Pluto and the - Recherche en IA spécialisée au MIT

Pluto and the
‘Planet Problem’:
Folk Concepts and
Natural Kinds in
Astronomy

Alisa Bokulich
Université de Boston

The Pluto controversy provides fertile new ground to revisit the traditional
philosophical problem of natural kinds and scientiªc change. Here I show
that further insight into the Pluto case is gained by drawing out some of the
striking analogies with what is termed the “species problem” in the philoso-
phy of biology. I argue that the taxon ‘planet’ can still be considered a natu-
ral kind term despite the fact that a) its meaning and extension have
changed over time, b) there are multiple scientiªcally compelling deªnitions of
planet, and c) many of those deªnitions include historical and/or relational
properties.

1. Introduction
Le 2006 decision by the International Astronomical Union to strip Pluto
of its status as a planet generated considerable uproar not only in scientiªc
circles, but among the lay public as well. After all, how can a vote by
424 scientists in a conference room in Prague undermine what every well-
educated second grader knows is a scientiªc fact?

The Pluto controversy provides a new and fertile ground in which to re-
visit the traditional philosophical problems of natural kinds and scientiªc
changement. Before engaging these philosophical problems, cependant, there are
two misguided reactions to the Pluto controversy worth dispelling from
the start. The ªrst misguided reaction is that this sort of classiªcatory
about-face is unprecedented in science. To the contrary, many such

This paper was ªrst presented as a talk in the Boston Colloquium for Philosophy of Sci-
ence in 2008; I am grateful to Richard Boyd and other audience members there for their
feedback. I would also like to thank Dr. Arielle Moullet of the National Radio Astronomy
Observatory for her help making sure the astronomy details cited in this paper are accurate
and up to date. Any remaining errors are, bien sûr, my own.

Perspectives on Science 2014, vol. 22, Non. 4
©2014 by The Massachusetts Institute of Technology

est ce que je:10.1162/POSC_a_00146

464

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

465

reclassiªcations can be found, Par exemple, in the biological sciences:
whales are not ªsh, koala bears are not bears, and more recently modern
birds are actually dinosaurs, hence the dinosaurs did not all go extinct. Dans-
deed, as we shall see later on, this isn’t even the ªrst time that the notion
of planet has undergone signiªcant revision. Science is in the business of
creating taxonomies, and with new scientiªc discoveries these taxonomies
and their extensions can change.

The second misguided reaction to the Pluto controversy is that it is
merely a matter of semantics. On this view, it is simply a question of how
we use words—nothing substantive—let alone scientiªc—is at stake in
the debate. As Shakespeare’s Juliet proclaims “What’s in a name? That
which we call a rose by any other name would smell as sweet.” The scien-
tists working in planetary science, cependant, are at pains to disagree. Pour
example, Alan Stern, who is head of NASA’s “New Horizons” Pluto-
Kuiper Belt mission, and his colleague Harold Levinson write,

Classiªcation schemes have played and continue to play a useful
role in astronomy because they allow us to put bodies into a struc-
ture with recognizable correlations of physical parameters. [. . .]
While the discussion that has gone on may at times have centered
on semantics, we believe that classiªcation schemes play an impor-
tant role in the scientiªc process. (Stern and Levinson 2002, 1)

Taxonomies, when they are done correctly, are not superºuous to science,
but rather play an important role in the organization of scientiªc knowl-
bord, the prediction of new phenomena, and the formulation of scientiªc
explanations. Some taxonomies are clearly better than others, and it
should go without saying that not any random collection of objects will be
useful for scientiªc reasoning.

Granting then that taxonomies are important to science, the next ques-
tion becomes what is it that distinguishes a scientiªcally useful taxonomy
from an unhelpful one? The traditional philosophical answer to this ques-
tion is that the successful scientiªc taxonomies are those whose taxons cor-
respond to natural kinds. On this view, there are objective groupings of
things in the universe into categories, and a natural-kind taxonomy is one
that correctly “carves nature at its joints,” to use Plato’s apt metaphor.1 As
we shall see, cependant, there is considerable disagreement in the philosoph-
ical literature concerning how precisely natural kinds should be under-
stood. On the traditional view, natural kinds are deªned by a set of neces-
sary and sufªcient properties that are possessed by all and only members of
the kind; these properties must be intrinsic (not relational or historical)

1. Plato, Phaedrus 265d–266a.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

466

Pluto and the ‘Planet Problem’

and together they deªne the unchanging essence of the kind.2 So, for ex-
ample, the chemical element potassium is taken to be a natural kind
whose essence is to have an atomic number of 19 (c'est, it has 19 protons
in its nucleus). The intrinsic property of having exactly 19 protons in its
nucleus is a necessary and sufªcient condition for being potassium, and all
and only those substances which are potassium have exactly 19 protons.

Following the work of Richard Boyd, cependant, there has emerged a
new conception of natural kinds that is rapidly gaining ground. On this
voir, a natural kind is deªned by what Boyd calls a “homeostatic property
cluster” (HPC) (Boyd 1999). HPC natural kinds are deªned as a family of
properties that are clustered in nature as a result of various underlying
homeostatic or causal mechanisms. Unlike the traditional view, ces
properties need not be intrinsic, but can be relational or historical as well.
Furthermore Boyd grants that the extensions of these kinds may have
vague boundaries, and this vagueness counts neither against the kind-
hood nor naturalness of the kind. What is essential is that there be what
he calls an ‘accommodation’ between these HPC kinds and the causal
structure of the world; such an accommodation is important because it is
what makes successful inductions and scientiªc explanations regarding
these kinds possible.

This philosophical question over how we are to understand natural
kinds underlies much of the contemporary debate in astronomy concern-
ing how to deªne ‘planet’. Although astronomers and planetary scientists
do not use the language of natural kinds, one of the chief issues dividing
the planet debate is between those who think that the term planet should
be deªned solely in terms of intrinsic properties (such as mass limits), et
those who think that historical properties (speciªcally how the body was
formed) and relational properties (whether the body is orbiting a star) sont
essential to the deªnition of the kind. As we shall see, there are many
striking similarities between the planet debate in astronomy and a long-
standing debate in the biology literature concerning how species should
be deªned, known as the “species problem.” By tracing this analogy be-
tween the species problem and what I am calling the “planet problem” in
some detail, the current controversy over Pluto can be proªtably placed
into broader philosophical context.

2. An Analogy to the Species Problem
By way of setting up the planet problem let me give a very brief overview
of the species problem. To those outside of biology and the philosophy of
biology, species would appear as quintessential examples of natural kinds.

2. A relational property is a property a thing has in relation to something else, while an

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

467

Since the Darwinian revolution, cependant, the realization that species arise
through a complex process of speciation challenges the traditional
essentialist view that there is some set of properties that all and only
members of given species possess. As Marc Ereshefsky explains,

A number of forces conspire against the universality and uniqueness
of a trait in a species. Suppose a genetically based trait were found
in all the members of a species. The forces of mutation, recombina-
tion and random drift can cause the disappearance of that trait in a
future member of the species. [. . .] Again biological forces work
against the uniqueness of a trait within a single species. Organisms
in related species inherit similar genes and developmental programs
from their common ancestors. [. . .] Another source of similar traits
in different species is parallel evolution. Species frequently live in
similar habitats with comparable selection pressures. (Ereshefsky
2007, 3)

Not only is it the case that there is rarely a set of unique properties de-
ªning a species, but contemporary biologist also deny that species can be
uniquely individuated on the basis of their genetic material. It turns out
that variability in DNA is often not a good indicator of morphological
variability nor of reproductive compatibility (Brookes 1998, 3). A strik-
ing example of this is the fact that there is a greater genetic difference be-
tween sibling species of drosophila fruit ºy than there is between humans
and chimpanzees (King and Wilson 1975). Despite the intuitions of many
philosophers, it does not appear that the essences of species are to be found
in their “microstructure.”3

The “species problem” refers to the fact that there is no single agreed-
upon way to deªne biological species. UN 1997 book by Michael Claridge
and collaborators surveyed more than a dozen different ways in which bio-
logical species can be deªned, each with its own scientiªc merits and limi-
tations. The most well-known perhaps is Ernst Mayr’s Biological Species
Concept (BSC), which deªnes species in terms of reproductive compatibil-
ity and an ability to produce fertile offspring. Problems with the BSC
deªnition of species include that it does not work for species that repro-
duce asexually, it is inapplicable to paleospecies, reproductive isolation

intrinsic property is a property a thing has in itself. Donc, Par exemple, I have the relational
property of being a mother, but the intrinsic property of being brown eyed.

3. This “microstructure” view of essences has been endorse at various points by philoso-
phers ranging from John Locke to Saul Kripke and Hilary Putnam; see Bird and Tobin
(2008) for a review.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

468

Pluto and the ‘Planet Problem’

occurs gradually making for vague boundaries, and hybrids—which are
quite common in the plant world—are often not weak or sterile.

Philip Kitcher (1984) has noted that one can distinguish two broad
“families” of approaches to deªning species. The ªrst approach emphasizes
morphology or structural similarities as being of paramount importance,
while the second approach emphasizes phylogenetic relationships—that
est, history and genealogy override similarity. In this latter category is in-
cluded cladistics, which deªnes species as comprising a common ancestor
and all its descendents. Advantages of the morphological approach (aussi
sometimes referred to as the classical or Linnean approach to species) dans-
clude that it is based on intrinsic and observable properties, while the dis-
advantages include that is not based on our current theoretical under-
standing of how species arise, it admits of vague boundaries, and has
difªculties handling things such as sexual dimorphism (c'est, females
of different species may be more structurally similar to each other than
they are to the males of their own species). Phylogenetic approaches, sur
the other hand, emphasize relational properties such as having a common
ancestor.

In an inºuential series of articles from the 1970s the biologist Michael
Ghiselin (1974) and the philosopher of biology David Hull (1978) argued
that species should not be thought of as natural kinds, classes, or univer-
sals at all; rather they should be thought of metaphysically as individuals.4
What is relevant here is not their positive argument that species should
be conceived of as individuals, but rather their negative arguments that
species are not the sort of thing that can be properly thought of as kinds.
Speciªcally one can ªnd the following four reasons cited for why species
are not kinds. D'abord, attempts to deªne species in terms of some ªxed set of
intrinsic properties fail. Deuxième, species have vague boundaries; c'est,
there are borderline cases where it seems no amount of information would
settle whether a particular organism or set of organisms does or does not
belong to the species. Troisième, species membership is spatio-temporally re-
stricted: species not only come into existence and go extinct, but they are
importantly historical entities, whose member organisms are connected
by a particular pattern of descent. To borrow an example from Joseph
LaPorte, even if one were to ªnd on Alpha Centauri an animal that is
genetically and morphological indistinguishable from horses here on
earth, they would still not be horses because they do not share the same

4. Species are individuals in the sense of having proper names, not having deªning
properties (intensions), there cannot be instances of them, and their constituent organisms
are parts not members (Ghiselin 1974, 536).

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

469

evolutionary history (Laporte 2004; see Hull 1978, 349 for a similar ex-
ample). This is in contrast with “proper” natural kinds, such as iron, que
are spatio-temporally unrestricted—any atom in the universe found to have
26 protons would count as iron. A fourth reason cited for why species can-
not be kinds is that species evolve whereas abstract kinds cannot (voir
Ghiselin 1987, 129; Ghiselin 1981, 303; and Laporte 2004, 9–10 for a
discussion).

More recently a number of philosophers, such as LaPorte (2004), Boyd
(1999), and Paul Grifªths (1999), have countered these objections, argu-
ing that they are not in fact obstacles to viewing biological species as nat-
ural kinds. Before examining the responses to these objections, cependant,
let us turn to the planet problem, where we will see that these same four
objections can be leveled against planet being considered a natural kind as
well. The striking points of analogy between the ‘species problem’ and the
‘planet problem’ are summarized in Figure 1 below.5 Although there is no
simple solution to the species problem that we can import to solve the
planet problem, the value of drawing out this analogy is three-fold. D'abord,
just as it was helpful for biologists and philosophers of biology to be able
to recognize, label, and explicitly discuss the species problem, it is simi-
larly helpful for those interested astronomy and planetary science to have
an explicit framing of the planet problem. Deuxième, debates in the astron-
omy literature, about whether historical and relational properties are ad-
missible in the deªnition of astronomical kinds, would be advanced by
recognizing that historical and relational properties have ªgured promi-
nently in the deªnitions of kinds in other sciences, such as biology. Troisième,
it is important for philosophers of science more generally to see that the
species problem may not in fact be an anomaly in science, but rather part
of a larger pattern that is more common, especially in what is termed the
“historical sciences.”

3. The Planet Problem
One can ªnd almost as many competing deªnitions of planet in the as-
tronomy literature as one can ªnd deªnitions of species in the biology lit-
erature. Most of these deªnitions for planet can be grouped into one of the
following two families of approaches: The ªrst group consists of those

5. One point of disanalogy between the species problem and the planet problem is that
one can distinguish the species category in general from speciªc species taxa (such as felis
catus), the latter of which consists of the concrete entities, whereas in the case of the taxon
‘planet’ the members are the concrete entities themselves (although there are planets with
different characteristics, they are not recognized as constituting their own taxanomic cate-
gories or subkinds in the same way.)

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

470

Pluto and the ‘Planet Problem’

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Chiffre 1. Analogies between the “Species Problem” and the “Planet Problem.”
Should we conclude that ‘species’ and ‘planet’ are not natural kind terms?

who think planet should be deªned solely in terms of intrinsic proper-
liens, while the second group emphasizes historical and relational prop-
erties, such as how the body was formed and whether it has dynamically
dominated the other bodies in its orbit. While the ªrst approach tends to
be favored by planetary geologists (more generally referred to as planetary
scientists), the second approach is favored by a group of astronomers gen-
erally referred to as dynamicists.

It is interesting that one of the few points of agreement between these
different groups is that the interior structure and composition of a planet
is irrelevant to the deªnition of its kind. In our own solar system, for ex-
ample, the composition of the planets varies signiªcantly. There are the
four “terrestrial” or “rocky planets”: Mercury, Venus, Terre, and Mars,
which are made of heavy elements with iron cores; then there are the two
“gas giants,” Jupiter and Saturn, which are largely made of hydrogen and
helium; ªnally, there are the two “ice giants,” Uranus and Neptune, lequel
are composed primarily of water, ammonia, and methane ices, with con-
siderably less hydrogen and helium gases than the other giant planets.
Other likely intrinsic properties such as having an atmosphere or having a
magnetic ªeld also fail to capture planethood, for example ruling out Mer-
cury and Venus respectively.

Perspectives on Science

471

Those who want to deªne planet in terms of intrinsic properties tend to
focus either on the shape of the object or on speciªc mass limits. The orig-
inal deªnition of planet proposed by the International Astronomical
Union on August 16, 2006—a proposal that was later voted down and re-
placed on August 24—focused on the object’s shape, speciªcally whether
the force of gravity of the body exceeds its material strength, forcing it to
take a spherical shape. The proposal speciªcally deªned planet as follows:

A planet is a celestial body that (un) has sufªcient mass for its self-
gravity to overcome rigid body forces so that it assumes a hydro-
static equilibrium (nearly round) shape, et (b) is in orbit around a
star, and is neither a star nor a satellite of a planet. (http://www
.astronomy2006.com/press-release-16-8-2006-nine.php)

In addition to being an intrinsic and easily observable property, an advan-
tage of this deªnition is that the threshold at which an object becomes
round is scientiªcally important insofar as it marks the transition between
an object that is geophysically dead to one that is active, having convec-
tion occur in its interior (Basri and Brown 2006, 198). Bien que le
shape of a body is correlated with its mass, it is not identical to a mass-
based deªnition of planet insofar as the mass at which an object become
round varies with the substance of the body; Par exemple, icy bodies can
reach hydrostatic equilibrium at much smaller masses than rocky bodies.
If this deªnition of planet had been adopted then not only would Pluto
still be considered a planet, but three more planets would have been added
to our solar system as well: Ceres, which is traditionally thought of as an
asteroid,6 Charon, which is traditionally thought of as a moon of Pluto,
and Eris, a trans-Neptunian object in the scattered disk.7 A number of ob-
jections have been raised against this deªnition of planet, cependant. D'abord, un
qualitative characteristic such as roundness forms a continuum, and hence
is vague with borderline cases. As one astrophysicist criticized, “[D]oes
gravity dominate the shape of a body if the cross-section deviates from hy-
drostatic equilibrium by 10% or by 1%? Nature provides no unoccupied
gap between spheroidal and nonspheroidal shapes, so any boundary would
be an arbitrary choice” (Soter 2006, 2513). Another objection to deªning

6. Ironically, when Ceres was ªrst discovered in 1801 it was believed to be a planet—
the eighth planet—however, it was soon realized that it was part of a vast population of
rocks that came to be known as the asteroid belt.

7. “Trans-Neptunian object” (TNO) is a general term used to refer to any object in our
solar system that orbits the sun at a distance greater than Neptune. It is divided into two
régions: the Kuiper Belt and, at the outer limit of the solar system, the Oort Cloud. Là
is a third class of TNOs called the “Scattered Disk,” which consists of highly inclined or
eccentric orbits that can be, at various times, in different regions.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

472

Pluto and the ‘Planet Problem’

a planet in terms of roundness was raised by Mike Brown, the Caltech as-
tronomer who was involved in the discovery of Sedna and Eris. Although
the International Astronomical Union claimed this deªnition would rec-
ognize only 12 planets, Brown argued that there were in fact 53 known ce-
lestial bodies that would satisfy this deªnition, with likely over a hundred
more to be discovered as astronomers continue to survey the Kuiper belt,
leading to almost 200 hundred planets in our solar system alone.8 This
deªnition, which Brown informally referred to as the “leave-no-ice-ball-
behind proposal,” was on his view too permissive in admitting members
to the kind.

A similar deªnition of planet based on intrinsic properties is in terms of
upper and lower mass limits. This approach has been defended by Stern
and Levison who deªne a planet as

[UN]ny body in space that satisªes the following testable upper and
lower bound criteria on its mass: [. . .] the body must: (1) Be low
enough in mass that at no time (past or present) can it generate en-
ergy in its interior due to any self-sustaining nuclear fusion chain
reaction (else it would be a brown dwarf or a star). And also, (2) Be
large enough that its shape becomes determined primarily by grav-
ville [. . .]. (Stern and Levison 2002, 4).

Although Stern and Levison formulate their deªnition in terms of mass
limits, they still rely on the hydrostatic equilibrium (“round shape”) con-
dition to deªne the lower mass bound. The difªculty, as was noted before,
is that the roundness condition is not in fact equivalent to a mass limit,
since it is a function of the density and compressive strength of the partic-
ular material. What sets this deªnition apart from the previous deªnition
is that it does not include what is known as the “circumstance” condition,
namely that a planet must be a body orbiting a star. Any body sufªciently
massive to be round and less massive than a star counts as a planet—no
matter where in the universe that body is.

By focusing on intrinsic properties alone, Stern and Levison have man-
aged to deªne the kind planet in a spatiotemporally unrestricted way,
however this leads to a number of counterintuitive consequences. Not
only does this deªnition admit the 100 or so objects recognized by the
previous deªnition of planet, but it admits at least 100 plus, y compris,
Par exemple, the Earth’s moon, and at least six other moons in our solar
système. Ironically our moon, which had long ago been considered a planet
(along with the Sun) on the geocentric Ptolemaic system, would—on this

8. Mike Brown (n.d.) The Eight Planets. http://web.gps.caltech.edu/(cid:2)mbrown/

eightplanets/.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

473

deªnition—once again be considered a planet (though of course without
the geocentrism). For many, cependant, any deªnition of planet that classi-
ªes a moon as a planet fails as an adequate deªnition.

In formulating their deªnition of planet Stern and Levison ªrst outline
a set of desiderata that they believe any deªnition for planet should satisfy
(Stern and Levison 2002, 3). What is particularly interesting about their
list of desiderata is that it is largely in accord with a very traditional view
of natural kinds. D'abord, they argue that an adequate deªnition of planet
should be physically based, “providing insight into the nature of what
planetary bodies are at their essence” (Stern and Levison 2002, 4). Deuxième,
it should be based on observable intrinsic properties, rather than relational
or historical properties (these latter sorts of properties are known in the as-
tronomy literature as “circumstance” and “cosmogony” respectively); troisième
it should be quantitative—that is, numerically based; fourth it should
uniquely classify any given body—no body should belong to more than
one astronomical kind unless one is a subkind of the other (in the natural
kinds literature this is known as a prohibition on cross-cutting categories);
and ªfth it should not allow a body to change its status as a function of
temps (Stern and Levison 2002, 3).9 To this list we can add that the kind
planet should be spatiotemporally unrestricted—a planet should be a
planet no matter where it came from or where in the universe it is found.
The approach to deªning planet exempliªed by Stern and Levison is in
sharp contrast to the other broad family of approaches for which the his-
torical origin of the body (known as cosmogony) and its dynamical cir-
cumstance are considered more important than any particular set of in-
trinsic properties. One prominent approach in this category deªnes a
planet simply on the basis of the physics of planetary formation. Steven
Soter, of the Astrophysics Department at the American Museum of Natu-
ral History, Par exemple, deªnes planet as the end product of secondary accre-
tion from a disk around a star or substar (aka brown dwarf). This well-
accepted model of planet formation is essentially a reªnement of the nebu-
lar hypothesis, ªrst proposed by Immanuel Kant in 1755.10 An interstellar
cloud of gas and dust initially collapses under gravitational attraction
to form a star. Because the nebula is rotating the remaining gas and dust
form a ºat pancake-like disk rotating around the star. Accretion is the
process by which these small particles collide and stick together form-
ing a number of small planetismals; the gravitational force of the larger

9. They identify two further desiderata, namely that it should be robust to new discov-
eries, and should be simple, comprising the fewest possible criteria (Stern and Levison
2002, 3).

10. See Kant’s “Universal Natural History and Theory of Heaven” (1755), especially

Part Two, Section One.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

474

Pluto and the ‘Planet Problem’

planetismals is then able to draw other smaller planetismals to it eventu-
ally becoming large enough to form a planet. The larger a planetismal
gets, the more quickly and effectively it can gather even more material to
it through gravitational attraction. As Soter describes, the “tendency of
disk evolution in a mature system [est] to produce a smaller number of rel-
atively large bodies (planets) in nonintersecting or resonant orbits, lequel
prevents collisions between them” (Soter 2006, 2513). He goes on to ex-
plain that “planets, deªned in this way, are few in number because the so-
lar system provides insufªcient dynamical room for many” (Soter 2006,
2514). This distinguishes planets from smaller objects such as asteroids,
comets, and Kuiper belt objects, which have not been able to clear other
planetismals from their orbital zones, and hence can continue to collide
with other objects.

The historical or “cosmogony” criterion for planethood, basé sur la façon dont
the object was formed, incorporates two other popular criteria for planet-
hood based on circumstance. The ªrst circumstantial criterion is whether
the body has cleared residual planetismals from its neighborhood either
through accretion or scattering, and the second circumstantial criterion is
whether the body is orbiting a star. Although these two circumstantial
criteria fall naturally out of the cosmogonist’s historical deªnition of
planet, they can be embraced independently of it. An advantage of the
dynamical clearing of planetismals criterion—whether as a part of the his-
torical criterion or independent of it—is that nature does provide a sig-
niªcant gap between those bodies that can and cannot clear their orbital
zone. Soter deªnes a parameter, (cid:3), which he calls the planetary discrim-
inant; it is deªned as the ratio of the mass of a body under consideration to
the aggregate mass of all the other bodies that share its orbital zone (Soter
2006, 2514–15). He uses this parameter as a quantitative test for whether
or not a particular body is the end product of disk accretion. Surveying the
major bodies in our solar system, he notes that there is an unoccupied gap
of four orders of magnitude between Mars, which has a (cid:3)-value of 5,100
and the closest contender Ceres, which has (cid:3)-value of .33. Having an un-
occupied gap of 4 orders of magnitude suggest that the division is not an
arbitrary or vague one, as in the case of speciªc mass limits, but rather a
division drawn by nature. The implications of this approach are that
there are only the eight traditional planets in our solar system—Pluto
with a (cid:3)-value of .07 does not make the cut. Unlike the mass-limits ap-
proach based on round shape, a potato-shaped body on Soter’s deªnition
would count as a planet as long as it had cleared its orbital zone. De la même manière
a very massive body (comparable to Mercury, Par exemple) if discovered in
the Oort cloud, would not count as a planet on this deªnition, since it

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

475

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

Chiffre 2. Mass, M., of a body versus the aggregate mass, m, in that body’s orbital
zone (in units of Earth masses). The solid lines bound the gap in the ratio of (cid:3) (cid:4)
M/m, which is a gap of 4 orders of magnitude between the values for Mars and
Ceres. Note that according to this dynamical quantity, Pluto clearly belongs in a
different category than the other planets. (From Soter 2006, figue. 3)

would not be the end product of secondary disk accretion, having cleared
its orbital zone.

Soter’s deªnition can also be applied to exoplanets—that is planets out-
side of our solar system. Focusing on dynamical criteria, Soter argues that
to be counted as a proper planet, an exoplanet should have non-overlap-
ping orbits with the other planets in its solar system, or be shielded from
collisions by a mean motion resonance. Of all the known multiple-planet
exosystems, all but three have non-overlapping orbits, and those three
seem to all have a 2:1 mean motion resonance that would prevent them
from colliding with their neighbors. Soter concludes “All known exo-
planets of main-sequence stars fall well above the gap [. . .][deªned by
the planetary discriminant (cid:3)] and would be classiªed as planets by the
criterion of dynamical dominance” (Soter 2006, 2518).

There has been some debate in the literature over whether the proper
characterization of this circumstantial criterion for planethood is in terms

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

476

Pluto and the ‘Planet Problem’

of having “cleared planetismals” or “gravitationally dominating” its or-
bital zone. Par exemple, it has been objected that Jupiter shares its orbit
with the Trojan asteroids, and so has not cleared its orbital zone, and hence
would seem to be disqualiªed as a planet on this deªnition.11 Advocates
of this interpretation respond that clearing planetismals refers only to
the period during the formation of the planet—not to objects that later
strayed into the planet’s path. Furthermore Jupiter clearly dominates the
dynamics of the Trojan asteroids, and although the Earth hasn’t “cleared”
the moon, it gravitationally dominates it as well.12 By contrast, Pluto does
not dominate the other Kuiper belt objects in its orbital zone, Ceres does
not dominate the asteroid belt, nor does Eris gravitationally dominate the
scattered disk. There is a clear dynamical difference between these two
groups of objects: c'est, the traditional eight planets on one hand and
Pluto, Ceres, and Eris on the other hand. Soter’s characterization of planet
in terms of his quantitative planetary discriminant parameter (cid:3) is able to
side step these debates by showing that what might appear to be border-
line cases on the clearing account are not in fact borderline at all, when the
right dynamical quantity is considered.

The most common objection to using circumstantial dynamical criteria
to deªne planet, such as whether it has cleared planetismals or whether it
orbits a star, is that these are highly contingent relational properties,
rather than intrinsic properties of the body itself. Par exemple, as Mike
Brown and Gibor Basri point out “the terrestrial planets would not have
been as effective in clearing planetismals without the inºuence of [le
much more massive] Jupiter” (Basri and Brown 2006, 202). One can
imagine two identical bodies, one of which gets classiªed as a planet be-
cause it keeps company with other massive planets that help it clear
planetismals from its orbital zone, while its twin, which happens not to
share such prodigious company, fails to meet this criterion for planethood.
In response to this objection Soter asks, “[W]hy is dynamical context any
less relevant [than intrinsic properties]? We refer to objects that orbit
planets as ‘moons’ even though two of them are larger than the planet
Mercury” (Soter 2006, 2518). Autrement dit, there is some precedent in
astronomy for making dynamical circumstance essential to the deªnition
of a kind.

A related objection to these sort of deªnitions is that the dynamical cir-
cumstance of a body can change. Some stellar systems might be unstable

11. Alan Stern quoted in Rincon, P.. (August 25th, 2006), Pluto Vote ‘Hijacked’ in Revolt,

BBC News. http://news.bbc.co.uk/2/hi/science/nature/5283956.stm.

12. Brun, Mike (n.d.) The Eight Planets. http://web.gps.caltech.edu/(cid:2)mbrown/

eightplanets/.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

477

due to the mutual gravitational inºuence of very massive planets on each
other, which can lead to a planet being ejected from the stellar system. Are
these ejected planets still planets? Can a body change the natural kind it
belongs to as a function of time?13 In many ways this is an analog of the
objection that, because species can evolve from one kind into another, ils
cannot properly be considered natural kinds. In the context of the species
problem, Ghiselin for example writes: “If species were not individuals,
they could not evolve” (Ghiselin 1987, 129). The intuition behind this
objection is that natural kinds are abstract objects with immutable es-
sences, hence cannot change. As Laporte rightly responds, cependant, “when
we say ‘Species evolve’, we do not mean that any abstract kind evolves;
[rather] we mean that successive members of a kind gradually become dif-
ferent from their ancestors” (LaPorte 2004, 9).

In the case of planets we do not have successive members of a kind, chaque
individual of which belongs to a single kind throughout its existence.
Plutôt, we have a single individual that changes the kind it belongs to as a
function of time. Néanmoins, when astronomers assert that a particular
planet has been ejected and is no longer a ‘planet’ but rather what is cur-
rently called a ‘free-ºoating object of planetary mass’, what they are assert-
ing is not that one abstract object has turned into another abstract object,
only that one concrete entity has changed such that it is now an instance
of a different kind than it was initially. A completely analogous situation
can occur for chemical elements, which are paradigmatic natural kinds.
Par exemple, an isotope of potassium known as potassium-40 (avec
19 protons) is naturally unstable and can transmutate into an atom of ar-
gon (argon-40) by having one of its protons convert into a neutron via pos-
itron emission or electron capture; alternatively that same atom of potas-
sium can also naturally transmutate into an atom of calcium (Ca-40) par
having one of its neutrons convert into a proton via beta-ray (high speed
electron) emission. In this situation we do not say that the abstract kind
‘potassium’ changed into another abstract kind ‘calcium’, but rather that
a particular atom that was an instance of the natural kind potassium is
now, through the afore-mentioned physical process, an instance of the
natural kind calcium. To require of a natural kind that instances of its
kind cannot change in this respect, is simply too strict of a requirement
for natural kindhood given the sort of world we live in. Even if we accept
that the ability to change kinds is not an obstacle to the natural kind
status of ‘planet’, there remain several other objections that need to be
considered.

13. Stéphanie Ruphy (2010) in her excellent article, “Are Stellar Kinds Natural

Kinds?” refers to this problem as “taxonomic nomadism.”

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

478

Pluto and the ‘Planet Problem’

4. Is ‘Planet’ a Vestige of Folk Astronomy or a Natural Kind?
Faced with the plurality of competing deªnitions for planet in the astron-
omy literature—each with its own weaknesses as well as strengths—one
might be easily be led to the view that planet is not in fact a natural kind
term at all, and should simply be eliminated from modern astronomy.
There are three general worries that can be raised against planet as a natu-
ral kind term that seem to speak in favor of this approach. The ªrst is that
the term ‘planet’ was not introduced in the context of a mature scientiªc
théorie, but rather was introduced in the vernacular, long before the struc-
ture of the solar system and the nature of the bodies it contains was under-
stood. De plus, both the deªnition and recognized extension of this term
have changed many times throughout history.14

Our word ‘planet’ comes from the ancient Greek word (planaytess)
(cid:5)(cid:6)(cid:7)(cid:8)(cid:9)(cid:10)(cid:11)(cid:12) meaning the “wandering stars,” since the points of light corre-
sponding to the planets seem to move against the background of the other
ªxed stars. It originally referred to the ªve planets visible to the naked
eye, which are Mercury, Venus, Mars, Jupiter, Saturn. In the Ptolemaic
geocentric model, ‘planet’ meant those bodies that orbit the Earth, lequel
included both the sun and moon as planets, making for a total of seven.
With Galileo’s discovery in 1610 of the four moons of Jupiter, which he
named the Medicean stars, the number of planets grew to eleven (moons
and planets were not clearly distinguished at this time). With the eventual
acceptance of the Copernican heliocentric model of our solar system the
Earth for the ªrst time came to be counted among the planets. By the end
de 1684, ªve satellites of Saturn were discovered, making for sixteen plan-
ets. It was only in the 18th century that the term ‘moon’ became clearly
distinguished from planet, and the number of planets dropped back down
to six. Dans 1781 William Herschel discovered Uranus, the seventh planet.
Twenty years later (1801) Giuseppe Piazzi discovered Ceres right where
the now-discredited Titius-Bode law predicted a new planet should be
trouvé. This was quickly followed by the discovery of Pallas, Vesta and
Juno, lequel, according to the best astronomy textbooks of the early nine-
teenth century brought the number of planets back up to eleven.15 Early
on Herschel had suggested that a new term ‘asteroid’ be introduced to de-
scribe Ceres and its cohorts, but it was only after the discovery of another
asteroid, Astrae, and the thirteenth planet, Neptune, which was discov-
ered in 1846, that Herschel’s distinction between asteroids and planets
took hold, and the number of planets dropped back down to eight. Dans
1930, Clyde Tombaugh discovered Pluto, which became ofªcially known

14. A very readable history of the planets can be found in Weintraub (2007).
15. For a fuller discussion see, Par exemple, Weintraub (2007), chapter 7.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

479

as the ninth planet. All seemed well with the solar system until, dans 2003,
Brown and collaborators discovered Sedna, a trans-Neptunian object that
is believed to be located in the Oort cloud, which marks the very outer re-
gion of our solar system. Although some heralded Sedna as the tenth
planet, the issue was really brought to a head in 2005 with Brown’s dis-
covery of Eris, a body which is 27% more massive than Pluto.16 Eris is a
“scattered disk” object, which is a dynamical class of objects that are be-
lieved to have been scattered from the Kuiper belt, where Pluto lies, et
now have highly eccentric orbits.17 As a body thought to be larger than
Pluto, it became difªcult to deny that Eris should be counted among the
planets if Pluto is. This brings us ªnally back to the 2006 decision of the
International Astronomical Union to demote Pluto, reducing the number
of planets in our solar system back down to eight. An abbreviated timeline
of the changing conception of planet can be found in Figure 3. As this
very brief history shows, both our understanding of the meaning and the
extension of ‘planet’ has undergone signiªcant changes. For many this
suggests that ‘planet’ is not in fact a natural kind term, but only a vestige
of folk astronomy. The solution to the planet problem on this view is sim-
ply to eliminate this folk concept from modern astronomy.

While some may take the fact that the meaning and extension of a term
has changed over time as an indication that it is not a natural kind, Je pense
it suggests just the opposite. If we understand natural kinds as those tax-
onomies that carve nature at its joints, then as scientists learn more about
the structure of the world, we would expect the deªnitions and extensions
of our natural kind terms to be responsive to these new scientiªc discover-
ies and insights. It is simply asking too much for our natural kind terms
to be immune to scientiªc progress. Indeed the fact that the term ‘planet’
has persisted through revolutionary changes in astronomy seems instead

16. Although astronomers are conªdent that Eris is 27–28% more massive than Pluto,
it is surprisingly difªcult to determine which one is larger. Initial measurements with the
Hubble telescope in 2006 suggested that Eris was signiªcantly larger than Pluto; cependant,
more accurate occultation measurements (timing Eris’s passage in front of a star) indicate
that its size is 2326 km across with an uncertainty of (cid:13)/- 12 km (Sicardy et al. 2011, 138).
This means it is approximately the same size as Pluto, which is thought to be 2306 km
with an uncertainty of (cid:13)/- 20 km. The difªculty in determining Pluto’s size is compli-
cated by the discovery in the mid-1980s that it has an atmosphere with a thick haze that
makes it appear larger. For a discussion see Brown’s http://www.mikebrownsplanets.com/
2010/11/how-big-is-pluto-anyway.html

17. Eris is inclined 44 degrees to the ecliptic where the planets are. Scattered disk ob-
jects are deªned dynamically by their eccentric orbits, rather than as a region of space,
since objects like Eris travel from the outer-most regions of our solar system (where it is
currently) to distances closer to the sun than Pluto.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

480

Pluto and the ‘Planet Problem’

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

Chiffre 3. Abbreviated timeline of the changing conception of ‘planet’.

to be evidence that it is a natural kind term in the way that other scientiªc
terms such as aether, phlogiston, and caloric are not.

There are two additional worries that speak against planet as a natural
kind term. The second worry is that the proposed deªnitions of planet do
not seem to ªt our traditional philosophical conception of what a natural
kind term should be like. Traditionally it is thought that natural kinds
should be speciªable solely in terms of intrinsic properties, which deter-
mine the necessary and sufªcient conditions for being a member of the
kind. On this traditional view, natural kinds should not admit of vague
boundaries, nor should they be spatio-temporally restricted—anything
with the relevant intrinsic properties should count as a member of
the kind no matter where it is or where it came from. Although Stern
and Levison’s proposed deªnition was formulated in terms of intrinsic
properties—mass limits—it failed to yield a categorically distinct kind in
so far as the mass limits admitted vague boundaries; in other words, sur
this deªnition any sharp distinction between planet and nonplanet must
be arbitrary since there exists a continuum of masses in nature. The alter-
native deªnitions for planet, such as the one proposed by Soter, do yield a
categorically distinct kind without vague boundaries, but do so only by
deªning planet in terms of historical and relational properties, tel que

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

481

how the body was initially formed and whether it is orbiting a star. Not
only do such kinds fail to be speciªed in terms of intrinsic properties, mais
they also yield kinds that are spatio-temporally restricted. To someone
wedded to the traditional conception of natural kinds, this would once
again suggest that ‘planet’ is not a natural kind term.

As we saw earlier, these same objections have been raised in the philos-
ophy of biology literature against species being counted as natural kinds.
The best current scientiªc deªnitions for species are also speciªed in terms
of relational rather than intrinsic properties, admit of vague boundaries,
and are spatio-temporally restricted. In response to this, many contempo-
rary philosophers of science, such Richard Boyd, Paul Grifªths, et
Joseph LaPorte, have argued that the problem is not with these deªnitions
of species, but rather with the traditional conception of natural kinds.
Grifªths, Par exemple, cogently argues that there is no reason why there
cannot be kinds with historical essences. Il écrit, “kinds are deªned by
the processes that generate their instances, and for many domains of ob-
projets, these processes are extrinsic rather than intrinsic to the instances
of the kind” (Grifªths 1999, 219). Plus généralement, Boyd has offered a new
theory of natural kinds in terms of what he calls homeostatic property
clusters. On Boyd’s conception, natural kinds are deªned by a cluster of
properties none of which may be necessary or sufªcient for membership
in the kind. The co-occurrence of the properties in the cluster is
brought about by homeostatic mechanisms, c'est, some properties tend
to bring about the existence of the other properties or there are underlying
causal mechanisms which tend to bring about the co-occurrence of these
properties.18 These property clusters can include relational and histori-
cal properties and the kind term deªned by these clusters may have vague
extensions. Important to all these thinkers is the view, emphasized early
on by Quine (1969), that the naturalness of the kind is determined by the
extent to which reference to that kind plays a central role in successful in-
ductions and scientiªc explanations.

Applying this approach to the debate over the deªnition of planet, nous
see that many of the objections to the recently proposed deªnitions of
planet can be dismissed as being motivated by an outdated and misguided
conception of natural kinds. With this new conception of natural kinds,
there is no a priori philosophical reason to reject these deªnitions based on
historical properties, such as how the planet was formed, or relational
properties, such as whether the body is orbiting a star or has cleared its or-
bital zone of planetismals. Consider, Par exemple, Soter’s deªnition of
planet (as the end product of secondary accretion from a disk around a

18. For a clear discussion of the nature of HPC kinds, see Boyd (1999), 143–44.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

482

Pluto and the ‘Planet Problem’

star): it relies precisely on the sort of underlying “homeostatic” mecha-
nisms that Boyd’s HPC account describes, namely that the mechanisms of
disk evolution tend to produce a small number of relatively large bodies in
nonintersecting orbits. It is these mechanisms, related to the historical or-
igin of the object, that give rise to the property cluster in the deªnition of
planet and that also make this deªnition of planet so useful for prediction
and explanation. Ainsi, deªnitions for planet that make use of such crite-
ria are candidates for being natural kinds in Boyd’s sense.

The important question, on this new conception of natural kinds, is not
whether these deªnitions of planet are speciªed in terms of unchanging
intrinsic properties, but rather, whether these deªnitions are facilitating
successful scientiªc predictions and explanations. Autrement dit, insofar
as these deªnitions are scientiªcally fruitful then they can be legitimately
counted as deªning natural kinds. As Boyd has stressed, “[t]he question of
just which properties and mechanisms belong in the deªnition of [le
kind] is an a posteriori question—often a difªcult theoretical one” (Boyd
1999, 143). Ainsi, on this view, whether planet is a natural kind is
largely an empirical question to be decided by astronomers.

Even if one accepts that neither the checkered history of the term
‘planet’ nor worries stemming from the traditional philosophical concep-
tion of natural kinds poses any real obstacle to the natural kindhood of
planet, there still remains the objection that there is simply no scientiªc
consensus on which deªnition of planet is the appropriate one. Just as in
the case of species, there seems to be a plurality of competing deªnitions
for planet each with some claim to legitimacy. In light of this plurality of
scientiªc deªnitions, I think there are three options available to those who
want to defend the natural kind status of planet.19 The ªrst option would
be to defend some version of what might, in analogy with the species
problem, be called “planet monism.” One could defend monism either
by declaring one of these deªnitions of planet to be the correct one, et
dismissing the others as incorrect, or by arguing that there is one correct
deªnition for planet, but that deªnition has simply not been discovered

19. In his recent book Scientiªc Enquiry and Natural Kinds P. D. Magnus devotes a sec-
tion to the Pluto case and also comes to the conclusion that planet meets the criteria for
natural kindhood. His discussion focuses on the IAU deªnition (not the competing deªni-
tions for planet in different accommodation contexts) and the objections he discusses are
also different (par exemple., the objection that ‘planet’ is dispensable because a narrower kind such
as ‘gas giant planet’ and ‘rocky planet’ would do the same work, and the objection that
there might be criteria that preserve the traditional nine planets). He also does not draw
out the detailed analogy with the species problem; hence, his discussion his complemen-
tary to the one I offer here.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

483

yet. The difªculty with monism, cependant, is that many of these deªni-
tions have compelling scientiªc considerations in their favor, and any at-
tempt to pick one would seem to be arbitrary.

A second approach to defending the kindhood of planet would be to
adopt what might be called “planet pluralism.” According to the plural-
ist, there are multiple, equally legitimate deªnitions for planet, each of
which makes some contribution to the prediction and explanation of some
domain of scientiªc phenomena. Donc, Par exemple, planetary geologists
might adopt the deªnition of planet in terms of mass limits based on hy-
drostatic equilibrium and nuclear fusion because they are interested in the
explanation and prediction of phenomena for which these sort of proper-
ties are the most relevant. While, on the other hand, the dynamicists are
typically interested in the prediction and explanation of different sorts of
phenomena for which dynamical history and the circumstance of the body
are more important. Both sets of properties correspond to real and impor-
tant features of the world, they just differ in their relative importance de-
pending on what explanatory scientiªc project one is engaged in. To use
Boyd’s terminology, each subªeld of astronomy places different accommo-
dation demands on the classiªcatory practices they deploy, and while this
might appear on the face of it to lead to some form of relativism, it does
pas, since all the properties appealed to on these various deªnitions are
real and objective features of the world. The central challenge for plural-
ism, cependant, is that if there are multiple equally legitimate deªnitions
for planet—each of which may vary in its extension—then is there any-
thing that legitimately unites these various deªnitions into the kind
planet?

There is a third approach to defending the natural kind status of planet,
and that is in terms of what might be called “planet integrationism.” This
approach deªnes planet as the conjunction of two or more different
deªnitions of planet. Integrationism is actually a form of monism in that
it argues that there is one correct deªnition for planet, but takes from plu-
ralism the insight that each of these deªnitions is latching on to some real,
scientiªcally important feature of the world. As the conjunction of various
conditions, planet integrationism tends to be the most stringent approach
to deªning planet, typically admitting a smaller number of members to
its kind than either deªnition would individually.

Interestingly it was the integrationist approach to deªning planet that
was ªnally adopted by the International Astronomical Union on August
24, 2006. According to the 2006 resolution of the IAU, planet is now
deªned as follows: “A ‘planet’ is a celestial body that (un) is in orbit around
the Sun, (b) has sufªcient mass for its self-gravity to overcome rigid body
forces so that it assumes a hydrostatic equilibrium (nearly round) shape,

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

484

Pluto and the ‘Planet Problem’

Chiffre 4. The current understanding of our solar system (as of 2013) is that it
consists of 8 planets and 5 dwarf planets (the latter of which include former-
planet Pluto and former-asteroid Ceres). Image source: Wikimedia Commons
(http://en.wikipedia.org/wiki/File:Solar_System_size_to_scale.svg) public domain.

et (c) has cleared the neighbourhood around its orbit” (http://www.iau
.org/iau0603.414.0.html). In this deªnition we see the integration of
three different possible deªnitions for planet—ªrst, the circumstantial
criterion that a planet must be in orbit around the Sun, second the plane-
tary geologist’s criterion based on mass limits, which was advocated by
Stern and Levison, and third the dynamicist’s condition that a planet must
have cleared its orbital zone of planetismals, as advocated by Soter. Le
IAU decided that any celestial body that meets the ªrst two conditions—
but not the third—be labelled a “dwarf planet.” Hence, although Pluto is
in orbit around the Sun and is sufªciently massive to have reach hydro-
static equilibrium, it has not cleared the many neighbouring Kuiper belt
objects in its orbit, and so is not a planet, but rather is a dwarf planet
(ironically dwarf planets are not a subspecies of planets on this deªnition).
According to this deªnition there are only eight planets: Mercury, Venus,
Terre, Mars, Jupiter, Saturn, Uranus, and Neptune. For a picture of our
current understanding of the solar system see Figure 4.

Although this may sound like a peaceful resolution to a long-standing
debate, in so far as this deªnition requires that a planet be in orbit
around the Sun—rather than any star or substar—this deªnition, strictly
speaking, only applies to planets in our solar system. As for the over
800 extrasolar planets discovered since 1990, there is still no ofªcial

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

485

deªnition of planet. It is possible that the issue of how to deªne planet in a
way that includes exoplanets will need to be revisited at a future IAU
meeting.

Having examined the tumultuous history of the concept of ‘planet,’ we
are now in a better position to understand the decision to demote Pluto
more speciªcally.

5. Was it Discovered that Pluto Is Not a Planet?
The problem at the heart of the Pluto controversy is that the traditional
deªnition of planet, which had been left rather vague, was—in light of
new astronomical discoveries—ceasing to be a scientiªcally useful concept.
C'est, the collection of objects gathered under the previous deªnition of
planet no longer seemed to constitute a natural group, either admitting
too many or too few objects, in what increasingly seemed to scientists to
be an arbitrary and haphazard way. While it is clear that there had been
a growing state of crisis in the ªeld of astronomy over the concept of a
planet, what is less clear is whether scientists discovered that Pluto is not
a planet. In raising this question I am not questioning astronomers’ deci-
sion to demote Pluto, but rather asking how we, as philosophers of sci-
ence, should classify this decision: Was it a scientiªc discovery or a con-
ventional stipulation?

D'une part, the fact that Pluto’s demotion was decided by means
of a majority vote by 400-some scientists at a conference in Prague seems
to speak in favor of it being merely a stipulation—not a discovery.20 On
the other hand, if it was merely a matter of convention, why was there a
scientiªc crisis at all? I want to argue that this dichotomy is a false one—
the decision to demote Pluto involved both discovery and stipulation. Al-
though there is no one discovery that forced astronomers to conclude that
Pluto is not a planet, there were a number of discoveries that were directly
relevant and precipitated this decision.

The ªrst important cluster of discoveries has to do with Pluto’s mass.
At the beginning of the twentieth century Percival Lowell had predicted
the existence of a new planet X based on perturbations in the orbit of Ura-
nus and Neptune. Based on these calculations it was determined that the
hypothesized planet must have a mass ten times that of Earth. Ironically it
was Clyde Tombaugh working at Lowell’s observatory, looking in the re-
gion of space where planet X was supposed to be, that led to the discovery
of Pluto. It slowly became apparent, cependant, that Pluto was not the
planet they had thought it to be: not only was Pluto’s orbit inclined 17.1°

20. Only 400 of the approximately 10,000 professional astronomers worldwide voted

on the resolution, which was not passed unanimously, but only by a majority.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

486

Pluto and the ‘Planet Problem’

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Chiffre 5. Estimated mass of Pluto (in units of Earth masses) as a function of
temps (from Dessler and Russell 1980).

to the ecliptic (while the next nearest inclined orbit is only 7° for Mer-
cury), but it was also realized that the mass of Pluto must be closer to one
Earth mass rather than ten. As this graph from an 1980 article by Alex
Dessler and Chris Russell shows, the estimated mass of Pluto has dropped
steadily every decade since its discovery.21 In 1978, when Pluto’s moon
Charon was discovered, it was determined that Pluto’s mass could be no
plus que .002 times the mass of the Earth. Dessler and Russell jokingly
subtitled their paper in which this graph appears “The Pending Disap-
pearance of Pluto.”

Not only is Pluto smaller than seven moons in our solar system (inclure-
ing the Earth’s moon), but beginning in 1992, it was discovered that
Pluto is not an isolated body but rather belongs to a vast collection of ob-
jects that make up the Kuiper belt, which includes an estimated 100,000

21. Note that the mass of Pluto is not actually changing—only scientists’ calculation of

the mass of Pluto.

Perspectives on Science

487

objects all over 100 kilometers in size.22 The discovery of the Kuiper
belt—a region of our solar system beyond the orbit of Neptune—
drastically changed scientists’ perception of Pluto, making it an object
much more analogous to Ceres in the asteroid belt (as we saw in the pre-
vious section, Ceres was also demoted from being a planet in the mid-
1800s when the many other objects in the asteroid belt were discovered).
Since the 1990s astronomers have not only discovered the Kuiper belt,
mais aussi, dans 1995, the ªrst brown dwarfs, which exist in the range of
masses above the largest gas giant planets but below that of stars, et
so are not capable of hydrogen burning nuclear fusion reactions in
their cores. The 1990s also saw the discovery of the ªrst exoplanets, que
is planets outside of our solar system, et en 1998 the ªrst free-ºoating
planets were discovered. All of these discoveries put pressure on astrono-
mers to make the deªnition of planet more precise and reconsider the
status of borderline objects such as Pluto. Arguably the straw that broke
the camel’s back was the discovery in 2005 of Eris—the ninth most mas-
sive body directly orbiting our sun, which is 27% more massive than
Pluto. Many other large trans-Neptunian objects such as Sedna, Make-
make, and Haumea, were also discovered around this time.23 While no one
of these many important scientiªc discoveries is “the discovery that Pluto
is not a planet,” they do show that the decision to redeªne planet and de-
mote Pluto was not solely a conventional one, but rather one for which
new scientiªc discoveries played a central and deªning role.

As LaPorte notes in the context of the species problem, faced with the
strain placed on a kind term by discoveries of the preceding sort, scientists
generally have one of three possible options available to them (LaPorte
2004, 66). D'abord, they can expand the taxon, granting that things formerly
thought not to belong to the extension of the term do belong to it after all.
This was the approach advocated by Stern and Levinson, that would have
kept Pluto as a planet, but also admitted 100 or so more objects as plan-
ets, including the Earth’s own moon. The second option that scientists
have in light of the strain imposed by such discoveries is to pare the unac-
ceptable taxon down, saying that things formerly believed to belong to
the extension of the term do not in fact belong to it. This is the approach
advocated by Soter and by the general assembly of the International Astro-
nomical Union, which pared the number of planets in our solar system
down to eight. Enfin, the third option that scientists have available
to them, when old taxonomic categories are faltering in light of new

22. Alan Stern (2012), PI of NASA’s New Horizons Pluto-Kuiper Belt Mission (http://

pluto.jhuapl.edu/overview/piperspective.php?page(cid:4)piperspective_08_24_2012).

23. Makemake and Haumea—like Pluto—are dwarf planets in the Kuiper belt.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

488

Pluto and the ‘Planet Problem’

scientiªc discoveries, is simply to declare the taxon a folk concept, unworthy
of continued scientiªc investigation. On this approach the term planet de-
scribes only an artiªcial kind, and while it might remain in ordinary dis-
course as a cultural term, it would be eliminated from the scientiªc vocab-
ulary of astronomers and planetary scientists—indeed the very notion of a
planetary scientist would become an oxymoron.

Although which of these three options scientists choose to adopt is
largely a matter of convention, what was not an option for the scientists
was to leave the traditional deªnition and extension of the term planet in-
tact, while having it remain a scientiªcally useful concept. As Soter notes,
“ad hoc deªnitions devised to retain Pluto as a planet tend to conceal from
the public the paradigm shift that has occurred since the 1990s in our un-
derstanding of the architecture of the solar system. [. . .] To be useful, un
scientiªc deªnition should be derived from, and draw attention to, the ba-
sic principles [of nature]» (Soter 2006, 2518). Natural kinds are a joint
venture between the taxonomic practices of human beings and the struc-
ture of nature, and as any practicing scientist will tell you, nature is often
quite successful in resisting the particular taxonomies we may wish to
impose.

6. Conclusion
As I have shown, new insight into the Pluto controversy is gained by
drawing out the striking analogy between what I have here termed the
‘planet problem’ in astronomy and the well-known species problem in the
philosophy of biology. I argued that the taxon planet can still be consid-
ered a natural kind despite the fact that a) its meaning and recognized ex-
tension have changed over time, b) there are multiple scientiªcally com-
pelling deªnitions of planet, and c) many of those deªnitions include
historical and/or relational properties. Barring these worries, it is ulti-
mately an empirical question to be decided by astronomers whether the
taxon planet is in fact scientiªcally useful for the organization, explana-
tion, and prediction of phenomena.

I also argued that the decision demote Pluto involved both discovery
and stipulation. The important new discoveries about the structure of our
solar system that have taken place since the 1990s have often been over-
looked in discussions about the Pluto case, which is often painted as noth-
ing more than a debate over semantics. Moving beyond the false di-
chotomy between discovery and stipulation is important for making
sense scientiªc episodes such as this, where there is a revision in an ac-
cepted scientiªc taxonomy.

Enfin, this analysis of the planet problem also suggests that the spe-
cies problem in biology, instead of being anomalous, may be more typical

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

489

in science than we had hitherto thought. If it turns out that the sort of
taxonomies that science ªnds most useful are often deªned by historical
and relational properties, are sometimes permissive of vague boundaries,
and are such that there can be multiple scientiªc deªnitions in different
accommodation contexts, then perhaps the correct conclusion to draw is
that the problem lies not with these scientiªc taxons, but rather with our
traditional philosophical conception of natural kinds.

Les références
Basri, G. and M. Brun. 2006. “Planetismals to Brown Dwarfs: What is a
Planet?.” Annual Review of Earth and Planetary Science 34: 193–216.
Oiseau, UN. and E. Tobin. “Natural Kinds.” The Stanford Encyclopedia of Philos-
ophy (Été 2010 Edition). Edited by Edward N. Zalta http://plato
.stanford.edu/archives/sum2010/entries/natural-kinds/.

Boyd, R.. 1999. “Homeostasis, Species, and Higher Taxa.” In Species: Nouveau
Interdisciplinary Essays. Edited by R. Wilson. Cambridge, MA: AVEC
Presse.

Brookes, M.. 1998. “The Species Enigma.” New Scientist 158: 2138.
Claridge, M., UN. Dawah, and M. Wilson. 1997. Species: The Units of

Biodiversity. Londres: Chapman & Hall.

Dessler, UN. and C. Russell. 1980. “From the Ridiculous to the Sublime:
The Pending Disappearance of Pluto.” EOS: Transactions American Geo-
physical Union 61 (44): 690.

Ereshefsky, M.. 2007. “Species.” The Stanford Encyclopedia of Philosophy
(Été 2007 Edition). Edited by Edward N. Zalta. http://plato
.stanford.edu/archives/sum2007/entries/species/.

Ghiselin, M.. 1974. “A Radical Solution to the Species Problem.” System-

atic Zoology 23: 536–44.

Ghiselin, M.. 1987. “Species, Concepts, Individuality, and Objectivity.”

Biology and Philosophy 2: 127–43.

Grifªths, P.. 1999. “Squaring the Circle: Natural Kinds with Historical
Essences.” In Species: New Interdisciplinary Studies. Edited by R. Wilson.
Cambridge, MA: AVEC Presse.

Hull, D. 1978. “A Matter of Individuality.” Philosophy of Science 45: 335–

60.

King, M.. C. et un. C. Wilson. 1975. “Evolution at Two levels in Humans

and Chimpanzees.” Science 188: 107–16.

Kitcher, P.. 1984. “Species.” Philosophy of Science 51: 308–33.
LaPorte, J.. 2004. Natural Kinds and Conceptual Change. Cambridge: Cam-

bridge University Press.

Magnus, P.. D. 2012. Scientiªc Enquiry and Natural Kinds: From Planets to

Mallards. Basingstoke, England: Palgrave MacMillan.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

2
2
4
4
6
4
1
7
8
9
9
4
1
p
o
s
c
_
un
_
0
0
1
4
6
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

490

Pluto and the ‘Planet Problem’

Quine, W. V. Ô. 1969. Ontological Relativity and Other Essays. New York:

Columbia University Press.

Ruphy, S. 2010. “Are Stellar Kinds Natural Kinds? A Challenging New-
comer in the Monism/Pluralism and Realism/Antirealism Debates.”
Philosophy of Science, 77 (5): 1109–20.

Sicardy, B. et autres. 2011. “Size, Density, Albedo, and Atmosphere Limit of
Dwarf Planet Eris from a Stellar Occultation.” European Planetary Science
Congrès 6: 137–38.

Soter, S. 2006. “What is a Planet?.” The Astronomical Journal 132: 2513–

19.

Arrière, S. UN. and H. Levison. 2002. “Regarding the criteria for planethood
and proposed planetary classiªcation schemes.” Highlights in Astronomy
12: 205–13. http://www.boulder.swri.edu/pkb/alan/papers.html.

Weintraub, D. 2007. Is Pluto a Planet? A Historical Journey through the Solar

System. Princeton, New Jersey: Princeton University Press.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d