ANRV273-EA34-07 ARI 21 March 2006 7:58

Planetesimals to Brown Dwarfs:

What is a Planet?
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

Gibor Basri1 and Michael E. Brown2

Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

1
Astronomy Department, University of California, Berkeley, California 94720-3411
2
Division of Geological and Planetary Science, California Institute of Technology, Pasadena,
California 91125

Annu. Rev. Earth Planet. Sci. Key Words

2006. 34:193–216
First published online as a extrasolar planets, Kuiper belt, defining planets, substellar objects,
Review in Advance on cosmogony, new planets
January 16, 2006

The Annual Review of

Abstract
Earth and Planetary Science The past 15 years have brought about a revolution in our understanding of our Solar
is online at
earth.annualreviews.org System and other planetary systems. During this time, discoveries include the first
Kuiper belt objects (KBOs), the first brown dwarfs, and the first extrasolar planets.
doi: 10.1146/
annurev.earth.34.031405.125058 Although discoveries continue apace, they have called into question our previous
perspectives on planets, both here and elsewhere. The result has been a debate about
Copyright 
c 2006 by
Annual Reviews. All rights the meaning of the word “planet” itself. It is clear that scientists do not have a widely
reserved accepted or clear definition of what a planet is, and both scientists and the public are
0084-6597/06/0530- confused (and sometimes annoyed) by its use in various contexts. Because “planet” is
0193$20.00 a very widely used term, it seems worth the attempt to resolve this problem. In this
essay, we try to cover all the issues that have come to the fore and bring clarity (if not
resolution) to the debate.

193
 ANRV273-EA34-07 ARI 21 March 2006 7:58

INTRODUCTION
Since prehistoric times, people have looked into the night sky and picked out the
Exoplanet: any planet planets. It was easy: They were the stars that moved (initially the Greeks considered
orbiting an object other the Sun and Moon to be planets too). Even when Galileo found that the planets
than our sun are other worlds (and that Earth is one, too), there was no real controversy about
Brown dwarf: an object what planets were. When Herschel spotted Uranus, he had no trouble claiming its
intermediate in mass planetary status (once its orbit was established). The real problems began at the start of
between stars and planets
the nineteenth century, with the discovery of Ceres (right where the Titus-Bode Law
(sometimes also called
“substellar object”) said it should be). Although thought a planet for a few years, it was soon reassigned
as an asteroid based on its size and mostly because of the company it kept (too many
similar bodies in similar orbits were discovered). Neptune was never in doubt, and
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

Pluto started off with only an inkling of trouble. As Pluto’s estimated size shrunk (from
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

like Mars to between Mercury and Ceres), people mostly shrugged their shoulders.
When, however, other Kuiper Belt objects (KBOs) were discovered during the 1990s
(in increasing numbers, including many “Plutinos” in the same resonant orbit, and
objects increasingly close in size to Pluto), the arguments used against Ceres as a
planet resurfaced against Pluto.
This past decade also saw the first discoveries of planets outside our Solar System
(exoplanets). The first exoplanets discovered are actually terrestrial, but were found
in orbit around a neutron star (Wolszczan & Frail 1992). These clearly have a very
different history from planets in our Solar System because their current orbits would
have been inside the supergiant star that preceded the central pulsar. The next large
group of (gas giant) exoplanets (well over 100 at this point) were discovered by the
radial velocity variability they induce in their host stars (Marcy & Butler 1998).
These objects are sometimes substantially more massive than Jupiter, and quite often
found in surprisingly (compared with our Solar System) close and/or elliptical orbits.
Recently, a few close-in exoplanets have also been found by the transit method—a drop
in light caused by the exoplanet passing between us and the host star (e.g., Burrows
et al. 2004). Distant exoplanets have also been found by gravitational microlensing
(e.g., Bond et al. 2004).
Brown dwarfs (“failed” stars) were first discovered at nearly the same time, and
then brown dwarfs with increasingly small masses began to be discovered (Basri 2000).
Inevitably, the puzzle of how to distinguish between exoplanets and brown dwarfs near
some transitional boundary appeared. This question sharpened as very-low-mass ob-
jects were discovered not orbiting anything else (e.g., Lucas & Roche 2000, Martin
et al. 2001). Systems were also found with exoplanets and brown dwarfs coexisting
in “planetary” orbits (Marcy et al. 2001). Astronomers and planetary scientists were
forced to admit that they do not really have a clear-cut definition of “planet.” Both
these debates and the “Pluto controversy” came as a surprise to the public, who did
not realize there was any problem defining “planet” (after all, we unquestionably
live on one!). We are just coming to the point where substantial numbers of terres-
trial exoplanets can be found. The Kepler Mission (NASA) will utilize the transit
method to accomplish this, beginning as early as 2008 (Borucki et al. 1997). Presum-
ably, their identity will not be controversial (as they will most closely resemble our

194 Basri · Brown

 ANRV273-EA34-07 ARI 21 March 2006 7:58

home world), although we should not underestimate the ability of Nature to hand us
surprises.
It is valid to ask why we scientists should even care what the definition of “planet”
is. Don’t we just study whatever is out there? One answer is that we use the word quite
a bit (as in the title of this publication), so it seems sensible to know what we mean by it.
Another reason is that it is a loaded word. The public is very interested in planets, and
learns about them as young schoolchildren. Certain kinds of discoveries (especially
of new planets) bring benefits (or at least attention) beyond that for discoveries of
similar objects that are not called “planets.” The latest example of this importance is
the current (as of this writing) controversy around the discovery of the “tenth planet”
(2003 UB313) by Brown et al. (2005). There are sometimes professional arguments
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

over whether the discovery merits such attention, based on the definition of “planet.”
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

We care, in part, because society cares.

“When I use a word”, Humpty-Dumpty said in a rather scornful tone, “it means just
what I choose it to mean, neither more nor less”.
“The question is”, said Alice, “whether you can make words mean so many different things”.
“The question is”, said Humpty-Dumpty, “which is to be master, that’s all.”
Lewis Carroll, Alice Through the Looking Glass (1872)

A good definition has several desirable characteristics. It should be succinct and

easily understood by the public, yet precise enough to be acceptable to scientists. An
ideal definition would not depend on specific knowledge or examples that will change
as we learn more, although the definition of planet has already changed multiple
times over the centuries of usage. Any definition should differentiate planets from
other objects they might be confused with, preferably based on observables (to allow
determination of whether a given object qualifies; it is even better if these observables
are quantitative). One might hope that a definition not be time or history dependent
(once a planet, always a planet), although there are proposed definitions for which that
is not the case. There is another characteristic of a good definition that, although much
less objective, also seems to be a major factor. The definition should have “cultural
support,” by which we mean that it cannot violate any strong preconceptions held
by a large number of people. The word planet has been around for a long time, and
scientists need to take care not to try to change too much the meaning of a word that
everyone already knows and uses.
Nature, of course, does not worry about definitions. Objects are made in various
contexts, with a continuum of masses, and using a number of formation mechanisms.
It is humans who wish to apply labels and to distinguish between different objects.
We must accept the fact that, in placing boundaries on a continuum, there might be
transitional objects that are quite similar to each other in many respects, yet which are
defined to lie on opposite sides of our imposed (and hopefully not too arbitrary) lines.
A scientific definition would find natural physical conditions for such boundaries. A
successful definition can only be arrived at through a consensus process. Our purpose
here is to lay out many of the arguments and considerations that go into it. Whether
some official definition is ever codified (or some consensus definition comes into

www.annualreviews.org • Planetesimals To Brown Dwarfs 195

 ANRV273-EA34-07 ARI 21 March 2006 7:58

wide use) depends on both scientists and the lay public having reached some common
understanding.
One major step is to decide the arenas in which difference is to be defined. There
are four basic arenas within which the planet definition debate is being played out.
These are characteristics, circumstances, cosmogony, and culture. A good deal of
the debate derives from the various weightings that different proposed definitions
give to each of these arenas, and the rest of the debate arises from disagreements
as to where lines should be drawn within them. In this article, we examine each
of the arenas and discuss the issues and controversies they contain. This is not a
traditional research review—it is more of an op-ed piece. We assume that the reader
has the background knowledge in planetary science to appreciate most of what we
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

are discussing. It would take far too much space to provide all that background, so
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

there is only a sparse bibliography, which is largely astronomical to bring current

developments to the predominantly planetary audience. At the end, we provide a
number of possible definitions for “planet.”

CHARACTERISTICS: PHYSICAL PROPERTIES

OF THE OBJECTS
In astrophysics, the definition of objects is usually based solely on their physical char-
acteristics, or observational characteristics if we do not yet understand the physical
objects. A modern star is an object whose luminosity is derived solely from hydrogen
fusion for some period of time. One need not specify how it formed, whether it is by
itself or in a binary system or cluster, or if it dies as a white dwarf or in a supernova
explosion. Prior to the twentieth century, it was a sun-like object, and prior to that a
fixed point of light on the celestial sphere.
For stars, the fundamental variable that determines the nature and fate of the ob-
ject is mass. In the case of planets, Earth is now unquestionably an example, but we
argue about how to define the upper and lower mass limits (or whether to use some
other scheme to set limits). We do know that if you keep reducing the mass, you even-
tually end up with rocks or snowballs, and that if you keep increasing the mass you
end up with brown dwarfs and then stars. The question in this arena is whether
there are observable, rational, informative, and/or compelling places to set the
limits.
From a physical point of view, the limiting objects (which everyone agrees are not
planets) differ in two different ways from planets near those limits. At the lower end,
the size and shape of rocks are quite different from those of small planets (while the
composition and luminous mechanisms can be similar). At the upper end, the amount
and mechanism of the luminosity of stars is somewhat different from those of large
planets (while the composition and size may be similar). It therefore seems sensible
to concentrate on the physical consequences of pressure support versus gravity at the
lower end, and the luminosity mechanisms at the upper end (while composition should
be left out in both cases). Of course, the luminosity of stars is also a result of the com-
petition between pressure support and gravity, so we are being consistent in that sense.
There are other physical transitions that depend on mass that we can also consider.

196 Basri · Brown

 ANRV273-EA34-07 ARI 21 March 2006 7:58

EFFECTS OF INCREASING MASS

Let us now briefly review the competition between pressure support and gravity,
from rocks to stars. Rocks are essentially supported by bound electron degeneracy
(in atoms and molecules), and gravity plays no role in the size or shape of the object.
This accounts for the bulk volume of macroscopic objects (like ourselves); without it
everything would collapse to nuclear densities (with sizes many orders of magnitude
smaller than before, as in a neutron star).
As one increases the mass of an object, several important transitions are reached.
Perhaps the first is the mass at which gravity is able to hold together a rubble pile.
It seems that the comet (Shoemaker-Levy 9) that impacted Jupiter in 1994 and the
comet Tempel 1 (determined by the Deep Impact mission) are such objects, along with
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

some small asteroids (based on their densities). This limit is clearly too low (as nobody
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

considers such objects planets). Next, the force of gravity can exceed the material
strength of the body and force the object to take the gravitational equipotential
surface for its shape. That shape, of course, is a sphere (in the absence of rotation).
One can estimate the size at which this transition occurs by equating the differential
pressure of a deviation from sphericity with the yield or compressive strength of the
material. For materials with the yield strength and density of stony meteorites, the
critical diameter is approximately 800 km.
This estimate is in accord with observations in the Solar System (Figure 1). Ceres,
with an average diameter of 941 km, is a gravitationally relaxed object (Thomas et al.
2005); although there is evidence it may have substantial water content in its mantle.
The next largest asteroids, Pallas and Vesta, with approximate diameters of 608 and
538 km, respectively, appear moderately elongated in Hubble Space Telescope (HST)

Figure 1
Small objects in the Solar System. Vesta (left) as imaged by HST (the lower image is a model
based on the data). Vesta (diameter = 538 km) is rocky and clearly nonspherical. Mimas
(middle image, from Cassini) is slightly smaller than Vesta (395 km) but icy and basically round,
although gravity has not caused the huge Herschel crater to smooth out to the level of smaller
craters. Ceres (right image, 941 km) has recently been found by HST to closely match the
gravitational equipotential surface (and has evidence for a differentiated interior).

www.annualreviews.org • Planetesimals To Brown Dwarfs 197

 ANRV273-EA34-07 ARI 21 March 2006 7:58

images. Outer Solar System bodies made of an ice-rock mixture (more ice than rock)
have lower yield strength and should become round at smaller sizes. Miranda is almost
spherical at 472 km, but retains large topological features from an apparently violent
youth; Mimas has a diameter of nearly 400 km but retains the huge Herschel crater.
The largest known substantially nonspherical icy body is Saturn’s satellite Hyperion,
with dimensions of 180 km × 140 km × 112 km.
While becoming spherical is perhaps the most obvious outward sign of increas-
ing mass, the interior of the body begins to undergo many interesting geophysical
transitions as the mass increases. An important transition from an object that is geo-
physically dead to one that is active is the transition to bodies large enough to have
convection in their interiors. The threshold for convection scales as the Rayleigh
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

number in the interior, which scales roughly as D5 times the thermal contrast. This
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

is likely to be an increasing function of D and a decreasing function of age (for bodies

of similar composition, where D is the diameter of the body). With such a strong
dependence on planetary radius, solid-state convection quickly becomes important
somewhere in the size approximately between the Moon (D = 3500 km) and Mars
(D = 6500 km). At a similar mass—a diameter of around 3000 km for a terres-
trial body—the average gravitational energy per atom exceeds ∼1 eV, which is the
typical energy of chemical reactions. A body this size has sufficient gravitational en-
ergy to substantially modify the initial chemical composition of its initial starting
materials.
One final threshold that occurs near this size range is when the central pressure of
the object begins to be large enough that the materials have significantly higher densi-
ties than their noncompressed forms. The central pressure at which such compression
occurs is approximately equal to the bulk modulus of the constituent material, which
is approximately 1 Mbar for rocky materials and 10 kbar for icy materials. If we as-
sume hydrostatic equilibrium and incompressible materials, the central pressure in a
body is P = (1.4 kbars)(ρ g/cc)2 (R/1000 km)2 . Pressures high enough for significant
compression are reached for a size of approximately 6000 km for rocky materials and
only 1000 km for ice-rock mixtures. The true sizes needed for compression will be
somewhat smaller, as the compression violates our incompressible assumption. The
estimates of the body size for which these important transitions are reached are ex-
tremely rough and can be greatly influenced by composition, temperature, state of
differentiation, and other details about the body. Nonetheless each of these transi-
tions is conceptually well defined and could equally well form the potential basis for
drawing boundaries.
After passing this mass range, there is not another significant qualitative change in
the relation between pressure and gravity until masses greater than Jupiter. There was
thought to be a significant “characteristic” boundary at about 10–15 Earth masses, in
the sense we did not think we would find “terrestrial” planets with larger masses. This
is because the supply of heavy elements in the protoplanetary disk is only approxi-
mately 1% that of hydrogen and helium (which we henceforth refer to as gas). It is
hard to gather a more massive object without also gathering a lot of gas. The planets
Uranus and Neptune are transitional objects in this sense. Very recently, however,
the discovery of an exoplanet (Sato et al. 2005; Figure 2) with nearly the mass of

198 Basri · Brown

 ANRV273-EA34-07 ARI 21 March 2006 7:58

Low-mass planets High-mass planets

Transiting exoplanets

1.00
Earth 10 Jupiter
Saturn

Mercury
Earth radii

Earth radii
Luna Uranus
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

Pluto

0.10

Ceres
1
Earth
Miranda

0.01
-5 -4 -3 -2 -1 0 1 -1 0 1 2 3 4
Log (Earth masses) Log (Earth masses)
Figure 2
The mass-radius relations for planets. This is essentially just a statement of bulk densities. For
low-mass planets, the lower line represents a bulk density of 6 gm/cc, whereas the upper line is
for 0.5 gm/cc. The planets lie between these lines depending on their composition: Rocky
planets with high concentrations of heavy elements have high densities, and icy or gaseous
planets have low densities. The relation changes at Jovian masses because degeneracy pressure
plays an increasing role (the density of the planets begins to increase rapidly). Furthermore, it
is difficult to construct very massive planets with primarily heavy elements (the behavior
shown for the high-density case is arbitrary between Uranus and Jupiter). The mass-radius
relations are only known for extrasolar planets (asterisks) that transit their stars and have radial
velocity data (giving the radius and mass, respectively, as a fraction of the stellar values).

Saturn but only two thirds its size has thrown this theoretical expectation into doubt
(it may have a rock/ice core of 60 Earth masses or more).
For massive gas giants, a significant change in the source of core pressure support
Degeneracy pressure: a
occurs. This change has been slowly brewing as one approaches higher masses because pressure term produced by
thermal support begins to play an increasing role. The gas giants have very hot cores the Fermi exclusion
compared with terrestrial planets because the pressures are so large at their center, principle operating on free
and this energy is able to free electrons even under such extreme pressures. Indeed, electrons of sufficient
density
electrons were already becoming freer in the metallic hydrogen interior of a planet
like Jupiter. At around two Jupiter masses, free electron degeneracy pressure begins
to become comparable to “normal” pressure terms.

www.annualreviews.org • Planetesimals To Brown Dwarfs 199

 ANRV273-EA34-07 ARI 21 March 2006 7:58

Until that point, adding mass to the planet made its radius grow. Free electron
degeneracy is different in two fundamental respects. First, it does not require thermal
input to make the electrons move faster (thereby providing more pressure). They now
do so because of the Fermi exclusion principle: There are enough electrons in a small
volume that some can be forced to occupy very high energy levels simply because
all lower ones are filled. This leads to the second difference, which is that producing
increased pressure requires increasing density (not more heat). Thus, adding mass
to objects supported in this way causes their radius to decrease (and therefore their
density to increase).
The mass boundary where objects begin to become smaller and denser is at ap-
proximately twice the mass of Jupiter (Figure 2). Such an object is thus the largest
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

planet one can have, although the peak in size extends to approximately five Jupiter
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

masses. A planet with ten Jupiter masses will actually be smaller than Jupiter, as are
brown dwarfs. One caveat here is that we must wait for the outer layers to cool off
before this is true because they are not supported by free electron degeneracy. This
is why the “hot Jupiters” shown in Figure 2 are actually larger than the peak size of
cold planets. We do not see the usual case of size increasing with mass again until
we reach true stars, whose nuclear fusion can get the core hot enough to lift the free
electron degeneracy. The onset of degeneracy (the largest size reached by nonfusing
objects) provides one natural place to draw a boundary.

SOURCES OF LUMINOSITY
We do not discuss luminosity sources for terrestrial planets, as nobody has proposed
using them to define limits at low masses. For gas giant planets, the very process of
cooling (radiating luminosity) allows the planet to contract, releasing gravitational
energy. This is the dominant source of luminosity for Jovian planets (and indeed
for brown dwarfs and stars in their prenuclear burning phases). The luminosity-
size history of the object can be simply derived by knowing how easily it radiates
contraction energy away. The contraction continues until another pressure term
(free electron degeneracy for very large planets) halts it. There is no nuclear fusion
at all.
The minimum mass required to cause substantial fusion is approximately 13 Jupiter
masses. What is fused at that mass is heavy hydrogen, or deuterium. All bodies form
with a supply of deuterium (produced in the Big Bang) at about 10−5 the abun-
dance of light hydrogen. All stars and brown dwarfs burn their deuterium during
the early part of their life; more massive objects burn it faster. For objects near
the fusion boundary, the deuterium-burning phase lasts more than 50 million years
(much longer than the total fusion lifetime of very-high-mass stars). At the bound-
ary mass, the deuterium burning does not fully halt the contraction or contribute
most of the luminosity at any time, but for just a little higher mass (such as 15
Jupiter masses), it basically does. The fusion limit constitutes a well-defined qualita-
tive and measurable (in principle) difference between objects that are otherwise quite
similar.

200 Basri · Brown

 ANRV273-EA34-07 ARI 21 March 2006 7:58

CIRCUMSTANCES: THE CONTEXT OF PLANETS

We now discuss the circumstances under which an object should earn the “planet”
label. It is well to remember that the question of whether orbits themselves are Orbital migration: a
a requirement is one of tradition (as we discuss below in Culture: The Common change in the size of a
Understanding of “Planet”). The requirement has been in place for the past four body’s orbit produced by
interactions between the
centuries or so (but not before then!). If Mercury were orbiting Jupiter rather than the
body and other bodies, or
Sun, it would clearly never have been considered a planet (as is the case for Ganymede), the body and a
whereas if our Moon were in Mercury’s place, it would be (barely) observable with protoplanetary disk
the naked eye and would have been considered a planet even before the Copernican
Revolution. Clearly, the common use of “planet” includes orbital circumstances.
Other considerations include whether the planet is “unique” or clearly a mem-
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

ber of a population (defined dynamically). The question of whether the orbit fits
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

our Solar System intuitions also keeps coming up. We also see below that circum-
stances can change. This immediately raises the question of whether it is the current
circumstances that matter or the full history of circumstances.

ORBITAL DYNAMICS OF PLANETARY SYSTEMS

AND PROTOPLANETARY DISKS
We first discuss the circumstances surrounding the formation of planets. Initially,
they form inside a massive gas disk (the ratio of gas to dust is generally 100:1 in the
interstellar medium). When particles are very small (up to the size of chondrules or
so), they are subject to Epstein drag with the gas. As they grow bigger, they are still
subject to gas drag, which arises because the gas has a slight radial pressure, causing it
to orbit slightly more slowly at a given radius. This causes bodies up to meters in size
to drift rather rapidly inward, especially in the inner planetary system, posing one of
the several problems in understanding how interstellar dust can reach planetesimal
size (kilometers).
There are several basic types of orbital migration that result from the interaction
between a forming planet and the gas disk (cf. Nelson & Papaloizou 2004). Two
are due to the production of waves and torques in the gas disk by the planet (they
are usually called Type I and Type III). They tend to operate on bodies with masses
between lunar and a few Earth masses, and they drive the bodies rapidly inward (on
short timescales compared with the disk lifetime). Another type of orbital migration
is due to the opening of a tidal gap in the disk by a massive planet (Type II). In this
case, the planet gets locked into the disk and shares its motion thereafter (inward
drift if in the accretion part of the disk). The timescale here is longer but still of
concern. All mechanisms tend to result in planets falling into the star (or getting
parked very close by, perhaps). On the positive side, this could explain the presence
of massive exoplanets close to their stars. On the negative side, it is actually hard to
retain planetary cores long enough to induce runaway gas accretion and build giants.
Of course, the outer regions of the disk are actually spreading, so planets locked out
there will migrate outward. The disk viscosity that drives all this is rather poorly
understood, so this topic remains one of active research.

www.annualreviews.org • Planetesimals To Brown Dwarfs 201

 ANRV273-EA34-07 ARI 21 March 2006 7:58

A related topic is the sweeping up of planetesimals (or smaller planets) as a giant

migrates inward, and another is the capturing of such bodies in outer orbital reso-
nances as they try to catch up to the giant. Both of these depend (in opposite ways) on
Hill sphere: the size of the
region around a planet whether the smaller bodies migrate faster than the large body. Much theoretical effort
where its gravity dominates is going into understanding these migrations better, and with more sophistication.
that of the central star Disks of planetesimals can also drive orbital modification. Saturn, Uranus, and
Neptune migrated substantial distances outward as they scattered planetesimals pre-
dominantly inward. Jupiter migrated slightly inward as it removed many of these plan-
etesimals from the Solar System. Dynamical friction can also play an important role in
the types of orbits that are produced and in the later mergers of planets and planetary
embryos to produce the “final” population. Orbits also can be changed through direct
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

gravitational scattering. Particularly when at least one of the planets is massive, other
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

planets can have their orbits substantially altered, or even be ejected from the system.
If there are several massive planets, quite a bit of system evolution can occur. It is
clear that the orbits of the planets can change dramatically during these early phases.
The next, and perhaps most important, transition occurs between objects that
are or are not massive enough to clear residual planetesimals through accretion or
scattering. Empirically, we can see that Ceres, for example, is not massive enough
to clear the asteroid belt of its many remaining planetesimals, whereas the other
terrestrial planets have certainly been successful in clearing their regions of influence.
As a cultural matter, collecting a few small bodies into orbital resonances has not
disqualified an object from being a planet (e.g., the Trojans for Jupiter or the Plutinos
for Neptune).
The most stringent criterion for clearing planetesimals is that the surface es-
cape velocity—which is the maximum velocity that one body can impart on another
through gravitational interaction—be greater than the local escape velocity from the
central star. A body of such mass will be able to scatter other small bodies beyond the
gravitational influence of the star. This criterion is met when the ratio of the planet to
central star mass is greater than the ratio of the planet’s distance from the star to the
radius of the star. In the Solar System, this threshold is only crossed for Jupiter, which
is thus the only planet capable of easily removing bodies from the Solar System.
For bodies less massive than Jupiter, clearing proceeds through a combination of
accretion and the eventual influence of larger bodies. A strict criterion for the size
at which this clearing happens is difficult to decide upon. One could argue that the
terrestrial planets would not have been as effective in clearing planetesimals without
the influence of Jupiter. In the outer Solar System, for example, planetesimals have
mostly been cleared inside of Neptune, where they can eventually be influenced by
Jupiter, but not outside of Neptune, where they have been pushed just to the limit of
Neptune’s influence. Empirically, however, it is easy to determine the largest bodies
in the Solar System that have not cleared their region: Ceres is still surrounded by
the asteroid belt, and Pluto and 2003 UB313 are still surrounded by the Kuiper belt.
To roughly estimate what mass Ceres or 2003 UB313 would have to have to clear
their orbital neighborhoods, one can examine the spacing of the “true planets” in
terms of their Hill radii (the Hill sphere is defined roughly as the size of the region
around a planet where its gravity dominates that of the central star). The terrestrial

202 Basri · Brown

 ANRV273-EA34-07 ARI 21 March 2006 7:58

planets that have cleared the planetesimals between them are separated by approxi-
mately 30 Hill radii each, whereas the giant planets are separated by approximately
10 Hill radii. An object in the asteroid belt that cleared the region between that object
and Mars would need approximately 1 Earth mass. In the Kuiper belt, coincidentally,
a similar mass is required. Although these estimates are rough, they suggest that the
largest belt objects, 2003 UB313 and Ceres, are not massive enough to clear out their
populations by about two to three orders of magnitude (more detailed calculations
agree, see Stern & Levinson 2002).
The clearing criterion divides the Solar System into two distinct classes. The eight
largest bodies have all cleared residual planetesimals from their vicinities, whereas
the others have not. This criterion is the basis for one possible dividing line between
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

objects that are planets and objects that are not. A planet could be, by definition, an
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

object that has cleared residual planetesimals, and thus cannot be part of a population
of smaller bodies.
As with all definitions, this one has advantages and disadvantages. The main ad-
vantage is that it correctly divides the Solar System into its dynamically important
constituents. Dynamically, there is no argument whether Ceres and 2003 UB313 nat-
urally belong to the asteroid and Kuiper belt populations rather than to the collection
of larger objects. One disadvantage is that there is no simple a priori mass threshold
above which an object would be classified as a “planet.” An interesting situation would
arise if, for example, a Mars-sized object were found in the inner Oort cloud. By the
dynamical definition, this object would not be considered a planet, yet most people
would instinctively feel that a body that size should indeed be called a planet.
For extrasolar planetary systems, there are likely to be a wide variety of unforeseen
dynamical circumstances, and those that have been seen have mostly been surpris-
ing. Many objects are found in eccentric orbits, some in mean motion resonances,
some with absurdly short orbital periods, and some with uncomfortably long ones
(given our Solar System preconceptions about how they form). Because of our current
mass-detection limits, all known exoplanets are well above the mass needed to clear
planetesimals from their neighborhood. Interestingly, few of the objects currently
known are massive enough to directly eject planetesimals because of detection biases
favoring short-period planets that are deep in the central stars’ gravitational wells. Of
particular interest will be new systems with terrestrial planets and no giant planets.
Are there any dynamical circumstances that should preclude an object from being
considered a planet in the circumstantial arena? As we did for increasing mass, we
search for natural dynamical transition points, and find only three. The first is between
bodies that are on stable orbits and those on unstable orbits. Even this distinction is
ambiguous. The Solar System itself is unstable on sufficiently long timescales, and it
is likely that planetary-sized bodies have been ejected from it in the past. All planetary
systems after their formation phase will be, by definition, at least marginally stable on
the timescale of their current lifetime, so there is no real need to distinguish between
stable and unstable systems. The second transition point, which we have concentrated
on, is whether the object can clear its neighborhood of smaller bodies. We have seen
that this does not depend only on mass, however, but also on circumstances (past and
present).

www.annualreviews.org • Planetesimals To Brown Dwarfs 203

 ANRV273-EA34-07 ARI 21 March 2006 7:58

The final and most obvious dynamical distinction is between bodies that are bound
to a massive central object (or objects) and those that are not. From the point of view
of orbital dynamics, at least, it makes no sense to talk of planets unless they are—or
were—bound. One question is whether the central object has to be a star, or whether a
brown dwarf (or neutron star) will do; most people seem willing to include all objects
capable of fusion. But if isolated massive nonfusing objects are not planets because
they do not orbit stars, what are the smaller objects that could be orbiting them?
Moons? Planets? Dynamical definitions also get into trouble when discussing ejected
planets. Is an object that was once orbiting a star and had the attributes of a planet but
is then ejected no longer a planet? It seems that the history of an object could play a
role in whether it is a planet, and that brings us to its earliest history, or cosmogony.
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

COSMOGONY: THE FORMATION OF PLANETS

This arena holds great importance for many scientists (although less so for the public).
It is perhaps the source of greatest disagreement among astronomers. There is a large
segment who hold that planets are by definition objects that “form in a disk around a
central star [perhaps] from the accumulation of planetesimals,” and that is the most
important defining property they have. The pure form of this position insists that if
the planet grows past the fusion limit, it is still a planet. Presumably, this position also
implies, if it turns out that Jupiter has no real core and formed suddenly during a rapid
gravitational instability in the protoplanetary nebula, that we have been mistaking a
sub-brown dwarf for a planet all this time. Such a claim, of course, would run afoul of
the culture arena. This difficulty is avoided by dropping the insistence on formation
through the specific mechanism of the accumulation of planetesimals.
The other element insisted upon is that the nucleation of the planet must take
place in a disk around a star (many accept a brown dwarf as a central object as well).
Objects (of any mass) that form in the centers of isolated disks (even if they ingest
many planetesimals) are not planets in this view. This cosmogonist group of scientists
is in contrast to another large group who think characteristics should receive the
greatest weight in the definition, and a third group who also give a lot of weight to
circumstances. The traditional position on planets (before 1990) is an amalgam of all
three arenas, heavily informed by our known Solar System at that time (culture).
The presence of disks in the formation of planetary (and stellar) systems is a given
in nature (cf. Mannings et al. 2000). There is essentially no chance that a collaps-
ing cloud can contain so little angular momentum that this will not happen. The
question is how massive and large the disk will be relative to the central object,
and whether it will be conducive to the subsequent formation of planetesimals and
planets within it. Evidence is quickly building that grain growth in disks is early
and robust, although the transition from large grains to planetesimals is much less
understood. Once planetesimals are present in large numbers, the path is clear to
larger bodies through pairwise accretion in what is called oligarchic growth (larger
bodies grow faster than smaller bodies—a runaway process). There are further merg-
ers between protoplanets and/or planets, and some may be ejected from the system
altogether.

204 Basri · Brown

 ANRV273-EA34-07 ARI 21 March 2006 7:58

In a circumstellar disk, objects above approximately 1 km in size can first begin to

cause enough gravitational perturbation to affect the dynamics within the disk. At this
size, we first call the bodies planetesimals, and the combined effects of gravitational
focusing and velocity evolution eventually causes runaway growth to the sizes of
protoplanets. In the simplest analytic planet-formation scenarios, these are objects
that have reached the isolation mass. That is, each protoplanet has swept out all
of the material within its Hill annulus and is accretionally isolated from the other
protoplanets. For the minimum-mass solar nebula at 1 AU, this isolation mass is
approximately equal to the mass of Mercury. In the Kuiper belt region, the Hill
annulus is significantly larger and the isolation mass is approximately seven Earth
masses. Any objects more massive than protoplanets must have undergone mergers
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

and are not pristine protoplanets any more.

Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

There is also an agreed upon standard model for gas giant planet formation.
This requires the growth of nongaseous planets up to 10–15 Earth masses while the
protoplanetary disk still has its gaseous component. Astronomers have found that gas
disks typically last approximately 5 million years, with the oldest lasting perhaps twice
that long. This sets the timescale over which the standard model must operate. When
the planetary core grows that massive, there is a rapid runaway gas accretion phase.
As the object grows, it opens a tidal gap in the disk, but it is now fairly clear that
accretion can continue even after the gap is opened (accretion streams spill into the
gap). The size to which the gas giant can grow is controlled by many factors (which
vary among systems) and is not really known.
The main competition to the standard model comes from rapid collapse scenar-
ios, which posit that disks that are cool and massive enough can form gas giants
directly through gravitational instabilities (Boss 1997). This resembles the means by
which close stellar binaries form. They also form in disks; typically, there will be a
circumstellar disk around each object, and often a circumbinary disk around both
of them (depending on their separation and the total angular momentum present).
As one shrinks the size of the secondary compared to the primary, the circumbinary
disk grows relatively larger and the disk around the secondary shrinks. The system’s
appearance slides smoothly over to what can be described instead as a small object
(surrounded by its own disk) embedded in a tidal gap in the circumstellar disk of the
primary. This picture is the same as in the rapid collapse scenario, and its resemblance
to binary star formation is one of the main factors invoked by those in the cosmogony
camp against including it for planets.
Of course, a small object surrounded by a small disk embedded in a gap in a much
larger disk is also a description that applies to a gas giant forming in the standard
scenario. There is clear evidence in the nature of the Galilean moon system that
Jupiter had such a disk. The time needed to form a planet is much less in the collapse
scenario, which also results in the lack of a denser planetary core. The rapidity is one
of its most attractive features; it avoids the problems due to orbital migration and the
dissipation of the circumstellar gas disk.
The collapse scenario is less well-accepted for a number of reasons. The main
empirical reason is that we know that three of our four Solar System giant planets
have cores, and Jupiter likely does, or did, too, giving no cause to look for an alternate

www.annualreviews.org • Planetesimals To Brown Dwarfs 205

 ANRV273-EA34-07 ARI 21 March 2006 7:58

mechanism. The main theoretical reason is that the conditions required in the disk
to cause the gravitational instability are rather extreme, and it has not been shown
that nature will actually produce them. Well-mixed disks are difficult to cool to the
point of instability (Bodenheimer 2004). One way around this is if the dust settles out,
making a cold dense layer at the disk midplane. In principle, some of the gas layer
could be depleted (by photoevaporation for systems in clusters with high-mass stars,
for example, or by the stellar wind), also producing some metallicity enhancement
relative to the central star. Mitigating against this are turbulent forces in the gas that
might keep the dust mixed in, and currently this problem is not understood enough
to make definitive conclusions.
There is also a scenario under discussion to bridge the grain-to-planetesimal gap
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

by gravitational instability (Youdin & Shu 2002). In a sufficiently settled dust disk, it
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

may be possible to get gravitational instabilities that can produce small (10–100 km)
bodies, which will be rock/ice rather than gas. This would be somewhat ironic because
one might end up insisting that gas giant planets must be produced by gathering
planetesimals and not by gravitational instability, but the planetesimals themselves
could be formed by direct collapse. Until there is more convincing evidence for
gravitational collapse at the right mass scales, however, it is possible to dismiss this
part of the discussion as purely theoretical.
There does seem to be one case where such evidence for direct gravitational
collapse of planetary mass objects may already exist, and that is the case of what
have been called free-floating planets (e.g., Lucas & Roche 2000, Martı́n et al. 2001).
Isolated objects are found at increasingly low masses, and some are apparently below
the fusion limit. There is no theoretical reason why the star-formation process (rapid
collapse from an interstellar cloud) should suddenly fail at the fusion limit. This is one
of the arguments given for insisting on planetesimals to define planets: Do we want
to suddenly change the name of objects forming by direct collapse just because one
(by now minor) source of luminosity (fusion) has given out? Another type of evidence
for the collapse scenario is discussed below—giant exoplanet candidates imaged far
from their stellar hosts.

CULTURE: THE COMMON UNDERSTANDING OF “PLANET”

Until the 1990s, there was not much controversy about the definition of “planet.” It
had been revised at the end of the seventeenth century to its current meaning (many
dictionaries still retain the astrological understanding of the word, too). We start our
section on culture by looking at a few dictionary definitions, which together capture
much of the general public impression.

The Oxford English Dictionary offers:

Mod. Astron. The name given to each of the heavenly bodies that revolves in approximately
circular orbits round the sun (primary planets), and to those that revolve round these
(secondary planets or satellites).
The primary planets comprise the major planets, of which nine are known, e.g., in order
of distance from the sun, Mercury, Venus, the Earth, Mars, Jupiter, Saturn, Uranus,

206 Basri · Brown

 ANRV273-EA34-07 ARI 21 March 2006 7:58

Neptune, and Pluto, and the minor planets or asteroids, the orbits of most of which lie
between those of Mars and Jupiter. [Emphases are in the original]

Dictionary.com offers:
A nonluminous celestial body larger than an asteroid or comet, illuminated by light from
a star, such as the sun, around which it revolves. In the solar system there are nine known
planets: Mercury, Venus, Earth, Mars, Jupiter, Saturn, Uranus, Neptune, and Pluto.

Merriam-Webster (online) offers:

b (1) : any of the large bodies that revolve around the sun in the solar system; (2) : a
similar body associated with another star.

It is interesting to note that the OED seems to include asteroids, and even moons,
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

in the general definition, and also includes “approximately circular orbits” (a nod to
cosmogony, unwitting or not). Apparently, neither the Kuiper belt nor exoplanets have
worked their way into this exalted source yet. The other sources want planets to be
“large” (bigger than asteroids or comets). There is an instance of a requirement on lu-
minosity sources. Pluto is definitely included in the first two sources, but the third has
a less Solar System–centric point of view. All require that planets orbit stars, and this
seems to be the strongest agreed-upon characteristics of planets wherever you look.
Culturally, there are a few additional important planetary characteristics that ap-
pear to have near universal support. Not surprisingly, all are difficult to quantify, but
most people have no difficulty understanding them. At one time, many people might
have suggested that planets need to be in orbits similar to those of the Solar System.
The fact that few worry about this aspect of exoplanets anymore is an interesting
indicator of how fast cultural perceptions can (and should) change in the face of new
information. One criterion that most people appear to hold is that planetary status
should be special in some way. In the Solar System, the number of planets is sufficiently
small that many people know them all. Any planetary definition that radically increases
the number of planets in the Solar System will destroy this important cultural aspect
of planets. In extrasolar planetary systems, we are unlikely to reach the point anytime
soon that we are in danger of violating this criterion. We are already comfortable
talking about extrasolar debris disks and their parent populations of small bodies.
A book is about to appear that treats the cultural history of the word “planet” in far
more detail than we can here. It is titled Is Pluto a Planet? A Historical Journey through
the Solar System, by David Weintraub, and it contains a superb summary of all the
historical meanderings and events that have led to the current situation. You might be
surprised to learn of (or be reminded of) all the definitions and planets themselves that
have been proposed and withdrawn over the centuries. The book also treats current
discoveries, and this debate, in some detail.

THE PLUTO CONTROVERSY

Within the Solar System, the question of “What is a planet?” has recently resurfaced
with the discovery of KBOs ( Jewitt & Luu 1993). When it became clear that Pluto
was not an odd isolated object on a unique orbit, but simply one of the largest known

www.annualreviews.org • Planetesimals To Brown Dwarfs 207

 ANRV273-EA34-07 ARI 21 March 2006 7:58

of a substantial population, some scientists invoked a circumstantial argument and

suggested that it is now wrong to still call Pluto a planet. They suggest that it should
be reclassified as the second largest known KBO. A similar reclassification happened
with Ceres and the asteroids once it was realized that they, too, are members of a
much larger population.
Traditionalists object that today Pluto has strong cultural status as a planet in a
way that the asteroids never had in the early 1800s. The characteristics planetary
definition, whereby all spherical primary objects orbiting stars should be considered
planets, poses a conundrum for traditionalists. It retains Pluto as a planet, but also
Ceres and perhaps tens to hundreds of KBOs. This has the cultural advantage of
saving Pluto, but that is strongly offset by the cultural disadvantage of suddenly
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

increasing the number of planets by as much as an order of magnitude. As noted

Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

above, sphericity is just one of the transitions that occur with increasing mass. The
more geophysically important boundaries are at significantly higher mass and would
exclude all of the asteroids and known KBOs.
Pluto’s situation clearly shows that there is no way to be scientifically consistent
and also satisfy tradition in this case. The dynamical definition, while scientifically
consistent, flies in the face of the apparent cultural desire to retain Pluto in the pan-
theon of planets. The spherical definition, while scientifically consistent and easily
applied, does perhaps more damage to tradition than the simple reclassification of
Pluto. No one has developed a logical definition of “planet” that retains the nine
historical planets and adds no new ones. There are only two solutions: give up on
culture or give up on scientific consistency. For scientists, the inclination is to re-
gard consistency as more important than culture (GB’s position), but in the realm
of planets where culture has a serious impact, it is worth considering giving up on
consistency (MB’s position). A final alternative would be to compose a consistent
general scientific definition, but agree to apply a cultural definition within our Solar
System.

ISOLATED OBJECTS WITHOUT FUSION

Beyond the Solar System, one of the reasons that the debate on “What is a planet?”
arose is the discovery of isolated objects that are probably, but not certainly (due
to current imprecision in determining their mass), incapable of fusion. The orig-
inal discoveries were dubbed “free-floating planets” by those in the characteris-
tics arena. Cosmogonists objected that such objects were probably not formed like
planets (although this is not an empirical claim). Clearly, however, if planets must
be in orbit around stars, those in the circumstances arena have a valid empirical
objection.
Modern star-formation theory muddies the waters by finding that objects can form
as part of a small-N cluster, whose components interact gravitationally for a while,
but where the lightest objects will be ejected from the group first. One might then
view them as ejected planets, although they were never in really stable orbits. We
have come to see, however, that the concept of stable orbits is questionable even in
the young Solar System.

208 Basri · Brown

 ANRV273-EA34-07 ARI 21 March 2006 7:58

OBJECTS IN UNEXPECTED ORBITS

It has become apparent during recent announcements of images of exoplanets that
circumstantial arguments are causing some scientists to reject the objects’ planetary
status. The objects found so far have at least several Jupiter masses (Neuhäuser et al.
2005) and are seen 50–100 AU from their stars [or in one case, from a brown dwarf
(Chauvin et al. 2005)]. Some astronomers reject these objects as planets because their
distance from their host stars suggests that they may not have formed in the standard
scenario (which is not thought to operate rapidly enough out there to beat the short
lifetime of gas disks). They are being called “sub-brown dwarf binaries,” and similar
names.
This is an example of circumstance being invoked with a subtext of cosmogony.
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

Presumably, this objection would be mitigated if the objects were found to be on

Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

highly eccentric orbits, suggesting that they obtained their large distance by gravita-
tional scattering with inner giant planets. It will be difficult to establish the orbital
parameters anytime soon, however. Gaps and warps are being detected in debris disks
at similar distances from their stars, which giant planets are invoked to cause. Dis-
comfort with these distant giants, which would not have highly eccentric orbits, has
not yet been expressed much (perhaps because the putative planets have not actually
been seen).
It is worth recalling the time of the first exoplanet discoveries (Marcy & Butler
1998). These have Jovian masses but are in very close orbits around their stars. Gas
giant planets simply were not supposed to be found there (no formation scenario works
in so close). As planets slightly further away were found (where tidal circularization
no longer could operate), their orbits were also found to be mostly eccentric (similar
to binary stars at similar separations). A few astronomers objected that whatever was
being found, it was not planets because massive planets do not have such orbits.
There was initially the possibility that brown dwarfs were being found with low
orbital inclinations (radial velocity only provides lower mass limits), but this was soon
removed as more objects were found. Fairly quickly, however, we came to understand
that eccentric giant planets could be found close in, and we had been biased by
our previous reliance on only the Solar System. This “objection on the basis of
expectation” is being applied once again to the distant massive objects. One wonders
whether objections to planetary status would still arise if the distant objects had less
than a Jupiter mass (despite the fact the standard model would still have a very hard
time producing them).

DEFINING PLANETS
We have now presented all of the reasons we know why the definition of “planet”
has become a contentious and difficult-to-solve dilemma. It is clear that agreement
cannot be reached on a definition before a consensus on the arenas of the definition
is reached. Here, the community seems divided into two broad camps. Those for
whom tradition is less important, while logical consistency and empirical verification
are most important, lean toward characteristics. They tend not to give the Solar

www.annualreviews.org • Planetesimals To Brown Dwarfs 209

 ANRV273-EA34-07 ARI 21 March 2006 7:58

System as great a weight compared with other planetary systems, and they feel that
recent discoveries call for a reopening of the whole question. The purists in this
camp reject any consideration of circumstances, cosmogony, or culture—they prefer
a purely mass-based definition. The primary problem for this approach comes from
culture; there is not a critical mass of cultural support for this position.
Using the influence of gravity to draw the lower mass limit places uncomfortably
(by tradition) small objects into the planetary domain. The number of known planets
in our Solar System instantly rises to something like 12–15 (with the likelihood of
many more to be found), and the extra ones are in the belts (e.g., Quaoar; Brown &
Trujillo 2004). Such a state of affairs is unacceptable to those who believe that the
clear circumstantial differentiation between planets and objects in populations should
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

be the basis for a division.

Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

It is also unacceptable to those for whom tradition (culture) is most important.

They are greatly influenced by the paradigm we have developed over the past few
centuries. They want a definition that preserves both the current list of planets and
our theories about how they were formed. Objects that do not fit comfortably into
these requirements should get other names, and definitions that seem to violate them
should be rejected. This can apply to all three arenas: Planets should not be too small
or too massive compared with the traditional list, they should orbit stars (without too
much company), and they should be formed from planetesimals in disks. It may be that
a majority falls into this camp at the moment (but no proper polling has been done).
The difficulty with the traditional position is in the details. No particular weighting
among the arenas has found general agreement. The Pluto problem poses a substantial
obstacle. Either tradition or logical consistency must be abandoned. It also seems
parochial to base a general definition of planets too closely on what we happen to
see here; it is quite clear by now that other planetary systems may vary widely from
the familiar. It is not clear when exoplanet discoveries should be allowed to dislodge
long-held beliefs derived from the Solar System. Alternatively, one could argue that
we will always know more about the Solar System than any other planetary system,
so basing definitions on the best-studied examples might be wise.
Insistence on a particular cosmogony is also problematic. It is not empirically
verifiable for most exoplanets; their histories remain largely out of reach. There
may be an overlap between formation mechanisms operating over a certain mass
range; there could also be an overlap between planets and brown dwarfs that form
with a given mechanism. We do not really know enough about planet formation
to be confident on these points. Purists in this arena, however, are willing to give
up verifiability. They are also willing to call fusing objects “planets” if formed by
nucleation in a disk, even if they have the same mass as brown dwarfs, and nonfusing
objects “sub-brown dwarfs” if they formed by direct gravitational instability (whether
in a disk or not). In some cases, they would be content to say that an object’s planetary
status cannot be confirmed or refuted empirically. This approach is perhaps more
philosophical than scientific (although science used to be natural philosophy).
We now become more specific about proposed definitions. MB invokes doses of
cosmogony and circumstance, with a strong nod to culture. GB has a preference for
characteristics, with a sprinkle of culture.

210 Basri · Brown

 ANRV273-EA34-07 ARI 21 March 2006 7:58

“That’s a great deal to make one word mean”, said Alice.

“When I make a word do a lot of work like that”, said Humpty-Dumpty, “I always
pay it extra.”
Lewis Carroll, Alice Through the Looking Glass (1872)

MIKE BROWN’S DISCUSSION

Cosmogony is for me the most satisfactory method for defining the difference be-
tween planets and stars/brown dwarfs, even if it is the hardest method to actually
use. In simple cases, like a single star and low-mass planet, for example, it is trivial
to distinguish between the star (which forms by collapse from an interstellar cloud,
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

even if surrounded by a disk after a while) and the planet (which is built later in the
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

star’s circumstellar disk). Planet definitions involving cosmogony get into trouble as
the mass of the secondary object increases. When is the secondary object a planet and
when is the system a double star? We have discussed above the cosmogonies that are
usually meant for planets, and how they differ from star-like formation. This scheme
works in the Solar System, but is difficult to apply empirically for exoplanets.
It seems we are left with an impractical task, but nature has apparently provided a
simple and remarkably powerful observational solution to this difficulty of separating
stars and planets by cosmogony. Radial velocity studies have found that objects in
orbits within a few AU with masses no larger than Jupiter are fairly common, but
the number decreases sharply as mass increases. Between roughly 5 and 60 Jupiter
masses, the mass regime known as the “brown dwarf desert,” there are few close stellar
companions known (yet they would be easier to find). Above that mass, the realm of
double stars is again well populated.
Nature has divided close stellar companions, by some mechanism, into low- and
high-mass populations, with a wide gap between the two. Such a clear dividing line
invites a classification scheme. Jupiter and the rest of the Solar System planets fit
firmly within the low-mass population. It is natural to call this low-mass population
“planets” and the high-mass population “stars.” Note that there need not have been
such a natural dividing line; the two populations could have merged smoothly (putting
us back into the cosmogony conundrum). As we learn more about the populations
more distant from the central star, we may have to change our ideas about where and
how to draw planetary dividing lines. The reason their status is uncertain is because
we truly are uncertain of their provenance, and uncertainty about what to call them is
perhaps the only honest response. We should feel no embarrassment about learning
more and modifying what we previously meant when we said “planet.”
To set the lower mass limit of planets, which for now only matters for the Solar
System, I prefer the dynamical definition that the object has cleared planetesimals
from its neighborhood, and thus is not part of a population. Any view of the Solar
System that sees Pluto or 2003 UB313 as tiny lonely outposts on the outer edge,
or Ceres as the sole miniplanet between Mars and Jupiter, misses one of the more
important points about the architecture of the Solar System and the importance of its
belts. The clearing definition appears to me to be the closest one can come to meshing
the cultural and scientific views of what should be called planets. It is consistent with

www.annualreviews.org • Planetesimals To Brown Dwarfs 211

 ANRV273-EA34-07 ARI 21 March 2006 7:58

the arguments from the past reassignment of Ceres from planet to asteroid. The only
casualty to the currently accepted system is that Pluto must go the way of Ceres and
be likewise reassigned. It seems a minor price to pay for a definition that is close to
satisfying culture while remaining scientifically rigorous.
Experience with the public after the discoveries of Quaoar and Sedna, however, has
slowly taught me not to ignore the importance of culture. Those in favor of absolute
scientific consistency regardless of cultural beliefs (which, until recently, included
me) should also argue that Madagascar, an island sitting on an isolated block of
continental crust in the Indian Ocean, should be called a separate continent. Though
their arguments would be scientifically sound, they would not get very far, even with
geologists. Likewise, it has become apparent in the past few years that the extent of
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

education and media makes it unlikely that the idea of Pluto as a true planet will ever
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

be abandoned. One can argue with culture forever and, apparently, get nowhere. I
thus finally concede: Pluto is a planet because we say it is and for no other reason.
If need be, we can give Pluto an adjective and call it an “historical planet.” All new
Solar System objects bigger than Pluto join the planet club by default. 2003 UB313,
a little larger than Pluto and spectrally similar, is a planet. 2005 FY9, a little smaller
than Pluto but spectrally similar, is not. This one hundred percent cultural definition
requires scientists to relinquish their desire to legislate a new and rigorous meaning
to the Solar System sense of the word “planet” and accept the meanings that culture
has been using all along. Planets are far too important to culture to expect that any
newly legislated definitions will have much sway.

Definition: (a) A planet is an object that is massive enough to clear planetesimals from
its orbital neighborhood and that is part of the empirically defined distinct group of
low-mass stellar companions with masses lower than approximately 5 Jupiter masses. (b)
In the Solar System, a planet is any of the nine historical planets plus any newly found
objects bigger than the smallest of these.

GIBOR BASRI’S DISCUSSION

I have summarized much of my thinking in an article in Mercury magazine (Basri
2003). Most objects in astrophysics are defined on the basis of their physical char-
acteristics, and now that planetary science is moving beyond the Solar System, that
seems like the right thing to do here as well. I do not favor cosmogony as part of the
definition for planets at all, partly because I do not think we know what we are talking
about to a sufficient level of understanding. More importantly, it is almost impossible
to apply empirically outside our Solar System. I do not accept a situation in which a
four-Jupiter-mass object seen 40 AU from a solar-type star might actually either be
a planet or be something else (one could get that circumstance with several different
histories). I do not think that the scientists who are trying to study such objects (and
the many other odd exoplanets) will stand for being told that there is little hope of
deciding whether or not they are studying planets.
Similarly, I do not much care for circumstantial definitions. I view them as just that,
and not related to a fundamental class of astrophysical object. In particular, I think it

212 Basri · Brown

 ANRV273-EA34-07 ARI 21 March 2006 7:58

is time to drop the great observational bias provided by our Solar System. Just as we
should not think the current crop of detected exoplanets is representative of exoplanets
in general, there is no reason to think that the Solar System is representative of
planetary systems in general.
I do find it expedient to make one exception. That is because of culture—the fact
that there is wide agreement that planets should orbit stars (which is a circumstance).
My definition, therefore, is given in two parts. One is purely based on mass (I would
call it the astrophysical part), and the other takes into account the one circumstance
I cannot see culture doing without. My mass-based limits are gravitational shaping
at the lower end and the ability for fusion at the upper end. I would say that these
define the domain of planetary-mass objects, which I have shortened to planemos. I
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

call all objects above this mass range fusors (a discussion of “What is a star?” was not
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

commissioned here). Then, planets are just planemos orbiting fusors. This scheme
includes Pluto and 2003 UB313 as planets, but also Ceres, Quaoar, Sedna, and other
yet-to-be-discovered and/or -named KBOs that are sufficiently round. The recent
work on Ceres (Thomas et al. 2005) only reinforces the idea that Ceres has all the
properties of miniplanet as defined here.
To deal with the cultural desire that planets be special, I greatly favor the use of
adjectives to distinguish between important classes of planets. I would call planets that
are not dynamically dominant miniplanets, thus diminishing the planetary status of
Pluto without eliminating it (and forcing logical consistency with the other miniplan-
ets). I would deal with the cosmogony argument (invoking some of MB’s reasoning)
by calling planets whose cores are becoming degenerate (those more massive than
approximately two Jupiter masses) superplanets. This places all the objects whose
provenance is easily questioned into a separate category, without fully robbing them
of their planetary status. It does not reduce the current count of exoplanets.
Free-floating objects in the right mass range are all planemos, but not planets,
and there is no need to determine their cosmogony to classify them. Some might in-
deed be ejected planets, but one would not call them that unless there were empirical
reasons for doing so. There are no (insuperable) empirical obstacles to determining
the status of objects in this scheme, and much of the cultural heat could be defused.
The large moons in our Solar System (including our own) are also planemos, but not
planets. Planetary scientists, as a practical matter, already study planemos rather than
exclusively planets. I believe this scheme answers all the questions and requirements
posed at the beginning, is relatively easy to implement, and could garner widespread
support (if enough people are prepared to reconsider their traditional preconcep-
tions). As scientists, I think we should remain consistent with our basically empirical
and logical roots. I also believe that if we lead on this issue, the public will follow.

Definitions:
Planet: A planemo that orbits a fusor.
Planemo: A round nonfusor.
Fusor: An object capable of core fusion.

One could of course drop the extra new words and combine everything into a
longer sentence. Legal disclaimers: “round” means “whose surface is very nearly on

www.annualreviews.org • Planetesimals To Brown Dwarfs 213

 ANRV273-EA34-07 ARI 21 March 2006 7:58

the gravitational equipotential”; “orbits” means “whose primary orbit is now, or was
in the past around”; and “capable” implies fusion is possible sometime during the
existence of the object by itself.

CLOSING REMARKS
A number of scientists have offered opinions on the definition of “planet.” Many,
including the IAU Working Group on Extrasolar Planets, have accepted the fusion
limit as setting the upper bound for planets. Some reject this because deuterium
burning is weak at the boundary or because they prefer cosmogony to an upper mass
bound. The idea of something like planemos has also been suggested by Nick Wolff in
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

the form of an acronym: PMOs (but with semiarbitrary mass boundaries, and favoring
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

cosmogony). Most reject free-floating objects as planets, but accept them as having
“planetary mass” (although some stick with a purely mass-based definition and call
them planets). The lower bound for planets is often suggested to be gravitationally
induced roundness, but arbitrary or other size limits that keep the Solar System
census unchanged have some support. The IAU has tried to avoid this issue but is
being forced to revisit it by the discovery of 2003 UB313. Perhaps by the time this
review appears a new ruling will have been made, but it is clear that no ruling could
satisfy everyone.
Many consider being a member of a belt population to be a disqualifying cir-
cumstance, and most of that group would drop Pluto’s planetary status. Others
would grandfather it in for tradition’s sake or refer to it with a dual status: small-
est planet/largest KBO. Stern & Levinson (2002) have proposed something very
similar to GB, but with a full range of astronomical adjectives transferring over: sub-
dwarf, dwarf, subgiant, giant, supergiant. The subdwarf boundary comes where they
calculate insufficient dynamical dominance, and their other boundaries are set at log-
arithmic intervals in Earth masses (corresponding roughly to terrestrial planets, ice
giants, gas giants, and superplanets).
It appears in the end that well-defined boundaries, empirical verifiability, or logical
consistency as requirements on a definition cannot necessarily overcome gut feelings
(based on culture and tradition) that certain objects just do not belong in the category
of “planets,” whereas others just do. The bases of such feelings are not always the
same, and they arise from the different weights given to the different arenas in which
the debate is held. This is the essentially human nature of the problem. There do
seem to be acceptable solutions, but agreeing on any one of them is difficult. We hope
that this essay makes the problem clearer, and that substantial further reflection by
everyone concerned, coupled with the ongoing flood of new discoveries, lights the
path to a consensus. In the meantime, the debate itself keeps the field fresh and the
public informed and engaged.

LITERATURE CITED
Basri G. 2000. Observations of brown dwarfs. Annu. Rev. Astron. Astrophys. 38:485–
519

214 Basri · Brown

 ANRV273-EA34-07 ARI 21 March 2006 7:58

Basri G. 2003. What is a planet? Mercury 32:27–34

Bodenheimer P. 2004. Planet formation in disks. Star formation in the interstellar
medium. ASP Conf. Ser. 323:329–36
Bond IA, Udalski A, Jaroszynski M, Rattenbury NJ, Paczynski B, et al. 2004. OGLE
2003-BLG-235/MOA 2003-BLG-53: a planetary microlensing event. Astrophys.
J. Lett. 606:L155–58
Borucki WJ, Koch DG, Dunham EW, Jenkins JM. 1997. The Kepler Mission: a
mission to determine the frequency of inner planets near the habitable zone of a
wide range of stars. In Planets Beyond the Solar System and the Next Generation of
Space Missions, ASP Conf Ser. 119:153–73. San Francisco: Astron. Soc. Pacific
Boss AP. 1997. Giant planet formation by gravitational instability. Science 276:1836–
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

39
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

Brown ME, Trujillo CA. 2004. Direct measurement of the size of the large Kuiper
belt object (50000) Quaoar. Astron. J. 127:2413–17
Brown ME, Trujillo CA, Rabinowitz DL 2005. Discovery of a planetary-sized object
in the scattered Kuiper belt. Astrophys. J. Lett. In press
Burrows A, Hubeny I, Hubbard WB, Sudarsky D, Fortney JJ. 2004. Theoretical
radii of transiting giant planets: the case of OGLE-TR-56b. Astrophys. J. Lett.
610:L53–56
Chauvin G, Lagrange A-M, Dumas C, Zuckerman B, Mouillet D, et al. 2005. Giant
planet companion to 2MASSW J1207334-393254. Astron. Astrophys. 438:L25–
28
Jewitt D, Luu J. 1993. Discovery of the candidate Kuiper belt object 1992 QB1.
Nature 362:730–32
Lucas PW, Roche PF. 2000. A population of very young brown dwarfs and free-
floating planets in Orion. MNRAS 314:858–64
Mannings V, Boss AP, Russell SS. 2000. Protostars and Protoplanets IV. Tucson: Univ.
Ariz. Press
Marcy GW, Butler PR. 1998. Detection of extrasolar giant planets. Annu. Rev. Astron.
Astrophys. 36:57–98
Marcy GW et al. 2001. Two Substellar Companions orbiting HD 168443. Astrophys.
J. 555:418–25
Martı́n EL, Zapatero Osorio MR, Barrado y Navascués D, Béjar VJS, Rebolo R.
2001. Keck NIRC observations of planetary-mass candidate members in the
sigma Orionis open cluster. Astrophys. J. Lett. 558:L117–21
Mazeh T, Zucker S, Pont F. 2005. An intriguing correlation between the masses and
periods of the transiting planets. MNRAS 356:955–57
Nelson RP, Papaloizou JCB. 2004. Orbital migration and disc-planet interactions.
In Extrasolar Planets: Today and Tomorrow, ASP Conf. Ser. 321:367–78. San Fran-
cisco: Astron. Soc. Pac.
Neuhäuser R, Guenther EW, Wuchterl G, Mugrauer M, Bedalov A, Hauschildt
PH. 2005. Evidence for a co-moving sub-stellar companion of GQ Lup. Astron.
Astrophys. 435:L13
Thomas PC, Parker JW, McFadden LA, Russell CT, Stern SA, et al. 2005. Differ-
entiation of the asteroid Ceres as revealed by its shape. Nature 437:224

www.annualreviews.org • Planetesimals To Brown Dwarfs 215

 ANRV273-EA34-07 ARI 21 March 2006 7:58

Sato B, Fischer DA, Henry GW, Laughlin G, Butler RP, et al. 2005. The N2K
Consortium. II. A Transiting Hot Saturn around HD 149026 with a Large Dense
Core. Astrophys. J. 633:465
Stern SA, Levinson HF. 2002. Regarding the criteria for planethood and pro-
posed planetary classification schemes. In Highlights of Astronomy, 12:205–13.
San Francisco: A.S.P.
Weintraub D. 2005. Is Pluto a Planet? A Historical Journey through the Solar System.
Princeton, NJ: Princeton Press. In press
Wolszczan A, Frail DA. 1992. A planetary system around the millisecond pulsar
PSR1257 + 12. Nature 355:145–47
Youdin AN, Shu FH. 2002. Planetesimal formation by gravitational instability. As-
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

trophys. J. 580:494–505
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

216 Basri · Brown

Contents ARI 20 March 2006 8:25

Annual Review
of Earth and
Planetary

Contents Sciences

Volume 34, 2006

Threads: A Life in Geochemistry

Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

Karl K. Turekian p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 1
Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

Reflections on the Conception, Birth, and Childhood of Numerical

Weather Prediction
Edward N. Lorenz p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p37
Binary Minor Planets
Derek C. Richardson and Kevin J. Walsh p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p47
Mössbauer Spectroscopy of Earth and Planetary Materials
M. Darby Dyar, David G. Agresti, Martha W. Schaefer, Christopher A. Grant,
and Elizabeth C. Sklute p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p83
Phanerozoic Biodiversity Mass Extinctions
Richard K. Bambach p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 127
The Yarkovsky and YORP Effects: Implications for Asteroid Dynamics
William F. Bottke, Jr., David Vokrouhlický, David P. Rubincam,
and David Nesvorný p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 157
Planetesimals to Brown Dwarfs: What is a Planet?
Gibor Basri and Michael E. Brown p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 193
History and Applications of Mass-Independent Isotope Effects
Mark H. Thiemens p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 217
Seismic Triggering of Eruptions in the Far Field: Volcanoes and
Geysers
Michael Manga and Emily Brodsky p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 263
Dynamics of Lake Eruptions and Possible Ocean Eruptions
Youxue Zhang and George W. Kling p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 293
Bed Material Transport and the Morphology of Alluvial River Channels
Michael Church p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 325
Explaining the Cambrian “Explosion” of Animals
Charles R. Marshall p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 355

vii
Contents ARI 20 March 2006 8:25

Cosmic Dust Collection in Aerogel

Mark J. Burchell, Giles Graham, and Anton Kearsley p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 385
Using Thermochronology to Understand Orogenic Erosion
Peter W. Reiners and Mark T. Brandon p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 419
High-Mg Andesites in the Setouchi Volcanic Belt, Southwestern Japan:
Analogy to Archean Magmatism and Continental Crust Formation?
Yoshiyuki Tatsumi p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 467
Hydrogen Isotopic (D/H) Composition of Organic Matter During
Diagenesis and Thermal Maturation
Arndt Schimmelmann, Alex L. Sessions, and Maria Mastalerz p p p p p p p p p p p p p p p p p p p p p p p p p p 501
Annu. Rev. Earth Planet. Sci. 2006.34:193-216. Downloaded from www.annualreviews.org

The Importance of Secondary Cratering to Age Constraints on

Access provided by Physical Research Laboratory on 09/05/23. For personal use only.

Planetary Surfaces
Alfred S. McEwen and Edward B. Bierhaus p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 535
Dates and Rates: Temporal Resolution in the Deep Time Stratigraphic
Record
Douglas H. Erwin p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 569
Evidence for Aseismic Deformation Rate Changes Prior to
Earthquakes
Evelyn A. Roeloffs p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 591
Water, Melting, and the Deep Earth H2 O Cycle
Marc M. Hirschmann p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 629
The General Circulation of the Atmosphere
Tapio Schneider p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 655

INDEXES

Subject Index p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 689

Cumulative Index of Contributing Authors, Volumes 24–34 p p p p p p p p p p p p p p p p p p p p p p p p p p p 707
Cumulative Index of Chapter Titles, Volumes 24–34 p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 710

ERRATA

An online log of corrections to Annual Review of Earth and Planetary Sciences

chapters may be found at http://earth.annualreviews.org

viii Contents