Synthese (2014) 191:3557–3573

DOI 10.1007/s11229-014-0462-x

A defense of indeterminate distinctness

Ken Akiba

Received: 8 July 2013 / Accepted: 14 April 2014 / Published online: 1 May 2014
© Springer Science+Business Media Dordrecht 2014

Abstract On the one hand, philosophers have presented numerous apparent examples
of indeterminate individuation, i.e., examples in which two things are neither deter-
minately identical nor determinately distinct. On the other hand, some have argued
against even the coherence of the very idea of indeterminate individuation. This paper
defends the possibility of indeterminate individuation against Evans’s argument and
some other arguments. The Determinacy of Identity—the thesis that identical things
are determinately identical—is distinguished from the Determinacy of Distinctness—
the thesis that distinct things are determinately distinct. It is argued that while the first
thesis holds universally and there is no case of indeterminate identity, there are reasons
to think that the second thesis does not hold universally, and that there are cases of
indeterminate distinctness.

Keywords Vague identity · Ontic vagueness · Modal logic · Evans’s argument

1 Introduction

Even among the generally difficult issues concerning vagueness and indeterminacy,
the issue of indeterminate (or vague) identity is one of the most confusing. On the
one hand, philosophers have presented numerous apparent examples of indeterminate
identity, i.e., examples in which two things seem to be neither determinately identical
nor determinately distinct. On the other hand, some have argued against even the
coherence of the very idea of indeterminate identity. But how can an idea that has
numerous concrete instances be incoherent?

K. Akiba (B)
Department of Philosophy, Virginia Commonwealth University, 915 W. Franklin St.,
Richmond, VA 23284-2025, USA
e-mail: kakiba@vcu.edu

123
3558 Synthese (2014) 191:3557–3573

In this paper, what is usually called ‘indeterminate (or vague) identity’ will be
called ‘indeterminate individuation’ for reasons which will become apparent shortly.
A typical example of indeterminate individuation involves two material objects. One,
say a, has mereologically and/or spatiotemporally vague boundaries, and the other, say
b, coincides with a except for some of its indeterminate parts. Thus, b has either precise
boundaries or at least less vague boundaries than a. We can precisify a in various ways,
making its boundaries sharper. One precisification of a is identical with b; but other
precisifications of a are not identical with b. Therefore, a and b are indeterminately
individuated; that is, they are neither determinately identical nor determinately distinct.
In symbols, ¬a = b ∧ ¬¬a = b, where ‘’ reads ‘determinately’.
For instance, in Weatherson’s (2003) example, Kilimanjaro (or k for short) is a
vague object, and it is indeterminate whether the atom Sparky is a part of Kilimanjaro.
Kilimanjaro+ (or k + ) is defined as the mereological sum (or fusion) of Kilimanjaro
and Sparky, whereas Kilimanjaro− (or k − ) is defined as k + minus Sparky. Then it
is indeterminate whether k = k + because in one scenario, in which Sparky is a part
of Kilimanjaro, k = k + , and in another scenario, in which Sparky is not a part of
Kilimanjaro, k = k − (= k + ).
In Williams’s (2008) example, one amoeba splits into two, which wander off to the
east and the west. Sue is the amoeba that survives the fission, Sally is the amoeba that
wanders off to the west, and Sandy is the amoeba that wanders off to the east. But it is
indeterminate whether Sue survives as Sally or Sandy. In one scenario, Sue = Sally;
in another, Sue = Sandy. Thus, it is indeterminate whether Sue = Sally.
Whereas Kilimanjaro is spatially indeterminate, Sue is temporally indeterminate.
Probably the most famous example of temporal indeterminacy is the story of the
Ship of Theseus. Similar examples have been given in, e.g., Geach (1962: ‘1001
cats’), Shoemaker (1963: ‘Brown and Brownson’; Shoemaker and Swinburne 1984:
‘Alpha Hall and Beta Hall’), Lewis (1993: ‘Fred’s house’), Edgington (2000: ‘two
mountains’), and Hawley (2002: ‘two mice’), among others.1
In spite of these numerous examples, however, the idea of indeterminate indi-
viduation has been strongly criticized, most famously by Evans (1978), and, more
recently, by Akiba (2000a, 2014), Williamson (1996, 2002), and Smith (2008). They
all argue that not only are those examples not genuine examples of indeterminate
individuation, but the very concept of indeterminate individuation is incoherent. They
deny that the situations presented in the alleged examples should be described as
¬a = b ∧ ¬¬a = b.
This paper aims to defend the notion of indeterminate individuation and its instances
against their arguments. It will argue that they all failed to show that the notion of
indeterminate individuation is incoherent, or that the alleged examples are not gen-
uine examples of indeterminate individuation. It won’t go so far as to show, once and
for all, that the notion is coherent, or that the alleged examples are genuine examples;
for instance, it won’t eliminate the possibility that the notion involves inconsistency
not revealed in those arguments. It is a defense of indeterminate individuation, and not
an argument for it, in that regard. In the next section Evans’s argument is examined in

1 The issue of vague identity is discussed in the papers in Akiba and Abasnezhad (2014, Sect. 5).

123
Synthese (2014) 191:3557–3573 3559

details. Section 3 is divided into four subsections, each dealing with 1. the argument
from stability; 2. Akiba’s analogical argument;2 3. Williamson’s balance argument;
and 4. Smith’s denial of indeterminate individuation, respectively.
We have mentioned ‘scenarios’, but let us introduce the term ‘precisifications’ to
refer to such scenarios, and sketch the framework we will be working in. In this paper,
a state of affairs is considered vague or indeterminate if and only if it can be under-
stood as one precise state of affairs3 but can be understood as another (incompatible)
precise state of affairs (in the ontic, not epistemic, sense of ‘can’). This ‘can’, then, is
interpreted in analogy with modality, as ‘there is a possible world, accessible from the
present world, at which ...’. Such possible worlds are called ‘(ontic) precisifications’.
A (proper ontic) precisification of a world (itself a precisification, possibly the actual
world) is a world which includes some precise states of affairs that are not included in
the original world. Every precisification is a precisification of itself but not a proper
precisification. The actual world, henceforth denoted as ‘@’, is also considered a pre-
cisification (of at least itself). Every precisification must be consistent, but not every
precisification needs to be complete (or maximal): it cannot contain both p and ¬ p but
need not contain either p or ¬ p for every state of affairs p (even though, presumably,
it contains p ∨ ¬ p). Thus, the world at which k = k + in the Kilimanjaro example, as
well as the world at which k = k − , is a precisification of the actual world, at which k
is vague. (These two worlds are henceforth referred to as w1 and w2.) Similarly, the
world at which Sue = Sally in the amoebae example, as well as the world at which Sue
= Sandy, is a precisification of the actual world, at which Sue is vague. Again, precisi-
fications are a species of possible worlds, broadly understood. This is a plausible idea
because, as we shall see in the next section, ‘determinately’  is usually considered
a modal operator in analogy with the alethic (or metaphysical) modal operator ‘nec-
essarily’ . By ‘p’, we shall mean ‘determinately, p’ or ‘it is determinate that p’,
and not ‘it is determinate whether p’ or “the sentence ‘ p’ is determinately true.”
Note that it is not assumed here that precisifications are monotone expanding (and
their contents cumulative or monotone increasing): it is not assumed that if w is a
precisification of w, w contains all the states of affairs contained in w and perhaps
more (if it is proper). On our conception of precisification it is at least theoretically
possible for there to be a state of affairs contained in w but not in w  (i.e., removed from
w in the process of precisification). This may seem counterintuitive at first; but, as we
shall see, it is not only possible but sometimes it happens. At this point, however, let us
keep in mind that the monotone expansion of precisifications is at least not assumed.
A point of emphasis in this paper will be the distinction between the following two
theses, the Determinacy of Identity and the Determinacy of Distinctness:

Determinacy of Identity (DI): Identical things are determinately identical, i.e.,

∀x∀y(x = y → x = y).

Determinacy of Distinctness (DD): Distinct things are determinately distinct,

i.e., ∀x∀y(¬x = y → ¬x = y).

2 Thus, we have changed our mind on this issue.

3 In an admissible manner—but this qualification is henceforth omitted for the sake of simplicity.

123
3560 Synthese (2014) 191:3557–3573

The view that DI does not hold, and that there are identical things that are not deter-
minately identical, will be called the view in favor of indeterminate identity, and the
view that DD does not hold, and that there are distinct things that are not determinately
distinct, will be called the view in favor of indeterminate distinctness. The view that
some pairs of objects are neither determinately identical nor determinately distinct
will be called the view in favor of indeterminate individuation. Thus, again, what is
usually called ‘indeterminate identity’ will be called ‘indeterminate individuation’ in
this paper. If DI and DD are both correct, then, since for any x and y, either x = y
or ¬x = y, x = y ∨ ¬x = y; this, however, is just the denial of the possibility
of indeterminate individuation, i.e., that for some x and y, ¬x = y ∧ ¬¬x = y
(= ¬(x = y ∨ ¬x = y)). Thus, DI and DD together imply that indeterminate
individuation is impossible. As you will see in the next section, Evans’s strategy to
refute indeterminate individuation, in essence, is, first, to prove DI, and, then, using
that result, to prove DD. Most past defenses of indeterminate individuation focused
on and tried to refute his proof of DI in the first part. This paper, in contrast, will grant
that there is nothing wrong in that proof, that DI is in fact a correct thesis, and, thus,
that there is no indeterminate identity. The main focus of the paper, instead, will be
on the proof of DD in the second part. It will be pointed out that while the proof of
DI relies only on logical and/or unproblematic principles and is thus irrefutable, the
proof of DD relies crucially on a problematic non-logical, modal principle, Axiom
B: ¬ p → ¬p. B holds universally only when the accessibility relation among
the relevant possible worlds is symmetric. There is, however, reason to think that the
relevant accessibility relation is not symmetric, and, thus, that B does not hold for
. Therefore—the paper concludes—Evans’s proof of DD and his attempt at refuting
indeterminate individuation fail; there still can be instances of indeterminate distinct-
ness. The other arguments also fail for various reasons, some related to DD.
This paper will embrace the following two methodological policies, which will
eliminate certain theoretical options that may otherwise exist for a defense of indeter-
minate individuation. First, we shall use only classical logic in the strictest sense of
the term, as well as its modal extensions. As is often pointed out, if we diverge from
classical logic, what counts as identity becomes unclear. For instance, Parsons (1987,
2000) and Parsons and Woodruff (1995) propose to use many-valued logic for the logic
of vagueness, and maintain that the original form of Leibniz’s Law, ∀x∀y(x = y →
∀P(P x → P y)), is, but its contrapositive, ∀x∀y(¬∀P(P x → P y) → ¬x = y),
is not, supported by our concept of identity. They try to block Evans’s argument by
rejecting the latter. Their claim is difficult to assess because Leibniz’s Law and its con-
trapositive are equivalent in classical logic—though not in their many-valued logic —
and, thus, are all but indistinguishable in our daily lives. Our concept of identity is
inseparably intertwined with our use of classical logic. If we drop classical logic, what
counts as identity will no longer be obvious, and the debate may degenerate into a
clash of intuitions.
Second, we shall not semantically ascend and discuss identity statements; we shall
only discuss identity as a relation among objects existing in the world, using, not men-
tioning, identity statements. Thus understood, the issue of indeterminate individuation
is an issue in ontic indeterminacy. Some people, most notably the so-called seman-
ticists (such as Lewis 1988), accept the truth of some indeterminate individuation

123
Synthese (2014) 191:3557–3573 3561

statements (i.e., statements of the form ‘it is indeterminate whether a is identical with
b’) without accepting the possibility of indeterminate individuation as a worldly fact.
They do this by assigning different denotations to singular terms such as ‘a’ and ‘b’ in
the descriptions of different precisifications. Whether they succeed in their attempts
or not, that is not a defense of indeterminate individuation as a worldly fact. Others
(such as Barnes 2009) employ counterpart theory for determinacy, according to which
a and b, deemed identical in one precisification, are not numerically the same a and
b, deemed distinct in another precisification, but just their counterparts (i.e., different
objects); that is, this theory also assigns different denotations to ‘a’ and ‘b’ in different
precisifications. Whether this view is plausible or not, it is not the view advocating
indeterminate individuation, either. The view advocating indeterminate individuation
ought to hold that one and the same object a is identical with one and the same b in
one precisification and distinct from it (not its counterpart) in another precisification.
The view advocating indeterminate individuation must thus assume transworld iden-
tity across different precisifications. Counterpart theory, just like semanticism, only
makes some indeterminate individuation statements true without committing itself to
the genuine possibility of indeterminate individuation.
These complications and confusions arise because philosophers—opponents as
well as proponents of indeterminate individuation—have a tendency to talk about
identity statements instead of identity relations. Evans was himself guilty of this
tendency. The best way to avoid this confusion is simply to embrace the policy of not
dealing with identity statements but dealing with identity relations directly, and that’s
the policy this paper will adopt. If we are talking about the object a, using and not
mentioning the name ‘a’ many times, we are of course talking about one and the same
object throughout the discourse, whereas if we are talking about the name ‘a’, there
is always a chance that each occurrence of the name has a different denotation.
Both policies upheld here should be acceptable to the critics of indeterminate indi-
viduation because these policies only make the defense of indeterminate individuation
more difficult. This paper’s contention is that even under those difficult, restrictive con-
ditions, a defense of indeterminate individuation is still possible. Many criticisms of
indeterminate identity in the literature, including those given in much of Williamson
(2002) and Smith (2008) we discuss, are actually directed to the use of the two strate-
gies we exclude, i.e., the use of non-classical logic and semantic ascent, which often
distract the critics’ attention from the main issues. By excluding them, this paper should
be able to focus on the genuine criticisms of indeterminate individuation. We shall
consider such criticisms in the next two sections, beginning with Evans’s argument.

2 Evans’s argument

Evans’s original refutation of indeterminate individuation was given in his famous one-
page paper, “Can There Be Vague Objects?” (Evans 1978). Even though the paper is
extremely influential, there are several serious flaws in it. We shall mention four. First,
Evans did not give any argument for his assumption that if there are vague objects,
their identity relations with some other, overlapping objects must be vague. So even if
he succeeds in his refutation of indeterminate individuation, it is unclear whether he

123
3562 Synthese (2014) 191:3557–3573

also has succeeded in his refutation of vague objects (which, as the title of the paper
indicates, was his ultimate goal). In this paper we shall ignore this defect, however,
because our interest is first and foremost in indeterminate individuation.
Second, as was alluded to in the last section, Evans did not treat the issue of inde-
terminate individuation purely ontically; instead, he semantically ascended, starting
his reductio argument thus: “Let ‘a’ and ‘b’ be singular terms such that the sentence
‘a = b’ is of indeterminate truth value, ...” (Evans 1978, p. 208). As a result, unneces-
sary complications arose in his original argument (see Lewis 1988). Salmon’s (1981)
similar argument avoids those complications (though his argument does not include the
second part of Evans’s argument for DD). The first part of the following presentation
is closer to Salmon’s argument.
Third, Evans thought that DI is already contradictory to the premise of the reductio
argument, ∃x∃y(¬x = y ∧ ¬¬x = y), even though it is clearly not, as you can
see below. Fourth, perhaps as a consequence of the third point, Evans did not spell out
his proof for DD and derivation of a contradiction, merely stating that a contradiction
is derivable in modal logic S5.
As a matter of fact, we do not need a modal logic as strong as S5 to derive DD
and a contradiction. What follows is perhaps one of the simplest proofs Evans could
have given. The only modal axiom used is B: ¬ p → ¬p, which is supported by
the symmetric accessibility relation.4 For a proof of DI, suppose, for some objects
a and b, a = b. Because a = a is a theorem, by the Rule of Necessitation (RN):
p ⇒ p, a = a. Then, by inserting P x = a = x to Leibniz’s Law:
∀x∀y(x = y → ∀P(P x → P y)), a = b → a = b (the proof of DI). For a proof
of DD, suppose, for reductio, that ¬a = b and ¬¬a = b at some w. Then, from
the second assumption, for some w  accessible from w (i.e., w Rw  ), a = b at w  .
Then, by DI, a = b at w  . Axiom B states that the accessibility relation among the
precisifications is symmetric: for any worlds w and w , if w  is accessible from w, w
is accessible from w (i.e., w Rw  ⇒ w  Rw). Thus, w in the present case is accessible
from w  (i.e., w  Rw). Then, since a = b at w  , a = b at w, contradicting ¬a = b
at w. Thus, ¬a = b → ¬a = b at w for any w (the proof of DD). Since for any
a and b, a = b ∨ ¬a = b, by both DI and DD, a = b ∨ ¬a = b. Therefore,
∃x∃y(¬x = y ∧ ¬¬x = y) cannot be true. QED.
Does this argument successfully show that indeterminate individuation is impossi-
ble? Our answer will be in the negative. But in order to justify the answer, we shall
proceed as follows. Since, again, this paper’s goal is a defense of indeterminate individ-
uation, it should be sufficient to present a plausible countermodel to Evans’s argument.
We shall do that first. Then we shall show that according to that model, his proof of
DI is valid but his proof of DD is not.
We have already touched upon the accessibility relation among precisifications.
But what is the nature of the accessibility relation? What does it represent? On our
view, it represents the process of precisification: for any precisifications w and w ,
w  is accessible from w if and only if w  is a precisification of w. In the previous
examples, the precisifications at which k = k + , k = k − , Sue = Sally, Sue = Sandy,

4 Williamson (1996) and Hughes and Cresswell (1996, p. 314) give proofs that use Axioms K and B (i.e.,
in the modal system KB), but K is in fact not necessary.

123
Synthese (2014) 191:3557–3573 3563

etc., are precifications of the actual world. If at all precifications of w, a state of affairs
p, e.g., k = k + ∨ ¬k = k + , holds, then p at w. If a state of affairs p, e.g., k = k + ,
holds at some precisifications of w but not at others, then ¬p ∧ ¬¬ p at w.
This interpretation of the accessibility relation among precisifications at least par-
tially determines the laws governing the relation. For instance, reflexivity, w Rw for
any w, may be considered to hold because every precisification may be regarded
as its own precisification (though not ‘proper’ precisification, at which some pre-
cisifying is actually done). Reflexivity supports Axiom T: p → p. Transitivity,
w Rw  ∧ w  Rw  ⇒ w Rw  , may also be considered to hold because if w  is a pre-
cisification of w and w is a precisification of w  , w  is a precisification of w. For
instance, if one part of Kilimanjaro is precisified at w , resulting in k + , and if one
part of k + is precisified at w  , resulting in k ++ , then k ++ itself is a precisification of
Kilimanjaro. Transitivity supports Axiom 4: p → p.
It is beyond the scope of this paper to determine exactly what the logic of ‘determi-
nately’  is. It is rather obvious, however, that symmetry, w Rw ⇒ w  Rw, should not
hold for . For if w  is a proper precisification of w, w cannot be a precisification of
w  . For instance, in the Kilimanjaro example, the actual world @, at which k is vague,
cannot be a precisification of the two worlds w1 and w2, at which k is precisified and
k = k + and k = k − , respectively. Symmetry supports Axiom B: ¬ p → ¬p. Thus,
Axiom B should not hold for . The situation will be even worse if reflexivity and
transitivity are to hold for  in addition to symmetry, for in that case the accessibility
relation will be an equivalence relation and the logic of  will be S5; so any two
worlds in an equivalence class are mutually accessible. But it makes little sense to say
that w1 and w2 in the Kilimanjaro example are precisifications of each other, since
one includes k = k + and the other includes ¬k = k + .5 As is well known, modal logic
S4 supports reflexivity and transitivity but not symmetry.
In fact, the current interpretation supports anti-symmetry, w Rw ∧ w  Rw ⇒ w =
w : if w  is a precisification of w, and w is a precisification of w , then those precisi-


fications cannot be ‘proper’; no genuine precisification must be done. A relation that

supports reflexivity, transitivity, and anti-symmetry is a partial order. So the current
interpretation supports the view that the relevant accessibility relation is a partial order.
Finally, Axiom K: ( p → q) → (p → q) and the Rule of Necessitation (RN):
p ⇒ p, which constitute the core of ‘normal’ modal systems, are acceptable on
the current interpretation. For any precisification w, if both p → q and p hold at any
precisification of w, then q must also hold at it, and if p holds at all precisifications,
then for any precisification w, p of course holds at any precisification of w.
The present interpretation of the accessibility relation is different from the more
popular interpretation, according to which w  is accessible from w if and only if w 

5 In fact, there is a simple quasi-Evans argument against indeterminate distinctness that uses not Axiom B but
Axiom 5: ¬p → ¬p, which is supported by the euclidean accessibility relation: w Rw  ∧ w Rw  ⇒
w  Rw  . Take any precisification w (possibly @). If, for some a and b, ¬a = b ∧ ¬¬a = b at w,
then there is a precisification w  of w (i.e., w Rw  ) at which a = b, and there is a precisification w  of
w (i.e., w Rw  ) at which ¬a = b. But, by DI, a = b at w  . By the euclidean accessibility relation,
w  Rw  ; thus, a = b at w  , contradicting ¬a = b. QED. However, this argument does not hold up because
the ‘is a precisification of’ relation cannot be euclidean, as we just saw. We thank Ali Abasnezhad (in
correspondence) for showing us a similar reductio proof in the modal system K45.

123
3564 Synthese (2014) 191:3557–3573

is only insignificantly different (for instance, perceptually indiscriminable) from w.6

If this interpretation is adopted, symmetry and Axiom B should hold, but transitivity
and Axiom 4 should not. Symmetry should hold, for if w  is only insignificantly
different from w, w should be only insignificantly different from w as well; transitivity
should not hold, for even if w is only insignificantly different from w and w is only
insignificantly different from w , w  may be significantly different from w. In this paper
we shall not decide which of the two interpretations is more relevant to the question
about indeterminate individuation.7 For our purposes it is sufficient to provide at least
one reasonable countermodel to Evans’s argument.
It is sometimes maintained that transitivity should not hold for ‘determinately’ 
because Axiom 4: p → p prevents the possibility of higher-order vagueness.
Sometimes we may want to say that p is determinate but that that’s not determinately
so, i.e., p ∧ ¬p; but Axiom 4 won’t allow that. However, even though Axiom 4
holds on our interpretation, Axiom 5: ¬p → ¬p does not; so even though we
cannot say that p is determinate but that that’s not determinately so, we can say that p is
not determinate but that that’s not determinately so, i.e., ¬p∧¬¬p. Thus, higher-
order vagueness, though somewhat restricted, can still exist on our interpretation.8
Now, what can we say about Evans’s proof from the present viewpoint? His proof
of DI is airtight. As has been pointed out (e.g., Williamson 1996; Akiba 2000a,b,
2004), it is nothing but the Marcus–Kripke proof of the Necessity of Identity (NI),
a = b →  a = b (Barcan 1947; Marcus 1961; Kripke 1971). The only modal rule
used is the Rule of Necessitation, which was just determined to hold. Thus, DI may
be considered a theorem.
The proof of DD, however, is a different story. It uses Axiom B in an essential
manner. As we have just seen, B should not hold for the accessibility relation inter-
preted as ‘is a precisification of’. Since Evans’s proof of DD relies on Axiom B and
the symmetry of the accessibility relation, it cannot be sustained. Contrary to the pop-
ular opinion, it is DD, not DI, that turns out to be problematic. Since we need both
DI and DD to refute the possibility of indeterminate individuation, there may still be
indeterminate individuation.
An implication of this is that there can be a precisification at which for some x and
y, ¬x = y and ¬¬x = y (the second conjunct of the indeterminate individuation
statement ¬x = y ∧ ¬¬x = y). In fact, in the Kilimanjaro example, ¬k = k + and
¬¬k = k + at @, and in the amoebae example, ¬Sue = Sally and ¬¬Sue = Sally at
@. p does not generally follow from p. Kilimanjaro and Kilimanjaro+ are distinct
at @ because they can be distinct after precisification (i.e., there is a precisification,

6 This interpretation is adopted most explicitly by Williamson (1999), although he does not commit himself
exclusively to it and considers other possibilities. Some, such as Dummett (1975), Salmon’s (1981), and
Greenough (2003), adopt the same interpretation or similar interpretations. Our interpretation is close to
Fine (1975), though we develop it in a different direction.
7 Nor shall we decide between our interpretation and the interpretation given in Barnes and Williams
(2011), which sets forth a generally ontic theory of indeterminacy but embraces S5 modality as the logic
of ‘determinately’.
8 See, e.g., Bobzien (2012) for a defense of Axiom 4. Field (2000, p. 19) argues that even S5 modality does
not eliminate higher-order vagueness in the most general sense.

123
Synthese (2014) 191:3557–3573 3565

w2, at which they are distinct) whereas Kilimanjaro cannot be distinct from itself. In
contrast, at w2, Kilimanjaro (= Kilimanjaro− ) and Kilimanjaro+ are not only distinct,
but determinately so. Generally, ¬x = y implies ¬x = y only when x and y are
precise in the relevant respects.
It may be mentioned in this context that the current interpretation of the accessibil-
ity relation does not support the strengthened form of the indeterminate individuation
statement, (¬x = y ∧ ¬¬x = y). It is sometimes argued (e.g., Williamson
2002, p. 292) that the advocates of indeterminate individuation ought to accept this
strengthened principle because if indeterminate individuation is possible, that possi-
bility should not depend on anything contingent. But, then, a contradiction is easily
derivable from this version with the use of only Axiom T, as follows:9 Suppose, for
some a and b, (¬a = b ∧ ¬¬a = b) at some w. Then, by Axiom T (reflexivity),
¬a = b ∧ ¬¬a = b at w. From the second conjunct, at some w accessible from
w (i.e., w Rw ), a = b. By DI, a = b at w  . But ¬a = b ∧ ¬¬a = b also at
w  —a contradiction.
However, the current interpretation of the accessibility relation does not support
the strengthened principle (¬x = y ∧ ¬¬x = y). For it implies that if
¬a = b ∧ ¬¬a = b (i.e., it is indeterminate whether a = b) at @, it carries
over to all of @’s precisifications, including those at which a and b are completely
precisified. The advocates of indeterminate individuation have no reason to accept such
a principle.
The common reason why some, such as Williamson, ascribe the strengthened
principle to the advocates of indeterminate individuation is that the original inde-
terminate individuation statement, ¬x = y ∧ ¬¬x = y, if true, cannot be
contingently true (contingent on what?); thus, it must be determinately true, i.e.,
(¬x = y ∧ ¬¬x = y). However, there is a conflation between alethic necessity
and determinacy in this reasoning. It may be the case that ¬x = y ∧ ¬¬x = y,
if true, is necessarily so, i.e., (¬x = y ∧ ¬¬x = y). However, determinacy is
different from alethic necessity. It is reasonable to think that Axiom B holds for alethic
necessity, and that the accessibility relation is an equivalence relation for alethic modal-
ity. Neither of those is the case for determinacy. Thus, (¬x = y ∧ ¬¬x = y)
should not hold even in the eyes of the advocates of indeterminate individuation. The
difference between alethic modality and determinacy will loom larger in the subse-
quent discussion.

3 Some other arguments

In sum, Evans’s original argument against indeterminate individuation does not hold
up. In the rest of this paper, we shall consider some other putative refutations of
indeterminate individuation, including Akiba’s, Williamson’s, and Smith’s; but we
begin with an argument which may be considered a way to rescue Evans’s argument.

9 Williamson (2002, p. 293) gives a proof that uses Axioms K and T (i.e., in the modal system KT), but K
is in fact not necessary.

123
3566 Synthese (2014) 191:3557–3573

3.1 The argument from stability

What has been revealed in our examination of the first half of Evans’s argument is
that we may have the combination of ¬a = b ∧ ¬¬a = b (i.e., it is indeterminate
whether a = b) and ¬a = b at @ for some objects a and b; and the former implies
that there is a precisification w1 of @ at which a = b (as well as a precisification w2
of @ at which ¬a = b). Many people, including Evans himself, find this combination
already contradictory, even though no formal contradiction is yet derived. Despite
Evans’s official claim, this apparent contradiction does not seem to be the same as the
one we derived in the last section using Axiom B and the symmetry of the accessibility
relation. Many people seem to feel that a more immediate contradiction is involved.
But what exactly is it?
Here is the potential derivation of the alleged contradiction, relying on the following
principle, called the stability principle (or Stability for short):

Stability: Once a state of affairs is included in a precisification, it will be included

in all of its further precisifications.

This principle is explicitly upheld in Fine (1975), from which its name is taken. It is
the same as the claim that precisifications are monotone expanding (or their contents
are monotone increasing), the claim mentioned in Sect. 1. If Stability is accepted,
then ¬a = b at @ will have to carry over to w1, contradicting a = b. Thus, the
origin of ¬a = b at @ and a = b at w1, ¬a = b ∧ ¬¬a = b at @, must be
denied. This argument may be called the argument from stability. We conjecture that
this is probably the source of many people’s gut feeling that ¬a = b at @ is already
contradictory to ¬a = b ∧ ¬¬a = b at @.
Clearly, the strength of the argument depends entirely on the plausibility of Stability
itself. But is it a plausible principle? Our answer is in the negative. Even though there
are many states of affairs that conform to the pattern prescribed in the principle, there
are also notable exceptions, and there is reason to think that the state of affairs ¬a = b
at @ is such an exception. So it does not carry over to w1 to produce a contradiction.
Instances of Stability are numerous. Everything that definitely holds at the actual
world @, such as ‘Michael Jordan is bald and tall’ and ‘Earth is the third planet from
the sun’, including all logical, mathematical, and/or necessary truths such as ‘ p ∨¬ p’,
‘2+3=5’, and ‘All bachelors are unmarried’, are all stable states of affairs and carry
over to all of @’s precisifications. Precisified states of affairs included not in the actual
world but in its precisifications, such as ‘Sue = Sally’ at w1 and ‘Sue = Sandy’ at w2,
‘k = k + ’ at w1, and ‘¬k = k + ’ at w2, are also generally stable and carry over to their
precisifications.
There are, however, also counterexamples to Stability. Take for example ‘Kiliman-
jaro is a vague object’ or ‘Kilimanjaro has vague boundaries’. This state of affairs ought
to be included in the actual world, but not in its precisifications at which Kilimanjaro is
completely precisified. Or consider ‘It is indeterminate whether Kilimanjaro is iden-
tical with Kilimanjaro+ ’, i.e., ¬k = k + ∧ ¬¬k = k + . As we have just seen at the
end of the last section, states of affairs of this kind are unstable and do not always
carry over to further precisifications. Generally, states of affairs about the indetermi-

123
Synthese (2014) 191:3557–3573 3567

nate features of the world and states of affairs involving vague objects as wholes and
not as superimpositions of precise objects will cease to carry over at some points, and
even the negations of those states of affairs will be included in precisifications. Such
exceptions to Stability have been known for years, but have often been dismissed as
anomalies (see, e.g., Lewis 1993).
Anomalies or not, unstable states of affairs exist, and exist for good reason. Thus,
Stability does not hold in full generality. But how about ¬a = b at @? Or, more
specifically, how about ¬k = k + at @? Should it be stable or unstable? The answer
is obvious: It should be unstable. That’s because it states that Kilimanjaro, a vague
object viewed as a whole, is distinct from Kilimanjaro+ , a (relevantly) precise object
viewed as a whole. This vague object is precisified into the one identical with k + at
w1, and the other identical with k − at w2. In contrast, again, k = k + included in
w1 and ¬k = k + included in w2 will carry over to their precisifications because k is
sufficiently precisified at them.
In sum, Stability does not hold in full generality, and does not hold in particular for
¬k = k + at @ and states of affairs of the same kind. Therefore, the argument from
stability fails.

3.2 Akiba’s analogical argument

Akiba (2000a,b, 2004) points out that Evans’s proof of DI is nothing but the Marcus–
Kripke proof of the Necessity of Identity and, as such, is perfectly legitimate; but he still
contends that the alleged examples are not genuine examples involving indeterminate
individuation. For instance, k and k + (or k − ), or Sue and Sally (or Sandy), are not
identical even at a precisification at which they coincide; thus, they are determinately
distinct at the actual world. Akiba defends his contention by drawing the following
analogy with a metaphysical case:

Following Gibbard (1975), suppose that Goliath is a statue and that Lumpl is
a single piece of bronze of which the statue is made; suppose, furthermore,
that Lumpl, as a single piece, is created and destroyed at the same time as
Goliath. Suppose someone contends that Goliath is thus identical with Lumpl.
One natural response to this contention ... is that they are not identical but only
coincident (or overlapping) throughout their existence, for they have different
dispositional or modal properties; for instance, Goliath can but Lumpl cannot
survive the loss of a small piece of the bronze, and Lumpl can but Goliath cannot
survive the flattening of the statue. That is, Goliath and Lumpl are coincident
(and observationally indistinguishable) in the actual world, but may not be in
some other possible worlds. Pace Gibbard, most of us do not take an example
like this to be an example of contingent identity; most of us take Goliath and
Lumpl to be nonidentical, even in the actual world. But then, by analogy, why
should we take ... Kilimanjaro and Kilimanjaro+ ... to be identical at least in
one scenario (or possible world) and, thus, indeterminately identical as a whole?
It seems more reasonable to say that ... Kilimanjaro and Kilimanjaro+ ... are
coincident but not identical in one scenario, while they are neither coincident

123
3568 Synthese (2014) 191:3557–3573

nor identical in another; thus, they are determinately distinct. (Akiba 2014, pp.
13–14)

Here the argument is that since the distinctness of Goliath (g) and Lumpl (l) at another
possible world implies their distinctness at the actual world, by analogy the distinctness
of k and k + at the actual world ought to imply their distinctness at its precisifications.
This argument fails in exactly the same way Evans’s argument fails. In the alethic
(or metaphysical) case, ¬g = l at some possible world accessible from the actual
world @; thus, ¬g = l at @. But g = g at @. Thus, by Leibniz’s Law, ¬g = l at
@. In the indeterminacy case, ¬k = k + at @. But this fact does not make ¬k = k +
true at any of @’s precisifications, ws, because even though @Rw at all such ws, we
need Axiom B and the symmetry of the accessibility relation to have w R@; without it,
the fact that ¬k = k + at @ cannot affect the conditions at ws and k = k + may be true
(and is indeed true in this case) at some of such ws (e.g., w1) whose precisifications
all include k = k + . Since the accessibility among precisifications are not symmetric,
Akiba’s analogical argument does not hold up.

3.3 Williamson’s balance argument

Williamson (2002) is mostly concerned with the aforementioned two strategies (i.e.,
the use of non-classical logic and semantic ascent), but he gives the following argument
against indeterminate individuation toward the end of his paper:

Recall the kind of situation which the notion of indeterminacy was introduced
to describe. Supposedly, the grounds for asserting something and the grounds
for asserting its negation are evenly balanced, and there is no room for further
considerations to decide the matter. If the relation between o and o* conforms
to this pattern, then the grounds for asserting that they are identical and the
grounds for asserting that they are not identical should be evenly balanced, and
there should be no room for further considerations to decide the matter. But, if
it is indeterminate whether they are identical, then the argument above [for the
contrapositive of DI that if they are not determinately identical, then they are
indeed not identical] does decide the matter, in favour of asserting that they are
not identical. The grounds are after all not evenly balanced. Thus the hypothesis
that it is indeterminate whether they are identical looks inconsistent with what
indeterminacy is supposed to be. (Williamson 2002, p. 292)

Again, in the situation we have been discussing, the actual world @ contains both
¬a = b ∧ ¬¬a = b and ¬a = b for some objects a and b (or Williamson’s o
and o*); and the former conjunction implies that there is a precisification w1 of @
at which a = b as well as a precisification w2 of @ at which ¬a = b. Williamson’s
argument, simply put, is that ¬a = b at @ breaks the balance between w1 and w2 in
favor of w2. So such a situation cannot obtain.
It is very difficult to make sense of Williamson’s argument from our viewpoint. He
talks as if the statement ‘it is indeterminate whether a = b (or o = o∗)’ describes
a single state of affairs that obtains only if the grounds for asserting a = b and the

123
Synthese (2014) 191:3557–3573 3569

grounds for asserting ¬a = b are evenly balanced; but in fact the statement is just a
conjunction of ¬a = b and ¬¬a = b, each of which may hold independently of
the other. For instance, what is the ground for asserting k = k + ? That’s the putative
fact that the atom Sparky belongs to Kilimanjaro (or s ∈ k for simplicity). What is the
ground for asserting ¬k = k + ? That’s the putative fact that the atom Sparky does not
belong to Kilimanjaro (¬s ∈ k). Insofar as there is a sufficient ground for an assertion
(of one or the other), it can be made, regardless of whether there is a ground for an
assertion of its denial. These grounds are included in the precisifications that include
k = k + and ¬k = k + , respectively: s ∈ k is included in w1 together with k = k + ,
and ¬s ∈ k is included in w2 together with ¬k = k + . Neither ground is included in
the actual world. Each conjunct ¬a = b or ¬¬a = b can hold true at the actual
world so long as its ground holds at some of its precisifications, regardless of whether
the ground for the other hold at other precisifications. There is no reason to think that
there ought to be any balance between the grounds. Furthermore, it is quite odd to take
¬k = k + at @ as a ground for anything because, again, the genuine grounds s ∈ k
and ¬s ∈ k are not included in @ but ¬k = k + is; so the status of the latter is quite
different from the status of the former.
For the state of affairs ¬k = k + at @ to be a ground for anything (and a ground
against k = k + at w1 in particular), it somehow must affect @’s precisifications (w1
in particular). That seems possible only if either the precisifications have access to @
or the state of affairs carries over to them. However, the former is possible only if the
accessibility relation is symmetric, and the latter is possible only if Stability holds for
the state of affairs; but, as we know, both possibilities have already been eliminated.

3.4 Smith’s denial

In the paper titled “ Why Sense Cannot Be Made of Vague Identity,” Nicholas Smith
(2008) tries to show exactly that. Against various attempts to make sense of vague
identity by assigning non-standard meanings to ‘=’, Smith argues that to understand
vague identity, we must have a mathematical model for identity statements, into which
the standard notion of identity is built; but that very notion eliminates the possibility
of indeterminate individuation. As we shall see, however, Smith ultimately fails to
answer why sense cannot be made of vague identity; he only makes a bald claim that
it cannot.
There is much in what Smith says we can agree with. We can wholeheartedly agree
that too many philosophers try to make indeterminate individuation statements true
by assigning non-standard meanings, but that that’s not the same as making sense
of indeterminate individuation. In this paper, however, because of the same concerns
we have decided not to talk about identity statements but to talk directly about the
identity relation; thus, most of Smith’s criticism does not apply to our defense. So
we can safely set it aside and ask Smith why the identity relation does not allow for
indeterminate individuation.
How does Smith show that we cannot make sense of indeterminate individuation?
In response to van Inwagen’s (1988) attempt to model indeterminate individuation,
Smith says thus:

123
3570 Synthese (2014) 191:3557–3573

We have been explicitly told that in a pairing {x, y}, x and y must be non-
identical. Now we are told that x and y are to be thought of as indefinitely
identical. I cannot make sense of this. If we are working with a standard set-
theoretic model, then x and y are simply non-identical; if we are not, then
unless we are given some other way to understand the presentation, we do not
understand it at all. The ordinary understanding of set-theoretic models rules out
van Inwagen’s proposed interpretation of his construction—yet no other way of
understanding the construction is presented. (Smith 2008, p. 7)

As you can see here, despite the title of his paper, Smith in fact does not give any reason
whatsoever why we cannot make sense of indeterminate individuation; all he says is
that we just cannot. All he says is that if x and y are “simply” non-identical, then they
cannot be indeterminately identical; that is, ¬x = y → ¬x = y. But this is nothing
but a statement of DD itself. So clearly Smith is not making any argument for it. Nor
does he make any argument in other parts of his paper. Perhaps Smith’s thought is that
if x and y are distinct simpliciter, they must be distinct in all contexts; in particular,
they must be distinct at all precisifications. Of course, if ¬x = y simpliciter implies
¬x = y at all precisifications, we cannot make sense of indeterminate distinctness; but
this implication is exactly what the advocates of indeterminate distinctness contend
with. They maintain, for instance, that k and k + are distinct members of the domain
of discourse, that ¬k = k + at the actual world, but that k = k + at some of its
precisifications.
Thus, Smith’s denial of indeterminate individuation is baseless. However, there is a
possible proposal against indeterminate individuation in the offing, inspired by Smith’s
thought, which may be worth considering even if it is itself not Smith’s own view:
Perhaps DD should be considered a primitive axiom, stemming from our concept of
identity. This way, we can dispense with the controversial Axiom B and the symmetry
of the accessibility relation. Is this proposal plausible?
It is not. Compare DD with DI. DI is not an axiom; it is a consequence of reasonable
axioms. But then why should DD be an axiom? Whence comes this asymmetry?
If we are asked why DI holds, we will reply thus: Suppose a and b are one and
the same thing, i.e., that a = b. It is a theorem, a universal truth, however, that
everything is identical to itself, and that a = a. So it must be the case that a = a
at all precisifications. But, then, since a = b, it must be the case that a = b at all
precisifications. Therefore, a = b. This justification is exactly what the proof of
DI provides. We cannot run a parallel argument for DD. Suppose a and b are distinct
things, i.e., that ¬a = b. But this is not a theorem, and here the universal truth of
a = a and b = b does nothing to help us to derive ¬a = b.
Compare DD also with the ND, ¬a = b → ¬a = b, which is considered a
theorem. If we are asked why ND holds, we will reply thus: Suppose, for reductio,
¬a = b and ¬¬a = b at some possible world w. Then, from the second claim, there
must be a possible world, w , at which a = b. By the Necessity of Identity, a = b at w
So, by the symmetry of the accessibility relation, a = b at w, contradicting ¬a = b at
w. For instance, since Goliath and Lumpl are distinct at the actual world, they cannot be
identical at any possible world. Thus, for any w, ¬a = b → ¬a = b at w. This proof,
of course, is analogous to the second part of Evans’s proof we have considered, and is

123
Synthese (2014) 191:3557–3573 3571

making use of Axiom B and the symmetry of the accessibility relation; but that is quite
acceptable in this case because the logic of alethic modality is usually considered S5,
for which the accessibility relation is an equivalence relation—reflexive, symmetric,
and transitive. This difference is the reason why Akiba’s analogical argument did not
hold up.
An important consequence of this line of thinking is that there may be worlds at
which it is not determinate that ¬a = b (i.e., ¬¬a = b) but it is necessary that
¬a = b (i.e., ¬a = b); indeed, the actual world in the Kilimanjaro example is
such a world, at which ¬k = k + , ¬¬k = k + , and ¬k = k + . A similar point
has indeed already emerged in our discussion of the strengthened Evans principle
(¬x = y ∧ ¬¬x = y) at the end of Sect. 2. A further consequence of this
difference between  and  is that precisificationally accessible worlds are not always
alethically accessible; for instance, w1 in the Kilimanjaro example, at which k = k + , is
precisificationally accessible from the actual world, but not alethically accessible. It is
sometimes taken for granted that any possible world ought to be alethically accessible.
But if we are correct, that cannot be the case. Needless to say, the converse relation does
not hold, either: there are alethically accessible worlds that are not precisificationally
accessible. The possible worlds at which Michael Jordan has much hair right now are
alethically accessible from the actual world, but not precisificationally accessible.
In sum, among the four theses, DI, DD, NI, and ND, DD is the only thesis that is
not derivable from other, independently plausible axioms. We do not have to justify
the others by simply insisting that they stem from our concept of identity; we have
better justifications, as spelled out above. In particular, we do not have to justify DI by
simply saying, ‘if x and y are identical simpliciter, then, by the meaning of identity, x
and y must be identical in all contexts’; we have a better story to tell. Thus, it is quite
ad hoc to insist that DD stems from our concept of identity and should be taken as a
primitive axiom.

4 Conclusion

This paper has distinguished two theses, the DI and the DD, and argued that while
the former is correct, there are reasons to think that the latter is not. The former is
derivable from independently plausible logical axioms, but the latter relies crucially on
Axiom B and the symmetry of the accessibility relation, which, however, do not hold
for determinacy if the accessibility relation is understood as the ‘is a precisification
of’ relation. Evans’s refutation of indeterminate individuation needs both theses; thus,
it cannot but fail. The argument from stability, a way to rescue Evans’s argument, also
fails because the principle on which the argument is based, the stability principle, does
not hold in full generality and does not hold for the relevant non-identity statements
in particular. Akiba’s, Williamson’s, and Smith’s arguments also fail for related and
unrelated reasons.
The defense of indeterminate individuation given in this paper is quite different
from most other defenses in the literature, for it agrees with the critics of indeterminate
individuation and disagrees with most of its defenders about the correctness of DI. The
paper argued that the refutations stumble not on DI but on DD. It is unfortunate that

123
3572 Synthese (2014) 191:3557–3573

so much attention has been paid to DI and so little to DD. Indeed, DD is often even
not clearly distinguished from DI (though Williamson 1996 is a clear exception). The
common name of the issue, the problem of vague (or indeterminate) identity, itself
indicates people’s tendency to conflate—or at least not clearly distinguish — the two
separate issues.
Two other sources of confusion have been mentioned and avoided in this paper: the
use of non-classical logic and semantic ascent. The excessive focus on DI and these
two strategies are not unrelated. Since Evans’s argument for DI is essentially (if not in
his original presentation) so simple, straightforward, and obviously valid in classical
logic, anybody who wants to refute it must resort to something beyond the ordinary.
Both strategies have serious drawbacks, as alluded to earlier. They also tend to deviate
critics’ attention away from the main issues. Although we have set it aside, most of
Williamson’s and Smith’s discussion actually concerns those strategies. Indeed, we
rarely see any defense of DD in the literature. This paper’s main point was that we
can defend indeterminate individuation within the standard, classical framework and
without employing those strategies by rejecting not DI but DD.

References

Akiba, K. (2000a). Vagueness as a modality. Philosophical Quarterly, 50, 359–370.

Akiba, K. (2000b). Identity is simple. American Philosophical Quarterly, 37, 389–404.
Akiba, K. (2004). Vagueness in the world. Noûs, 38, 407–429.
Akiba, K. (2014). Introduction. In Akiba and Abasnezhad (2014, pp. 1–21).
Akiba, K., & Abasnezhad, A. (Eds.). (2014). Vague objects and vague identity. Dordrecht: Springer.
Barcan, R. (1947). The identity of individuals in a strict functional calculus of second order. Journal of
Symbolic Logic, 12, 12–15.
Barnes, E. (2009). Indeterminacy, identity and counterparts: Evans reconsidered. Synthese, 168, 81–96.
Barnes, E., & Williams, J. R. G. (2011). A theory of metaphysical indeterminacy. In K. Bennett & D. W.
Zimmerman (Eds.), Oxford studies in metaphysics (Vol. 6, pp. 103–148). Oxford: Oxford University
Press.
Bobzien, S. (2012). If it’s clear, then it’s clear that it’s clear, or is it? Higher-order vagueness and the S4
axiom. In B. Morison & K. Ierodiakonou (Eds.), Episteme, etc.: Essays in honor of Jonathan Barnes
(pp. 189–212). Oxford: Oxford University Press.
Dummett, M. (1975). Wang’s paradox. Synthese, 30, 301–324.
Edgington, D. (2000). Indeterminacy de re. Philosophical Topics, 28, 27–44.
Evans, G. (1978). Can there be vague objects? Analysis, 38, 208.
Field, H. (2000). Indeterminacy, degree of belief, and excluded middle. Noûs, 34, 1–30.
Fine, K. (1975). Vagueness, truth and logic. Synthese, 30, 265–300.
Geach, P. (1962). Reference and generality. Ithaca, NY: Cornell University Press.
Gibbard, A. (1975). Contingent identity. Journal of Philosophical Logic, 4, 187–222.
Greenough, P. (2003). Vagueness: A minimal theory. Mind, 112, 235–281.
Hawley, K. (2002). Vagueness and existence. Proceedings of Aristotelian Society, 102, 125–140.
Hughes, G. E., & Cresswell, M. J. (1996). A new introduction to modal logic. London: Routledge.
Kripke, S. (1971). Identity and necessity. In M. Munitz (Ed.), Identity and individuation (pp. 135–164).
New York: New York University Press.
Lewis, D. (1988). Vague identity: Evans misunderstood. Analysis, 48, 128–130.
Lewis, D. (1993). Many, but almost one. In K. Campbell, J. Bacon, & L. Reinhardt (Eds.), Ontology,
causality, and mind (pp. 23–38). Cambridge: Cambridge University Press.
Marcus, R. (1961). Modality and intensional languages. Synthese, 13, 303–322.
Parsons, T. (1987). Entities without identity. Philosophical Perspectives, 1, 1–19.
Parsons, T. (2000). Indeterminate identity. Oxford: Oxford University Press.

123
Synthese (2014) 191:3557–3573 3573

Parsons, T., & Woodruff, P. (1995). Worldly indeterminacy of identity. Proceedings of the Aristotelian
Society, 95, 171–191.
Salmon, N. (1981). Reference and essence. Princeton: Princeton University Press.
Shoemaker, S. (1963). Self-knowledge and self-identity. Ithaca, NY: Cornell University Press.
Shoemaker, S., & Swinburne, S. (1984). Personal Identity. Oxford: Blackwell.
Smith, N. J. J. (2008). Why sense cannot be made of vague identity. Noûs, 42, 1–16.
van Inwagen, P. (1988). How to reason about vague objects. Philosophical Topics, 16, 255–284.
Weatherson, B. (2003). Many, many problems. Philosophical Quarterly, 53, 481–501.
Williams, J. R. G. (2008). Multiple actualities and ontically vague identity. Philosophical Quarterly, 58,
134–154.
Williamson, T. (1996). The necessity and determinacy of distinctness. In S. Lovibond & S. G. Williams
(Eds.), Essays for David Wiggins: Identity, truth and value (pp. 1–17). Oxford: Blackwell.
Williamson, T. (1999). On the structure of higher-order vagueness. Mind, 108, 127–143.
Williamson, T. (2002). Vagueness, identity and Leibniz’s law. In A. Bottani, M. Carrara, & P. Giaretta
(Eds.), Individual essence and identity (pp. 273–304). Dordrecht: Kluwer.

123