SUSAN C.

H A L E

SPACETIME AND THE ABSTRACT/CONCRETE

DISTINCTION*

(Received 24 November, 1986)

The distinction between abstract and concrete entities has a twofold

purpose in contemporary philosophy of mathematics. First, this dis-
tinction is meant to divide all of our ontology into two fundamentally
different kinds. 1 Second, this distinction is used to separate entities
which are epistemicaUy troublesome in that we cannot interact With
them causally, i.e. abstract entities, from entities which do not pose this
particular epistemic problem, i.e. concrete entities.
The standard objection to platonist theories of the foundations Lof
mathematics, which posit some abstract entities such as numbers, sets,
and functions, bears witness to the epistemological use of the abstract/
concrete distinction. The standard objection is that it is at best unclear
how we could have epistemic access to abstract entities; perhaps, and
far worse for the platonist, epistemic access to abstract entities is
a priori impossible.
This objection rests crucially on two premises, which are:

(a) " . . . for X to know that [a sentence] S is true requires some

causal relation to obtain between X and the referents of the
names, predicates, and quantifiers of S"; 2 and,
(b) abstract entities cannot enter into causal relations?
The abstract/concrete distinction can fulfill its twofold purpose in con-
temporary philosophy of mathematics only if it can satisfy conjointly
the following three conditiofis:

C 1. Clarity Condition: it can be clearly drawn;

C2. Ontological Significance Condition: it can be shown to
divide our ontology into two fundamentally different kinds;
and,
C3. Epistemological Significance Condition: it can be shown to
have some epistemological significance.

PhilosophicalStudies 53 (1988) 85--102.

9 1988 byD. ReidelPublishing Company.
86 SUSAN C. HALE

It should be noticed that not all three of these conditions must be

met in order for the abstract/concrete distinction to be a legitimate
ontological distinction. Only C1 and C2 must be satisfied for this
purpose.
Until very recently, it has been assumed that the abstract/concrete
distinction is so commonly understood that there is no need to state it
clearly and precisely before drawing conclusions, ontic or epistemic,
from it. However, the distinction has been drawn in quite a variety of
ways, none of which is obviously equivalent to any other. 4
As David Lewis has pointed out, " . . . the historically and etymologi-
cally correct thing to mean if we talk of 'abstract entities'" is that
"abstract entities are abstractions from concrete entities." 5 It was in this
sense that 'abstract' was used, for but one instance, in the dispute
between Locke and Berkeley over the doctrine of abstract ideas. But, as
Lewis also notices, this is " . . . by no means the dominant meaning in
contemporary philosophy". 6 There is no One dominant meaning in
philosophy now. Rather, many versions of the distinction are used and
these different versions are often taken uncritically to be equivalent.
Among these are the following:

(1) Abstract entities are not in spacetime whereas concrete entities

are; 7
(2) Abstract entities cannot participate in causal networks but
concrete entities can; 8
(3) Abstract entities have only relational properties while concrete
entities have some intrinsic properties; 9
(4) Abstract entities are universals and concrete entities are partic-
ulars; 10
(5) Abstract entities are sets and concrete entities are individuals; 11
(6) Abstract entities are never indiscernible from one another
whereas concrete entities (sometimes) are; 12
(7) Abstract entities are human constructions or creations, whereas
concrete entities exist independently of human minds or lan-
guage; 13
(8) Abstract entities are types and concrete entities are tokens; 14
(9) Negative terms apply, but positive ones do not, to abstract
entities; 15
(10) Concrete entities are known by observation whereas abstract
 THE ABSTRACT/CONCRETE DISTINCTION 87

entities are known in some other way, e.g., by abstraction, by

intuition, a priori; 16
(11) Concrete entities "can be pinned down by pointing" but ab-
stract entities cannot be ostended; iv
(12) A n individual is concrete if and only if it is exhaustively
divisible into concreta, i.e., into fully determinate parts, and an
individual is abstract if and only if it contains no concretum.~8

In the rest of this p a p e r I shall focus only on the first version of the
distinction that I have catalogued above, viz., the version according to
which abstract entities are not in spacetime and concrete ones are. For
brevity, I shall call this "the spacetime criterion". The main reason for
my choice is that this version seems to have the greatest credence
among philosophers today. I shall demonstrate that, while it may be
possible for the spacetime criterion to meet the clarity condition, it is
no trivial matter to draw this distinction clearly. Rather, adopting a
clear and precise statement of the spacetime criterion commits one to
controversial claims about the nature of spacetime and about whether
or not commitment to spacetime points can be abandoned in favor of
arbitrarily small spacetime regions.
Perhaps it will seem to the reader that what it means to say that an
entity is in s p a c e t i m e is so obvious that time spent clarifying this is
wasted. I would like to dispel this appearance of obviousness by looking
at a test case which has recently gained considerable importance in
current philosophy of mathematics, viz., spacetime points.
In Science Without Numbers, Hartry Field argues for the possibility
and attractiveness of doing physics without ontic commitment to
mathematical entities, i.e. without quantifying over mathematical entities.
His argument for the possibility of such an enterprise relies crucially on
two premises, Which are:
(1) Mathematics does not have to be true in order to be applicable to the natural
world; it needs only to be conservative over science. Roughly, that is, if we add a
(good) mathematical system S to a set of nominalistically statable scientific asser-
tions N, we will not be able to derive any nominalistically statable conclusions from
N + S that we cannot derive from N alone; 19
and
(2) Typical platonistic physical theories, such as Newtonian gravitational theory, can be
nominalistically reformulated with a purely concrete ontology which has the same
structural features as the ontology of mathematized science.2~
88 SUSAN C. H A L E

In defending this second claim, Field provides a reformulation of

N e w t o n i a n gravitational theory which he takes to be nominalistic. Since
the core of N e w t o n i a n gravitational theory is the N e w t o n i a n theory of
spacetime, Field begins his p r o g r a m with a reformulation of this theory
of spacetime. His strategy is to base the axiom system for N e w t o n i a n
s p a c e t i m e on the axiom system for affine g e o m e t r y d e v e l o p e d by L. W.
Szczerba and A. Tarski. T h e uniqueness t h e o r e m of the Szczerba-
Tarski a x i o m system explains invariance u n d e r affine transformations.
Field restricts the Szczerba-Tarski axiom system by providing a unique-
ness t h e o r e m which explains the invariance of the laws of N e w t o n i a n
mechanics only u n d e r Galilean transformations, which are a special
case of affine transformations. H e p r o c e e d s to replace descriptions of
the structure of this system by m e a n s of o r d e r e d quadruples of real
n u m b e r s with descriptions which refer to spacetime points, 2~ which he
regards as a special case of spacetime regions, " . . . n a m e l y as regions of
minimal size". 22
A l t h o u g h Field writes of abstract entities as "causally isolated f r o m
everything we o b s e r v e " and as "existing s o m e w h e r e outside of space-
time", 23 he uses neither of these versions of the a b s t r a c t / c o n c r e t e
distinction when defending the claim that ontic c o m m i t m e n t to space-
time points is nominalistically acceptable. Instead he uses one of the
directly epistemological versions catalogued above. H e writes:

Perhaps it is a bit odd to use the phrase 'physical entity' to apply to spacetime
points. But however this may be, spacetime points are not abstract entities in any
normal sense. After all, from a typical platonist perspective, our knowledge of mathe-
matical structures of abstract entities (e.g. the mathematical structure of real numbers)
is a priori; but the structure of physical space is an empirical matter. That is, most
platonists who believe current physical theory believe that it is a priori true that there
are real numbers obeying the usual laws, and that it is a high-level empirical hypothesis
... that there are lines in space which (locally anyway) are isomorphic to the real
numbers. No platonist would identify the real numbers with the points on any physical
line: for one thing, it would be arbitrary which line one picked to identify the real
numbers with, and arbitrary which point on the line to identify with 0 and which with 1;
but more fundamentally, to make any such identification would be to identify the real
numbers with something we can know about only empirically. 24

I will not dispute Field's claim that spacefime points are c o n c r e t e by

this directly epistemological criterion. Rather, I would like to use
spacetime points as a test case for the clarity of the s p a c e t i m e criterion
(a criterion which Field also uses, as we have seen above).
 THE ABSTRACT/CONCRETE DISTINCTION 89

FOUR READINGS OF THE SPACETIME CRITERION

A. 'Is Located in Spacetime'

What do we mean by 'in' when we say that an entity is in spacetime?

An immediate suggestion is that an entity is in spacetime if and only if it
is located in spacetime. Are spacetime points concrete entities by this
reading of the spacetime criterion?
In post-Newtonian physics spacetime is a four-dimensional manifold
in which material events are located. Spacetime points are the funda-
mental event-locations in this manifold. Yet it makes no sense to say
that a location is in itself. Since spacetime points are the fundamental
locations, we cannot find some other reference point by which to locate
spacetime points in the manifold. Thus, because of their role in locating
events in spacetime, spacetime points themselves cannot be said to be
located in spacetime; they are locations, not located. On this reading of
the spacetime criterion, spacetime points are abstract if they exist.
On this reading of 'in spacetime' spacetime points are analogous to
Wittgenstein's famous case of the standard metre bar in Paris, " . . . of
which one can say neither that it is one metre long, nor that it is not one
metre long ...-.25 Of this example Wittgenstein writes that his con-
siderations should not be taken as ascribing any extraordinary property
to the standard metre but rather as marking " its peculiar role in the
language-game of measuring with a metre-rule". 26 The conceptual
situation with spacetime points is closely analogous; their special role of
locating material events in spacetime shows that they are not them-
selves located in spacetime.
Let us suppose that we wish to find a reading of the spacetime
criterion according to which spacetime points, if such there be, are
concrete entities, and let us look at several different readings of 'in
spacetime' to see if we can find one which will serve this purpose.

B. 'Is Spatiotemporally Extended'

An alternative reading of 'is in space' was suggested by Einstein, viz., 'is

spatially extended'. 27 To put Einstein's suggestion in more modern
terminology, we could read 'in spacetime' as 'is spatiotemporally
90 S U S A N C. H A L E

extended'. Since Einstein proposed this replacement prefatory to

attempting to show that general relativity does not imply that space is
absolute, but allows a relationalist theory instead, we might guess that
'is spatiotemporally extended' will not suffice to admit spacetime points
as concrete entities.
We can see that this guess is correct as follows. Since spacetime
points are the fundamental event-locations, they must serve to locate
idealized mass points. These idealizations are unextended. Material
events are constituted of these idealized mass points as a whole is
constituted of its parts. Each unextended mass point is coincident
with a spacetime point, so spacetime points themselves must also be
unextended. Thus, this reading of the spacetime criterion classifies
spacetime points as abstract.

C. 'Is Part of Spacetime'

Another reading of 'in spacetime' is that an entity is in spacetime if and

only if it is a part of spacetime. 28 Since spacetime points are the
fundamental event-locations, they are the fundamental parts of which
spacetime is composed, assuming that spacetime is indeed composed of
parts. Thus, on this reading of the spacetime criterion, spacetime points
are concrete.
However, this reading appears to fail as an adequate necessary
condition for concreteness, for it does not classify physical objects or
material events as concrete, since physical objects and material events
are not, at least on the ordinary conception, themselves parts of
spacetime. I will consider a reply to this objection below.

D. 'Is Part of Spacetime or Is Located in Spacetime'

A fourth suggestion for clarification of the spacetime criterion is that
an entity is in spacethne if and only if either it is part of spacetime or it
is located in spacetime. 29 This reading admits both spacetime points
and material entities which are not parts of spacetime as concrete,
while classifying paradigmatically abstract entities such as numbers as
abstract.
Is Field free then to endorse this fourth suggestion as the correct
 THE ABSTRACT/CONCRETE DISTINCTION 91

reading of the spacetime criterion without further argument? I shall

argue that he is not, for if he were to adopt the latter reading he would
thereby commit himself to substantivalism9
The two major competing theories about the nature of spacetime
since the days of Newton and Leibniz have been substantivalism and
relationalism. According to substantivalists, spacetime itself is an entity
with a structure and with features over and above the features of
material events. According to relationalists, spacetime itself is not an
entity but rather spatio-temporal talk is a way of talking about the
systematic structure of relations which ordinary material events bear to
one another.
Two substantivalist positions can be distinguished. First, there is the
less extreme position according to which spacetime is "one of the many
real existents constituting the world", as Lawrence Sklar puts it. 31
Second, there is a more radical substantivalist position according to
which spacetime and its parts are the only existent entities. Proponents
of this second position, which I shall call 'supersubstantivalism', attempt
to reduce any apparent material event to the region of spacetime
occupied by that material event.
Let us look at our four suggested readings of 'in spacetime' again,
to see whether their adoption commits one to substantivalism, super-
substantivalism, or relationalism.
The first suggested reading was that an entity is in spacetime if and
only if it is located in spacetime. It may look as if adoption of this
reading commits one to substantivalism, but this is not the case. This
appearance arises from the locution 'located in', which seems to imply
both an occupier and an occupied when i{ is used in a true assertion.
The occupier here would be some concrete entity, by definition, and
the occupied would be spacetime itself. However, one can adopt this
reading of the spacetime criterion and yet avoid commitment to sub-
stantivalism in the following way. One may deny that 'in' is used here in
the sense of mere physical inclusion and then one may explicate
'located in' by claiming that its use in a true assertion indicates that a
material event bears some~ spatiotemporal relation to other material
events. Thus, this reading of the spacetime criterion is compatible with
relationalism.
The debate over whether such a relationalist may or may not avoid
92 SUSAN C. HALE

commitment to relations, and therefore the debate over whether or not

a nominalistically acceptable relationalism can be had, involves a
variety of physical and philosophical issues, which I cannot hope to
settle here. If the relationalist abandons talk of material events and
limits himself to talk of material objects only, he can avoid such
commitment, for one can specify a spatiotemporal relation which every
material object bears to some other material objects, viz., distance.
There is a unique distance interval between the centers of mass of any
two material objects which have mass. If we admit material objects
without mass, and neutrinos might turn out to be such objects, then a
unique distance interval can be found by specifying the interval between
nearest point and nearest point or between furthest point and furthest
point. Thus, the relationalist can explicate his talk of material objects in
spacetime without quantifying over relations and limit the values of his
quantificational variables to material objects.
However, it is unclear that the relationalist can specify any spatio-
temporal relation or set of spatiotemporal relations which hold between
every event and some other event and which hold only between
spatiotemporally located events. Thus, it is unclear whether or not the
relationalist can talk of events in spacetime without quantifying over
relations. One suggestion for such a set of relations focuses on purely
temporal relations, as follows: An event x is in time if and only if there
is an event y such that x is before y or x is after y or x is simultaneous
with y or x temporally overlaps y. However, if one believes in events
with temporal but not spatial duration, e.g., Cartesian mental events,
then this set of purely temporal relations will not serve as an explication
of 'is in spacetime'. But if/one refuses commitment to events with
temporal but not spatial duration, then specification of this set of
temporal relations will allow one to limit his quantification to first-order
quantification over material events and be able to avoid commitment to
relations.
This reading of :in spacetime' is also compatible with substantivalism,
but not without a price. As we have seen earlier, this reading classifies
spacetime points as abstract; so a substantivalist who wants to adopt
this reading will have to find a replacement for spacetime points as
fundamental event-locations, since substantivalists like Field want
spacetime and its parts to be concrete. Substantivalists might be able to
 THE A B S T R A C T / C O N C R E T E D I S T I N C T I O N 93

find such a replacement by paraphrasing talk about spacetime points as

limit-talk about arbitrarily small spacetime regions. H o w e v e r , they
might not.
Such a strategy was developed, in a different context, by A. N.
W h i t e h e a d in Process and Reality. In discussing extensive connection,
he defines o n e connective relation, viz., covering, as follows:

Definition 11. An abstractive set a is said to 'cover' an abstractive set/3, when every
member of the set a includes some members of the set/3.32

C o m m e n t i n g on this definition, he writes:

It is to be noticed that each abstractive set is to be conceived with its members in serial
order, determined by the relation of inclusion. The series starts with a region of any
size, and converges indefinitely towards smaller and smaller regions, without any
limiting regions. When the set a covers the set/3, each member of a includes all the
members of the convergent ~tail of t3, provided that we start far enough down in the
serial arrangement of the set/3?3

Later he defines 'point' as follows:

Definition 16. A geometrical element is called a 'point', when there is no geometrical

element incident in it. 34

H e defines 'geometrical element' as follows:

Definition 13. A geometrical element is a complete group of abstractive sets equivalent

to each other, and not equivalent to any abstractive set outside the group. 35

M u c h m o r e exegetical detail would be needed for the reader to under-

stand Whitehead's strategy fully, but no m o r e is n e e d e d for m y p u r p o s e
here. Whitehead's remarks o n Definition l I a n d Definitions 13 and 16
taken together b o t h illustrate one p r o b l e m a substantivalist will face if
he attempts to a d o p t this strategy; for Whitehead, points, or : m o r e
accurately, point-replacements, are abstractive sets (whose m e m b e r s are
regions). 36 It is uncontroversial that sets are abstract entities, even
paradigmatically so; thus, it is uncontroversial that point-replacements
are abstract o n this strategy. While c o m m i t m e n t to these abstract
entities m a y n o t be unjustifiable in and of itself, this c o m m i t m e n t
cannot be justified b y a substantivalist; for substantivalists claim that
spacetime and its parts are existent, concrete entities. P e r h a p s a clever
substantivalist could find a way to a d o p t arbitrarily small spacetime
94 SUSAN C. HALE

regions as a replacement for spacetime points without thereby com-

mitting himself to sets, but it is difficult to imagine how convergence
could be explicated without commitment to sets or sequences.
There is a further problem with this sort of replacement which is
troublesome to substantivalists and relationalists alike. This problem is
that any such replacement is supposed to provide an adequate basis for
physical geometry. We have noticed earlier that spacetime points must
serve to locate idealized mass points and that idealized mass points are
unextended. Arbitrarily small spacetime regions, no matter how small,
never converge on a point-limit; hence, arbitrarily small spacetime
regions are extended. So they Cannot serve as locations for idealized
mass points. Therefore, a system of physical geometry in which there
are no spacetime points, but arbitrarily small spacetime regions instead,
is an inadequate basis for physical theory unless physical theory can be
reformulated without idealized mass points. As far as I know, no such
reformulation has been attempted yet. 37
This strategy to make the first suggested reading of 'in spacetime'
consistent with substantivalism, then, is no more than a promissory
note, and we have two good reasons for s~specting that this note cannot
be cashed. First, the problem with idealized mass points shows that we
have good reason to doubt that replacement of points by limit-regions
can provide an adequate basis for physical geometry at all. Second, the
problem with commitment to arbitrarily small regions as sets shows that
we have good reason to doubt that any such replacement will be
consistent with substantivalism.
The second suggested reading of 'in spacetime' was Einstein's
suggestion, viz,, 'is spatiotemporally extended'. Since this reading of the
spacetime criterion implies the same thing about the categoreal status of
spacetime points as our first proposal does, similar reasoning in this
case will show that this reading is consistent with either relationalism or
substantivalism with spacetime points replaced by limit-regions if such a
replacement can be accomplished.
The third suggestion was that an entity is in spacetime if and only if
that entity is a part of spacetime. As was noted above, this reading
seems to imply that spacetime points and regions are concrete entities
but that material events are not concrete. But someone who wishes to
adopt this reading of the spacetime criterion can avoid this apparently
 THE ABSTRACT/CONCRETE DISTINCTION 95

absurd conclusion by identifying material entities that do not appear to

be themselves parts of spacetime with the spacetime regions which they
occupy. However, this identification commits.one to supersubstantival-
ism, the view that spacetime is not only a real existent but, more
strongly, that spacetime and its parts are the only real existents. There-
fore, one who adopts this reading of the spacetime criterion must
embrace either supersubstantivalism or the absurd conclusion that
material events are abstract entities.
The fourth suggestion was that an entity is in spacetime if and only if
either that entity is located in spacetime or that entity is a part of
spacetime. This reading admits both spacetime points (and other
regions of spacetime) and material events which are not parts of
spacetime as concrete, without committing one to supersubstantivalism.
But this version of the spacetime criterion is not free of commitments
about the nature of spacetime, for it commits one to the less radical
substantivalist position according to which spacetime and its parts are
real existents of the material universe because it classifies all spacetime
parts as concrete.
It might appear to the reader that such commitments are fairly
innocuous. I shall attempt to dispel this appearance by showing that
nontrivial questions must be answered before the substantivalism-rela-
tionalism debate can be justifiably settled. This debate is quite complex,
and I shall not attempt to give a complete or terribly sophisticated
catalogue of all the arguments for each position which can be found in
the literature. Nor shall I attempt to determine which theory is the
better theory of the nature of spacetime. My exposition will be limited
to the amount of detail needed to show that commitments about the
nature of spacetime are hardly innocuous.
The major argument for substantivalism is that it fits better with the
language of scientists than relationalism does. As Sklar puts it, the
relationalist
is unable to account in his theory for expressions commonly taken as intelligible in
"ordinary and scientific discourse. The relationist cannot account for the notion of an
unoccupied spatiotemporal location; or make sense of the notion of a universe totally
devoid of material things . . . . 38

The relationalist can answer this charge in at least two different ways.
First, he can simply deny that the notion of empty spacetime is
96 SUSAN C. HALE

coherent, and assert that the fact that the phrase 'empty spacetime' is
bandied about in ordinary and scientific discourse does not establish its
intelligibility.
Second, the relationalist may invoke possibilia in reply to this
objection. H e m a y take unoccupied spatiotemporal locations in a
n o n e m p t y universe as possible but nonactual relations to material
events and interpret talk about empty spacetime as talk about the
lawlike structure that would govern entities in this spacetime were there
any. If the relationalist argues this way, the debate may then focus on
the legitimacy of talk of possibilia.
The major argument for relafionalism has two steps. First, a vefifica-
fionist principle is asserted. Sklar puts this principle as follows:

The meaningful assertion of the existence of some entity or feature of the world
requires that the presence or absence of that entity or feature, or a change in that
feature, have some observational consequences.39

Second, it is argued that c o m m i t m e n t to substa:,xivalism implies the

truth of statements which are meaningless according to the principle
quoted above. For example, if spacetime itself were a substance then it
would be meaningful to ask what the position of the material world in
spacetime is. This question, it is then argued, has no meaningful answer
according to the above principle, assuming that the spatiotemporal
relations of material events relative to one another would remain
constant if the material world w e r e to change positions relative to
spacetime itself.
A version of this argument, which follows, was given by Leibniz in a
letter to Samuel Clarke:
Motion does not indeed depend upon being observed; but it does depend upon being
possible to be observed. There is no motion, when there is no motion to be observed:
And when there is no change that can be observed, there is no change at all. The
contrary opinion is grounded upon the assumption of a real absolute space.4~

The substantivalist may reply to this argument either by denying that

the verificationist principle is sound scientific methodology, or by
denying that c o m m i t m e n t to spacetime itself has no observationally
meaningful consequences. It is this latter course which Newton t o o k
when he argued, with this famous bucket argument, that the absolute
acceleration of a physical object, relative only to space itself, has
consequences which can be observed. 41
 THE ABSTRACT/CONCRETE DISTINCTION 97

A current major controversy is whether or not relationalism is

compatible with general relativity. Implicit in these debates is the
principle that our ontological commitments ought to be consistent with
our best scientific theories, and that in the physical realm general
relativity is our best theory.
Three positions can be taken in this debate, and all three are.
First, one may argue that substantivalism is inconsistent with general
relativity; Einstein argued this. Second, one may claim that both
substantivalism and relationalism are consistent with general relativity;
Sklar suggests an argument to the first half of this conclusion. Third,
one may argue that relationalism is inconsistent with general relativity;
Field argues this.
Einstein claimed that, according to general relativity:

There is no such thing as an empty space, i.e. a space without field. Spacetime does not
claim existenceon its own,but only as a structuralqualityof the field. 42

His argument for this claim is based on taking an empty spacetime to

be equivalent to a field-flee spacetime, and is as follows. The spacetime
appropriate to general relativity is the Riemannian generalization of the
pseudo-Euclidean four-dimensional Minkowski spacetime of special
relativity; this generalization allows for curvature of the manifold. How-
ever, the Riemannian generalization admits field-free, flat Minkowski
spacetime as a special case for which the equation

(I) ds 2 = dXl 2 ~ dx22 -t- dx32 - dx42

has an "objective metrical significance", in Einstein's words. 43

However, if the spacetime described by (I) is subjected to an
arbitrary continuous transformation of the co-ordinates then ds comes
to be expressed in the new system by

(Ia) ds 2 ~" gikdXidXk 44.

(Ia) d~scribes the general field law of pure gravitation in general

relativity, where 'gik' is a function denoting the pure gravitational field.45
If we try to imagine a field-free, and thus empty, spacetime, then we
must be able to imagine removing the function 'gik' from the field
equation. However, this we cannot do, writes Einstein, "for the func-
tions gik describe not only the field, but at the same time also the
topological and metrical structural properties of the manifold".46
98 SUSAN C. HALE

Field has argued that relationalism is inconsistent with field theories,

as contrasted with action-at-a-distance theories. Since Field takes "the
metric or gravitational fields characteristic of general relativity" to be a
paradigm example of a field theory, 47 his argument can be read as an
argument that general relativity and relationalism are incompatible, and
is probably best so read. Here Einstein and Field directly oppose one
another.
Field argues that field theories, by definition, presuppose substan-
tivalism. He defines a field theory as a "theory that employs causal
predicates that apply to spacetime points or . . . regions directly". 48 He
presents a relationalist objection to this definition, which asserts that
spacetime and fields have been conflated therein. But Field dismisses
this objection as trivializing relationalism, since this view requires that
"we have to view field theories as postulating entities whose geometric
properties are exactly the same as the geometric properties that the
substantivalist ascribes to spacetime". 49 Not only is this a trivialization,
according to Field, but also it forces us to "alter the way we think about
field theories". 5~ However, this alleged trivialization does seem to
correspond better to how Einstein thought about field theories, as the
above quotes from Einstein show.
Although most of Sklar's discussion of general relativity is concerned
with the role of absolute motion in the theory rather than with the place
of general relativity in the more general debate between substantivalists
and relationalists, he does give a more general argument for the
consistency of substantivalism with general relativity. Roughly, his
argument is that since the Riemannian generalization does not uniquely
determine one metric field, empty spacetimes are admitted; Minkowski
spacetime is one such empty spacetime. Furthermore, he claims that
empty Minkowski spacetime plays a crucial role in general relativity in
that it is the "natural choice of background spacetime to which to add
the contributions of mass-energies to the metric field . . . . ,, 51
Einstein and Sklar disagree about the philosophical significance of
the role that Minkowski spacetime plays in general relativity. Einstein
dismisses flat Minkowski spacetime as a "special c a s e . . , which in itself
has no objective significance". 52 It is important to notice that Einstein
distinguishes between objective metrical significance and objective
significance in itself; he admits that Minkowski spacetime has the
 THE ABSTRACT/CONCRETE DISTINCTION 99

former but denies that it has the latter. Thus, for Einstein, the mathe-
matical place of Minkowski spacetime in general relativity does not
show that the existence of an empty spacetime is consistent with general
relativity. This claim of Einstein's is connected integrally with his
general views on the epistemology of physics and its relation to the
ontology of physics. This dispute, then, can only be adjudicated against
the background of an overall assessment of the epistemology of physics
and its relation to ontology.
I have not attempted to solve the issue of whether substantivalism or
relationalism is the better theory of the nature of spacetime, nor have I
attempted to give a complete and sophisticated survey of the arguments
for and against each position. Rather, I have tried simply to demon-
strate that the issue is very complex.
From my discussion, we can extract at least five questions which
must be answered before the issue can be settled. These are:

(1) Is Leibniz's verificationist principle sound scientific methodol-

ogy?
(2) Does the assertion of the existence of spacetime have observa-
tional consequences?
(3) What is the role, both scientific and inferential, of absolute
acceleration in physics, and is this role justified?
(4) What is the theoretical and epistemological role of Minkowski
spacetime, and other spacetimes, in general relativity, and how
are these roles related to the ontology of physics?
and
(5) Are fields distinct from spacetime manifolds? If so, should we
posit both or only one; and, if only one, which one?

Until we have answered these five questions, we cannot justifiably

adopt either the third or fourth suggested readings of the spacetime
criterion, since both readings carry commitments to substantivalism.
Questions about the nature of spacetime are matters of nontrivial
debate in the philosophy of physics and ought not be decided on the
grounds of obstinate insistence on a particular version of the abstract/
concrete distinction.
The first and second readings were shown to be free of commitments
100 SUSAN C. H A L E

to relationalism, if o n e can replace spacetime points with arbitrarily

small spacetime regions; otherwise, their a d o p t i o n carries c o m m i t m e n t
to relationalism. A l t h o u g h we have received p r o m i s s o r y notes on this
replacement, no one has yet s h o w n that it will work. F u r t h e r m o r e , there
are reasons to d o u b t b o t h that it will w o r k at all, as the p r o b l e m with
idealized mass points shows, and that it will w o r k consistently with
substantivalism, since any such replacement apparently c o m m i t s o n e to
sets, which are paradigmatically abstract entities.
These considerations, I believe, suffice to show that clarification
of the spacetime criterion is not any trivial matter. Rather, it is a
controversial and contingent matter, since it carries c o m m i t m e n t s a b o u t
the nature of spacetime which can only be justifiably decided o n the
basis of o u r knowledge of the contingent truths of physics. Given this
situation, I suggest that we would do best to stop using this version of
the a b s t r a c t / c o n c r e t e distinction and either focus o u r efforts o n trying
to decide which of the other versions I have catalogued above is best
a d o p t e d or else stop talking of the abstract and the c o n c r e t e as if this
were meaningful discourse. I suspect that it is not.

NOTES

* Earlier versions of this paper were read at seminars at The University of North
Carolina at Chapel Hill, at the North Carolina Philosophical Society meeting in
February 1986, and at the Eastern Division meeting of the American Philosophical
Association in December 1986. I am grateful for the comments I received on those,
and other less formal, occasions. My greatest debts are to Catherine Elgin, David
Lewis, William Lycan, Michael Resnik, Jay Rosenberg, and George Schiesinger.
J See Lewis, D.: 1986, On the Plurality of Worlds (Basil Blackwell, Oxford), p. 81;
Shelton, L. V.: 1980, 'The Abstract and the Concrete: How Much Difference Does This
Distinction Mark?' (Unpublished paper, read at the 1980 meetings of the American
Philosophical Association/Eastern Division), p. 1.
2 Benacerraf, P.: 1973, 'Mathematical Truth', Journal of Philosophy 70, reprinted in
Paul Benacerraf and Hilary Putnam, eds.: 1983, Philosophy of Mathematics: Selected
Readings (Cambridge University Press, Cambridge), p. 412.
3 'Mathematical Truth', p. 414.
4 As far as I knew, Lewis, On the Plurality of Worlds, pp. 81--86, and Shelton, 'The
Abstract and the Concrete: How Much Difference Does This Distinction Mark?, pp.
1--22, are the only philosophers who have noticed what a muddle this distinction is.
5 On the Plurality of Worlds, p. 85.
6 Ibid.
7 Dnmmett, M.: 1973, Frege: Philosophy of Language (Harper & Row, New York), p.
491; Field, H: 1980, Science Without Numbers (Princeton University Press, Princeton),
p. 43; Kitcher, P.: 1978, 'The Plight of the Platonist' Nofis 12, p. 119; On the Plurality
of Worlds, p. 83; Resnik, M. D.: 1980: Frege and the Philosophy of Mathematics
 THE ABSTRACT/CONCRETE DISTINCTION 101

(Cornell University Press, Ithaca), p. 26; 'The Abstract and the Concrete: How Much
Difference Does This Distinction Mark?', p. 11; Steiner, M.: 1975, Mathematical
Knowledge (Cornell University Press, Ithaca), p. 110.
s Frege: Philosophy of Language, p. 491; Science Without Numbers, p. 43; Kitcher, P.:
1983, The Nature of Mathematical Knowledge (Oxford University Press, New York), p.
102; On the Plurality of Worlds, p. 83; Mathematical Knowledge, p. 110.
9 Mathematical Knowledge, p. 131; for a discussion of this see 'The Abstract and the
Concrete: How Much Difference Does This Distinction Mark?', pp. 18-- 19.
to On the Plurality of Worlds, p. 83.
11 Ibid.
12 Ibid.
~'~ 'The Abstract and the Concrete: How Much Difference Does This Distinction
Mark?'p. 2.
14 E. M. Adams and others suggested, this to me in discussion of an earlier version of
this paper which I presented at the North Carolina Philosophical Society meeting in
February 1986.
~5 Schlesinger, G. N.: 1985, The Intelligibility of Nature (Aberdeen University Press,
Aberdeen), p. 38.
16 'The Abstract and the Concrete: How Much Difference Does This Distinction
Mark?', p. 2. Q
17 Quine, W. V.: 1981, 'Things and Their Place in Theories', in W. V. Quine, Theories
and Things (Harvard University Press, Cambridge), p. 16~
18 Goodman, N.: 1977, The Structure of Appearance, 3rd ed. (D. Reidel, Dordrecht), p.
178.
19 Science Without Numbers, pp. 10--13.
20 Ibid.., Chs. 6--8,
21 Ibid.,pp. 55--56.
= Ibid.,p. 37.
23 Ibid., p. 43.
24 Ibid.,pp. 31--32.
25 Wittgenstein, L.: 1958, Philosophical Investigations, 3rd ed. Translated by G, E. M.
Anscombe (Macmillan, New York), sec. 50, p. 25.
26 Philosophical lnvestigations, p. 25.
27 Einstein, A.: 1961, Relativity: The SlJecial and the General Theory, 15th ed. Trans-
lated by Robert W. Lawson (Bonanza, New York), p. vi.
28 David Lewis suggested this reading to me in conversation.
29 It will be noticed that this reading is merely a disjunction of readings (A) and (C). A
disjunction of readings 03) and (C). would draw the abstract/concrete distinction
equivalently, and choice between these two disjunctive readings is arbitrary.
30 Field admits this in Science Without Numbers, pp. 32--36. He argues extensively
for substantivalism in Field, 'Can We Dispense with Spacetime?, PSA 1984, 2, pp.
33--90.
31 Sldar, L.: 1974, Space, Time, and Spacetime (University of California Press,
Berkeley), pp. 166--167.
32 Whitehead, A. N.: 1957, Process and Reality (Harper & Row, New York), pp. 454--
55.
33 Process and Reality, p. 455.
34 Ibid., p. 456.
3s Ibid.,pp. 455--56.
36 Ibid., p. 4 5 4 .
37 Michael D. Resnik suggested this to me in conversation.
38 Space, Time, and Spacetime, pp. 232--233.
39 Ibid.,p. 173.
102 S U S A N C. H A L E

40 Alexander, H. G., ed.: 1956, The Leibniz-Clarke Correspondence (University of

Manchester Press, Manchester), p. 74. Quoted in Sklar, p. 174.
41 Space, Time, and Spacetime, pp. 182--183.
42 Relativity: The Special and the General Theory, p. 155.
43 Ibid.,p. 153.
44 Ibid.,p. 154.
45 Ibid.,p. 155.
46 Ibid.
47 'Can We Dispense with Spacetime?', p. 40.
48 Ibid.
49 Ibid, pp. 41--42.
50 Ibid., p. 41.
5~ Space, Time, and Spacetime, p. 220.
52 Relativity: The Special and the General Theory, p. 155.

BIBLIOGRAPHY

Alexander, H. G., ed.: 1956, The Leibniz-Clarke Correspondence (University of Man-

chester Press, Manchester).
Benacerraf, P.: 1973, 'Mathematical truth', Journal of Philosophy 70, reprinted in
Benacerraf, P., and Putnam, H., eds.: 1983, Philosophy of Mathematics: Selected
Readings, 2nd ed. (Cambridge University Press, Cambridge), pp. 403--420.
Dummett, M.: 1973, Frege: Philosophy of Language (Harper & Row, New York).
Einstein, A.: 1961, Relativity: The Special and the General Theory, 15th ed. Lawson, R.
W., ed. (Bonanza, New York).
Field, H.: 1980, Science Without Numbers (Princeton University Press, Princeton).
Field, H.: 1984, 'Can we dispense with spaeetime?', PSA 1984, 2, pp. 33--90.
Goodman, N.: 1977, The Structure of Appearance, 3rd ed. (Reidel, Dordrecht).
Kitcher, P.: 1978, 'Theplight of the platonist', Nofis 12, pp. 119--136.
ditcher, P.: 1983, The Nature of Mathematical Knowledge (Oxford University Press,
New York).
Lewis, D.: 1986, On the Plurality of Worlds (Basil Blackwell, Oxford).
Quine, W. V.: 1981, 'Things and their place in theories', in Quine, W. V.: Theories and
Things (Harvard University Press, Cambridge), pp. 1--23.
Resnik, M. D.: 1980, Frege and the Philosophy of Mathematics (Cornell University
Press, Ithaca).
Schlesinger, G. N.: 1985, The Intelligibility of Nature .(Aberdeen University Press,
Aberdeen).
Shelton, L. V.: 1980, 'The abstract and the concrete: How much difference does this
distinction mark?' unpublished paper read at the 1980 Eastern Division meetings of
the American Philosophical Association.
Sklar, L.: 1974, Space, Time, and Spacetime (University of California Press, Berkeley).
Steiner, M.: 1975, Mathematical Knowledge (Cornell University Press, Ithaca).
Wittgenstein, L.: 1958, Philosophical Investigations, 3rd ed. Anseombe, G. E. M., trs.
(Macmillan, New York).

D e p a r t m e n t o f Philosophy,
The University o f North Carolina at Chapel Hill,
CaMwell Hall O09 A ,
Chapel Hill, N C 27514,
U.S.A.