Join a Princeton astrophysicist to learn how solving the mysteries of the universe

Topic Subtopic
sparks a new understanding of our world. Science & Mathematics Astronomy

Introduction

Introduction to Astrophysics
“Pure intellectual stimulation that can be popped into
the [audio or video player] anytime.”

to Astrophysics
—Harvard Magazine

“Passionate, erudite, living legend lecturers. Academia’s

best lecturers are being captured on tape.”
—The Los Angeles Times
Course Guidebook
“A serious force in American education.”
—The Wall Street Journal
Professor Joshua N. Winn
Princeton University

Joshua N. Winn is a Professor of Astrophysical Sciences

at Princeton University. After earning his PhD in Physics
from MIT, he held fellowships from the National Science
Foundation and NASA at the Harvard-Smithsonian Center
for Astrophysics. Professor Winn was a participating
scientist in NASA’s Kepler mission and is a coinvestigator
in NASA’s Transiting Exoplanet Survey Satellite mission. He
has authored or coauthored approximately 200 scientific
articles and is the recipient of the MIT Buechner Faculty
Teaching Prize.

THE GREAT COURSES ®

Corporate Headquarters
4840 Westfields Boulevard, Suite 500
Chantilly, VA 20151-2299
USA
Guidebook

Phone: 1-800-832-2412
www.thegreatcourses.com
Professor Photo: © Jeff Mauritzen - inPhotograph.com.
Cover Image: ESO/L. Calçada/CC BY 4.0

Course No. 1360 © 2018 The Teaching Company. PB1360A

Published by
THE GREAT COURSES

Corporate Headquarters

4840 Westfields Boulevard | Suite 500 | Chantilly, Virginia | 20151‑2299

[phone] 1.800.832.2412 | [fax] 703.378.3819 | [web] www.thegreatcourses.com

Copyright © The Teaching Company, 2018

Printed in the United States of America

This book is in copyright. All rights reserved. Without limiting the rights under copyright reserved above, no
part of this publication may be reproduced, stored in or introduced into a retrieval system, or transmitted, in
any form, or by any means (electronic, mechanical, photocopying, recording, or otherwise), without the prior
written permission of The Teaching Company.
 JOSHUA N. WINN, PHD
Professor of Astrophysical Sciences
Princeton University

J
oshua N. Winn is a Professor of Astrophysical Sciences at Princeton
University. He studied physics as an undergraduate at the Massachusetts
Institute of Technology (MIT), spent a year as a Fulbright Scholar at
Cambridge University, and then returned to MIT to pursue a doctorate. While
in graduate school, Professor Winn worked in medical physics, condensed matter
physics, and astrophysics and wrote for the science section of The Economist.
After earning his PhD, he held fellowships from the National Science Foundation
and NASA at the Harvard-Smithsonian Center for Astrophysics. He was on
the MIT Department of Physics faculty for 10 years before joining the faculty
of Princeton.

i
 PROFESSOR BIOGRAPHY

Professor Winn’s research goals are to explore the properties of planets around
other stars, understand how planets form and evolve, and make progress on the
age-old question of whether there are other planets capable of supporting life.
His research group uses optical and infrared telescopes to study exoplanetary
systems, especially those in which the star and planet eclipse each other.

Professor Winn was a participating scientist in NASA’s Kepler mission, which

had as its goal the detection of earthlike planets, and he is a coinvestigator in
another NASA mission called the Transiting Exoplanet Survey Satellite, which
was launched in 2018. He has authored or coauthored approximately 200
scientific articles, mainly on the topic of exoplanetary science. Over the years,
Professor Winn and his research group have pursued topics in stellar astronomy,
planetary dynamics, radio interferometry, gravitational lensing, and photonic
bandgap materials.

Professor Winn has taught numerous subjects in physics and astronomy to

both undergraduate and graduate students. He has received the MIT Buechner
Faculty Teaching Prize and the MIT School of Science Graduate Teaching Prize
for excellence in teaching.

Professor Winn’s other Great Course is The Search for Exoplanets: What
Astronomers Know. 

ii
 TABLE OF CONTENTS

INTRODUCTION

Professor Biography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
Course Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

GUIDES

1 Zooming Out to Distant Galaxies . . . . . . . . . . . . . . . . . . . . . . 4

2 Zooming In to Fundamental Particles . . . . . . . . . . . . . . . . . . 14
3 Making Maps of the Cosmos . . . . . . . . . . . . . . . . . . . . . . . . 26
4 The Physics Demonstration in the Sky . . . . . . . . . . . . . . . . . . 40
5 Newton’s Hardest Problem . . . . . . . . . . . . . . . . . . . . . . . . . 52
6 Tidal Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Quiz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

7 Black Holes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
8 Photons and Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
9 Comparative Planetology . . . . . . . . . . . . . . . . . . . . . . . . . 111
10 Optical Telescopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
11 Radio and X-Ray Telescopes . . . . . . . . . . . . . . . . . . . . . . . 137
12 The Message in a Spectrum . . . . . . . . . . . . . . . . . . . . . . . 149
Quiz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164

iii
 TABLE OF CONTENTS

13 The Properties of Stars . . . . . . . . . . . . . . . . . . . . . . . . . . 166

14 Planets around Other Stars . . . . . . . . . . . . . . . . . . . . . . . . 178
15 Why Stars Shine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
16 Simple Stellar Models . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
17 White Dwarfs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
18 When Stars Grow Old . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Quiz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244

19 Supernovas and Neutron Stars . . . . . . . . . . . . . . . . . . . . . . 246

20 Gravitational Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
21 The Milky Way and Other Galaxies . . . . . . . . . . . . . . . . . . 274
22 Dark Matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
23 The First Atoms and the First Nuclei . . . . . . . . . . . . . . . . . . 306
24 The History of the Universe . . . . . . . . . . . . . . . . . . . . . . . 319
Quiz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334

SUPPLEMENTARY MATERIAL

Quiz Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336

Important Numerical Values . . . . . . . . . . . . . . . . . . . . . . . . . . 341
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
Image Credits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346

To go back to the page you came from, PRESS Alt + ← on a PC or ⌘ + ← on a Mac.

On a tablet, use the bookmarks panel.

iv
 INTRODUCTION TO
ASTROPHYSICS

A
stronomy is a fascinating subject because the universe is full of such
wonders as black holes, exploding stars, and colliding galaxies and
because new discoveries are being made at a rapid pace. While it is
possible to appreciate astronomy with images and qualitative descriptions, the
goal of this course is to gain access to the deeper level of beauty and understanding
that astrophysics—the application of the laws of physics to comprehend celestial
phenomena—can provide. This not only gives a greater appreciation for the
wonders of the universe, but also allows for the perception of hidden regularities
and connections between phenomena. For example, the relationship between
a white dwarf and a neutron star is analogous to the relationship between an
atom and an atomic nucleus.

This course conveys the quantitative foundations of astrophysics with the hope of
both satisfying and stimulating curiosity about the subject. The most important
prerequisites are the desire to understand deeply, the capacity and patience
for learning new things, and a sense of wonder. To gain the most from the
course, a good background in freshman-level classical mechanics and calculus is
needed; logarithms, trigonometry, and vectors are employed throughout. After
completing this course, you will have a firmer grip on the universe and an
enhanced ability to solve problems in the physical sciences.

Astrophysics spans more orders of magnitude in space, time, and mass than any
other science, and the first 3 lectures of this course provide a unifying structure
to help comprehend the vast range of scales, based on orders of magnitude and
logarithmic charts. The first lecture zooms out from spatial scales of human
beings to the entire observable universe. The second lecture zooms in to the
fundamental particles and the 4 fundamental forces of nature. The third lecture
is about how astrophysicists establish the locations of objects in 3 dimensions,
despite being stuck in an arbitrary location within just one of countless galaxies.

1
 COURSE SCOPE

The next topic is gravity. Lectures 4 through 7 apply the law of gravity to
understand the motion of planets, the destructive power of tidal forces, and the
existence of black holes. A feature of this section is a detailed examination of
the relationship between Kepler’s laws of planetary motion and Newton’s laws
of motion and gravity, a topic usually reserved for more advanced courses.

Attention then turns to a different force of nature: electromagnetism. Lecture

8 is on photons, the basic unit of electromagnetic radiation. The properties of
photons are compared with those of ordinary particles, and many important
formulas are introduced. Lecture 9 provides an immediate application of these
concepts to understand the basic properties of the planets in the solar system.
The next 3 lectures are about telescopes. They discuss the fundamental purpose
of telescopes and the differences between telescopes in the radio, optical, and
x-ray domains of the electromagnetic spectrum. Lecture 12 takes a deep dive
into spectroscopy, the main way to learn about the physical conditions of a star,
planet, nebula, or galaxy.

Then begins a sequence of lectures about stars and their planets in which the
following questions are addressed: How are the properties of stars determined?
What has been discovered about planetary systems around other stars? Why do
stars shine, and how long do they last? What are the conditions like at the center
of a star? What happens when a star runs out of fuel? How can the existence of
stars that are millions of times denser than the Earth be explained?

After such detail is spent on understanding stars, they are destroyed in lecture 19,
which is about supernovas and their causes. Then comes a highlight of the
course, in lecture 20, about gravitational waves. The seminal first detection of
colliding black holes is examined in detail, starting with the original data and
culminating in a calculation of the masses of the black holes and their distance
from Earth. Even as recently as 2015, this lecture could not have been written.

2
 COURSE SCOPE

The last 4 lectures zoom out to gain a perspective on galaxies and the universe as
a whole. Lecture 21 not only features dazzling images of galaxies—the orchids
of the universe—but also introduces the mind-bending astrophysical concept
of the galaxy as a collisionless fluid of stars. The topic of galaxies is developed
further in lecture 22, including active galaxies, in which material is funneling
into a central black hole, and the mystery of dark matter. Finally, lectures 23 and
24 present the quantitative basis of the modern creation story: the big bang. The
course ends at the frontier of astrophysics and particle physics, with the discovery
of what may turn out to be an entirely new force of nature. 

3
Lecture 1

ZOOMING OUT TO
DISTANT GALAXIES

T he words “astrophysics” and “astronomy” are

basically interchangeable these days, but there
is a subtle intellectual distinction. Astrophysics is the
application of the laws of physics to understand celestial
phenomena. Occasionally, we even discover a new law
of physics by studying what’s out there. In contrast,
astronomy can be defined as the careful observation
of heavenly bodies—a cultural activity dating back
thousands of years that only gradually became
scientific. The ancient Babylonians, the Chinese, and
the Mayans were all accomplished astronomers, but
they weren’t astrophysicists. Compared to astrophysics,
no other science spans such a vast range of scales—
from nanometers to billions of light-years and from the
radiation of a single electron to the output of trillions
of suns.

4
 Lecture 1 — Zooming Out to Distant Galaxies

Astrophysics began in the 17 th century with Isaac

Newton, who explained the motions of the planets
with his shiny new equations relating force, mass,
acceleration, and gravity.
The actual word “astrophysics” is more recent. It’s
from the mid-19 th century, after the invention of
photography and spectroscopy. These techniques
allowed us to go beyond looking through telescopes
with our eyes; now we could make more objective records,
detect fainter sources, and connect our observations to laboratory
experiments with light, heat, and atoms.

PUTTING THE UNIVERSE

INTO PERSPECTIVE
®® It’s difficult to put the whole universe into perspective. Even if we scale
everything down by a factor of a billion, the nearest star to the Sun would
be 25,000 miles away, and our next-door neighbor galaxy would be 20 billion
miles away. No matter how much we try to scale things down to a manageable
size, we still get mind-boggling numbers. The problem is there’s no one scale
factor that will put all the phenomena from
the cosmos to the microworld into a mentally
comprehensible map.

®® If we start with a map of a building near

Washington DC, for example, and a scale
bar representing 100 meters, then when
we expand our field of view by a factor of
10—making the scale bar 1000 meters, or 1
kilometer—we can take in the whole city.
5
 Lecture 1 — Zooming Out to Distant Galaxies

®® If we expand our field of view by another

factor of 10, we start to see regional features,
such as the Chesapeake Bay. And by taking
another step, the scale bar becomes 100
kilometers, and we can see the entire mid-
Atlantic Seaboard.

®® Expand another factor of 10 and we can see

the entire Earth, hanging in empty space. At
this point, we’ve zoomed out from hundreds
to millions of meters.

®® This brings up the issue of units of

measurement. The standard metric unit
for length is meters, including millimeters,
kilometers, and so on. The scale bar we
have just used is 1 million meters long, or
1 megameter. Another way to write that
is with scientific notation: 10 6 meters,
because 10 to the sixth power (1,000,000)
is 1 million.

®® But when we’re thinking about entire

planets, meters are not very convenient;
it’s better to measure things in units of
the radius of the Earth. One Earth radius
is defined as 6378 kilometers. That way,
we can say that the planet Neptune has a
radius of about 4 Earth radii, and Jupiter’s
is about 11. These numbers are much easier
to comprehend than however many millions
of meters. That’s why the Earth radius is a
handy unit; it’s written as R Å, where Å is the
astronomical symbol for the Earth.

6
 Lecture 1 — Zooming Out to Distant Galaxies

®® The next useful unit that we’ll need is for the size of stars. The Sun’s radius is
about 700 million meters, or a little more than 100 times bigger than Earth.
The solar radius, the unit of choice when dealing with stars, is written as R €,
where € is the astronomical symbol for the Sun.

®® Once we get to a scale of 10 13 meters, most of the other planets come into
view. We’ve reached the scale of planetary systems, for which the traditional
unit is the radius of Earth’s orbit around the Sun—a unit called the
astronomical unit (AU). It’s about 215 solar radii, or 150 billion meters. With
the astronomical unit, the solar system can easily be described. Mercury is
about 2 ⁄ 5 of an AU from the Sun; Jupiter is out at 5.2 AU.

®® When we expand the scale again, beyond the solar system, we find ourselves
in empty space for quite a while, until we get to 10 16 meters, at which point
some of the neighboring stars come into view. A good unit to use on this
scale is the light-year, or the distance light travels in 1 year, which is just
short of 10 16 meters. For example, the nearest star, Proxima Centauri, is 4.2
light-years away.

®® In practice, astrophysicists don’t use light-years. Instead, the preferred unit is

called the parsec, and it’s about 3.3 light-years. The typical distance between
stars is 1 or 2 parsecs.

®® From here, we need to zoom out 4 more factors of 10—4 more orders of
magnitude—until the architecture of the Milky Way Galaxy comes into
view, at around 10 20 meters. At this stage, we just keep using parsecs, but
with metric prefixes, such as “kilo-” for 1000. The diameter of a typical spiral
galaxy is 10 or 20 kiloparsecs.

®® It takes a few more orders of magnitude to start seeing neighboring galaxies.

The typical spacing between galaxies is a few megaparsecs, or millions
of parsecs.

7
 Lecture 1 — Zooming Out to Distant Galaxies

®® After another step, the galaxies group together to form clusters of galaxies,
joined by what look like filaments, or webs of galaxies. And when we keep
increasing the scale bar all the way to 10 26 meters, the universe starts to look
like random static, with nowhere different from anywhere else. The natural
scale at this stage is the gigaparsec, or billions of parsecs.

®® That’s the end of the line—the largest spatial scales about which we have any
direct knowledge. By zooming out 26 orders of magnitude, we have a view
of the entire observable universe.

EXPRESSED EXPRESSED IN THE

SCALE UNIT IN METERS PREVIOUS UNIT

Planets RÅ 6.4 × 10 6 m —
Stars R€ 7.0 × 10 8 m 109 R Å
Planetary Systems AU 1.5 × 10 11 m 215 R €
Between Stars pc 3.0 × 10 16 m 206,265 AU
Galaxies kpc 3.0 × 10 19 m 1000 pc
Between Galaxies Mpc 3.0 × 10 22 m 1000 kpc
Observable Universe Gpc 3.0 × 10 25 m 1000 Mpc

LOGARITHMIC MAPS AND CHARTS

®® Another tactic that astrophysicists use to cope with all of these orders of
magnitude is by making logarithmic maps. Taking the logarithm of a number
means expressing the number as a power of 10 and then plucking out the
exponent. For example, 1000 is 10 to the third power, written as 10 3, so the
logarithm of 1000 is 3. The log of 1 million is 6.

8
 Lecture 1 — Zooming Out to Distant Galaxies

®® This also works for numbers smaller than 10. The number 1 is equal to 10 to
the 0 th power, written as 10 0, so the log of 1 is 0; 1 ⁄ 10 is 10 to the −1 power,
written as 10 −1, so the log of 1 ⁄ 10 is −1; and so on.

®® A logarithmic map is an ordinary map based on a single scale factor. For

example, 1 inch on the map might be 1 kilometer in real life. But on a
logarithmic map, the scale factor changes when moving from one end to the
other. The first inch might correspond to 1 meter in real life, but then the
second inch is 10 meters, then 100 meters, 1000 meters, then 10 4, then 10 5,
and so on. Mathematically, with every inch, the logarithm of the scale factor
is increased by 1 unit.

®® Besides maps, there are other logarithmic charts, such as logarithmic time
lines as well as more abstract logarithmic charts that help make sense of things
that range over many orders of magnitude.

®® For example, our galaxy is full of objects ranging widely in mass and size.
Among other things, there are asteroids, moons, planets, and stars. Let’s say
that we go around our galaxy and measure the mass and radius of everything
smaller than the Sun. To compare all these things, we can make a chart
of mass versus radius, with mass on the horizontal axis and radius on the
vertical axis.

®® Each data point shows

the mass (in units of
Earth masses) and
radius (in units of Earth
radii) of a single object.
This chart shows a
relationship between
radius and mass: The
more massive the object,
the bigger the radius,
which makes sense.

9
 Lecture 1 — Zooming Out to Distant Galaxies

®® But there’s a problem with this chart. Because we need to make the axes
range high enough to encompass the very largest objects—millions of Earth
masses—the more numerous, smaller objects end up crammed in close to
the origin and the details are difficult to see.

®® If we remake this chart with logarithmic axes, the horizontal axis is still
telling us the mass, but now each tick mark represents a factor of 10. Likewise,
the vertical axis still tells us the radius, but on a logarithmic scale.

®® This logarithmic chart shows all the data clearly and, even better, some
patterns that were hidden in the ordinary chart. There are 4 different groups,
differing in the relationship between mass and radius. In each of these 4
zones, we can fit the data, at least approximately, with a straight line.

10
 Lecture 1 — Zooming Out to Distant Galaxies

®® On a regular x-y chart, a straight line means that y = ax + b, where a is the slope
of the line and b is a constant, the y -intercept. That’s a linear relationship. But
on a logarithmic chart, a straight line means that there’s a linear relationship
between the logs of the variables: log x = a log y + b.

®® In this case, the log of the radius (R) equals a times the log of the mass plus a
constant: log R = a log M + b.

®® Solving for R results in the following: R ∝ M a .

®® This kind of relationship is called a power law, where one variable is

proportional to another raised to some power. And evidently, when
Mother Nature creates objects in our galaxy, she uses 4 different power law
relationships between radius and mass.

®® Let’s measure the slope—the value of a —in each of the 4 zones.

11
 Lecture 1 — Zooming Out to Distant Galaxies

®® For the lowest-mass objects, the slope is about 1 ⁄ 3. That means R ∝ M  1 ⁄ 3. This
low-mass regime is closest to the one where we have some direct experience:
small things, such as rocks and boulders. And the relation between radius
and mass of everyday objects depends on the density of whatever material
the object is made of. We can understand objects that have masses less than 1
Earth mass; they behave like rocks.

®® In the second zone, the slope is about 1 ⁄ 2, so R ∝ M  1 ⁄ 2. This means that the
more massive objects have bigger radii than we would expect if they all had
the same density. The more massive objects are less dense. The most massive
objects in this zone are a lot less dense than rock; this makes sense because
these are gaseous planets.

®® In the third zone, the slope is 0. The size hardly changes at all, even when
the mass is increased by a factor of 100. In everyday life, when we pack more
mass onto a ball, the ball gets bigger; apparently this is not the case for balls
between 100 and 10,000 Earth masses. The more massive objects are much
denser than the less massive versions. Part of the reason these objects are
increasingly dense is gravitational compression: They are so massive that
their own gravity compresses them to higher densities than usual. The other
part of the explanation is an effect called quantum degeneracy pressure. The
objects in this zone are sometimes called Jovian planets, though toward the
higher-mass end, the traditional term is brown dwarfs.

®® For the highest-mass objects, the slope is about 1, which means that radius
is proportional to mass. These are stars—objects for which gravitational
compression is so strong that nuclear fusion ignites at the center, creating lots
of heat and pressure. This same nuclear fusion also produces the light that
stars are famous for; it’s what makes stars shine.

12
 Lecture 1 — Zooming Out to Distant Galaxies

https://apod.nasa.gov/apod/astropix.html

https://www.youtube.com/
watch?v=0fKBhvDjuy0.

https://www.khanacademy.org/

13
Lecture 2

ZOOMING IN
TO FUNDAMENTAL
PARTICLES

U sually, when something is called “astronomical,”

it means it’s really big. But just as crucial for
astrophysics are the orders of magnitude smaller
than human scales, because the smallest and largest
scales of the universe are deeply connected. In
this lecture, you will be exposed to the realm of
fundamental particles, including electrons, protons,
neutrons, and neutrinos. In addition, you will learn
about the 4 fundamental forces of nature: gravity,
electromagnetism, the strong nuclear force, and the
weak nuclear force.

14
 Lecture 2 — Zooming In to Fundamental Particles

GRAVITY
®® Gravity—probably the most familiar of the 4 fundamental forces of nature—
is what keeps us pinned to the surface of the Earth.

®® Every mass attracts every other mass, according to Newton’s law of gravity,
which says that the force is proportional to the product of the 2 masses
and inversely proportional to the square of the distance between them. The
constant of proportionality is G, Newton’s gravitational constant, which has
a value of 6.7 × 10 −11 N m 2 kg −2.

®® Let’s suppose that we have a big mass, M, held fixed

at the origin of our coordinate system and that
m is free to move—like a planet orbiting a star.
This means that m will feel a pull toward
the origin.

®® To convey the direction of the force, we’ll

use vector notation. A vector is a quantity
with a magnitude and direction. We put
an arrow (→) over the F to remind us it’s a
vector. And, by convention, a “hat” ( ) on top means it’s a unit vector, or a
vector with magnitude of 1, so all it’s doing is specifying a direction. r points
in the direction of increasing r —that is, away from the origin. But the force
is toward the origin, which is why there is a minus sign. The acceleration
vector points in the opposite direction as r.

®® The potential energy associated with the gravitational force also varies as
the product of masses, but it goes inversely with r as opposed to r 2. Again,
it’s negative. But does this make sense? If we let m fall toward the origin, r
shrinks, and according to the formula, the potential energy becomes more
negative, which implies positive energy must be showing up somewhere else,
because the total energy is conserved. And that does make sense: The kinetic
energy, 1 ⁄ 2mv 2, is increasing as the mass accelerates toward the origin. The
gravitational potential energy is being converted into kinetic energy.
15
 Lecture 2 — Zooming In to Fundamental Particles

Logarithmically Zooming In
On the human scale, things are measured in meters. Zooming in to a tenth of a
meter, we center on the human face, and as we keep narrowing our
field of view to a hundredth of a meter, 10 −2, we stare into the
human’s eye.
Another factor of 10, to the millimeter scale, and we can
fit through the pupil of the eye and dive inside. At 10 −4
meters, we can see the blood vessels in the retina, and
by the time we hit 10 −5, we can see individual blood cells.
Once we reach 10 −6 meters, a millionth of a meter—called a
micron—we can see individual bacteria, each one a few microns
across. We’ve zoomed in to the size of the wavelength of light. Light is a wave,
an oscillating pattern of electric and magnetic fields—but it’s hard to
tell this on human scales, because the wavelength is only about
half a micron. On this scale, though, light waves bend and
spread, like water waves, and it’s impossible to focus them
sharply. That’s the phenomenon of diffraction.
After another few orders of magnitude, at 10 −8 meters,
we start to see that the water that surrounds us is not a
continuous fluid. It’s made of individual molecules.
When we zoom in to 10 −9 meters—that is, the scale of
nanometers—molecules don’t look solid. Instead, they look fuzzy, and they’re
in constant motion, jiggling and vibrating. They’re getting knocked around by
other molecules. The energy of all those random motions is what we perceive as
heat on human scales. The hotter the material, the more vigorously the
molecules are bouncing around.

16
 Lecture 2 — Zooming In to Fundamental Particles

Zooming in closer, we see the individual atoms that make up molecules. For
example, a single water molecule is made of 2 hydrogen atoms and 1 oxygen
atom. Like any atom, oxygen has a nucleus, which has a positive electrical charge,
and is surrounded by orbiting electrons, which have a negative electrical charge.
Because of the opposite charges, the nucleus and electrons are
attracted to each other.
Before we can make out any details of the nucleus, we
need to go 4 orders of magnitude below the atomic
scale. The diameter of the oxygen nucleus is about
5 × 10 −15 meters, or 5 femtometers. At this scale, if the
nucleus were a marble, the electron cloud would be the
size of a football field. We can see now that the nucleus
is actually a cluster of 16 little marbles: 8 are protons and 8
are neutrons.
Zooming further, inside the proton, things become very hectic. There are quarks
within a sea of particles called gluons, and everything is in motion, with particles
appearing and disappearing.

ELECTROMAGNETISM
®® The second fundamental force of nature is electromagnetism. The relevant
equation here is Coulomb’s law, which says that the electrical force goes as the
product of charges divided by r 2 and is very similar in form to Newton’s law
of gravity. For the proportionality constant, we’ll use the Greek letter eta (η).
Numerically, η is 9 × 10 9 N m 2 ⁄ C 2, where the Coulomb (C) is the standard
unit of charge. In those units, the electron and the proton both have a charge
of magnitude 1.6 × 10 −19, which can be represented as e.

17
 Lecture 2 — Zooming In to Fundamental Particles

®® Notice that the force law has a plus sign this time, not a minus sign. When the
product of charges is positive—that is, when they’re both the same sign—the
force is repulsive, pushing the charges apart. When the charges have opposite
signs, like an electron and a proton, they attract.

®® The Coulomb energy is the potential energy associated with the electric
attraction or repulsion. As in the case of gravity, it varies as 1 ⁄ r.

®® The Coulomb force explains why the electrons of an atom are attracted to
the nucleus. But there must be something else going on, because why don’t
the electrons fall all the way down onto the nucleus, neutralize it, and come
to rest?

®® We might ask the same question about the Earth: If it’s attracted to the Sun,
why doesn’t it fall in and burn up? The answer in this case is that the Earth has
a nonzero angular momentum—a sideways velocity—and the gravitational
acceleration just keeps turning its velocity vector around in a circle.

®® When we look closely at an atom, we might expect to see the electrons

whirling around the nucleus, like a miniature solar system, but we don’t.
Instead, the electrons look indistinct; there’s an electron cloud surrounding
18
 Lecture 2 — Zooming In to Fundamental Particles

the nucleus. That’s because electrons, like

all fundamental particles, obey the rules The cloud is called a wave
of quantum theory, the counterintuitive
function because the equation
laws of motion and interaction. These
that governs the size and shape
rules are more exact and fundamental
than Newton’s laws of motion.
of the cloud—how it moves and
interacts with other clouds—
®® Quantum theory says that when we resembles the equation for
measure the location of an electron, or ordinary waves. And like regular
any fundamental particle, we get a specific waves, the wave function can
answer. But when we’re not measuring take the form of a pattern moving
it—when we’re not forcing the question through space with a certain
of where it is—the electron spreads out speed. It can even interfere with
into a cloud. And there’s no way to predict other wave functions, producing
exactly where we’ll find it when we do
fringes, like when water
measure it. All we can say is we’re likely
waves overlap.
to find it somewhere in this cloud, or
wave function.

®® In the case of an electron near the nucleus of an atom, the wave function isn’t
moving; it’s trapped by the electrical attraction to the nucleus. It’s like a sound
wave reverberating inside an organ pipe or the vibrating surface of a drum.
And the wave function obeys Heisenberg’s uncertainty principle: If you try to
pin down a particle’s location, by trapping the wave function in a tiny volume,
the particle’s momentum—mass times velocity—becomes more uncertain.

HEISENBERG'S UNCERTAINTY PRINCIPLE

~
x p
2
h
~= 6.6 ⇥ 10 34
J·s
2⇡

19
 Lecture 2 — Zooming In to Fundamental Particles

®® In this mathematical relationship, Δx is the spatial extent of the cloud and

Δp is the extent of the momentum cloud (the range in the possible values of
momentum the particle might have if you measure it).

®® You can’t make both Δx and Δp as small as you might want; their product
is always at least ℏ ⁄ 2, a fundamental constant of nature. The h is Planck’s
constant, 6.6 × 10 −34 joule-seconds, and the small bar through it is shorthand
for h ⁄ 2π.

®® The uncertainty principle explains why atoms are stable. Even if you drop
an electron directly onto a proton, with 0 angular momentum, it doesn’t fall
down and come to rest. That would imply that Δx and Δp are both 0—and
this can’t be. Instead, the wave function strikes a balance between Dx and Dp.

®® The proton exists as a wave function, too, but there’s a big difference. Even
though the proton and electron have charges of equal magnitude, the proton
is much more massive—by a factor of 1800. This ends up causing the proton’s
wave function to be much smaller than the electron’s.

®® Neutrons are nearly the same size and mass as protons but without any
electrical charge. They’re neutral. But if the protons in a nucleus have positive
charge and the neutrons are neutral, then there aren’t any negative charges,
so what’s holding this cluster of “marbles” together? Shouldn’t the protons
repel each other and fly apart?

THE STRONG NUCLEAR FORCE

®® This brings us to the third fundamental force of nature: the strong nuclear
force. This is a very short-range force that acts between nucleons—protons
and neutrons. It’s a complicated force with no simple equation. It depends on
how many nucleons are present, which kinds, whether they’re spinning, and
other things. And it only acts over femtometers. Beyond that, it’s negligible.

20
 Lecture 2 — Zooming In to Fundamental Particles

®® Think of it this way: In a stable nucleus, all the “marbles” are coated with a
layer of “glue” strong enough to withstand the electric repulsion. That’s the
strong force. The strong force is also why the “marbles” are rigid: The force
is attractive up until the point of contact, but then it becomes repulsive, so
it’s very difficult to compress a nucleus.

THE WEAK NUCLEAR FORCE

®® Neutrinos—neutral particles that have a much smaller mass than the
electron by at least a factor of a million—interact mainly through the fourth
fundamental force of nature: the weak nuclear force.

®® The weak force is a short-range force, like the strong force, but it’s not like
any sort of “glue.” In fact, it’s kind of a stretch to call it a force; it’s more like
a special power nucleons have to change identities. A neutron can change
into a proton, or vice versa. For example, a neutron sitting all by itself will
spontaneously turn into a proton within about 10 minutes.

®® The total electrical charge should be conserved, so the new proton’s positive
charge has to be balanced by negative charge somewhere else. What happens
is the weak force conjures up an electron along with the proton and they sail
away in nearly opposite directions.

®® You’d expect them to be exactly opposite, because in addition to charge,

momentum has to be conserved. The initial momentum of the stationary
neutron was 0, so you’d think the proton and electron would have equal
and opposite momenta. But when you measure them, they’re not exactly
opposite. The reason is that the weak force also produces a neutrino that
sails away at nearly the speed of light, carrying just enough momentum so
it all adds up to 0.

21
 Lecture 2 — Zooming In to Fundamental Particles

COMPARING FORCES
®® What sets the nuclear forces apart is you only notice them on femtometer
scales. In contrast, gravity and electromagnetism are long-range forces, acting
on all scales, and their force laws look similar: They both go as 1 ⁄ r 2. But
there are major differences between gravity and electromagnetism, starting
with the fact that electromagnetism is much stronger.

®® Say we have 2 protons separated by some distance r. What’s the ratio between
the force of electric repulsion and the force of gravitational attraction?

®® To find out, we divide the Coulomb force by the gravitational force. The r 2
terms cancel, and when we plug in the numerical values of all the constants,
we find a ratio of 10 36 power. In other words, the electrical repulsion is
unimaginably stronger than the gravitational attraction.

9.0 ⇥ 109 N m2 C 2
1.6 ⇥ 10 19
C

⌘e2
Fe ⌘e2
⇡ 1036
r2
= Gm2p
=
Fg Gm2p
r2

2 27
6.7 ⇥ 10 11
N m2 kg 1.7 ⇥ 10 kg

®® But if gravity is really so pathetic, why is it the most familiar force of nature?
The reason is that gravity is always attractive; it’s never repulsive. There’s no
such thing as negative mass.

22
 Lecture 2 — Zooming In to Fundamental Particles

®® Electromagnetism is different. Here, particles can be positive or negative, and

when they merge together, the result is neutral. It doesn’t feel any electric force.
So, what happens is that all the positive and negative charges in the universe
attract one another. They quickly find each other, getting pulled together with
tremendous force, forming tiny structures with no net charge: atoms. This is
why the incredible strength of the electric force is hidden from us.

Richard Feynman compared the situation of the proton and electron pulling on each
other inside the atom to a pair of Olympic arm wrestlers pulling on each other’s arms
with tremendous force. They’re equally matched; their clenched hands aren’t moving.
From far away, you might not even be aware of their intense effort.

®® Once neutral atoms form, all that’s left of electric forces are the slight
imbalances that arise because the negative charge, the electron cloud, is more
spread out than the positive charge, the nucleus.

®® In fact, what we perceive as everyday forces—our feet pushing on the ground,

our hands pulling on a rope, our knuckles knocking on a door—are all
complex manifestations of the residual forces that are left over from the
combination of electromagnetism and quantum theory.

®® Gravity, on the other hand, never gets cancelled. That’s why, when we zoom
out to astronomical scales, gravity is the dominant force. That’s why gravity—
even though it’s weak—sculpts the properties of planets, stars, and galaxies.

®® Electromagnetism is more than just attraction and repulsion. There are

magnetic fields, which come from moving charges, and there is electromagnetic
radiation, which comes from accelerating charges. Whenever you accelerate a
charge—speed it up, slow it down, whirl it around—it radiates. It takes some
of its own energy and flings it outward at the speed of light. The radiated
energy takes the form of photons.

23
 Lecture 2 — Zooming In to Fundamental Particles

This chart of distance scales from the previous lecture now

includes 3 key units discovered during this lecture’s zoom-
in: the micron (same order of magnitude as the wavelength
of visible light), the Bohr radius (the atomic scale), and the
femtometer (the nuclear scale).

EXPRESSED EXPRESSED IN THE

SCALE UNIT IN METERS PREVIOUS UNIT

Nucleus fm 10 −15 m —
Atom a0 5 × 10 −11 m 52,000 fm
Visible Light µm 10 −6 m 18,900 a 0
Human m 1m 10 6 µm
Planets RÅ 6.4 × 10 6 m 6.4 × 10 6 m
Stars R€ 7.0 × 10 8 m 109 R Å
Planetary Systems AU 1.5 × 10 11 m 215 R €
Between Stars pc 3.0 × 10 16 m 206,265 AU
Galaxies kpc 3.0 × 10 19 m 1000 pc
Between Galaxies Mpc 3.0 × 10 22 m 1000 kpc
Observable Universe Gpc 3.0 × 10 25 m 1000 Mpc

24
 Lecture 2 — Zooming In to Fundamental Particles

https://apod.nasa.gov/apod/

https://www.youtube.com/
watch?v=0fKBhvDjuy0

https://www.youtube.com/watch?v=jfSNxVqprvM

25
Lecture 3

MAKING MAPS
OF THE COSMOS

I f we want to explore the whole universe, we’d

better have a good map. This lecture is about how
astronomers locate objects in the universe.

If you want to specify the location of an object in 3

dimensions, you use Cartesian coordinates; you choose
an origin and pick 3 perpendicular directions to be the
x, y, and z axes. Instead of this system, astronomers
use latitude, longitude, and elevation. They use a
spherical polar coordinate system—specifically, an
Earth-centered coordinate system—and extend
the concepts of latitude, longitude, and
elevation up into the heavens.

26
 Lecture 3 — Making Maps of the Cosmos

Imagine a giant transparent sphere that is centered on the Earth and marked with
grid lines of latitude and longitude. The latitude lines tell us how far we are from
the celestial equator—the projection of the Earth’s equator up into the sky—and
the longitude lines tell us how far east or west we are from the
celestial equivalent of the prime meridian. That way,
when we look at a distant star, we can read
off the star’s angular coordinates by seeing
where it appears relative to the grid.
That leaves only the third dimension:
the distance to the star, the celestial
equivalent of elevation, which is much
trickier to measure.

ANGULAR COORDINATES
®® Say there are 2 stars that happen to be located along nearly the same line of
sight from the Earth, so they appear close together on the celestial sphere.
If they’re too close, they blend together and appear as a single point of light,
rather than 2. What determines whether we can perceive the double star?

®® Generally, it depends on how good the telescope is, but we can be more
specific than that. This question pinpoints the basic dilemma of astronomy—
that all we have is light. With few exceptions, our only source of knowledge
is the electromagnetic radiation that happens to hit the spinning ball of rock
we live on. So, we need to understand the physics of light.

27
 Lecture 3 — Making Maps of the Cosmos

®® Imagine a telescope as a big lens pointed at a star straight overhead that

focuses the starlight into a tight spot on a camera. If there’s another star in
a slightly different direction, then ideally, the lens focuses its light onto a
different spot in the image, which shows 2 dots: star 1 and star 2.

®® The problem, though, is we

can’t focus light into as small
a point as we might like. The
stars blend together when
the angle between them,
∆θ, is of order λ ⁄ D, where
λ is the wavelength of light
and D is the diameter of the
lens—or mirror, or whatever
we’re using to collect and
focus light.

®® The reason for this inevitable

blurring is the phenomenon
called diffraction, a consequence
of the wave nature of light.
Light is an electromagnetic
wave, a traveling pattern of
oscillating electric and
magnetic fields.

®® So, we can imagine the

light from star 1 as an ocean
wave, a traveling pattern of crests and
troughs of electromagnetic energy, with a wavelength—a
separation between crests—of λ. This wave passes through the diameter D
of the telescope and then a lens or mirror responds to that pattern of energy
by redirecting it toward a camera.

28
 Lecture 3 — Making Maps of the Cosmos

®® The energy gets redirected to a different

position on the camera if the starlight comes
in from a different angle, tilted by ∆θ. But
if ∆θ is tiny, how could the lens or mirror
possibly tell the difference? The wave energy
is smeared out, with a spatial extent of λ.
So, the telescope still sees a crest filling
the opening. The optical system responds
essentially the same way as before, directing
the energy to the same spot on the detector.

®® As ∆θ increases, at what point do the waves

from star 2 start to look different from star
1? The answer is when you no longer have a
crest extending across the opening. The tilt
is large enough that there’s a crest at one
end and a trough at the other end. For this
minimum value of ∆θ, it’s at least possible
for the optics to distinguish the 2 waves.

®® Because the distance between a crest and a

trough is half of λ,

λ ⁄2
tan ∆θmin =  .
D
®® In practice, ∆θmin is a tiny number. For visible light, λ is about half a
micron while D is usually a meter or more. So, we can use the small-angle
approximation: Whenever an angle is small, the tangent is nearly equal to the
angle itself, expressed in radians. In this case, this means that

∆θmin ≈ λ .
2D
®® That’s the minimum angular separation we can reliably measure with a
telescope of diameter D.

29
 Lecture 3 — Making Maps of the Cosmos

®® Keep in mind that this is just an order-of-magnitude relationship, based

on the rough idea that we need to tilt the waves enough to get a shift of
order λ across the telescope opening. To get this exactly right, we’d need to
calculate the diffraction pattern of light after it passes through a circular
hole of diameter D.

®® What that more complex calculation shows is that the starlight gets focused
into a blob surrounded by a pattern of rings, and the diameter of the blob is
1.22 λ ⁄ D. So, if 2 stars are closer than that, their blobs merge together. That’s
why the usual definition of this diffraction limit is 1.22 λ ⁄ D.

λ ⁄D is a dimensionless number—a length divided by a length. That means the

angle is measured in natural units, or radians.
In ordinary life, we measure angles in degrees, with a right angle being 90° and
360° going around the whole circle. Using the number 360 is a tradition going
at least as far back as the ancient Babylonians, who used a base-60 counting
system. That’s also why we divide 1 degree into 60 arc minutes and 1 arc minute
into 60 arc seconds.
It’s simpler to calculate in radians, the system in which a right angle is π ⁄ 2
and the whole circle is 2π.

®® Because the diffraction limit is proportional to λ ⁄ D, the way to improve

our angular resolution is to increase D: build a bigger telescope. And that
works, to a point. But in practice, there are other reasons why our images
are blurry besides diffraction. Maybe our lens isn’t polished perfectly or our
mirror has defects. And then there’s the constantly fluctuating atmosphere,
which scrambles the directions of light rays at the level of an arc second, even
at our best mountaintop observatories. So, even with a large telescope, we
usually cannot achieve the ultimate diffraction limit. That’s one reason why
we launch telescopes into space, above the atmosphere.

30
 Lecture 3 — Making Maps of the Cosmos

MEASURING THE DISTANCE

FROM EARTH
®® Let’s return to the original problem of locating things in space. The really
tough problem—historically the most difficult problem in astronomy—is
measuring that third dimension: the distance from Earth.

®® Imagine we discover some new galaxy. We measure its angular extent on

the celestial sphere. With only that information, we can’t tell if the galaxy is
relatively small and nearby or huge and billions of light-years away.

®® So, we’re on the Earth and an object is located a distance d from Earth. The
object has a true size of S and an angular size of α —that is, the rays arriving
from opposite sides of the object have an angle of α between them.

®® We use the small-angle approximation again, because in practice α is tiny

(maybe just a few microradians). That means, α ≈ S ⁄d, or, equivalently, S = αd.
So, if all we know is α, we can’t figure out S.

®® A similar situation arises when we measure the brightness of a source. For a

given brightness, we can’t tell if the source is intrinsically luminous and far
away or if it’s actually intrinsically faint and happens to be nearby.

31
 Lecture 3 — Making Maps of the Cosmos

®® Suppose the luminosity of the source is L . That’s the power—the energy

per unit time—that the luminous object is pouring out into space. We
can measure L in watts, for example, and all that power spreads out as the
radiation goes farther from the source.

®® The Earth is far away, at a distance d. By the time the light reaches Earth,
it’s been spread out over a huge sphere of radius d.

®® We can’t measure L directly. Instead, we have a telescope with a certain

collecting area, and we measure the power received by the telescope. We then
calculate the power per unit area, which is called the flux, F.

®® The following equation, representing the flux-luminosity relation, is an

example of an inverse square law: The flux goes down as the inverse square
of the distance.
L
F =
4⇡d2
®® If we measure F, we need to know d in order to deduce L . And we need to
know L , the true luminosity, if we want to figure out what’s physically going
on to produce that radiation.
32
 Lecture 3 — Making Maps of the Cosmos

®® Measuring the distance to an astronomical object is a crucial problem we need

to solve. And we do so in 4 different steps: radar, parallax, standard candles,
and standard explosions.

RADAR
®® Build a giant radio transmitter, aim at a nearby planet, and fire. If you hit
the target and your receiver is sensitive enough, you can detect the echo—the
reflected radio waves. The echo is delayed by a time interval ∆t = 2d ⁄ c, where
2d is the round-trip distance and c is the speed of radio waves (that is, the
speed of light). And because we know the speed of light, we can calculate d.

®® With the world’s biggest transmitters, we can measure the distances to

Mercury, Venus, Mars, and even some asteroids. That allows us to make
maps of the solar system with a precision of a few parts in 10 billion, an
astonishing level of detail.

®® But unfortunately, this method is limited to relatively nearby objects. You

can show that the amplitude of the echo falls off like 1 ⁄ d 4. To go beyond the
solar system, we need other techniques. We move on to the second step of
the 4-step solution: parallax.
33
 Lecture 3 — Making Maps of the Cosmos

PARALLAX
®® Parallax is based on simple geometry. Hold out your arm and raise a finger.
Next, close your left eye and look at your finger and the scene in the
background. Then, switch eyes: Open your left and close your right eye.
It looks like your finger jumped! That’s because your right eye views from
a slightly different angle, so it sees your finger projected against a different
part of the background scene.

®® That’s parallax. If you measure that shift in angle, as well as the distance
between your eyes, you could use trigonometry to calculate the distance to
your finger.

®® In astronomy, we take advantage of the Earth’s motion around the Sun. We

take a picture of a nearby star and then wait 6 months for the Earth to go
halfway around and take another picture. That’s like closing your left eye
and opening the right. The nearby star, like your finger, will appear to have
shifted in position relative to the more numerous background stars.

®® Let’s do the math. Here’s the

Earth going around the Sun in a
nearly circular orbit, with a radius
of 1 AU. A star is at distance d.
We need to point our telescope
in a certain direction to see the
star. But 6 months later, we need
to point in a slightly different
direction. In fact, as the year
progresses, the star will seem
to move in a little circle on the
celestial sphere, with an angular
radius of α.

®® Using the small-angle approximation, α = 1 AU ⁄ d, or d = 1 AU ⁄ α.

34
 Lecture 3 — Making Maps of the Cosmos

®® And because we already know the value of the astronomical unit very precisely
from radar ranging, whenever we measure α, we can calculate d.
↵arcsec = 206,265 ↵
206,265 AU
d=
↵arcsec

1 pc ⌘ 206,265 AU
1 pc
d=
↵arcsec

®® For simplicity, the star is shown directly above the plane of Earth’s orbit.
In general, though, the star will be off to the side somewhere. That doesn’t
change the basic idea. It just means that the star will appear to move in an
ellipse, rather than a circle, and the parallax angle is the semimajor axis of the
ellipse. And if the star is right on the ecliptic—the projection of the Earth’s
orbit onto the celestial sphere—it’ll go back and forth along a straight line.

As is the case with the diffraction limit, the equation α = 1 AU ⁄ d works when α is
expressed in radians. But what if we want to use arc seconds?
One radian works out to be 206,265 arc seconds, so if we’re expressing α in arc
seconds, the right side of the parallax equation becomes 206,265 AU ⁄ d.
The tradition at this point is to define a new unit of distance, the parallax second,
or parsec, equal to 206,265 AU. That way, the numbers are easier:
d = 1 parsec ⁄ α in arc seconds.
The parsec is a handy unit for measuring the distances between stars, and it
happens to have the same order of magnitude as the light-year.
1 parsec ≈ 3.3 light-years

35
 Lecture 3 — Making Maps of the Cosmos

®® Parallax is by far our reliable method for measuring distances to stars. But as
the distance gets larger, eventually the parallax angle becomes too small to
measure—if for no other reason than the diffraction limit.

®® Right now, our best parallax measurements come from a space telescope
called Gaia, launched by the European Space Agency in 2013. It measured
parallaxes as small as 0.0001 of an arc second, good enough to make maps
of the galaxy out to 10,000 parsecs, or 10 kiloparsecs.

®® That’s impressive. But to go beyond our galaxy—and there’s a lot beyond

our galaxy—we need to take another step in the quest to measure distances.

STANDARD CANDLES
®® Our 2 best ways to measure more distant objects both rely on the flux-
luminosity relation derived earlier: F = L ⁄ 4πd 2, where L is the power an object
emits and F is the power per unit area measured by Earthlings. We can
measure F, but we can’t figure out L unless we also know the distance, d.

®® But suppose there were some light source out there for which we already
knew L . In that case, we could calculate the distance by rearranging the
flux-luminosity equation.

L
F =
4⇡d2
r
L
d=
4⇡F

®® RS Puppis is an example of a category of stars called Cepheid variables—so

called because the first known example was in the constellation Cepheus and
because they vary in brightness. They pulse, getting brighter and fainter, in
an endless cycle, with a period of typically a few weeks.
36
 Lecture 3 — Making Maps of the Cosmos

®® The average luminosity of a Cepheid can be predicted accurately from the

period of the pulsations, as discovered by Henrietta Leavitt in 1912. Stars
that pulse more slowly are intrinsically more luminous.

®® One reason we know this

to be true is that some
Cepheids are close enough
for parallax measurements,
so we can determine
their luminosities. And
among that collection,
we observe that L is
linked to the pulsation
period, P. A schematic
chart of luminosity
versus period shows this
increasing relationship.

®® So, if we spot a Cepheid a megaparsec away, in some other galaxy, we can’t

measure its parallax, but we can measure the pulsation period. We just
monitor the flux and see how long it takes to rise and fall. Then, we use the
period-luminosity relationship to determine L and then calculate d, which is
the distance to the galaxy where the Cepheid resides.

Cepheid variables were what first allowed us to zoom out beyond the Milky Way.

®® Cepheids are examples of what are called standard candles — sources for
which we somehow know the luminosity. Even though Cepheids have been
used for more than a century to map out our galactic neighborhood, we don’t
know exactly why they are such good standard candles.

37
 Lecture 3 — Making Maps of the Cosmos

®® With our best telescopes, Cepheids can be seen out to a distance of around 50
megaparsecs. But if we want to go farther—to reach out to gigaparsecs—we
need to use standard explosions.

STANDARD EXPLOSIONS
®® In the 1980s, astronomers realized that a certain category of exploding stars,
or supernovas, produce fireballs that all have nearly the same peak luminosity.
They’re called Type Ia supernovas, and they all explode with nearly the same
energy. They’re predictable enough so that if you measure the color and
duration of the afterglow, you can determine its luminosity to within a few
percent. We know this because we’ve spotted Type Ia supernovas in nearby
galaxies that also have Cepheids in them.

®® Here’s a chart showing data

for some nearby Type Ia
supernovas. The horizontal
axis is time (in days) and
the vertical axis shows
the measured luminosity
of the explosion. Notice
that all the supernovas
rise to about the same
level, with differences that
correlate with the duration:
The faster the explosion
fades, the lower the peak
luminosity is.

®® When we measure the rise and fall of flux from a really distant Type Ia
supernova, we can match the observed duration of the event to one on this
chart and then read off the peak luminosity.

38
 Lecture 3 — Making Maps of the Cosmos

®® The great advantage of using Type Ia supernovas as standard explosions is

that they’re as bright as 5 billion Suns, bright enough to see even when they
happen in galaxies that are extremely far away.

®® We don’t know for sure what causes Type Ia supernovas. They’re almost
certainly exploding white dwarfs, but the trigger for the explosion remains
a topic of active research. What is clear, though, is that we can use them to
measure cosmological distances. They’re the last step in our quest.

39
Lecture 4

THE PHYSICS
DEMONSTRATION
IN THE SKY

T o the naked eye, the planets of the solar system are

points of light that move from night to night relative
to a fixed background of stars. It was this celestial
physics demonstration that led Isaac Newton to his law
of gravity. The planets move in response to the Sun’s
gravity, without any friction. And each one is a different
distance from the Sun, so by observing them, you
can figure out how gravity depends on distance. The
goal of this lecture—and the next one—is to solve this
physics problem in the sky.

The name “planet” comes from an ancient Greek word for “wanderer,” because
the planets wander through the constellations.

40
 Lecture 4 — The Physics Demonstration in the Sky

KEPLER’S FIRST LAW

®® The 17th-century astronomer Johannes Kepler pointed
out 3 patterns in the motion of the planets.
Kepler’s first law is that the planets trace out
ellipses as they go around the Sun. The
orbits look like circles, but they’re not;
they’re slightly flattened into ovals. And
the Sun is not at the center.

®® Before getting to the physics of why the

orbits are ellipses, let’s address the geometry
of ellipses. Let’s start simple, though, with a
circle. Mathematically, a circle is defined as the
set of all points that are the same distance from
some chosen center.

Wouldn’t it be simpler if the orbits of planets were circles? The ancient Greeks
certainly thought so. And later, even when the data got better and proved to
be inconsistent with circular motion, theoreticians didn’t abandon circles—they
doubled down on them. They had the planets move in circles, the centers of
which were themselves moving in circles. These are the “epicycles” that became
the basis for the Ptolemaic model for the solar system, which prevailed up
until Kepler.

®® An ellipse has 2 focus points, or foci,

and for all the points on the ellipse,
the distance to the first focus plus
the distance to the second focus
is a constant. And that constant
is equal to the length of the long
axis of the ellipse, the major axis.
41
 Lecture 4 — The Physics Demonstration in the Sky

®® If a circle has a radius of a, the equivalent for an ellipse is the radius along the
major axis, called the semimajor axis, which can also be labeled a. With an
ellipse, we also have the distance between the center and either focus, which
can be whatever we want, as long as it’s smaller than a. Tradition dictates
that we express that distance as ae, where e —or eccentricity—is a number
smaller than 1.

®® When e is 0, the foci coincide at the center and we have a circle of radius
a. As e gets larger—and closer to 1—the foci separate and we get a more
elongated ellipse.

®® The area of a circle is πa 2. For an ellipse, area is

p
area = ⇡a2 1 e2 .

®® Next, we need the mathematical equation for an ellipse in polar coordinates.

We start by introducing a coordinate system. We’ll put the origin at one of
the foci—for example, the one on the right—and we’ll lay down x and y axes
along the major and minor axes. To specify the points on the ellipse, we use
polar coordinates: r is the distance from the origin and θ is the angle measured
counterclockwise from the x axis.

®® What is the equation for r as a function of θ?

®® To find the equation, we start with

the fact that at any point, the
sum of distances to the foci is
equal to the length of the major
axis, 2a. We can write that as
r + r ′ = 2a , where r is the distance
to the focus at the origin and r ′ is the
distance to the other focus.

®® But we want the equation to be purely in terms of r and θ, not r ′. To get rid
of the r ′, we use the law of cosines.
42
 Lecture 4 — The Physics Demonstration in the Sky

®® The Pythagorean theorem says c 2 = a 2 + b 2, where a, b, and c are the lengths
of the sides of a right triangle. The law of cosines is the generalization to any
triangle, with γ being the angle across from the c side.

c 2 = a 2 + b2

c 2 = a2 + b2 2ab cos

®® Let’s apply the law of cosines to this

triangle, with r′ as the c side.

That’s our equation!

43
 Lecture 4 — The Physics Demonstration in the Sky

®® What Kepler noticed—his first law—is that all the planets move in ellipses,
with the Sun not at the center but at one of the foci.

®® Each planet has its own semimajor axis and eccentricity. For the Earth, a
is about 93 million miles, or 150 million kilometers, or 1 AU. The Earth’s
orbital eccentricity is 0.017, which is quite small. All the planets in the solar
system have small eccentricities. That’s why it took so long to notice that the
orbits are not circles. Mercury has the most eccentric orbit, with e = 0.21.

®® Some of the planets around other stars that have been discovered over the
last few decades have larger eccentricities, some even larger than 0.9. These
incredibly elongated orbits were one of the big surprises of exoplanetary science.

KEPLER’S SECOND LAW

®® Kepler’s second law is about how fast the planets move. When they’re close
to the Sun, they move faster, in a specific way. As the planet moves, the line
joining the planet and the star—the planet’s radius vector—sweeps out area
at a steady rate.

®® When the planet is near the

star, the radius is shorter
but the planet moves
faster, so the swept-
out sector has a large
angular width; when
the planet is far away,
the radius is longer but the
planet is slower, making a tall
and skinny sector. Kepler found
that all the sectors have the same area.

44
 Lecture 4 — The Physics Demonstration in the Sky

®® Let’s put the planet at an arbitrary position and say it moves for an infinitesimal
time interval dt. It sweeps out a thin sector spanning an angle of dθ, with
an area of dA.

®® What is the area of the sector? The planet moves in both the radial direction,
or the r direction, and the perpendicular direction, or the θ direction—the
direction of increasing θ. It’s the motion in the θ direction that sweeps out
area; purely radial motion doesn’t sweep any area.

®® In time dt, the planet moves in the θ direction by an amount rdθ, using the
small-angle approximation. So, the swept-out sector is basically a skinny right
triangle with sides of r and rdθ. The area of that triangle is
1
dA = r · rd✓.
2
®® That leaves out a tiny corner piece of the
sector whose dimensions are dr and rdθ, the
product of 2 tiny numbers. In the limit of
infinitesimal dt, that piece is vanishingly
small compared to the rest of the triangle.
This means that we can write
1
dA = r · rd✓.
2
dA 1 d✓
®® And dA ⁄ dt, the rate at which area is swept out, is = r2 .
dt 2 dt
®® If that rate is a constant, as Kepler observed, then dθ ⁄ dt must be proportional
to 1 ⁄ r 2. In other words, θ advances at a rate that varies as 1 ⁄ r 2.

45
 Lecture 4 — The Physics Demonstration in the Sky

KEPLER’S THIRD LAW

®® Kepler’s third law is about total time required to go all the way around: the
orbital period. The bigger the orbit, the longer the period.

®® In this logarithmic chart, the horizontal axis shows the semimajor axis in
AU and the vertical axis shows the period in Earth years. So, the point
representing Earth is at 1 AU and 1 year, and the other points are for the
other planets.

®® It’s striking—they all fall on a single straight line! It has a slope of 3 ⁄ 2; if we
move 2 units to the right, the line goes up 3 units. Because this is a log plot,
that means
3
log P = log a + constant.
log a + const.
2

®® This in turn means that P / a3/2. That’s Kepler’s third law.

46
 Lecture 4 — The Physics Demonstration in the Sky

®® Unlike Kepler, we now know that other planets have orbiting bodies, or
moons. The data points for the 4 biggest moons of Jupiter all lie on a straight
line, too—with the same slope of 3 ⁄ 2. But interestingly, it’s not the same
line as the one defined by the planets. It’s shifted up, to a longer period for
a given semimajor axis.

®® Why is P proportional to a 3 ⁄ 2, and why do the moons of Jupiter have a larger
proportionality constant?

NEWTON’S LAWS OF
MOTION AND GRAVITY
®® Kepler didn’t understand why his laws—which should really be called
patterns—hold. That task fell to Isaac Newton.

®® Newton’s law of motion is that the force acting on a body equals the mass of
that body times its acceleration:
d~v
F~ = m~a , or F~ = m .
dt
GM m
®® Newton’s law of gravity is F
~g = r̂, where M is the Sun’s mass and m
r2
is the planet’s mass.

®® How do these laws relate to Kepler’s laws? Kepler’s second law is the most
fundamental, so let’s start there. The key concept is the conservation of
angular momentum.

®® But before getting to angular momentum, let’s consider momentum and velocity.
Momentum is mass times velocity, p = mv, and the velocity has 2 components.
In time dt, the velocity takes the planet from one position to another, changing
both r and θ, so the velocity has a radial component—toward or away from the
origin—and an angular component in the perpendicular direction.

47
 Lecture 4 — The Physics Demonstration in the Sky

®® The radial component, vr, is equal to dr ⁄ dt, and the

angular component, vθ, is rdθ —the distance moved in
the direction of increasing θ —divided by dt.

®® Angular momentum, L , is defined as rmvθ. Only the

angular component—the sideways component—
of the velocity matters. And because we have an
equation for vθ, we can write
d✓
L = r · mv✓ = mr2 .
dt

®® In vector language, L
~ = ~r ⇥ m~v .

®® The cross product is the way to pick out only the perpendicular component;
it has a magnitude of r times the component of mv that’s perpendicular to r.

®® In some circumstances, angular momentum is conserved. It doesn’t change

with time, even if the body is changing in other ways. The classic example
is the twirling figure skater who pulls in her arms, effectively reducing her
r —which means her dθ ⁄ dt must increase to compensate. That’s why she
twirls faster.

®® Angular momentum is conserved whenever there’s no net torque—no force in

the θ direction. This is certainly true for the planets; the only force is gravity,
which is in the radial direction, toward the Sun.
d✓
®® So, as a planet goes around, even though r and v are always changing, r2 = const.
dt
is a constant and therefore
d✓ 1
/ 2.
dt r

®® That’s Kepler’s second law! This shows that Kepler’s second law is a
consequence of the conservation of angular momentum. Just as the ice skater
twirls faster when she pulls in her arms, the planets twirl faster when they
approach the Sun.
48
 Lecture 4 — The Physics Demonstration in the Sky

®® This is an important result, with implications beyond planetary motion. It

helps explain why material speeds up as it spirals into a black hole, why a star
spins faster when it contracts in size, and why a young star is surrounded by
a spinning disk of material within which the planets are formed.

®® In Kepler’s third law, why is the orbital period proportional to the 3 ⁄ 2 power
of the semimajor axis?

®® We’ll answer this question in 2 stages: First, we’ll prove Kepler’s third law
for a circular orbit, and then, in the next lecture, we’ll prove it for the general
case of elliptical orbits.

®® Imagine a planet moving in a circle of

radius a with some constant speed
v. Over a full orbital period, P, the
planet travels all the way around the
circle. Therefore, v must equal the
circumference of the circle, 2πa, divided
by P. Or, equivalently, P = 2πa ⁄ v.

®® One reason why P increases with a is that the

circumference of the circle gets bigger. There’s a longer way to go.

®® In addition, when a is larger, v is lower; the planet moves more slowly because
the gravitational attraction is weaker. This increases P even more so that at
the end of the day P goes as a 3 ⁄ 2.

®® In a time dt, the planet advances by a small

angle dθ, which corresponds to an arc length
of adθ. So,
d✓
v=a .
dt

49
 Lecture 4 — The Physics Demonstration in the Sky

®® During that same time interval, the velocity vector rotates by the same
angle dθ. The change in the velocity vector is vdθ, so the magnitude of the
acceleration, the rate of change of velocity, dv be v d✓ .
accel. must
= =
dt dt
®® Let’s combine the equations by solving the first one for dθ ⁄ dt and then
inserting the answer, v ⁄ a, into the second equation. This gives
⇣v⌘ v2
acceleration
=v = .
a a

®® We just proved that to keep a body moving at speed v in a circle of radius a, an

inward acceleration—a centripetal acceleration—of v 2 ⁄ a needs to be supplied.

®® In the case of a planet, that acceleration is provided by the Sun’s gravitational

2
force, GM .= v
a2 a
2
®® We set that equal to v 2 ⁄ a: GM = v .
a 2 a

®® Then, we solve for v, finding

GM
= v2
a
r
GM
v= .
a

®® We can insert this into our previous expression for the period, and we find
that P is proportional to a 3 ⁄ 2.
2⇡a r
P = GM
v v=
a
r
a
P = 2⇡a
GM
2⇡
P =p a3/2
GM

50
 Lecture 4 — The Physics Demonstration in the Sky

®® We also find that the proportionality constant

P =is p2⇡ . a3/2
GM
®® It decreases with the mass of the attracting body. That’s why there was a
vertical offset between the data for the planets and for the moons of Jupiter.
The Jovian moons have a longer period for a given a because Jupiter is less
massive than the Sun.

Kepler’s third law is the most reliable way we have to measure the mass of just
about anything in astrophysics. The basic idea is that to measure an object’s
mass, we need to watch other things moving in response to its gravity. It works
for stars, planets, black holes, neutron stars, and entire galaxies. It even works—in
a sense—for measuring the mass of the entire universe.

51
Lecture 5

NEWTON’S
HARDEST PROBLEM

I magine that the year is 1660. The law of gravity is

unknown. You’ve just read Johannes Kepler’s books
and puzzled over the 3 patterns he observed in the
motion of the planets. Can you use those patterns to
figure out the law of gravity? This is a tough problem
that took Isaac Newton years to solve. But you have an
advantage: calculus.

Kepler’s Laws in Equation Form

Kepler’s law says that the orbits of the planets are ellipses with the Sun at one
focus, so if we use a polar coordinate system with the Sun at the origin, the path
of the planet, r (θ) is

 
.

52
 Lecture 5 — Newton’s Hardest Problem

This is the equation for an ellipse that we derived in the previous lecture, in terms
of the semimajor axis, a, and the eccentricity, e.
Kepler’s second law says that the line from the Sun to the planet sweeps out area
at a steady rate. In the previous lecture, we showed that this implies that

is a constant—a certain area per unit of time—specific to each planet. For the
Earth, the numerical value is π (AU) 2 per year, because the Earth’s orbit is
approximately a circle of radius 1, which has a total area of π.
More generally, is equal to the area of the ellipse divided by the orbital
period, P :

Kepler’s third law says that P is proportional to a 3/2.

THE VELOCITY VECTOR

®® We already know from laboratory experiments that force equals mass times
acceleration, but we don’t yet know the equation for the force of gravity. To
obtain a clue, we need to calculate the acceleration of a planet that obeys
Kepler’s laws.

®® To calculate acceleration, first we need to know the planet’s position as a

function of time. Then, we’ll take the time derivative to get the velocity.
Then, we’ll take another time derivative to get the acceleration.

53
 Lecture 5 — Newton’s Hardest Problem

®® Kepler’s first law tells us the position, but not as a function of time—it’s a
function of angle, θ. All the time information is the second and third laws.
So, we need to combine the equations somehow.

®® Another problem is that we wrote Kepler’s first law in polar coordinates, but
with vectors, such as acceleration, it’s easier to take derivatives in Cartesian
coordinates, x and y. So, let’s convert to Cartesian coordinates.

®® In general, when the polar coordinates are r and θ, the x coordinate is r cosθ,
and y = r sinθ.

®® So, for our planet,

®® And we get a similar equation for y:

®® We can do the same thing with

unit vectors. In the polar coordinate
system, r is a vector of length
1 pointing in the direction of
increasing r. That means r changes
orientation as the planet moves
around; it always points away from
the origin. At any point, though, we
can write r as

r̂ = x̂ cos ✓ + ŷ sin ✓.

54
 Lecture 5 — Newton’s Hardest Problem

®® Now let’s calculate the velocity by taking the time derivative of x and y.
Because they’re written as functions of θ, not time, we need to use the chain
rule. vx, the x component of velocity, is dx ⁄ dt, which is dx ⁄ dθ times dθ ⁄ dt. To
get dx ⁄ dθ, we use standard tools of calculus. Because x has functions of θ in
the top and bottom of the expression, we use the quotient rule. We take the
derivative of the top times the bottom, minus the top times the derivative of
the bottom, over the bottom squared. And we can simplify a bit.

®® For dθ ⁄ dt, we need Kepler’s second and third laws, the ones relating to time.
Let’s consolidate them by writing the P in the second law in terms of a using
the third law, which says that P equals some constant, K, times a 3 ⁄ 2. Because
the second law has a 1 ⁄ 2 on the left side and a π and a square root on the
right side, we can cancel out the 1 ⁄ 2 and the π and fit the K under the square
root and write the third law as follows.

55
 Lecture 5 — Newton’s Hardest Problem

®® Now we have the ingredients we need to calculate the velocity. We plug in

the expressions we just derived, which leads to an equation in terms of r and
θ. To put everything in terms of just one variable, θ, we insert the ellipse
equation for r (θ) and simplify.

®® The factor in front of sinθ is a constant—it doesn’t depend on θ or time—and

it has units of velocity. To make the equation look even simpler, let’s name
that constant v 0 and simplify.
s
K
= sin ✓
a(1 e2 )

vx = v0 sin ✓

®® That leaves the other component of velocity,

vy, which we calculate as dy ⁄ dθ times dθ ⁄ dt.
The steps are similar to the ones for vx.

56
 Lecture 5 — Newton’s Hardest Problem

®® What does all this mean? Let’s find out by tracking the planet’s velocity vector
over a full orbit. We’ll plot vx on the horizontal axis and vy on the vertical
axis. This kind of chart is called velocity space; each point specifies a velocity,
rather than a position.

®® The equations tell us that vx starts at 0 and vy starts at v 0 + e. Then, as θ

increases, vx goes negative and vy shrinks. When we keep going, we find that
the tip of the velocity vector moves in a circle!

57
 Lecture 5 — Newton’s Hardest Problem

®® We can prove it algebraically, too, by showing that our equations imply the
equation for a circle in velocity space with radius v 0 and centered at the point
(0, ev 0).

®® While a planet moves in an ellipse, its velocity vector traces out a circle.

THE ACCELERATION VECTOR

®® The x component of acceleration, ax, is dvx  ⁄ dt, which we can use the chain
dvx as dvx d✓ .
rule, as we did previously,a to=write = = v cos ✓
d✓
x 0
dt d✓ dt dt
®® Substituting for v 0 and dθ ⁄ dt,
s p
K Ka(1 e2 ) K
= cos ✓ = cos ✓.
a(1 e2 ) r2 r2

®® The y component works the same way. We take the θ derivative to get sinθ
and then plug in dθ ⁄ dt, leading to

dvy dvy d✓ d✓
ay = = = v0 sin ✓
dt d✓ dt dt
s p
K Ka(1 e2 ) K
= sin ✓ = sin ✓.
a(1 e2 ) r2 r2

®® In vector notation, we just learned that acceleration is

K K
~a = (x̂ cos ✓ + ŷ sin ✓)= r̂.
r2 r2

58
 Lecture 5 — Newton’s Hardest Problem

®® It all hangs together, if the Sun is pulling the planet toward it with a force
whose strength varies as 1 ⁄ r 2. We just “discovered” the law of gravity by
following Kepler’s 3 clues.

®® We now see that the constant K that appeared in Kepler’s third law sets the
overall strength of the Sun’s gravitational force. If we further assume that
the force is proportional to the mass of the attracting body, we can write K
as GM€, where G is a fundamental constant of nature.

THE CONSERVATION LAWS

®® Now let’s return to a modern stance, in which we already know the law of
gravity and want to understand some other aspects of planetary motion.
Specifically, let’s consider the 2 big conservation laws: the conservation of
angular momentum, L , and energy, E.

®® Both L and E remain constant throughout a planet’s elliptical orbit, even

while the planet is moving and changing speed. So, we should be able to
derive expressions for L and E purely in terms of constants— G, M, m, a,
and e.

®® First, let’s do this for angular momentum. In general, L = mrvθ. Remember,

only the angular component of the velocity matters. Because vθ = , we
can also write L as .

®® And previously in this lecture, we consolidated Kepler’s second and third

laws into one equation:

d✓ p
r2 = Ka(1 e2 ) .
dt

59
 Lecture 5 — Newton’s Hardest Problem

®® If we multiply this by m and substitute GM for K, we arrive at a new formula

for angular momentum that is purely in terms of constants:
p
L = m GM a(1 e2 ) .

As an immediate application of this new formula for angular momentum, we

can take care of a piece of unfinished business: Kepler’s third law. In the previous
lecture, we proved it for a circular orbit. Now we can prove it for the general case
of an elliptical orbit.

60
 Lecture 5 — Newton’s Hardest Problem

®® Energy—the other conserved quantity—

has 2 parts: kinetic, 1 ⁄ 2mv 2 , and 1 GM m
E= mv 2
potential, −GMm ⁄ r. 2 r

®® Their sum must be equal to some

combination of the constants G, M, m,
a, and e. Let’s figure out what it is.

®® Because energy is constant, we can calculate it at any point in the planet’s

orbit and get the same answer, so let’s make life simple by choosing θ = 0.
That’s when the planet makes its closest approach to the Sun and r = a(1 − e).

®® We can figure out the velocity with another application of our new angular
momentum formula. In general, L = mrv?.

®® Here, atLθ= = 0,
mrv? is simply v, because at that point, the velocity vector is
totally perpendicular to the radius vector: r is in the x direction and v is in
the y direction. So, at θ = 0, L = ma(1 − e)v.

®® We solve for v and plug in our new expression for L .

p
L m GM a(1 e2 )
v= =
ma(1 e) ma(1 e)

®® Then, we insert that expression

for v into the energy equation
and simplify. The algebra leads
to a cascade of cancellations and
a result that’s refreshingly simple.

61
 Lecture 5 — Newton’s Hardest Problem

®® All the terms related to eccentricity cancelled out—it turns out that energy
depends only on the semimajor axis of the ellipse, not its eccentricity. If
you have a nearly circular orbit with radius 1 AU, like the Earth’s, and you
compare it to a planet on a highly elliptical orbit, with a = 1 AU and e = 0.9,
they’ll both have the same energy.

®® They’ll also have the same orbital period (1 year) because Kepler’s third law
says P depends on a, but not on e. The planet on the elliptical orbit whips
around the Sun near its closest approach, and moves more slowly when it’s
far away, and the 2 effects cancel each other exactly to give the same period
as the Earth. It’s an interesting coincidence.

A GRAPHICAL APPROACH TO
UNDERSTANDING ORBITS
®® Imagine a particle of mass m that is gravitationally attracted to a larger mass,
M. We give our particle some initial position and velocity, which in turn
corresponds to some values of angular momentum, L , and energy, E.

®® As before, the energy is

1 GM m GM m
E= mv 2 .=
2 r 2a

®® And the v 2 has 2 components, radial and angular: vr 2 and vθ 2.

®® Let’s now bring in the conservation of angular momentum.

62
 Lecture 5 — Newton’s Hardest Problem

®® So, we can rewrite the energy equation as

1 L2 GM m
E= mvr2 + .
2 2mr2 r

®® That’s an interesting way to write it, because the second term is purely a
function of r, making it look sort of like the potential energy, even though it’s
really part of the kinetic energy. That’s the basis of a neat trick: We define an
effective potential energy, Ueff , equal to the highlighted term below.
1 L2 GM m
E= mvr2 + .
2 2mr2 r

®® That way, we can write

1
E= mv 2 + Ue↵ (r).
2 r

®® The reason this helps is because now the energy equation only depends on
r and vr, which is the time derivative of r. So, even though we live in a
3-dimensional world, the motion of the particle is governed by a single-
variable equation! That’s what makes it easy to understand graphically.

®® Let’s plot Ueff as a function of r.

For small r, the 1 ⁄ r 2 is dominant,
and it’s positive, so Ueff shoots up
to infinity. For large r, the 1 ⁄ r is
dominant, and it’s negative, so
as r grows, the potential dives
down to negative values and rises
toward 0 as r goes to infinity. It
makes a bowl shape. The exact
shape of the bowl depends on L ,
how much angular momentum
we give the particle.

63
 Lecture 5 — Newton’s Hardest Problem

®® The trajectory of the particle

depends on E , how much total
energy we give it. First, let’s
consider the case in which E is
negative—the negative potential
energy dominates the positive
kinetic energy. We’ll plot E as a
dashed horizontal line.

®® Because the difference between

E and Ueff is equal to 1 ⁄ 2mvr 2,
which is always a positive
number, the particle’s radius (r)
must be confined to the region
where E is bigger than V—that
is, where the dashed line is higher
than the solid line: inside the
bowl of the effective potential.

®® Furthermore, at locations where E − Ueff is large, that means vr is large, too,

so the particle is moving quickly in the radial direction. Wherever Ueff gets
close to E, the particle must be slowing down. When the lines cross, vr is 0
and r is momentarily staying still.

®® All this means that the particle’s radial motion can be understood qualitatively
by imagining we drop a marble in this bowl, starting at one of the intersection
points. The marble starts at rest, rolls to the bottom and speeds up, rolls up
to the same height on the other side, stops briefly, then drops down again,
and keeps oscillating.

®® Likewise, the r value of our particle will grow, then shrink, and then grow
again, as it’s whirling around. That makes sense: We already know that the
particle follows an ellipse, with a distance to the origin that gets bigger and
smaller as it goes around.

64
 Lecture 5 — Newton’s Hardest Problem

®® And if we happen to put the particle right at the lowest point in this
bowl, it will just stay there. That corresponds to a circular orbit, with an
unchanging radius.

®® This graph can teach us other things, too. For example, we’ve just seen that
for a given angular momentum, a circular orbit has the minimum possible
energy—the low point in the bowl. Whenever you drain energy out of an
orbit, with friction or some other process that leaves angular momentum
alone, the orbit will circularize.

®® In addition, we see that it’s impossible for the particle to ever reach r = 0.
That’s because of the first term in the effective potential, L 2 ⁄ 2mr 2, which
makes an infinitely high barrier, guarding the origin. The only exception
would be if L , the angular momentum, is exactly 0. Then, there’s no barrier.

®® In plain language, to make a direct hit on the origin, you need to be dropped
straight in, with no sideways motion. If you have any angular momentum at
all, you’ll orbit the attractor—you won’t hit it.

®® When the total energy is positive, rather than negative, the dashed
line intersects the solid line only once, near the center. So, the particle
approaches the origin and then turns around and flies away, slowing down
but never returning.

®® That’s an unbounded trajectory—like what happened in 2017, when an

interstellar asteroid (which was later named ‘Oumuamua) approached the
Sun at high speed and with a lot of energy. It flew in from parsecs away, and
its radial coordinate, r, shrunk to a quarter of an AU before the Sun deflected
it in a different direction and it flew away. It was the first time anyone had
detected an asteroid from some other star system encountering our own.

®® The plot of the effective potential tells us what’s happening to the radial
coordinate, but it doesn’t tell us what’s happening to its θ coordinate. We
need keep in mind that while r is changing, the particle is also moving in
the perpendicular direction.
65
 Lecture 5 — Newton’s Hardest Problem

®® What’s happening to the θ

coordinate for a bound orbit
with negative total energy? The
conservation of energy tells us that
there will be some minimum and
maximum radius for the particle
that is set by the intersection
points between the energy (the
dashed line) and the effective
potential (the solid curve).

®® And the conservation of angular momentum tells us that the perpendicular

velocity is L ⁄ mr, so it whirls faster when r is small and slower when it’s
farther out. That’s Kepler’s second law—which holds for any central force,
not just gravity.

®® So, you can play this trick—defining an effective 1-dimensional potential—

for any central force law, whether the force goes as 1 ⁄ r 3, or  r, or whatever.

®® In general, the particle whirls around, going from the minimum to the
maximum radius and back again, in accordance with Kepler’s second law.
The trajectory makes a beautiful pattern called a rosette orbit that fills in the
space between the minimum and maximum distance.

®® But for the special case of the inverse square law, there’s a remarkable
coincidence: The trajectory comes around and repeats, making an ellipse.
Just about any other force law—any other power of r —leads to infinitely
looping rosettes instead of a fixed geometric shape.

®® Another exception is if the particle is attached to the origin with an ideal

spring, with force proportional to r ; then, its trajectory is also an ellipse, but
in that case, the origin is the center of the ellipse instead of the focus.

66
 Lecture 5 — Newton’s Hardest Problem

®® Why is it that the actual force law chosen by Mother Nature is one of the
exceptional cases that gives ellipses? It turns out that this coincidence is related
to the fact that for the specific case of the 1 ⁄ r 2 law, there is a third conserved
quantity besides energy and angular momentum. Here’s the equation for this
additional constant of motion.
~
~v ⇥ L
~e = − r̂
GM m

®® The equation takes the planet’s velocity vector, crosses it with the angular
momentum vector, divides by GMm, and then subtracts the r unit vector.
The result is called the eccentricity vector.

®® The time derivative of this quantity is 0. The eccentricity vector is constant

in time, even while v and r are changing throughout the orbit.

®® The magnitude turns out to be the orbital eccentricity, and the direction of the
vector specifies the orientation of the ellipse—it points along the major axis.
Working that out is another way to prove that Newton’s law of gravity implies
Kepler’s first law (as opposed to what we did, which was demonstrate that
Kepler’s laws imply an inward acceleration
going as 1 ⁄ r 2).
Energy is conserved because
®® Whenever there’s a conserved quantity, such the laws of physics don’t
as energy or angular momentum, there’s a change with time. Angular
corresponding symmetry in nature—a sense momentum is conserved
in which nature is mathematically simpler whenever the situation has
than it could have been. This is called rotational symmetry.
Noether’s theorem, after Emmy Noether.

®® It turns out that the equations governing the motion of a particle under
the force of gravity from another particle are mathematically equivalent—
through a complicated change of variables—to the equations for a particle
moving freely, without any force, on the surface of a 4-dimensional sphere.
And it’s the perfect symmetry of that 4-dimensional sphere that leads to the
conservation law for the eccentricity vector.
67
Lecture 5 — Newton’s Hardest Problem

68
Lecture 6

TIDAL FORCES

T his lecture will address 3 questions. First, the major

rings around planets are all in the range of about
2 to 2.5 times the radius of the planet. Why is that?
Second, all of the giant planets have moons. Why are
the large moons spherical while the smaller ones can
have irregular shapes? Third, all the large, spherical
moons are far from the planet—well outside the rings—
whereas the small moons are found all over, including
in and among the rings. Why are the large moons only
found in wide orbits? To answer these questions, our
discussion of gravitational orbits needs to be expanded.

69
 Lecture 6 — Tidal Forces

UNDERSTANDING RINGS AND MOONS

®® So far, we have assumed that the orbiting body is a point mass—an
infinitesimally small mathematical point. We’ve ignored the fact that a planet,
or a moon, is a real object with a nonzero size.

®® This is important because the force of gravity depends on distance; it gets

weaker as you get farther away from the attracting mass. For example, the
side of the Earth facing the Sun—the dayside—is pulled harder than the
nightside. The differences in gravitational forces from one part of a body to
another are called tidal forces.

®® Suppose we have a planet of mass M that has a moon with a small mass, m,
whirling around in a circular orbit of radius r.

®® Actually, let’s start with an even simpler case. If we drop the moon, starting
from rest, it will accelerate downward and crash into the planet. It doesn’t
orbit because we didn’t give it any angular momentum.

®® Now let’s give the moon a nonzero size. We could make

it a sphere, but the math would get too hairy. To keep
things simple, let’s just take one step beyond the point-
mass approximation. We’ll model the moon not as 1 point
mass, but as 2 point masses—2 rigid spherical “rocks,”
each of mass m, with their centers separated by some small
distance ∆r.

®® When we let go of the rocks, they both fall onto the planet.
They do not stay together as they fall. The inner rock is
closer to the planet, so it feels a stronger gravitational
force than the outer one, leading to a larger acceleration
and causing it to pull away from the outer rock. Our
“moon” breaks apart as it falls.

70
 Lecture 6 — Tidal Forces

®® In general, the magnitude of the force from the planet is

GM m
F (r) = .
r2

®® Let’s calculate the difference in force, ∆ F, experienced by the rocks. Because

∆r is very small compared to r,
dF 2GM m
F = Fout Fin ⇡ r= r.
dr r3

®® The minus sign means that the force weakens with distance.

®® We’ve learned that the part of the “moon” closest to the planet is pulled harder
by an amount proportional to ∆r, the size of the moon, and M, the mass of the
planet and inversely proportional to the cube of the moon’s orbital distance.
Those are the hallmarks of tidal forces: They grow with the size of the body—
the mass of the attractor—and fall off as the third power of distance.

®® If we want to keep our “moon” intact, we need to supply a force to counteract

∆ F: the gravitational force between the rocks, which are attracted to each
other with a “self” gravitational force of
Gm2
Fself = .
( r)2

®® If we want them to stay together as they fall, the magnitude of ∆ F must be

smaller than Fself. That leads to an inequality that can be simplified as follows.
| F | < Fself
2GM m Gm2
3
r<
r ( r)2
2M m
<
r3 ( r)3

®® If m is big enough and ∆r is small enough—that is, if the rocks are massive
and closely packed—we can satisfy this inequality and they hold together
as they fall.
71
 Lecture 6 — Tidal Forces

®® But there’s a 1 ⁄ r 3 on the left side. As time goes on and r gets smaller, the left
side grows rapidly and, at some point, overwhelms the right side. Let’s call
that minimum orbital distance rmin, which we can solve for by setting the 2
sides equal to each other. Inside of rmin, the moon will break apart.
2M m
3 =
rmin ( r)3
✓ ◆1/3 ✓ ◆1/3
2M M
rmin = r ⇡ 1.26 r (two rocks)
m m

®® Our model of a moon as 2 point masses is not very realistic. You can do
a similar calculation for a model in which the moon is a big blob of fluid
that can deform and flow in response to tidal forces. That takes more
mathematical horsepower, but the result is similar. In fact, it’s the same as
our equation, but with ∆r representing the average radius of the moon, and
the 1.26 is replaced by 2.44.

✓ ✓◆1/3◆1/3
2M M
rmin ⇡
= 2.44 r r (fluid body)
m m

®® It’s more traditional to write the equation in terms of the densities and sizes
of the bodies, rather than masses. We can replace the planet mass, M, with
volume times density, and we can do the same for the moon, m. With those
substitutions and a touch of algebra, the minimum radius comes out to be
2.44 times the planet radius, times the cube root of the density ratio.

72
 Lecture 6 — Tidal Forces

®® This was first worked out by Édouard Roche, so the minimum distance is
known as the Roche limit: the distance within which tidal forces overcome
the gravitational binding force of a moon modeled as an idealized fluid body.

®® What if the moon is orbiting, instead of just falling in? Is there still a
Roche limit? Yes. It’s the same, and in that case, rmin refers to the minimum
orbital distance.

®® Suppose a moon made from a bundle of rocks is going in a circular orbit.

The rocks closer to the star have slightly smaller orbital distances, so, by
Kepler’s third law, they have shorter orbital periods. The outer rocks have
longer periods.

®® Therefore, unless the moon’s self-gravity is strong enough, the rocks drift
apart over time, with the inner ones moving ahead of the outer ones. The
moon gets shorn into pieces and strung out into an arc around the star.
Eventually, the arc reaches all the way around the planet, making a ring. This
may be where planetary rings come from!

®® The number 2.44 is a pretty good match to the observed sizes of the rings
of the giant planets, which range up to 2 or 2.5 times the radius of the
planet. The density ratio is always of order unity, because the densities of
the moons and planets are of the same order of magnitude, a few grams per
cubic centimeter. So, the numbers fit the story.

All the planets have a Roche limit, including Earth. The Moon’s mean density is
around 3 grams per cubic centimeter, typical of rocks. The Earth’s is higher, around
5.5, because the Earth’s stronger gravity compresses its interior and because the
Earth has more iron in its core. Given those numbers, the Roche limit comes out
to about 3 Earth radii. Our Moon is at a distance of 60 Earth radii, so it’s not in any
danger of tidal destruction.

73
 Lecture 6 — Tidal Forces

®® We derived the Roche limit by setting the tidal force, which tries to pull
the body apart, equal to the attractive gravitational force trying to hold it
together. But there are other ways for a body to hold together besides gravity.
There are also chemical or material forces that give rocks their rigidity. The
silicon atoms in a rock aren’t held in place by gravity; they’re stuck together
with chemical bonds, which are ultimately electromagnetic forces at the
atomic level.

®® The Roche limit is only relevant for objects that are held together mainly by
gravity. And we shouldn’t take the factor of 2.44 too seriously; that’s for the
ideal case of a frictionless fluid. Material forces allow a body to come closer
than this official limit.

®® Why can we find little moons,

smaller than about 500 kilometers,
What happens when a body violates the
nestled right within the rings that
Roche limit?
are inside the Roche limit? It
In 1994, a comet named Shoemaker- must be because those objects are
Levy-9 crashed into Jupiter. By the time held together mainly by material
it hit Jupiter, tidal forces had broken it forces, not gravity. In fact, you
into fragments, each of which punctured can tell if an astronomical object
is gravitationally bound by just
Jupiter’s clouds in a different place,
looking at it. If gravity is the
making a series of brown spots.
dominant force, it’ll be a sphere.
The comet broke apart because it was a That’s because each piece is
loose conglomeration of rocks and chunks attracted to every other piece. Left
of ice; there wasn’t much holding the to its own devices, gravity draws
chunks together besides gravity. everything inward toward the
center of mass, smoothing out any
lumps to make a perfect sphere.

®® In contrast, chemical and material forces are very local; they act only between
neighboring molecules or surfaces in direct contact. So, a body held together
by those forces can be any shape—an egg, a potato, a person—depending on
the history of how the pieces came together.
74
 Lecture 6 — Tidal Forces

®® But if we make the object bigger and bigger, gravity becomes more important
and eventually dominates over chemical and material forces.

®® Let’s do an order-of-magnitude calculation to estimate how big a body has

to be for gravity to mold the shape into a sphere.

®® Suppose we have a rock with a characteristic size of R and a mass of M. Now

let’s make it slightly larger by adding a single silicon atom. Which is more
important: the molecular forces that bind the silicon to the minerals on the
surface or the gravitational attraction of the atom to the entire mass of rock?

®® It’ll be easiest to compare the relevant amounts of energy. The energy levels
of electrons in atoms are always on the of order of a few electron volts. The
energy scale for material forces tends to be an order of magnitude lower,
because rocks aren’t perfect crystals—they’re ragged collections of crystals,
and the interactions between them are weaker. So, let’s say the energy released
is, on average, 0.1 eV per silicon atom.

®® Meanwhile, the gravitational potential energy that’s released when we add

mass m is of order GMm ⁄ R. For gravity to be more important, we need that
to be much larger than 0.1 eV. That gives us a condition on the rock’s mass
over radius, M ⁄ R. We’re wondering about the critical size, so let’s write M as
volume times density and solve for R.

75
 Lecture 6 — Tidal Forces

®® The density of rock is around 3 grams per cubic centimeter, or 3000 kilograms
per cubic meter, and the mass of a silicon atom is about 28 proton masses, or
4.7 × 10 −26 kilograms. Plugging in those numbers along with the constants
leads to a critical radius of about 600 kilometers.

®® Based on this calculation, we would expect objects much larger than that
to be sculpted into spheres by gravity, while much smaller objects can have
irregular shapes. And this is what we observe among the moons of Saturn.

Why are the major rings of the giant planets all within about 2.5 planetary radii?
Because that’s the approximate location of the Roche limit.
Why are the large moons spherical? Because they’re big enough for gravity to
dominate over material forces.
Why do the large moons orbit well outside of the rings? Because otherwise tidal
forces would break them into smaller pieces.

OCEAN TIDES
®® In addition to helping us understand rings and moons, tides are also relevant
to planets, stars, black holes, and entire galaxies. A more down-to-earth
example of tidal forces is ocean tides.

®® The Earth’s gravity pulls on the Moon, and the Moon’s gravity pulls on the
Earth. That means the Moon exerts tidal forces that, left unopposed, would
tear the Earth apart by squeezing it along the direction to the Moon. The
Earth’s gravity prevents that from happening. But there’s more to it than that.

76
 Lecture 6 — Tidal Forces

®® The vectors in this image represent the gravitational

force of the Moon at different points in space surrounding
the Earth. Close to the Moon, the vectors are longer
because the force is stronger. There’s also
some variation in direction, because all
the vectors point straight at the center of
the Moon.

®® But importantly, the Moon is orbiting the

Earth. To see the Earth and Moon sitting
still, as they are in this image, we must
be in a frame of reference that’s rotating
along with the orbit, once a month.

®® We’re allowed to do physics in rotating frames, but the price we pay is that we
must insert a fictitious force: the centrifugal force. In this case, the centrifugal
force on the Earth points away from the Moon, with a strength such that
at the center of the Earth, the centrifugal force cancels out the gravitational
force exactly. That’s why the Earth is sitting still in this frame of reference.

®® Let’s add the centrifugal and gravitational forces and

replot the net force vectors. There’s no net force at the
center of the Earth. The net force points toward the
Moon on the near side and away from the Moon on the
far side, where the centrifugal force is larger than
the gravitational attraction. And in between
there are sideways forces.

®® Imagine that the Earth is a frictionless

sphere surrounded by a thin layer of F gravity
water. What would happen to the water?
It would feel these net forces and flow F centrifugal
around the surface to form 2 bulges, one
on the near side and one on the far side.

77
 Lecture 6 — Tidal Forces

®® Then, if the frictionless Earth were to rotate, sliding underneath the layer
of water, an observer on the surface would see the ocean rise in height, then
fall, rise, and fall again over the course of a full day. That’s why we observe
2 high tides and 2 low tides in 1 day.

®® The Earth is not a frictionless sphere. There’s lots of friction, and there are
continents, underwater mountains, and all kinds of things that make the
picture more complicated. That’s why we need tide tables.

®® The tidal forces from the Sun are also significant.

That’s why the maximum height of the tide varies
with the phase of the Moon. When the Sun, Moon,
and Earth are along a line, the Sun and Moon work
together and produce unusually high tides called
spring tides. When the Sun and Moon are at right
angles, the contrast between high and low tides is
reduced, and they are called neap tides.

®® From the relative heights of the spring and neap tides,

we can determine that the Sun’s tidal forces are not
quite as strong as the Moon’s—they’re only about
half as strong.

®® Tidal forces are proportional to the mass of the

attracting body over the cube of the distance, so
the ratio of solar to lunar tidal forces is equal to the
Sun’s mass divided by the Moon’s mass, times the
cube of the distance to the Moon over the distance
to the Sun.
✓ ◆3
Fsun Msun dmoon
=
Fmoon Mmoon dsun

78
 Lecture 6 — Tidal Forces

®® Suppose we don’t already know the masses and orbital distances. Can we learn
something interesting about the Sun, or Moon, by observing spring and neap
tides? Yes, we can—if we also know about total solar eclipses. The stunning
thing about total eclipses is the Moon blots out the Sun almost exactly, rim
to rim. They have the same angular radius in the sky.

®® Because the angular radius, ∆θ, is equal to the true radius divided by distance,
the observation of total eclipses tells us that
Rsun Rmoon
✓= = .
dsun dmoon

®® Or, equivalently, the ratio of distances is equal to the ratio of radii. So, in
our tidal force equation, we can replace the cube of the distance ratio by the
cube of the radius ratio.
✓ ◆3
Fsun Msun Rmoon
=
Fmoon Mmoon Rsun

®® This is the ratio of the average densities of the Sun and Moon!

®® The fact that the Sun’s tidal force is about half the Moon’s tells us that the
Sun’s average density is half that of the Moon. The Moon looks like a rock, so
its density is about 3 grams per cubic centimeter, from which we can deduce
that the Sun’s average density is around 1.5 grams per cubic centimeter.

79
QUIZ
LECTURES 1–6

1 In our neighborhood of the Milky Way, the typical spacing between stars is 1 pc,
and the relative velocities between neighboring stars are of order 20 km ⁄ sec.
Imagine a scaled-down model in which stars are replaced by grains of sand
(1 mm). What is the typical spacing between the grains of sand, and how fast
are they moving? [LECTURE 1]

2 Your weight on a planet is the product of your mass and

CLICK to navigate.
the planet’s surface gravitational acceleration, GM ⁄ R 2.
Suppose you visit different planets that all have the
same density. Write a proportionality to express how
your weight (W ) depends on planet mass (M ). [LECTURE 1]

To go back to the page you came from, PRESS Alt + ← on a PC or ⌘ + ← on a Mac.

On a tablet, use the bookmarks panel.

3 Make a list that compares the 4 fundamental forces of nature. Think of at least
one way in which each force has a direct impact on everyday life. [LECTURE 2]

4 Suppose you start out with the mass of an electron and double your mass every
day. How many days would elapse before you have the mass of the entire Milky
Way Galaxy (approximately 1 trillion solar masses)? [LECTURE 2]

5 Try to measure the angular resolution of your eye in arc seconds. One way is
to prick 2 holes in a piece of cardboard a few millimeters apart and illuminate
them from the back with a flashlight. Then, in a large, dark room, see how far
away you can still tell that there are 2 holes instead of one. [LECTURE 3]

80
 Quiz FOR Lectures 1– 6

6 The flux we receive from a star is proportional to 1 ⁄ r 2. But the flux we receive
from a distant asteroid is approximately proportional to 1 ⁄ r 4. Why? What is
the key difference? [LECTURE 3]

7 Think about the problem of establishing Kepler’s laws of planetary motion based
on naked-eye observations of the night sky. What would be some of the major
difficulties? [LECTURE 4]

8 Neptune’s moon Triton has an orbital period of 5.877 days and a semimajor
axis of 354,759 km. Calculate the mass of Neptune in units of Earth masses.
[LECTURE 4]

9 If a spacecraft in a circular orbit fires its rocket, boosting the speed of the
spacecraft in the forward direction, how does the total orbital energy change?
How does the angular momentum change? What are the resulting changes to
the shape of the orbit? [LECTURE 5]

10 Halley’s comet has an orbital period of 75.32 years and an orbital eccentricity of
0.96714. How close does Halley’s comet come to the Sun in AU? [LECTURE 5]

11 Some have argued that the definition of a planet should include the requirement
that the shape is a sphere. What are the merits and faults of this definition?
Think about which objects in the solar system would be classified as planets
and how this definition could apply to planets around other stars. [LECTURE 6]

12 Express the Roche limit as a minimum orbital period, rather than a minimum
orbital distance. You should find that the minimum orbital period is independent
of the properties of the central body. [LECTURE 6]

Go to page 336 for solutions.

CLICK to navigate.

81
Lecture 7

BLACK HOLES

I n the popular imagination, black holes are like

vacuum cleaners, sucking up everything around
them. That’s not quite true—or at least it’s no truer than
it is for the Sun. The Sun doesn’t suck in all the planets.
To fall into the Sun, or a black hole, you need to have
nearly 0 angular momentum—hardly any transverse
velocity—so that you follow a radial trajectory all the
way down to r = 0. Black holes do swallow things, but
that only happens when there’s some way for orbiting
material to get rid of its angular momentum.

82
 Lecture 7 — Black Holes

The Gravitational Field of a Spherical Object

Newton’s law of gravity, F = GMm ⁄r 2, gives the magnitude of the force
between 2 masses. Specifically, it’s the force between 2 point masses—idealized
mathematical points with nonzero mass but zero size.
When we allow the body feeling the force to have a nonzero size, it feels tidal
forces, differential gravitational forces that tend to pull the body apart. But what
happens when we allow the body producing the force—such as the Earth—to
have a nonzero size?
The Earth isn’t a point mass; it’s a sphere. When we’re standing on the surface,
our bodies are gravitationally attracted to every cubic centimeter of material
in the entire Earth, pulling all at once. And the rock just below our feet is much
closer to us, and pulls harder, than the same amount of rock on the other side of
the Earth.
So, you might think it would be a mess to calculate the net force from the entire
Earth; you’d have to add the forces exerted by each cubic centimeter, taking into
account the huge variations in distance and direction from place to place.
But Isaac Newton taught us that when we’re outside a spherically symmetric
object, we can calculate the gravitational force by pretending it’s a point mass
endowed with the total mass of the real object.
What happens if we’re inside an object with a spherically symmetric mass
distribution? Let’s say we dig a tunnel in the Earth down to a radius r from the
center. It turns out the net gravitational force at the bottom of the tunnel can be
calculated by pretending that all the mass down deeper, with radius less than r,
is replaced by a point mass at the center and by ignoring all the exterior mass!
This is because although the nearby atoms are pulling harder, there are more
atoms farther away pulling in the other direction, and the effects cancel
each other.

83
 Lecture 7 — Black Holes

Therefore, if we were inside the Earth, we’d feel a gravitational force, which we can
calculate by pretending all the interior mass is concentrated at the center. If the
Earth had a constant density, ρ, the total interior mass would be

The gravitational force is proportional to the interior mass divided by r 2, so it

goes as r 3⁄r 2—it’s proportional to r. As we dig further down, the force declines in
strength, and at the center of the Earth, there’s no gravitational force at all.

THE MASS AND RADIUS

OF A BLACK HOLE
®® A black hole has no surface; it’s a real-life point mass. Its mass is concentrated
into a single point in space, so as you approach, the gravitational force gets
stronger, without bound—all the way down to r = 0, where it becomes infinite.

®® A black hole is an infinitely deep gravitational pit. And the pit has slippery
sides; if you get too close, you inevitably fall in.

®® Let’s say we’re cruising around the galaxy and find ourselves a distance of
r 0 from a black hole. In a panic, we turn away from the hole and fire our
thrusters, burning all the fuel in a desperate burst and giving the spaceship
a velocity of v 0 directed away from the hole. Will we escape?

84
 Lecture 7 — Black Holes

®® We can figure this out based on the conservation of energy. The initial energy
of the spaceship is the kinetic energy plus the gravitational potential energy,
where M is the mass of the black hole and m is the mass of the ship.
1 GM m
E= mv 2
2 0 r0

®® At some later time, we make it out to some larger distance, r1, with a slower
speed, v1, because the black hole’s gravity has been slowing us down. So, the
energy is
1 GM m
= mv 2
2 1 r1

which we can set equal to the initial energy.

1 GM m
E= mv 2
2 0 r0

1 GM m
= mv 2 →0
2 1 r1

®® In the case where we just barely escape, r 1 approaches infinity and the
potential energy approaches 0. In addition, v1 approaches 0, because we had
just enough energy to make it out, with no leftover kinetic energy. So, the
total energy in this case must be 0, giving us a simple equation, which we
can solve for v 0, giving

1 GM m
mv 2 =0
2 0 r0
1 GM m
mv 2 =
2 0 r0
r
2GM
v0 =
r0

®® That’s the formula for the escape velocity. If we have enough fuel to go at
least that fast, we can escape. Otherwise, we fall in.

85
 Lecture 7 — Black Holes

®® But here’s the thing about a black hole: Because there’s no surface, there’s
no minimum value for r 0. There’s nothing equivalent to the Earth’s surface,
within which the gravitational force starts getting weaker. So, the escape
velocity grows without bound as r 0 approaches 0.

®® At some point, the escape velocity exceeds the speed of light—the fastest
speed it’s possible for anything to attain—and in that case, no amount of
fuel would be enough. To find the value of r 0 where that happens, we set the
escape velocity equal to c, the speed of light, and solve for r 0.
r
2GM 2GM
v0 = = c −! r0 = = RS
r0 c2

®® So, even though the black hole itself has 0 size, 2GM ⁄ c 2 is sort of a radius—
the radius of no return. It’s called the event horizon or the Schwarzschild
radius (R S), after Karl Schwarzschild, the first person to solve Einstein’s
equations of general relativity exactly for the case of a point mass.

®® Let’s find the Schwarzschild radius of a black hole with the mass of the Earth
(MÅ ). In this case, MÅ is 6 × 10 24 kilograms, giving a Schwarzschild radius
of 9 millimeters. What that means is that if we could somehow compress
the entire Earth down to the size of a marble, we would have a black hole.

®® For the mass of the Sun (M€ ), the Schwarzschild radius comes out to be 3
kilometers, so we can write the formula as a scaling relation.
✓ ◆
M
RS = 3 km
M

86
 Lecture 7 — Black Holes

Imagine that the Sun is somehow compressed to a radius of 3 kilometers, turning

it into a black hole. What would happen to the solar system? How long would we
have before Earth fell into the black hole?
Actually, the Earth wouldn’t fall in. And the planets’ orbits wouldn’t change. It’s
a corollary of Newton’s theorem about spherically symmetric mass distributions,
which turns out to also hold true in relativity. The force from the Sun is the same
as the force from a black hole of equivalent mass.

EVIDENCE FOR BLACK HOLES

®® Since the mid-1990s, a group at UCLA led by Andrea Ghez has been watching
a cluster of bright stars near the constellation Sagittarius in the center of our
galaxy. These stars are moving fast, and not just in random directions. They’re
being deflected by the gravity of a giant mass at the center. Some stars are
orbiting the center in Keplerian ellipses.

®® One star in particular, named S0-2, made a complete orbit over 16 years.
But in optical and infrared images, there’s hardly any light coming from the
focus of the ellipse. S0-2 and the neighboring stars are being attracted to an
unremarkable spot in the center.

®® Let’s figure out how massive the attractor must be based on observations of
S0-2. The images of S0-2 show that the angular size, ∆θ, of the long axis
of the ellipse is about a quarter of an arc second. And we know from other
observations that the distance, d, to the galactic center is 8 kiloparsecs. With
that information, we can calculate the semimajor axis, a.

2a
✓=
d
✓ ◆
1 1 1
a= ✓d = arcsec (8,000 pc) = 1000 AU
2 2 4

87
 Lecture 7 — Black Holes

®® We also observe that S0-2 takes 16 years to go around, so P = 16 years. And
whenever we know both P and a, we can apply Kepler’s third law to find the
mass. First, let’s rearrange Kepler’s third law to solve for the mass.
2⇡ 3/2 2 3
P =p a −! M = 4⇡ a
GM G P2

®® At this stage, we could plug in numbers, looking up the numerical value of

G, converting a to meters and P to seconds, and so forth, but let’s make it
easier by converting the equation into a scaling relation.

®® Because M is proportional to a 3 ⁄ P 2 and we know the answer is 1 solar mass

for the Earth, we can write
⇣ a ⌘3 ✓ P ◆−2
M = (1 M ) .
1 AU 1 year

®® This is convenient for our problem because we already know a and P in those
units. The mass of the mysterious attractor, in units of solar masses, is

®® In other words, lurking at that nondescript spot in the center—named

Sagittarius A*—is something with a mass of 4 million Suns. And it’s all
crammed into a space that must be smaller than S0-2’s distance of closest
approach, about 100 AU, which is not much bigger than the solar system.

®® Astrophysicists have not been able to think of anything that could have so
much mass within such a small space without either glowing very brightly
or quickly collapsing under its own gravity. So, it’s either a black hole or
something even more exotic, beyond our current physical understanding.

88
 Lecture 7 — Black Holes

®® How close are these stars getting to the Schwarzschild radius? Are they
in danger of falling in? The Schwarzschild radius is about 12 million
kilometers—which is only about 0.1 AU. The orbit of S0-2 has a minimum
distance of 100 AU, so it’s safe.
✓ ◆
M
RS = 3 km ⇡ 12 ⇥ 106 km ⇡ 0.1 AU
M

OPTICAL AND RADIO TELESCOPES

®® Every once in a while, a star or cloud of gas falls into a black hole. The
evidence for tidal disruption events takes the form of bright outbursts from
the centers of distant galaxies, but those galaxies are all so far away that our
telescopes can’t resolve the details, so there’s no hope of making images that
show the star falling in.

®® In our own Milky Way, could we watch a star getting torn to shreds? Let’s
figure out how big a telescope we would need.

®® The fundamental limit on the angular resolution of an image, ∆θmin, is the

diffraction limit, which is of order λ ⁄ D, where λ is the wavelength of light
and D is the diameter of the telescope.

λ
✓min ⇠
D

®® How big does D need to be to resolve details on the scale of the Schwarzschild
radius, 0.1 AU? We can find out by setting λ ⁄ D equal to the angular size
of the Schwarzschild radius. That’s 0.1 AU divided by 8000 parsecs, the
distance to the galactic center. For λ, we insert half a micron, which is typical
of visible light. After the necessary unit conversions, the diameter comes out
to be 8000 meters.

89
 Lecture 7 — Black Holes

®® The world’s largest optical telescopes are 10 meters across; in this case, we
would need one that is 8 kilometers across. That is crazy. Nevertheless, there’s
an effort underway to make images sharp enough to resolve the Schwarzschild
radius of Sagittarius A*. The trick is to use a radio telescope instead of an
optical telescope. The project is called the Event Horizon Telescope.

®® At radio wavelengths, it’s possible to combine the information from widely

separated telescopes to mimic the performance, in terms of angular resolution,
of a much larger telescope. So, if we connect telescopes in different parts of
the world, we can get the necessary angular resolution—at least in principle.
This technique, called very long baseline interferometry, has become routine
for observations at wavelengths of meters and centimeters, but getting it to
work at a millimeter is still a challenge.

ORBITING A BLACK HOLE

®® If you get too close to a black hole, within the Schwarzschild radius, you fall
in, and if you’re far away, the orbits are the same as they would be around
a star, or planet, or any “normal” astronomical body. But what happens in
between those extremes—when you’re orbiting near a black hole but not right
at the Schwarzschild radius?

®® To answer this question, we’ll use the effective potential energy, a concept
we used previously to analyze planetary motion. In Newtonian gravity, the
equation for total energy is as follows, where Ueff is the sum of the real
potential energy and a term that depends on angular momentum but can be
written in a form that makes it resemble a type of potential energy.

1
E= mv 2 + Ue↵ (r)
2 r
GM m L2
Ue↵ (r) = +
r 2mr2

90
 Lecture 7 — Black Holes

®® This gives us a graphical way to understand how the radial coordinate changes
with time as a body whirls around an attractor.

®® In general relativity, there is an extra term in the effective potential energy.

GM m L2 L 2 RS
Ue↵ (r) = +
r 2mr2 2mr3

®® When we replot Ueff for a black hole, we need to add the contributions of the
−1 ⁄ r term, the 1 ⁄ r 2 term, and the −1 ⁄ r 3 term. As r gets smaller, the new 1 ⁄ r 3
term grows the fastest of all, and it’s negative, so the effective potential dives
downward at a small radius.

®® To make the numbers come out nice, the vertical axis shows Ueff ⁄ mc 2 and
the horizontal axis shows the radius in units of the Schwarzschild radius.
For concreteness, a value of angular momentum equal to 2.1mc times the
Schwarzschild radius was used.

91
 Lecture 7 — Black Holes

®® Below 1 Schwarzschild radius, there’s a pit in the potential. We can visualize

what’s going to happen to the radial coordinate r by imagining we drop
a marble onto this curve starting from an initial height E , given by the
total energy.

®® If the marble is far away from the

Schwarzschild radius, then it rolls back
and forth. That means r oscillates back
and forth, as was the case in Newtonian
gravity; the particle follows an orbit with
some minimum and maximum distance.

®® But if we let go of the marble inside the

Schwarzschild radius, it falls into the pit,
regardless of energy.

®® And even if we start the marble farther

out in the bowl, if the energy is high
enough, it will go over the inner wall of
the bowl and fall into the pit.

®® This means that in general relativity,

you can hit the origin even if you do
have some nonzero angular momentum.
There’s no more barrier here, sealing off
the singularity, as there was in Newtonian
gravity. It’s as though when you get close
enough, the hole reaches out and grabs
you. It absorbs both your mass and your
angular momentum. In other words,
black holes—rotating point masses—
can rotate.

®® This curve was drawn for a specific value of angular momentum. Different
values lead to different shapes for the bowl.
92
 Lecture 7 — Black Holes

®® As we lower the angular momentum, the bowl gets pounded down on the left
side until it’s not even a bowl any more. There’s no more minimum, which
means there’s no place where the particle can sit still at a constant radius. A
circular orbit is impossible.

®® This is a famous result in general relativity. It’s called the innermost stable
circular orbit (ISCO). For a nonrotating black hole, the ISCO turns out to
be 3 times the Schwarzschild radius. For a rotating black hole, the ISCO can
be larger or smaller, depending on which way it’s rotating.

®® Let’s return to a case with more angular momentum, in which the effective
potential energy curve still looks like a bowl. If you give the particle a modest
amount of energy, the radial coordinate oscillates back and forth, just as it
does for a planet going around the Sun.

93
 Lecture 7 — Black Holes

®® For a planet, the trajectory is an elliptical orbit with a minimum distance of

a(1 − e) and a maximum distance of a(1 + e)—but we only get ellipses for the
very specific case of a force that goes as 1 ⁄ r 2. And in general relativity, the
force law doesn’t go exactly as 1 ⁄ r 2, due to that extra term in the potential.
So, we don’t get perfect ellipses.

®® Close to the black hole, the orbits are not even approximately ellipses. They’re
rosettes—whipping around fast as they approach the black hole, then slowing
down as they recede, and coming back again from a different angle.

~
~v ⇥ L
~e = − r̂
GM m

®® If you’re relatively far away, the orbits are very nearly ellipses—but not quite.
The orientation of the ellipse gradually wheels around in space. It’s an effect
called apsidal precession.

®® In general relativity, the eccentricity vector—that combination of velocity,

momentum, and position that turns out to be conserved in classical gravity—
is no longer conserved. It slowly changes with time. It rotates around the
direction of the angular momentum vector.

®® So, if you looked carefully enough at planets orbiting a black hole, you would
see that they’re not quite obeying Kepler’s first law. And it’s not just black
holes. Once you’re outside a spherical mass distribution, the gravitational
effects are the same, no matter whether you’re orbiting a planet, star, or
black hole.

®® That means Kepler’s first law must be at least a little bit wrong in the solar
system, too. The planetary orbits should be precessing. And they are. The
effect is strongest for Mercury, the planet with the smallest orbital distance.

94
Lecture 7 — Black Holes

95
Lecture 8

PHOTONS AND
PARTICLES

W hen it comes to the task of solving the

mysteries of the cosmos, the only clues that
astrophysicists get—with few exceptions—are from the
trickle of radiation that arrives at the Earth from distant
sources. In this lecture, you will become familiar with
electromagnetic radiation.

ELECTROMAGNETIC WAVES
®® In classical physics—that is, not quantum physics—electromagnetic radiation
takes the form of waves. Waves result whenever some physical quantity varies
smoothly throughout space and time—such as the height of the surface of a
pond or the pressure of the air in a room—that, when disturbed, produces
an oscillating pattern.

96
 Lecture 8 — Photons and Particles

®® No matter what kind of wave it is, the wavelength, λ, can be defined as the
distance between maximum values of whatever is waving—for example,
between the crests of a water wave. And the frequency, ν, can
be defined as the rate at which the pattern oscillates, or how
many times per second the height of the water bobs up
and down, completing a cycle. When wavelength and λ
frequency are multiplied, the result is the wave’s
phase velocity—the speed with which the pattern
moves—which, for electromagnetic radiation, is
the speed of light, c.

®® But with light, it’s less obvious what’s waving.

It’s not the air, or anything material; it’s
the more abstract oscillations of electric
and magnetic fields. The electric field is a
vector whose magnitude wobbles back
and forth, while the magnetic field
does the same thing, but tilted by
90°—and the whole pattern moves
in the direction perpendicular to λ⌫ = c
both vectors.

®® The wavelengths of visible light range from 0.4 to 0.7 microns. The
corresponding frequency, ν, is just under 10 15 cycles per second, or Hertz.
But that’s just a tiny slice of the whole spectrum. Toward longer wavelengths,
there’s infrared radiation, and when the wavelength exceeds 1 millimeter,
they are called microwaves, and then radio waves. Likewise, on the short-
wavelength end is ultraviolet radiation, and when the wavelength is shorter
than 10 nanometers, they are called x-rays, and then gamma rays (see image
on the following page).

®® All of these names and boundaries are arbitrary; they’re based on the different
technologies humans use to study electromagnetic radiation. Mother Nature
makes no sharp distinctions.

97
 Lecture 8 — Photons and Particles

Wavelength (m)

Frequency (Hz)

Energy (eV)

®® Waves are produced by a disturbance, such as dropping a stone in water

or clapping your hands. For electromagnetic waves, the disturbance is the
acceleration of an electrically charged particle.

®® When a charge accelerates, it radiates. And it goes backward, too: Charges

can absorb electromagnetic energy, causing them to accelerate.

®® These facts come from solving James Clerk Maxwell’s equations of

electromagnetism.

®® The pattern and energy of the waves depend on the details of the acceleration.
An electron slowing down in a block of lead produces a certain pattern while
an electron whirling in a magnetic field produces a different one. And an
electron falling from one orbit to another inside an atom produces yet another
kind of radiation.

®® But if the charges are moving randomly—if there are zillions of charges
rattling around, colliding with each other, and they’ve been at it long enough
to reach a steady state—there’s an enormous simplification. The radiation
takes on a universal character, depending only on the temperature.
98
 Lecture 8 — Photons and Particles

INDIVIDUAL PROPERTIES OF
PARTICLES AND PHOTONS
®® Quantum theory teaches us that if we look closely enough at electromagnetic
energy, we’ll see it comes in tiny lumps, called photons. In principle, we could
count the photons arriving from the Sun—like counting raindrops falling
from the sky or grains of sand in an hourglass—but photons have some weird
properties that normal particles do not have.

®® Think of a box full of tiny particles whizzing around and knocking into
each other. This is an idealized model of a gas. Each particle has a mass, m,
and a speed, v —from which the particle’s kinetic energy and momentum
can be computed. (The symbol ε is used for the energy of a single particle to
distinguish it from the energy of the entire gas, E.)

PARTICLES PHOTONS

MASS m zero
SPEED v c
1 1 hc
ENERGY ✏= mv 2 = m(v
✏= 2
x+h⌫vy2=+ vz2 )
2 2 λ
✏ h⌫
MOMENTUM p = mv p= =
c c

®® Photons have 0 mass. In addition, they always travel in empty space at the
same speed, c, which is equal to 3 × 10 8 meters per second. And even though
they have 0 mass, photons have energy and momentum. (The constant of
proportionality, h, is Planck’s constant, which is equal to 6.6 × 10 −34 joule-
seconds.)

99
 Lecture 8 — Photons and Particles

COLLECTIVE PROPERTIES
OF PARTICLES
®® Let’s say the box of particles has a volume V and is filled with a certain
number of particles, N. The number density, n, is defined as the number per
unit volume.

®® Into the box a total energy of E is injected. NUMBER DENSITY

That’s the sum of all the kinetic energies of all N
n=
the particles—which is constant in time, because V
the box is sealed tight. If the particles are allowed
to knock around for a long time, their positions
and velocities become randomized; a particle could turn up anywhere in the
box with equal probability. And they’ll come to share the total energy more
or less equally.

®® That’s the conceptual basis of temperature. The temperature, T, is defined

to be proportional to the average energy per particle. For an ideal gas, the
proportionality constant turns out to be 3 ⁄ 2k, where k is Boltzmann’s
constant, which is equal to 1.4 × 10 −23 joules per Kelvin.

®® The general rule is that the average energy is

1 ⁄ 2kT times the number of independent ways a AVERAGE ENERGY
particle can store or exhibit energy—the number 3
of degrees of freedom. Our particles can move in h✏i = kT
2
3 dimensions, so the kinetic energy has 3 terms,
each of which counts as a degree of freedom.

1 1  
✏= mv 2 = m vx2 + vy2 + vz2
2 2

100
 Lecture 8 — Photons and Particles

®® So, the average energy per particle is 3 ⁄ 2kT.

ENERGY DENSITY
®® This means that the energy density, u, which is 3
u= nkT
the total energy per unit volume, is equal to 3 ⁄ 2 2
times the number density, n, times kT.

®® So, the temperature of a gas is a scale for the energy associated with the
random motion of the particles.

®® In addition to energy, the particles have momentum. And the scale for that
is pressure. The particles are constantly knocking into the walls of the box,
or any surface that might be inserted in the gas. Those
collisions exert a force on the surface—that’s pressure.
PRESSURE

®® To simplify the math, let’s imagine a universe in which P = nkT

particles can move in only one direction (back and forth).

®® When a particle with speed v hits the wall, it reflects back with speed v in
the opposite direction. Its momentum changes from +mv to −mv, a change
of −2mv. Because momentum is conserved, the wall must have absorbed
a momentum of +2mv. It feels a push. And that keeps happening as more
particles hit the wall. In a time ∆t, how much momentum does the wall absorb?

®® Let’s say all the particles have the same speed, v, but half are moving right
and half are moving left. The particles that hit the wall are the ones moving
to the right that start within a distance of v∆t from the wall.

®® If we focus on an area ∆ A of the wall, that singles out a box of volume v∆t∆ A.
So, the total momentum absorbed by the wall will be the momentum from
each collision (2mv) times the number of collisions (n ⁄ 2, the number density
of particles moving to the right) times the volume of the box, v∆t∆ A.
⇣n⌘
p = (2mv) (v t A) = nmv 2 t A
2

101
 Lecture 8 — Photons and Particles

®® Force is momentum per unit time and pressure is the force per unit area, so
to get the pressure, we divide our equation by ∆t and ∆ A, giving nmv 2. And
because mv 2 is twice the kinetic energy, ε, we can also write it as 2nε.

p/ t 1
Pv = = nmv 2 = 2n mv 2 = 2n✏
A 2

®® We just assumed that all the particles have the same speed, v, but that’s not
true. There is a range of speeds. So, we should replace ε by its average value,
which would be 1 ⁄ 2kT in a 1-dimensional universe.
1
P = 2nh✏i = 2n kT
2
P = nkT

If you have a background in ®® We just derived the ideal gas law:

chemistry, you might be used to Pressure is proportional to number
density and temperature. We did it for a
seeing P = nkT as P = nRT. These
1-dimensional gas, but in 3 dimensions,
equations are the same. The R
you end up getting the same equation.
is just k in chemists’ units; they
like to count particles in units ®® Suppose we pop a tiny hole of area A in
of moles. Physicists tend to just the wall. Gas particles will start leaking
count the particles individually. out and the gas will lose energy. What’s
the rate of energy loss?

®® In our 1-dimensional universe, the number of particles that leak out in time
∆t is equal to the number density of right-moving particles, n ⁄ 2, times the
volume of that same box, v∆t∆ A. Each one has energy ε, giving
⇣n⌘
E=✏ (v t A).
2

®® Let’s divide by ∆ A and ∆t to give the power per FLUX

unit area—the flux—of escaping energy.
F / nhv✏i / T 3/2

102
 Lecture 8 — Photons and Particles

®® Again, because there’s a range of speed and energies, we need to take the
average. The important thing is that it’s proportional to nvε, which also turns
out to be proportional to T (3 ⁄ 2).

power E/ t 1
F = = = nhv✏i
area A 2

COLLECTIVE PROPERTIES OF PHOTONS

®® Photons, unlike particles, do not collide with each other; they sail right
through each other. The only things photons interact with are charged
particles. In order to randomize the positions and energies of photons, you
need to have charged particles—which are themselves in thermodynamic
equilibrium—around. So, we’ll fill our box with charged particles, colliding
all the time, producing momentary accelerations and thereby producing and
absorbing photons.

®® For an ideal gas, the average energy per particle is

3 ⁄ 2kT. For the photons, it turns out to be 2.7kT. AVERAGE ENERGY
That’s not so weird. They’re both proportional h✏i ⇡ 2.7 kT
to temperature; it’s just a different numerical
constant in front.

®® Things get weird, though, with the number

density. For an ideal gas in a closed box, n is NUMBER DENSITY
constant; the particles don’t spontaneously ✓ ◆3
3.9 kT
appear out of nowhere or vanish. Even if we add n⇡
hc
energy, speeding up the particles, their number
stays the same.

103
 Lecture 8 — Photons and Particles

®® But photons do appear out of nowhere—whenever a charged particle

accelerates. The particle flings away some of its own energy in the form
of photons. Likewise, a photon vanishes when its energy is absorbed by a
charged particle.

®® So, for photons, we shouldn’t expect n to be a constant. If we inject energy

into the gas, speeding up the particles, the magnitude of their accelerations
will rise and they will produce more photons. It
turns out that the number density of photons rises
as temperature to the third power. ENERGY DENSITY
4 4
u= T
®® For the ideal gas, energy density, u, is proportional c
to nT. For photons, n varies as T  3, and if energy
density varies as nT, then we might expect
the energy density of photons to go as T  4. And it does. The constant of
proportionality is traditionally written (4 σ) ⁄ c, where σ is the Stefan-
Boltzmann constant.

®® For the gas, pressure equals nkT, the ideal gas law. For PRESSURE
photons, again, n goes as T  3, so we might expect pressure 4 4
to go as T  4, and it does. In this case, the proportionality P = T
3c
constant is (4σ) ⁄ 3c.

®® For the gas, flux—the power per unit area that would emerge from a tiny
hole in the box—is n times the average of vε, which is proportional to T  3 ⁄ 2.
For photons, n scales with T  3, v is always c, and ε is proportional to kT, so
we might guess that flux is proportional to T  4, and it is.

We never notice radiation pressure because at ordinary temperatures, it’s negligible. In

principle, though, when you’re at the beach, the sunlight hitting your body pushes you
down into the sand—a little. Inside really massive stars, though, the radiation pressure
can be so large that it blows the star apart.

104
 Lecture 8 — Photons and Particles

®® This result—the flux of electromagnetic radiation from a

body at temperature T is proportional to T  4 —is important FLUX
enough to deserve its own name: the Stefan-Boltzmann law. F = T4

®® The constant of proportionality is σ, the one that also

appeared in the equations for energy density and pressure. It’s not a new
fundamental constant; it’s a certain combination of h, c, and k. But it occurs
so frequently that the abbreviation is helpful.

2⇡ 5 k 4 8 2 4
σ= = 5.7 ⇥ 10 Wm K
15h3 c2

COMPARING DISTRIBUTIONS
OF ENERGIES
®® For the case of the gas, the average energy is 3 ⁄ 2kT. If we pick a particle
at random, we expect its energy to be about 3 ⁄ 2kT—but not exactly—
depending on its recent history of collisions. Likewise, the speed of a given
particle is always fluctuating.

3
h✏i = kT
2

®® A fundamental rule that emerges from classical statistical physics is that the
probability to find a particle in a state with energy ε is proportional to e −ε  ⁄ kT,
an exponential function called the Boltzmann factor.

Boltzmann factor: prob. (state) / e

prob.(state) E/kT

®® This means that the energy will almost always be of order kT. Much larger
energies are vanishingly rare, because of the exponential falloff. The particles
tend to share the energy equally; there’s little chance that one particle is going
to end up with a disproportionate share of the total energy.
105
 Lecture 8 — Photons and Particles

®® From that basic rule, it’s possible to ✓ 1 ◆

mv 2
derive the probability distribution prob.(v) / v 2 exp − 2
kT
for the energy, or the speed, of

Relative probability
a particle in an ideal gas. It’s
called the Maxwell-Boltzmann
distribution, and it’s the product of
the Boltzmann factor and a factor
of v 2 that comes from counting the 0 500 1000 1500
Speed [m s−1]
number of possible states for the
particle to have energy E.

®® The horizontal axis is speed (in

meters per second), and the vertical
✓ 1 ◆
axis is the relative probability mv 2
prob.(v) / v 2 exp − 2
that you’ll find a particle to have kT
Relative probability

that speed; in other words, it’s the

fraction of particles that have that
speed at any given time.

®® The function depends on the

particle mass and the temperature. 0 500 1000 1500
Speed [m s−1]
For this chart, the case of nitrogen
molecules at 300 Kelvin—basically,
air at room temperature—was
chosen. The distribution rises from
0 to a peak at around 400 meters ✓ 1 ◆
mv 2
per second and then falls off again. prob.(v) / v 2 exp − 2
kT
r
2kT
Relative probability

vpeak =
®® If the temperature is increased, m
the peak spreads out and moves to
higher velocities.

®® If the temperature is decreased, the

peak moves to the left. In general, 0 500
Speed [m s−1]
1000 1500

the most probable speed is  2kT ⁄ m.

106
 Lecture 8 — Photons and Particles

®® Let’s switch to photons. Imagine trapping some photons from the Sun in a
reflecting box and measuring the wavelength of each one. Once we collect
enough photons, we find that there’s a peak at around 0.6 microns. That’s the
most popular wavelength to have. The shape of the function looks like the
Maxwell-Boltzmann distribution, but it’s different in detail, because photons
are not your everyday particles. It is called a Planck spectrum.

®® One of the curves is for 5800 Kelvin, approximately the temperature of the
Sun’s outer layers. The bright star Vega is hotter—closer to 9500 Kelvin—
so its spectrum is shifted toward higher energies, which means shorter
wavelengths. And the faint, nearby star Proxima Centauri is only about 3000
Kelvin, so its photons generally have lower energies and longer wavelengths.

®® If we pop a tiny hole in our box, the total flux that leaks out is σT  4. But the
contributions to that flux from photons of various wavelengths is the flux
density—power per unit area per unit wavelength.

107
 Lecture 8 — Photons and Particles

®® When we measure the flux density of the Sun as a function of wavelength, we

find it’s a pretty good fit to a theoretical Planck spectrum. It peaks at around
2 kilowatts per square meter per micron at a wavelength of half a micron.

®® It doesn’t fit exactly because the Planck spectrum describes the radiation
you get from particles that have been knocking around long enough to reach
a constant temperature. They’re in thermodynamic equilibrium. It’s often
called a blackbody spectrum, because the derivation relies on the material
being a perfect absorber of photons and therefore “black.”

®® The Sun, or any other real object, does not meet those criteria exactly. The
Sun is not all at one temperature; it gets hotter as you go deeper. And the
Sun’s material is not perfectly absorbing. But the spectrum of the Sun and
other stars are nevertheless reasonably well described by the Planck function,
which is what allows us to say that the Sun is approximately a blackbody.

®® Let’s look at the Planck spectrum on logarithmic axes. That way, we can let
the wavelength scale range over a factor of 1000, from ultraviolet to infrared,
and we can let the flux density scale over a factor of a trillion.

108
 Lecture 8 — Photons and Particles

®® As the temperature increases, the curve lifts up vertically: Hotter sources

produce more radiation at all wavelengths. The area under each curve—the
integral of flux density over wavelength—is the total flux, which is equal to
σT  4. That’s the Stefan-Boltzmann law. Double the temperature and the flux
rises a factor of 16.
R
F = F dλ = σT 4

®® As the temperature increases, the peak of the spectrum shifts to shorter

wavelengths. This makes sense because we expect the typical photon energy,
hc ⁄ λ, to be of order kT, implying λ should be of order hc ⁄ kT—inversely
proportional to temperature.
hc
⇠ kT
peak

®® When we do the math exactly, we find that the peak of the spectrum occurs
when λ is about 1 ⁄ 5 of hc ⁄ kT. That’s called Wien’s law. We can also write it
as a scaling relation.
✓ ◆ 1
hc T
peak ⇡ ⇡ 10 µm
5kT 300 K

109
 Lecture 8 — Photons and Particles

We are constantly bathed by photons whose spectrum follows the Planck

function with an accuracy better than 1 part in 10,000 and a temperature of 2.7
Kelvin. According to Wien’s law, that corresponds to a wavelength of 1 millimeter,
which is in the microwave band of the spectrum.
Why is the universe permeated with this microwave blackbody radiation? It’s a
clue that at some point in the past, the universe itself was a “gas” of particles at
a single temperature—in thermodynamic equilibrium—long before it became the
place we know today, with tiny pockets of extreme heat and vast expanses of
frigid cold.
This so-called cosmic microwave background radiation is some of the best
evidence we have for the big bang.

https://phet.colorado.edu/en/
simulation/radiating-charge.

110
Lecture 9

COMPARATIVE
PLANETOLOGY

T his lecture will analyze the planets from an

astrophysicist’s perspective. Mercury, Venus, Earth,
and Jupiter—a quartet of planets spanning a wide range
in orbital distance, size, mass, temperature, and type
of atmosphere—make for an interesting comparative
study. The lecture will address the following questions:
Why are these planets so different? Why doesn’t
Mercury have any atmosphere? Why is Venus so much
hotter than Earth? Why is Jupiter so huge?

111
 Lecture 9 — Comparative Planetology

COMPARING ORBITAL DISTANCE,

RADIUS, MASS, AND TEMPERATURE
a [AU] R [R Å] M [M Å] T [K]

EARTH 1 1 1 288
VENUS 0.72 0.95 0.81 737
MERCURY 0.39 0.38 0.055 100−725
JUPITER 5.2 11.2 317 124

®® Earth orbits the Sun at 1 AU; that’s its orbital distance (a). Its radius (R) is
1 Earth radius, and its mass (M ) is 1 Earth mass. The surface temperature
averaged over the whole globe (T ) is about 59° Fahrenheit, or 288 Kelvin.
And it has an atmosphere composed mainly of nitrogen and oxygen.

®® In many ways, Venus is like Earth’s twin sister, except its atmosphere is
mainly carbon dioxide and it’s much thicker than Earth—almost 100 times
as massive. It’s also scorching hot, with an average surface
temperature of 737 Kelvin, 2.6 times hotter than Earth.

®® Mercury, the closest planet to the Sun, has a

semimajor axis of 0.39 AU. Mercury is relatively
small, 38% the size of the Earth and 5.5% of the
mass. It looks like the Moon: barren and pocked
with craters. Also like the Moon, Mercury has no
atmosphere to speak of. Its surface temperature
varies from a freezing 100 Kelvin at night to a
broiling 725 Kelvin at noon.

®® Going the other direction—beyond Earth and past Mars—Jupiter is out at

5.2 AU. Jupiter is a different beast from Mercury, Venus, and Earth. It’s a gas
giant planet. It’s basically all atmosphere; it’s a ball of hydrogen and helium
gas with no solid surface. As you dive deeper into the planet, it keeps getting
112
 Lecture 9 — Comparative Planetology

denser, until eventually the pressure becomes so

high that hydrogen liquefies and even turns into
a metal. For Jupiter’s surface temperature, we
can use the Stefan-Boltzmann law to define
an effective temperature—the temperature of
the outermost layers of gas—which comes out
to be 124 Kelvin. That’s colder than it ever
gets on Earth.

WHAT DETERMINES THE

TEMPERATURE OF A PLANET?
®® Why is the Earth’s average surface temperature 288 Kelvin and not much
higher or lower? The Earth absorbs sunlight. But it doesn’t retain all that
solar energy. It radiates. As the Earth’s temperature rises, it radiates more
power, rising as T  4.

®® At some point, the radiated power equals the incoming solar power—at which
point the Earth stops heating up. It reaches radiative equilibrium, with no
net gain or loss of energy.

®® Let’s calculate the temperature of a planet that has

achieved radiative equilibrium. There’s some
flux, Fin, of incoming solar radiation,
and the planet is a big target with
cross-sectional area πR 2 that
intercepts that flux. So,
the incident power is

Pin = Fin · ⇡R2 .

113
 Lecture 9 — Comparative Planetology

®® In equilibrium, this must equal the outgoing power. For simplicity, let’s
assume the entire surface of the planet is radiating like a blackbody at a
single temperature T, so, according to the Stefan-Boltzmann law, the flux is
σT  4, and the total radiated power is σT  4 times the total surface area, 4πR 2.

Pout = σT 4 · 4⇡R2

®® In radiative equilibrium, we set power in equal to power out and solve for
T, giving

Pin = Pout −! Fin = 4 T 4

✓ ◆1/4
Fin
T =
4

®® Fin, the solar flux at the orbital distance of the planet, is equal to the luminosity
of the Sun spread out over a giant sphere with a radius equal to the orbital
distance, a.

L
Fin =
4⇡a2

®® We can simplify this if we also approximate the Sun as a blackbody and

replace L with the following.

L 4⇡R2 σT 4 R2 σT 4
Fin = = =
4⇡a2 4⇡a 2 a2

®® That way, we get some nice cancellations, leading to our final answer.
✓ ◆1/4 ✓ ◆1/4 r
Fin R2 T 4 R
T = = =T
4 4 a2 2a

114
 Lecture 9 — Comparative Planetology

®® If we evaluate the numbers for the case of Earth, we get 280 Kelvin. That’s
not far from the true average temperature (T ) of 288 Kelvin—an encouraging
sign that our calculations captured the essential physics. We can write the
result as a scaling relation.

280 K
=p
a/(1 AU)

®® This makes it easy to apply to the other planets in the quartet. We can add
a column to the chart, called equilibrium temperature (Teq), with the result
of this calculation.

a [AU] R [R Å] M [M Å] T [K] Teq [K]

EARTH 1 1 1 288 278

VENUS 0.72 0.95 0.81 737 328
MERCURY 0.39 0.38 0.055 100−725 446
JUPITER 5.2 11.2 317 124 122

®® This formula works well for Jupiter, too. But the calculated temperature for
Venus is more than 400° too cold. And what about Mercury, where there’s
no single temperature? Clearly, we’re missing something important.

®® What we’re missing are the effects of the

atmosphere. Venus has a very thick atmosphere, Atm.
and Mercury has none.
EARTH N2, O2
®® When we derived the equilibrium temperature, VENUS CO2
we equated the power from the Sun to the power
MERCURY none
radiated by the entire surface of a spherical
planet. We assumed that, somehow, the surface JUPITER H2, He
maintains a constant temperature.

115
 Lecture 9 — Comparative Planetology

®® For the Earth and Jupiter, that’s not a bad assumption, because the global
circulation of the atmosphere tends to smooth out temperature differences. But
on Mercury, there’s no way for heat to flow quickly around the surface, so the
dayside gets cooked and the nightside freezes. Therefore, our approximation
of a constant temperature is inappropriate.

®® Let’s recalculate the temperature, using the opposite approximation: no heat

transfer. Every square meter of the surface has its own temperature. For
simplicity, let’s target the hottest-possible temperature—which is at local
noon, when the Sun is directly overhead.

®® The incoming solar flux, Fin, is the same

as before, but the outgoing flux, Fout,
is the blackbody radiation at the local
surface temperature, σT  4, or
✓ ◆1/4
Fin
−! Ts = .

®® It’s the same as before, except we’re missing a 4 that used to be in the
denominator—which means the result will be without the 2.
r
R
=T
a

®® So, the temperature we calculate under the assumption of local reradiation

is higher by a factor of 2 .

®® In addition, for Mercury, the semimajor axis, a, is 0.39 AU. But if we want
the absolute hottest temperature, we should plug in the distance of closest
approach to the Sun, a(1 − e), where e is the orbital eccentricity.
s
R
Ts, max = T
a(1 e)

116
 Lecture 9 — Comparative Planetology

®® With that correction, we get 708 Kelvin, which agrees pretty well with
the data.

®® Venus has plenty of atmosphere, so a constant surface temperature is a

reasonable approximation. This time, our calculation didn’t work because
we ignored the possibility that the atmosphere—not just the surface—absorbs
and emits radiation.

®® Venus’s atmosphere is full of carbon dioxide molecules. What kinds of

photons do those molecules absorb and emit?

®® For something to absorb a photon, there has to be somewhere the photon’s

energy can go. There must be something ready to accept the few electron
volts that a solar photon has to offer. Usually, the things that can absorb a
few electron volts are electrons; they use the energy to jump between different
orbits in atoms and molecules.

®® But the problem is that carbon dioxide is electronically stable. The electrons
are happy where they are, so it takes tens of electron volts before they can be
persuaded to jump into higher orbits. That makes carbon dioxide transparent
to visible light. The same is true of nitrogen and oxygen.

®® But molecules do have other ways to absorb energy, besides shifting around
electrons. The atoms are joined by chemical bonds that act sort of like springs,
which can absorb energy and start vibrating and rotating. And the energies
of those motions are on the order of a few hundredths of an electron volt.

®® Objects that radiate photons with those kinds of energies are at room
temperature. For a blackbody, at 300 Kelvin, the characteristic energy, kT,
is 1 ⁄ 40 of an electron volt, and the typical wavelength is 10 microns—
infrared radiation.

117
 Lecture 9 — Comparative Planetology

®® So, the Sun’s photons, with energies of a few electron volts, sail through
the atmosphere as if it weren’t there and get absorbed by the surface. The
surface temperature rises to a few hundred Kelvin and starts radiating
infrared photons, which can’t just escape into space. They get absorbed by
the atmosphere.

®® All this means we need to modify our calculation of the equilibrium temperature.

®® Rays of sunlight are striking the planet’s surface with a flux of Fin. The surface
is at temperature Ts, and it radiates infrared photons with a flux of σTs 4.

®® When we were analyzing Mercury, we set the surface flux equal to Fin, but
now let’s include an atmosphere. We’ll use the simplest-possible model: a
layer of gas completely transparent to visible light and completely opaque to
infrared. All the upward flux from the surface gets absorbed.

®® Next, the atmosphere heats up and starts radiating. It reaches some equilibrium
temperature, Ta, and thereby produces a flux of σTa 4, both upward (into space)
and downward (back to the surface).

®® In equilibrium, the flux radiated away from the surface must equal the flux
hitting the surface—from both the Sun and the atmosphere. Therefore,

Ts4 = Fin + Ta4.

®® The atmosphere also reaches

equilibrium between outgoing
and incoming flux:

2 Ta4 = Ts4.

®® The 2 is there because the atmosphere radiates from both the top and
bottom surfaces.

118
 Lecture 9 — Comparative Planetology

®® This time, we have 2 equations instead of one. To solve for Ts, the surface
temperature, we double the first equation and then use the second one to
eliminate the σTa 4.

−! 2 Ts4 = 2Fin + 2 Ta4

2 Ts4 = 2Fin + Ts4
Ts4 = 2 Fin

®® There’s an extra factor of 2 on the right side of the equation, compared to the
case of no atmosphere, so in this model, the atmosphere increases the surface
temperature by a factor of 4 2 , or about 19%.

®® This is the famous greenhouse effect—the same one that has everyone worried
on Earth. If we make the atmosphere more opaque to the Earth’s own thermal
radiation, by pumping out megatons of CO2, then the surface will heat up.

®® For Venus, the greenhouse effect is much larger than 20%. It’s more like
220%!

WHY ARE THE PLANETS’

ATMOSPHERES SO DIFFERENT?
®® The answer to the question of why planets’ atmospheres are so different has
3 essential physical ingredients: the equilibrium temperature, the Maxwell-
Boltzmann distribution, and the escape velocity.

®® The Maxwell-Boltzmann distribution tells us the distribution of speeds of the

gas molecules and how it depends on temperature. The peak is at a speed of
r
2kT
vth = .
m

119
 Lecture 9 — Comparative Planetology

✓ 1 ◆
®® That’s the most probable mv 2
prob.(v) / v 2 exp − 2
speed— vth, the speed due kT
to thermal fluctuations. r

Relative probability
2kT
vth =
m
®® The thermal speed is the
most probable one, but
there’s a distribution of
velocities, a spread of an
order of magnitude or so.
0 500 1000 1500
Speed [m s ] −1

®® This means that for hot

gases and lightweight
molecules—high T and
low m —there’s a danger that some molecules will be moving faster than
the planet’s escape velocity. Deep in the atmosphere, that doesn’t matter,
because after a few nanoseconds, the molecule will bash into another one,
randomizing its speed again. But at the top of the atmosphere, where the
density is low, the fastest-moving particles might escape into space and
never return.

®® Let’s consider hydrogen gas, H2, with a mass of 2 proton masses. At the
Earth’s average surface temperature of 288 Kelvin, the thermal velocity is 1.5
kilometers per second, which means a small fraction of hydrogen molecules
have speeds as high as this speed.
r
2kT
vth = = 1.5 kmss−11 for H2 at 300 K
1.5 km
m
r
2GM
®® Meanwhile, the escape velocity is
vesc = , which for Earth is 11 kilometers
R
per second.
r r
2GM 2GM
vesc = vesc = 11 km s−1 for Earth
R R

®® So, if there were hydrogen gas in Earth’s atmosphere, the random jostling
of those molecules would cause some of them to leave—and by the time a
billion years goes by, almost all the hydrogen would be gone.
120
 Lecture 9 — Comparative Planetology

®® To be sure a planet can hold on to a molecule, let’s require the escape velocity
to be at least 10 times higher than the thermal velocity. That leads to an
inequality that we can solve for m, the mass of the molecule. It will be useful
to express that mass in units of proton masses.
r r
vesc > 10 vth −! 2GM 2kT
> 10
R m
GM kT
> 100
R m
m kT R
> 100
mp GM mp

®® Running the numbers for the case of the Earth at room temperature, 300 K,
we get a minimum molecular mass of 4.
✓ ◆✓ ◆
T R/M
>4
300 K R /M

®® That’s the mass of helium—2 protons and 2 neutrons. So, this calculation
suggests the Earth can’t retain hydrogen and helium is a marginal case. But
nitrogen and oxygen are plenty heavy. N2 has a
molecular mass of 28, and for O2, it’s 32. mmin [mp]

EARTH 4
®® Let’s evaluate the minimum mass for the other
planets. For Venus, it’s 11; it’s higher than Earth’s VENUS 11
because Venus is hotter. Venus can’t retain MERCURY 66
hydrogen and helium either, but holding on to
JUPITER 0.06
CO2, with a molecular mass of 44, is no problem.

®® Mercury is hot and has a low escape velocity, so it can’t hold anything lighter
than 66. That rules out all the common molecules.

®® For Jupiter, we get a number that is less than 1, meaning that it’s massive and
cold enough to retain even the lightest gases.

121
 Lecture 9 — Comparative Planetology

®® By comparing thermal velocity to escape velocity, we can understand some of

the patterns in planetary atmospheres. But keep in mind that real atmospheres
are complicated.

Venus and Earth are the same size and have the same mass, so why is Venus’s
atmosphere 100 times more massive than Earth’s?
The current thinking is that Venus and Earth originally did have similar
atmospheres, but Venus underwent a runaway greenhouse effect—a positive
feedback loop that ultimately led to complete evaporation of all the surface water.
Venus was left as a blistering hot, dry planet smothered in carbon dioxide.

WHY IS JUPITER SO MASSIVE?

®® Jupiter is massive enough to retain hydrogen and helium. But why is it
so massive?

®® Hydrogen and helium are the most common elements in the universe—by
far. Carbon, nitrogen, and oxygen are only a percent or 2 by mass of the total
inventory of atoms. The Sun and all the stars are basically giant spheres of
hydrogen and helium. So, then, why aren’t the planets like that, too? Why
aren’t all planets like Jupiter?

®® It’s because planets form in a very different way from stars. Stars are what
happens when a cloud of gas collapses under its own gravity into a ball of gas
dense and hot enough to ignite nuclear fusion reactions.

122
 Lecture 9 — Comparative Planetology

®® Planets form out of the stuff left over after star formation. A new star is
surrounded by a rotating disk of hydrogen and helium gas. Planets are
thought to start from the microscopic flecks of heavier elements that are
mixed in with this gas—dust grains. Over time, gravity causes the dust to
settle down into a layer that is thin and dense enough that the dust grains
start colliding and stick to each other. Over millions of years, they grow to
the size of planets.

®® The details of this process, called planetesimal formation, are still hazy, but
what is clear is that the hydrogen and helium don’t participate. They’re too
lightweight; their high thermal speeds prevent them from condensing.

®® Why aren’t all planets rocky? Where did Jupiter come from?

®® If a rocky planet gets big enough, its escape velocity will start to exceed
10 times the thermal velocity of the gas. That allows the planet to start
attracting and retaining gas from the huge reservoir of hydrogen and helium
all around it.

®® In the solar system, this process had no trouble making objects as large as
Venus and the Earth. Those planets can’t hold on to hydrogen. They’re too
hot, and their escape velocities are too low. How much farther from the Sun
would we need to move them so they could retain hydrogen?

®® Let’s take our equation for the minimum mass of molecules that can be
retained—which is proportional to temperature—and combine that with
our equation for the equilibrium temperature in terms of orbital distance.

m kT R
> 100 r
mp GM mp m kT R R
r > 100
R mp GM mp 2a
T =T
2a

123
 Lecture 9 — Comparative Planetology

®® Then, we can plug in the mass of hydrogen and solve for a , the orbital
distance. This comes out to be 3.4 AU. That’s right in between the orbits of
Mars and Jupiter—that is, at the dividing line in the solar system between the
inner rocky planets and the outer gas giant planets. So, a rocky planet needs
to be far away from the Sun to be cold enough to start attracting hydrogen
and have a chance of becoming a gas giant.

®® Of course, in reality, planet formation is more complicated than we’ve

accounted for in our calculations.

Why are hydrogen and helium by far the most common elements in the
universe? Why is everything else so much rarer?
This observation, like the cosmic microwave background radiation, is another
pillar of evidence supporting the theory of the big bang.

http://nssdc.gsfc.nasa.gov/planetary/planetfact.html.

124
Lecture 10

OPTICAL
TELESCOPES

M ost people think that the main purpose of a

telescope is to magnify, or to make distant
objects seem closer. But for professional astronomers,
magnification is one of the least important reasons to
build a big telescope. As you will learn in this lecture,
the most important reasons are to collect more light
and to improve angular resolution, spectral resolution,
and temporal resolution.

125
 Lecture 10 — Optical Telescopes

COLLECTING MORE LIGHT

®® You can think of the light from distant sources as a gentle rain of photons
falling onto the Earth’s surface. When we look up, our eyes collect the rain
that happens to fall through our pupils. When we use a telescope, we’re using
a bigger bucket to collect the rain.

®® Consider Proxima Centauri, the closest star

to the Sun. The rain coming from Proxima
Centauri amounts to about 100 photons
per square centimeter per second. The
pupils of our eyes are only a fifth of a
square centimeter, and our retinas don’t
respond to 90% of the photons that get
in. Even worse, we only have about a
tenth of a second to detect a signal. Our
visual system can’t accumulate signals for
longer than that; it’s like a camera with a
hardwired shutter speed.

®® Putting all that together, the average number of photons from Proxima
Centauri that enter our eyes and get detected during a tenth of a second
is 0.2. That’s less than 1, which is why Proxima Centauri is invisible to the
naked eye.

Photon flux = 100 ⁄ cm2 s

# detected = 100 ⁄ cm2 s × 0.2 cm2 × 0.1 × 0.1 s = 0.2

®® With a telescope and a digital camera, though, we can boost all the factors
in this calculation. We can increase the collecting area, detect nearly all the
photons, and use whatever shutter speed we want.

®® The reason this is so important is that the rain of photons is not completely
steady. There are fluctuations.
126
 Lecture 10 — Optical Telescopes

®® Let’s say we’re looking at a star that’s spraying our telescope with 100
photons per second, on average. The key here is the phrase “on average.” In
reality, the number of photons arriving each second is a random number.
Whenever we’re dealing with events that occur at random times but with a
well-defined average rate—such as radioactive decays, earthquakes, or the
arrival of photons from a distant star—the relevant piece of mathematics is
the Poisson distribution.

®® Let’s say N is the average number of events we expect to occur in some time
interval. Then, the probability we will actually observe k events is
Nk
prob.(k) = e N
.
k!

®® That’s the Poisson distribution.

®® Let’s plot it for the case of N = 100 photons. The horizontal axis is k, the number
of photons detected, and the vertical axis is the corresponding probability.

127
 Lecture 10 — Optical Telescopes

®® The most likely outcome is 100. But any value between 90 and 110 is also
likely. Mathematically, the mean value of k, which is written as ákñ, is equal
to N. And the standard deviation of k, or σk —a measure of the spread in the
likely values—is equal to  N.

hki = N
p p
k = h(k − hki2 i = N

®® In this case, the mean is 100 and the standard deviation is 10, which means
there’s about a 68% chance that the number of photons in our image will be
between 90 and 110.

®® This implies that our flux measurement could deviate from the average by
about 10 units out of 100, or 10%. In other words, the signal-to-noise ratio
of our measurement is 10. The signal is the average flux of the star, and the
noise refers to the random deviations from the average.

®® What if we’re looking at a fainter source

that only delivers an average of 25 photons
to our detector? Let’s replot the Poisson
distribution for N = 25. Now the mean is
down to 25, and the standard deviation
is 5, so the signal-to-noise ratio is 25 ⁄ 5,
or 5.

®® In general, the maximum-possible signal-to-noise ratio is

signal N p
= p = N.
noise N

®® In fact, it’s usually worse than that, because the light from your favorite star
might be blended together with other sources of light—from other nearby
stars in the sky that you can’t resolve with your telescope or from the faint
glow of the Earth’s atmosphere. The photons from those other sources, called
sky noise, also contribute to the fluctuations.

128
 Lecture 10 — Optical Telescopes

®® When sky noise is significant, we need to revise our signal-to-noise equation

to be

signal N
=p
noise N + Nsky ,

where Nsky is the number of photons from other sources besides the star.

®® The inevitable fluctuations in the photon count, called Poisson noise, is an

unforgiving fact of life in astronomy. There are many ways a measurement
can be wrong, but even if you have perfect equipment and make no mistakes,
you can’t eliminate the Poisson noise.

®® But there are ways to decrease it. We can increase the exposure time, thereby
increasing both N and Nsky. We can use a camera that responds to a wider
range of photon energies, or wavelengths, boosting the photon count. But
often, these measures are not enough, and we have no choice but to build a
bigger telescope.

This chart shows the diameter

of the largest optical telescope
in the world over the centuries,
starting from Galileo’s telescopes,
which were only 10 centimeters
across, to today’s largest
telescopes, which use mirrors
that are 10 meters across.
The vertical axis is logarithmic,
which means that the upward linear trend represents exponential growth.
Telescopes have been doubling in size every 40 years or so.

129
 Lecture 10 — Optical Telescopes

IMPROVING ANGULAR RESOLUTION

®® Another advantage of big telescopes is they improve our angular resolution:
our ability to measure the direction from which photons are coming and
make sharper images.

®® The diffraction limit is the tightest-possible focus we can achieve given that
light is a wave. The smallest angle we can resolve is about 1.22 λ ⁄ D, where λ
is the wavelength and D is the diameter of the primary mirror or lens.

®® Visible light has a wavelength of about half a micron, and our eyes have a
D of about 5 millimeters. The diffraction limit works out to be 1.2 × 10 −4
radians, or 25 arc seconds.

0.5 ⇥ 10 6 m 4
✓min = 1.22 = 1.2 ⇥ 10 rad = 0.4 arcmin
5 ⇥ 10 3 m

®® But astronomers can do better by increasing D to 10 meters, an improvement

by a factor of 2000. In that case, the equation gives Δθmin of only 0.013 arc-
seconds.

0.5 ⇥ 10 6 m 8
✓min = 1.22 = 6.1 ⇥ 10 rad = 0.013 arcsec
10 m

®® There’s a catch, though. Fluctuations in the temperature and density of air

produce a blurring effect of order 1 arc second, even on a high mountaintop,
so we can’t take full advantage of the large D to improve our angular
resolution—at least not easily.

®® There are 2 ways around the problem: Put your telescope in space, above the
air, like the famous Hubble Space Telescope; or stay on the ground and use a
technique called adaptive optics, which is when you put a deformable mirror
somewhere in the light path of your telescope. Many mechanical actuators
are mounted on the backside of the mirror, which can apply tiny localized
forces under computer control, pulling and pushing by a fraction of a micron.
130
 Lecture 10 — Optical Telescopes

®® The goal is to distort the mirror in just the right way to reverse the distorting
effects of the air. You have a separate camera stare at an extremely bright star,
which you know should appear as a sharp point in the image. But it doesn’t,
because of the turbulent air; it looks like a big blotch.

®® The computer measures the shape of that blotch and uses an algorithm
that tells it how to distort the mirror to turn the blotch into a point. And
it all has to happen within a few milliseconds, because the atmosphere is
constantly changing.

®® Adaptive optics allows us to get close to the diffraction limit even with a
10-meter telescope.

®® Better angular resolution also gives us another way to reduce the Poisson
noise. The reason that helps is it improves our ability to separate the star’s light
from the other light sources; it reduces Nsky in our signal-to-noise equation.
So, in situations where the sky noise is the main problem, improving the
angular resolution leads to a higher signal-to-noise ratio.

IMPROVING SPECTRAL RESOLUTION

®® A third advantage of telescopes is they improve our spectral resolution: our
ability to measure the energies, or wavelengths, of the incoming photons.

®® Our eyes can sense millions of different colors, making very fine distinctions
between photons with wavelengths ranging from about 0.4 to 0.7 microns.
But as miraculous as color vision is, our eyes can’t analyze those different
hues in any quantitative way.

®® With a telescope, we can distinguish colors in several different ways. We can

put colored filters in the light path, admitting only the light within a narrow
range of wavelengths. Then, we can take exposures through different filters
and combine them.
131
 Lecture 10 — Optical Telescopes

®® If we want to make finer distinctions, we can put a prism in the light path,
which deflects light by an amount depending on wavelength. That way, the
light from a star, instead of appearing as a point in our image, shows up as a
little stripe of light, a minirainbow.

®® If there are other stars nearby, we don’t want all those rainbows to overlap.
That would be confusing. So, before the beam of light reaches the prism, we
interrupt it with an opaque sheet with a slit in it and position the slit so that
the light from our favorite star, or galaxy, goes through the slit, and all the
other sources of light are blocked.

®® That’s spectroscopy. The camera takes a picture of the spectrum of light so

that each pixel in the image records how many of the arriving photons have
a wavelength in a particular range. Sorting the photons by wavelength is a
powerful technique for understanding the nature of a light source.

132
 Lecture 10 — Optical Telescopes

®® Modern spectrographs differ in detail from the simple prism-and-camera

design. Instead of a prism, they’re more likely to use a diffraction grating, a
series of regularly spaced grooves in an otherwise flat surface. The grooves
have a different reflectivity from the surrounding material, and when a light
beam bounces off, the reflections from the stripes interfere with each other in
such a way that the direction of the reflected beam depends on wavelength.

IMPROVING TEMPORAL RESOLUTION

®® The fourth advantage of telescopes is they improve our temporal resolution:
our ability to measure the time of arrival of a burst of photons or any more
gradual variations in their rate of arrival.

®® Our eyes have a sort of built-in exposure time of about 20 milliseconds. For
astronomy, a fixed exposure time would be a severe limitation. Cameras allow
us to choose whichever exposure time is most appropriate for the purpose.

®® With short exposures, we can witness events that happen quickly. Today,
there are astronomical cameras capable of capturing scenes on the scale of
milliseconds, or even microseconds. This has led to the discovery of all sorts
of fascinating, rapidly varying sources.

®® Cameras also allow us to accumulate light for much longer than our eyes—to
build up a high signal-to-noise ratio.

®® The best way to think of a research telescope is a machine for measuring

the properties of photons from celestial bodies: the direction it came from,
its energy or wavelength, and the time of its arrival. There is also one more:
polarization, one of a photon’s wavelike properties. The polarization is the
direction in which the electric field of the wave is oscillating.

133
 Lecture 10 — Optical Telescopes

®® The traditional system of units that optical astronomers use to measure the
flux of a source—the power per unit area arriving at the Earth in the form of
light—is the magnitude scale. (It would be logical to express flux in standard
metric units, or watts per square meter, but astronomers hardly ever do that).
The apparent magnitude, m, of a source is defined as
✓ ◆
F
m= 2.5 log ,
F0

where F is the flux and F 0 is a reference flux.

®® The choice was made to set F 0 equal to the flux of Vega, a bright star in the
northern sky. That way, we don’t have to worry about calibrating our camera
to measure starlight in watts per square meter. We can just compare the flux
of our star to the flux of Vega. Then, if we ever want to convert to standard
units, we just look up the flux of Vega in watts per square meter, which has
already been measured.

®® Because of the minus sign, brighter objects have lower magnitudes. It makes
the magnitude scale like a ranking system for flux: a first-magnitude star
is brighter than a second-magnitude star, which is brighter than a third-
magnitude star, and so on.

®® When we invert the magnitude equation, we find

F = F0 · 10 0.4m
.

®® So, a star with 0 magnitude has F = F 0 —that is, the star has the same flux as
Vega. And a star with magnitude 1 is fainter by a factor of 10 −0.4, or about
40%.

m = 0 −! F = F0
0.4
m = 1 −! F = F0 · 10 ⇡ 0.40 F0

134
 Lecture 10 — Optical Telescopes

®® If we go down to a fifth-magnitude star, we find that increasing the magnitude

by 5 units corresponds to lowering the flux by a factor of 100.

m = 5 −! F = F0 · 10−0.4⇥5 = F0 · 10−2 = 0.01 F0

®® In addition to flux, we can use magnitudes to represent color. One way to

quantify the color of a light source is to measure its flux through a colored
filter—for example, blue—that only admits short-wavelength photons. Then,
measure its flux through a red filter and take the ratio, F B ⁄ F R . Objects that
are intrinsically blue will have a higher ratio than red objects.

®® The color index is defined as the difference between the 2 corresponding

apparent magnitudes.

color index = mB − mR

®® When we write that difference in terms of fluxes, using the definition of

apparent magnitudes and then use some standard log properties to rearrange
things, we see that the color index is a logarithmic scale for the flux ratio,
F B ⁄ F R. For this pair of filters, a high color index means a high blue magnitude
relative to red, and because high magnitude means faint, a source with a high
blue-red color index is relatively red.

FB F
= −2.5log − 2.5log R
FB,0 FR,0

FB F
= −2.5log − 2.5log B,0
FR FR,0

®® We can also use magnitudes to represent luminosity—the intrinsic power

of a source independent of its distance from Earth. For a source at distance
d that emits equally in all directions, the relation between luminosity and
flux is F = L ⁄ 4πd  2.

135
 Lecture 10 — Optical Telescopes

®® Instead of expressing the luminosity of a source in watts or as some multiple

of the Sun’s luminosity, we sometimes express it as an absolute magnitude,
defined as the apparent magnitude the source would have if we could
magically position it 10 parsecs away. (This number is an arbitrary choice;
the point is that if we put everything at a common distance, differences in
apparent magnitude signify differences in luminosity.)

®® Usually, we use m for apparent magnitude and M for absolute magnitude.

To find the relationship between them, first we need to compute the flux
the source would deliver to Earth if it were parked 10 parsecs away. Let’s call
that F10. Because flux is proportional to 1 ⁄ d   2,
✓ ◆2
d
F10 = F
10 pc .

®® Then, we form an apparent magnitude based on F10. We divide by the reference

flux, take the log, and multiply by −2.5. That’s the absolute magnitude.

136
Lecture 11

RADIO AND X-RAY

TELESCOPES

D ifferent parts of the electromagnetic spectrum

go by different names, but for the purposes
of astronomy, the spectrum can be divided into 3
parts. The infrared, visible, and ultraviolet are lumped
together and called optical. Everything with lower
energies and longer wavelengths is called radio,
and everything with higher energies and shorter
wavelengths is called x-rays. These are the 3 main
cultures in observational astronomy. Optical, radio, and
x-ray astronomers use different technologies, different
units, and different jargon. In this lecture, you will learn
about radio and x-ray astronomy.

137
 Lecture 11 — Radio and X-Ray Telescopes

MAKING AN IMAGE WITH

CAMERAS AND MIRRORS
®® Among the reasons that we build telescopes is to detect invisible radiation—to
go beyond the visible range and explore all the other orders of magnitude.

®® The analogy from the previous lecture of a telescope as a bucket for collecting
rain fails in one major respect: A bucket collects rain no matter what direction
it’s coming from, whether it’s falling vertically or coming in at a slant. For
a telescope, that would be bad; we’d have no idea where the photons were
coming from.

®® We want to point the telescope at something—a star or a galaxy—and only

collect the photons coming from a narrow range of directions. We want to
make an image, in which each pixel of the image corresponds to a different
point on the sky.

®® Mathematically, an image is a mapping between the direction the photons are

coming from and the position on some surface. Ideally, we want the number
of photons that hits each point on the surface to be proportional to the rate
of photons arriving at the Earth from a certain direction in space.

®® But how does a telescope perform that mathematical mapping? How does it
sort the photons by incoming direction?

®® Conceptually, the simplest way to make an image is with a pinhole camera,

where no lens or mirror is needed.

®® Think of a completely dark room except for a small hole in one wall that
is admitting light, and there’s a screen on the opposite wall. The light that
hits a given spot on the screen must have come from the specific direction
defined by the line from the hole to that spot, which means there is a 1-to-
1 correspondence between the x and y coordinates on the screen and the
direction from which light enters the hole.

138
 Lecture 11 — Radio and X-Ray Telescopes

®® Actually, there would be a 1-to-1 correspondence if diffraction didn’t exist.

Photons coming from angles differing by less than about λ ⁄ D radians get
blurred together in the image because of diffraction of electromagnetic waves.

®® How can we achieve the tightest-possible focus? Let’s say the distance from the
hole to the screen is F, the so-called focal length, and the hole has diameter D.

®® Let’s start with a big hole and imagine the light from a distant star goes
straight through the hole. Ideally, we want the image of the star to be a
pinpoint on the screen, but because the hole is so big, the image is a luminous
circle of diameter D.

®® To shrink the circle down into a pinpoint, we should reduce the diameter of
the hole. But at some point, D becomes so small that the diffraction limit
starts to dominate—the image of the hole stops being a crisp circle and starts
fuzzing out. If we keep reducing D beyond that point, the image gets worse,
because the diffraction limit goes as λ ⁄ D.

®® The optimal case, the tightest focus, is when the angular diameter of the
circular image, D ⁄ F, is equal to the diffraction limit, which is roughly λ ⁄ D.

D
⇠
F D
p
®® Solving for D, D
D⇠~ F.

®® We’ve just learned that the ideal size for the hole is on the order of the
geometric mean of the focal length and the wavelength.

®® There’s one big drawback to using a pinhole camera. The optimal hole is so
small that hardly any light gets through. So, while it can make sharp images
over a wide field of view, the images are faint, with a low signal-to-noise ratio.
Even a daylight scene might require an exposure lasting minutes or hours.

139
 Lecture 11 — Radio and X-Ray Telescopes

®® This makes pinhole cameras totally useless for optical astronomy, where the
light levels are down by many orders of magnitude. But variations on the
pinhole camera do find use in astronomy—when there is no other choice.

®® Very-high-energy photons—x-rays and gamma rays—punch straight through

most materials. We can’t easily build mirrors or lenses to redirect their
trajectories and focus them. But we can drill a hole in a layer of a dense
metal, such as lead or tungsten, that is thick enough to block them. Then,
for the screen, we can use germanium crystals or other materials that produce
an optical flash when a high-energy photon hits it.

®® To mitigate the problem of the small hole, astronomers drill several widely
spaced holes. This makes the pattern on the screen confusing, because now it’s
the overlap from lots of different pinholes, but if you observe the same scene
multiple times with the camera in different orientations, computer algorithms
can disentangle the information and reconstruct the scene.

®® This is called a multiple-pinhole camera,

and there’s another variation called a coded-
aperture telescope that has been used by
x-ray astronomers.

®® In the optical and radio domains,

photons don’t pack as much energy,
so we can use mirrors to focus them.
This allows us to collect much more
light than a pinhole camera.

®® At optical wavelengths, we could use

a lens instead of a mirror, but lenses
have a problem. Because they’re made of glass, they act like prisms, even
when you don’t want them to; the amount they deflect the light depends at
least slightly on its wavelength. This introduces chromatic aberration: What
should be a white point in the image turns into a multicolored blob.

140
 Lecture 11 — Radio and X-Ray Telescopes

®® That’s one reason why astronomers use reflective optics: glass mirrors coated
with aluminum or silver.

®® Another reason is that a big mirror is easier to support than a lens. Only one
side of the mirror, the reflective side, matters, so you can lay it shiny-side up
on a stable supporting structure—as opposed to a lens, which you can only
grip on the rim or else you’ll block the light.

®® The mirror has to be curved to focus light. One possibility is to use a mirror
whose surface has the shape of a parabola, in which all the photons coming
in along the symmetry axis get bounced to the same point: the focal point.
So, a parabolic bucket not only collects the photons but also redirects and
concentrates them into a small area, where we can put a detector.

®® Parabolic dishes are used in radio telescopes.

The dish is pointed straight at a source of
radio waves and the radio static coming
from that direction is measured. To make
an image, we can slew the dish through a
range of directions, recording the intensity
of the static as we go.

®® At shorter wavelengths, it’s more common

to use the imaging property of the parabola,
in addition to the focusing property.
Photons arriving head-on, straight down
the symmetry axis of the parabola, all end up at the focal point. The ones
coming from a slightly different angle, off axis, don’t get directed to the focal
point, but they do get concentrated near a different point that is displaced to
the side from the focal point.

®® For small angles, the displacement is proportional to the incoming angle—

which is just what we want to make an image: a mapping between incident
direction and location in a surface.

141
 Lecture 11 — Radio and X-Ray Telescopes

®® To capture the image, we insert a 2-dimensional detector with many separate

pixels, each of which can record the intensity at a point in the focal plane.
For optical astronomers, the detector of choice is the charge-coupled device,
which uses a thin layer of pure silicon. Photons hit the silicon and knock loose
some electrons, which can be trapped and counted by electronics mounted
on the silicon surface.

®® Optical astronomers rarely use parabolas. It would be wonderful if all the

photons from a given direction get sent to a single point in the focal plane
but that’s not the case. Only the on-axis light is focused perfectly. Off the
axis, the rays hitting different parts of
the parabola land in slightly different
locations in the focal plane.

®® This causes image distortions, or

aberrations. Stars near the middle of
the image appear pretty sharp, but
away from the center, they look like
little cones, or comets—which is why
this type of distortion is called a coma.

®® Perhaps there is a better shape than

the parabola, one that will focus light
to a single point no matter which
way it’s coming from. Unfortunately,
in 1856, James Clerk Maxwell and
Ernst Abbe proved that there’s no
shape for a mirror, even in principle,
that lacks aberrations.

®® Instead, astronomers reduce aberrations

by using multiple mirrors. The light
hits a primary mirror and bounces to a
secondary mirror, which is also curved,
before going to the detector.
142
 Lecture 11 — Radio and X-Ray Telescopes

®® Having more than one surface that can be adjusted gives the optical designer
the freedom to reduce whatever kind of aberration is most worrisome. Mirrors
with elliptical or hyperbolic cross sections, instead of parabolic ones, can be
used. Even 3 mirrors can be used.

RADIO ASTRONOMY
®® What distinguishes radio astronomy from optical and x-ray astronomy is that
the wavelengths are long. This has important implications.

®® First, it’s easier to build a focusing mirror. As a rule of thumb, the mirror
needs to be polished with an accuracy of at least a tenth of a wavelength—
that is, it needs to conform to the shape of a parabola, or whatever surface
has been chosen, to within λ ⁄ 10. That’s much easier when λ is a meter than
when it’s a millionth of a meter. This is why the world’s largest telescopes
are radio telescopes.

The big dish at Arecibo,

Puerto Rico (pictured), is
305 meters across.
In 2016, China completed
the construction of a dish
that is 500 meters across.

143
 Lecture 11 — Radio and X-Ray Telescopes

®® Another implication of the long

Radio astronomy grew out of wavelengths of radio photons is
electrical engineering. Karl Jansky, that their wave nature is much more
conspicuous than their particle nature.
a radio engineer at Bell Telephone
In fact, nobody ever calls them “radio
Laboratories, was the first person
photons”—always radio waves. We
to detect radio static coming from
can use purely wave-based methods
astronomical sources in 1931. to detect, amplify, and combine radio
For nearly a decade after waves. It’s not like counting photons;
that, the world’s leading radio it’s more like following the rise and
fall of ocean waves.
astronomer—essentially the only
radio astronomer—was Grote Reber,
®® This means that Poisson noise does
an engineer who built his own
not apply in radio astronomy; the
backyard radio telescope as a hobby statistics are different. It also allows
and mapped all the brightest radio for the possibility of interferometry—
sources in the sky. an amazing trick to improve our
angular resolution.

®® You’d think that radio images would have terrible angular resolution because
the diffraction limit is λ ⁄ D and λ is a million times bigger than in the optical
range. But, in fact, radio astronomers enjoy the best angular resolution of all.
This is because they can combine the waves from 2 widely separated receivers.

®® Let’s say we have 2 dishes that

are a distance B (for baseline)
apart and there’s a radio source
straight overhead sending
down waves, a pattern of crests
and troughs. Both dishes
detect the waves and amplify
them, without losing track of
the phase, and send the signals
to a central computer called
a correlator.
144
 Lecture 11 — Radio and X-Ray Telescopes

®® The correlator merges the

2 signals, which combine
constructively because
the crests from straight
overhead hit the dishes at
the same time, resulting in
a big signal at the output.

®® If the source is not straight

overhead, the waves are
coming in at some small
angle Δθ. In this case, the
crests will hit one of the
dishes first and then the other. In the small-angle approximation, the extra
distance the wave must travel to hit the second dish is BΔθ. If that is equal
to half a wavelength, then the waves will interfere destructively; a crest plus
a trough gives 0 signal at the output.

®® The output of the correlator can tell us whether the source is straight overhead
or displaced by an angle of λ ⁄ 2B. We can resolve details in our radio image
on that angular scale.

®® That’s just like the diffraction limit, except B is not the diameter of a single
dish; it’s the separation between the dishes. We can put them kilometers
apart, if we want, and achieve the same angular resolution as an enormously
large telescope.

®® What if we increase θ some more, so that BΔθ is an entire wavelength? At

that point, we return to constructive interference. This means that with only
the output of the correlator, we can’t tell whether a source is directly overhead
or at an angle of λ ⁄ B from the vertical—or 2 λ ⁄ B, or 3λ ⁄ B. They all lead to
constructive interference.

145
 Lecture 11 — Radio and X-Ray Telescopes

®® We can get around this problem by

building more than 2 dishes. Each
pair of dishes is sensitive to angles
that form a pattern of fringes on the
sky, with the fringes perpendicular
to the line between the dishes. If
the correlator is getting information
from many different pairs of
telescopes, with baselines in different directions, then the crisscrossing fringe
patterns allow us to pinpoint the source to a specific location on the sky so
that we can make a proper image.

The Very Large Array Radio Telescope in New Mexico is an interferometer with 27 radio dishes
that can be moved along railroad tracks to spread them apart as much as 36 kilometers. It can
achieve the same angular resolution as a 36-kilometer radio dish!
The catch is that it doesn’t have the sensitivity of such a large dish, so it doesn’t collect nearly
as much radiation as the bigger dish would, but it does solve the angular resolution problem.

146
 Lecture 11 — Radio and X-Ray Telescopes

X-RAY ASTRONOMY
®® The Earth’s atmosphere is totally opaque to x-rays. X-ray photons are energetic
enough to knock apart molecules and rip off electrons, so they slam into a
molecule and get stopped. They never make it down to the ground.

®® That’s great news for life on Earth; it keeps us from getting fried. But it’s bad
news for x-ray astronomy—a field that couldn’t get started until the 1960s,
when advances in the space program made it possible to put telescopes above
the atmosphere. The sky turned out to have a glittering display of x-ray
sources, which are now understood to be related to black holes, neutron stars,
supernovas, and other fascinating phenomena.

®® Focusing x-rays is difficult. We

can’t just use a regular mirror; the The pioneers of x-ray astronomy were
photons would penetrate through from high-energy particle physics
it instead of getting reflected. The and cosmic-ray physics. Two of these
coded-aperture mask referenced pioneers were Bruno Rossi and Riccardo
previously is one solution. But there Giacconi, who convinced NASA to launch
is another. an x-ray telescope in 1962.
®® It turns out that x-rays will reflect
from metals as long as they strike
the surface at a very grazing angle, like skipping a stone off the surface of a
pond. X-ray mirrors look like polished metal cylinders. But they’re not exactly
cylinders. The sides are slightly sloped so that x-rays can skip off the surface
and land on a charge-coupled device for detection.

®® It’s a serious challenge, though, to polish the surface of an x-ray mirror to

an accuracy of λ ⁄ 10 when lambda is just a few nanometers—it needs to be
smooth at the atomic scale! But if we want sharp x-ray images, we have to
figure out how to do it.

147
 Lecture 11 — Radio and X-Ray Telescopes

®® In 1999, NASA launched the Chandra X-ray Observatory, which makes the
sharpest x-ray images of any facility. The Chandra mirrors are glass coated
with iridium. To boost the
collecting area, Chandra
uses multiple mirrors,
nesting small ones inside
the larger ones. And there’s
a second set of hyperboloid
mirrors to reduce the
image aberrations. But the
time to research, develop,
and implement the mirror
technology was more than
20 years, and the mission
cost billions of dollars.

148
Lecture 12

THE MESSAGE IN
A SPECTRUM

T his lecture will take a deep dive into spectroscopy,

which is one of the most important things we
can do with a telescope. We can sort photons by
wavelength or, equivalently, by energy. Spectroscopy
is our main way to learn about the physical conditions
of a star, planet, nebula, or galaxy—its temperature,
pressure, and composition.

149
 Lecture 12 — The Message in a Spectrum

THE SOLAR SPECTRUM

®® In the spectrum of sunlight, along with the continuous ribbon of color,
there are dark lines. Sunlight is missing certain colors. For example, there’s a
conspicuous line in the red part of the spectrum and another one in the yellow.

®® The pattern of lines looks sort of

The dark lines in the Sun’s spectrum
random, but not completely random—
were first observed in 1802, but the almost as though it were a secret code.
message wasn’t decoded until the In fact, it is a code. It contains a message
1920s because it was written in the from the atoms and ions in the Sun’s
language of quantum theory, which outer layers, which are broadcasting
wasn’t developed until the 1920s. information about their temperature,
abundance, and much more.

®® Setting aside quantum physics, we can imagine an atom as a tiny solar system
in which the attraction is provided by the electrical force instead of gravity.
Electrons orbit the nucleus, just as planets orbit the Sun.

®® The classical theory of electromagnetism says that any accelerating charge will
radiate. And an electron orbiting a nucleus is accelerating. It’s a centripetal
acceleration; it feels an inward force. That’s what keeps it bound to the
nucleus. Therefore, an orbiting electron should emit electromagnetic waves,
which carry away energy.

150
 Lecture 12 — The Message in a Spectrum

®® In the case of planets, the total orbital energy—kinetic plus gravitational

potential—is
GM m
E= ,
2a

where a is the orbital distance.

®® A similar formula applies to the case of an electron orbiting a proton under

the influence of the electrical attraction:
⌘e2
E= ,
2a
where η is the Coulomb constant and e signifies the magnitude of the charge
on both the proton and the electron.

®® If the system is gradually losing energy by radiating electromagnetic waves, E

is getting lower. It becomes more negative. This means a must be shrinking.
The electron should spiral inward toward the nucleus.

®® This should happen to planets, too. An accelerating planet produces

gravitational waves that fly away at the speed of light. In principle, this causes
planets to spiral inward and crash into the Sun. But the timescale over which
the orbit shrinks is orders of magnitude longer than the age of the universe,
making gravitational waves irrelevant to planetary motion.

®® But electromagnetism is stronger than gravity, and atoms are smaller than
planetary systems. When you calculate how long it should take for an electron
to spiral inward and crash into the nucleus, it comes out to be on the order of
10 nanoseconds. In other words, all the atoms in the universe should collapse
within 10 nanoseconds. But they don’t, so there must be something wrong
with the classical model of the atom.

®® What’s wrong is that it ignores quantum theory. Electrons are not just
particles; they have wavelike properties, too. An electron in an atom is like a
wave that’s trapped near the nucleus.
151
 Lecture 12 — The Message in a Spectrum

®® Because the electron’s wave nature—its wave function—is confined by the

electrical attraction to the nucleus, it is a standing wave, and the possibilities
for its energy are restricted to a discrete set. Those energies are found by
solving an equation: the time-independent Schrödinger equation.

~2 @ 2 ⌘e2
=E
2m @x2 r

®® We get a discrete set of energies, instead of a continuum, which means there’s

some lowest-energy state—a minimum orbital distance—called the ground
state. In the context of gravity, that would be like a certain distance from the
Sun inside which planets can’t exist.

®® For hydrogen, the energy levels of the electron obey a simple equation:
13.6 eV
En = ,
n2

where n is a whole number. The ground state is −13.6 eV.

®® These levels can be represented on

an energy diagram in which height is
proportional to energy. Because the
energies vary as 1 ⁄ n 2, they bunch up
near 0 as n increases.

®® For bigger atoms with more electrons

and for molecules with more than
one nucleus, the energy diagrams
are more complicated, but the point
remains that the energy levels are
discrete and there’s a ground state.

152
 Lecture 12 — The Message in a Spectrum

®® This explains why atoms and molecules are stable: the electrons in the ground
state can’t radiate. To do so, they would need to lose energy, and they’re
already in the minimum energy state.

®® It also explains the dark lines in the Sun’s spectrum. An electron can jump
from one level to a higher one, but only if it absorbs just the right amount of
energy, such as from a passing photon whose energy is equal to the difference
between 2 energy levels.

®® Suppose there’s an electron in the n = 2 level of hydrogen. To jump to n = 3,

the electron needs to absorb an energy—based on our equation—of 1.89 eV.
✓ ◆
1 1
E3 E2 = 13.6 eV = 1.89 eV
32 22

®® A photon with 1 eV won’t work; there’s not enough energy. A photon with
2 eV also won’t work; the electron either absorbs a whole photon or does not
absorb a photon at all. It can’t carve off a fraction of the photon’s energy.

®® Because photon energy, E, equals hc ⁄ λ and λ equals hc ⁄ E, an energy of 1.89
eV corresponds to a wavelength of 0.656 microns.

hc
= = 0.656 µm
E3 E2

®® This is in the red part of the spectrum. In fact, it’s exactly the shade of red
that was missing from the solar spectrum!

®® The dark lines are called absorption lines. It’s a bit of a misnomer; the
atoms are emitting those special photons just as much as they’re absorbing.
The electron can fall back down from n = 3 to 2, releasing a photon with
1.89 eV.

153
 Lecture 12 — The Message in a Spectrum

®® Because we know the energy levels of atoms from laboratory experiments and
computer calculations, whenever we measure the wavelengths of absorption
lines, we can learn the identities of the atoms responsible for those lines. Each
atom produces a bunch of lines with a known pattern.

STELLAR SPECTRA
®® Let’s look at some stellar spectra in a different way. Instead of rainbows, we’ll
plot intensity versus wavelength across the visible range of the spectrum.

®® If the Sun were a perfect blackbody, the spectrum would be a smooth curve
peaking at around half a micron, but a real spectrum has lots of divots—the
absorption lines. We see now that the star is not completely black at those
wavelengths; there’s some intensity, but it’s lower by as much as 30% than
the surrounding spectrum.

154
 Lecture 12 — The Message in a Spectrum

®® For example, there is a dip at

0.656 microns from hydrogen
and one at 0.589 microns
from sodium.

®® Let’s compare the solar

spectrum with the spectra
of other stars. For that, we’ll
need to use a logarithmic
intensity axis to capture the
wide range of intensities on a
single chart.

®® First, let’s pick a bluish-white star, such as m Andromedae.

®® We can tell it’s bluer than the Sun because the intensity rises toward shorter
wavelengths, the blue end of the spectrum. And the absorption lines are
different, too. We hardly see the sodium line, and the hydrogen line is deeper.
The deep lines at shorter wavelengths are also from hydrogen; they represent
jumps from n = 2 to 4, 5, and 6.
155
 Lecture 12 — The Message in a Spectrum

®® Does that mean µ Andromedae has more hydrogen than the Sun and the
Sun has more sodium? That’s what some astronomers used to think. The
truth is subtler.

®® A big clue came when the pioneers of astronomical spectroscopy discovered

that stellar spectra can be sorted into a single sequence. The following is a
graph of the spectra of some representative stars in that sequence. From one to
the next, the shape of the spectrum and the pattern of lines change smoothly.
As we work our way downward on this plot, the hydrogen lines go away and
the sodium line builds up.

®® This pattern can be shown as a chart in which the x-axis is the position of
a star in this sequence and the y -axis is the strength of various absorption
lines—how dark they are in the spectrum. At one end, some helium lines are
strongest; then, as you move down, the hydrogen lines take over. Later, the
lines of sodium and calcium rise and fall in strength.

156
 Lecture 12 — The Message in a Spectrum

®® At the very end of the sequence, there are lines from titanium oxide, which
is closely related to the active ingredient in some sunscreen lotions. Does this
mean that the stars at this end of the sequence are made out of sunscreen
lotion? No.

®® It turns out that all the stars are made of basically the same ingredients in
the same proportions. What’s changing as we move along the sequence is the
temperature of the star’s outer layers.

The pattern of absorption lines in spectra encodes

temperature. The first person to crack this code
was Cecilia Payne (later Payne-Gaposchkin)
in 1925. By interpreting stellar spectra with
the newly discovered laws of quantum
theory, she could deduce the temperature
and composition of a star based on its
spectrum. She was the first person to
figure out what stars are made of.

157
 Lecture 12 — The Message in a Spectrum

®® To produce an absorption line, not only do we need photons with the right
energy, but we also need a supply of atoms with electrons sitting in the lower
energy level ready to absorb those photons. And the supply of such atoms
will depend on temperature.

®® In general, stars are about 75% hydrogen by mass and 24% helium, with the
remaining percent or so from heavier elements, especially carbon, nitrogen,
and oxygen. Sodium, calcium, and titanium oxide are only present in minute
quantities, but they happen to have electron energy levels in the right places
to produce strong lines when the temperature is right.

NEBULAR SPECTRA
®® A nebula is an interstellar cloud of gas and dust. There are many different
kinds, and they are found all over the galaxy.

®® The Orion Nebula is a glowing, colorful, complex cloud that appears in the
sword of Orion. At 400 parsecs away and 7 or 8 parsecs across, it’s one of the
nearest big star-forming regions, where clumps of gas are contracting under
their own gravity, eventually becoming stars.

®® The spectrum of the Orion Nebula looks different from that of a star. It’s
totally dark, except for a few bright spikes at particular wavelengths.

ORION NEBULA SPECTRUM

SUN SPECTRUM

158
Lecture 12 — The Message in a Spectrum

159
 Lecture 12 — The Message in a Spectrum

®® The key difference between the star and the nebula is density, not temperature.
The density of gas in the Orion Nebula is about 10 billion times less than
in a star.

®® The low density means our usual assumption of thermal equilibrium between
atoms and photons breaks down. The atoms in the nebula are in thermal
equilibrium; they collide and share energy, achieving a uniform temperature
of about 10,000 Kelvin for the case of Orion.

®® But photons don’t interact as frequently. If an atom inside the nebula emits
a photon, it won’t always crash into another atom and get absorbed. There’s
a good chance it will just sail off into space.

®® In the nebula, collisions between atoms push the electrons up to higher energy
levels. They eventually fall back down and emit photons, which escape the
nebula and land in our telescope, having never interacted with anything
during its journey across 400 parsecs.

®® The photons we detect, then, only have the energies that are emitted by
electron transitions within the atoms of the nebula. This is called an emission-
line spectrum.

®® The situation is different in a dense medium, like a star, where a photon

emitted by one atom almost always gets absorbed by a neighboring atom.
That atom will then reemit another photon, but not necessarily with the
exact same energy; the high density and high frequency of collisions cause the
energy levels to be blurred out. Photons can also get scattered from charged
particles—electrons or ions—which alter the photon’s energy in a continuous
way, producing more blurring of the spectrum.

®® We can imagine the photons as pinballs getting knocked around in a pinball

machine. All the bumpers, flippers, and kickers are the atoms and ions. The
directions and energies of the photons are constantly changing.

160
 Lecture 12 — The Message in a Spectrum

®® Given enough time, the atoms and photons come into thermal equilibrium,
and the photons attain a Planck spectrum. The little bit of light that leaks
out will have the spectrum of a continuous rainbow.

®® Quantitatively, the key concept separating the nebula and the star is the
mean free path: the average distance a particle travels between interactions.
The mean free path of a visible photon in air might be 10 kilometers on a
clear, dry day. That means photons from 10 kilometers away can propagate
unimpeded all the way to the eye.

®® The mean free path of the molecules in air is much shorter, on the order of a
tenth of a micron. They’re constantly colliding with each other. The photon
mean free path is longer because visible photons don’t have the right energy
to be absorbed by nitrogen or oxygen.

®® In general, a medium is transparent if its spatial extent is substantially smaller

than the mean free path of photons, such as a piece of glass or a diffuse
interstellar gas cloud like the Orion Nebula. Instead of using the word
“transparent,” astronomers use the term “optically thin.” The idea is that
from the point of view of a photon, the medium is thin; there’s little chance
a photon will hit something when it goes through.

®® Conversely, when the spatial extent is larger than the mean free path,
astronomers say it is optically thick, or opaque.

®® A related concept is optical depth, which is defined as the distance through

some medium divided by the mean free path. It’s a dimensionless number
that is often written as τ.

®® Whenever you’re peering into an optically thick medium, the photons

reaching your eye originate from an optical depth of order 1.

161
 Lecture 12 — The Message in a Spectrum

®® That’s what we mean when we refer to the “surface” of Jupiter or of the Sun.
Both objects are gaseous spheres; there’s no solid surface. The gas gradually
gets denser as you go from the outside to the inside. But we can define the
“surface” in such cases as the level at which the optical depth is of order 1.
The more technical term is the photosphere.

®® The mean free path is not the same for all photons. It depends strongly on
wavelength. Photons with a wavelength just right to be absorbed by the
surrounding atoms have a much shorter mean free path than photons with
some random wavelength.

The Basic Rules of Spectroscopy (Kirchhoff’s Laws)

1 Optically thick sources emit a continuous Planck spectrum. That’s because
the density is high enough for the photons to reach thermal equilibrium with
the atoms.
2 Optically thin sources, such as the Orion Nebula, produce emission lines.
Photons don’t interact with atoms on their way out, so there’s no way for
them to reach equilibrium.
3 If a hot, optically thick source is surrounded by a cooler, optically thin layer,
that’s when you get an absorption spectrum.

®® Now you can understand why it’s a little too simplistic to say that the dark
lines in a spectrum come from absorption. It’s true that those photons are
getting absorbed, but they’re getting emitted at the same rate. It’s more
accurate to say that the dark lines are there because the star is opaquer at
those wavelengths—we can’t see as far down into the stellar fog.

162
 Lecture 12 — The Message in a Spectrum

An image of the Sun with an appropriate telescope shows that it gets fainter near
the rim of the circle—the so-called limb of the solar disk. It’s not an optical illusion;
the radiation from the limb really is less intense than the radiation from the
center. This phenomenon is called limb darkening.
At first, you might surmise that this phenomenon is a result of the fact that the
Sun is a sphere, not a flat circle, and that there is a shadowing effect. But what
causes limb darkening is that the Sun, like all stars, is a gaseous sphere that’s
hotter and denser on the inside than the outside.
It’s a neat effect, and a useful way to test models
for the Sun’s photosphere. By observing limb
darkening in different colors, we can learn
about how temperature increases with
depth and how the overall opacity of the
Sun varies with wavelength.

163
QUIZ
LECTURES 7–12

1 Does the observation of the motion of stars around Sagittarius A* prove the
existence of a black hole? If not, what further evidence would be required?
[LECTURE 7]

2 A star falls directly into a black hole and is tidally disrupted. But if the black
hole is too massive, this happens inside the Schwarzschild radius and the tidal
disruption event cannot be observed. What is the maximum black hole mass
for which we could observe the tidal destruction of a Sunlike star? [LECTURE 7]

3 The star Sirius has a radius 1.7 times larger than the Sun and an effective
temperature of 9940 Kelvin. What is the total luminosity of Sirius relative to
the Sun? [LECTURE 8]

4 What is the peak in the spectrum of thermal radiation from your own body? At
what rate (in watts) does your body radiate energy? [LECTURE 8]

5 Which properties of the solar system can be understood from basic physical
principles? Which properties were contingent on events happening in the remote
past with no fundamental explanation? [LECTURE 9]

6 Life on Earth seems to require liquid water. This motivates the definition of the
habitable zone of a star as the range of orbital distances where a planet would
have a surface temperature from 0°C to 100°C. What are the boundaries of the
Sun’s habitable zone? How might your answer change depending on the planet’s
atmosphere? [LECTURE 9]

164
 Quiz FOR Lectures 7–12

7 What are the advantages and disadvantages of putting an optical telescope in

space? [LECTURE 10]

8 A diffraction-limited telescope of diameter 0.5 m is used to make a 10-second

exposure of a star at a wavelength of 2 µm. During the exposure, the detector
records 300 photons from the star. In the resulting image, the sky brightness is
1000 photons per square arc second. Calculate the signal-to-noise ratio of the
detection of the star. [LECTURE 10]

9 What is the main obstacle to building a telescope optimized for ultraviolet

observations? How about a radio telescope operating at 100 MHz? [LECTURE 11]

10 The Atacama Large Millimeter/submillimeter Array (ALMA) is a radio

interferometer in northern Chile. It consists of 66 antennas that observe at
wavelengths ranging from 0.3 to 9.6 mm. The maximum separation between
antennas is 16 km. What is ALMA’s best-possible angular resolution?
[LECTURE 11]

11 Imagine a hypothetical star for which temperature decreases with depth (i.e., it
is hotter on the outside than the inside). How would the appearance of the star
differ from that of the Sun? What would its spectrum look like? [LECTURE 12]

12 Which of the following objects do you expect would have a spectrum resembling
a blackbody? Venus, neon sign, human body, incandescent lamp, LED lamp,
interior of an oven, interior of a freezer, radio antenna, lava erupting from a
volcano, lightsaber. [LECTURE 12]

Go to page 337 for solutions.

165
Lecture 13

THE PROPERTIES
OF STARS

F or hundreds of years, we’ve known that the stars in

the sky are distant Suns, producing vast quantities
of light and heat. But are there different kinds of
stars? Why are they so bright and hot? Will they keep
shining forever? The answers to these questions came
only after astronomers learned to measure the basic
properties of stars: luminosity, temperature, radius,
and mass.

166
 Lecture 13 — The Properties of Stars

MEASURING LUMINOSITY,
TEMPERATURE, AND RADIUS
®® To measure a star’s luminosity—its total power output—we can use the flux-
luminosity relationship. First, we measure distance to the star, d, maybe with
parallax. Then, we measure the flux, F, the power per unit area we detect
with our telescope. Finally, we calculate the luminosity as

L = 4⇡d2 F .

®® We get the effective temperature (Teff ), the temperature of its photosphere,

from spectroscopy. We can measure the strength of the absorption lines and
place it on the spectral sequence, which is a temperature scale.

®® Let’s plot L versus Teff using logarithmic axes. To get oriented, we’ll mark the
location of the Sun, which has an effective temperature of about 5800 Kelvin
and a luminosity of 1—that is, 1 solar luminosity.

®® Next, let’s plot the data for

the 1000 brightest stars in the
sky, which range in effective
temperature from around
3000 to 30,000 Kelvin and
range in luminosity from
around 1  ⁄ 
3 to 100,000
times the Sun’s. The Sun is
way below average; it’s one of
the least luminous stars on
the chart.

®® Many of the data points follow a diagonal stripe from the lower left to the
upper right called the main sequence. The upward slope means stars with
hotter photospheres are more luminous. That makes intuitive sense. But
there’s also a bunch of points higher up on the left—very luminous stars but
with relatively cool photospheres. These are called giants.
167
 Lecture 13 — The Properties of Stars

®® Because stars are approximately blackbodies, we can use the Stefan-Boltzmann

law: The luminosity equals the total surface area, 4πR 2, times the blackbody
flux, σTeff 4.

®® And because we’re using a logarithmic chart, let’s take the log. The result is
an equation that connects the location of a point in the chart with the radius
of the star.

log L = log(4⇡σ) + 2 log R + 4 log Te↵

®® The stars in the upper left of the chart are called giants because they’re 10 to
100 times bigger than the Sun.

®® The stars on the main sequence don’t all have the same size. Stars on the
faint, cool end are the size of the Sun or smaller, and stars on the luminous,
hot end are up to 10 times bigger than the Sun.

®® This type of chart, luminosity versus temperature, is called a Hertzsprung-

Russell (HR) diagram after the astronomers who first drew these diagrams
in the early 1900s.

168
 Lecture 13 — The Properties of Stars

®® But their charts were a little different. Following astronomical tradition, they
plotted absolute magnitude instead of luminosity, and instead of effective
temperatures (which they didn’t know at that time), they plotted a color index.

®® The apparent magnitude is a logarithmic measure of flux: −2.5 times the log of

flux relative to Vega. The minus sign makes bright stars have low magnitudes,
and the 2.5 means that a difference of 5 magnitudes corresponds to a flux
ratio of 100.
✓ ◆
F
m= 2.5 log
F0

®® The absolute magnitude is a log scale for luminosity. It’s equal to the apparent
magnitude minus 5 times the log of the distance divided by 10 parsecs.
✓ ◆
d
M =m 5 log
10 pc

®® A color index is the difference between 2 apparent magnitudes measured

through different filters. It’s a log scale for a flux ratio. For example, the B −
V color index is the apparent magnitude measured through a standard blue
filter minus the apparent magnitude measured through a so-called visual
filter, which is centered in the middle of the visible range.

B V = m(blue) m(visual)

®® And because it’s all referenced to Vega, the color index of Vega is 0. For the
Sun, B − V happens to be 0.66.

169
 Lecture 13 — The Properties of Stars

®® The color index is a proxy for effective temperature—blue stars are hot and
red stars are cool—and the color index is easier to measure. You don’t need
spectroscopy; you just need a pair of colored filters for your camera.

®® A traditional HR diagram shows the absolute V magnitude against B − V.

It has the same features as the previous chart, but everything is flipped left
to right. That’s because a low value of B − V means the star is relatively
bright in B compared to V—it’s blue. And a star with high B − V is red.
So, the horizontal axis goes from blue to red, hot to cool. It’s a backward
temperature scale.

There’s a big difference between the nearest stars and the brightest stars. If we
pick a certain volume of space and count all the stars inside, we find that small,
faint stars are most common.
Bigger and more
luminous stars are
rare. To find them,
we need to look far
away. But because
they’re so bright,
they dominate our
naked-eye view of the
night sky. That’s why
Betelgeuse, Arcturus,
Aldebaran, and many
other famous stars
are giants.
In contrast, most of the nearest stars are invisible without a telescope.

170
 Lecture 13 — The Properties of Stars

MEASURING MASS AND SIZE

®® We now know how to find the luminosity, temperature, and radius of a star,
but if we want to understand stars, we need to know mass, which determines
a star’s gravitational pull and internal pressure as well as the amount of fuel
available to keep it shining.

®® How can we learn a star’s mass? To answer this question, we look toward
the constellation of Perseus, in which there is a red star called Algol. Every
3 days, Algol drops in flux by 30% and stays that way for 10 hours before
going back to normal. We can tell the difference by eye.

®® It turns out that Algol is actually a pair of stars that orbit each other and
periodically eclipse one another. Pairs of stars—called binaries—are common.
About half of the points of light in the night sky represent the combined
light of more than one star, but our eyes lack the angular resolution to see
them separately.

®® Most binary stars do not show eclipses, though. Only a small percentage
happen to have orbits oriented nearly parallel to our line of sight so that the
stars cross directly in front of each other, blocking each other’s light and
making the whole system appear fainter.

®® Eclipsing binaries like Algol allow us to measure stellar masses. We use

Kepler’s third law, which can be written as
GM 2
a3 = P .
4⇡ 2

®® In the planetary context, a was the semimajor axis of the planet’s orbit, P was
the period, and M was the mass of the Sun.

171
 Lecture 13 — The Properties of Stars

®® When discussing Kepler’s laws in lecture 4, we said that the Sun sits still at
one focus of the elliptical orbit. But the truth is that the Sun moves a little
bit. This follows from Newton’s third law: Every reaction is accompanied by
an equal and opposite reaction. If the Sun pulls on a planet, the planet must
pull back on the Sun with equal force.

®® But because acceleration equals force over mass and the Sun is so much more
massive than the planets, the Sun’s acceleration is much smaller than that of
the planets. That’s why we often neglect the Sun’s motion.

®® We can’t do that when 2 stars of comparable mass are pulling on each other.
We need to solve the 2-body problem. We’ll find that both stars travel in
elliptical orbits with a focal point at the center of mass of the system.

®® The center of mass is the average location of all the components of a system
weighted by mass. For 2 stars with vector positions r1 and r 2, the center of
mass is located at

~ com = m1~r1 + m2~r2 ,

R
m 1 + m2

where the m1 + m2 is the total mass.

®® Let’s solve the 2-body problem. We have 2 stars, each of which obeys Newton’s
second law, F = ma. For star 2,
d2~r2
m2 = F~ ,
dt2

where F is the gravitational force exerted by star 1. We can write a similar

equation for star 1, except this time it’s −F because of Newton’s third law:
The forces must be equal and opposite.

d2~r1
m1 = F~
dt2

172
 Lecture 13 — The Properties of Stars

®® When we add these equations, the left side can be rewritten as the second
derivative of m1r1 + m2r 2, which is proportional to the center of mass vector.

d2~r1 d2~r2 d2
m1 2
+ m2 2 = 0 −! (m1~r1 + m2~r2 ) = 0
dt dt dt2
~ com
d2 R
=0
dt2

®® We’ve just proven that the location of the center of mass does not accelerate.
The 2 stars pull each other around, but the center of mass moves in a straight
line at a constant speed.

®® This invites us to work in a reference frame that moves along with the center
of mass. In other words, we’ll choose the origin of our coordinate system to
be at the center of mass so that

m1~r1 + m2~r2 = 0.

®® Next, let’s go back to our equations of motion and strategically subtract

them. We’ll multiply the first equation by m1 and the second by m2 and then
subtract. The product of the masses times the second derivative of the relative
separation, (r 2 − r1), is equal to the total mass times F.

173
 Lecture 13 — The Properties of Stars

®® Let’s use the shorthand r for the relative separation, and for F, let’s plug in
Newton’s law of gravity:
d2~r Gm1 m2
m1 m2 = (m1 + m2 ) r̂,
dt2 r2

where r is the unit vector in the r direction.

®® The product of masses cancels out, and we’re left with a simple equation:
d2~r G(m1 + m2 )
= r̂.
dt2 r2

®® This looks just like the equation of motion for a single mass—which is great
because we already solved the 1-body problem in lectures 4 and 5. The only
difference is that what used to be M, the mass of the Sun, is now replaced by
the total mass of both bodies.

®® We previously proved that this equation leads to Kepler’s 3 laws, so Kepler’s

laws also apply to binary stars.

¯¯ The first law says that the vector r traces out an ellipse, although in this
case the origin is the center of mass, not the Sun.

¯¯ The second law applies, so the relative speed of the stars will rise as they
approach each other.

¯¯ The third law also applies. The only differences are that the M in Kepler’s
third law is the total mass of both stars and the a is the semimajor axis
of the relative orbit.

G(m1 + m2 ) 2
a3 = P
4⇡ 2

®® That means if we can measure a and P for a binary star system, we can
calculate the total mass.

174
 Lecture 13 — The Properties of Stars

®® We’d also like to know m1 and m2 individually, not just their total. For that,
we use the fact that the center of mass is at the origin:

m1~r1 + m2~r2 = 0,

which implies
m1
−! ~r2 = − ~r1.
m2

®® And because r = r 2 − r1, we can replace r1 with r 2 − r and then solve for r 2
again, which leads to

m1
−! ~r2 = − (~r2 − ~r)
m2
✓ ◆
m1 m1
~r2 1 + = ~r
m2 m2
m1
~r2 = ~r
m1 + m2

®® We already knew that the relative separation r traces out an ellipse, but now
we know that star 2 itself moves in an ellipse but scaled down—by an amount
that depends on the mass of star 1.

®® When you do the same thing for star 1, you get

m2
~r1 = ~r
m 1 + m2 .

®® The minus sign means that star 1 is moving in an ellipse with the opposite
orientation.

®® Each star moves in an ellipse, and the 2 ellipses have a common focus: the
center of mass. The size of each ellipse is proportional to the other star’s
mass. The ratio of semimajor axes, a2 ⁄ a1, equals m1 ⁄ m2. This makes sense;
the heavier body doesn’t move as much.
175
 Lecture 13 — The Properties of Stars

®® As they move, the stars are always on opposite sides of the center of mass.
They also have equal and opposite momenta. We can see that by taking the
derivative of our center-of-mass equation, giving

m1~r1 + m2~r2 = 0
m1~v1 + m2~v2 = 0
mm~
1v
~1 =
+ mm
2~v22~v=
2 0

®® Their velocities are always in opposite directions, and the ratio of their speeds
tells us the mass ratio: v1  ⁄ v2 = m2  ⁄ m1. The heavier star moves more slowly.

®® If we can measure the relative sizes of the orbits, or orbital speeds, we learn the
mass ratio, and if we measure the orbital period, we can use Kepler’s third law
to learn the total mass—giving us enough information to solve for m1 and m2.

®® But how do we measure the sizes of the orbits and orbital period? If our
telescope has good enough angular resolution to see both stars as distinct
points of light, we can track them over the course of at least 1 full orbit.

®® Some binaries are close enough to resolve, but not many. In general, we need to
rely on a more indirect method, based on spectroscopy: Doppler spectroscopy.

®® The Doppler effect is the shift in wavelength that you observe whenever
the source of the waves is moving. For example, when a car is speeding
past you, the pitch is higher when the car
is coming at you and lower when DOPPLER SHIFT
it goes away. The sound waves v
from the approaching car are = r
c
compressed, and shorter
sound waves mean higher
pitch. But when the car
is speeding away, the
waves get stretched out,
and longer waves mean
lower pitch.
176
 Lecture 13 — The Properties of Stars

®® The same thing happens when you have a moving source of light. The
fractional shift in wavelength is equal to vr  ⁄ c, where c is the speed of light
and vr is the radial velocity: the component of the velocity along the line
between the star and us. That’s the only component of the velocity that
produces a Doppler shift.

®® When a star is moving toward us, the wavelengths of all its absorption lines
get shifted toward the blue end of the spectrum. When it’s moving away, the
shift is toward the red end. If we measure that shift, we can calculate the
radial velocity.

®® Even though the 2 stars of Algol are blended together in our images, the
spectrum of that single point of light reveals 2 different sets of absorption
lines. They’re shifted in wavelength with respect to each other because the
stars are moving at different speeds, and as the stars go around, we can watch
those lines shift back and forth.

®® By using eclipsing binaries, we can measure stellar masses and make precise
measurements of the sizes of the stars by tracking the changes in brightness
during eclipses. Even though all we see is a point of light, we can nevertheless
measure the masses and sizes of both stars, sometimes to within a few percent.

177
Lecture 14

PLANETS AROUND
OTHER STARS

T he discovery of planets around other stars—which

we call extrasolar planets, or simply exoplanets—
didn’t get underway until the mid-1990s. That’s when
technology advanced to the point that exoplanets
could be detected.

For exoplanets, the approach of direct imaging—which would involve making

a sharp image of a nearby star and then searching for any points of light going
around the star—turns out to be nearly impossible. This is because planets are so
small and faint compared to stars.
In fact, an Earth twin is more than a billion times fainter than its star. To order of
magnitude, the planet is like a firefly buzzing a few meters away from a giant
searchlight in Las Vegas and we’re trying to detect it using a telescope in New
York City.

178
 Lecture 14 — Planets around Other Stars

DOPPLER SPECTROSCOPY
®® As with binary stars, the Doppler and eclipse techniques are where most of
our knowledge about exoplanets comes from. The physics is the same as with
binary stars, but there are some practical differences because of the extreme
contrast between planets and stars.

®® For example, a spectrum only shows the absorption lines from the star, not
the planet. The planet is way too faint. That means we can track the star’s
radial velocity with the Doppler method, but not the planet’s. But let’s find
out if we can still learn something interesting.

®® We’ll use the equations from the previous lecture, but instead of m1, we’ll use
m for the star, and instead of m2, we’ll use mp for the planet.

®® For a circular orbit, the maximum amplitude of the star’s radial velocity is
the circumference of the star’s orbit around the center of mass, divided by the
period, times the geometrical factor of sin(I ) that picks out the line-of-sight,
or radial, component of the star’s velocity.

2⇡a?
v?,r = sin I
P

®® Through Doppler spectroscopy, we can measure that amplitude, and we can

measure the orbital period. That means we can calculate asin(I )—which is
not very interesting in itself but does tell us something about the planet’s mass.

®® Recall that both the star and the planet are orbiting the center of mass, and
from the definition of the center of mass, we know that ma = mpap. We
can use that fact to eliminate a from the equation.
2⇡ap mp
= sin I
P m?

®® Next, let’s bring in the information about the orbital period. We’ll use Kepler’s
third law, which connects the orbital period with the total mass and the total
orbital separation, a + ap. In this case, though, the total mass is very nearly
179
 Lecture 14 — Planets around Other Stars

equal to the mass of the star. And because the star doesn’t move as much as
the planet, the total separation is nearly equal to ap. That gives an expression
for ap that we can substitute into the radial velocity equation.
G(m? + mp ) 2
a3 = P
4⇡ 2
Gm? 2
a3p ⇡ P
4⇡ 2
✓ ◆1/3
2⇡ Gm? 2 mp
⇡ 2
P sin I
P 4⇡ m?

®® After tidying up, we notice that the radial velocity amplitude is proportional
to the planet mass.
✓ ◆1/3
2⇡G mp
⇡ 2/3
sin I
P m?

®® We don’t have enough information to solve for both masses, but if we already
have a reliable estimate for the star’s mass—based on its similarity to other
stars for which we can measure the mass—then we can calculate the planet’s
mass from this equation.

®® Actually, because sin(I ) is on the right side, we can calculate mpsin(I ), but
not mp by itself. That’s not ideal, but at least it gives us a lower bound on the
planet mass. If mpsin(I ) is 1 Earth mass, then mp must be 1 or larger, because
the sine of an angle is always 1 or smaller.

®® So, Doppler spectroscopy reveals the orbital period and the minimum
mass of the planet. In addition, the Doppler method reveals the orbital
eccentricity from the way the radial velocity varies over the course of a full
period. A circular orbit will show a sinusoidal variation, but an eccentric
orbit will be faster in some parts and slower in others, leading to a skewed,
nonsinusoidal signal.

180
 Lecture 14 — Planets around Other Stars

PLANETARY ECLIPSES
®® When the planet eclipses the star, we get even more information. Whenever
we detect eclipses, we know we must be viewing the orbit nearly edge on,
which implies that sin(I ) is very close to 1. In that case, the Doppler signal
tells us the planet mass without any ambiguity.

®® For an Earth-mass planet orbiting a solar-mass star with a period of 1 year,

the star’s orbital velocity is 9 centimeters per second.

✓ ◆1/3
2⇡G mp
v?,r ⇡ 2/3
P m?
✓ ◆−1/3 ✓ ◆−2/3
P m? mp
⇡ 9 cm s−1
1 year M M⊕

®® To get the corresponding wavelength shift, we need to divide by the speed of

light, resulting in 3 parts in 10 billion.

v?,r 10
= = 3 ⇥ 10
c

®® Nobody has ever achieved this level of precision; we have not yet detected an
Earth twin with this method.

®® But if we make the planet 10 times more massive than the Earth or shrink
the orbital period by a factor of 1000, then the star’s velocity increases to
about 1 meter per second, which can be achieved with present technology.

®® The good news is that there are many planets more massive than the Earth,
with periods much shorter than a year.

®® While we can calculate the radii of stars based on eclipse durations, this
doesn’t work well for planets. Instead, we take advantage of the fact that the
planet is dark—essentially black compared to the star. So, when the planet
181
 Lecture 14 — Planets around Other Stars

goes in front of the star, it’s as though the stellar disk has a circular black spot.
The fraction of starlight being blocked is the area of the planet’s silhouette
divided by the area of the stellar disk, which is equal to the square of the
radius ratio between the planet and the star.
✓ ◆2 ✓ ◆2
F ⇡Rp2 Rp Rp /R?
= = = 8.4 ⇥ 10−5
F ⇡R?2 R? R⊕ /R

®® If we were on an alien planet monitoring the Sun as the Earth crossed in front,
the Sun would appear to get fainter by 84 parts per million. That’s not easy
to detect, but at least now we’re working with parts per million, not billion.

®® Measuring fluxes with a precision of parts per million is very difficult from
beneath the Earth’s atmosphere, but we can do it using space telescopes. This
is why almost all of our knowledge about Earth-sized planets comes from
space telescopes.

®® Eclipsing planets are astrophysical treasures, just like eclipsing binaries.

Unfortunately, they’re rare treasures. It takes a special coincidence to be
viewing the planetary system from just the right angle.

®® To find the odds of this happening, let’s draw the imaginary celestial sphere
surrounding a star with a planet. The star illuminates the sphere, but the
planet casts a shadow, and as it orbits, the
planet’s shadow traces out a band on the
celestial sphere.

®® If there are observers all over the

galaxy monitoring the star from
all possible directions, the only
ones who see eclipses are the ones
inside the shadow band. So, the
probability of seeing eclipses is
equal to the area of the shadow band
divided by the total area of the sphere.
182
 Lecture 14 — Planets around Other Stars

®® To calculate that probability, let’s assume that R, the radius of the star, is
much smaller than a, the radius of the planet’s orbit, which in turn is much
smaller than d, the distance to the celestial sphere.

®® From the planet’s point of view, the angular width of the shadow band is
equal to the angular diameter of the star, 2R ⁄ a. The shadow band makes
a thin ribbon around the celestial sphere with a width equal to the angular
width, 2R ⁄ a, times the distance, d. The area of the shadow band is equal
to its circumference, 2πd, times its width, and if we divide that area by the
total area of 4πd   2, we find the eclipse probability to be R ⁄ a.
✓ ◆
2⇡d ⇥ 2R? d/a R? R? /a
prob. = = = 0.005
4⇡d2 a R /1 AU

®® If we plug in 1 solar radius for R and 1 AU for a, the probability is 1 ⁄ 215,

which is approximately 0.005, or about half a percent. This means we need to
monitor hundreds of Sunlike stars with Earthlike planets before we’re likely
to find even 1 that eclipses.

®® Even worse, the eclipses are brief and easy to miss. The maximum duration
is the diameter of the star divided by the planet’s orbital speed, which we
can also calculate in terms of the orbital period and the star’s average density,
using Kepler’s third law.
✓ ◆1/3 ✓ ◆−1/3
2R? P ⇢?
max. dur. = = 13 hours
vp 1 year ⇢

®® For the Earth crossing the Sun, the maximum duration is 13 hours.

®® The goal, then, is to monitor the brightness of hundreds of stars, waiting for
the one day each year when one of the stars might dip in brightness by 84
parts per million.

183
 Lecture 14 — Planets around Other Stars

®® That was precisely the goal of a NASA mission called Kepler, a 1-meter space
telescope that spent 4 years monitoring several hundred thousand stars, a
few thousand of which were bright enough that Kepler could have detected
eclipses by planets just like the Earth.

®® And Kepler did find some—a few dozen—but it was a struggle because the
eclipse signals are so tiny. The primary mission ended in 2013, and since then,
astronomers have been arguing about which of those signals represent real
planets and which are other types of astronomical phenomena or simply noise
fluctuations that happen to mimic a planetary signal.

®® In 2018, NASA launched a new spacecraft called

TESS, which stands for Transiting Exoplanet Survey A transit is the
Satellite, that will also detect eclipsing exoplanets but, passage of a small
this time, around brighter stars that are closer to the body in front of a
Earth. That will make any newly discovered planets larger body.
easier to confirm and study.

®® TESS isn’t specifically designed to find Earth twins. That quest is only part of
what makes exoplanetary science so interesting. Even though Kepler only found
a few dozen potentially Earthlike planets, it found thousands of planets very
different from Earth—different from any of the planets in the solar system.

THEORIES AND RULES OF

PLANETARY SCIENCE
®® Let’s review the expectations we had prior to the discovery of exoplanets based
on the patterns we see in the solar system.

®® In the solar system, all the planets have nearly circular orbits, and all the
orbits are aligned with each other: They lie flat, in a single plane, to within
a few degrees. Those 2 patterns are evidence for planet formation within a
flat, circular disk of material swirling around a newborn star.
184
 Lecture 14 — Planets around Other Stars

®® A third pattern is that all the small, rocky planets are close to the Sun and the
gas giants are farther away. This is the observation that led to the theory of
core accretion, in which all planets start out rocky. They can only transform
into gas giants far away from the star—beyond the so-called snow line, where
it’s cold enough for water, methane, and ammonia to freeze, providing more
solid material to build a planet massive enough to be able to accrete hydrogen
gas and puff up to become a gas giant.

®® According to core accretion theory, the most massive planets, which produce
the biggest signals, should have very wide orbits with periods of decades,
making them hard to detect with the Doppler or eclipse techniques. But this
is not what we actually found.

®® A logarithmic chart of orbital distance against planet size for all the known
eclipsing planets shows that for solar system planets, the snow line is at
around 3 or 4 AU. The data for exoplanets shows that there are thousands
of them, many of which are very close to the star, at distances of a tenth or
even a hundredth of an AU.

185
 Lecture 14 — Planets around Other Stars

®® The chart shows many planets that are as large as Jupiter, but instead of
having orbits at 5 AU, like the real Jupiter, theirs are smaller than a tenth of
an AU. These are called hot Jupiters, and their existence seems to contradict
the story about core accretion and the necessity for giant planets to form
beyond the snow line.

®® However, this chart is not a representative sample of exoplanets. It’s a sample

of detected planets, and as such, it exaggerates the abundance of the planets
that are easy to detect—namely, large planets in tiny orbits. The chart makes
it seem like hot Jupiters are very common, but in fact, they only occur around
half a percent of Sunlike stars.

®® Likewise, the absence of exoplanets with orbits wider than 1 AU on this chart
does not imply that such planets are intrinsically rare; all it means is that we
have a hard time detecting them. The eclipse probability is too low.

®® The most commonly detected planets have sizes in between those of Earth
and Neptune and periods of less than a year. We find them in tiny multiplanet
systems, such as Kepler-11, which has 6 planets scrunched inside what would
be Venus’s orbit around the Sun. These compact multiplanet systems occur
around 30% of Sunlike stars, but they were unanticipated by theorists.

®® How did that happen? What was so wrong with the theory of planet formation?

®® Some theorists say that the theory of planet formation was correct but
incomplete; nobody had thought hard enough about what might happen after
planets form. Their orbits can get rearranged and shrink through interactions
between planets and the gaseous disk, or close encounters between planets,
or tidal effects from a passing star.

®® Other theorists are ready to abandon the idea that the snow line plays a crucial
role in giant planet formation. To make progress, we need to keep looking
for patterns among the exoplanets.

186
 Lecture 14 — Planets around Other Stars

®® We can change the chart so that it shows orbital eccentricity against orbital
distance (semimajor axis) and plot the data for planets found through the
Doppler method, rather than eclipses, because it’s the Doppler data that
reveal the eccentricity.

®® The solar system planets have low eccentricities—nearly circular orbits. But
the exoplanets have eccentricities that range all the way up to 0.9! Another
supposed rule of planetary science is broken.

®® Another pattern in the chart is that the planets with the smallest orbits tend
to have small eccentricities—more circular orbits. That’s one pattern about
exoplanets that we actually do understand. It’s a consequence of tidal forces.

187
 Lecture 14 — Planets around Other Stars

®® When a planet is close to a star, it gets slightly stretched by tidal forces. If

the orbit is eccentric, the stretching varies with time: It’s stronger when the
planet approaches the star and weaker when it recedes. The friction associated
with that constant stretching converts orbital energy into heat, and when
an orbit loses energy without changing its angular momentum, the orbit
becomes circular.

For more information about exoplanets, check out Joshua Winn’s Great Course
The Search for Exoplanets: What Astronomers Know.

https://exoplanetarchive

188
Lecture 15

WHY STARS SHINE

T he Earth intercepts less than a billionth of the Sun’s

total luminosity—which is 3.8 × 10 26 watts—yet
even that tiny fraction amounts to 10,000 times more
power than all the world’s power plants put together.
In addition, the Sun has been shining for 4.6 billion
years, based on evidence from primitive meteorites.
And geologists have found evidence that liquid-water
oceans have also existed for billions of years, so the
Sun’s luminosity couldn’t have been much lower in the
past or else the oceans would have frozen. With these
numbers, the Sun’s total energy output since it formed
can be calculated by multiplying the Sun’s luminosity by
its age, giving 5.5 × 10 43 joules of energy. Where did all
that energy come from? The answer is nuclear fusion.

189
 Lecture 15 — Why Stars Shine

Chemical reactions involve electrons changing orbits or getting traded between

nuclei, and the energy scale for those transitions is always on the order of 1
electron volt per proton mass, or 100 megajoules per kilogram. A chemical fuel
with orders of magnitude better than that cannot be found.
On the other hand, nuclear reactions involve rearrangements of nucleons—
protons and neutrons. The typical energies involved with those rearrangements
are millions of electron volts. That’s what gives the Sun its longevity.

A CRASH COURSE IN NUCLEAR PHYSICS

®® An atomic nucleus is the tiny bundle of protons and neutrons at the center
of a cloud of electrons. In a neutral atom, the number of protons equals the
number of electrons because they have equal and opposite electric charges.
The number of neutrons is about the same as, or somewhat higher than, the
number of protons.

A carbon atom has 6 electrons around a nucleus of 6 protons and 6 neutrons. Some
carbon nuclei, though, have 7 neutrons. We say that carbon has 2 main isotopes:
nuclei with the same number of protons but different numbers of neutrons.
Why do we call them both carbon if their nuclei are different? It’s because
elements are named by chemists, not nuclear physicists. The chemical properties
of an atom depend almost entirely on the number of electrons, which is equal to
the number of protons. The neutron count doesn’t matter much.
But it matters a lot for nuclear reactions. When carbon is involved in a nuclear
reaction, we need to know which isotope we’re talking about. The convention is
to specify the total number of nucleons—protons plus neutrons—which is often
called the atomic mass (A). For example, carbon-12 has 6 protons and 6 neutrons,
so A = 12, and it’s written like this:  12C.

190
 Lecture 15 — Why Stars Shine

®® A big difference between chemical and nuclear reactions is the spatial scale
over which the changes occur. The smaller scale of nuclei and the millionfold
more energy released are consequences of the strength of the strong nuclear
force, and the short range over which it acts, compared to electromagnetism.

®® In chemistry, a key concept is the heat of formation: the amount of energy

released if you make a molecule from its elemental constituents. For example,
if you mix hydrogen and oxygen to make water, you release 2.5 electron volts
per water molecule. That’s the heat of formation of water.

®® The analogous concept in nuclear physics is the nuclear binding energy,

defined as the energy released if you make a nucleus out of free protons and
neutrons. Or, equivalently, it’s the energy required to completely disintegrate a
nucleus. Carbon-12, for example, has a binding energy of 92 million electron
volts (MeV).

®® The binding energies of all the most stable isotopes have been measured in
laboratory experiments. On a chart of the measured binding energy, B, versus
atomic mass, A, the x-axis ranges from hydrogen at A = 1 to uranium at 238.

®® The data form an upward

sloping, nearly straight 2000
line. That makes sense:
B = binding energy [MeV]

The bigger the nucleus, 1500

the harder we need to work
to tear it apart. The slope, 1000
B ⁄ A, tells us how hard—
based on this chart, it’s 500
about 2000 MeV ⁄ 250, or
8 MeV per nucleon. 0
0 50 100 150 200 250
A = atomic mass

191
 Lecture 15 — Why Stars Shine

®® But despite appearances, the

line is not exactly straight. To
8
get a clearer view, let’s plot B ⁄ A
(instead of B) versus A. 6

B/A [MeV]
62
Ni, 58Fe, 56Fe
®® Now we see that it’s not as 4 most stable nuclei
simple as 8 MeV per nucleon.
2
Small nuclei have a relatively
low binding energy per nucleon. 0
B ⁄ A rises to a maximum value 0 50 100 150 200 250
at an atomic mass of around 60. A = atomic mass

®® Why does the binding energy curve have this shape? Think back to our
mental model of a nucleus as a cluster of marbles. Each one is coated with
glue—that’s the strong nuclear force. If a marble is surrounded by other
marbles, it takes about 8 MeV of energy to overcome all the glue and pull it
out. That’s the average value of B ⁄ A.

®® But that’s not all there is to it. Some marbles are near the surface of the cluster.
They’re not surrounded by marbles in all directions, so they’re not glued as
tightly in place. That lowers the binding energy relative to the average.

®® This “surface penalty” is especially bad for small nuclei, because they have
a higher surface-to-volume ratio. The surface area of a sphere goes as radius
squared, and volume goes as radius cubed, so the surface-to-volume ratio
varies as 1 divided by the radius, meaning it gets worse for small nuclei.

S 1
−! /
V R

®® Another problem is that some of the marbles—the protons—repel each other.

The electric repulsion works against the glue, lowering the binding energy.
So, there’s a penalty for having protons.

192
 Lecture 15 — Why Stars Shine

®® Then why would a nucleus want to have any protons? Why don’t we see nuclei
made entirely from neutrons?

®® It’s because of another effect called symmetry energy. The “glue” sticking
nucleons together is a little stronger when it’s between a proton and a neutron
than it is between 2 protons or 2 neutrons. That increases the binding energy
of nuclei with lots of proton-neutron pairs.

®® Now we can understand the shape of the binding energy curve. Light nuclei
suffer from high surface area; that’s why their binding energy is lower than
average. Very heavy nuclei also have low binding energies, because they suffer
from an internal conflict: As you keep adding nucleons to a nucleus, you’d
like to avoid the proton penalty and stick with neutrons, but to satisfy the
desire for symmetry, you need to add protons. The net effect is to destabilize
the nucleus.

®® Notice that there are some spikes in the binding energy curve. Let’s get a
better look at them by zooming in on the lightest nuclei.

6
B/A [MeV]

0
0 5 10 15 20 25
A = atomic mass

193
 Lecture 15 — Why Stars Shine

®® These enhancements in binding energy come from aspects of the strong force
that favor even numbers of protons and even numbers of neutrons. Helium-4,
for example, has 2 of each, so it has especially high binding energy. Also in
this category are beryllium-8, carbon-12, oxygen-16, and neon-20.

EXPLAINING THE SUN’S

ENERGY: NUCLEAR FUSION
®® We know from its spectrum that the Sun is mostly hydrogen. Could it be
using hydrogen as a fuel, fusing it into helium?

®® The stable nucleus of hydrogen, H-1, is just a proton, with a binding energy of
0. Helium-4 has a binding energy of 7 MeV per nucleon. So, hydrogen fusion
releases 7 MeV per nucleon, which is equivalent to 670 million megajoules
per kilogram.

®® But is that enough to explain the Sun’s total energy output? Let’s calculate
the required mass of hydrogen: the Sun’s total energy output divided by the
energy per kilogram released by fusion.

5.5 ⇥ 1043 J
1 = 8.2 ⇥ 1028 kg
670,000,000 MJ kg
= 0.04 M

®® By converting only 4% of its hydrogen into helium, the Sun can release
enough energy to shine for billions of years.

®® There is a close relationship between mass and energy, which Einstein

taught us about with his most famous equation, E = mc  2. This equation
means that neither energy nor mass is conserved. Interactions can change
one into the other, and the exchange rate is E = mc  2. Mass and energy turn
out to be the same kind of stuff at a fundamental level; it’s their combination
that’s conserved.
194
 Lecture 15 — Why Stars Shine

®® An important implication is that when an interaction releases energy—

such as when fusing hydrogen—the total mass of all the particles involved
must decrease.
E
m=
c2

®® This happens during chemical reactions, but we never notice it. The changes
in mass are only at the level of parts per billion. But for nuclear reactions, the
changes are more substantial.

®® Let’s calculate the fractional change in mass, ∆m ⁄ m, for hydrogen fusion. We
take the 7 MeV per nucleon that is released during fusion and divide by c  2
to get the corresponding mass loss and then divide by the mass of a proton
to make it a fraction.

m 7 MeV/c2
= = 0.0075
m mp

®® This means that a helium nucleus is less massive, by 0.75%, than the sum
of 2 isolated protons and 2 isolated neutrons. This difference in mass can be
measured in laboratory experiments.

®® All of this implies that the Sun, like anything that radiates energy, is losing
mass. And we can calculate how much. In time dt, the Sun emits energy dE
and loses mass dE ⁄ c  2. So, the rate of mass loss is dE ⁄ dt divided by c  2, and
dE ⁄ dt is the Sun’s luminosity. Dividing the Sun’s luminosity by c  2, we get

dM dE/c2 L
= = 2
dt dt c
= 4 ⇥ 109 kg s 1.

195
 Lecture 15 — Why Stars Shine

®® This means that every second, 4 billion kilograms of mass are vanishing
from the Sun, having been converted into energy and flung out into space.
But don’t be too concerned. The Sun won’t disappear any time soon. Its total
mass is 2 × 10 30 kilograms, so even after 10 billion years, the Sun will lose
only about 0.07% of its mass in this way.

PROCESS OF CONVERTING
PROTONS INTO HELIUM
®® Even after it became clear that the Sun’s luminosity has a nuclear origin, the
sequence of nuclear reactions was the subject of confusion and debate for
decades. We can’t expect 4 protons to crash into one another all at the same
time. Such 4-body collisions are always vanishingly rare. The only plausible
reactions are 2-body collisions. So, you’d think the first step would be to
smash 2 protons together to make helium-2. But there are a few problems.

®® The first one is that protons repel each other. How do we get them to approach
each other closely enough for the strong force to take over?

®® We can represent the problem schematically by plotting the potential energy

of a proton as a function of its distance from another proton. As we bring
the protons together, the electric potential energy (Ee ) rises; it’s ηe 2 ⁄ r, where
η is the Coulomb constant and e is the proton’s charge.

⌘ e2
Ee =
r

®® But if they manage to get within 1.5 femtometers, or 1.5 × 10 −15 meters, the
strong force takes over and the energy drastically decreases. The marbles come
close enough to touch—and stick.

196
 Lecture 15 — Why Stars Shine

®® To find the maximum height of the potential energy barrier, we evaluate the
electrical potential energy of 2 protons separated by 1.5 femtometers, which
comes out to be 1 MeV.

⌘ e2
= 1 MeV
1.5 fm

®® So, if we want the proton to

stick, we need to throw it at
the other proton with at least
1 MeV of kinetic energy. If
it only has a quarter of an
MeV, the repulsion will
cause it slow down, halt at
5.5 femtometers, and then
fly away.

®® Do the protons at the center of the Sun have at least 1 MeV of kinetic energy?

®® We have good reason to believe that the temperature at the center of the Sun
is 15 million Kelvin, and we know that the average kinetic energy per particle
in a gas at temperature T is 3 ⁄ 2kT.

T = 15 ⇥ 106 K

3
hEk i = kT = 0.002 MeV
2

®® According to this calculation, the Sun is way too cold! The protons should
have no chance of coming close enough to stick. So, how do they even get
started making helium?

®® The answer is that quantum theory helps the protons.

197
 Lecture 15 — Why Stars Shine

®® The time has come to set aside our mental model of protons as marbles.
Like any fundamental particle, protons also have a wavelike nature and are
described by a wave function—a cloud of probability extending over a region
of space.

®® Unlike classical particles, wave functions do not simply halt when they reach
a potential energy barrier. They leak through. The amplitude of the wave
function decreases exponentially as it leaks through, so the portion that gets
through the barrier is tiny. Nevertheless, there’s a chance for a proton to
tunnel through the barrier—even though, officially, it doesn’t have enough
energy. This phenomenon is called quantum tunneling, and it’s governed by
the Schrödinger equation.

®® Let’s perform an order-of-magnitude calculation to establish how much energy

a proton needs to have a decent chance of tunneling through the barrier.

®® The key concept is the de Broglie wavelength, λd, defined as h ⁄ p, which is
Planck’s constant over momentum. You can think of λd as the spatial extent
of a particle’s cloud of probability. When you observe the particle, it won’t
necessarily be located where you would expect based on classical physics, but
you’ll find it within about 1 de Broglie wavelength of the expected location.
198
 Lecture 15 — Why Stars Shine

®® This means that we don’t need the protons to get within 1.5 femtometers of
each other. It’s enough that their wave functions overlap to within about 1
de Broglie wavelength. That turns out to be much easier; it doesn’t require
as much energy.

®® The de Broglie wavelength is defined in terms of momentum, but we’re not

interested in momentum. Instead, we’re interested in kinetic energy, 1 ⁄ 2mv 2.

1 p2 p
Ek = mv 2 = −! p = 2mEk
2 2m

®® So, we can rewrite the de Broglie wavelength as

h h
d = =p .
p 2mEk

®® For the protons to approach within a de Broglie wavelength, the required

energy is
⌘ e2
Ek > .
λd

®® This leads to an equation with energy on both sides, which we can solve for
energy and then plug in the numbers.
p
⌘ e2
2mp Ek
Ek >
h
p ⌘ e2 p
Ek > 2mp
h
✓ 2 ◆2
⌘e
Ek > 2mp
h

Ek > 0.0025 MeV

®® The answer is 3 orders of magnitude lower than 1 MeV, the required energy
for classical particles. And it’s the same order of magnitude as the average
kinetic energy of protons at the center of the Sun.
199
 Lecture 15 — Why Stars Shine

®® So, by relying on quantum tunneling, protons can fuse even at the “low”
temperature of 15 million Kelvin.

®® An important aspect of quantum tunneling is the exponential falloff in the

wave function as it penetrates a barrier. Because of that, the probability to
tunnel through a barrier depends very sensitively on energy. Small changes
in energy lead to large changes in the fusion rate.

®® Rates of thermonuclear reaction—where the nuclei rely on random kinetic

energy to come together and fuse—depend very strongly on temperature.
The rate of proton fusion in the Sun varies as T 4. Other important reactions
vary as T 8, or even higher powers. In practice, this means that each reaction
has a sharply defined ignition temperature below which the power output is
negligible and above which it’s substantial. This has important consequences
for stellar structure.

®® But there’s another problem: Helium-2, the nucleus made of 2 protons, is

extremely unstable! So, after all that work bringing the protons together, the
nucleus falls apart and we’re back to square one.

®® What saves the day is the weak nuclear interaction, which can convert protons
into neutrons and vice versa.

®® Two protons approach each other, and right when their wave functions merge,
there’s a slight chance that one of them will change into a neutron, spitting out
a positron and a neutrino in the process. The proton and the newborn neutron
make a nucleus of hydrogen-2, also known as deuterium, which is stable.

®® However, the interaction that allows this transformation is weak. The

probability is low. If you follow a single proton bouncing around in the core
of the Sun, you’ll have to wait several billion years, on average, before you
see it fuse into deuterium.

200
 Lecture 15 — Why Stars Shine

®® But after that, things start happening faster. Within a few seconds, the
deuterium smacks into another proton and sticks, forming helium-3 and
emitting a gamma ray photon.

®® From there, there are 3 different paths that lead to helium. In the simplest
one, given enough time—about 20,000 years—a helium-3 nucleus will find
another one and collide, knocking away 2 protons. What’s left is 2 protons
and 2 neutrons: that’s helium-4, the end point, which is stable and interacts
no further.

What’s the average power output per cubic meter in the core of the Sun?

The answer, 300 watts per cubic meter, is not very impressive. A kitchen toaster
emits 1000 watts of heat.
This is why we should not envision the Sun as a blast furnace with nuclear bombs
going off all the time. Instead, it’s more like a collection of toasters spaced apart
every few meters.
What makes the Sun so luminous is that it’s so big—there are a lot of toasters.
And, just as important, the overlying material, the outer 90% of the Sun, is
optically thick; it traps the energy for a long time, keeping the interior toasty hot.

201
Lecture 15 — Why Stars Shine

202
Lecture 16

SIMPLE STELLAR
MODELS

A ll stars are powered by nuclear fusion. But despite

this common power source, stars show a wide
range of properties. For stars on the main sequence of
the Hertzsprung-Russell diagram, there are patterns,
and this lecture will consider these relationships.

The biggest star you can see with

your naked eye is Betelgeuse, the
star at Orion’s upper-left shoulder.
It’s about 1000 times larger than
the Sun! If the Sun were swapped
for Betelgeuse, it would engulf all
the planets out to Jupiter.

203
 Lecture 16 — Simple Stellar Models

CENTRAL PRESSURE
®® How long would it take for a sphere of mass M and radius R to collapse under
its own gravity?

®® Consider an atom at the surface. At time 0,

gravity pulls inward, imparting an
acceleration of GM ⁄ R 2. As time goes
on, the atom falls to a radius r (t).
The entire mass is also contracting,
so the mass M pulling on the atom
is always the same. But because
r (t) is decreasing, the acceleration
is increasing.

®® Acceleration is the second time derivative of position, so we need to solve

the second-order differential equation for r (t) and find the time it takes for
the atom to reach r (t) = 0. Instead of doing this, though, we can perform an
order-of-magnitude estimate.

®® Let’s use T to represent the time to collapse, and instead of trying to solve
for r (t), let’s replace the functions on both sides of the equation with simple
expressions that we have reason to believe have the correct orders of magnitude.

®® On the left side, we have acceleration. What’s the typical acceleration of a

falling atom? We know that the average speed must be R ⁄ T because it travels
a distance R in a time T. But acceleration is change in speed.

®® The atom starts from rest and reaches the average speed somewhere along the
way—we don’t know where, exactly, but maybe we won’t be so far off if we
assume it’s at time T ⁄ 2. That implies an inward acceleration of magnitude
R ⁄ T divided by T ⁄ 2, or 2R ⁄ T 2.

®® Meanwhile, on the right side, we have GM ⁄ r 2. Let’s replace that function
with the value it takes at the halfway point: r = R ⁄ 2.
204
 Lecture 16 — Simple Stellar Models

®® By making those substitutions, we convert the differential equation into a

simple algebraic equation, which we can solve for T.

r
⇡ R3
®® If you solve this problem exactly, you get T = , which is only off
2 2GM
by a factor of π ⁄ 2.

®® When we evaluate this free-fall time for a sphere with the Sun’s mass (2 ×10 30
kilograms) and radius (7 × 10 11 meters), we get about half an hour.
s
⇡ R3
⇡ 0.5 hour
2 2GM

®® So, if gravity were the only force acting, the Sun would collapse into a black
hole in half an hour—which is not very long. So how does the Sun survive
for billions of years?

®® There’s a force that opposes gravity: the force that arises from differences
in pressure.

®® The pressure of a gas is the force per unit area produced by all the microscopic
collisions with the gas particles. When gravity tries to compress a sphere of
gas, the contraction brings the particles closer together, increasing the rate of
collisions. The contraction also releases gravitational potential energy, which
heats the gas, making the collisions more energetic.

205
 Lecture 16 — Simple Stellar Models

®® Both effects cause the interior pressure to increase, leading to a net outward
force that eventually becomes strong enough to oppose gravity and halt
the collapse.

®® The equation that expresses that balance between pressure and gravity is
called the equation of hydrostatic balance.

®® Think about water pressure in the ocean. The pressure at the ocean floor
is higher than it is at the surface. So, pressure is not just one number; it’s a
function of height, P (z).

®® Consider a horizontal layer of water at height z with an infinitesimal thickness

of dz and a cross-sectional area A. Let’s calculate the gravitational force and
the pressure forces acting on the water in this layer.

®® There are 2 forces: gravity and pressure. Gravity pulls down with a force
dmg, where g is the gravitational acceleration near the surface of the Earth
and dm is the mass of the infinitesimal layer, which can also be written as
density times volume, ρAdz.

206
 Lecture 16 — Simple Stellar Models

®® The water in the layer feels inward pressure from every direction. The
horizontal forces cancel out, though; the right side gets pushed just as hard
as the left side.

®® But the vertical forces don’t cancel, because pressure decreases with increasing
height. The pressure on the top of the water layer is weaker than the pressure
on the bottom. So, the net effect of pressure (Fp) is an upward force given by
the difference in pressure times the area.

®® If the water isn’t moving—if it’s in balance—then the downward force of

gravity (Fg) must equal the upward force from pressure (Fp).

Fp = Fg

®® This leads to an equation in which the A’s cancel out.

®® This can be rearranged to form the equation of hydrostatic balance.

®® For stars, we need to make a few changes. One is trivial: We replace z with
r, the radial coordinate. More importantly, when we consider a whole star,
the density and the gravitational acceleration are not constants—they vary
from the center to the surface—so ρ becomes ρ(r) and g is replaced by the
more general formula.

207
 Lecture 16 — Simple Stellar Models

®® A subtler point is that M is now a function of r, too, so we write it as Mr. It

represents all the mass located interior to r. That’s from Newton’s theorem:
We only need to consider the interior mass when calculating the force from
a spherically symmetric object. Mr grows from 0 at r = 0 to the total mass of
the star when r equals the stellar radius, R.

®® Now let’s use the equation of hydrostatic balance to estimate the Sun’s central
pressure, Pc. In the spirit of order-of-magnitude estimates, we can replace the
derivative dP ⁄ dr with a simple ratio because the pressure decreases from Pc
to nearly 0 as we go from the center to the surface.

®® And we need to replace the function on the right side with an expression
representative of the function’s value, so let’s replace ρ(r) with the average
density. For r, let’s use half the total radius, and for Mr, let’s use half of the
total mass. That turns the differential equation into an algebraic equation,
which we can solve for Pc.

®® When we plug in the Sun’s mass, radius, and average density of 1.4 grams per
cubic centimeter, we get 5 × 10 14 newtons per square meter—that’s 5 billion
times higher than the air pressure on Earth.

208
 Lecture 16 — Simple Stellar Models

CENTRAL TEMPERATURE
®® Let’s estimate the central temperature, Tc. The temperature of a gas is closely
related to pressure, as encoded in the ideal gas law: P = nkT, where n is the
number of particles per unit volume, k is Boltzmann’s constant, and T is
the temperature.

®® Instead of number density, we will write this in terms of mass density, ρ,

because that’s what appears in the equation of hydrostatic balance. The
relationship is simple: Number density equals mass density divided by the
average mass per particle.

®® The ideal gas law tells us that central pressure equals central density, times
k, times central temperature, divided by mavg. And we also have our order-of-
magnitude expression for central pressure that we obtained from the equation
of hydrostatic balance. Let’s set them equal.

®® That leaves us with an equation where on one side we have the central density
and on the other we have the average density. Those aren’t the same—so we
make an educated guess. The central density is higher than average, perhaps
twice as high. When we replace ρc with 2ρavg, the ρ’s cancel out and we can
solve for Tc.

209
 Lecture 16 — Simple Stellar Models

®® It’s time to plug in some numbers for the Sun. The only thing we don’t already
know is the average mass per particle. For simplicity, let’s say that the Sun is
made of pure hydrogen and it’s hot enough to be ionized, so there are equal
numbers of protons and electrons. Both particles contribute to the number
density, but the electrons have essentially 0 mass compared to the protons, so
the average particle mass is half a proton mass. Plugging that in, along with
the mass and radius of the Sun, we get T = 1.2 × 10 7 Kelvin.

®® That’s the same order of magnitude that is sufficient to ignite hydrogen fusion.

PROPERTIES OF MAIN-SEQUENCE STARS

®® The formula for central temperature is true regardless of the star’s
energy source.

®® For thermonuclear fusion, the reaction rates increase rapidly with temperature.
For each reaction, there’s a sharply defined ignition temperature. An
implication is that we should expect all hydrogen-fusing stars to have the
same central temperature, regardless of the mass or size of the star.

®® Once a star achieves hydrostatic balance, the central temperature always

hovers just above the ignition temperature. That’s why, for stars fusing
hydrogen in their cores, Tc is a constant, independent of M or R . Our
equation now says that M ⁄ R is proportional to a constant, or, equivalently,
R is proportional to M.

210
 Lecture 16 — Simple Stellar Models

®® This is exactly what we observed in our chart of radius versus mass for
main-sequence stars! Now we understand the reason for that pattern: Main-
sequence stars are all fusing hydrogen at their centers.

®® What about the observed pattern between luminosity and mass? The data
tell us that L goes as M  3. How can we understand that?

L ∝ M 3

®® If we make the usual approximation that stars radiate nearly as blackbodies,

then L = 4πR 2σTeff4.

®® How does that depend on mass? We know that R is proportional to M. But

what about Teff ?

®® It’s important to remember that Teff is the temperature of a star’s outer

layers—the photosphere, not the center. On the main sequence, the
central temperatures are all the same—about 10 7 Kelvin—but the effective
temperatures are not. They range from 3000 to 30,000 Kelvin. To understand
the relationship between central and effective temperatures, we need to think
about how the energy flows from the center to the surface.

®® Fusion reactions produce very energetic photons: x-rays and gamma rays. But
they don’t make it far, because the Sun is opaque. It’s optically thick. This is
because the interior is dense and because it’s ionized.

®® Charged particles, especially electrons, interact strongly with photons. So,

the photons are nearly trapped; they transfer their energy to the surrounding
gas, heating it up. Photons are constantly getting scattered, absorbed, and
reradiated with different energies.

211
 Lecture 16 — Simple Stellar Models

®® But the energy doesn’t stay in the core forever. That’s because the temperature
can’t be constant throughout the whole star. The core is superhot, and above
the photosphere is the cold vacuum of space, so temperature must be a
function of r. And heat will naturally flow from high to low temperature—
from the core to the photosphere.

®® The rate of heat flow depends on the temperature gradient. The larger the
difference in temperature between the core and photosphere, the more rapidly
heat will flow. The temperature gradient rises until the rate of heat flowing
upward and escaping is equal to the rate at which energy is being released by
fusion. At that point, the star achieves a steady state and T (r) stops changing.

®® To be more quantitative, we’ll model the heat flow as a diffusion process.

We’ll assume that heat is being transferred in the form of photons, but the
photons can’t travel far; they’re like pinballs, bumping around from particle
to particle.

®® Each time the ball hits a “bumper,” it collides and flies off with the same speed
in a random direction until it hits the next one. After the first collision, it
follows a straight path represented by the vector r1. After the second collision,
it advances by r 2, and so on. After the N th collision, the location of the ball,
r, is the sum of r1, r 2, and so forth, up to rN.

r 1 = r 1 + r 2 + … + rN

®® Now let’s imagine playing a zillion pinball games. What is the average
distance the ball goes after N collisions? To calculate that, let’s square the
equation for r —that is, let’s take the dot product of r with itself. Then, we
expand that product.

212
 Lecture 16 — Simple Stellar Models

®® Interestingly, all the dot products average to 0. That’s because the vectors have
random directions. Any 2 vectors are just as likely to point the same way as
opposite ways, so each dot product is just as likely to be positive as negative,
and the average is 0.

®® The r 2’s don’t average to 0, but they all have the same average value because
the “bumpers” have random locations. The average of each r 2 term will be the
square of ℓ, the mean free path. Because there are N terms, the sum is Nℓ 2.

®® Taking the square root, we find that the distance from the origin—the so-
called root-mean-square distance—is equal to the square root of N times ℓ. In
other words, the distance grows as the square root of the number of collisions.

®® Let’s rewrite that in terms of elapsed time, t. The number of collisions, N,

equals t divided by the mean time between collisions—which is equal to the
mean free path divided by the speed of the pinball (c, because our pinball is
standing in for a photon).

®® The distance grows as the square root of ℓct. Contrast this to the case of a
ball rolling in a straight line without any “bumpers”; the distance would be
proportional to t, not the square root of t. The dependence on the square root
of time is a distinctive feature of diffusion.

®® If the Sun were optically thin—a transparent sphere of radius R —the photons
could fly free and unimpeded from the core to the surface. The time to reach
the photosphere (tfree )—would equal R ⁄ c.

213
 Lecture 16 — Simple Stellar Models

®® For the Sun, that comes out to be 2.3 seconds.

®® But real stars are optically thick, and the photons have to play pinball. To see
how long that takes, let’s take our diffusion formula and set the root-mean-
square distance equal to the stellar radius. Solving for t gives a diffusion
time of R 2 ⁄ ℓc.

®® To evaluate this for the Sun, we need to know ℓ, the mean free path. That’s
not something we’re equipped to calculate from first principles; it’s a job for a
quantum mechanic with a good computer. The outcome of those calculations
is that the Sun’s mean free path averaged over its whole interior is about
1 millimeter.

®® If we insert the Sun’s radius for R, 1 millimeter for ℓ, and the speed of light
for c, we get a diffusion time of 50,000 years.

®® A star’s opacity dramatically slows the progress of photons and spreads out
their arrival times relative to the case of 0 opacity. The ratio of the diffusion
time to the free-flying time is R ⁄ ℓ. For the Sun, that ratio is of order 1 trillion.

214
 Lecture 16 — Simple Stellar Models

®® Now we can understand the connection between the central and effective
temperatures.

®® For the Sun, that’s the fourth root of 10 −12, which is 10 −3.

The Sun’s core is of order 10 7 degrees because that’s when nuclear fusion
becomes possible, and the photosphere is 1000 times cooler because the Sun’s
radius is a trillion times larger than the mean free path.

®® We also have all the tools we need to understand the mass-luminosity relation.

L ∝ M 3

®® This is what we observed to be true for main-sequence stars.

215
Lecture 17

WHITE DWARFS

U nlike ordinary stars, white dwarfs—planet-sized

objects with the mass of the Sun—do not fuse
hydrogen. In fact, they don’t fuse anything; there’s no
internal power source. They’re held up against the force
of gravity by Heisenberg’s uncertainty principle.

DEGENERACY PRESSURE
®® For ordinary stars, what prevents gravitational collapse is gas pressure. The
higher temperature and density toward the center of a star lead to increasing
pressure with depth. The resulting outward force opposes the inward pull of
gravity, allowing the star to achieve hydrostatic balance. That balance can
be sustained because the nuclear reactions keep the center hot; they replenish
the energy that diffuses outward and radiates away into space.

®® When a star runs out of nuclear fuel, the radiated energy isn’t being replenished
anymore. The pressure drops. Hydrostatic balance is lost. The star contracts.

216
 Lecture 17 — White Dwarfs

®® When the core is compressed to a high enough density, quantum theory

becomes relevant. The star comes up against the Pauli exclusion principle:
No 2 electrons can occupy the same exact quantum state—same position,
energy, and angular momentum.

The Pauli exclusion principle is the underlying reason for the most basic fact of
chemistry: When you add an electron to an atom, you can’t put it in the same state as
the other electrons that are already there. Instead, adding electrons leads to wider orbits
and more complex atoms, leading to the rules and patterns governing the periodic table
of the elements.

®® In a white dwarf, the electrons are squeezed so tightly that the Pauli exclusion
principle prevents them from occupying the same location in space. That
creates a pressure—unrelated to temperature—that pushes them apart. It’s
this quantum pressure that keeps gravity at bay.

®® Let’s calculate the order of magnitude of the effect. The Heisenberg uncertainty
principle says that for any particle, the minimum-possible product between
the uncertainties in momentum and position is of order ℏ, which is Planck’s
constant (h) divided by 2π.

( p x)min ⇠ ~

®® Imagine compressing a cold gas to a density of n particles per cubic centimeter.

Each particle is confined in a small cube with volume 1 ⁄ n. The uncertainty
in each coordinate of the particle’s location, ∆ x, has been squeezed down to
1 divided by the cube root of n.

particles 1 cm3
n V =
cm3 n particle

1
x=
n1/3

217
 Lecture 17 — White Dwarfs

®® Heisenberg taught us that the momentum of each particle will not remain 0;
it will inevitably range up to values of ℏ ⁄ ∆ x, or ℏ times the cube root of n.

®® And because momentum is mass times velocity, the velocity will range up to

~ 1/3
v⇠ n
m ,

where m is the mass of each particle. So, at high density, even if the gas is
cold and there’s no energy source, the particles will be moving. That creates
a pressure that can resist further compression.

®® To find out how much pressure, we’ll use a result from lecture 8, when we
derived the ideal gas law (which is P = nkT ): The pressure of a gas is nm times
the average value of v2. The same logic applies here. To order of magnitude,
we can replace the v in the pressure equation with the ∆v that arises from
Heisenberg’s principle, because the particles have speeds that range from 0
to several times ∆v.
~ 1/3
v⇠ n
m
P = nmhv 2 i
✓ ◆2
~ 1/3
Pdeg ⇠ nm n
m
~2 5/3
Pdeg ⇠ n
m

®® This is called degeneracy pressure.

218
 Lecture 17 — White Dwarfs

There are 3 things to notice about the formula for degeneracy pressure.
1 It’s independent of temperature. That’s why degeneracy pressure can
support a white dwarf even when there is no internal energy source to keep
the gas hot.
2 Degeneracy pressure varies inversely with particle mass. That means in a
neutral gas of electrons and ions, it’s the electrons—the low-mass particles—
that produce almost all the degeneracy pressure.
3 Degeneracy pressure varies as the 5/3 power of density. That’s a stronger
dependence than ideal gas pressure, which varies as density to the first
power. So, if we crank up the density, eventually degeneracy pressure will
dominate over gas pressure.

BUILDING A STAR USING

DEGENERACY PRESSURE
®® In the previous lecture, we constructed models for main-sequence stars by
setting the outward force from the pressure gradient equal to the inward
force of gravity. Let’s do the same thing now, but using degeneracy pressure
instead of ideal gas pressure.

®® We need to satisfy the equation of hydrostatic balance:

®® Let’s replace this differential equation with the algebraic approximation we

derived in the previous lecture: the central pressure is of order the average
density times 2GM ⁄ R.

219
 Lecture 17 — White Dwarfs

®® But this time, instead of using the ideal gas law, let’s set Pc equal to the
degeneracy pressure.
~2 5/3 2GM
n ⇠ ⇢avg
m R

®® Because degeneracy pressure varies inversely with mass and electrons are
1800 times less massive than protons, the electrons provide almost all the
degeneracy pressure. So, for m, we’ll insert the electron mass, and for n, we’ll
insert the electron density.

~2 5/3 2GM
n ⇠ ⇢avg
me e R

®® In addition, the mass density, ρ, is dominated by the ions. While the electrons
provide most of the pressure, the ions provide most of the mass. That makes
the relationship between n and ρ subtler than before.

®® Let’s suppose our white dwarf is made of carbon and oxygen. Carbon has 6
electrons, 6 protons, and 6 neutrons, for an atomic mass of 12. Oxygen has
8 electrons and an atomic mass of 16. In both cases, there are twice as many
nucleons as electrons.

®® That means for every 1 electron, there are 2 nucleons. If the electron number
density is n, the mass density is 2mpne, where mp is the proton mass. We can
use that relation to rewrite ne in terms of ρ.

®® Another subtlety is that the left side of the equation refers to the central
density, but the right side refers to the average density. If we make the crude
assumption that the central density is twice the average density, that will
cancel the 2 on the right side of the equation.

220
 Lecture 17 — White Dwarfs

®® Next, let’s bring all the constants over to the left side of the equation and the
stellar properties to the right side. The right side becomes

®® M, R, and ρ are not all independent; ρ is M divided by 4 ⁄ 3πR 3. For simplicity,

let’s just say it’s M ⁄ R 3.

®® Then, the right side simplifies to

®® We can solve for R, leading to an expression for radius as a function of mass.

It looks a little neater if we express the stellar mass in units of the proton mass.

®® There’s something strange about this result: R is inversely proportional to the

cube root of M. In other words, increase the mass and the radius decreases.

®® Although counterintuitive, this is the case because when you add mass,
you increase the gravitational compression. For degeneracy pressure to
compensate, the particle velocities need to increase. And from Heisenberg’s
principle, the way to force ∆v to increase is to make ∆ x decrease—shrink the
spacing between particles. So, you end up with a smaller, denser star.

221
 Lecture 17 — White Dwarfs

®® Let’s run the numbers for an object with the mass of the Sun. But because we
derived this formula with blatant disregard for factors of 2 and π, the exact
answer turns out to be bigger by about a factor of 3. When you make that
correction and plug in all the numerical constants, you get an R of about
5800 kilometers. And studies of large samples of white dwarfs confirm the
amazing truth that the bigger the mass, the smaller the size.

THE ORIGIN OF WHITE DWARFS

®® A star like the Sun has enough hydrogen to last about 10 billion years. At
that point, the core has been converted entirely to helium. Fusion stops, the
gas pressure starts to decrease, and the core contracts.

®® This causes the density to rise and, less obviously, causes the core to heat up.
Gravitational contraction releases gravitational potential energy, which is
converted into heat. Eventually, the core gets dense and hot enough to ignite
helium fusion.

®® At a temperature of around 10 8 Kelvin,

This is a remarkable and helium nuclei have enough kinetic energy to
seemingly paradoxical property collide and make beryllium-8—which is a
of all stars: When they lose very unstable nucleus, but if there are enough
energy, their cores heat up. helium nuclei around, there’s a chance that
(Usually, adding energy heats one of them will hit the beryllium nucleus
something up.) The jargon hard enough to make carbon-12. So, carbon
starts accumulating and sometimes fuses
is that gravitationally bound
with helium to make oxygen-16.
systems, like stars, have a
negative heat capacity.

222
 Lecture 17 — White Dwarfs

®® Meanwhile, strange things are happening higher up in the star. The outer
part of the star swells up to 100 times its usual size—the star becomes a
giant. Eventually, the outer layers rise so high and become so weakly bound
to the star that they get pushed out into space by radiation pressure. The
hot, dense core becomes fully exposed to the universe after billions of years
of being hidden away.

®® By this point, the core is so dense that degeneracy pressure is sufficient to

oppose gravity, so it stops contracting and there’s no way to ignite any further
nuclear reactions. Once the helium runs out, the core becomes an inert ball of
carbon and oxygen. That’s a white dwarf. It starts off white-hot but gradually
cools and fades to black over billions of years. This is the fate that awaits
the Sun.

®® While the white dwarf is still hot, all those ultraviolet and x-ray photons stream
out and light up the surrounding gas. What used to be the star’s outer layers
now form an optically thin cloud, which
displays an emission-line spectrum.
The white dwarf is surrounded
by a wispy, glowing sphere.
It’s called a planetary
nebula (even though
it has nothing to do
with planets).

223
 Lecture 17 — White Dwarfs

WHITE DWARF STRUCTURE

®® There was a flaw in our earlier calculation regarding the theory of white dwarf
structure. An assumption was made that’s not always true.

®® We started by saying that when Heisenberg’s principle starts to become

relevant, the uncertainty in momentum is ℏ times the cube root of the
number density. And because ∆p equals m∆v, then ∆v also varies as n 1 ⁄ 3.

®® This equation implies that if we keep increasing the density, we can make ∆v
as high as we want. But that can’t be true. The universe has a maximum speed:
the speed of light. So, as ∆v approaches c, our formula must break down.

®® Let’s figure out when this problem occurs. We’ll set ∆v equal to, let’s say, half
the speed of light and solve for the critical value of n.

®® To get the corresponding mass density, we multiply by 2 proton masses, which

numerically comes out to be 7.3 million grams per cubic centimeter. Above
that density, we need to take relativity into account.

®® Because density increases with mass, there must be some critical mass for a
white dwarf above which the electrons start moving close to the speed of light.

224
 Lecture 17 — White Dwarfs

®® To calculate that critical mass, we’ll start with the mass-radius relationship
we previously derived.

®® This implies that density is proportional to M ⁄ R 3. And for a white dwarf, R is
proportional to M −1 ⁄ 3. Together these imply that density goes as mass squared.

®® The density of a 1-solar-mass white dwarf is about 3 times lower than the critical
density. To raise the density by a factor of 3, we need to increase the mass by
a factor of 3, or 1.7. We expect the critical mass to be around 1.7 solar masses,
which is not that high. It seems realistic that the core of a star could exceed 1.7
solar masses. To repair our calculation, we need a new formula for degeneracy
pressure for the case in which the particles are moving close to the speed of light.

®® When we derived the ideal gas law in lecture 8, we started by imagining a

gas in which particles move in only 1 direction. Let’s do the same thing,
but this time all the particles move with same speed, c. What pressure do
they exert on the wall? When a particle hits the wall, it reflects, changing its
momentum from p to −p. Because momentum is conserved, the wall absorbs
a momentum of 2p.

®® Now consider the momentum absorbed by an area ∆ A of the wall. In a time

∆t, it absorbs a momentum of

®® The first factor, 2p, is the momentum transferred per collision. The second
factor, n ⁄ 2, is the number density of rightward-moving particles. And the
third factor is the volume of space within which a particle is close enough to
hit the wall within a time ∆t.

225
 Lecture 17 — White Dwarfs

®® Pressure is force per unit area, or momentum per unit time and area, so we
divide by ∆t∆ A, giving pnc, where p is the momentum of each particle.

pressure P = pnc

®® For degenerate particles, we know from Heisenberg’s principle that ∆p is

of order ℏ times the cube root of n. Inserting that for p, we find that the
degeneracy pressure is of order ℏcn 4 ⁄ 3. That’s the formula we need for particles
moving near the speed of light.

®® The nonrelativistic formula went as n 5 ⁄ 3. This new one goes as n 4 ⁄ 3; the
pressure doesn’t increase as rapidly with density. This will be turn out to be
bad news for a massive white dwarf.

®® When you pile more mass onto a white dwarf, increasing the gravitational
force, the only way it can reestablish hydrostatic balance is to become denser.
That way, the particles are pressed closer together and Heisenberg’s principle
leads to higher velocities and higher degeneracy pressure.

®® But above the critical mass, the particle speeds can’t increase any more;
they’re already at the speed of light. The degeneracy pressure starts rising
more slowly with density. The white dwarf has a harder time fending off
gravitational collapse.

®® Let’s see if we can still achieve hydrostatic balance. We’ll take our equation
expressing the condition of hydrostatic balance,

Pc ~ ρ avg 2GM ,
R

and insert the new formula for degeneracy pressure.

226
 Lecture 17 — White Dwarfs

®® We’ll aim straight for the scaling relation between M and R, ignoring all the
2s and πs and neglecting the difference between ρc and ρavg. As before, we’ll
replace the electron number density with ρ ⁄ 2mp, but we won’t bother with
the 2, and then we’ll bring all the physical constants over to one side and the
stellar properties to the other. And because we’re interested in the mass-radius
relation, we’ll replace ρ with M ⁄ R 3.

®® The R s cancel out. There’s no radius anymore in the equation—just

mass. Apparently, there’s only 1 possible mass for which we can achieve
hydrostatic equilibrium.

®® Solving for M leads to an elegant collection of fundamental constants.

Plugging in the values of those constants, the number turns out to be
2.2 × 10 57, and the mass is 1.8 solar masses.

⇠ 1.8 M

®® A more rigorous calculation that keeps track of all the 2s and πs gives a
somewhat smaller answer of 1.4 solar masses. This glorious combination of
fundamental constants and the critical mass of 1.4 is called the Chandrasekhar
limit, named after astrophysicist Subrahmanyan Chandrasekhar.

MCh = 1.4 M

227
 Lecture 17 — White Dwarfs

®® The Chandrasekhar mass means that relativistic degeneracy pressure can

only balance gravity for that 1 specific mass.

®® On this chart of radius versus mass, the curves are exact solutions of the
equation of hydrostatic equilibrium for a cold degenerate gas. One curve
shows the relationship we derived for low-mass white dwarfs: R goes as M −1 ⁄ 3.
But this curve ignores the effects of relativity. When we include relativistic
effects in the theory, we get the other curve.

®® At low masses, the 2 curves are similar, but as the mass approaches the
Chandrasekhar limit, the relativistic curve starts dropping. The radius is
shrinking; gravity is winning. And at 1.4 solar masses, the radius drops to 0!

228
 Lecture 17 — White Dwarfs

What actually happens if the core of a massive star exceeds 1.4 solar masses?
Here’s a quote from the end of Chandrasekhar’s most famous paper:
It must be taken as well established that the life-history of a star of
small mass must be essentially different from the life-history of a star
of large mass. For a star of small mass, the natural white-dwarf stage
is an initial step towards complete extinction. A star of large mass
cannot pass into the white-dwarf stage, and one is left speculating on
other possibilities.

229
Lecture 18

WHEN STARS
GROW OLD

D ark nebulas—clouds of hydrogen and helium, along

with a small percentage of heavier elements—
are the raw material for stars. Dark clouds are stable
for millions or billions of years, but occasionally
something causes part of a cloud to get a little too
dense—maybe from a collision with a different cloud
or a nearby supernova explosion. When that happens,
the increased gravity of the dense region attracts the
surrounding gas, amplifying the density still further.
The cloud begins collapsing. It fragments into clumps
of different sizes, which eventually become stars with
different masses.

230
 Lecture 18 — When Stars Grow Old

THE CORE OF A STAR

®® When examining the core of a star, the 2 properties we need to keep track of
are its temperature and density. To do so, let’s make a logarithmic chart with
central density plotted against central temperature.

®® The Sun’s core has a density of around 100 grams per cubic centimeter and
a temperature of 10 7 Kelvin. That’s the ignition temperature for hydrogen
fusion. We expect all main-sequence stars to have that same core temperature
because they’re all fusing hydrogen into helium. On the chart, the hydrogen-
fusion line is a vertical line going through the Sun. Any main-sequence star
will have core properties falling somewhere on the line.

®® The ignition temperature of helium is higher, about 10 8 Kelvin. That’s

because helium has a higher electric charge than hydrogen, so it takes more
energy to overcome the electrical repulsion and get the nuclei to fuse into
carbon and oxygen. On the chart, any star that’s fusing helium in its core
will appear close to a vertical line at 10 8 Kelvin.

®® Likewise, fusing carbon

and oxygen to make heavier
elements requires an even
hotter temperature, around
10 9 Kelvin. In general,
the heavier the element,
the hotter the ignition
temperature. (We could
keep drawing vertical lines
at higher temperatures for
fusing even heavier elements.)

®® With these landmarks in place, let’s consider the fate of the Sun. It’s happily
fusing hydrogen into helium at 10 7 Kelvin and will stay at that point while
the hydrogen lasts. But what happens when the hydrogen runs out?

231
 Lecture 18 — When Stars Grow Old

®® Without fusion to replenish the energy that the star is radiating away, the core
begins contracting. Its density increases. Contraction also liberates gravitational
potential energy, which is converted to heat, raising the temperature. (Recall
that when gravitationally bound systems lose energy, they heat up.)

®® As the temperature and density rise, the point on the chart that represents the
Sun moves up and to the right. We can figure out the exact trajectory of that
point by using the fact that the star is still very nearly in hydrostatic equilibrium.

®® The reason for this is that the energy loss is so slow that at any instant, the
star is extremely close to equilibrium. The pressure may be changing over
millions of years, but the star’s mass distribution adjusts itself on a timescale
of hours to strike a temporary balance between gravity and pressure. This
kind of situation is called a quasi-equilibrium.

®® As a result, we can calculate a star’s path on the chart by using the equation
of hydrostatic balance. In a previous lecture, we got the result that M ⁄ R
is proportional to central temperature by setting the pressure required for
hydrostatic balance equal to ideal gas pressure. We used this equation to
show that for stars on the main sequence, which all have the same Tc, R is
proportional to M.
M
/ Tc
R

®® Now we’re talking about stars that aren’t fusing hydrogen anymore; there’s
no reason for the central temperature to be a constant. But as long as we’re
near equilibrium, it will still be true that that M ⁄ R is proportional to Tc.

®® Let’s write that in terms of central density instead of radius. We’ll start by
cubing the equation and writing the left side as M  2(M ⁄ R 3)—because that
way, we recognize that second factor as being proportional to density.
M3
/ Tc3
R3
✓ ◆
2 M
M / Tc3
R3
232
 Lecture 18 — When Stars Grow Old

®® Technically, M ⁄ R 3 is proportional to the average density, not necessarily the

central density, but it’s a decent approximation to assume that the central
density is always higher than average by the same factor, in which case we
can rewrite the equation as

M 2 ⇢c / Tc3.

®® Solving for ρ and taking the log, we get a straight line with a slope of 3.

⇢c / Tc3 M 2

log ⇢c = 3 log Tc 2 log M + const.

®® That’s the line the core of the Sun will follow once there’s no hydrogen left—
toward a higher density and a hotter temperature.

233
 Lecture 18 — When Stars Grow Old

STELLAR EVOLUTION AT THE CORE

®® When a star condenses from a glob of galactic gas, it starts out with low
density and low temperature. Gravity draws it together. The core contracts
and heats for millions of years, until the center reaches 10 7 degrees. Then,
the hydrogen ignites. The energy erupting from fusion in the core comes
into balance with the energy being lost up at the surface. The star joins the
main sequence. That’s where the Sun is now, and where it will remain for
another 5 billion years.

®® Then, the core runs out of hydrogen; it’s nearly pure helium. Fusion fizzles
out. So, the core starts contracting again, proceeding farther up a diagonal
line with a slope of 3.

®® The temperature rises until it hits 10 8 degrees. That’s hot enough to ignite
the helium. This new power source halts the contraction and allows the star
to linger a while—maybe for tens of millions of years.

®® The helium-fusing phase doesn’t last as long as the hydrogen-fusing phase

because helium fusion isn’t as efficient, so there’s isn’t as much energy to be
liberated by fusing helium—only a tenth as much energy per unit mass as
from fusing hydrogen.

®® Next, the helium gets used up. The core becomes pure carbon and oxygen.
Fusion fizzles out. Gravitational contraction resumes. The star moves farther
up the diagonal line.

®® If it reaches a billion degrees, the carbon and oxygen ignite, again pressing
the pause button on the collapse. The core fuses heavier and heavier nuclei,
each one lasting for less and less time, burning with increasing desperation.

®® But there’s something important missing from the log chart, something that

will alter the fate of the Sun: degeneracy pressure.

234
 Lecture 18 — When Stars Grow Old

®® The equation governing the Sun’s trajectory—that diagonal line with a slope
of 3—came from assuming that the star is supported by ideal gas pressure. As
the density keeps rising, though, at some point degeneracy pressure becomes
more important than gas pressure.

®® When does that happen? We can find out by setting the equation for
degeneracy pressure equal to the equation for ideal gas pressure. To keep
things simple, we won’t keep track of all the constants; we’ll just keep track
of how ρ depends on T.

®® After some algebra, we see that on the boundary between gas pressure and
degeneracy pressure, ρ varies as T 3 ⁄ 2, which corresponds to a straight line with a
slope of 3 ⁄ 2 on the log chart.
To find the y-intercept, we’d
need to keep track of all the
constants; if you do that, you
find that the boundary line
runs from about 10 2 to 10 8
grams per cubic centimeter
as the temperature runs
from 10 6 to 10 10. Above
that boundary, degeneracy
pressure dominates.

235
 Lecture 18 — When Stars Grow Old

®® However, in the previous lecture, we learned that degeneracy pressure doesn’t

always go as ρ 5 ⁄ 3. When the density approaches the critical density—several
million grams per cubic centimeter—the pressure starts going as ρ 4 ⁄ 3. The
change in behavior is because the electrons are squeezed so tight that they’re
moving close to the speed of light.

®® So, we need to modify the

boundary line. Above the
critical density, the slope of the
boundary is defined by setting
ρ 4 ⁄ 3 to be proportional to ρT,
the ideal gas pressure. In that
case, when we solve for ρ, we
get that it goes as T 3.

®® The slope is twice as high as

before. The effect is to make
a kink in the boundary line at
high density.

®® Now that we’ve established the

domain of degeneracy pressure,
let’s restart our solar-mass star at
low density and temperature. It
starts contracting and works its
way up the line, pausing to fuse
hydrogen and then contracting
until the helium ignites.

®® But now we see that after the helium runs out, it doesn’t advance to carbon
ignition. Contraction brings it into the degeneracy zone, and degeneracy
pressure halts the contraction. Because degeneracy pressure doesn’t depend
on temperature or the energy content of the gas, the core can achieve a stable
and eternal balance between pressure and gravity—no fusion required. The
star becomes a white dwarf.
236
 Lecture 18 — When Stars Grow Old

®® The star is still losing energy through radiation, so the temperature decreases,
meaning the point on the chart moves straight leftward. The white dwarf
cools off and goes dark over billions of years.

®® All of that was for a star of 1

solar mass. Now let’s back up
to the beginning and follow
the progress of a less massive
star—for example, one that
is a tenth of a solar mass.
What changes?

®® The trajectory of a star on

this chart was dictated by
the equation representing
hydrostatic balance.

log ⇢c = 3 log Tc 2 log M + const.

®® The −2log M is part of the

y -intercept; the lower the
mass, the higher the intercept.
So, when we decrease M, the
line shifts upward.

®® Let’s follow that new line.

The star forms with low
density and temperature and
contracts, getting hotter and
denser, but it crosses into the
degeneracy zone even before
hydrogen can ignite. It ends up as a sort of failed star, which cools off into
oblivion—like a white dwarf, but without ever having fused hydrogen. Such
an object is called a brown dwarf, a name that has nothing to do with the
color of the star.
237
 Lecture 18 — When Stars Grow Old

®® So, the fate of an object

of low mass is to become
a brown dwarf or a white
dwarf. What about a high-
mass star—for example,
one that is 10 solar masses?
This time, relative to the
Sun, the trajectory will be
shifted down by 2 units.

®® Shifting down takes us

away from the degeneracy
zone. In fact, because of the kink in the boundary, the trajectory of a 10-solar-
mass star never enters the degeneracy zone. The reason is that the core of
a massive star is hotter at any given density; it must be hotter to generate
enough pressure to maintain equilibrium. The high temperature ensures that
gas pressure overwhelms degeneracy pressure, all the way up to the critical
density and beyond.

®® If we dial the mass back

down, somewhere in
between 1 and 10 solar
masses, there’s a special case
in which the star’s path just
skims the boundary. This
special mass turns out to be
the Chandrasekhar mass—
the maximum-possible
mass of a white dwarf.

®® A star of higher mass will

keep contracting, igniting
heavier and heavier nuclei, burning ever hotter and trying ever more
desperately to prevent gravitational collapse. But it can’t hold out forever.

238
 Lecture 18 — When Stars Grow Old

®® Think about the binding energy

curve. Fusion only releases energy
for nuclei on the left side of the 8

peak. More massive nuclei prefer

6

B/A [MeV]
fission to fusion. So, once you fuse 62
Ni, 58Fe, 56Fe
your way up to iron and nickel, 4 most stable nuclei
at a temperature approaching
2
10 billion Kelvin, nuclear fusion
is spent. 0
0 50 100 150 200 250
A = atomic mass
®® Once the star crosses that
boundary, which is marked in
the log chart with a thick black
line, gravity acts unopposed.
In less than a second, the
core collapses.

®® Another possible source of

pressure is radiation pressure,
which is the pressure arising
from photons, which, like gas
particles, carry momentum and
thereby exert a pressure. But
it obeys a different formula than the ideal gas law: Radiation pressure is
proportional to temperature to the fourth power, a much stronger dependence
on temperature.

4 4
Prad = T
3c

®® This means that even though radiation pressure is negligible in Sunlike stars,
if we make the core hot enough, radiation pressure will dominate. To figure
out where that happens in the chart, we should set the ideal gas pressure equal
to radiation pressure and solve for ρ versus T.

239
 Lecture 18 — When Stars Grow Old

®® The result is that the zone where radiation pressure dominates is in the lower-
right portion of the chart, with a boundary of slope 3. A star supported
by radiation pressure
throughout its interior
turns out to be unstable;
the intense radiation from
such a star would push
itself apart. That helps
explain why we almost
never find stars more
massive than about 100
solar masses. Such stars are
in the radiation zone.

STELLAR EVOLUTION AT THE SURFACE

®® We start the story of a solar-mass star, as before, with a clump of low-density
gas being pulled together by gravity. It starts out large, in the upper right
of a Hertzsprung-Russell diagram, which plots luminosity versus effective
temperature. For about 10 million years, it contracts while maintaining a
nearly constant effective temperature of 3000° or 4000°.

®® Then, it swings to hotter

temperatures. The underlying
reason is that convection
stops being as important—a
nonobvious consequence
of the fact that deep down
inside, hydrogen has begun
fusing. The star reaches the
main sequence. It settles in
for 10 billion years of steady
hydrogen fusion.
240
 Lecture 18 — When Stars Grow Old

®® When the innermost 10% of the star’s mass has

been converted to helium, the star leaves the At its maximum size, the
main sequence. It starts the shell-burning phase, Sun will grow 200 times
in which fusion occurs only within a layer of larger than it is today,
hydrogen surrounding the helium core. Because
engulfing Venus and
the shell is sitting on top of a contracting core,
roasting the Earth. Its
the shell gradually becomes denser and hotter,
causing it pump tons of power into the outer
luminosity will be 2000
layers of the star, causing them to expand—a times higher than it is today.
lot. The star swells up to become a red giant.

®® Meanwhile, the core is still contracting.

So, the star develops a split personality.
It becomes like a star within a star:
a very dense, hot core within a much
larger puffball of cooler gas.

®® Eventually, the core contracts enough

for helium to ignite. In stars like the
Sun, this is an explosive process called
the helium flash—a sudden burst of
fusion, related to the fact that the core
is nearly degenerate when this happens.

®® Then, the star settles in for a steady

phase of helium fusion. Now that
fusion is happening in the core again,
the size and luminosity go back
down. The star takes up a position on
what could be considered the helium
main sequence, officially called
the horizontal branch. It’s like the
hydrogen-burning main sequence but
at higher luminosity.

241
 Lecture 18 — When Stars Grow Old

®® The star hangs out on the horizontal

branch until the core runs out of
helium. Then comes a replay of
the same events as before. The star
starts shell-burning again, this time
with a helium shell surrounding an
inert core of carbon and oxygen.
This causes the star to swell up
and become a giant star for the
second time.

®® At this point, the star also starts

shedding. The radiation pressure
in the outer layers gets intense
enough to push the gas out into
space—a strong wind erupts from
the surface. A star like the Sun
loses about half its mass during
this stage. This means that by the
time nuclear fusion halts, the star
has expelled its outer layers and
revealed the hot, dense core inside:
a white dwarf.

The mass of a star is what determines its fate. A white dwarf can’t exist with a
mass larger than 1.4 solar masses, the Chandrasekhar mass. It would collapse
under its own weight. But because of all the mass shedding that happens during
the giant phases, a white dwarf can emerge from a star that was initially much
larger than the Chandrasekhar mass.
That makes for a complicated relationship between a star’s initial mass and its
ultimate fate.

242
Lecture 18 — When Stars Grow Old

243
QUIZ
LECTURES 13–18

1 For some binaries, we can observe both stars move around the center of mass. For
others, we can see only one star moving and the other star is too faint. Which
properties of the stars can we learn from each type of system? [LECTURE 13]

2 DI Herculis is an eclipsing binary star system with a period of 10.55 days. The
radial velocity amplitudes of the stars are 110.7 and 126.6 km/s. What are the
masses of the 2 stars? [LECTURE 13]

3 Which is more interesting: finding planets that resemble Earth or finding

planetary systems that are fundamentally different from the solar system?
[LECTURE 14]

4 Pretend you are viewing the solar system from 10 pc away in a random direction.
How likely would it be that you could observe transits of Venus? How often
would they occur, and how long would each transit last? How much would the
Sun appear to fade during the transits? [LECTURE 14]

5 What are the useful aspects of a mental model of a nucleus as a cluster of

marbles? What are the main limitations? [LECTURE 15]

6 Suppose, contrary to fact, that the Sun’s observed luminosity came from the
combustion of coal, which releases 30 MJ ⁄ kg. How many kilograms of coal
would need to be burned each second? By what fraction would the Sun’s mass
decrease each year? [LECTURE 15]

244
 Quiz FOR Lectures 13–18

7 The lifetime of an energy-releasing process is equal to the amount of fuel divided

by the rate at which the fuel is burned. How does the lifetime of a main-
sequence star depend on its mass? Which live longer: low-mass or high-mass
stars? [LECTURE 16]

8 Make an order-of-magnitude estimate of the central pressure of the Earth in

N ⁄ m 2. [LECTURE 16]

9 Describe the significance of the Chandrasekhar mass in your own words.

[LECTURE 17]

10 The stars known as 40 Eridanus B and C form a binary with a period of 247.9
years and a semimajor axis of 34.3 AU. The ratio of distances of B and C from
the center of mass is 0.37. The absolute magnitude of star B is fainter than the
Sun by 4.77 units, and its effective temperature is 16,900 K. Find the mass,
radius, and luminosity of star B. [LECTURE 17]

11 Why is the helium-burning phase so much shorter in duration than the

hydrogen-burning phase? Why is the carbon-burning phase even shorter in
duration? [LECTURE 18]

12 Make your own Hertzsprung-Russell diagram with logarithmic axes ranging

from 100,000 to 1000 Kelvin and 10 −4 to 10 6 solar luminosities. Locate the Sun.
Draw straight lines to represent 0.1, 1.0, and 10 solar radii. Sketch the future
evolutionary track of the Sun. [LECTURE 18]

Go to page 338 for solutions.

245
Lecture 19

SUPERNOVAS AND
NEUTRON STARS

T he Magellanic Clouds—2 patches of light that can be

seen on a dark night in the Southern Hemisphere—
are neighboring galaxies caught by the gravitational
pull of our galaxy. On February 22, 1987, someone
pointing a telescope at part of the Large Magellanic
Cloud would have seen a star named Sanduleak −69
202 with only 1 day to live.

246
 Lecture 19 — Supernovas and Neutron Stars

CORE-COLLAPSE SUPERNOVAS
®® Sanduleak −69 202 is a star of nearly 20 solar masses that was shining for 10
million years. Then, the core ran out of hydrogen and started contracting,
surrounded by a shell of burning hydrogen. Soon, the core became hot and
dense enough to fuse helium into carbon.

®® The helium lasted 500,000 years. But eventually the core was completely
converted into carbon and oxygen, surrounded by a shell of burning helium,
which was itself surrounded by a shell of burning hydrogen.

®® The core contracted enough to ignite the carbon, but fusing heavier elements
doesn’t produce as much power. The nuclear binding energy curve starts
flattening out as the peak is approached. So, the supply of carbon lasted only
a few hundred years.

®® The core contracted some more,

igniting heavier elements,
getting ever denser and
hotter, with the ashes
from one reaction
serving as the fuel for
the next one. The core
started to look like an
onion, with concentric
shells of fusion.

®® After just a few years, the core

worked its way up to silicon, forging
it into iron. This last, most desperate phase of fusion lasted only 1 day. Now
the core is nearly all iron, one of the most stable nuclei. Fusing iron doesn’t
release energy; it requires energy. So, there’s no more nuclear power available
to prevent the core from gravitational collapse.

247
 Lecture 19 — Supernovas and Neutron Stars

®® And despite shedding lots of material up at the surface from the star’s radiation
pressure, the core is still more massive than the Chandrasekhar limit of 1.4
solar masses—which means electron degeneracy pressure is powerless to
prevent collapse.

®® So, the floor drops out. All the material in the core falls freely toward the
center. The free-fall time is about half a second for an Earth-sized object.
r
R3
T ⇠
2GM
s
R⊕3
T ⇠ ⇠ 0.5 sec
2G · 5 M

®® But the star didn’t just collapse to

form a black hole. It exploded,
releasing a staggering amount
of energy.

®® This kind of event, a

core-collapse supernova,
only happens about once
per century in a galaxy
like the Milky Way. So,
the astronomers of 1987
were excited to see one so
relatively nearby.

®® They observed it at all possible

wavelengths over many years. This
allowed them to estimate the total energy
released by the explosion in all detectable forms, including light, x-rays, and
the kinetic energy of the shock waves that slammed into the surrounding gas.
They came up with a figure of order 10 44 joules. Astronomers had already
seen lots of core-collapse supernovas in more distant galaxies—not in nearly
as much detail—but the total energy always tends to be of order 10 44 joules.
248
 Lecture 19 — Supernovas and Neutron Stars

Within a few seconds, a core-collapse supernova releases about as much energy

as is radiated by the Sun over billions of years, and it releases even more energy
in a hidden form.

NEUTRINOS AND NEUTRON STARS

®® By 1987, experimental physicists had built a new generation of neutrino
detectors—huge underground vaults of liquid surrounded by equipment that
can spot the tiny flashes of light that happen whenever a high-energy neutrino
makes a rare collision with an atomic nucleus somewhere in the liquid.

®® These new detectors weren’t designed to study neutrinos from the Sun. The
new purpose was to measure the decay of protons, which—if it happens—
would produce neutrinos. Protons are very stable, but some of the particle
physics theories that were in vogue predicted that very rarely a proton
would decay.

®® None of these experiments ever spotted a decaying proton. But some of them
did detect a burst of neutrinos on February 23, 1987. The total number of
neutrino detections that could be attributed to the supernova was 25.

®® After correcting for how many neutrinos weren’t detected because they sailed
right through the Earth, physicists estimated that the total energy of all the
neutrinos produced in the supernova explosion was of order 10 46 joules—100
times more than all the light and heat!

®® This teaches us that we should think of a core-collapse supernova as a neutrino

explosion that, as an incidental by-product, produces a small fireworks
display. Where do all these neutrinos come from?

®® Whenever you see neutrinos, you know the weak nuclear interaction has been
up to something. What happened in Sanduleak −69 202—and what happens
in all core-collapse supernovas—is that the core of the star contracts so much
249
 Lecture 19 — Supernovas and Neutron Stars

that the positive ions and negative electrons are crushed together, packing so
close that even the weak interactions, with their very short range, can mix
them up and transmute them into neutrons.

®® As it turns out, the end point of those transmutations—the most stable

mixture at those high densities—is mainly neutrons, with only a minority
of protons and electrons. The star becomes a neutron star.

®® When nucleons are pressed together tighter than a femtometer, the strong
nuclear interactions become important, preventing further compression.
If we think of a nucleus as a cluster of rigid marbles, a neutron star is a
cluster of 10 57 marbles, and the gravity is so strong that the marbles are
noticeably squished.

®® If the mass of the core of the collapsing star is too large, more than about 2
solar masses, the marbles shatter. The strong force gives way, and the core
collapses all the way to become a black hole—a point with 0 volume endowed
with the mass of the star.

®® But less than 2 solar masses and the strong force is just strong enough to
prevent this fate. That’s when you get a neutron star.

®® We can estimate the size of a neutron star by analogy with our previous
work on white dwarfs. What we did then was set the pressure required for
hydrostatic balance equal to the pressure from electron degeneracy. That led
to a relationship between mass and radius, with radius being proportional to
1 divided by the cube root of mass.

✓ ◆ 1/3
~2 M
Rwd ⇠
G me m2p mp

mn

250
 Lecture 19 — Supernovas and Neutron Stars

®® The formula for the radius has a factor of the electron mass (me) in the
denominator. That traces back to our decision to consider only the electrons
and ignore the degeneracy pressure from the nucleons. This made sense at
the time, because degeneracy pressure varies inversely with particle mass and
a nucleon is 1800 times more massive than an electron.

®® But if most of the electrons are gone because they merged with the protons,
then it’s the neutrons that provide the degeneracy pressure. If we repeated our
calculation for neutron stars, the electron mass in the denominator would be
replaced by the neutron mass (mn).

®® The consequence is that the radius is lower, by a factor of 1800, for a given
mass. We saw that electron degeneracy leads to stars the size of Earth, which
has a radius of 6400 kilometers. Neutron degeneracy should lead to stars the
size of just 3 or 4 kilometers.

®® Although this line of reasoning does give the right order of magnitude, in
reality most of the internal pressure in neutron stars is from the strong nuclear
force, not just degeneracy pressure. That’s something we didn’t have to worry
about for white dwarfs. But for neutron stars, it’s a
big worry, and it makes calculating the mass-radius
relationship difficult.
A neutron star is a city-
®® Most theorists who have tried arrive at a radius in sized object that packs
the neighborhood of 10 kilometers for a neutron star the mass of 1.5 Suns.
with a typical mass of 1.5 solar masses.

®® A neutron star has a comparable mass to a white dwarf but a radius 1000 times
smaller, so the density of a neutron star is higher by a factor of 1000 3, or 10 9.

While a cubic centimeter of a white dwarf has the mass of several metric tons, a cubic
centimeter of a neutron star has the mass of a billion tons.

251
 Lecture 19 — Supernovas and Neutron Stars

SUPERNOVA REMNANTS
®® Once the collapse of a star’s core has been halted by the strong force, there are
still several solar masses of material in the outer layers of the star—material
that is now falling directly onto the surface of the newborn neutron star.
It rebounds from the surface, producing a shock wave that travels out at
thousands of kilometers per second, driving the gas outward.

®® This produces a lot of heat and a burst of nuclear reactions. And these
reactions are not like the steady reactions at the center of the Sun, which
gradually work their way up to heavier elements over billions of years; these
are sudden, violent, uncontrolled reactions that reach all the way up to iron
and even a little beyond.

®® The elements that are forged range across the periodic table from carbon,
nitrogen, and oxygen up to yttrium and zirconium. And the explosion sprays
them all over the galaxy.

®® A neutron star is born at a

temperature of hundreds of
thousands of degrees—so
hot that there’s hope of
detecting its thermal
radiation even though
it’s so tiny.

®® Cassiopeia A is the
fading remnant of a
supernova explosion
in the 17 th century,
a lthough nobody
seems to have noticed
it at the time because it’s
behind some thick clouds of
dust in the Milky Way.
252
 Lecture 19 — Supernovas and Neutron Stars

®® But we can see it clearly in x-ray images, which show all the gas that was
pushed out into space and heated to hundreds of thousands of degrees. And
right in the center is a tiny dot, representing x-rays from the neutron star.

®® The Crab Nebula, 6500 light-years away in the constellation of Taurus, is

a different kind of supernova remnant. The neutron star isn’t just quietly
cooling off. It’s ejecting energetic particles that energize the surrounding
gas, producing a blue-green haze that permeates the nebula, along with lots
of x-rays.

THE DISCOVERY OF NEUTRON STARS

®® The neutron particle was discovered by physicist James Chadwick in 1932. Just
2 years later, astronomers Walter Baade and Fritz Zwicky dreamt up the idea
of a neutron star and proposed that a supernova is the transformation of an
ordinary star into a neutron star. Then, in 1967, neutron stars were detected!

®® But the way in which neutron stars revealed themselves wasn’t anticipated by
anyone and remains to this day a poorly understood phenomenon. They pulse.

®® Jocelyn Bell (now Bell Burnell) and Antony Hewish discovered radio sources
that emit regular pulses, like a clock. The Crab pulsar, for example, emits a
pulse every 33 milliseconds.

®® What we think is happening is that the neutron star is shining narrow beams
of radiation from points on its surface—probably the 2 points along the axis
of the star’s magnetic field. And the magnetic axis need not be aligned with
the rotation axis. So, as the star rotates, the beams of radiation swing around,
like the beam of a lighthouse. If one of those beams happens to sweep across
the Earth, we see a pulsar.

253
 Lecture 19 — Supernovas and Neutron Stars

®® While the Sun only rotates once a month and white dwarfs tend to rotate
about once a day, the Crab pulsar spins around 30 times per second! The
reason neutron stars spin so fast is the conservation of angular momentum.
The angular momentum of a spinning body
is Iω, where ω is the angular frequency—how
many radians per second—and I is the moment Imagine a city-sized
2
of inertia, which is MR  times some constant that sphere, more massive
depends on how the mass is distributed. For a than the Sun, spinning
sphere of uniform density, the constant is 2 ⁄ 5. as fast as the tires of
an Indy 500 race car.
2
L = I! = M R2 !
5

®® Before it collapses, the core has an initial radius of Rinit and spins with angular
velocity ωinit. Then, gravity compresses the core to a radius of R final.
2 2
Rinit !init = Rfinal !final

®® What happens to its angular velocity? We can figure out by appealing to the
conservation of angular momentum. During the collapse, the star doesn’t lose
mass and it’s not getting torqued from outside, so we can equate the initial
and final values of R 2ω. Therefore,
✓ ◆2
!final Rinit
= .
!initial Rfinal

®® Because R shrinks by a factor of 1000, the angular velocity increases by

1000 2, or 1 million.
✓ ◆2
104 km
= ⇠ 106
10 km

®® A rotation period of 1 day becomes a period of 100 milliseconds.

Pinit = 1 day −! Pfinal = 100 ms

254
 Lecture 19 — Supernovas and Neutron Stars

®® If you observe a pulsar long enough, you’ll usually see that the rotation is
slowing down very slightly. Every year, the Crab pulsar’s rotation period
grows by 13 microseconds. If it’s slowing down, it must be losing energy.
Let’s figure out how much.

®® The kinetic energy of a rotating object is

✓ ◆2
1 2 1 2⇡
Ek = I! = I
2 , 2 P

which we can write in terms of the rotation period, P, using ω = 2π ⁄ P.

✓ ◆2
1 2 1 2⇡
Ek = I! = I
2 2 P

®® If P changes, then E must be changing. Let’s take the time derivative to see
the connection. Then, let’s plug in the parameters of the Crab pulsar: M is
1.5 solar masses, R is 10 kilometers, P is 33 milliseconds, and dP ⁄ dt is 13
microseconds per year. After converting units and multiplying through, the
answer is 5 × 10 31 joules per second—which is a good match to observations
of the total luminosity of the Crab Nebula.

®® The rotating pulsar is flinging away high-energy particles from its surface
that fly off and energize the surrounding nebula. The neutron star is acting
like a giant flywheel, a reservoir of rotational kinetic energy that’s being
tapped to light up the surrounding gas. In that sense, the Crab is a rotation-
powered nebula.
255
 Lecture 19 — Supernovas and Neutron Stars

®® Once the period and the period derivative have been measured for a pulsar,
that’s usually enough to predict future pulse times very accurately. In that
sense, pulsars make good clocks.

Some pulsars rival the accuracy of the world’s best atomic clocks.
A category of pulsars called millisecond pulsars spin unusually rapidly even for
neutron stars and are incredibly stable. By timing them, we can measure all
sorts of subtle physical effects, including the slight warpages of space and time
predicted by general relativity.

SUPERNOVA NOMENCLATURE
®® Originally, the word “supernova” simply meant a really energetic explosion—
to be distinguished from a “nova,” a milder but still impressive explosion.

®® Over time, different flavors of supernovas were recognized. A spectrum of

the explosion at its brightest phase sometimes reveals evidence for hydrogen;
others have no hydrogen. Those without hydrogen were called Type I
supernovas, and the ones with hydrogen were called Type II.

®® The Type I supernovas turned out to show some variety, too. Some of them
have lots of silicon in their spectra while others don’t. Type I was subdivided
into Ia, Ib, Ic, and so on, depending on which elements were on display.

®® The most meaningful physical distinction is not between Type I and Type
II, but between Type Ia supernovas and everything else.

®® The transformation of an ordinary star into a neutron star—called a core-

collapse supernova—can be Type II, Ib, or Ic, depending on the details of
how the massive star shed its outer envelope before the core collapsed.
256
 Lecture 19 — Supernovas and Neutron Stars

®® Type Ia supernovas have a fundamentally different origin, one that’s less

well understood. All the theories involve a white dwarf that for some reason
becomes too massive for its own good.

®® A striking thing about Type Ia supernovas is that they all have nearly the
same peak luminosity; they’re much more standardized than core-collapse
supernovas. That’s what makes Type Ia supernovas useful for measuring
distances to faraway galaxies. They function as standard candles, or “standard
explosions.” Measure the maximum flux and infer the distance. This
commonality of Type Ia supernovas might be because the exploding mass is
always nearly the same—the Chandrasekhar mass.

®® Type Ia supernovas also leave behind remnants that are just as impressive as
core-collapse supernovas.

There haven’t been any supernova

explosions detected in our
galaxy since Kepler's
Supernova in 1604. We’re
overdue for one.

257
Lecture 19 — Supernovas and Neutron Stars

258
Lecture 20

GRAVITATIONAL WAVES

O ne way that neutron stars announce themselves

to astronomers is as radio pulsars. Another way is
through x-rays. In the early days of x-ray astronomy, in
the 1970s, some of the strongest sources were found
to be binary stars—or, at least, Doppler spectroscopy
revealed 1 star that was being pulled around by a
massive companion, but there was no light from any
second star, just x-rays.

259
 Lecture 20 — Gravitational Waves

ACCRETION DISKS
®® Imagine an ordinary star and a neutron star in a tight orbit. The normal star
swells up into a giant star and gets distorted by the strong tidal forces from
the neutron star until it violates the Roche limit.

®® Material from its outer layers spills out, but it can’t fall directly onto the
neutron star; it has too much angular momentum. So, it swirls around the
neutron star, forming a vortex, like the water swirling down the drain of a
bathtub. In this context, the vortex is called an accretion disk.

®® Gradually, the material loses angular momentum and falls toward the neutron
star, releasing a lot of gravitational potential energy along the way. The gas
heats up to tens of millions of degrees and glows with x-rays. We can calculate
how much energy is released by following the progress of a small amount of
gas, with mass m, as it spirals down the drain.

260
 Lecture 20 — Gravitational Waves

®® The spiraling is a slow process, so at any moment the gas follows a nearly
circular orbit. If the gas starts at some large distance, where the orbital energy
is practically 0, and descends to the neutron star of radius R, the energy
released is given by the usual formula for orbital energy,

GM m
E=
2R ,

where M is the mass of the neutron star.

In terms of energy,
®® Accretion disks are widespread throughout gravitational accretion
astrophysics. You find them around not just neutron outdoes nuclear fusion
stars but also white dwarfs and black holes. Even
by more than an order
a normal star can accrete mass from a companion,
of magnitude.
though it doesn’t release nearly as much energy
because the material doesn’t have as far to fall.

®® For normal stars and white dwarfs, the accretion disk glows in the visible and
ultraviolet range. For neutron stars and black holes, it’s x-rays.

MERGING PAIRS OF BLACK HOLES

®® Until the early 20 th century, gravity was conceived as a force that one mass
exerts on another. A mass can just reach across space and pull on other masses.
But Einstein taught us that masses don’t directly pull on each other. Instead,
a mass distorts the surrounding space; it curves it, the way a bowling ball
curves the surface of a trampoline.

®® Then, other masses follow the curvature of space instead of just coasting by. If
you roll a marble on the trampoline, it curves around the bowling ball. From
high above, it might look like the bowling ball is pulling on the marble. But
the connection is indirect.

261
 Lecture 20 — Gravitational Waves

®® One implication is that gravity doesn’t act instantly. If a mass changes location,
there’s a slight delay as the stretching of space is transmitted outward. That
allows for the possibility of gravitational waves.

®® If you’re in a large swimming pool and someone does a cannonball dive on the
other side of the pool, it takes a while for the waves to reach you. Likewise,
when black holes crash into each other, there’s a delay before the resulting
distortions in space reach us. But while a water wave might travel a few meters
per second, gravitational waves travel at 300 million meters per second—the
same speed as light.

®® Another difference is what happens when the wave reaches you. If you’re in
the water and a wave reaches you, you bob up and down. But if you’re in space
and a gravitational wave reaches you, you get stretched—first vertically, then
horizontally, then back to vertical, and so on. You feel oscillating tidal forces.

®® The stretching motion is unimaginably small. Even the strongest gravitational

waves that pass through the Earth stretch things by only a few parts in 10 21.

®® But even though gravitational waves are incredibly weak, they have been
detected by a team led by scientists from MIT and Caltech at the Laser
Interferometer Gravitational-Wave Observatory (LIGO). LIGO takes
advantage of the wave nature of photons to sense the tiny changes in length
induced by a gravitational wave. They use electromagnetic waves, from a
laser, to detect gravitational waves.

®® They built tunnels in 2 perpendicular directions; let’s call them north and
west. A gravitational wave changes the relative lengths of the tunnels: One
gets stretched while the other gets scrunched, and vice versa. So, the goal is
to continuously monitor the relative lengths of the tunnels.

262
 Lecture 20 — Gravitational Waves

®® The way it’s done is to send laser light through a beam splitter, essentially a
half-silvered mirror that reflects half the light west and transmits the other
half north. At the end of each tunnel is a highly reflective mirror that bounces
the light back to the beam splitter. From there, the light either goes back
toward the laser or bounces to a photodetector.

®® If the 2 laser beams arrive at the

detector in phase—that is, if each
crest arriving from the west is
met by a crest arriving from the
north—then the beams interfere
constructively. The combined light
has a higher amplitude than either
beam separately, and the detector
registers a large signal.

®® If the 2 optical paths differ by half a wavelength, then the beams interfere
destructively. Each crest from the west is met by a trough from the north and
the signals cancel out—no light gets to the detector.

®® The experimenters arrange for destructive interference and work as hard as

they can to maintain that condition, keeping all the equipment stable. They
call this fringe lock.

®® Then, a gravitational wave comes through, stretching one tunnel and

squishing the other, back and forth. This breaks the fringe lock. Some light
reaches the detector, oscillating in intensity at the same frequency as the
incoming gravitational wave.

®® But the stretching effect is minuscule. The strength of a gravitational wave is

quantified by its strain, defined as the fractional change in length it produces,
∆ L ⁄ L , which is at best 1 part in 10 21.
L
strain =
L

263
 Lecture 20 — Gravitational Waves

®® One thing that helps is to make the tunnels very long—increase L as much as
possible so that ∆ L is also larger and easier to detect. The LIGO tunnels are
4 kilometers long. In addition, there are extra mirrors that bounce the light
back and forth hundreds of times, which has the same effect as lengthening
the tunnels.

®® Another challenge is making sure the mirrors are kept isolated from external
vibrations. The slightest tremor—from a distant earthquake, a passing truck,
or even sound waves—would knock the mirrors out of alignment. So, the
experimenters pump all the air out of the tunnels and use the world’s best
shock absorbers.

®® And even then, even in fringe lock, the detector is never totally dark. There
are always some vibrations unrelated to gravitational waves that simply can’t
be filtered out.

®® The solution for this was to build 2 interferometers at different sites: one in
Hanford, Washington, and the other 3000 kilometers away in Livingston,
Louisiana. The idea is that the random fluctuations will be different at the
2 sites. You only get excited when you see the same signal at both locations
at the same time—or at least nearly the same time.

®® Depending on where the wave is coming from, it will arrive at one site earlier
than the other, but still within 10 microseconds, the time it takes to travel
3000 kilometers at the speed of light.

®® In fact, we can use the measured delay to help figure out where the wave
is coming from, although not very well. We can’t pinpoint the location of
the source on the sky; by measuring the delay, you can only restrict the
possibilities to a certain circle on the sky.

®® The solution for this was to have a third facility, which is called VIRGO and
is located in Italy, build an interferometer. With 3 detectors, the position of
the source on the sky can be triangulated.

264
 Lecture 20 — Gravitational Waves

The Historic First Detection of Gravitational Waves

This chart shows the signal of the gravitational wave measured at LIGO’s Hanford
site in September 2015. The horizontal axis is time. The vertical axis is the strain,
expressed as a multiple of 10 −21.

Before about 0.33 seconds, the data look like random noise. Then, an oscillation
starts to build, gets stronger and more rapid, and then disappears at 0.43
seconds. That signal represents the ripples of gravitational waves that arrived at
the Milky Way from a remote galaxy where billions of years ago a pair of black
holes were spiraling together.
Space was churning furiously as the black holes orbited around one another.
The resulting gravitational waves carried energy away from the system at the
expense of the orbital energy. That made the orbit shrink, causing the black holes
to speed up, which increased the rate of energy loss more. This positive feedback
loop explains the quickening oscillations in the Hanford signal.
Then, the 2 event horizons merged, the 2 black holes became 1, and the
waves stopped.

265
 Lecture 20 — Gravitational Waves

The data from Livingston tell the same story: The signal is consistent with the
Hanford data after accounting for the different locations and orientations of the
2 interferometers.

THE MASSES OF THE BLACK HOLES

®® Another way of viewing the data is a frequency-time diagram, in which the
horizontal axis is still time, but the strain is indicated by color and the vertical
axis is frequency—the rate of oscillations at a given time.

®® Based on the maximum frequency of the signal, we can calculate the masses
of the 2 black holes.

266
 Lecture 20 — Gravitational Waves

®® To calculate the masses accurately requires the machinery of general relativity,

but we can get an approximation with the help of Kepler’s third law, which
allows us to calculate the total mass of a binary if we know both the orbital
period and the orbital separation.

®® To keep things simple, let’s consider a pair of black holes with the same mass
on a circular orbit with period P and separation a.

267
 Lecture 20 — Gravitational Waves

®® The data show that the gravitational wave rose in frequency from 50 to 300
hertz, which corresponds to wave periods ranging from 1 ⁄ 50 to 1 ⁄ 300 of
a second.

®® In our scenario, the wave period is equal to half of the orbital period. That’s
because after half an orbit, the black holes switch places, and because they’re
identical, the system looks the same again. So, the pattern of gravitational
waves must repeat every half an orbital period. We conclude that the orbital
period, P, shrunk to a minimum of 1 ⁄ 150 of a second.

®® What about the orbital separation? We can’t just read that off the chart, but
we can reason that the period was shortest just before the final merger, when
the 2 event horizons started touching each other. At that moment, a was equal
to about twice the Schwarzschild radius.
2 3
®® 2M
Now we can apply Kepler’s third law. The total mass,M is 4⇡ a 2 .
tot, =
G P
✓ ◆3
4⇡ 2 2GM
®® For
2Ma, =
we insert 2 · .
GP 2 c2

®® Now M appears on both sides of the equation.

✓ ◆3
4⇡ 2 2GM
2M = 2·
GP 2 c2

®® We solve for M and then run the numbers, using P = 1 ⁄ 150  of a second.

44 ⇡ 2 G2 M 3
=
P 2 c6
2 c6 P 2
= M2
44 ⇡ 2
G2
p
2 c3 P
M= = 7.6 ⇥ 1031 kg = 38 M
16⇡ G

268
 Lecture 20 — Gravitational Waves

®® The mass comes out to be 38 solar masses. This agrees pretty well with the
more sophisticated analysis by the LIGO team, which found the 2 black hole
masses to be 36 and 29 solar masses.

MAXIMUM STRAIN
®® The other key piece of information from the chart is the maximum strain,
10 −21, which we can use to determine the distance to the galaxy where the
merger took place. But first we need a formula telling us how strain decreases
with distance.

®® To determine distances, we often rely on the flux-luminosity relation:

F= L ,
4π d 2

which we derived based on the conservation of energy: All the power from
the source gets spread out over a giant sphere of surface area 4πd 2. The same
is true of gravitational waves.

®® But LIGO is not a flux detector. The response of an interferometer is not

based on the flux of energy carried by the waves; it responds to the oscillations
themselves. It directly measures the wave amplitude—the strain.

®® Because amplitude varies as the square root of flux, the amplitude of a wave
varies as 1 divided by distance, not distance squared. So, we expect the
formula for the strain of a gravitational wave to be something divided by d.

p 1
amplitude / flux /
d

269
 Lecture 20 — Gravitational Waves

®® And what is that something? That’s a job for a general relativist. But we can
make an educated guess. Gravitational waves are produced whenever a mass
is shaking around, so we might expect the amplitude to depend on the mass
and on how fast it’s shaking.

®® When black holes merge—when their event horizons touch—how fast

are they moving? It turns out they’re moving close to the speed of light, c.
Therefore, in the strain formula, we expect the numerator to depend on M
and c. And because this is gravity, G should also make an appearance.

L (??)
=
L r

®® Now let’s think about units. The formula we’re seeking is for the strain, which
is unitless; it’s a fractional change in length. And the denominator is distance,
so the numerator must also be a distance.

®® To build a quantity with units of distance from M, c, and G, we use GM ⁄ c 2.

That’s proportional to the Schwarzschild radius, which has units of distance.

L ? GM
⇠ 2
L c d

®® In short, we expect the strain to be of order GM ⁄ c 2d. We can use this fact
to calculate the distance the gravitational waves traveled before they reached
the LIGO detectors in September 2015. We solve for d and insert 10 −21 for
the strain and 39 solar masses for M, giving a distance of 1.8 billion parsecs.

38 M

GM/c2
d⇠ ⇠ 1.8 ⇥ 109 pc
L/L
21
10

270
 Lecture 20 — Gravitational Waves

®® A more accurate analysis, based on general relativity,

gives 0.4 billion parsecs. We got the right order The current best
of magnitude. estimate is that a
galaxy like the Milky
®® A billion parsecs is a long way away. This tells us 2 Way hosts 1 merger
things: Mergers between giant black holes must be between giant
rare if the first one we found was so far away, and the black holes every
total energy released in gravitational waves must be 100 million years.
tremendous to be detectable at such a distance.

ENERGY RELEASED
®® We can calculate the energy released in gravitational waves by asking how
much the orbital energy decreases when the black holes start in some wide
orbit and spiral inward until their event horizons touch. The answer is
GM  2 ⁄ 2a, where a, the orbital separation, is twice the Schwarzschild radius.

GM M GM 2
E= =
2a 2 · 2RS The energy released in
gravitational waves is
®® We can simplify that, and when dust settles, the 100 times more than
answer is 1 ⁄ 8Mc 2. The efficiency is 1 ⁄ 8. If we
the energy released in a
plug in 39 solar masses, ∆ E comes out to be about
core-collapse supernova.
10 48 joules.

GM 2 c2
=
4 2GM

1
= M c2 ⇠ 1048 J
8
39 M

271
 Lecture 20 — Gravitational Waves

®® And almost all that energy is released within a tenth of a second, so for that
brief interval, the luminosity is of order 10 49 watts. This can be expressed
as 10 23 times the Sun’s luminosity, or 10 12 times the luminosity of all the
stars in the Milky Way Galaxy. These 2 merging black holes were—for a
moment—emitting more power than a trillion galaxies!

MERGING NEUTRON STARS

®® Another momentous discovery came in August 2017, when LIGO and
VIRGO detected the gravitational waves from a pair of merging neutron stars.

®® In that case, the frequency-time diagram showed a longer signal, lasting 30

seconds instead of just a fraction of a second. That’s because neutron stars
are only around 1 solar mass, instead of 40, and it takes longer for them to
radiate away all their orbital energy.

272
 Lecture 20 — Gravitational Waves

®® Astronomers detected the afterglow of the merger—a new source of light,

x-rays, and radio waves emanating from a distant galaxy.

®® Colliding black holes produce no electromagnetic

radiation; they’re black. But colliding neutron stars It’s now thought that
produce a fireball. And the observations of the almost all the gold in
afterglow confirmed the predicted characteristics of the universe is thrown
the fireball. off like sparks from
colliding neutron
®® The fireball models also predict that the conditions
stars. Each collision
are so extreme that a series of exotic nuclear reactions
takes place—reactions that require a massive dose produces a few Earth
of neutrons—forging many of the heaviest elements masses of gold.
in the periodic table, such as bismuth, thorium,
uranium, and gold.

273
Lecture 21

THE MILKY WAY

AND OTHER GALAXIES

F rom a given point on Earth, we only see a section of

the Milky Way—which our ancestors thought looked
like spilled milk—but we can piece together an all-sky
view, which shows a flattened distribution of stars, like a
disk. When we look along the “galactic equator,” we’re
looking within the plane of the disk at multitudes of
distant stars. When we look above and below the disk,
we hardly see any stars by comparison. We also see a
brightening in the middle of the map. That’s the center
of the galaxy, where the disk thickens into a bulge, a
sort of elliptical blob of stars.

274
Lecture 21 — The Milky Way and Other Galaxies

MILKY WAY
275
 Lecture 21 — The Milky Way and Other Galaxies

At the center of the Milky Way is the black hole Sagittarius A*, with the mass of
4 million suns. It’s surrounded by the stellar bulge, which is a few kiloparsecs
wide and has a mass of 20 billion suns. Then there’s the disk, with a total mass of
around 70 billion suns. The Sun is in the disk about 8 kiloparsecs from the center.

Outside the plane of the disk there aren’t as many stars, but there are some, and
they’re arranged in a more spherical distribution, called the stellar halo. That’s
also where we find the globular clusters, each with up to a million stars packed
into a tight ball.

276
 Lecture 21 — The Milky Way and Other Galaxies

WHAT DOES OUR GALAXY LOOK LIKE?

®® We have no way to obtain a picture of the Milky Way from afar, but we can
look at neighboring galaxies, which are probably like ours. Andromeda is
the nearest big galaxy to ours. It’s “only” 2.5 million light-years away, or 0.8
megaparsecs. The Milky Way probably looks something like Andromeda.

ANDROMEDA GALAXY

277
 Lecture 21 — The Milky Way and Other Galaxies

®® Looking farther away, we can find galaxies that we’re viewing from higher
above the disk, giving us a better view of the spiral patterns of stars, gas,
and dust.

M74

®® Very often, the spiral arms appear bluer than the bulge. That’s because the
disk is where new stars are formed—including blue, massive stars. The bulge
is mainly older stars, so all the blue, massive stars are gone; they already
evolved into red giants. The spiral arms are also dotted with colorful nebulas,
marking the locations of star-forming regions.

278
 Lecture 21 — The Milky Way and Other Galaxies

M81

®® A spectrum of a disk galaxy reveals that The Sun makes a complete orbit
the disk is rotating. You can tell from the of the Milky Way every 200
Doppler effect that half of the galaxy is million years, traveling at about
approaching us, so it’s blueshifted, while 220 kilometers per second.
the other half is receding, so it’s redshifted.

®® In addition to disk galaxies, we find a lot of galaxies that look like featureless
blobs of stars—all bulge and no disk. They’re called elliptical galaxies. The
stars in elliptical galaxies tend to be older and redder than in spirals, and they
don’t all rotate together or show much coordinated motion. Instead, the stars
are orbiting more randomly in all directions.

279
 Lecture 21 — The Milky Way and Other Galaxies

®® Those are clues that elliptical galaxies are older than disk galaxies. They’ve
had more time for stars to age and for their orbits to randomize. Current
thinking is that when 2 disk galaxies collide, the resulting train wreck
becomes an elliptical galaxy.

SPIRAL AND ELLIPTICAL GALAXIES

®® There are elliptical galaxies that are cut through with dark dust lanes, which
are more characteristic of spirals. There are disk galaxies that don’t have any
spiral arms. There are disk galaxies in which the spirals attach to the bulge
through an elongated bar of stars. And there are some spirals that don’t seem
to have any bulges at all.

NGC 5128

280
 Lecture 21 — The Milky Way and Other Galaxies

®® Galaxies have more diverse forms than stars. They’re like the orchids of
the universe. Both their formation and their evolution are more contingent
on circumstances than for stars—and we shouldn’t expect to be able to
understand them with simple equations. But we can make some progress.

®® The first step is to think of a galaxy as a fluid, or a gas, but instead of being
made of atoms or molecules, it’s made of stars. That’s a mind-bending idea.
A star is millions of times more massive than the entire Earth, but galaxies
are trillions of times larger still. So, when we zoom out to galactic scales, we
can treat stars as the microscopic particles of a fluid. In a way, we’re reverting
to the original description of the Milky Way as a stream of milk!

®® The fluidlike nature of galaxies is especially clear in images where 2 galaxies

are interacting with each other. In some cases, the tidal gravity from one
galaxy causes stars to spray out of the other galaxy, forming long, thin
streams, like a fountain.

®® In other cases, the collision of 2 galaxies launches spherical waves in the

pattern of stars, which look like wispy bubbles and curtains.

®® But there are major differences between an ordinary fluid and a fluid of stars.
One of them relates to the mean free path, the average distance a particle
travels before it collides with another one.

®® The mean free path of a nitrogen or oxygen molecule in air is about 0.1
microns. In the Sun, the mean free path of a photon is about a millimeter.
In both cases, the mean free path is much smaller than the scales we’re
usually interested in—that’s why we can assume that the particles rapidly
exchange energy and achieve thermal equilibrium and that each position can
be associated with a well-defined temperature.

®® What about galaxies? How far does a star travel, on average, before it hits
another star?

281
 Lecture 21 — The Milky Way and Other Galaxies

®® The defining criterion for the mean free path is nσℓ = 1, where ℓ is the mean
free path, n is the number density of the particles, and σ is the cross section,
the area of the “target” a particle has to hit for an interaction to occur.

®® In our neighborhood of the Milky Way, the density of stars is about 1 star
per cubic parsec. What about the cross section, σ?

®® Imagine throwing a star at another star of the same radius, R. For a collision,
the centers of the stars must come within a distance of 2R of each other. So,
the target area is a circle of radius 2R, and the collision cross section is π
times 4R 2.

σ = 4⇡R2

®® Using the radius of the Sun, the mean free path comes out to be 2 × 10 14
parsecs.

1 1 14
`= ⇠ 2 ) ⇠ 2 ⇥ 10 pc
nσ (1 pc−3 )(4⇡R

®® The size of the Milky Way is “only” 2 × 10 4 parsecs. According to this
calculation, the Sun could cross the Milky Way 10 billion times before it
hit another star!

®® This calculation is a little unfair, because stars don’t have to physically touch
in order to deflect each other and exchange energy. So, the true cross section
for stellar interactions is larger than 4πR 2. But even when you take that into
account with a more complex calculation, the fact remains that a typical star
in the Milky Way is unlikely to ever hit another star.

®® Stellar collisions only occur in unusually dense groupings, such as globular

clusters, or when they’re caused by effects other than long-range gravitational
forces, such as energy loss due to tidal forces or gravitational radiation.

282
 Lecture 21 — The Milky Way and Other Galaxies

Wouldn’t you love to see a galaxy collision up close?

The Andromeda Galaxy, the nearest big spiral galaxy, is headed straight for the
Milky Way at 110 kilometers per second. Over the next 4 billion years, Andromeda
will loom larger and larger in the sky until the 2 galaxies begin merging into a
single galaxy.
But there’s no reason to be concerned. Even during this galactic pileup, the
probability that the Sun will have a close encounter is extremely low.

IDEAL VERSUS COLLISIONLESS GAS

®® So, we can think of a galaxy as a “gas” of stars—but it’s not an ideal gas. It’s
a collisionless gas. A star’s trajectory is determined not by close encounters
with other stars, which are vanishingly rare, but rather by the combined
gravitational field of the entire galaxy.

®® The spatial distribution of all the stars determines the galaxy’s gravitational
field, which then determines the motions of the stars and changes their
spatial distribution. Everything is tangled up on large scales rather than small
scales. That allows for more complicated behavior than a gas, including all
the wonderful patterns and instabilities we see, such as spiral arms, bulges,
and bars.

®® Another big difference between an ideal gas and a collisionless gas relates
to temperature.

®® The temperature of an ideal gas is defined to be proportional to the average

kinetic energy associated with the random motions of the gas particles. The
average value of 1 ⁄2mv2 is 3 ⁄2kT, where k is Boltzmann’s constant.
⌧ �
1 3
mv 2 = kT
2 2

283
 Lecture 21 — The Milky Way and Other Galaxies

®® We can try to do something similar for stars in a galaxy. We can define the
dynamical temperature (Tdyn) to be proportional to the average of 1 ⁄2mσv2,
where m is the mass of a star and σv is the velocity dispersion, or the spread
in velocities of the stars at a given location.
⌧ �
1 2
m v / Tdyn
2

®® We can measure the velocity dispersion by obtaining a spectrum of all the

starlight from a given location in the galaxy and looking at the width of the
absorption lines—the spread in wavelength. This “blurring” of the absorption
lines is produced by all the different Doppler shifts of individual stars at
that location.

®® The concept of dynamical temperature gives us a different way to think

about galaxies. The stars in both elliptical and disk galaxies have speeds
of hundreds of kilometers per second. The difference is that ellipticals are
dynamically hot: The stars fly around randomly, with a large dispersion.
Disks are dynamically cold: The stars rotate all together, with less variation
in velocity between neighboring stars.

284
 Lecture 21 — The Milky Way and Other Galaxies

®® A spiral galaxy, then, is analogous to a cold, swirling glass of milk, while an

elliptical galaxy is more like a hot puff of smoke.

®® But the analogy with ordinary temperature breaks down if you push it too far.
For example, in a highly elongated elliptical galaxy, the velocity dispersion is
higher when you measure the components of velocity along the long axis as
opposed to the short axis. So, the dynamical temperature at a given location
depends on direction! That’s pretty weird.

®® And with galaxies, there’s no exact equivalent of thermal equilibrium. If you

stir up an ordinary gas, it eventually returns to a steady state of constant
temperature with the velocities of the gas particles approaching a Maxwell-
Boltzmann distribution.

®® But in a galaxy, there is no steady state. In the fullness of time, a system

of gravitationally interacting particles that can exchange energy with its
surroundings will tend to eject some particles and all the rest collapse toward the
center. It’s a mathematical phenomenon called the gravothermal catastrophe.

THE VIRIAL THEOREM

®® Imagine a bunch of particles flying around with different speeds and
trajectories but held together by their mutual gravity so they can’t escape.
The virial theorem says that if you watch the system for a long time, then
on average, regardless of the details, the total kinetic energy will be equal to
−1 ⁄ 2 of the total potential energy.

1
hEk i = hEg i
2
The virial theorem is derived in the video lecture.

285
 Lecture 21 — The Milky Way and Other Galaxies

®® The virial theorem is a mathematical expression of the concept that stars have
a negative heat capacity—that when they lose energy, they heat up. And it
applies to any isolated, gravitationally bound collection of particles, whether
it’s atoms in a star, stars in a galaxy, or a cluster of thousands of galaxies.

®® Let’s say we’re studying an elliptical galaxy. We’d love to know the orbits of
all the stars and how they change with time, but that’s a job for a trained
galactic dynamicist.

®® Thanks to the virial theorem, we know that regardless of the details, the
kinetic energy is −1 ⁄ 2 of the gravitational potential energy on average.

1
hEk i = hEg i
2

®® From that, we can write a useful order-of-magnitude expression. For the

kinetic energy, we’ll write 1 ⁄ 2Mσv2, where M is the total mass of the galaxy
and σv is the velocity dispersion.

®® For the potential energy, we’ll write GM2 ⁄ R.

®® In general, there will also be some numerical factor like 3 ⁄ 5 or π ⁄ 8 that
depends on how the stars are arranged, but our expression captures the right
order of magnitude.

®® Then, we can solve for M, giving σv2R ⁄ G.

®® This is useful because we can measure R based on the galaxy’s angular size
and its distance, and we can measure σv by looking at the widths of the
absorption lines in the spectrum. Then, this equation allows us to estimate
M, the mass of the galaxy—which we otherwise wouldn’t be able to measure.

286
 Lecture 21 — The Milky Way and Other Galaxies

For any gravitationally bound system, if we measure the overall size of the
system and the spread in velocities, we can estimate the total mass.
To do that, we do have to assume that the observed state of the system is
representative of its long-term average. This would be false if we were observing
the initial formation of the system or if it were about to collapse. But subject to
that caveat, the virial theorem can be used to measure the masses of elliptical
galaxies, the bulges of disk galaxies, star clusters, and even the black holes that
reside at the centers of galaxies.

GALAXY DYNAMICS
®® We now know to think of a galaxy as a collisionless gas of stars. The stars
hardly ever have close encounters; they interact through long-range, collective
forces. But what about the interactions of galaxies with each other?

®® The Hubble Space Telescope shows us that when we look in any direction
in the sky, we see galaxies galore—about 1 big galaxy per cubic megaparsec.
So, when we zoom way out to the gigaparsec scale, entire galaxies play the
role of the fundamental particles of the universe. So, should we think of the
universe as a collisionless gas of galaxies?

®® It turns out the answer is no. Galaxies often interact and collide. The typical
spacing between galaxies is about 50 times the size of an individual galaxy.
That’s much more closely packed than the stars within an individual galaxy,
where the spacing between stars is millions of times larger than an individual
star. Calculations of the mean free path show that over a billion years, a
typical galaxy has a few-percent chance of smacking into another galaxy.

287
 Lecture 21 — The Milky Way and Other Galaxies

®® This implies that if you look at hundreds of images of galaxies, you’ll see
some close encounters. And we do. We see pairs of galaxies in the midst of
collisions that last hundreds of millions of years. The tidal force of each galaxy
on the other causes their stars to spill out in curved arcs or to get distorted
into blobby, cometlike shapes.

PORPOISE GALAXY

288
 Lecture 21 — The Milky Way and Other Galaxies

®® In galaxies that look like elliptical galaxies surrounded by a ring of stars, the
ring is probably a spiral galaxy that got too close and the tidal gravitational
forces strung it out into a complete circle—sort of like a galactic-scale version
of the rings of Saturn.

HOAG'S OBJECT

289
 Lecture 21 — The Milky Way and Other Galaxies

®® The bubbles and shells we see around some galaxies are thought to be an
aftereffect of galaxy mergers. The smaller galaxy oscillates inside the bigger
one for a while, causing stars to be ejected in spurts before the 2 galaxies
merge completely.

®® Mergers were ubiquitous early in the history of the universe, when galaxies
first formed. All large galaxies, including ours, formed through the mergers
and accretion of smaller galaxies.

®® Many elliptical galaxies were probably created by the mergers of 2 disk

galaxies. Computer simulations can show the progress of simulated galaxies
over billions of years as 2 disk galaxies approach each other, collide, emit
streams and shells of stars, and then gradually come together and sink in
their combined gravitational well, forming an elliptical.

®® In other words, when 2 dynamically cold galaxies hit each other, the collision
converts all that galactic-scale kinetic energy into the random motions of
individual stars, resulting in a dynamically hot elliptical galaxy. In that sense,
galaxy collisions are like 2 globs of cold milk crashing together, splattering,
and heating up.

290
Lecture 22

DARK MATTER

M 87 is an elliptical galaxy with a tiny blue streak

near its center. A close look shows that there’s
something extraordinary about this galaxy: The center
is glowing brightly, and it’s emitting a beam of light.

M87

291
 Lecture 22 — Dark Matter

ACTIVE GALAXIES
®® M87 is an example of an active galaxy. There’s a supermassive black hole at
the center, just as there is in all big galaxies, but what’s different here is that
the black hole is actively accreting gas. Gas is funneling toward the black
hole. The gas slowly loses energy and angular momentum, causing it to spiral
inward, speed up, and heat up. By the time it’s within a fraction of a parsec
of the black hole, it’s glowing brightly.

®® After it crosses the event horizon, we never see it again—and the black hole
gets slightly more massive. But just before that, at the innermost edge of the
accretion disk, 2 powerful jets of plasma are being launched up and down,
perpendicular to the disk.

292
 Lecture 22 — Dark Matter

®® Why this happens is only dimly understood.

®® One thing that’s clear is the force propelling the jets is electromagnetic.
By this point, the gas is a hot, ionized plasma, and like many plasmas, it’s
prone to developing a tangled internal magnetic field. As the plasma and
the magnetic field approach the black hole, they whirl around in a frenzy,
inducing electric fields that can accelerate charged particles vertically.

®® The electromagnetic field becomes so strong that

electrons and positrons emerge spontaneously Positrons are the antimatter
out of pure energy, a phenomenon called pair equivalent of electrons;
production. All the newborn charged particles
they have the same mass
fly away from the disk at nearly the speed of
but opposite charge.
light. And the underlying energy supply for all
these fields and particles might be the black hole.

®® Black holes can rotate. Accreting black holes absorb so much angular
momentum from the spiraling material that they probably rotate close to
the speed of light. This leads to a relativistic effect in which the surrounding
space starts spinning, too, and if there’s plasma there, the rotational energy
can be converted to magnetic energy.

®® So, the jets of an active galaxy might be rotation-powered, in the same way
that the luminosity of the Crab Nebula comes from the rotation of the
neutron star at its center.

®® This so-called Blandford-Znajek mechanism relies on concepts from plasma

physics and, by itself, doesn’t explain why the ejected particles end up forming
such a narrow beam. To explain that, theorists figure that the accretion disk
is not like a thin, flat plate surrounding the black hole; it’s more like a fat
inner tube, or torus, which restricts the escaping charged particles to narrow
cones surrounding the black hole.

293
 Lecture 22 — Dark Matter

®® There are many different types of active galaxies that go by different names,
depending on how luminous they are and what angle we’re viewing them from.

®® For example, an optical image of Cygnus A shows a galaxy with an irregular

shape. When we look with radio and x-ray telescopes, we see that the galaxy
is surrounded by a haze of x-rays, and at radio wavelengths, we see 2 jets
that puff out into galactic-scale fireballs. This type of active galaxy is called
a radio galaxy.

CYGNUS A

294
 Lecture 22 — Dark Matter

®® Other active galaxies look more

prosaic. This Hubble image shows 2
stars along with some galaxies in the
background. But in fact, the star in
the middle of the image isn’t a star.
It’s an active galaxy that is millions of
times farther away than the star. The
giveaway is its spectrum, which shows
emission lines from the hot gas in the
accretion disk.

®® The accretion disk is 100,000 times more luminous than all the stars of the
galaxy put together. With that bright light shining in our faces, we can’t even see
the galaxy. This kind of active galaxy is called a quasi-stellar object, or quasar.

How can the light from a single accretion disk overwhelm an entire galaxy of
stars? How long could such a beacon possibly shine before running out of energy?
Accretion disks glow because gravity pulls the gas inward, speeding it up,
converting gravitational potential energy into kinetic energy. Then, within the
vortex of material, a lot of that kinetic energy is dissipated as heat.
When a small mass dm falls from far away, the energy released is GMdm ⁄ 2R,
where M is the black hole mass and R is the inner edge of the accretion disk,
which is about 3 times the Schwarzschild radius (RS)—the location of the
innermost stable circular orbit.
To calculate the resulting luminosity, energy per unit time, we divide by the time
interval dt over which the mass is falling and simplify.

295
 Lecture 22 — Dark Matter

Then, we solve for dm ⁄dt, the rate at which the black hole must be fed to produce
a given luminosity.

An entire galaxy has a luminosity of order 10 billion suns, and if the accretion disk
is 100,000 times brighter, then L is 10 15 solar luminosities. The value of dm ⁄dt
required is 3.5 × 10 23 kilograms per second, or 6 solar masses per year.
That’s not so much. Just toss in 1 star every few months and the radiation from
the accretion disk will overwhelm the light of the surrounding 10 billion stars. Such
is the power of gravitational accretion.

®® The Milky Way is not currently an active galaxy. Our black hole, Sagittarius
A*, is in between meals. But all big galaxies, including ours, go through
phases of activity and inactivity, depending on the supply of gas to the central
black hole.

®® There’s evidence that the Milky Way was last active a few million years ago. It’s
based on a gamma ray image of the entire sky made by the Fermi Gamma-ray
Space Telescope. There are 2 regions of excess gamma rays above and below
the center interpreted as the fading “bubbles” of high-energy particles that
were spewed out by relativistic jets during an episode of accretion sometime
during the Pliocene Epoch on Earth.

296
 Lecture 22 — Dark Matter

®® Active galaxies might be part of the explanation for the connection between
the mass of a black hole and the mass of the surrounding galaxy. When a
galaxy grows by accreting a smaller galaxy, the central black hole feasts on
the fresh supply of gas.

®® Then, the powerful jets from the newly activated accretion disk push back on
the infalling gas, expelling it, or at least preventing it from accreting. In this
scenario, galaxy growth is self-regulating: If the galaxy starts growing too fast,
the black hole gets out of control, producing outflows that put the damper on
any further growth. This type of negative feedback between the black hole and
the surrounding galaxy could explain why their properties are linked so tightly.

GALAXY CLUSTERS
®® Galaxies don’t have totally random locations. They’re clustered. If you start in
one galaxy, you’re more likely to find another galaxy within a few megaparsecs
than if you’d started at a random point in the universe.

®® A galaxy cluster is a gravitationally bound system, with each galaxy following

a complex orbit determined by the combined gravitational potential of all
the galaxies in the cluster. And the virial theorem can be used to estimate
the mass of the entire cluster.

®® This was first done by an astronomer named Fritz Zwicky in the 1930s in a
study of the Coma cluster. He obtained spectra of individual galaxies and saw
that some were redshifted and some were blueshifted relative to the average.

®® From the spread in these Doppler velocities, he measured the velocity dispersion
to be 1000 kilometers per second. He also measured the cluster radius, so he
could estimate the mass using the formula from the previous lecture.

2
vR
M⇠ ⇠ 5 ⇥ 1013 M
G

297
 Lecture 22 — Dark Matter

®® The answer he got was about 5 × 10 13 solar masses. The Coma cluster has
about 1000 galaxies, so Zwicky’s calculation implied that the average galaxy
mass is 5 × 10 10 solar masses. But based on the luminosities of the individual
galaxies, his best estimate for the average mass was a few times 10 8.

®® In other words, the mass of the cluster seemed to exceed the sum of the masses
of all the stars inside the galaxies by a factor of 100.

®® This was the first clue that the overwhelming majority of the mass in the
universe is not luminous. There’s something in the cluster that is exerting
gravitational forces on the galaxies, making them move fast. But we don’t
know what. We do know it’s invisible at all wavelengths. And it’s dark. This
is the famous dark matter problem.

®® The numbers have changed since Zwicky’s day; these days we think the dark
matter outweighs normal matter by a factor of 5 or 6. And we have lots of
other evidence for dark matter.

®® Starting in the late 1970s, it became clear that dark matter pervades individual
galaxies, too—not just the space between them.

DARK MATTER
®® For an elliptical galaxy, we can estimate the total mass in 2 ways: by
measuring the velocity dispersion of the stars and using the virial theorem
or by measuring the total luminosity of the galaxy and calculating what total
mass of stars is needed to produce that much light. The virial mass always
exceeds the luminous mass by a large factor.

298
 Lecture 22 — Dark Matter

®® We can perform a similar test with disk galaxies. We perform Doppler

spectroscopy of the starlight at different distances from the center of the
galaxy. That way, we determine the rotation velocity as a function of radius.
That function, V (r), is called the galaxy’s rotation curve.

®® What would we expect the rotation curve to look like? Let’s pretend, contrary
to fact, that the galaxy is a point mass M. Then, the situation is just like a
planet going in a circle around a star: We set the centripetal acceleration equal
to the gravitational acceleration and solve for V, giving the square root of
GM ⁄ r. So, we’d expect the rotation velocity to decline with increasing radius.
r
V2 GM GM
= 2 −! V (r) =
r r r

®® A real galaxy, though, does not have a central dominant point mass; the black
hole is tiny compared to the combined mass of the stars. So, we need to calculate
the gravitational force from the whole distribution of mass within the galaxy.

®® Newton taught us that if we’re at a distance r within a spherical mass

distribution, we’re allowed to ignore all the exterior mass and pretend that
all the interior mass is lumped together as a single point at the center. But
disk galaxies aren’t spheres, and Newton’s theorem doesn’t apply to flat disks.

®® For a disk, the formula for the rotation curve is much messier, so what
physicists tend to do is assume the galaxy is a sphere anyway and trust that
the answer will resemble the right answer even if it differs in detail.

®® We’ll describe the mass distribution with a function, Mr, that tells us how
much total mass is enclosed within a sphere of radius r. As r increases, Mr
rises, too, because we’re enclosing more and more material until we get far
enough away that we’re enclosing all the mass. Then, in the limit of R, Mr
levels off to a constant, the total mass of the galaxy.

lim Mr = Mtot
r!1

299
 Lecture 22 — Dark Matter

®® To find the velocity of an orbiting star, we appeal to Newton’s theorem.

We simply take our previous result and replace the constant M with the
function Mr .
r
GMr
V (r) =
r

®® In the 1970s, measurements of galaxy rotation curves became more accurate

and extended to larger r. Nearly everyone expected that Mr would level off
once r was larger than 10 kiloparsecs, outside the visible disk of stars. That
far away, Mr should stop increasing and the rotation velocity should start
declining as 1 divided by the square root of r.

®® But that’s not what was observed. Instead, V was found to keep rising! In
many galaxies, it levels off to a constant value—but it never decreases, even
well outside the disk!

®® This means that if V is observed to be a constant, then we can solve for Mr

and find that it grows in proportion to r.

V 2r
V (r) = const. −! Mr =
G

®® Way out there, where there are hardly any stars, as we increase r, we still
enclose more and more mass. It’s the dark matter again.

®® More sophisticated analyses using the best-available data show that the dark
matter does in fact form a nearly spherical mass distribution and extends out
to hundreds of kiloparsecs!

®® We are led to the stunning conclusion that a spiral galaxy is just a little bit
of flotsam spinning around at the center of a much larger and more massive
entity, called the dark matter halo.

300
 Lecture 22 — Dark Matter

®® What is the dark matter? That’s one of the most important unanswered
questions in astrophysics. Astronomers and physicists have tried for decades to
detect dark matter in some other way—besides its gravitational influence—
and have failed. Theorists have tried to dream up new forms of matter that
could avoid detection in all these ways.

®® The idea currently in favor is that dark matter is composed of one or more
hitherto unknown fundamental particles—particles that feel and exert
gravity but that otherwise interact very weakly, if at all, with normal matter.

THE HUBBLE DIAGRAM

®® We can measure how fast a galaxy is moving toward or away from us from
its Doppler shift. The fractional shift in wavelength is equal to the radial
velocity divided by the speed of light.

vr
=
c

®® If the wavelengths are being stretched to larger values—toward the red end
of the spectrum—Δλ is positive and the galaxy is moving away from us.
Likewise, if the spectrum is blueshifted, the galaxy is coming toward us.

®® When we do this for the few dozen brightest galaxies in the sky, we find that
the radial velocities are all over the place; some are coming toward us while
others are zooming away. But when we do this for fainter galaxies, a strange
thing happens: They’re almost all redshifted, speeding away from the Milky
Way. This was first discovered by Vesto Slipher in the early 20 th century.

®® But it gets even more interesting when we measure the distance to each galaxy.
That’s much harder.

301
 Lecture 22 — Dark Matter

®® One way to do it relies on Cepheid variable stars, which pulsate, glowing

brighter and then fainter with a regular period. And the period is found to be
closely linked to the luminosity, so if we measure the period, we can calculate
the luminosity. They act as standard candles. We can use the flux-luminosity
relation and solve for the distance, d.
rr
L LL
F = 2
−!
−! dd==
4⇡d 4⇡F
4⇡F

®® But individual Cepheids can’t be detected beyond around 50 megaparsecs.

To reach out farther, we use standard explosions: Type Ia supernovas, which
always seem to explode with the same energy (or at least nearly the same).
They’re predictable enough so that if you measure the color and duration
of the afterglow, you can calculate its luminosity to within a few percent.
And they’re bright enough to see all the way across the universe, billions of
light-years away.

®® When we measure radial

velocities and distances to
lots of galaxies and plot
one against the other, we
see a straight line: V is
proportional to d.

®® A diagram of this type,

published by Edwin
Hubble in 1929, was the
first clear evidence for
the big bang—because
V proportional to d is
what you expect from
an explosion.

302
 Lecture 22 — Dark Matter

®® Imagine a bunch of particles are at a single location at time 0. Then there’s

an explosion, imparting each particle with a random velocity. Some particles
are shot out at high speed; others are shot out at lower speeds. At some later
time, where will each particle be? The higher the speed, the farther it will
get. Distance equals velocity times time, so at any given time, we’d observe
d and V to be proportional.

®® More formally, suppose the i th particle has velocity vi. The distance it travels
between time 0 and the present time, t 0, is vi times t 0. That implies vi equals
ri divided by t 0. Velocity is proportional to distance, and the constant of
proportionality—the slope of the line in Hubble’s diagram—is 1 divided by
the time since the explosion.

®® The slope is called the Hubble constant, H0, and its value is about 70
kilometers per second per megaparsec. Taking the reciprocal and converting
megaparsecs to kilometers and seconds to years gives a value for t 0 of 14 billion
years since the big bang.

®® This agrees well with the best estimate of 13.8 billion years that comes from
more sophisticated analyses—but that’s partly a coincidence. The story isn’t
as simple as galaxies coasting away from a single point. Their velocities need
not be constant; after all, gravity acts to slow the galaxies down and pull
them back together.
303
 Lecture 22 — Dark Matter

Why are the galaxies rushing away from us? Does that mean we’re sitting at the
center of the universe—the site of the big bang?
Even though it might seem like it at first glance, the Hubble diagram doesn’t
imply that there is any sort of privileged galaxy at the center. Edwin Hubble
would have published a similar diagram no matter which galaxy he lived in.
Let’s say you’re inside a lump of raisin bread dough, sitting on a raisin, with other
raisins all around. The oven comes on and the dough expands—the bread rises.
You’ll see all the raisins receding from you, with velocity proportional to distance.
And if you hop over to a different raisin, you’ll still see all the other raisins
receding from you, with velocity proportional to distance. There’s no unique
center of the expansion.

This result is proven more formally in the video lecture.

But if we turn back the clock, the galaxies come closer and closer together.
Doesn’t that mean they’ll all land on a single point?
Not necessarily. For one thing, the theory of general relativity describes the big
bang as the expansion of all of space from a condition of infinite density, not an
explosion that took place at a single location.
In the world of baking, we need to imagine an infinite loaf of raisin bread. If we
reverse the clock and watch the bread “unrise,” the raisins get closer together, but
it’s still infinite in all directions. We just bring in more and more raisins into our
field of view, and the bread gets denser until at time 0 it’s infinitely dense.

304
Lecture 22 — Dark Matter

305
Lecture 23

THE FIRST ATOMS

AND THE FIRST NUCLEI

T he universe began as a collection of fundamental

particles, hot and dense. Space itself was
expanding, causing the density to drop, along with
the temperature. With these premises and knowledge
of fundamental physics, we can make quantitative
predictions about the formation of the first nuclei and
the first atoms that are confirmed through observations
of light element abundances and the cosmic microwave
background radiation.

306
 Lecture 23 — The First Atoms and the First Nuclei

HUBBLE’S LAW AND THE

COSMOLOGICAL SCALE FACTOR
®® The observation that the farther away the galaxy, the faster it’s receding—
Hubble’s law—is one of the pillars of evidence supporting the big bang theory.

®® In equation form, Hubble’s law is v = H0 d, where d is the distance to the

galaxy, v is the velocity with which it’s receding, and the proportionality
constant is the Hubble constant (H0).
1 1
H0 = 70 km s Mpc

®® Hubble’s law implies that v increases without bound as we look farther and
farther away. But that can’t be right. The galaxies can’t exceed the speed of
light, can they?

®® It’s tricky. What we observe directly is not velocity but rather the Doppler
shift of a galaxy’s spectrum.

®® The spectrum of a nearby star has an absorption line at a wavelength of

0.656 microns. That’s the H-alpha line, which comes from electrons jumping
between the second and third energy levels of hydrogen. We call 0.656
microns the rest wavelength because the star isn’t moving very fast relative
to us. It’s the same wavelength we’d observe in a physics laboratory.

307
 Lecture 23 — The First Atoms and the First Nuclei

®® But when we observe the starlight from a galaxy far, far away, the pattern of
lines is shifted to longer wavelengths. Now, the H-alpha line is at, for example,
0.689 microns, 5% longer than usual.

obs rest
z= =
rest

0.689 0.656 v
z= = 0.05 =
0.656 c

®® The redshift, z , is defined as ∆ λ ⁄ λ, the observed wavelength minus the rest
wavelength divided by the rest wavelength. In this case, z is 0.05. And if we
interpret that shift as an ordinary Doppler shift, z is equal to v ⁄ c.

®® An even more distant galaxy shows the H-alpha line at 1.3 microns, which
is twice as large as usual! That would seem to imply that v is twice the speed
of light, contradicting the fundamental principle that nothing can travel
faster than light.

®® The formal resolution of this apparent paradox is that the expanding universe
needs to be described with general relativity, in which the “velocity” of a
distant galaxy is not a meaningful concept. There’s no unique way to define
the relative velocity between 2 objects in widely separated locations when
space is changing with time. So, the velocity cz that we would naively compute
for a distant galaxy has no physical significance.

®® Another way to think about it is that the redshift of a distant galaxy is not
an inherent property of the galaxy; it doesn’t depend on how fast the galaxy
is moving with respect to anything else. Instead, the redshift is something
that happens to photons as they travel from that galaxy to our telescopes. It’s
a consequence of expanding space.

308
 Lecture 23 — The First Atoms and the First Nuclei

®® So, we need a mathematical model for expanding space. For simplicity,

imagine that the universe is 1 dimensional, with all the galaxies strung out
along a line. Let’s use a ruler with some tick marks to keep track of their
locations and put the Milky Way at the origin.

0 1 2 3 4 5 6 7 8

®® Now imagine that our linear universe is expanding. The galaxies stay on the
same tick marks, but the physical distance between the ticks is growing with
time—the ruler is expanding. To emphasize the point, we put a stationary
ruler beneath the expanding ruler.

0 1 2 3 4

0 1 2 3 4 5 6 7 8

®® To describe the expansion, we introduce a function, a (t), called the

cosmological scale factor, defined as the factor by which the tick marks have
expanded by the time t. It’s proportional to the distance between tick marks.

309
 Lecture 23 — The First Atoms and the First Nuclei

®® If the galaxies were all coasting away from each other at constant speeds,
a(t) would be proportional to t. But that’s not necessarily true. For example,
gravity acting alone to pull the galaxies back together would cause a(t) to
vary as t 2 ⁄ 3.

®® But for now, let’s not commit to any specific function; we’ll just leave it as
a(t). And because we can use whatever units we want to measure distances,
we’ll choose to measure the distances between galaxies relative to their current
distances— a = 1 at the current time, t 0. In the past, a was less than 1, and in
the future, a will be greater than 1.

a(t0 ) = 1

®® We can track the position of a galaxy using either the stationary ruler or the
expanding ruler. On the stationary ruler, the galaxy’s coordinate changes
with time; we’ll use r for this physical coordinate.

®® But on the expanding ruler, a galaxy’s coordinate stays the same. Each
galaxy stays close to whichever tick mark it started on. The tick marks on
the expanding ruler are called comoving coordinates; it’s a coordinate system
that expands along with space. We’ll use s for the comoving coordinate to
distinguish it from the physical coordinate, r. The relationship between them
is r = a(t) s.

®® Now let’s see what Hubble’s law looks like in this new language. Consider a
galaxy at a comoving distance of s. At any time t, the physical distance, r, is
a(t) s. The recession velocity, v, is dr ⁄ dt, which is equal to da ⁄ dt times s. To
put that purely in terms of physical distance, we’ll replace s with r ⁄ a.
r = a(t) s
 �
dr da da r
v= = s=
dt dt dt a(t)

310
 Lecture 23 — The First Atoms and the First Nuclei

®® Then, we can rearrange the equation.

✓ ◆
1 da
v= r
a dt

®® That’s Hubble’s law: Velocity is proportional to distance. In our new

language, the constant of proportionality—the Hubble constant—is equal
to (1 ⁄ a)(da ⁄ dt).

®® However, we see now that the Hubble constant need not be a constant in
time; it could be changing, over billions of years, depending on a(t). That’s
why purists refer to (1 ⁄ a)(da ⁄ dt) as the Hubble “parameter,” H, and reserve
the name Hubble “constant” and the symbol H0 for the currently measured
value of 70 kilometers per second per megaparsec.

INTERPRETING GALAXY REDSHIFTS

®® Suppose at some time t in the past that a distant galaxy emits photons that
travel for billions of years through an expanding universe and end up inside
our telescope at time t 0, the present day.

®® It helps to conceptually divide the journey into lots of tiny steps and pretend
there are alien astronomers all along the way who are all observing the light
from that same galaxy.

®® The first alien is at a physical distance of dr from the source—and, crucially,

we’ll let dr be such a short distance that the subtleties of relativity can’t
possibly matter. Because Hubble’s law applies to everybody, the alien observes
the galaxy to be receding with a small velocity, dv = H dr.

311
 Lecture 23 — The First Atoms and the First Nuclei

®® Because the velocity is small, we can rely on the familiar nonrelativistic

formula for the Doppler shift, dλ ⁄ λ = dv ⁄ c, which we can write as

d dv H dr
= = .
c c

®® We learned that H equals (1 ⁄ a)(da ⁄ dt), so let’s make that replacement.

✓ ◆
1 da dr
=
a dt c

®® We can also replace dr ⁄ c by dt, the time it takes for the light to travel a
physical distance dr.
✓ ◆
1 da
= dt
a dt

®® Those replacements lead to a simple equation:

d da
= .
a

®® The fractional change in wavelength equals the fractional change in the scale
factor during the time interval dt.

®® The same logic applies to the second alien down the line, who observes an
additional fractional wavelength shift equal to the fractional change in the
scale factor during the second time interval dt.

®® To calculate the wavelength that we observe at the end of the journey, we

need to integrate all the infinitesimal shifts the photons experience along
the way. On the left side, we integrate from the original wavelength, λrest, to
the observed wavelength. On the right side, we integrate from a(t), the scale
factor back when the light was emitted, to the current value, a = 1.
Z Z 1  ��  ��
obs
d da obs
obs 11
= −!
−! lnln ==lnln
rest a(t) a rest
rest a(t)
a(t)

obs 1
=
rest a(t)
312
 Lecture 23 — The First Atoms and the First Nuclei

®® The wavelength gets stretched by the same factor the universe has expanded
throughout the photon’s journey.

®® This implies that the redshift, z , is equal to 1 ⁄ a − 1. Or, equivalently,

1 ⁄ a = 1 + z.
obs rest 1
z= = 1
rest a(t)

1
=1+z
a(t)

®® This is the interpretation of the redshift we’ve been seeking. When we observe
a galaxy to have a redshift of 2, instead of saying the galaxy is rushing away at
twice the speed of light, it makes more sense to say the universe has expanded
by a factor of 3 since the light was emitted.

THE COSMIC MICROWAVE BACKGROUND

®® In the 1960s, Robert Wilson and Arno Penzias were radio astronomers
employed by Bell Labs, where they had access to a microwave antenna that
was unusually well shielded from terrestrial interference. But despite that
shielding, they observed a persistent source of static and became frustrated.
After painstakingly ruling out equipment problems, they concluded that
space is awash with microwaves.

®® Over time, observations showed that this ever-present radiation amounts to

400 photons per cubic centimeter flying in every direction with the spectrum
of a nearly perfect blackbody.

®® This cosmic microwave background (CMB) radiation is the most perfect

blackbody known. No other light source in nature, or in any laboratory,
matches the Planck function so closely over such a wide range of wavelengths.

313
 Lecture 23 — The First Atoms and the First Nuclei

®® Interestingly, the implied temperature of the CMB that we obtain by fitting

the data with a Planck function is 2.72548 Kelvin, hovering just a few degrees
above absolute zero.

®® What produced all those photons with such a perfect thermal spectrum? Why
are they everywhere? And why is the temperature so cold?

®® In our study of spectroscopy, we learned that blackbody radiation comes from

optically thick materials in thermal equilibrium, the conditions in which
the photons are constantly randomizing their energies through collisions,
absorptions, and emissions by charged particles.

®® But the universe isn’t optically thick! The Earth isn’t suspended in a fog; the
night sky is black. The universe is transparent. Photons can travel for billions
of years without hitting anything, straight from some distant galaxy to our
telescopes. And the universe certainly isn’t all at the same temperature; space
is very cold, and stars are very hot.

®® To make sense of the cosmic blackbody spectrum, we are led to the conclusion
that the universe used to be optically thick—it used to be much denser.
The existence of the CMB is another pillar of evidence supporting the big
bang theory.

®® At early times, the universe was like the interior of a star—just ions and
electrons, everywhere, hot and dense enough to glow with blackbody
radiation, like the inside of a kiln. But then, over time the universe expanded,
the density dropped, and at some point the universe became transparent. All
those photons were still there, but the chance of getting absorbed or scattered
had become negligible.

®® After that, the photons kept sailing along in straight lines, and they’re still
there, billions of years later. But, like all photons propagating across the
expanding universe, their wavelengths have been stretched.

314
 Lecture 23 — The First Atoms and the First Nuclei

®® A blackbody spectrum has an important mathematical property. If you start

with a collection of photons whose wavelengths follow a blackbody spectrum
with temperature T and then stretch the wavelengths of all the photons by
the same factor, a, then the transformed collection of photons will still have
a blackbody spectrum—but with a colder temperature: T ⁄ a.

®® In other words, the expansion of the universe preserves the blackbody

spectrum of the photons, even after there’s no way to maintain thermal
equilibrium—the photons aren’t interacting with anything anymore. But the
temperature of that blackbody spectrum drops in proportion to 1 ⁄ a.

®® That’s why the CMB has such a low temperature today. When the universe
was like the inside of a star, it glowed at visible wavelengths. But since then,
the universe has expanded, stretching the wavelengths from microns into
millimeters—into the microwave region of the electromagnetic spectrum.
Given that the CMB temperature is 2.7 Kelvin today and that it varies as
1 ⁄ a, in general the temperature is 2.7 ⁄ a(t).

2.7 K
T (t) =
a(t)

THE EPOCH OF RECOMBINATION

®® If we were studying water vapor instead of the universe and allowed water
vapor to expand and cool, we know that once it cools to 100° Celsius, it would
condense into liquid, and if it kept cooling to 0°, it would freeze. The universe
has 2 special temperatures, too, and when it cooled to those temperatures,
important things happened.

®® The most recent of these 2 events was the formation of the first atoms. At early
times, the universe was too hot for atoms to exist. Electrons and ions existed
separately, forming a plasma. If an electron did happen to combine with an
ion to form an atom, it was quickly blasted apart by one of the countless
high-energy photons flying around.
315
 Lecture 23 — The First Atoms and the First Nuclei

®® But as the universe cooled, the photon energies dropped. At some point,
the probability for an atom to get ionized by a photon dropped to nearly 0,
which allowed the force of electrical attraction to bring
electrons and ions together to combine permanently.
This time period is called the epoch of recombination.
The universe had
to cool all the way
®® This was also when the universe became transparent,
because photons don’t interact as strongly with neutral down to 3000
atoms as they do with electrons and ions. So, when we Kelvin before
look at the CMB, we’re seeing photons that have been atoms could form.
traveling since the epoch of recombination.

THE EPOCH OF NUCLEOSYNTHESIS

®® When we turn back the clock starting from the epoch of recombination—
with ions, electrons, and light everywhere—the scale factor gets smaller, and
the density and temperature rise.

®® Nuclei are no longer stable. If nucleons

In 1925, Cecilia Payne (later Payne- do come together to make helium or
Gaposchkin) correctly deduced lithium, they get blasted apart by a
that the stars are mostly made of gamma ray arriving soon afterward.
hydrogen and helium. Today we The universe is a plasma of bare protons,
know that this is because of the neutrons, and electrons.
events that took place during the
®® This time period is called the epoch
first half hour after the big bang:
of nucleosynthesis. Only at the end
the epoch of nucleosynthesis. of this epoch was it cool enough for
nuclei heavier than hydrogen to form—
and persist.

316
 Lecture 23 — The First Atoms and the First Nuclei

®® At the beginning of this epoch, the universe

was a mixture of protons, neutrons, and The temperature during the
electrons heated in a bath of gamma rays. epoch of nucleosynthesis,
As it cooled below a few billion Kelvin,
when it was too hot for heavy
the photons lost the power to disintegrate
elements, must have been
nuclei, freeing up the strong nuclear force
billions of degrees.
to bind protons and neutrons together into
heavier elements.

®® But as the universe expanded, the density of protons and neutrons decreased,
making collisions less likely. And the neutrons started disappearing, because
free neutrons spontaneously decay into protons, which repel each other,
making it difficult to make nuclei. At this point, the abundances of different
elements don’t change any more—at least not until much later, when stars
form and start fusing heavier elements at their cores.

®® One of the triumphs of modern cosmology—a third pillar supporting

the big bang theory—is that the results of these calculations match the
measured abundances of the light elements hydrogen, helium, lithium, and
their isotopes.
317
 Lecture 23 — The First Atoms and the First Nuclei

The theory of nucleosynthesis makes a bold prediction: The universe should be

awash with neutrinos that were released by primordial nuclear reactions and
have been sailing through the universe ever since.
In other words, just as there is a cosmic microwave background dating from the
formation of the first atoms, there should be a cosmic neutrino background from
the formation of the first nuclei.
The predicted energies of the cosmic neutrinos are so low that they seem nearly
impossible to detect, but several experimental groups are working toward
that goal.

318
Lecture 24

THE HISTORY
OF THE UNIVERSE

T his lecture will attempt to provide an explanation

of the entire history of the universe. History means
what happened over time, so we need to add the
dimension of time to our discussion. We’ve seen that
a good way to describe the expanding universe is
with the cosmological scale factor, a(t). The distance
between any pair of galaxies grows in proportion to a,
but what equation determines how a varies with time?
It’s called the Friedmann equation, and this lecture will
approach it in stages.

319
 Lecture 24 — The History of the Universe

DESCRIBING THE UNIVERSE

WITH CLASSICAL MECHANICS
®® Let’s assume that the entire universe is an enormous sphere of uniform density.
It’s only if we look closely that we see the specks of dust spread throughout
its volume—those are galaxies. What determines how
an arbitrary galaxy at a distance r from the center
moves with time?

®® The only force is gravity, and from Newton’s

theorem, the gravitational acceleration is
directed inward with a magnitude GMr ,
accel. =of 2
where Mr is the total mass interior to r. r

GMr
accel. =
r2

®® This equation is the same one we solved during our study of planetary motion
and black holes. This case is simpler, though, because there’s just one variable,
r, instead of r and θ. The trajectory is purely radial. So, in this model, the
motion of the galaxy is the same as that of a spaceship near a black hole with
no more fuel and no angular momentum.

®® Even without solving the equation, we can guess what’s going to happen. If
the sphere starts from rest, the galaxy will fall inward. All the interior galaxies
will fall, too, so Mr will remain constant as the sphere contracts.

®® If the initial condition is a Hubble expansion—an expanding sphere with

initial velocity proportional to distance—then gravity will slow it down.
Whether a galaxy eventually gets pulled back or escapes to infinity depends
on how initial speed compares to the escape velocity.
r
2GMr
vesc =
r2

320
 Lecture 24 — The History of the Universe

®® We can visualize the possible outcomes with the graphical method, as we did
for planets and black holes. We start by writing the total energy of the galaxy:
1 GM m
E= mv 2 .
2 r

®® We rearrange that to write the kinetic energy as the difference between the
total energy and the potential energy:
✓ ◆
1 GM m
mv 2 = E .
2 r

®® (Previously, we had another term, L 2 ⁄ 2mr 2, but in this case, L = 0.)

®® Then, we sketch the

potential energy as
a function of r and
make a horizontal
line at the level E .
The square of the
speed at any location
is proportional to the
difference between
the 2 lines: E minus
the potential energy.

®® Suppose E is positive
and a galaxy starts
with an initially outward speed. As the galaxy advances with time, the
difference between E and the potential energy shrinks, so the galaxy slows
down. As r goes to infinity, the potential energy becomes irrelevant and the
speed approaches the square root of 2E ⁄ m. That describes a universe that
expands forever, coasting at a constant speed.
r
2E
E > 0 −! lim v(t) =
t!1 m
expands forever

321
 Lecture 24 — The History of the Universe

®® If the energy is negative, then

the lines cross at a certain
point. In this case, the galaxy
advances to that point, where
it stops, turns around, and falls
toward the origin. This would
be a universe that expands
for a while and then ends up
collapsing into a black hole.

®® In between these 2 cases is

a special case when the total
energy is exactly 0. That’s like a spaceship with an initial speed exactly equal
to the escape velocity. It corresponds to a universe that keeps expanding but
at an ever-decreasing rate.
E = 0 −! lim v(t) = 0
t!1
expands at ever-decreasing rate

®® In this model, the fate of the universe depends on its total energy. If we could
measure the total energy of the universe, we’d be able to determine its fate.
There’s a problem, though. On scales of gigaparsecs, we need to describe the
universe with general relativity, not classical mechanics.

CALCULATING CRITICAL DENSITY

AND THE AGE OF THE UNIVERSE
®® To prepare for general relativity, let’s dress up the energy equation in different
clothing. First, let’s divide through by m, converting everything to units of
energy per unit mass. And because E ⁄ m is a constant, let’s just call it k.
✓ ◆
1 GM m 1 2 E GM
mv 2 = E v = +
2 r 2 m r

322
 Lecture 24 — The History of the Universe

®® Next, let’s bring in the scale factor. Instead of r (t), we’ll write a(t) r 0, where
r 0 is an arbitrarily chosen distance scale—for example, 100 megaparsecs,
the distance from the Milky Way to the Coma galaxy cluster. With that, the
velocity, dr ⁄ dt, becomes da ⁄ dt times r 0.

r(t) = a(t) r0

dr da
v(t) = = r0
dt dt

®® Finally, instead of the enclosed mass, we’ll write the equation in terms of the
density of the universe, ρ(t). We’ll replace Mr with
4⇡(ar0 )3
M= ⇢.
3

®® We make all those substitutions and then tidy up by multiplying both sides
by 2 and dividing by (ar 0) 2.

✓ ◆2
1 da G 4⇡(ar0 )3
r0 =k+ ⇢
2 dt ar0 3
✓ ◆2
1 da 2k 8⇡G
= + ⇢
a dt a2 r02 3

®® The quantity on the left side, (1 ⁄ a)(da ⁄ dt), is the Hubble parameter, H. And
with that change of notation, we have derived the classical Friedmann equation.

2k 8⇡G
H2 = + ⇢
a2 r02 3

®® In this new guise, the equation relates the cosmological scale factor and its
time derivative to the overall density of the universe at any given time. This
allows us to rephrase our statements about the fate of the universe in terms
of its density.

323
 Lecture 24 — The History of the Universe

®® The critical case of 0 total energy corresponds to k = 0. In that case,

8⇡G
k = 0 −! H 2 = ⇢.
3

®® Another way to put it is that there’s a critical density of

3H 2
−! ⇢c = .
8⇡G

®® If the actual density equals the critical density, the universe expands forever
at ever-decreasing speed. If the density is higher, the universe collapses. And
if it’s lower, the universe ends up coasting at constant speed.

®® To figure out what’s going to happen to our universe, we need to measure

the density and compare it to the critical density. The current value of the
critical density is
3H02
−! ⇢c,0 = ,= 9 ⇥ 10 30
g cm 3
8⇡G

where H0 is the Hubble constant, 70 kilometers per second per megaparsec.

Plugging that in gives a critical density of 9 × 10 −30 grams per cubic centimeter.
A more interesting way to express that is 5.5 proton masses per cubic meter.

®® That seems like a pretty low bar for the universe to jump over to achieve the
critical density. But we need to remember that the universe is gigantic, and
most of it is empty space. To measure the average density, we need to assess a
representative volume of the universe that is large enough for entire galaxies
to be like specks of dust.

®® When astronomers did that throughout the 1980s and 1990s, they found
that the universe does have an average density on the order of a few proton
masses per cubic meter. Even the dark matter, it turns out, is very dilute. This
remarkable result suggested that the universe is in that perfectly balanced
state, with 0 total energy.

324
 Lecture 24 — The History of the Universe

®® But as the measurements got better, a density equal to the critical density
was ruled out. The actual average density of matter is only 30% of the
critical density.
⇢0 = 0.30 ⇢c,0

®® Why should the density be of the same order of magnitude as the critical
density but not quite equal?

®® Many theorists suspected that the density really is equal to the critical density
but the measurements were off—maybe astronomers were still missing a lot
of the dark matter. Let’s see where that logic leads.

®® Let’s solve the Friedmann equation and find a(t) for the special case of k = 0.
✓ ◆2
1 da 8⇡G
H2 = = ⇢
a dt 3

®® Both a and ρ are functions of time, but they’re linked by the fact that ρ is
mass over volume, and because the total mass isn’t changing, ρ must be
proportional to 1 ⁄ a 3.
✓ ◆2 ✓ ◆2
1 da 1 da
®® This implies that / , or, equivalently, a is
= aconst.
constant.
a dt a3 dt

®® From there, we take the square root and then integrate to find that a3/2 / t,
or a / t2/3.

®® We’ve just learned that the cosmological scale factor grows with time, but not at
a constant rate; expansion at a constant rate would imply that a is proportional
to t. Gravity decelerates the expansion, making it go as t 2 ⁄ 3 instead.
✓ ◆2/3
t
®® We can write aa =
as .
t0

®® And we can calculate the value of t 0, the current age of the universe, based
on the measured value of the Hubble constant.

325
 Lecture 24 — The History of the Universe

1 da 1 2 t 1/3 1 da 1 2 t 1/3
®® In general, H = ,=which
· inH2/3
this
= case = is · .
a dt a 3 t a dt a 3 t2/3
0 0

®® When we take that derivative and evaluate it at time t 0, the left side is the
Hubble constant, H0.
1/3
1 2 t0 2
H0 = · =
1 3 t2/3 3t0
0

3
®® This means that t0 = = 9.3 ⇥ 109 years .
2H0

®® Plugging in 70 kilometers per second per megaparsec for H0, the age of the
universe comes out to be 9.3 billion years.

®® But there’s a problem: The Sun may be only 5 billion years old, but some
other stars in our galaxy appear to be 13 billion years old. How could stars
be older than the entire universe?

®® These 2 issues—the density not quite equaling the critical density and getting
the wrong age for the universe—are both artifacts of our oversimplified
model. To model the universe correctly, we need to use general relativity. The
relativistic version of the Friedmann equation solves these problems, but in
a shocking and disturbing way.

RELATIVISTIC EFFECTS

In general relativity, when we work out the problem analogous to the expanding
sphere of uniform density, we find 3 features that don’t show up in the
classical Friedmann equation.

326
 Lecture 24 — The History of the Universe

1 The simplest new feature is an extra constant on the right side: Λ/3, where
Λ is called the cosmological constant. It’s a constant of integration that we
get when deriving the Friedmann equation from Einstein’s more general
field equations.

2 A subtler change is that ρ isn’t just mass over volume. Relativity teaches us
that energy and mass are related, E = mc 2, so particles that are essentially
pure energy, like photons and neutrinos, also affect the expansion of the
universe. We have to understand ρ as the density of matter plus the energy
density of all the radiation divided by c 2.

3 The subtlest change is that k, the constant representing energy per unit
mass in our classical model, acquires a deeper interpretation. It specifies the
curvature of space.

®® If k = 0, space is flat, and geometry adheres to its normal principles

(i.e., parallel lines never meet and the sum of the angles in a
triangle is 180°). In a 2-dimensional universe, if k = 0, then
the universe is like an endless flat sheet of paper.

®® But if k isn’t 0, space is curved, and the usual

rules of geometry don’t apply.

↵ + β + γ = 180

327
 Lecture 24 — The History of the Universe

®® If k is less than 0, the universe is curved

like the surface of a sphere, where 2
lines that start out parallel at the
equator eventually cross at the
north pole and a triangle drawn by
connecting 3 points with the shortest-
possible paths has the sum of its angles
as more than 180°.

®® If k is greater than 0, the universe is curved like an

infinite saddle, where parallel lines always diverge and
triangles have angles that sum to less than 180°.

®® If you were trapped on a giant, featureless

surface, how could you tell if you were living
on a flat sheet, a sphere, or a saddle? One way
would be to draw a triangle and measure the
angles to see if they add up to 180 or not.

®® The real universe has 3 dimensions of space, making the

curvature difficult or impossible to visualize, but the logic is
the same. And this experiment has been done over the past few decades. Of
course, nobody actually went around the universe in a rocket ship with a
marker drawing triangles. The experiments are less direct; they’re based on
the cosmic microwave background radiation.

®® From these experiments, the result is clear: k is equal to 0 within a few percent.
Our universe is flat. This implies the universe has exactly the critical density.

®® But this finding seems to contradict the fact that measurements of the total
amount of matter in the universe on the largest scales imply that the density
is only 30% of the critical density. From just those measurements, we would
have expected k to be less than 1 and the universe to be curved like a saddle.

328
 Lecture 24 — The History of the Universe

®® Let’s set aside that contradiction for now.

8⇡G ⇤
®® If k really is 0, the Friedmann equation simplifies to H 2 = ⇢+ .
3 3

®® Physically, that constant term, Λ ⁄ 3, doesn’t make sense. The best way to
see that is to factor out 8πG so that the Λ term is being added directly to ρ.
✓ ◆
8⇡G ⇤
H2 = ⇢+
3 8⇡G

®® This makes clear that a constant value of Λ has the same effect in the equation
as a density that is constant in time.

®® This is crazy, because the density of anything should decrease as the universe
expands. If a grows by a factor of 2, then the mass density decreases by a
factor of 8. But the Λ term would stay the same, like a type of mass that can’t
be diluted—as if each new cubic centimeter of the expanding universe came
into existence filled with new mass, or energy.

®® It seems like Λ is a purely mathematical

artifact. Any “reasonable” universe Einstein originally thought Λ was a
must have Λ = 0—in which case the negative number so that it would
Friedmann equation becomes the same cancel out ρ and zero out the right
as the classical equation.
side of the equation. But that’s
because he was working before
8⇡G
−!H 2 =
⇤⇤==00 −!
3
⇢ Hubble. Einstein thought that the
universe was stationary, so he
®® We’re back to the same equation that we wanted da ⁄dt to equal 0.
started with! It seems like none of the Years later, after Einstein learned
relativistic effects matter.
about the evidence for an expanding
universe, he regretted this decision.

329
 Lecture 24 — The History of the Universe

THE MYSTERY OF DARK ENERGY

®® In the mid-1990s, astronomers started getting good at discovering Type
Ia supernovas in distant galaxies. These act like standard candles: We can
determine the distance based on the flux-luminosity relationship, and we can
measure the redshift of the galaxy where the supernova took place. Each new
supernova adds a new data point to the Hubble diagram—and astronomers
were adding them at ever-greater distances and higher redshifts.

®® The distance, d, divided by c tells us how much time has elapsed since the
supernova went off. And the redshift, z , tells us the value of the cosmological
scale factor at that time.
d
t=
c
1
=1+z
a

®® So, we can convert the redshift-distance data into a chart of a versus t. Our
solution of the Friedmann equation said that a should be growing like t 2 ⁄ 3.
We can plot that curve.

®® We can also plot some

other cases. If gravity
were irrelevant and
the universe were just
coasting along, we’d see
a is proportional to t.
And if the density were
higher than the critical
density, the universe
would recollapse.

®® However, none of these models turned out to fit supernova data; the data
points for supernovas are higher than any of the curves.

330
 Lecture 24 — The History of the Universe

®® If we want connect the data points to the present day, when a = 1 and the
slope is H0, we need to draw a curve that bends upward. The scale factor is
not just increasing; the rate of increase has grown with time. In other words,
the expansion of the universe is accelerating.

®® That seems absurd. It’s easy to understand why the expansion rate might
slow down, from the attraction of gravity. But to speed it up, you’d need
some kind of antigravity! It doesn’t make much sense, but that’s where the
data have led us.

®® This is the top unsolved problem in astrophysics. Cosmologists, astronomers,

and particle physicists have united in the effort to understand what’s going
on. The phenomenon—the force, or substance, that propels the expansion of
the universe with ever-increasing speed—has become known as dark energy,
in analogy with dark matter. But it’s just a label; its true nature is unknown.

®® What might have something to do with it is Λ, that integration constant we

so casually discarded on the advice of Albert Einstein. So, Λ might have a
physical meaning after all.

®® Let’s return to the Friedmann equation, but this time we’ll retain Λ and see
what happens.

®® In an expanding universe, as time goes on, ρ decreases. Matter and radiation

get diluted. But Λ, being constant, persists. Eventually, we reach a point at
which we can neglect the ρ term altogether.
✓ ◆2
1 da 8⇡G ⇤
H2 = = ⇢+
a dt 3 3
✓ ◆2
1 da ⇤
⇢ = 0 −! =
a dt 3

331
 Lecture 24 — The History of the Universe

®® This implies that is a constant—an equation for which the solution is

exponential growth.
r
1 da ⇤
=
a dt 3
p
a = e ⇤/3(t t0 )

®® That could be what we’re observing today: a transition in the history of the
universe when the gravitational attraction of ordinary matter has been overcome
by a universal repulsive force, represented by the cosmological constant.

®® Fitting the supernova data to an upwardly bending line also has the effect of
increasing the age of the universe; the curve doesn’t cross a = 0 until 14 billion
years ago. So, the cosmological constant solves the problem of the stars that
appeared to be older than the universe.

®® It also explains how the universe can be flat even though the density of matter
is less than the critical density. We previously found that in the Friedmann
equation, Λ ⁄ 8πG acts like a density that gets added to ρ. According to the
data, ρ is 30% of the critical density and the Λ term makes up the other 70%.
✓ ◆
8⇡G ⇤
H2 = ⇢+
3 8⇡G

®® Even though we don’t understand dark energy, once we invoke it, everything
fits together nicely.

332
 Lecture 24 — The History of the Universe

This logarithmic chart of a versus t is what results when we solve the Friedmann
equation, including the effects of matter, radiation, and dark energy—all at
the numerical levels
consistent with
data from Type Ia
supernovas, the cosmic
microwave background,
and numerous
other sources.

333
QUIZ
LECTURES 19–24

1 What would be the effects on society if a nearby supernova occurred in modern

times that was bright enough to be seen in the daytime? [LECTURE 19]

2 Suppose a carbon white dwarf accretes material from a companion until it

reaches the Chandrasekhar mass, triggering a nuclear explosion in which all
the carbon is converted to iron. How much energy is released? How does this
compare with the gravitational binding energy of a typical white dwarf? Note:
The atomic masses of carbon-12 and iron-56 are 12.0000 and 55.9349 times
the mass of a proton. [LECTURE 19]

3 Does the detection of gravitational waves leave any room for doubt about the
existence of black holes? What further proof would be needed? [LECTURE 20]

4 Check the latest news for discoveries from the LIGO and VIRGO projects. How
many black hole mergers have been detected? How many neutron star mergers?
Have there been any other types of detections, such as a merger between a black
hole and a neutron star, or a supernova? [LECTURE 20]

5 Look up the following galaxies: Whirlpool, Triangulum, Messier 63, Messier 94,
UGC 12591, the Cartwheel, the Sombrero, Centaurus A, NGC 3370, and NGC
4038. Which is your favorite? A good reference is the Astronomy Picture of the
Day (https://apod.nasa.gov). [LECTURE 21]

6 The radius of influence of a black hole is defined as GM ⁄ σ 2, where M is the

mass of the black hole and σ is the velocity dispersion of the bulge of stars within
which the black hole resides. Why does this definition make sense? What is the
radius of influence of a black hole with the mass of 10 8 suns in a bulge with
σ = 160 km ⁄ s? [LECTURE 21]

334
 Quiz FOR Lectures 19 –24

7 Look up the Bullet cluster, which is often cited as evidence that dark matter
is real and that the problem is not with the law of gravity. What is the basic
argument, and is it convincing? [LECTURE 22]

8 For a galaxy with a flat rotation curve (i.e., a rotation velocity independent of
distance), how does the mass density vary with radius? [LECTURE 22]

9 Try to think of some other explanation for the observation that the Earth is
surrounded by a blackbody spectrum of photons with a temperature of 2.7 K.
How might these other possibilities be tested or ruled out? [LECTURE 23]

10 Pair production refers to the process by which a photon can spontaneously

transform into an electron and a positron. For this to occur, the photon must
have an exceeding 2mec 2, where me is the mass of an electron. By what factor
has the universe expanded since it was hot enough for the average photon to
undergo pair production? [LECTURE 23]

11 Prior to 1998, it seemed to many scientists that a flat, matter-dominated universe

was the most beautiful possible cosmological model because of its simplicity.
Can the modern cosmological model be considered beautiful? Should we expect
human aesthetics to be a reliable guide to the fundamental properties of the
universe? [LECTURE 24]

12 The energy density of radiation varies as 1 ⁄ a 4, rather than 1 ⁄ a 3. (The extra factor
of 1 ⁄ a is because the wavelengths are stretched in proportion to a.) Solve the
Friedmann equation to find a(t) for a flat universe dominated by radiation. What
is the implied age of the universe, given the measured value of H0? [LECTURE 24]

Go to page 339 for solutions.

335
 QUIZ SOLUTIONS

LECTURES 1–6

1 44 km and 0.9 m ⁄ sec.

2 In general, your weight is proportional to M ⁄ R 2. Constant density implies R is

proportional to M 1 ⁄ 3, and therefore your weight, W, is proportional to M ⁄ (M 2 ⁄ 3),
or M −1 ⁄ 3.

3 Answers will vary.

4 The final mass divided by the initial mass is 2 n, where n is the number of
days. The ratio of the mass of the Milky Way to the mass of an electron is
r = 2.2 × 10 72. For this to equal 2n, we need n = log2(r) = log(r) ⁄ log(2) = 240 days.

5 Professor Winn tried this and found his eyes to have an angular resolution of
approximately 50 arc seconds, with corrective lenses.

6 The difference is that the star emits light while the asteroid reflects sunlight. The
flux of sunlight reaching the asteroid varies as 1 ⁄ r 2, and the fraction of that light
reaching the Earth varies approximately as 1 ⁄ r 2, giving a net dependence of 1 ⁄ r 4.

7 Answers will vary.

8 17.2 Earth masses.

9 The energy and angular momentum increase. The orbit becomes elliptical,
with the location of the rocket burn becoming the distance of closest approach.

336
 QUIZ SOLUTIONS

10 From Kepler’s third law, the semimajor axis is 17.834 AU. The distance of closest
approach is a(1 − e) = 0.59 AU.

11 Answers will vary.

12 Pmin depends only on the density of the orbiting body. It is approximately 12.6
hours divided by the square root of density, expressed in g ⁄ cm 3.

Click here to go back to the quiz.

LECTURES 7–12
1 Answers will vary.

2 By setting the Roche radius equal to the Schwarzschild radius, the limiting black
hole mass is found to be 320 million solar masses.

3 L is proportional to R 2T  4, giving about 25.

4 About 10 µm and 845 W for a typical human. Note, though, that your body
also absorbs radiation from the ambient air.

5 Answers will vary.

6 Assuming the planet absorbs all the incident sunlight and radiates in all directions
equally, the surface temperature is (5777 K)(R ⁄ 2a) 1 ⁄ 2. The corresponding
habitable zone is from 0.56 to 1.04 AU. In reality, the atmosphere will lead to a
higher surface temperature at a given orbital distance. The planet’s reflectivity
also plays a role.

7 Answers will vary.

337
 QUIZ SOLUTIONS

8 The angular diameter of the stellar image is 1 arc second. The star’s image is
a circle of radius 0.80 square arc seconds, containing about 800 photons from
the sky. Therefore, the signal is 300 and the noise is  1100, giving a signal-to-
noise ratio of 9.

9 The atmosphere absorbs ultraviolet radiation. At 100 MHz, terrestrial

interference from FM radio is severe.

10 Divide the shortest wavelength by twice the maximum baseline, giving 9.4
nanoradians, or 0.002 arc seconds.

11 The star would be limb-brightened, instead of limb-darkened, and would show

an emission-line spectrum.

12 Venus, human body, incandescent lamp, oven, freezer, lava. Probably not the
lightsaber, although Kylo Ren’s lightsaber may be an exception.

Click here to go back to the quiz.

LECTURES 13–18
1 When we can track the motion of both stars, we learn their individual masses.
When only one star is visible, we can only learn the total mass.

2 5.15 and 4.52 solar masses.

3 Answers will vary.

4 The transit probability is 0.65%. Transits would occur every 225 days and
last a maximum of 11 hours. The Sun would appear to get fainter by 75 parts
per million.

338
 QUIZ SOLUTIONS

5 Answers will vary.

6 1.3 × 10 19 kg ⁄ sec and 0.00021.

7 Lifetime is proportional to M ⁄ L . For a main-sequence star, L is proportional to

M  3, so lifetime is proportional to 1 ⁄ M  2. Less massive stars live longer.

8 Evaluating the equation of hydrostatic balance for a sphere of uniform density

gives an estimate of 2.7 × 10 11 N ⁄ m 2. More sophisticated modeling gives a
density of 3.6 × 10 11 N ⁄ m 2.

9 Answers will vary.

10 0.47 solar masses, 0.013 solar radii, and 0.012 solar luminosities.

11 There are 2 reasons: The efficiency of burning decreases as the atomic mass of
the fuel approaches that of iron, and the stellar luminosity rises during the later
phases of evolution.

12 Answers will vary.

Click here to go back to the quiz.

LECTURES 19–24
1 Answers will vary.

2 The explosion releases 3 × 10 44 joules, as compared to 3 ⁄ 5 GM  2 ⁄ R ~ 2 × 10 43

joules of binding energy. Hence, enough energy is released to blow the white
dwarf into smithereens.

3 Answers will vary.

339
 QUIZ SOLUTIONS

4 Answers will vary.

5 Answers will vary.

6 The radius of influence is where the velocity dispersion that would be produced
solely by the black hole is comparable to the actual velocity dispersion of the
surrounding stars. For the given parameters, it is 17 pc.

7 Answers will vary.

8 The density varies as 1 ⁄ r 2, assuming a spherical mass distribution.

9 Answers will vary.

10 In general, T = 2.7 K divided by a . We need 2.7kT to equal 2m ec 2 , giving

T = 4.4 × 10 9 Kelvin, and 1 ⁄ a = 1.6 billion.

11 Answers will vary.

12 a(t) = (t ⁄ t 0) 1 ⁄ 2. The age of the universe is t 0 = 1 ⁄ 2H0 = 7 billion years.

Click here to go back to the quiz.

340
 IMPORTANT NUMERICAL VALUES
CONSTANTS, UNITS, AND LAWS
astronomical unit (AU). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.496 × 10 8 km = 215.1 R sun
Bohr radius (a 0 ). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.29 × 10 −11 m
Boltzmann constant (k). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.381 × 10 −23 J ⁄ kg
Coulomb’s constant (η). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.988 × 109 N m 2 ⁄ C 2
electron mass (me). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.11 × 10 −31 kg
Hubble constant (H0). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 km ⁄ s ⁄ Mpc
Newton’s gravitational constant (G ). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.673 × 10 −11 m 3 ⁄ kg ⁄ s 2
parsec (pc). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.26 light-years = 206,265 AU
Planck’s constant (h). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.626 × 10 −34 J s
proton mass (mp ). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.67 × 10 −27 kg
speed of light (c). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.998 × 10 8 m ⁄ s
Stefan-Boltzmann constant (σ). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.670 × 10 −8 W ⁄ m 2 ⁄ K 4
Wien’s law. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . λmaxT = 2.9 mm K

OTHER VALUES
distance to the Andromeda Galaxy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 778 kpc
distance to the center of the Milky Way Galaxy.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.0 kpc
distance to the Coma galaxy cluster. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 Mpc
effective temperature of the Sun. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5777 K
luminosity of the Sun. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.826 × 10 26 W
mass of the Earth.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.974 × 10 24 kg
mass of the Sun.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.989 × 10 30 kg
radius of a proton.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.88 × 10 −15 m
radius of the Earth. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6378 km
radius of the Sun. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.955 × 10 5 km

341
 BIBLIOGRAPHY
Abbott, B., et al. “Observation of Gravitational Waves from a Binary Black
Hole.” Physical Review Letters 116 (2016): 061102. Reported the first detection
of gravitational waves.

Bartusiak, M. Einstein’s Unfinished Symphony: The Story of a Gamble, Two Black

Holes, and a New Age of Astronomy. Yale University Press, 2017. A casual and
friendly history of our understanding of gravitational waves and the effort to
detect them.

Blundell, K. Black Holes: A Very Short Introduction. Oxford University Press,

2016. Concise account of general relativity and black holes for first-year physics
students (i.e., expect some equations).

Carroll, B., and D. Ostlie. An Introduction to Modern Astrophysics. 2 nd ed.

Addison-Wesley, 2006. A voluminous undergraduate textbook on astronomy
and astrophysics.

Carter, J., et al. “Kepler-36: A Pair of Planets with Neighboring Orbits and
Dissimilar Densities.” Science 337 (2012): 556–559. The journal article reporting
the discovery of an extraordinary transiting planet system with 2 planets located
close to one another, leading to chaotic motions.

Choudhuri, A. Astrophysics for Physicists. Cambridge University Press, 2010.

Pursues all the major topics of this course at the graduate level. Especially good
on fluid physics and plasma physics, 2 very important topics that fell outside
the scope of this course.

Chown, M. Solar System: A Visual Exploration of All the Planets, Moons and Other
Heavenly Bodies That Orbit Our Sun. Black Dog & Leventhal Publishers, 2011.
Beautiful photographs and technical illustrations are used to expose the basic
properties of the solar system.

342
 Bibliography

de Pater, I., and J. J. Lissauer. Planetary Sciences. Cambridge University Press,

2015. A standard text on planetary science, aimed at upper-level undergraduates
or beginning graduate students.

Doyle, L., et al. “Kepler-16: A Transiting Circumbinary Planet.” Science 333,

6049 (2011): 1602–1606. The original journal article reporting the discovery
of the first transiting circumbinary planet.

Fleisch, D., and J. Kregenow. A Student’s Guide to the Mathematics of Astronomy.

Cambridge University Press, 2013. An elementary guide to some basic equations
that occur throughout astronomy for students who are less comfortable
with mathematics.

Levin, J. Black Hole Blues and Other Songs from Outer Space. Anchor, 2017. The
story of the Laser Interferometer Gravitational-Wave Observatory (LIGO), based
on interviews with the key personalities.

Maoz, D. Astrophysics in a Nutshell. 2 nd ed. Princeton University Press, 2016.

Excellent and concise textbook meant for physical science students with no
prior exposure to astrophysics. Covers many of the same topics as this course,
occasionally at a somewhat higher level of sophistication.

Perryman, M. The Exoplanet Handbook. Cambridge University Press, 2014. A

concise and comprehensive overview of exoplanet-observing techniques and
accomplishments. Written for the working astronomer, but some sections can
be understood by anyone who has studied college physics.

Phillips, A. C. The Physics of Stars. 2 nd ed. Wiley, 1999. Strongly physics-oriented
treatment of stellar structure, stellar evolution, and stellar remnants. The author’s
discussion of radiative diffusion was an inspiration for this course.

Prialnik, D. Theory of Stellar Structure and Evolution. 2 nd ed. Cambridge

University Press, 2009. Outstanding treatment of how stars work and how
they change with time. The author’s treatment of the core’s-eye view of stellar
evolution is especially insightful.
343
 Bibliography

Ricker, G., et al. “Transiting Exoplanet Survey Satellite.” Journal of Astronomical

Telescopes, Instruments, and Systems 1 (2014): 014003. https://doi.org/10.1117/1.
JATIS.1.1.014003. Overview of the NASA Transiting Exoplanet Survey Satellite
(TESS) mission, launched in 2017, whose purpose is to study and discover
thousands of new transiting exoplanets.

Rybicki, G. B., and A. P. Lightman. Radiative Processes in Astrophysics. Wiley-

VCH, 2004. The standard graduate-level work on the subject.

Ryden, B. Introduction to Cosmology. 2 nd ed. Cambridge University Press, 2016.

Convivial and clear introduction to the subject.

Ryden, B., and B. Peterson. Foundations of Astrophysics. Addison-Wesley, 2010.

A good college textbook on astrophysics as well as a source of inspiration in
designing this course.

Scharf, C. The Zoomable Universe: An Epic Tour through Cosmic Scale, from Almost
Everything to Nearly Nothing. Scientific American/Farrar, Straus and Giroux,
2017. A richly illustrated guide to the orders of magnitude of the universe.

Shapiro, S. L., and S. A. Teukolsky. Black Holes, White Dwarfs, and Neutron
Stars. The standard graduate-level work on the subject.

Thorne, K. Black Holes and Time Warps: Einstein’s Outrageous Legacy. Norton,
1995. A book for the general public about general relativity written by one of
the winners of the Nobel Prize for the detection of gravitational waves. Though
the field has changed quite a bit since 1995, this is still a great read.

Tyson, N., M. Strauss, and J. Gott. Welcome to the Universe: An Astrophysical

Tour. Princeton University Press, 2016. Covers the same ground as this course,
at a much less technical level and with many creative flourishes.

344
 Bibliography

Winn, J., and D. Fabrycky. “The Occurrence and Architecture of Exoplanetary

Systems.” Annual Reviews in Astronomy and Astrophysics, 53 (2015): 409–447.
https://doi.org/10.1146/annurev-astro-082214-122246. A review article written
for astronomers. Discusses the shapes, spacings, and orientations of exoplanet
orbits as well as planets in binary star systems.

INTERNET RESOURCES

Astronomy Picture of the Day. https://apod.nasa.gov/apod/astropix.html. Check

this site daily and enjoy a continuous and casual astronomy education.

Eames Office. “Powers of Ten TM (1977).” Published on August 26, 2010.

YouTube video, 9:00. https://www.youtube.com/watch?v=0fKBhvDjuy0. Classic
video that inspired the first few lectures of this course.

Khan Academy. “Introduction to Logarithms.” Khan Academy course. https://

www.khanacademy.org/math/algebra2/exponential-and-logarithmic-functions/.
Review of the key properties of logarithms and exponentials used throughout
this course.

NASA Exoplanet Archive: A Service of NASA Exoplanet Science Institute. https://

exoplanetarchive.ipac.caltech.edu/. The most comprehensive catalog of known
exoplanets, including their sizes, orbits, distances from Earth, etc.

Obreschkow, Danail. “Cosmic Eye (Original HD Portrait Version 2011).

Published on January 11, 2012. YouTube video, 3:09. Modern update of the
classic “Powers of 10” video. https://itunes.apple.com/app/id519994935.

PhET. “Radiating Charge.” https://phet.colorado.edu/en/simulation/

radiating-charge. Interactive simulation of electromagnetic radiation from an
accelerating charge.

Planetary Fact Sheets. http://nssdc.gsfc.nasa.gov/planetary/planetfact.html. This

website presents lots of basic data for the planets in the solar system.

345
 IMAGE CREDITS
main title . . . . . . . . . . . . . . . . Vikrant Agarwal / EyeEm/Getty Images
lecture titles . . . . . . . . . . . . . . . . . . . . ©vi73777/iStock/Getty Images
5. . . . . . . . . . . . . . . . . . . . . . . . . ©GeorgiosArt/iStock/Thinkstock

5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Google Earth

6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Google Earth

16 . . . . . . . . . . . . . . . . . . . . . . . . . ©Science Photo Library - CDC/

Brand X Pictures/Getty Images

16 . . . . . . . . . . . . . . . . . . . . . . ©Steve Allen/Stockbyte/Getty Images
27 . . . . . . . . . . . . . . . . . . . . ©Wavebreakmedia/iStock/Getty Images
28. . . . . . . . . . . . . . . . . And1mu/Wikimedia Commons/CC BY-SA 4.0
77 . . . . . . . . . . . . . . . . . . . . ©OstapenkoOlena/iStock/Getty Images
78 . . . . . . . . . . . . . . . . . . . . ©OstapenkoOlena/iStock/Getty Images
97. . . . . . . . . . . . . . . . . And1mu/Wikimedia Commons/CC BY-SA 4.0
108 . . . . . . . . . . . . . . . . . . Sch/Wikimedia Commons/CC BY-SA 3.0
112 . . . . . . . . . . . . . . NASA images courtesy Science Operations Center
at the Johns Hopkins University Applied Physics Laboratory.
113 . . . . . . . . . . . . . . . . . . . . . . . . . ©MarcelC/iStock/Thinkstock

126 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ESA/Hubble & NASA

143 . . . . . . . . . . . . . . . David Broad/Wikimedia Commons/CC BY 3.0

146 . . . . . . . . . . . . . . . . . . . . . . . . . . ©Glowimages/Getty Images
148 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NASA/CXC
150 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NASA
154 . . . . . . . . . . . . . . . . BélaBéla/Wikimedia Commons/CC bY-SA 4.0
157 . . . . . . . . . . . . . . . . Smithsonian Institution/flickr/Public Domain
158 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NASA/STSci
158 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NASA
159 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NASA/STSci
163 . . . . . . . . . . . . . . . NASA/SDO & the AIA, EVE, and HMI teams
203 . . . . . . . . . . . . . . . . . . . . . . . ©Astrobobo/iStock/Getty Images

346
 Image Credits

223 Adam Block/Mount Lemmon

. . . . . . . . . . . . . . . . . . . . . . . . . .

SkyCenter/University of Arizona
248 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ESA/Hubble, NASA
252 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NASA/CXC/SAO
257 . . . . . . . . . . . . . . . . . . . . . . NASA/CXC/NCSU/M.Burkey et al
260 . . . . . . . . . ©MARK GARLICK/Science Photo Library/Getty Images
267 . . . . . . . . . . . . . . . . . . . . . . . . . . . . LIGO Lab/Caltech/MIT
272 . . . . . . . . . . . LIGO Scientific Collaboration and Virgo Collaboration
275 . . . . . . . . . . . . . . . . . . . . . . . ESA/Gaia/DPAC/CC BY-SA IGO
277 . . . . . . . . . . . . . . . . . . . . . . . ©blackphobos/iStock/Thinkstock
278 . . . . . . . . . . . . . . . . . . . . . . . . . ©Stocktrek Images/Thinkstock
279 . . . . . . . . . . . . . . . . . . . . . . . . ©azstarman/iStock/Thinkstock
280 . . . . . . . . . . . . . . . . . . . . NASA, ESA, and the Hubble Heritage
(STScI/AURA)-ESA/Hubble Collaboration
288 . . . . . . . . . . . . . . . . . . . . . . . . . NASA, ESA, and The Hubble
Heritage Team (STSci/AURA)
289 . . . . . . . . . . . . . . . . . . . . NASA and The Hubble Heritage Team
(STScI/AURA); Acknowledgment: Ray A. Lucas (STScI/AURA)
291 . . . . . . . . NASA, ESA and the Hubble Heritage Team (STScI/AURA)
292 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NASA/CXC/M.Weiss
294 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NASA/STScI
295 . . . . . . . . . . . . . . . . . . . . . . Charles Steidel (California Institute
of Technology, Pasadena, CA) and NASA/ESA

347