe is very important for all discussions of close binaries. In
the following we shall call the distance of M 2 from L x the length l2 and the
distance of M1 to Lx the length lx. As long as the radius of M 2 is less than l2
and the one of M1 is less than lx nothing will happen, the stars are well
separated. But suppose that M 2 is a white dwarf with M ~ 0.8 M o , and M x is
an M star with 0.5 M o . Then M 2 is well within its Roche lobe, but M1 is
much larger and may actually have a radius which is larger than lx. In this
case the star reaches beyond Lx and the outer layer of the star is pulled over
by M 2 . The star loses mass to M 2 . Such a system is called a semi-detached
system. The material will not generally fall straight into the star M 2 but has
some angular momentum and therefore spirals around the smaller star and
forms a disk. We call this an accretion disk.
In the disk the matter does not flow in a strictly orderly way. The incoming
material creates turbulence and friction. So some of the energy is transformed into heat, especially at the point where the incoming stream hits the
disk. At that place, the so-called hot spot is formed, (see for instance Shu
1982, p. 194). But also within the disk there is some friction and the angular
Fig. 17.20. Equipotential surfaces in a binary system are shown. The thick line
shows the Roche lobe with the Lagrange point Lx. Near the stars the
equipotential surfaces are nearly spheres. Outside of the Roche lobe they are
rather complicated. The arrows show the direction of net force working on a
particle at that location. At the Lagrange point L the direction changes
abruptly. (Adapted from Shu 1982.)
190
17 Explosive stars
momentum will slowly be destroyed. The matter then falls into the white
dwarf, as shown in Fig. 17.21.
This is the picture which people currently have about a prenova. What is
seen is the light of the blue white dwarf plus the light of the hot disk and hot
spot, making the object rather too bright for a white dwarf.
Suppose now that Mx is a white dwarf and it collects matter coming from
M 2 , which is mainly hydrogen. The white dwarf itself consists mainly of
carbon with a thin helium layer on the outside. (We will see in Volume 3 why
this is so.) No nuclear burning takes place in the white dwarf. It actually cools
down while losing energy at the surface. The white dwarf M x now accretes a
layer of hydrogen from M 2 on its surface. The thicker this layer becomes, the
higher will be the temperature at the bottom of this hydrogen layer. (For
white dwarfs the temperature in the interior is still very high, possibly around
107 K). We remember that in a white dwarf the density is extremely high.
Once the hydrogen atoms are no longer directly at the surface of the white
dwarf they get into very high density regions. At such high densities the
nuclear reactions start at lower temperatures than they would at solar
densities. When the temperature at the bottom of the hydrogen layer
becomes just high enough to start burning the hydrogen, nuclear energy is
released. The layer heats up, the nuclear reactions work faster and the energy
production increases, which results in a further increase of the temperature.
Because of the high densities in the white dwarf gas the gas is degenerate,
which means the gas pressure does not increase with increasing temperature.
Therefore the hydrogen burning layer remains in hydrostatic equilibrium in
spite of the increasing temperature, and the nuclear reactions proceed still
faster leading to an exponential increase in temperature. If the temperature
becomes high enough the degeneracy of the gas will be removed. (For each
Mass-transfer stream
Fig. 17.21. The Roche lobe is shown for a binary system in which the larger
left-hand star fills its Roche lobe. At the singular point Ll mass can flow from
the large star to the small one, which is much smaller than its section of the
Roche lobe. (From Shu 1982.)
17.2 Novae
191
density there is a well-defined temperature for which this will happen.) Once
this temperature is reached, the gas pressure increases again with increasing
temperature and the pressure forces exceed the gravitational forces. The nuclear burning region expands and cools due to the expansion. The nuclear
reactions slow down. A new equilibrium with a non-degenerate burning
zone is set up. It is believed that the runaway burning in the accreted
hydrogen and the subsequent expansion are the origin of the nova
explosions. We must, however, admit that there are still serious problems for
the theoreticians who actually want to reproduce a nova explosion on the
computer. A large carbon over-abundance is required to make the nuclear
reactions work fast enough and the chemical abundances observed in a nova
explosion are not reproduced theoretically. Nevertheless this scenario seems
to be the only viable one proposed so far.
18
Our sun
18.1
Introduction
The sun is, of course, the star nearest to us and is, therefore, the beststudied star. We have mentioned the sun several times as an example when
we talked about distances of stars, effective temperatures and masses of stars,
as well as angular radii. For all these studies we considered the sun to be just
one of the normal stars, which it most probably is. It is the most thoroughly
studied normal star. The sun is also the only star for which we can get high
spatial resolution, which enables us to observe fine details on the surface
which we will not be able to observe on other stars, at least not in the
foreseeable future. These high spatial resolution studies of the sun reveal
many features and processes which may well be also going on in other stars,
but which we are not able to study in any other celestial object. Some of these
features are the solar chromosphere and corona, though ultraviolet
observations by means of satellites now permit us also to study global
properties of these outer layers of stars other than the sun.
Another such phenomenon, which we can study in detail only in the sun, is
the solar activity, which means flares, sunspots, and the whole solar cycle of
activities. Again, observations with very high resolution and very sensitive
receivers now permit us to study global effects of activity in other stars which
seem to show activities similar to the sun. We would never know how to
interpret these observations if we did not have the sun nearby so that we can
study the solar surface in detail.
18.2
The surface of the sun
How does the sun look at higher spatial resolution? Not at all
smooth and homogeneous, as one might expect for a hot gas ball.
First we see some large, dark spots, the so-called sunspots (see Fig. 18.1).
We will talk about those later. Outside of the sunspots, we see a network
193
18 Our sun
194
called the solar granulation. It seems to be a network of dark structures with

brighter blobs in between. According to what we discussed previously, dark
means less radiation, which means a lower temperature. So we see regions
with different temperatures on the solar surface. The spots are the coolest
regions. The effective temperatures of the spots are about 3700-4500 K. For
the granulation, the temperature difference between the dark and the bright
region is only ~ 200-400 K, with an average temperature of about 5800 K,
as we saw previously.
We will understand the solar granulation better after we have looked at a
solar spectrum with very high spectral and spatial resolution (see Fig. 18.2)
*
/
'r
i
s
*
Fig. 18.1. A photograph of a small section of the sun is shown with very high
spatial resolution. We see the dark umbra of a sunspot and the somewhat
brighter penumbra around the umbra. In addition, we see small dark pores.
The rest of the solar surface is not smooth at all; it is covered by a pattern of
bright so-called 'granules' imbedded in the intergranular network. See also
Fig. 18.4. (From Unsold 1982; photograph by Schwarzschild using the
stratospheric telescope.)
18.2 The surface of the sun
195
Because of the very high wavelength resolution, the spectral line is now
rather broad. The interesting thing is that it is no longer a straight line but
rather wiggly. Remember that in Section 9.2 we talked about the Doppler
effect? That the spectral lines are shifted towards longer wavelengths when
the light source is moving away from us, and towards shorter wavelengths
when the light source is moving towards us? This is obviously what we see in
these wiggly lines: some parts of the solar disk are moving towards us while
other parts are moving away from us. Moving towards us means rising in the
solar photosphere, and moving away from us means falling into the solar
Fig. 18.2. A narrow wavelength band of the solar spectrum is shown with very
high spectral resolution and very high spatial resolution along the spectrograph
slit (see Fig. 18.4). At such high spatial and spectral resolution the solar spectral
lines are not at all straight lines. They are quite wiggly due to the Doppler
shift, which is different for different spots of the sun seen at the different
locations along the spectrograph slit. The continuous spectrum between the
lines also looks streaky. It is brighter at the locations of the bright spots on the
sun - the granules - and darker in between at the intergranular regions. When
looking at the wiggles of the line profile we see that the Doppler shift goes
preferentially to the left in the bright regions and preferentially to the right in
the dark regions. In the photograph the wavelength increases toward the right.
For the dark regions the velocities are therefore pointed away from us, while
for the bright granules they are pointed toward us. (From Zirin 1966.)
196
18 Our sun
photosphere (see Fig. 18.3). The spectrum in Fig. 18.2 was taken by having
the slit of the spectrograph cover a small area on the solar disk, as shown in
Fig. 18.4. Along the lengths of the slit we see the spectra of different elements
of the granulation. At the point where we see a bright granulation element,
we have more light in the spectrum; at places where we have a dark portion
of the granulation pattern in the slit, we have less light in the spectrum.
- Observer
Fig. 18.3. Solar material moving towards us is rising in the solar photosphere,
while material moving away from us is falling into the solar surface.
% *s
I \
h*
i
-.. *
jr
Fig. 18.4. A small section of the solar surface with the solar granulation is
shown. Such an image of the sun is projected onto the spectrograph slit, which
is outlined in the figure. Along the slit different bright and dark spots are seen.
The spectrograph produces a spectrum of each spot along the slit. In this way a
spectrum with high spatial resolution along the slit can be obtained, as seen in
Fig. 18.2.
18.3 The outer layers of the sun
197
A careful look at the spectrum in Fig. 18.2 shows that bright streaks,
showing the spectrum of a bright granule, usually show a blueshift of the
lines i.e., a wiggle toward shorter wavelengths, while dark streaks, showing
the spectrum of dark areas on the solar surface, show lines with wiggles
towards longer wavelengths, at least in most cases.
The solar granulation then tells us that we have a pattern of upward and
downward moving columns, the upward moving gas columns being brighter
and therefore hotter than the downward moving ones. The diameters of
these columns appear to be 500-700 km, but we have to remember that the
turbulence in our Earth's atmosphere, limits our resolution to 1/2-1 second
of arc. One second of arc corresponds to 700 km on the solar disk in the
center of the sun. If the granulation has, indeed, a smaller size pattern, we are
not able to resolve it. Satellite telescopes, which avoid the problem of seeing,
so far, have had too small an aperture to resolve much smaller structures.
The space telescope will be too light sensitive to be used on the sun which
would burn it out.
18.3
The outer layers of the sun
18.3.1 The solar chromosphere

Before the time of rockets and satellites, the outer layers of the sun
could only be observed during solar eclipses. The sun proper emits about
10000 times more light than the chromosphere and one million times more
than the corona, of which we show a picture in Fig. 18.5.
Fig. 18.5(a). The solar corona as seen during a solar eclipse. On the right-hand
side a solar X-ray picture, taken shortly after the eclipse, is photomontaged on
the shadow of the moon. X-ray emission is strong in regions of large solar
streamers. (From Gibson 1973.)
198
18 Our sun
The deepest of these outer layers becomes visible during solar eclipses just
after second contact, when the solar disk proper is just covered up (see
Fig. 18.6). This layer is called the chromosphere because it looks rather
colorful, reddish. The spectrum of the chromosphere looks very different
from that of the photosphere. The Fraunhofer lines turn into emission lines
and the lines in the chromospheric spectrum are brighter than the
neighboring continuum. The spectrum is also called the flash spectrum
because it is visible for only a few seconds. In Fig. 18.7 we show such a flash
spectrum. This spectrum was taken with a slitless spectrograph so we see an
image of the solar limb for each emission line. (Usually we see an image of the
slit of the spectrograph illuminated by the star which gives us spectral lines
which are images of the spectrograph slit in the different colors.)
When the moon covers up more of the deeper chromospheric layers, the
Fig. 18.5(6). The solar corona during sunspot minimum and maximum. (From
Gibson 1973.)
1 st contact
2nd
3rd
4th contact
Fig. 18.6. We show the phases of a solar eclipse, with the numbering of the
contacts given. The angular size difference between the sun and the moon is
exaggerated.
It*
Fig. 18.7. A solar flash spectrum is shown taken shortly after third contact. The image of the solar chromosphere is seen in
the different spectral lines. The corona is visible all around the moon's shadow, the coronal lines at 5303 A and 6374 A are
therefore seen as complete circles. (From Unsold 1938. Photograph by Davidson & Stratton 1927.)
200
18 Our sun
higher, less luminous layers become visible. Lines of more highly ionized
particles become visible, showing the increasing temperatures of the higher
chromospheric layers. In fact, one can study the temperature and pressure
stratification of the chromosphere by first studying the spectrum from the
highest layers, observed last. Knowing what the emission is from these
layers, we can subtract this contribution from the spectrum of the next
deeper spectrum to get the emission from the deeper layer only, etc. Of
course, we have to take into account that we always look tangentially along
the solar limb. Our line of sight always cuts through all the layers (see
Fig. 18.8), but knowing what the emission of the highest layers is, we can take
that into account.
In Fig. 18.9 we show the temperature structure in the solar chromosphere
and in the overlying transition layer, as derived from such studies of the solar
chromosphere during solar eclipses. The temperature rises rather steeply in
these layers until it reaches more than a million degrees in the corona. The
region with temperatures between 30 000 K and 106 K is called the transition
region (from the chromosphere to the corona).
Fig. 18.8. When observing the solar chromosphere during a solar eclipse, our
line of sight always cuts through all layers of the chromosphere and corona.
O)
10
log (h-h 0 )
12
Fig. 18.9. The temperature in the outer layers of the sun is shown as a function
of height above a reference level, h0, called the 'surface' of the sun. (From
Bdhm-Vitense 1985.)
201
The sun is a rather cool star, therefore the photospheric light is quite dim at
short wavelengths, which means in the ultraviolet. At these wavelengths the
emission from the solar chromosphere is actually stronger than the light
from the solar disk because the chromosphere has a higher temperature. This
ultraviolet light can not be seen from the surface of the Earth because the
atmosphere absorbs all the ultraviolet light, but from rockets and satellites it
can be measured. We can therefore now also observe the chromospheric
spectrum without an eclipse. This fact also enables us to observe chromospheres and transition layer emission from other stars for which previously no
chromospheric or coronal light emission could possibly be observed. We did
not know whether our sun was the only star to have a chromosphere and
corona, or whether all cool stars have these high-temperature outer layers.
We know now that cool main sequence stars with spectral types earlier
than M5 have chromospheres and coronae, and all cool giants and the cool
main sequence stars have at least chromospheres.
183.2
The solar corona
After second contact during solar eclipses, the chromospheric

spectrum remains visible for about 20 s, after which time the whole
chromosphere is also covered up by the moon and only the corona remains
visible. Photographs of the solar corona show that it looks very different at
different times (see Figs 18.5 and 18.6), depending mainly on the number of
sunspots which are visible on the disk at that time, of course before or after
the eclipse. During sunspot maximum the corona is large and looks almost
circular, while during minimum it is only extended in the equatorial region,
the region where the spots are mainly found. Large rays are seen showing us
the directions of the magnetic field lines in the corona. In the equatorial
regions the magnetic field lines show loops, while in the polar regions the
magnetic field lines point radially outwards and the field lines look 'open.'
The spectrum of the corona shows spectral lines from ions which have lost
many electrons, up to 20, indicating temperatures up to 2 or even 3 million
degrees.
Such high temperature plasma also emits X-rays which can also be
observed from rockets and satellites. We can actually take X-ray photographs of the sun, as seen in Fig. 18.10. This picture shows that the X-ray
emission from different coronal regions is very different. It is generally larger
in regions of large sunspot groups. There are large regions on the solar
surface where the X-ray emission is so small that the X-ray photograph looks
dark. These regions are called coronal holes. They have actually nothing to
202
18 Our sun
do with holes, but are just too cool or have too low densities to emit X-rays.
They occur mainly in regions of 'open' field lines which reach very far out
from the sun before they close, and from which matter can more easily escape
from the solar surface than in regions with closedfieldlines, i.e. in the loops.
183.3
The energy source and the heating of the chromosphere and

corona
We said earlier that heat transport can only be from higher

temperature regions to cooler ones. Since the sun loses energy at the surface,
which has to be replenished by heat transport from the deeper layers, the
temperature in the sun has to decrease outwards. We cited the absorption line
spectrum as proof of such a temperature gradient outwards. How then can
the temperature increase in the very outer layers of the sun? How can the
solar radiative heat transport be from the relatively cool photosphere to the
high temperature corona? The answer to the latter question is that the heat
does not flow to the high temperature corona, it only passes through it. The
hot corona contains so little material that the photospheric radiation does
not notice the existence of this material. There is essentially no absorption of
photospheric radiation in the corona. The optical depth of the corona is of
Fig. 18.10. An X-ray image of the sun showing a coronal hole extending right
across the solar surface. In this region the X-ray emission is very low. (Photo:
NASA, courtesy of L. Golub and Harvard-Smithsonian Center for
Astrophysics. From Gibson 1973.)
203
the order of 10" 1 0 . The average optical depth of the chromosphere is about
10~ 4 . These layers do not get their energy from the radiation which comes
from the deeper layers and is absorbed in these layers, as is the case for the
solar photosphere. What then is the heating mechanism for the chromospheres and coronae? This question has stimulated an intense study, especially
during the last decade when the International Ultraviolet Explorer (IUE)
satellite and the X-ray satellite, the Einstein Observatory, were launched.
The answer to the question for the specific sources of energy in the different
layers can not yet be given with certainty; the discussions are still going on.
I think it is generally agreed that the ultimate energy source is the so-called
hydrogen convection zone which is situated just below the visible layers of
the photosphere. We saw the granulation picture. The Doppler shifts of the
wiggly lines show velocities of the order of l-2km/s, which means about
5000 km/h. These velocities are generated in the hydrogen convection zone
(see Volume 2). In these strong velocity fields, much noise is generated, which
means acoustic waves, which in the low density chromospheric layers
develop into shock waves. They may also lead to the generation of
magnetohydrodynamic waves in the general solar magnetic fields and
especially in the strong magnetic fields around the sunspots.
These forms of energy are not heat energy, but are in brief called
'mechanical energy.' Waves are influenced by the temperature structure, but
they can propagate in any direction regardless of the temperature gradient.
Their propagation is mainly influenced by the density gradient and by the
magnetic fields. The acoustic waves may dissipate some energy in the low
chromospheric layers. When they develop into shock waves they dissipate
their energy much faster. Magnetohydrodynamic waves cannot dissipate
their energy as fast in the chromosphere. They may be the vehicle by which
mechanical energy can be transported into the corona where they may be
partly transformed into waves similar to acoustic waves which may possibly
heat the corona, though other heating mechanisms are also possible. We do
not yet completely understand the details of these wave generations and
dissipations, but we do know that there is enough energy in these waves to
replenish the energy loss in the chromosphere and corona of the sun and
similar stars. As in the photosphere, the resulting temperature
stratification is determined by a balance between the energy input and the
energy loss. In the high layers coronal temperatures around 1 or 2 million
degrees are reached. These high temperature regions are the origin of the socalled 'solar wind'. The solar wind energy increases steeply with increasing
temperature of the corona. It takes out this energy and is therefore a very
efficient cooling mechanism.
204
18 Our sun
18.3.4 The solar wind

The solar wind was first discovered by L. Biermann who studied the
tails of comets. The ionized parts of these tails are always pointed away from
the sun, as if something pushes them out. He calculated that the radiation of
the sun alone was not strong enough to do it, so there had to be some
material flying away from the sun at high speeds, pushing the comet tail out.
When rockets studied the motions and magnetic fields in the Earth's
magnetosphere, they found that the magnetic field is deformed by a stream of
matter coming from the sun at high speeds, 500 km/s. This stream got the
name 'solar wind.' The speeds are really more than a wind, but the stream has
a very low density, ~ 10~ 2 3 gcm~ 3 .
In solar surface regions with closed magnetic field lines the solar wind has
to fight the magnetic forces. Ionized material cannot flow across magnetic
lines of force. (Charged particles spiral around the lines of force). The solar
wind cooling is therefore less efficient in regions with closed lines of force,
where the temperatures therefore remain high. In regions of'open' field lines
the cooling by the wind is much more efficient and the temperature is
expected to be lower in such regions of the stellar surface. This explains the
existence of coronal holes, the regions of low X-ray emission in the corona.
18.4
The active sun
18.4.1 The Butterfly diagram

In a previous photograph of the sun (Fig. 18.1) we saw a dark
sunspot. Each of these sunspots lasts for about a month and then disappears
while new ones show up. The number of sunspots, and their position, varies
with time.
The following diagram (Fig. 18.11) called the 'Butterfly' diagram shows the
position of sunspots on the sun as a function of time. The ordinate is the
latitude on the sun. The number of sunspots shows a cycle of approximately
11 years duration. For a new cycle, a few spots show up first at rather high
latitudes, ~40, and the next new spots occur closer to the equator. The
sunspots themselves do not move toward the equator, but the zone does.
While the last spots from a given cycle are still present near the equator, the
first spots of the new cycle occur at higher latitudes.
What are these dark spots? Why are they dark? The first insight was gained
when Hale found in 1908 that the sunspots were regions of large magnetic
fields.
18.4 The active sun
205
Fig. 18.11. The solar latitude of sunspot positions is shown as a function of

time. The first sunspots of a new cycle occur at latitudes around 40. Later in
the cycle sunspots occur at lower latitudes. The last sunspots of a cycle occur
rather close to the equator, while the first spots of the new cycle appear already
at latitudes around 35. Because of its appearance this diagram is called the
butterfly diagram. (From Unsold 1938.)
aph slit
. 18.12. Sunspot spectra obtained with the magnetograph are shown. The vertical direction is the space coordinate along the
ctrograph slit. The dark regions show the position of the sunspot. The horizontal coordinate is the wavelength of the light. The
t seen in two adjacent spectra comes from the same area on the sun but is of opposite direction of circular polarization (see
apter 16). The Zeeman effect due to the magnitude field in the sunspots is clearly seen in the shift of the spectral lines of the
ctra of opposite direction of polarization, indicating magnetic fields of several thousands Gauss. (From Kiepenheuer 1953.)
18.4 The active sun
207
18.4.2 The magnetic fields in sunspots

In Section 15.2 we discussed how we can measure magnetic fields in
stars by means of a magnetograph first constructed by Babcock. This
instrument makes use of the Zeeman effect: if a light source is put into a
strong magnetic field, the spectral lines split up in several components, the
amount of splitting increasing with the strength of the magnetic field. The
wavelength difference between the various components of a line can then be
measured and the field strength determined. The direction of the field can
also be determined because the pattern of the components, i.e., their relative
strengths and the way the light is polarized, changes with the direction of the
field relative to the line of sight.
Measurements of the Zeeman effect in sunspots revealed that sunspots
have magnetic fields of several thousand Gauss in the darkest part which is
called the umbra (see Fig. 18.1). In Fig. 18.12 we see a spectrum of a sunspot
in which one strip shows the spectrum of one direction of polarization,
and the adjacent strip shows the other direction of polarization. The next
pair of spectra shows the adjacent area of the sunspot. The line shifts of two
adjacent spectra clearly show the presence of a magnetic field.
The field appears to be mainly longitudinal, which means vertical in the
umbra, but curves around in the penumbra. We do not understand at all
why there is such a sharp boundary between umbra and penumbra and the
surrounding area, nor do we understand the difference in brightness between
umbra and penumbra.
When studying the polarity of the magnetic fields, it is found that whenever
there are two spots closely associated, they have opposite polarities. In large
sunspot groups, one also finds opposite polarities in the two halves of the
group. The interesting thing is that during one sunspot cycle the north
Fig. 18.13. During one 11-year sunspot cycle the magnetic north polarity is
always in the preceding spot on the northern hemisphere and in the following
spot in the southern hemisphere. In the next 11-year cycle the polarities are
reversed.
208
18 Our sun
polarity is always on the preceding spot (preceding with respect to the solar
rotation) and the south polarity on the following spot for one hemisphere
and opposite for the other hemisphere. During the next cycle the polarities
are reversed (see Fig. 18.13). In reality, the sunspot cycle is therefore
~ 22 years if we include the magnetic polarity variations.
18.4.3 Differential rotation of the sun
The rotation rate of the sun can best be measured by measuring the
sunspot velocity across the disk (see Fig. 18.14). The average rotation period
of the sun is nearly 26d, almost identical to the orbital period of the moon.
However, the spots closer to the equator rotate somewhat faster than those
further away. This is called the differential rotation of the sun.
18.4.4 Possible explanation of the sunspot cycle
At solar temperatures and densities several of the heavy elements like
iron and silicon lose one electron, therefore the solar material contains a
large number of free electrons. This makes it a very good conductor for
electrical currents. Such gas with charged particles is also called a plasma. A
plasma can not move through magnetic lines of force because the electrons
always have to spiral around the magnetic lines offeree. If a plasma wants to
move perpendicularly to a magnetic field it has to take the magnetic lines of
force along. We say the magnetic lines of force are frozen into the material.
Based on this information we can give the following very simplified
explanation of the solar cycle.
We have just seen that the solar equator is rotating faster than the higher
latitudes. Due to this differential rotation, the magnetic lines of force are
wound up as shown in Fig. 18.15. The direction of the lines offeree is reversed
in the northern and southern hemispheres.
If such a magnetic 'rope' breaks through the solar surface in different
regions, pairs of magnetic spots are seen for which the direction of polarity is
always the same (see Fig. 18.16). Since in the northern and southern
hemisphere the field directions along the 'ropes' are opposite, the northsouth polarity of the spot pairs is also reversed.
In the next solar cycle the overall solar magnetic field is reversed and the
polarities are therefore all reversed but remain opposite in the northern and
southern hemispheres.
We want to emphasize that this is a very simplified picture of the processes
involved, but we can at least understand the main features in this way. To do
18.4 The active sun
209
it better we have to apply the dynamo theory which for non-rigid rotation
gives periodic changes of the overall magnetic field.
18.4.5 The temperatures of the sunspots
Since the sunspots are dark, the gas in these magnetic regions must
be cooler than in the surroundings. What makes the sunspots cool? To cool
L
L
I
k<
V *;-;
k %
A
k
k
1
llr
i
A
1
A *
A
A *". A
Ik <- 1
JL
ML mL
J
Am to
' Jm
Fig. 18.14. The rotation rate of the sun can be seen in the motions of the
sunspots around the sun. Spots close to the equator move somewhat faster than
those at larger latitudes. An average rotation period of 28 days is found. (From
Abell 1982.)
210
18 Our sun
the gas we must either take out more energy at the surface or put in less
energy at the bottom. Astronomers cannot quite agree on the correct
explanation. The suggestion by Biermann (1941) was that less energy is being
fed into the sunspot gas at the bottom. We said earlier that below the surface
of the sun wefinda convection zone. The moving material transports energy
to the surface. In the strong magnetic fields of the sunspot, the material
cannot move easily, so the convection probably transports less energy to the
surface. That could leave the sunspots cooler.
On the other hand, Parker (1974) maintains that this cannot be the correct
explanation. He thinks there must be additional cooling. It cannot be by
light radiation, since the sunspots emit less light than the surroundings.
Perhaps the magnetic field leads to some kind of magnetohydrodynamic
waves which transport energy out of the sunspots? Much more research
needs to be done in this field.
If Bierman is correct, then only stars with convection zones could have
dark spots. If stars without convection zones have dark spots, then it seems
that Parker must be right. There seems to be observational evidence that
Fig. 18.15. The magnetic lines of force on the sun are 'frozen' into the material.
They are wound up due to the differential rotation of the sun. The horizontal
components of the magnetic lines of force are then in opposite directions on the
northern and southern hemispheres.
Fig. 18.16. When the magnetic ropes seen in Fig. 18.15 break through the solar
surface, sunspot pairs of opposite polarities are seen, always with the same
orientation in one hemisphere but opposite in the southern and northern
hemisphere.
18.4 The active sun
211
stars without convection zones may also have dark spots. It is not known
whether those might be magnetic spots.
18.4.6 Stellar activity and stellar rotation
We saw in Chapter 13 that stellar rotation rates decrease abruptly
for stars cooler than spectral type F5. For these cooler stars the rotational
velocities are in fact so small that they cannot be measured from the line
widths anymore. It appears that we can measure the rotation rate of these
stars from starspot activities similar to the one observed for the sun. In active
solar regions where a large number of spots are seen, we also see in the center
of the strong Ca + lines at 3933 and 3968 A a weak emission line. This line is
stronger in active regions than for the overall solar spectrum. When, due to
solar rotation, these active regions rotate to the back side of the sun where we
cannot see them, then in the overall spectrum of the sun the emission in the
center of the Ca + lines becomes weaker. For the sun this effect is difficult to
measure because the emission is so weak, but for many cool stars this
emission is much stronger.
When we measure the stellar Ca + K line emission at 3933 A as a function
of time, we can frequently see variations which are semi-periodic. This is
understandable if the star is rotating. When the activity region disappear
behind the star, the emission is reduced; when the spots appear again on the
other limb, half a rotation later, the emission becomes strong again. When
the spot finally disappears, the emission from this region cannot be seen any
more, but in the meantime there may be another active spot group at another
place on the star for which the emission can be observed. But this group may
already be further advanced on the stellar disk, so there will be a phase shift
between the rotation of the first spot and the one for the second spot. This
causes deviations from strictly periodic behavior. Nevertheless, rotation
periods have been measured in this way. It is found that stellar chromospheric activity generally increases with increasing rotation rates of the stars.
Only early F stars seem to be an exception, the reason for this is not yet
understood.
19
Interstellar absorption
19.1
Introduction
Even though this volume is dedicated only to stellar observations

and the theory of stellar structure, we have to talk a little about the
interstellar material, which means the material which is between the stars,
because the light of the stars, especially of those which are very far away, has
to pass through this interstellar material before it reaches us. Just as in the
Earth's atmosphere, the star light is influenced by interstellar absorption,
also called interstellar extinction. We cannot observe all the light which is
emitted from the star because the interstellar gas and the interstellar grains,
called dust, have absorbed part of the light. Just as for the Earth's
atmosphere, we have to correct for interstellar absorption. For the Earth's
atmosphere we can determine the extinction if we observe the star at different
zenith distances, which means for different path lengths, through the
atmosphere. For interstellar extinction we cannot do this since the path
length through the interstellar medium is always the same. We have to find
other ways to determine the influence of the interstellar extinction.
Basically there are two components of interstellar extinction: the absorption by the grains and the absorption by the gas. These components are
usually, but not necessarily, associated. In the following paragraphs, we will
discuss both components and see how we can find out about interstellar
absorption.
19.2
The interstellar dust
19.2.1 General appearance
When looking at photographs of the Milky Way, we see regions with

a large number of stars and, right next to them, regions where there are
hardly any stars.
213
214
19 Interstellar absorption
Dark patches intermixed with bright matter can be seen in many parts of
the sky (see, for instance, Fig. 19.1). On the small-scale photograph of the
Milky Way seen in Fig. 19.2 a dark lane cuts the Milky Way almost into two
halves.
The question is, are the stars distributed so inhomogeneously, or is there
something covering up the stars? When we try to find external galaxies, we
Fig. 19.1. In the sky, dark regions with very few stars are seen next to brighter
regions with very many stars. In this picture we see the so-called Horsehead
Nebula. (From Abell 1982.)
^k%
* ^D
Fig. 19.2. A composite photograph of our Milky Way is shown. The border
lines for different photographic plates can clearly be seen. We also see large
dark lanes cutting through the Milky Way almost cutting it in two halves.
(From Shu 1982.)
19.2 The interstellar dust
215
have trouble finding any in the plane of the Milky Way, which means in the
plane of our Galaxy. For this reason the plane of the Galaxy, or of the Milky
Way, is called the zone of avoidance. This strongly suggests that there is
indeed absorption in the plane of the Galaxy because the external galaxies
have no reason to care about the plane of our Galaxy. How can we decide
with certainty whether the dark regions are indeed regions of strong
absorption and not holes in the Milky Way? We can find out by statistics, by
means of the so-called Wolf diagram.
19.2.2 The Wolf diagram
Let us look at a certain region in the sky and thereby observe all the
stars in a given solid cone with an opening Aco. All these stars are projected
against the given area in the sky (see Fig. 19.3). We study only stars of a given
kind, say only AOV stars or, in other words, stars with spectra, like Vega. All
such stars have the same absolute magnitude, M v . (There are more luminous
AO stars around like giants and supergiants, but we saw that they can be
recognized from their spectra.)
Let us assume that the stars are distributed homogeneously in this cone
and that the number of AOV stars per pc 3 is n. In any volume V9 the number of
AOV stars is then N = n-V. We now consider the part of the spherical shell at
distance d with thickness Ad, which is cut out by the cone (see Fig. 19.3). The
whole sphere is described by a solid angle 4TC. The volume of the part of the
sphere with solid angle opening Aco is therefore given by a fraction ACO/4TT,
which means
V=47id2AdA(o/4n.
(19.1)
Fig. 19.3. Observing all the stars in a given area in the sky corresponds to
observing all stars within a cone with an opening of solid angle Aco, whose
size is determined by the extent of the observed region in the sky.
216
The number of AOV stars in this volume is then

N(d) = n-V= n-d3'Aco'^.
(19.2)
These stars all have an apparent visual magnitude

mv = M v + 5 1 o g d - 5 ,
(19.3)
or, for a change in distance by A log d we find a change in apparent visual

magnitude by
d. (19.4)
(A given Amv corresponds to a given A In d = Ad/d = Amv/2.171.) For a given
mv we have then, according to (19.3)
where d is obtained in pc. M v is known from nearby AOV stars like Vega; mv is
measured.
The number of AOV stars with apparent magnitudes between m v and
mv + Amv is given by the number of stars with distances between d and
d + Ad, for which the relation between Amv and Ad/d is obtained from
(19.4). From (19.2) we find
logAT[mv(d)] = logn + 3-logd + log(Ao;-Alnd),
(19.6)
if we assume a homogeneous stellar density with n stars per unit volume,

which means n = constant. For a given choice of A In d corresponding to a
chosen Amv for the magnitude interval for which we count the stars, all terms
on the right-hand side of (19.6) are constant except for the log d term, which
depends on the mv for which we count the stars. We can therefore rewrite
(19.6) in the form
log JV[mv(d)] = 3-logd 4- const,
(19.7)
with
const = log Aco + log yrfr
+ log n.
Replacement of logd by means of (19.5) now yields

log JV [mv(d)] = 0.6 mv + Const,
(19.7a)
where the new Constant is different from the constant in (19.7) because it also
includes the term stemming from the absolute magnitudes of the stars
studied.
217
In a plot of log N versus m v we should therefore find a straight line (see

Fig. 19.4). This is the Wolf diagram.
Assume now that there is an absorbing cloud at distance d = d0, for which
mv = m vo - Suppose this cloud absorbs 75% of the light of the stars behind it,
leaving 25% of the light, which means decreasing the light by a factor of 4,
corresponding to an increase in magnitude by 1.5 for all stars behind the
cloud. For a given distance, of course, the number of stars would not change;
the N(d) for d>d0 would remain the same, but the apparent magnitude of
these stars would be increased by 1.5. The relation between log JV and rav
would then be as shown by the dashed line in Fig. 19.4. For the figure we also
assumed that the cloud would extend over a distance Ad, corresponding to
Amv = m vl mv0. Without this finite extent to the cloud, the short, dashed
line in the figure would have to be plotted horizontally because the mv of the
stars would either be changed by 1.5 or not at all. If the cloud is extended, the
stars in the cloud have intermediate shifts, depending on their position in the
cloud.
If at a larger distance d = dY there should be another cloud, we may find
another shift in the distribution, as indicated by the dash-dotted curve in
Fig. 19.4. Such a diagram can therefore not only tell us whether there is a
cloud or not, but also where it is, how extended it is, and how much it
absorbs.
log N{my)
Two clouds
my{d)
Fig. 19.4. A schematic Wolf diagram is shown. For homogeneous steller

densities we expect a linear relation between the log of the number of stars of a
given kind, log N (with a given absolute visual magnitude) and the apparent
visual magnitudes (see the solid line). If there is an absorbing intersteller cloud
between us and the stars the apparent magnitudes of the stars behind the cloud
are all increased by an amount determined by the absorption in the cloud. The
relation between log N and mv is shifted as shown by the long dashed line. A
second cloud behind the first one may give an additional shift (dash-dotted line).
218
What would we see if there was not an absorbing cloud, but instead there
was a 'hole' in the star distribution, say, between the distances d2 and d 3 ? If
these distances correspond to apparent magnitudes mv2 and mv3, then all the
stars with mv2 < mv < mv3 would be missing. For these magnitudes N would
be zero. The stars at larger distances would generally not be affected, but
would probably have the same average density. The star counts for larger mv
would therefore be the same. The relation between log N and mv would then
look like the one shown in Fig. 19.5, where we have assumed that two holes
exist: one at mv = mv2 and another one at mv = mv4. In this case the Wolf
diagram looks very different from the one with the clouds. Such a Wolf
diagram therefore permits us to distinguish between 'holes' in the stellar
distribution and an absorbing interstellar cloud. In Fig. 19.6 we show an
example of an observed diagram, clearly showing that interstellar clouds are
obscuring the stars.
19.2.3 Interstellar reddening
The study described in the preceding section can, of course, also be
done with the blue and ultraviolet magnitudes in the same way. We can then
find out how much the same clouds absorb in the blue or in the violet. Such
studies have shown that the absorption is larger in the blue and still larger in
the ultraviolet. For most clouds we find
AmB = 1.3-Amv and Am u = 1.53 Amv
log
(19.8)
/V(A77 V )
i
i
i
^
i
!
i
j
i
i
i
i
i
i
i
i
A77 V2
A77 V4
Fig. 19.5. We show the Wolf diagram expected if there are holes in the number
density of stars at distances corresponding to apparent visual magnitudes mV2
and m V4, but which would be normal again for distances corresponding to
apparent visual magnitude mV3 and mV5, respectively. In this case we would see
no stars with apparent magnitudes between mV2 and mV3 or with magnitudes
between mV4 and m V5 . There would be 'holes' in the log N(mY) plot.
219
Since the B and U magnitudes are more increased than the V magnitudes,
the colors of the stars change due to the interstellar absorption. We find,
roughly,
AmB - Amv = A(B - V) - Amv(1.31 - 1) = Am v 0.31,
(19.9)
or
(19.10)
= 3.2-A(B-V)
or
(19.11)
1 and
Amv
A ( U - B ) / A ( B - V ) = 0.72,
(19.12)
but small differences may exist between different clouds.

In order to find the relation between the color change A ( B - V ) or
A(U B), we can, of course, also directly compare the colors of the AOV stars
Schematic
3
SMon
10 +1 12 13
15 16 1Fm
Fig. 19.6. Below the schematic Wolf diagram, already shown in Fig. 19.4, we
see actually observed Wolf diagrams for a region around the star $ Ophiuchi
and around the star S Monocerotis. The stars behind the cloud are dimmer by
2.3 and 2.0 magnitudes due to the absorption by the interstellar material
in the clouds. (From Becker 1950).
220
in front of the cloud with those behind the cloud and determine the color
changes and the ratio of A(B - V) to A(U - B).
In the far infrared the interstellar absorption becomes very small, actually
close to zero. If we call a far infrared magnitude min, then the color change
A(mv min) for stars behind the cloud gives directly the change in visual
magnitude for the stars behind the cloud. In this way the ratio of Amv to
A(B V), called R, is determined. It usually turns out to be R = 3.2, but may
occasionally have values around 5 or perhaps even 6.
The A(B V) and A(U B) due to interstellar absorption are usually
called E(B V) and E(U B), where E stands for color excess.
Once we have established that we do indeed have absorption in the
interstellar medium, and once we know that the interstellar medium causes
reddening, we can use this reddening effect to correct for the interstellar
absorption. If we know what the true U B and B V colors are for
unreddened stars, we can plot the two-color diagram for these true colors
(see Fig. 19.7). In this figure we have also plotted the observed colors for
some stars. For a given U B the reddened stars have a much larger B V.
We know now that for any change in B V by a value E(B V) the change
in U - B is E(U - B) with E(U - B) = 0.72 x E(B - V). This relation defines
a direction in the two-color plot. In Fig. 19.7 we have indicated the direction
in which the stellar colors move in the two-color diagram due to interstellar
reddening. For an E(B V) = 0.3, star D would appear in the position of star
C. We can then trace backwards from where the star in position C came in
the two-color diagram before it was reddened by tracing the observed colors
E(B-V)
U-B
E(U-B)
0-
-0.3
0.6
-B-V
Fig. 19.7. The two-color diagram for the true colors of the stars is shown
schematically by the solid line. Also shown is a point (*) for the observed colors
of a star C. The direction of the reddening line E(U - B)/E(B - V) = 0.72 is also
drawn in. Following the direction to the 'true' color line of normal stars tells
us what the unreddened colors are of the observed star (point D).
221
back along the reddening line to the true color line. This gives us the true
colors of the star. Since Amv = 3.2 x E(B V), we can also derive by how
much the apparent magnitude of the star has been changed due to the
interstellar absorption and can correct for it. Suppose we have measured for
an AOV star a B - V = 0.10 and U - B = 0.07. Its true B - V = 0 as we know
for an AOV star. Its E ( B - V ) is therefore E ( B - V ) = 0.10 and therefore
Amv = 0.32. If we measured mv = 8.32, its true apparent magnitude would have
been mv0 = 8.0 if there had not been any interstellar absorption. If we want to
determine the distance to the star we can use (19.3), but we have to use, of
course, the apparent magnitude which the star would have had without the
interstellar absorption, i.e. mv0. With M v = 0.5 we find a distance modulus
mv My = 7.5, which puts the star at a distance of about 316 pc.
On average, we find an interstellar absorption of about one magnitude in
the visual for a distance of 1 kpc. The distance to the galactic center is about
8 kpc. Even without the dense dark clouds in that direction, the stars close to
the center would be expected to have an average reduction in brightness of 8
magnitudes due to the average interstellar extinction. Their colors are
expected to be changed by E(B V) = 2.4.
19.2.4 Determination of the 'true colors of the stars
How can we determine the 'true' color line for the stars? We can, for
instance, measure the colors of nearby stars. Chances are that they are nearly
unreddened. There are, however, no O or B stars nearby. If there are no
nearby stars of a given kind, the best thing is usually to measure stars which
are above the galactic plane, if possible. Stars in the plane are usually more
reddened than those above the galactic plane because the dust is confined to
a rather thin disk (see Fig. 19.8). In fact, it is this observation which tells us
Line of sight
Galactic plane
*Dust layer
Line of sight
Fig. 19.8. For a given distance stars in the Galactic plane are more reddened
than stars outside of the Galactic plane because the line of sight passes through
the dust layer for a longer distance.
222
that the dust is confined to a rather thin disk. It is distributed inhomogeneously, but rarely reaches higher than about 100 pc.
Unfortunately, the O and B stars are also strongly concentrated toward
the galactic plane, since they are formed from large dust clouds. So they lie
entirely within the dust layer. But the dust layer is patchy and for some stars
the E(B V) is smaller than for others, so we can tentatively assume that the
bluest ones have the 'true' color. For some stars there is a way to check this.
Suppose we have observed a few nearby A or F stars so that we know their
true colors. If the O and B stars happen to be near A or F stars, for instance,
in the clusters, we can then measure the reddening for the A and F stars and
hope that they will be the same for the hot O and B stars. In young clusters
the extinction is, however, patchy and may be different for different stars, as,
for instance, for the Pleiades, in which we can see patches of dust, seen as
bright nebulosity (see Fig. 19.9), which is very inhomogeneous. The best way
would be to measure colors for binaries with one O or B star and the other
component being an A or F star. Unfortunately, there are not many binaries
with O and B type components around for which the companion colors can
also be measured, mainly because O and B stars are so short-lived, and also
Fig. 19.9. On a long-exposure photograph of the Pleiades star cluster patches

of nebulosity show up in front of the stars, due to intersteller dust. This
interstellar dust appears bright because it reflects the starlight in our direction.
223
because they are so bright that A or F stars can only be measured if they are
well separated from the hotter star. We may sometimes find stars behind the
O and B stars for which we may expect at least as much reddening as for the
early type star. We then get an upper limit for the reddening. By putting all
the available data together we can get a rather good value for the 'true'
colors.
19.2.5 The wavelength dependence of the interstellar extinction
For nearby unreddened stars the whole energy distribution can be
measured, for instance, by so-called scanner observations which measure the
intensities in wavelength bands of 10 or 20 A. If scanner observations are
then also made for the same kind of star at a known distance, the flux ratio of
the two stars, corrected for the difference in distance, tells us the interstellar
extinction law, because
J2\A)
-TA(interstellar)
(19.13)
Here TX (interstellar) is the optical depth of the interstellar medium between

us and the star (see Fig. 19.10). fx(X) is the flux of the unreddened star, and
f2(X) the flux of the reddened but otherwise similar star after correction for
the difference in distance.
With
Amx = - 2 . 5 - A l o g / , ,
(19.14)
we find
mA(2) - mx(l) = - 2.5 [log f2(X) - log f, (A)]
A(2)
(19.15)
because star 1 is unreddened, so T A (1) = 0.

We then obtain xx(X) for every wavelength and can measure how the
interstellar extinction varies with the wavelength L For the wavelength band
Star
I
h-H
MA)
Observer
Fig. 19.10. The radiative flux F{X) from a star passes through an interstellar gas
cloud and is reduced by the absorption in the cloud. The observed flux
F2(A) = Fl{k)-Q~Xk, where Ft(A) is the flux we would have received without
the absorption in the cloud.
224
3900 < I < 10000 A, the wavelength region observed with most telescopes,
we find nearly xx oc 1/1 For the atomic absorption of hydrogen and similar
atoms or ions we find KX OC A3 for the absorption from each atomic level. The
interstellar absorption has a very different wavelength dependence. The
reason is that the interstellar absorption is mainly due to dust grains. The
light gets scattered by these little particles. If the particles have a size
comparable to the wavelength of the light, the light is not completely blocked
out by the grains, but can get around the grain, although not the full amount.
1iiiiiiiiiiiiiii
m = 1.5
15
20
Fig. 19.11. Extinction curves for spherical grains are shown according to Mie's
theory. The normalized scattering cross-section Q = <j/{na2) is plotted as a
function of p = 2x\m 1|, where a is the radius of the dust particle, a the actual
cross-section, and x measures the ratio of the particle size to the wavelength, m
is the refractive index. (From Dyson & Williams 1980.)
225
Think about the sound waves with wavelengths of meters which can get
around the corner of a house, or the radio waves (100 m) which can also get
around houses.
If one calculates the scattering on little spheres, as has been done by Mie,
one finds a scattering coefficient, as seen in Fig. 19.1, where the scattering
and absorption cross-section is plotted as a function of p = 2x | m 11, where
2na
and a is the radius of the scattering particles; m is the refractive index; | m 11
is of the order 0.1-0.5 for different substances. This then yields p to be about
p~-x
(19.16)
The actual scattering and absorption cross-section for the particles is

G = na2Q,
(19.17)
such that Q gives the ratio of the scattering and absorption cross-section to
the geometrical cross-section.
From Figure 19.11 we see that for l ^ p ^ 4 the Q is approximately
proportional to p ~ a/A, or in other words, for X < a < 4A, or for dust particle
sizes of the order of a few wavelengths of the light the extinction cross-section
is roughly proportional to 1//L If a 104 A, we expect a linear increase of
the absorption cross-section a with 1/1 for wavelengths between 10000 and
2500 A. When p becomes very small, which means for very large A, the
extinction goes to zero.
For very long wavelengths the interstellar grains do not absorb or scatter,
which means that the interstellar dust clouds become transparent to radio
waves. (If you put a football into the path of a sound wave, it does not
influence the sound wave.) Therefore, for radio waves the Milky Way dust is
transparent. Already in the infrared the extinction becomes very small,
almost zero. For very short wavelengths the cross-section for absorption and
scattering does not increase any more but remains, finally, about constant
with wavelength. Grains much larger than the wavelength of the light simply
block the light which hits them. Only the light beams passing between the
grains get to us. This fraction of the light is the same for all wavelengths.
19.2.6 The ultraviolet extinction
Since we have rockets and satellites, we can also measure the
extinction curves in the ultraviolet. Fig. 19.12 shows the approximate
226
dependence on I/A: the right-hand end of the curve corresponds to the

wavelength of 1215 A, which is the wavelength of the hydrogen Lya line,
which is the strongest hydrogen line because it corresponds to the transition
of the electron from the ground level to the next higher level, namely the level
with the main quantum number n = 2. The interstellar medium absorbs
strongly in all the hydrogen Lyman lines originating from the ground level.
Therefore, we cannot measure the dust extinction in the wavelength region of
these lines. The overall run of the extinction curve in the ultraviolet does not
look like that of the Mie curve seen in Fig. 19.11. There is a hump around
2200 A, which is about 400 A wide. For shorter wavelengths we find first
a decrease and then again a rather steep increase of the absorption.
This behavior may be understood if we have a mixture of particles of
different sizes. The smaller particles then become important for shorter
wavelengths. The hump needs a special interpretation. Most astronomers
believe it is due to veryfinegraphite particles, although this is not yet certain.
When interpreting these extinction curves, we also have to keep in mind
that the Mie theory considers only spherical dust grains. We do not know
how the cross-sections change for odd particle shapes.
CD
UJ
10
Fig. 19.12. Interstellar extinction curves, mA - m v , per E(B - V) = 1, also written

as (Ax - 4 V )/E(B - V), or E(/l - V)/E(B - V), are shown as a function of I/A,
where X is the wavelength of the light measured in \i. (\\i = 104 A). The bump
occurs at 2200 A. The average extinction curve is shown as the dashed line.
Low extinctions in the ultraviolet are relatively rare and do not influence the
average very much.
The interstellar extinction for the short wavelengths does not look like Mie
curves seen in Fig. 9.10. (From Dyson & Williams 1980.)
193 The interstellar gas
227
The interesting part is that this extinction curve is not the same everywhere
in the sky. In regions of star formation it looks different. What changes is the
size of the hump relative to the rest, and also the degree of increase for very
short L In dense interstellar clouds, where star formation is going on, the
increase at short wavelengths is much smaller and often the hump is less
high. In Fig. 19.12 we compare some of these curves with the standard curve.
People try to understand this by taking into account the strong ultraviolet
radiation of the newly formed hot stars. They speculate that this radiation
destroys the small grains, resulting in a smaller increase of the extinction for
the short wavelengths.
In the external galaxy next to us, the Large Magellanic Cloud (LMC), the
increase towards the ultraviolet is even steeper than shown in Fig. 19.12. The
absorbing particles responsible for this increase must be relatively more
abundant in the LMC than in the Galaxy. If these were the small grains we
wonder why the many young, hot stars in the LMC do not destroy these
small grains.
It seems that the interstellar medium in the LMC is, on average, very
transparent. There seem to be very few large grains in the LMC.
19.3
The interstellar gas
The interstellar gas is of little importance for the observations of cool

stars, except for the observations of the transition layer and coronal lines and
the X-ray observations.
For the hot stars it is important that the hydrogen in the interstellar gas
absorbs all light being emitted at wavelengths shorter than 912 A, except
for the very closest hot stars, which are, however, all white dwarfs.
Fortunately, cool atomic hydrogen gas can absorb at all wavelengths,
which means continuously over all wavelengths, only for X ^ 912 A. For
wavelengths shorter than 912 A all hydrogen atoms in the ground level can
contribute to this absorption. At 912 A the absorption coefficient per
hydrogen atom is about 10~17cm2. With an average density of about
1 hydrogen atom per cm3 in the interstellar medium (with laboratory
standards this is, of course, an extremely good vacuum), there are about
3 x 1019 atoms between us and a star at 10 pc distance. The optical depth
xx along the line of sight at 912 A is then 1^300, and the amount of
radiation which we would receive is
jx
,-300
where / A0 is the flux which we would have received without any hydrogen
absorption. This means we would not be able to measure anything.
228
Fortunately, there are some directions in the sky for which the hydrogen
density along the line of sight is only about 10" 3 atoms cm" 3. In this
direction we may then still see far ultraviolet light from hot stars within a
distance of about 50 pc. There are, however, no O stars and very few early B
stars in this direction within 50 pc, but with an appropriate camera on a
satellite those few stars could be observed at these short wavelengths, which
would be very important to do.
Our estimates show that generally we cannot expect to receive any of the
far ultraviolet light of the O and B stars, unless we go to extremely short
wavelengths, namely the X-ray region, where the absorption coefficient of
hydrogen is much smaller.
For the O stars most of the radiation is, however, emitted at the
unobservable wavelengths. Therefore, we can only observe a relatively small
fraction of the total light emitted by these stars. It is therefore very difficult to
determine accurate effective temperatures for these stars when we see only a
minor fraction of the emitted energy.
The interstellar gas is also quite important for the interpretation of X-ray
observations. At X-ray wavelengths the hydrogen and helium absorption
coefficients are much lower than the absorption coefficient, around 900 A,
but they are still large enough to absorb a large fraction of the X-rays emitted
by hot stars and by extragalactic sources. In Fig. 19.13 we show the
calculated dependence of the absorption coefficient on wavelength for the
short wavelengths, including the extreme ultraviolet and the X-ray region.
In the optical spectral region the interstellar gas is seen by means of its
interstellar absorption lines. Lines of neutral particles, of ionized particles,
and of interstellar molecules are seen. In the interstellar medium collisions
are very rare. The radiation density usually is also quite low. There are,
therefore, very few processes to excite the atoms or ions to higher states of
excitation. The interstellar lines therefore usually are due to absorption
processes from the ground state of the atoms or ions. Only in very dense
regions may we see lines originating from states with very low but non-zero
excitation energies. Interstellar absorption lines can usually be recognized
easily because only absorption lines from the ground state are seen. In
addition, the interstellar lines in the optical region are generally quite sharp.
For rapidly rotating hot stars, they can easily be recognized because of this.
In Fig. 19.14 we show several spectra in which the interstellar absorption
lines from the gas can be seen as sharp lines, while the stellar absorption lines
are broadened by rotation.
For these interstellar absorption lines we can also measure the line
strengths from which we can determine the number of absorbing atoms
19.3 The interstellar gas
229
along the line of sight. If we do that for all the lines which can be observed in
the optical and in the ultraviolet spectral region, we can determine the
relative abundances of the elements along the line of sight. The elements are
mostly singly ionized. (Once an ultraviolet photon ionizes an atom, the
remaining ion finds it hard to meet another electron with which to
recombine because the densities are so low in the interstellar medium.) If we
want to study relative abundances, we usually have to look at the lines of the
singly ionized ions. There are exceptions, though, for atoms which have
rather large ionization energy, like nitrogen or oxygen, and which would
need far ultraviolet photons for ionization. These are mainly neutral. Such
abundance analysis and the comparison with the abundance of hydrogen
III I I I I
MM
UJ
IO 3
IO2
IO
WAVELENGTH (Angstroms)
Fig. 19.13. The calculated interstellar extinction is shown also for wavelengths
between 100 and 912 A, which generally cannot be observed. For wavelengths
shorter than about 20 A the extinction becomes smaller than the dust extinction
in the visual spectral region. Therefore, X-rays can penetrate the interstellar
medium, while radiation with wavelengths between about 100 and 900 A
generally cannot. (From Cruddace etal. 1974.)
230
showed that most of the heavy elements are depleted in the interstellar
medium with respect to hydrogen as compared to the solar material.
Obviously, most of the heavy elements are locked up in grains. Clearly,
grains cannot form from hydrogen gas. Only a small amount of mass in the
form of grains can lock up all the heavy atoms because there are so few of
them to begin with. If we want to determine the number of hydrogen atoms
absorbing at /1912 A from the observed abundance of, say, Ca + along the
line of sight using the abundance ratio of Ca to hydrogen, we have to take
this into account.
For all stellar observations we have to be aware of the effects of interstellar
absorption, which may vary considerably from one place to another.
Fig. 19.14. Sharp absorption lines due to absorption by Ca+ ions in the
interstellar gas are seen in the spectra of the stars { Ophiuchi (top), P Cygni, v
Saggitarii and a Cygni. In the spectrum of v Sagittarii a sharp line due to the
ionized molecule CH + (present also in the interstellar medium) is also seen.
(From Greenstein 1951.)
Appendix A
Problems
Chapter 1
1. The altitude of a star is the angular height of the star above the horizon; it changes during
the day and night because of the Earth's rotation. What is the largest altitude which the stars
can reach for an observer at a latitude of 35 if the declinations of the stars are as follows:
Star
S
-22
+30
+90
2. One siderial day is the true rotation period of the Earth, as can be seen from the apparent
rotation of the sky. There are about 366.25 siderial days in a year. We experience one day less
per year because the Earth orbits the sun once a year. A star with right ascension a = 8h.00 is in
the meridian at 8:00 hours siderial time. At this time it has the greatest altitude. The difference
between siderial time and right ascension is called the hour angle. For the star with 5 = 22
plot the altitude as a function of the hour angle for an observer at 35 latitude. For how many
hours is the star above the horizon?
3. Why can we see some stars only during the winter and some only during the summer?
Chapter 2
1. A star has a proper motion of \ib = 0.1 arcsec and fia = 0 per year. (/xa is the proper motion in
the direction of right ascension, [id in the direction of declination.) The star has a radial
velocity vr = 20 km/s. What is its velocity relative to us, if it is at (a) a distance of 20 pc? and (b)
at a distance of 100 pc?
2. Another star has a declination of 3 = + 60. It has fia = 0.1 sec and \ib = 0.5 arcsec per year,
vr = 20 km/s. How large is its total velocity relative to us if it is at a distance of 20 pc?
Chapter 3
1. You want to measure the U, B, V colors of an unknown star. You have a photometer whose
counts per minute are proportional to the fluxes which you receive. You measure the standard
star, say Vega (you can of course also take another star whose apparent magnitudes you know
accurately) for 1 minute each at different zenith distances z, and obtain the following counts
for the different filters:
z
V
B
U
20 10000 8000 5000
40 9460 7247 4223
60 8063 5451 2597
231
232
Appendix A
Determine T V , T B , TV for airmass 1.

2. After you have determined the extinction correction for the Earth's atmosphere you
measure the fluxes for the unknown star. It is much fainter than your standard star, therefore
you integrate for 10 minutes each. You measure the following counts:
V
z = 40 3784 2898 1689

Determine the m v , mB, mv.
Which kind of a star is it if it is a main sequence star?
3. For a star with m v = 3.0 the trigonometric parallax was measured to be 0.03 arcsec. What is
the distance of this star? What is its absolute magnitude M v ?
Chapter 5
1. For an unknown star with m v = 5.5 you have measured the colors and found
(U B)o = 0.0, (B V)o = 0.0. Assuming it is a main sequence star, at what distance is it?
What should be its parallax angle nl Could it be measured?
For another star you measured ( U - B ) o = 0.12, ( B - V ) o = 0.63, m v = 7.75, at what
distance is this star if it is a main sequence star?
2. Suppose you were to observe a visual binary system. You find that both stars have
( U - B ) o = 0.12 and ( B - V ) o = 0.63. The magnitude of the fainter star is m v = 8.5 and
the brighter star has m v = 5.5. Which kind of a star is the brighter one? What is the
distance modulus of these stars?
Chapter 6
1. Vega has M v = 0.5. How large is M bol ?
Calculate the luminosity of Vega. Remember that M bolo = 4.75. If Vega has a mass of
2.7 M o what is the ratio of L/M for Vega? Compare with the solar value.
2. After correction for interstellar extinction the star T Scorpii has B V = 0.30. Its M v =
4. How large is its M bol? Use BC = 3.0. Calculate its luminosity L. If its mass is 15 MQ what
is its ratio of L/M! Make a plot of L/M for the sun, Vega and T SCO. If we approximate the
relation by L oc Mp how large is P approximately?
Chapter 8
1. For the star a Orionis
Michelson interferometer.
correction for this star is
of a Orionis? Remember
S=1.38x H
an angular diameter of 0.048 arcsec was measured with the

Its average visual magnitude is m v = 0.8. The bolometric
about BC = 1.50. What is the average effective temperature
that for the sun mbol = 26.85, and the solar constant is
2. For Vega an angular diameter of 3.24 x 10" 3 arcsec was measured by Hanbury Brown.
The bolometric correction for Vega is BC = 0.15. What is the effective temperature of Vega?
The parallax angle of Vega was measured to be n = 0.123 arcsec. What is the radius of Vega?
Use these values of R and Teff to calculate the luminosity of Vega and its absolute bolometric
magnitude. Compare this value with the value obtained from its M v and the bolometric
correction. How could the difference perhaps be explained?
Appendix A
233
3. For Sirius, Hanbury Brown measured an angular diameter 6 = 5.89 x 10~ 3 arcsec. The
apparent visual magnitude of Sirius is mv = 1.46. The bolometric correction is BC ^0.15.
What is the Teff of Sirius? Remember that for the sun mbol = 26.85, and the solar constant
S=1.38. x 10 6 ergcm 2 s" 1 .
4. Sirius has a faint companion, Sirius B, which has an apparent visual magnitude m v = 8.68.
Sirius is at a distance of 2.65 pc. Calculate the absolute visual magnitude of Sirius B.
The energy distribution of this star shows that it has Teff = 26 000 K. The bolometric
correction for this temperature is BC = 2.75. For the sun BC = 0.07. Calculate the M bol for
Sirius B. Calculate the luminosity of this star and its radius. Which kind of a star is this?
Remember L 0 % 4 x 10 3 3 ergs~ 1 .
From the binary orbits we determine that the mass of Sirius B is M (Sirius B) ~ 1 M o .
Calculate the average density of Sirius B.
Chapter 9
1. A spectral line whose laboratory wavelength is Xo = 4343.0 A is seen in a stellar spectrum
at a wavelength of 4344.2 A. What is the line of sight velocity component of the star relative
to us? Is the star moving away from us or towards us?
2. For a binary system the ratio of the radial velocities for the two stars was measured to be
i?r(l)/i;r(2) = 1.5. One of the stars is a solar type star, which means a G2V star. What is the mass
of the other star?
3. For an eclipsing binary with a period of 30 days the total duration of the eclipse is 8 hours.
We see a flat minimum for 1 hour and 18 minutes. The radial velocity amplitude for star 1 is
i;rl = 30 km s~* and for star 2 it is vr2 40 km s~ l . For a circular orbit and i = 90 how large
are the radii of the stars? Estimate the masses of the stars.
Chapter 10
1. In Fig. A.I we show low resolution stellar spectra of some stars. The wavelengths for a few
strong lines are given. Estimate the spectral types of these stars.
Chapter 11
1. In stellar spectra we see mainly absorption lines. What does this tell us about the
temperature stratification in stars?
2. In the solar spectrum we see a strong line at 4303.177 A. We suspect that this is an Fell
line, i.e. a line from the Fe + ion. Check from the Multiplet tables of Mrs Moore Sitterly which
other Fell lines should be present if this line identification is correct. Which lines should be
stronger and which lines should be weaker than the line at 4303.177 A?
Chapter 13
1. For A stars we usually observe very broad lines due to rotational broadening. We call the
equatorial rotational velocity vrot. If for a given star the full line width from edge to edge is
0.8 A for a line at 4343 A, how large is i?rotsini?
For magnetic stars we can measure the rotation period. Suppose that for the observed star a
rotation period of 3 days is found, how large is sin i ?
234
Appendix A
4101.7
4340.5
4861.3
Fig. Al. A series of stellar spectra showing dark absorption lines in a bright
continuum. The wavelengths of the strongest lines are given at the bottom in A.
Laboratory emission line spectra are seen on both sides of the stellar spectra.
They are used to determined the wavelengths of the stellar absorption lines.
Chapter 14
1. Suppose that for a magnetic star we observe for a normal Lorentz triplet a shift of the two
circularly polarized components by 0.2 A from the central component. How large is the
longitudinal magnetic field component?
2. For an average peculiar A star i;rot sin i is about 70 km s" 1 . How large a magnetic field
would such a star need to have before we could measure the splitting of the Lorentz triplet for
a line at 4500 A?
Chapter 16
1. A S Cephei star in the Galaxy has mv = 3.7 and a pulsation period of 5.6 days. How far
away is it?
2. In the Andromeda Galaxy which has a distance modulus of mY M v = 24.1 we can now
just barely observe RR Lyrae stars which have an average absolute magnitude of M v = 0.6.
What are the apparent visual magnitudes of these stars in the Andromeda galaxy?
Appendix A
235
3. The RR Lyrae stars in the globular cluster M3 have an apparent visual magnitude of
m v = 15.6. How far away is the globular cluster? If we take the plane of our Galaxy as
the reference plane then M3 has an altitude of 77 above this plane. We call this angle the
galactic latitude b. How high, above the Galactic plane is M3, which means how large is z in
Fig. A.2 below.
Chapter 17
1. If the supernova 1987A in the Large Magellanic Cloud keeps expanding with its present
velocity of 10000 km s~ * for another year, how large would it be in pc after 10 years? How
large would it then be in seconds of arc? The presently adopted distance modulus to the LMC
is mv My = 18.3.
2. If in the near future one of the Pleiades stars should become a supernova with the same
intrinsic brightness as the supernova 1987A in the LMC which reached mv = 2.9, what would
be the apparent brightness of such a Pleiades supernova? If it should also expand like the
LMC supernova at 10000 km s~ * how large would it be after 10 years as measured in arcsec?
Compare the size and brightness with the moon.
Chapter 18
1. In the solar granulation we see temperature differences of up to about 400 K. If the effective
temperatures of a dark and a bright spot are 5600 K and 6000 K respectively what is the
intensity ratio of the two spots at 5000 A if they radiate like black bodies?
2. In the solar granulation we see velocity differences of about 2 km s" 1 . How large is the
Doppler shift for such velocities for a wavelength of 5000 A? Compare this with the
rotational broadening of the solar spectral lines. From the motion of the sunspots we
determine that the rotational period for the sun is 27 days.
Chapter 19
1. For a main sequence star of spectral type B we measure B V = 0.3, U B = 0.36.
What are its true colors? How large is its color excess? Its apparent magnitude is mv = 5.0.
What would be its mVo, mBo, mUo magnitudes, and its (B - V)o and (U - B)o colors? Use
R = Am v /(B - V) = 3.2. How far away is this star?
Galactic plane
Earth
Fig. A2
Appendix B
Some important astronomical
quantities
I. Atomic quantities
Electron mass m = 9.105 x 10~ 28 g
Proton mass mH = 1.672 x 10~ 24 g
Planck's constant h = 6.624 x 10" 27 ergs
Elementary charge e = 4.802 x 10 ~ 1 0 esu
Velocity of light c = 2.998 x 1010 cms" 1
Stefan-Boltzmann constant o = 5.672 x 10" 5 ergcm" 2 s~ 1 K"
Boltzmann constant k= 1.38 x l O ' ^ e r g K " 1
Gas constant Rg = 8.314 x 107 erg K " 1 mol" 1
* ion (H)= 13.60 eV
Xion(He) = 24.58 eV
+
Zion (He ) = 54.4eV
%ion(Fe) = 7.90eV
Absorption edges for Hydrogen continua

Lyman continuum: 912 A
Balmer continuum: 3647 A
Paschen continuum: 8208 A
H" continuum: 17 000 A
Helium Lyman continuum: 504 A
Helium+ Lyman continuum: 228 A
Hydrogen lines
Lya: 1215.67 A
Ly: 1025.72 A
Lyy:972.5A
Ha: 6562.82 A
HP: 4861.33 A
Hy: 4340.47 A
H<5: 4101.74 A
Some important resonance lines of other elements
He = He I
He+=HeII
Ca = C a I
I = 591.42 A
303.78 A
4226.73 A
237
238
Appendix B
Ca + = Ca II
Na = N a I
C = CI
C+=CII
C + + + = C IV
O=OI
K = KI
Ba = BaI
Ba+=BaII
Sr = SrI
Sr+=SrII
Mg = M g I
Mg + = Mg II
3933.66; 3968.47 A
5889.95; 5895.92 A
1657.00 A
1335.68; 1334.52 A
1548.20; 1550.77 A
1302.17; 1304.86; 1306.02 A
7664.91; 7698.98 A
5535.48 A
4554.03 A
4607.33 A
4077.71 A
4571.10; 2852.12 A
2795.52; 2802.70 A
II. Earth
1 Siderial day = rotation period of the Earth: 23h56m04M
1 year = 365.256 solar days
Inclination of equatorial plane to ecliptic: 2327'
1 astronomical unti 1 a u = 1.496 x 10 13cm
Mass of the earth M e = 5.98 x 1027 g
Equatorial radius R@ = 6.371 x 107 cm = 6371 km
III. Sun
Rotation Period P o = 25.38 days
Mass M o = 1.98 x 10 33g
Radius RQ = 6.96 x 1010cm
Gravitational acceleration gQ = 2.74 x 10 4 cms~ 2
Mean density p = 1.41 g
Angular radius R/d = 959.63 arcsec = 0.0046526 radians
Solar constant S= 1.38 x 10 6 ergcm" 2 s~ 1
Luminosity L o = 3.96 x 10 33 ergs~ 1
Surface flux %FQ = 6.41 x 10 6 ergcm" 2 s" 1
Effective temperature Teff = 5800 20 K
Apparent visual magnitude m v o = 26.78
Apparent bolometric magnitude mbol o = 26.85
B - V = 0.63 0.02
U - B = 0.13 0.02
Absolute visual magnitude M v o = 4.82
Absolute bolometric magnitude Mbol o = 4.75
IV. Some other important quantities
l e V = 1.602 x 10" 1 2 erg
1 parsec = 3.2598 light years = 3.084 x 1018 cm
gravitational constant G = 6.67 x 10~8 dyncm 2 g~2
0 Kelvin = - 273.16 centigrade
mv = 0 corresponds to nfx = 3.6 x 10" 9 erg cm " 1 s ~ l/A
or TT/A = 3.6 x 10" 1 6 wattcm" 2 /A at 5500 A.
References
I. Other textbooks
All introductory astronomy textbooks.

Also:
Dufay, J. (1964). Introduction to Astrophysics: The Stars (Translated by O. Gingerich).
Dover Publications, New York.
Novotny, E. (1973). Introduction to Stellar Atmospheres and Interiors. Oxford University
Press, Oxford.
Shu, F. (1982). The Physical Universe. University Science Books, Mill Valley, California.
Swihart, T. L. (1968). Astrophysics and Stellar Astronomy. John Wiley and Sons, Inc. New
York.
Unsold, A. & Bascheck, B. (1983). The New Cosmos, 3rd ed. Springer Verlag, Berlin.
Voigt, H. H. (1975). Abriss der Astronomie, Wissenschafts Verlag, Bibliographisches
Institut, Mannheim.
Walker, G. (1987). Astronomical Observations. Cambridge University Press, Cambridge.
II. Some Star catalogues
Bonner Durchmusterung (BD), 4th ed. (1968). Universitats Sternwarte Bonn, Verlag F.
Diimmlers.
Henry Draper Catalogue, A. J. Cannon & Pickering, E. C. (1918-24). Annals of the
Astronomical Observatory of Harvard College, Vol. 91-99.
Smithsonian Astrophysical Observatory Star Catalogue (SAO) (1966). Positions and proper
motions of 258, 997 stars for the epoch and equinox of 1950.0. U.S. Government
Printing Office.
The Bright Star Catalogue (BS or HR), 3rd ed., Hoffleit, D. & Jascheck, C. (1982). Yale
University Observatory, New Haven.
General Catalogue of Variable Stars, Kukarkin et al. (1969 ff.). Astronomical Council of
the Academy of Sciences, Moscow. 2nd edition: Kholopov etal. 1985 ff, NAUKA
Publishing House, Moscow.
III. Some Reference books for stellar data
Allen, C. W. (1968). Astrophysical Quantities. 3rd ed. University of London. The Athlon
Press.
Landolt Bornstein (1982). Numerical Data and Functional Relationships in Science and
Technology, Group VI, Vol. 2, Astronomy and Astrophysics, ed. K. Schaifers and H. H.
Voigt. Springer Verlag, Berlin.
239
240
References
Blanco, V. M., Demers, S., Douglass, G. G. & Fitzgerald, M. P. (1968). Photoelectric

Catalogue: Magnitudes and Colors of Stars in the UBV and Uc, B, V systems.
Publications of the United States Naval Observatory, 2nd Series, Vol. 21.
Jamar, C , Maceau Hercot, D., Monfils, A., Thompson, G. L, Houziaux, L. & Wilson, R.
(1976). Ultraviolet Bright Star Spectrophotometric Catalogue. European Space Agency
SR 27, SR 28.
Moore, C. E. (1959). National Bureau of Standards, Technical Note No. 36 (data on
spectral lines). U.S. Government Printing Office.
Strand, K. Aa. (1963). Basic Astronomical Data, Vol. Ill of Stars and Stellar Systems.
Gen. ed. G. P. Kuiper & B. M. Middlehurst. The University of Chicago Press,
Chicago.
Thompson, G. I., Nandy, K., Jamar, C , Monfils, A., Houziaux, L., Carnochan, D. C. &
Wilson, R. (1978). Catalogue of Stellar Ultraviolet Fluxes. The Science Research
Council.
IV. References Quoted in the Text
Abell, G. O. (1982). Exploration of the Universe, 4th edn. Saunders College Publishers,
Philadelphia.
Allen, C. W. (1968). Astrophysical Quantities, 2nd edition. University of London, the
Athlon Press.
Arp, H. C. (1958). In Handbook of Physics, Vol. 51, p. 75. Springer Verlag, Berlin.
Babcock, H. (1960). In Stellar Atmospheres, ed. J. Greenstein. Volume VI of Stars and
Stellar Systems, Gen. ed. G. P. Kuiper & B. M. Middlehurst, p. 282. The University of
Chicago Press, Chicago.
Becker, W. (1950). Sterne und Stern Systeme, 2nd edition. Theodor Steinkopff, Dresden.
Biermann, L. (1941). Vierteljahresschrift, d. Astronomischen Gesellschaft, 76, 194.
Bidelman, W. P. & Keenan, P. C. (1951). Astrophysical Journal, 114, 473.
Binnendijk, L. (1960). Properties of Double Stars. University of Pennsylvania Press,
Philadelphia.
Bohlender, D. A., Brown, D. N., Landstreet, J. D. & Thompson, I. B. (1987).
Astrophysical Journal, 323, 325.
Bohm-Vitense, E. (1984). Science, 223, 777.
Bohm-Vitense, E. & Canterna, R. (1974). Astrophysical Journal, 194, 629.
Bohm-Vitense, E. & Johnson, P. (1977). Astrophysical Journal Supplement Series, 35, 461.
Borra, E. F., Landstreet, J. D. & Thompson, I. B. (1983). Astrophysical Journal
Supplement Series, 53, 151.
Bray, R. J. & Loughhead, R. E. (1974). The Solar Chromosphere, p. 224. Chapman &
Hall, London.
Burnham, R. (1978). Burnham's Celestial Handbook, Vol. II & Vol. III. Dover Publ. Inc.,
New York.
Clayton, D. D. (1983). Principles of Stellar Evolution and Nucleosynthesis. The University
of Chicago Press, Chicago.
Cohen, M. & Kuhi, L. V. (1979). Astrophysical Journal Supplement, 41, 743.
Cox, J. P. (1980). Theory of Stellar Pulsations. Princeton University Press, Princeton,
New Jersey.
Cruddace, R., Paresce, F., Bowyer, S. & Lampton, L. (1974). Astrophysical Journal, 187,
500.
Deutsch, A. J. (1958). In Handbook of Physics, Vol. 51, p. 689. Springer Verlag, Berlin.
Deutsch, A. J. (1952). Transaction of the International Astronomical Union, 8, 801.
References
241
Dyson, J. E. & Williams, D. A. (1980). The Physics of the Interstellar Medium. John
Wiley & Sons, New York.
Eddington, A. S. (1926). The Internal Constitution of Stars. Dover Publication 1950,
p. 277.
Gibson, E. G. (1973). The Quiet Sun. NASA, Washington, D.C.
Greenstein, J. (1951). In Astrophysics ed. J. A. Hynek, p. 526. McGraw Hill Book
Company, Inc., New York.
Greenstein, J. (1986). Sky and Telescope, 72, 349.
Hanbury Brown, R. (1968). In Annual Review of Astronomy and Astrophysics, 6, 13.
Hanbury Brown, R. (1974). The Intensity Interferometer. Taylor & Francis Ltd. London.
Herbig, G. H. (1969). In Nonperiodic Phenomena in Variable Stars, ed. L. Detre, p. 75.
Dordrecht: Reidel.
Herbig, G. H. & Jones, B. F. (1981). Astronomical Journal, 86, 1232.
Johnson, H. L. & Morgan, W. W. (1953). Astrophysical Journal, 111, 313.
Joy, A. H. (1960). In Stellar Atmospheres, ed. J. Greenstein. Vol. VI of Stars and Stellar
Systems, Gen. ed. G. P. Kuiper & B. M. Middlehurst, p. 653. The University of
Kiepenheuer, K. O. (1953). In The Sun, ed. G. P. Kuiper, p. 322. The University of
Lamia, E. (1982). In Landolt-Bornstein: Numerical Data and Functional Relationships in
Science and Technology. New Series Volume 2, Astronomy and Astrophysics ed. K.
Schaifers and H. H. Voigt, p. 35. Springer Verlag, Berlin.
Ledoux, P. & Walraven, Th. (1958). In Handbook of Physics, Vol. 51, p. 353. Springer
Verlag, Berlin.
Morgan, W. W., Keenan, P. C. & Kellman, E. (1943). An Atlas of Stellar Spectra.
Astrophysical Monographs, Chicago.
Morgan, W. W , Abt, H. A. & Tapscott, J. W. (1978). Revised MK Spectral Atlas for
Stars Earlier than the Sun. Yerkes Observatory, University of Chicago and Kitt Peak
National Observatory.
Paczinski, B. (1985). In Astronomy and Astrophysics, ed. M. S. Roberts, p. 143. The
American Association for the Advancement of Science.
Parker, E. N. (1974). Solar Physics, 36, 249.
Pettit, E. (1951). In Astrophysics, a Topical Symposium, ed. J. A. Hynek, p. 259. McGrawHill Book Company, New York.
Popper, D. (1980). Annual Review of Astronomy and Astrophysics, 18, 115.
Ridgway, S. T., Wells, D. C. & Joyce, R. R. (1977). Astronomical Journal, 82, 414.
Sandage, A. & Walter, M. (1966). Astrophysical Journal, 143, 313.
Schmidt-Kaler, Th. (1982). In Landolt-Bornstein: Numerical Data and Functional
Relationships in Science and Technology, New Series, Volume 2, Astronomy and
Astrophysics, ed. K. Schaifers & H. H. Voigt, p. 1. Springer Verlag, Berlin.
Schwartz, R. D. (1983). Annual Review Astronomy and Astrophysics, 21, 209.
Shu, F. H. (1982). The Physical Universe. University Science Books, Mill Valley, Cal.
Smart, W. M. (1977). Textbook on Spherical Astronomy. 6th ed., Cambridge University
Press, Cambridge.
Unsold, A. (1938). Physik d. Sternatmosphdren. Springer Verlag, Berlin.
Unsold, A. (1982). The New Cosmos. Springer Verlag, Berlin.
Zirin, H. (1966). The Solar Atmosphere, p. 282. Blaisdell Publishing Company, Waltham
Massachusetts.
Zwicky, F. (1965). In Stellar Structure, ed. L. H. Aller & D. B. McLaughlin, p. 367,
Vol. VIII of Stars and Stellar Systems, Gen. ed. G. P. Kuiper and B. M. Middlehurst.
University of Chicago Press, Chicago.
Index
(Volume 1)
Page numbers printed in italic type indicate the main source of information
absorption
coefficient, earth's atmosphere 17 ff
lines 89 ff
interstellar 213 ff, 230
abundances of chemical elements 103, 105
ages, stellar 109
angular radius 42, 49 ff, 59
astronomical unit 11
Baade-Wesselink method 157 ff
Balmer lines 89, 101, 171
barium stars 144ff
beta decay 144
black body 21 ff
binaries 67 ff, 142, 184tt
eclipsing 73, 78 ff, 157
visual 76
bolometric correction 44
Bunsen 89,92
butterfly diagram 205
Calcium lines 90, 141 ff
Cepheids, delta 153
Cepheids, beta 155
chromosphere, spectrum 102, 148
classification, spectral 89 ff
coherence of light 55
clusters
galactic 33 ff
globular 36 ff
color
excess 107
B-V, U-B 16, 26, 27
magnitude diagrams 31 ff
colors, stellar 14, 15, 221
corona, stellar 102, 197 ff
crab nebula 165, 167, 173 ff
declination 2
delta Scuti stars 153
diffusion, radiative 139, 140, 143
dipole, magnetic 132
distance
modulus 29
solar 9ff
distances, stellar 11 ff
Doppler shift 7, 70
dust, interstellar 214 ff
eclipse, solar 198
ecliptic 2
emission lines 89, 147
energy
nuclear 179
gravitational 179
energy distribution
black body 24
stellar 91
solar 20
equatorial plane 2
equipotential surface 188
extinction
interstellar 223 ff
earth's atmosphere 16 ff
flux, radiative
23
giants 32, 38, 95

granulation 194
halo, galactic 109
helium lines 102, 171
Herbig-Haro objects 149
horizontal branch 37 ff
Hyades 34 ff
hydrogen lines 89 ff, 101
hydrostatic equation 136 ff
interference 54
interferometer
Hanbury Brown 53 ff
Michelson 50 ff
interstellar bullet 151
interstellar extinction 223 ff
interstellar reddening 220 ff
jet
151
Kepler
laws 10, 67 ff
supernova 167
Kirchhoff s law 89
243
244
Index
Lagrange point, Lx 188

Large Magellanic Cloud 161, 165, 227
Light echo 173ff
Line identification 100
long period variables 156
Lorentz triplet 117
luminosity
solar 43
stellar 41 ff
Lyman
lines 101
continuum, interstellar 227 ff
magnetic field
earth 117
stellar 125 ff
lines 120
reversal 133, 135
magnetic pressure 134
magnetograph, Babcock's 118, 122
magnitudes
absolute 14, 28 ff
apparent 13
bolometric 44
main sequence 31, 38
mass-luminosity relation 85 ff
masses, stellar 67 ff, 74, 85, 86
metallic lines 103 ff
metallic line stars 140ff
Mie theory 224
Milky Way 213
molecules 104 ff
nebulae, reflection
neutron
capture 144
star 181
Novae 183 ff
222
oblique rotator 131

occultation, Lunar 58 ff
orbits, binaries 75 ff
optical depth, earth's atmosphere
parallaxes
photometric 38
trigonometric 11
parsec, definition 12
P Cyni profiles 148, 172
peculiar A stars = Ap stars 127 ff
penumbra, sunspot 194
period-luminosity relation 161 ff
photosphere 203
Planck function 23
Pleiades 33 ff
polarization, circular 118
populations, stellar 105, 107 ff
precession, earth's 1
prenova 188 ff
presupernova 175 ff
proper motion 7
pulsar 180
17, 19
pulsating stars
quadrupole
153 ff
133
radial velocities 7, 71 ff
radius, solar 42, 238
radii, stellar 80, 158
rare earth elements 128
right ascension 2
Roche lobe 189
rotation
differential solar 206 ff
earth 1, 3
stellar 111 ff, 115, 143
RR Lyrae stars 153
RV Tauri stars 154
sedimentation 138, 139, 143
seeing 49, 197
solar constant 27
spectra, stellar 89 ff
spectral types 90 ff
Stefan-Boltzmann Law 23
subdwarfs 107
subgiants 33, 38
sun 193ff
sunspots 119, 193 ff
sunspot cycle 204 ff
supergiants 36, 95
supernovae 165 ff
supernova remnants 178
temperature
effective 46, 63 ff, 86
solar 24, 64
Wien 24
thermodynamic equilibrium
tidal bulge 188
tidal force 187
transition layer 102, 148
triangulation 9
T Tauri stars 147 ff
Tycho de Brahe, supernova
umbra, sunspot
22
167
194
acoustic 203
magnetohydrodynamic 203
white dwarfs 33, 97, 146
Wien's Law 24
wiggly lines 195
wind, stellar 143
Wolf diagram 215 ff
W Vir stars 153
X-rays 197, 201, 202
X-ray absorption, interstellar
Zeeman effect 117
Zenith distance 17
ZZ Ceti stars 155
229

Bohmvitense e Introduction To Stellar Astrophysics Volume 1

Uploaded by

Document Informationclick to expand document informationBohmvitense e Introduction to Stellar Astrophysics Volume 1

Document Informationclick to expand document information

Copyright:

Available Formats

Bohmvitense e Introduction To Stellar Astrophysics Volume 1

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Bohmvitense e Introduction To Stellar Astrophysics Volume 1

Uploaded by

Copyright:

Available Formats

Introduction to stellar astrophysics

Introduction to Stellar Astrophysics:

The right of the

CAMBRIDGE UNIVERSITY PRESS

Published by the Press Syndicate of the University of Cambridge

Proper motions of stars

Distances of nearby stars

The brightnesses of the stars

Color magnitude diagrams

The luminosities of the stars

Angular radii of stars

Effective temperatures of stars

Masses and radii of stars

Understanding stellar spectra

Stars with peculiar spectra

The topic of this volume is stellar astronomy or more accurately stellar

The coordinate system

If we want to study stars, the first thing to look at might be their

1.2 Direction of the Earth's rotation axis

Direction of the Earth's rotation axis

From the previous discussion it is apparent that we have to know

Visibility of the sky

Fig. 1.3. A long-exposure photograph with a fixed position of the telescope

1.3 Visibility of the sky

Proper motions of stars

2 Proper motions of stars

of a star in a direction perpendicular to the line of sight will lead to a

Distances of nearby stars

The distance of the sun

The distances to nearby objects on Earth are determined by

3 Distances of nearby stars

wherefc(Venus)and b(Earth) are the semi-major orbital axis of Venus and

3.2 Trigonometric parallaxes of stars

both semi-major axes. The second equation is provided by a measurement of

If we measure d, then (3.4) provides the second equation needed to determine

Trigonometric parallaxes of stars

3 Distances of nearby stars

The brightnesses of the stars

The apparent magnitudes

The brightness of a star is a quantity which can be measured easily, at

4 The brightnesses of the stars

The colors of the stars

We have emphasized that the apparent brightnesses or magnitudes

4.2 The colors of the stars

4 The brightnesses of the stars

abbreviate mv by V. So V means the same as m v .

Correction for the absorption of radiation in the Earth's

When we measure the radiative energy arriving at the telescope we

4.3 Radiation absorption in the Earth's atmosphere

Dividing by JA and remembering that d/A//A = d(ln JA) we find

Here we have defined the so-called optical depth TA by

Surface of the Earth

4 The brightnesses of the stars

where xx is the variable optical depth along the path s.

where xXt is the optical depth measured perpendicularly through the

where xXt is now independent of z.

4.3 Radiation absorption in the Earth's atmosphere